Transportation Research in India: Practices and Future Directions 9811696357, 9789811696350

Citation preview

Springer Transactions in Civil and Environmental Engineering

Akhilesh Kumar Maurya  Lelitha Devi Vanajakshi  Shriniwas S. Arkatkar  Prasanta K. Sahu Editors

Transportation Research in India Practices and Future Directions

Springer Transactions in Civil and Environmental Engineering Editor-in-Chief T. G. Sitharam, Indian Institute of Technology Guwahati, Guwahati, Assam, India

Springer Transactions in Civil and Environmental Engineering (STICEE) publishes the latest developments in Civil and Environmental Engineering. The intent is to cover all the main branches of Civil and Environmental Engineering, both theoretical and applied, including, but not limited to: Structural Mechanics, Steel Structures, Concrete Structures, Reinforced Cement Concrete, Civil Engineering Materials, Soil Mechanics, Ground Improvement, Geotechnical Engineering, Foundation Engineering, Earthquake Engineering, Structural Health and Monitoring, Water Resources Engineering, Engineering Hydrology, Solid Waste Engineering, Environmental Engineering, Wastewater Management, Transportation Engineering, Sustainable Civil Infrastructure, Fluid Mechanics, Pavement Engineering, Soil Dynamics, Rock Mechanics, Timber Engineering, Hazardous Waste Disposal Instrumentation and Monitoring, Construction Management, Civil Engineering Construction, Surveying and GIS Strength of Materials (Mechanics of Materials), Environmental Geotechnics, Concrete Engineering, Timber Structures. Within the scopes of the series are monographs, professional books, graduate and undergraduate textbooks, edited volumes and handbooks devoted to the above subject areas.

More information about this series at https://link.springer.com/bookseries/13593

Akhilesh Kumar Maurya · Lelitha Devi Vanajakshi · Shriniwas S. Arkatkar · Prasanta K. Sahu Editors

Transportation Research in India Practices and Future Directions

Editors Akhilesh Kumar Maurya Department of Civil Engineering Indian Institute of Technology Guwahati Guwahati, Assam, India Shriniwas S. Arkatkar Department of Civil Engineering Sardar Vallabhbhai National Institute of Technology Surat, Gujarat, India

Lelitha Devi Vanajakshi Department of Civil Engineering Indian Institute of Technology Madras Chennai, Tamil Nadu, India Prasanta K. Sahu Department of Civil Engineering Birla Institute of Technology and Science Hyderabad, Telangana, India

ISSN 2363-7633 ISSN 2363-7641 (electronic) Springer Transactions in Civil and Environmental Engineering ISBN 978-981-16-9635-0 ISBN 978-981-16-9636-7 (eBook) https://doi.org/10.1007/978-981-16-9636-7 © Transport Research Group of India 2022 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Contents

1

Reflections on 10 Years of TRG’s Journey . . . . . . . . . . . . . . . . . . . . . . . Ashish Verma

2

Development in Aggregate Quality Characterization Approaches for Pavement Construction . . . . . . . . . . . . . . . . . . . . . . . . . Bharat Rajan and Dharamveer Singh

3

Research Trends in Materials and Design of Asphalt Pavements . . . Nikhil Saboo and Animesh Das

4

Driving Behavior Modeling in Mixed Traffic Conditions: Developments and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sanhita Das, Bhargava Rama Chilukuri, Shriniwas S. Arkatkar, and Akhilesh Kumar Maurya

5

Pedestrian Flow Characteristics Over Different Facilities: Findings and Way Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arunabha Banerjee and Akhilesh Kumar Maurya

1

11 31

47

79

6

Emerging Traffic Data Collection Practices Under Mixed Traffic Conditions: Challenges and Solutions . . . . . . . . . . . . . . . . . . . . 101 Anuj Kishor Budhkar, Gowri Asaithambi, Akhilesh Kumar Maurya, and Shriniwas S. Arkatkar

7

Planning for Suitable Walk-Access Infrastructure Components in Various Classes of Urban Bus Stop Catchment Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 Subhojit Roy and Debasis Basu

8

Crowd Management Guidelines for Mass Religious Gatherings . . . . 151 Ashish Verma, Harihara Subramanian Gayathri, P. S. Karthika, Nipun Choubey, and Tarun Khandelwal

9

A New Framework for Comprehensive Mobility Plans in India . . . . 167 Vajjarapu Harsha and Ashish Verma v

vi

Contents

10 Transit-Oriented Development (TOD) as a Sustainable Transport Strategy for Metropolitan Cities . . . . . . . . . . . . . . . . . . . . . . 183 Manoranjan Parida, Phani Kumar Patnala, Robert Hrelja, and Ravi Sekhar Chalumuri 11 Design and Evaluation of Public and Non-motorized Transport Systems for Sustainability in Indian Cities . . . . . . . . . . . . . 203 Madhu Errampalli 12 Speed-Based Safety Evaluation of Horizontal Curves in Rural Highways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Tushar Choudhari, Gourab Sil, and Avijit Maji 13 A Global Perspective of Railway Security . . . . . . . . . . . . . . . . . . . . . . . . 233 Malavika Jayakumar, Aparna Joshi, Avijit Maji, and Prasanta K. Sahu 14 What Drives the Battery-Electric-Bus Introduction in Indian Setting: Operators Perspective and Way Forward . . . . . . . . . . . . . . . . 249 Bandhan Bandhu Majumdar, Prasanta K. Sahu, and Dimitris Potoglou 15 Sustainable Freight Transportation Planning and Policies for a Logistics-Driven India: Current State and Future Ahead . . . . . 265 Agnivesh Pani, Prasanta K. Sahu, and Bandhan Bandhu Majumdar 16 Automated Sensors for Indian Traffic: Challenges and Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 Lelitha Devi Vanajakshi and Shriniwas S. Arkatkar 17 Planning for Traditional and Special Area Requirement Travel Modes—Existing Scenario and Way Forward . . . . . . . . . . . . . 305 Udit Jain, Dharitri Kahali, Vivek R. Das, and Shriniwas S. Arkatkar 18 Air and Water-Based Transportation in India—Identifying the Research Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 Adnan Pasha and Rajat Rastogi

Editors and Contributors

About the Editors Prof. Akhilesh Kumar Maurya is currently Professor in the Department of Civil Engineering at Indian Institute of Technology (IIT) Guwahati, India. He received his Ph.D. degree in Civil Engineering from Indian Institute of Technology (IIT) Kanpur and M.Tech. degree in Computer Aided Design from Indian Institute of Technology (IIT) Roorkee. He is active academically and professionally in the area of traffic flow modeling, driver behavior, road safety audit and accident analysis, traffic data collection and analysis studies since over a decade. He has published more than 100 technical papers in international journals and conferences. He has received the DAAD fellowship at Technische Universität Darmstadt. He is currently President of “Transportation Research Group (TRG)” of India, member of World Conference on Transport Research Society (WCTRS) and the life member of Indian Roads Congress. He is also certified Road safety Auditor by International Road Federation (India) and Australian Road Research Board. Apart from several Indian institutions, he has also delivered invited lectures at various international institutes like Technische Universität Berlin, Technische Universität Darmstadt, University Duisburg-Essen and National University of Singapore. Dr. Lelitha Devi Vanajakshi is a Professor in the Transportation Division of the Department of Civil Engineering at Indian Institute of Technology (IIT) Madras. She holds a Ph.D. in the area of Transportation Engineering from Texas A&M University, USA. Her teaching and research interests are in the area of transportation systems with an emphasis on traffic flow modelling, traffic operations, and intelligent transportation systems. Dr. Shriniwas S. Arkatkar is currently serving as an Associate Professor in the Department of Civil Engineering at SVNIT Surat, India. Recently, he is appointed as ‘Adjunct Professor’ at the Department of Civil Engineering, Ryerson University, Ontario, Canada. Prior to this, he worked in the Department of Civil Engineering

vii

viii

Editors and Contributors

at Birla Institute of Technology and Science (BITS) Pilani, India. He has over 10 years of experience in teaching, research, and consultancy in the field of traffic and transportation engineering. He obtained his Ph.D. from the Department of Civil Engineering, Indian Institute of Technology (IIT) Madras. He completed his bachelors in Civil Engineering in 1999 and masters in the area of Urban Planning in 2001, from Visvesvaraya National Institute of Technology (VNIT), Nagpur, India. Dr. Arkatkar has published more than 120 research papers in journals and conference proceedings. He has experience in diverse fields of transportation such as traffic flow modelling, traffic safety, intelligent transportation systems (ITS), transportation planning, traffic operations, and traffic simulation applications. He is actively involved as Executive Secretary, Transportation Research Group of India (TRG) and as member of SIGs of WCTRS, Indian Roads Congress (IRC), Governing Council member in the Institute of Urban Transport (IUT), Ministry of Housing & urban Affairs (MoHUA). Dr. Prasanta K. Sahu is an Assistant Professor in the Department of Civil Engineering, Birla Institute of Technology and Science (BITS) Pilani at Hyderabad, India. He holds a Ph.D. in Transportation Systems Engineering (TSE) from Indian Institute of Technology (IIT) Bombay and Master’s in TSE from Indian Institute of Technology (IIT) Kanpur. Dr. Sahu is an Adjunct Professor in Price Faculty of Engineering at University of Manitoba, Canada. Dr. Sahu serves as a Handling Editor of Transportation Research Record – The flagship journal of Transportation Research Board, USA. Also, he serves as Guest Editor to Research in Transportation Economics. Dr. Sahu is the Standing Committee Member of AT015-Freight Transportation Planning and Logistics, Transportation Research Board, USA. He has been a pioneering contributor to freight transportation planning and policy in India with numerous high-quality publications aiming to improve the logistic efficiency of goods movement. Dr. Sahu is currently leading an associate research center (ARC) of Center of Excellence on Sustainable Urban Freight Systems (CoE-SUFS) at BITS Pilani. He also focuses on green mobility planning, quality of life, demand management schemes such as congestion pricing and transport emissions. Multiple projects led by Dr. Sahu in these research areas are currently ongoing under the sponsorship of funding schemes such as Global Challenges Research Fund (GCRF), Government of the United Kingdom and Scheme for Promotion of Academic and Research Collaboration (SPARC), MHRD, Asian Smart Cities Research Innovation Network—La Trobe University Australia, UK-India Education and Research Initiative (UKIERI), Government of India, Public Works Department, Government of Telangana. Dr. Sahu is a life member of Institute of Urban Transport (IUT) India and Transportation Research Group (TRG) India and Indian Road Congress (IRC). Additionally, he is a visiting faculty to University of Regina, Regina, Canada and Cardiff University, Cardiff, Wales, UK.

Editors and Contributors

ix

Contributors Shriniwas S. Arkatkar Department of Civil Engineering, Sardar Vallabhbhai National Institute of Technology, Surat, India; Department of Civil Engineering, National Institute of Technology Surat, Ichchanath, Surat, Gujarat, India Gowri Asaithambi Department of Civil and Environmental Engineering, Indian Institute of Technology Tirupati, Setttipalli, Andhra Pradesh, India; Department of Civil Engineering, Indian Institute of Technology Guwahati, Assam, India Arunabha Banerjee Department of Civil Engineering, IIT Guwahati, Guwahati, Assam, India Debasis Basu School of Bhubaneswar, Jatani, India

Infrastructure,

Indian

Institute

of

Technology

Anuj Kishor Budhkar Department of Civil Engineering, Indian Institute of Engineering Science and Technology Shibpur, Howrah, West Bengal, India Ravi Sekhar Chalumuri Transportation Planning and Environment Division, CSIR—Central Road Research Institute, New Delhi, India Bhargava Rama Chilukuri Department of Civil Engineering, IIT Madras, Chennai, India Nipun Choubey Transportation Systems Engineering (TSE), Department of Civil Engineering, Indian Institute of Science (IISc), Bengaluru, India Tushar Choudhari Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India Animesh Das Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur, India Sanhita Das Department of Civil Engineering, IIT Roorkee, Roorkee, India Vivek R. Das Department of Civil Engineering, Dayananda Sagar College of Engineering, Bangalore, India Madhu Errampalli Transportation Planning and Environment (TPE), CSIRCentral Road Research Institute (CRRI), New Delhi, India Harihara Subramanian Gayathri Transportation Systems Engineering (TSE), Department of Civil Engineering, Indian Institute of Science (IISc), Bengaluru, India Vajjarapu Harsha Department of Civil Engineering, Indian Institute of Science (IISc), Bangalore, India Robert Hrelja Faculty of Culture and Society, Department of Urban Studies, Malmo University, Malmo, Sweden

x

Editors and Contributors

Udit Jain Department of Civil Engineering, Visvesvaraya National Institute of Technology, Nagpur, India Malavika Jayakumar Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra, India Aparna Joshi Department of Civil Engineering, Birla Institute of Technology and Science Pilani, Hyderabad, Telangana, India Dharitri Kahali Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, India P. S. Karthika Transportation Systems Engineering (TSE), Department of Civil Engineering, Indian Institute of Science (IISc), Bengaluru, India Tarun Khandelwal Transportation Systems Engineering (TSE), Department of Civil Engineering, Indian Institute of Science (IISc), Bengaluru, India Avijit Maji Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra, India Bandhan Bandhu Majumdar Birla Institute of Technology and Science, Pilani, Hyderabad Campus, India Akhilesh Kumar Maurya Department of Civil Engineering, IIT Guwahati, Guwahati, Assam, India; Department of Civil and Environmental Engineering, Indian Institute of Technology Tirupati, Setttipalli, Andhra Pradesh, India Agnivesh Pani Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India Manoranjan Parida Centre for Transportation Systems, Indian Institute of Technology Roorkee, Roorkee, Uttarakhand, India Adnan Pasha Research Scholar, Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, India Phani Kumar Patnala Department of Civil Engineering, University of Manitoba, Winnipeg, Manitoba, Canada Dimitris Potoglou School of Geography and Planning, Cardiff University, Wales, UK Bharat Rajan Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India Rajat Rastogi Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, Uttarakhand, India Subhojit Roy School of Bhubaneswar, Jatani, India

Infrastructure,

Indian

Institute

of

Technology

Editors and Contributors

xi

Nikhil Saboo Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, India Prasanta K. Sahu Department of Civil Engineering, Birla Institute of Technology and Science Pilani, Hyderabad, Telangana, India Gourab Sil Department of Civil Engineering, Indian Institute of Technology Indore, Indore, India Dharamveer Singh Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India Lelitha Devi Vanajakshi Department of Civil Engineering, Indian Institute of Technology Madras, Chennai, India Ashish Verma Founding and Immediate Past President Transportation Research Group of India (TRG), Indian Institute of Science, Bangalore, India; Transportation Systems Engineering (TSE), Department of Civil Engineering, Indian Institute of Science (IISc), Bengaluru, India

Chapter 1

Reflections on 10 Years of TRG’s Journey Ashish Verma

1.1 Evolution With the completion of 10 years of TRG on May 28, 2021, it is a good time to reflect on its journey so far and the way forward. We should look back from where we started and understand: how was the journey so far, what were the lessons learnt, and what are the challenges ahead? It will not only give us a sense of value and belonging for TRG but will also motivate us to work harder in future toward its growth and fulfilling its aims and objectives. Like every important and new initiative, the idea of TRG also evolved and developed over a long period of about 3 years before it actually came into existence as a registered society. The very first meeting to discuss such an initiative of collaboration was proposed and held during First Indo-US Symposium on Advances in Mass Transit and Travel Behaviour Research (MTTBR-08), which was organized by Dr. Ashish Verma (jointly with Prof. Ram Pendyala who was with Arizona State University, USA, that time) at IIT Guwahati during February 12–15, 2008. This meeting was held on February 13, 2008, at 7:00 p.m. in the Conference Room of IIT Guwahati Guest House and was attended by 12 learned colleagues from 7 older IITs and that is where this very important idea of forming a society by the name Transportation Research Group of India (TRG) emerged with its focus on furthering transportation research and academics in the country, and it was agreed to pursue and work on it further. I still remember the delicious snacks (prepared by the guest house caterers) that I arranged before the meeting for the attendees which, I personally feel, resulted in a very positive atmosphere and led to this great idea of TRG during the meeting, clearly highlighting how important these small aspects (satisfying taste A. Verma (B) Founding and Immediate Past President Transportation Research Group of India (TRG), Indian Institute of Science, Bangalore 560012, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_1

1

2

A. Verma

Fig. 1.1 Group Photo of MTTBR-08 Delegates

buds ☺) could be. Subsequent to this, more meetings in smaller groups were held on different occasions to discuss the formation of TRG and take it further. However, this whole effort of forming TRG saw lots of ups and downs over a period of 3 years. The set of colleagues who were given initial responsibility of organizing the 1st CTRG somehow could not push it further because of which we were at ground zero again after initial momentum. But, we somehow managed to stick to the idea and believed that it will certainly take off one day. Eventually, the momentum picked up again in 2010 and the establishment of TRG was finally announced in December 2010, and it got formally registered as a society on May 28, 2011, which is remembered and celebrated by us as the Foundation Day of TRG (see Fig. 1.1). However, 2011 proved to be the most challenging year for TRG so far, after all it had to have its due share of teething problems. We were not only struggling with getting the basic setup in place for TRG, but also we had already announced the 1st CTRG to be held at Bangalore in December 2011. Adding to this was the turbulence created by a few colleagues that we saw and experienced within our fraternity, which, when I look back now, feels that every great idea has to probably go through it. In spite of all these odds, our noble intentions, motivations, and approach laid such a strong foundation of this whole idea of TRG that we successfully managed to overcome all those teething problems. I must highlight here the strong cohesiveness with which the founding members of TRG stood together and sailed the boat of TRG past the turbulent water successfully. Eventually, during December 7–10, 2011, at Taj Vivanta Yeshwantapur, Bangalore, 300 + delegates from 15 countries witnessed the emergence of a first-ever quality, regular, peer-reviewed research conference from India in the shape of 1st CTRG, and since then there is no looking back. With 2nd, 3rd, 4th, 5th, and the upcoming 6th version of CTRG, we have not only maintained

1 Reflections on 10 Years of TRG’s Journey

3

and reinforced the basic tenets of CTRG but have only taken them to higher levels. Today, with the kind of feedback and responses we have received in the past, I can say with full confidence that CTRG is among the best research conferences in the transportation domain across the world and has created its own brand image, which all Indian colleagues in transportation should feel proud of. The following have been the highlights of past CTRGs: • • • • • • • • • •

Five Conferences 2011, 2013, 2015, 2017, and 2019, upcoming 2021. The conferences covered the full domain of transportation. Consistent participation of 300–400 participants from 12–15 countries. It is fast becoming a regular pilgrimage for transport professionals in India. More than 200 double-blind peer-reviewed technical paper presentations in each conference. Past conferences have provided a wide range of Executive Courses, Tutorials, Workshops, Technical Tours, Keynote Sessions, and Special Sessions. No distinction between oral and poster presentation in terms of quality/rating of papers to make both the formats equally attractive. Selected papers are considered for possible publication in TiDE and Springer. Other papers have been published with indexed proceedings either with Elsevier or with Springer. Appreciated worldover for quality papers and keynote speakers. No frills conference, focus on high technical standards, and standard venue location and facilities (see Fig. 1.2).

Fig. 1.2 Glimpses of Past CTRGs

4

A. Verma

1.2 Checking the PDF File Further, we also managed to roll out (in October 2014) the first-ever truly international peer-reviewed journal (TiDE) coming from India in the field of transportation and published by leading international publisher Springer, again something that the transportation fraternity in India have craved for so long. Moreover, TiDE provides a quality platform for research work done on unique and challenging transportation issues of Developing Economies. A distinguished set of colleagues from across the world are leading this journal and will certainly take TiDE in the future to a level akin to the best of transportation journals. The growth trajectory of TiDE in numbers is also a reflection of its rise and stature today. In this journey, we have also successfully received support from various national and international bodies, like TRB, ATPIO, ASCE, WCTRS, EASTS, CSIR-CRRI, DST, etc. We have also signed MOUs with other international associations like ATPIO for undertaking collaborative activities and will work toward strengthening these relationships. The following is the summary of activities done by TRG in the past 10 years: • 1st CTRG, Bangalore, December 7–10, 2011. • Seminar on “Planning for Pedestrian Mobility and Safety”, March 24, 2012. • 1st Foundation Day Seminar on “Research in Sustainable Transportation Planning”, May 28, 2012. • Seminar on “Vehicle Telematics and ITS”, Hotel Lalit Ashok, Bangalore, November 30, 2012. • “High Speed Rail Symposium”, Mysore, May 9–10, 2013, in association with SJCE, Mysore. • 2nd Foundation Day Workshop on “ITS for Sustainable Transportation System and Choices”, May 28–29, 2013, in association with Intel Corporation. • TRG’s Special Session at 13th WCTR in Rio on “Status and Challenges of Transportation Research in India”, July 15–18, 2013. A Special Issue has been prepared for “Case Studies in Transport Policy”, Elsevier. • 2nd CTRG, Agra, December 12–15, 2013. • TRG’s 3rd Foundation Day Seminar on “Research & Development in Transportation”, organized by Directorate of Urban Land Transport (DULT), Govt. of Karnataka, Bangalore, in association with TRG, on May 28, 2014. • Indo-Dutch “Strategic Bi-cycle Planning Workshop”, organized by Directorate of Urban Land Transport (DULT), Govt. of Karnataka, Bangalore, in association with TRG during June 26–28, 2014, at the Golden Palms Hotel, Golden Palms Avenue, Off Tumkur Road, Bangalore. • Presentation of TRG activities at ATPIO annual meets (January 2012, 2013, 2014, 2015, 2016, 2017, 2018, and 2019) at Washington DC, during TRB annual meet. • TRG Newsletter. • Launch of TRG’s Official Journal “Transportation in Developing Economies (TiDE)”, on October 10, 2014, at Dept. of Civil Engg., IIT Delhi, (Published by Springer).

1 Reflections on 10 Years of TRG’s Journey

5

• • • • • • •

AIRDEV-2015, November 4–6, 2015, IISc Bangalore. 4th Foundation Day Seminar, May 28, 2015. 3rd CTRG held in Kolkata during December 17–20, 2015. 6th Foundation Day Seminar, May 28, 2017. 4th CTRG held in Mumbai during December 17–20, 2017. Starting of bidding process for CTRG after 4th CTRG. Establishment of 9 Technical Committees of TRG (TCTs) with effect from June 20, 2019. • 5th CTRG held in Bhopal during December 18–21, 2019. • TRG Webinar Series, May–July 2020—40 webinars delivered. • 6th CTRG to be held in Trichy during December 2021 (see Fig. 1.3). I thank all the Founding and Outgoing Board Members of TRG who gave wholehearted support to me and sacrificed their personal and professional time to work for this great endeavor of TRG. I also thank our Patrons and many transportation colleagues of Indian origin in other countries who have helped and brought support for TRG in every possible way.

Fig. 1.3 Launch of TRG’s Official Journal “Transportation in Developing Economies (TiDE)”, on October 10, 2014, at Dept. of Civil Engg., IIT Delhi

6

A. Verma

1.3 Mission, Vision, and Objectives Let’s recollect with what mission, vision, and objectives we started TRG. Mission To aid India’s overall growth through focused transportation research, education, and policies in the country. Vision • To provide a unique forum within India for the interchange of ideas among transportation researchers, educators, managers, and policymakers from India and all over the world, with the intention of covering all modes and sectors of transport (road, rail, air, and water; public and private; motorized and non-motorized) as well as all levels (urban, regional, inter-city, and rural transport) and for both passengers as well as freight movement, in India. At the same time, to also address the transportation-related issues of safety, efficiency, economic and social development, local and global environmental impact, energy, land use, equity and access for the widest range of travelers with special needs, etc. • To serve as a platform to guide and focus transportation research, education, and policies in India toward satisfying the country’s needs and to assist in its overall growth. Objectives • To conduct regular peer-reviewed conferences in India, so as to provide a dedicated platform for the exchange of ideas and knowledge among transportation researchers, educators, managers, and policymakers from India and all over the

1 Reflections on 10 Years of TRG’s Journey

•

• •

•

7

world, from a perspective which is multi-modal, multi-disciplinary, multi-level, and multi-sectoral, but with an India-centric focus. Initially, this conference has been held every 2 years; however, the frequency may change as per the decision of the society from time to time. To publish a peer-reviewed journal of good international standards that consider and recognize quality research work done for Indian conditions, but which also encourage quality research focused on other developing and developed countries that can potentially provide useful learning lessons to address Indian issues. To conduct other activities such as seminars, training and research programs, meetings, and discussions as decided by the society from time to time, toward fulfilling the mission and vision of the society. To identify pertinent issues of national importance, related to transportation research, education, and policy through various activities of the society, and promote transportation researchers, educators, managers, and policymakers in an appropriate manner to address the same. To collaborate with other international societies and organizations like WCTRS, ASCE, TRB, etc., in a manner that works toward fulfilling the mission and vision of the society.

In short, we aspire to see TRG playing a similar role in future in India, as what bodies like TRB have done in the USA in the long run.

1.4 Creation of Standing Technical Committees of TRG (TCTs) Did we work toward fulfilling our aims and objectives of TRG? I must say we did make considerable progress, but there is still a long way to go and a lot of work to be done by the hands which will carry TRG ahead in future. While we have done several activities under TRG since its inception, one very important milestone that we have achieved recently is the creation of standing Technical Committees of TRG (TCTs). The TCTs would not only handle all the technical aspects of CTRG related to their theme/track, as per the guidelines framed by TRG from time to time, but will also be expected to carry out independent technical activities related to their theme/track under the aegis of TRG and in consonance with the mission, vision, and objectives of TRG. The following are the 9 TCTs that were established by TRG with effect from June 20, 2019: TCT-A01: Pavements and materials; TCT-B01: Traffic flow theory, operations, and facilities; TCT-C01: Transport planning, policy, economics, and project finance; TCT-D01: Travel behavior and transport demand; TCT-E01: Environment (including energy) and sustainability in transportation; TCT-F01: Transportation safety and security;

8

A. Verma

TCT-G01: Transport and mobility networks (including public transportation, freight, and logistics); TCT-H01: Emerging travel technologies (ITS and IOT); TCT-I01: Other transportation modes (including NMT) and pedestrian.

1.5 Summarizing TRG’s Achievements in 10 Years The following points summarize what we managed to do so far: • Started both, a regular peer-reviewed research conference and a peer-reviewed international journal. • Establishment of standing technical committees of TRG (TCTs). • Managed support and association with major national and international transportation research-related bodies. • Improved networking and collaboration of transportation professionals in India and abroad. • Focused workshops, seminars, and brainstorming activities together with government and private entities on specific topics of relevance to the country and practice. • Representation of TRG and transportation research activities in India at various international forums like WCTR, ATPIO annual meets, special issue publications in journals, etc.

1.6 Areas to Work On What we have achieved so far is only a good start. There are still many areas that we need to work/improve upon. Some of them are highlighted as follows: • Improving the membership of TRG. • Improving the financial support from various entities to sustain the activities of TRG. • Improving collaboration and engagement with government and private organizations for various activities of TRG. • Developing guidelines and standards for India on various aspects of the transportation domain. • Setting up a permanent Secretariat office and support staff for managing administrative works of TRG. • Contributing to nation-building through our collective expertise in our subject domain of transportation.

1 Reflections on 10 Years of TRG’s Journey

9

1.7 Vision for Future • TRG should eventually become an advisor to the nation in every aspect and domain of transportation. • Through its collective wisdom and activities, TRG should guide the innovation, research, and capacity building in the country in the full domain of transportation. • Taking the lead, TRG should act as a platform for transportation research and capacity building in Developing Economies, not just India. • TRG should provide good decision support to various government programs related to transportation, and engage with government agencies, in this regard, at all levels.

Chapter 2

Development in Aggregate Quality Characterization Approaches for Pavement Construction Bharat Rajan and Dharamveer Singh

2.1 Introduction The rapidly changing development activities across the world lead to an increase in tire pressure, gross vehicle weight, heavy traffic volumes, and introduction of a new vehicle with unusual axle configuration on pavements. Notably, aggregate as the largest volumetric fraction among pavement materials plays a key role in the durability of these new generation demand-based pavements. The word ‘aggregate’ is applicable for a wide range of materials, which includes fragmented quarried rock, gravel, sand, and recycled/byproduct like recycled concrete aggregate (RCA) and recycled asphalt pavement (RAP), overburnt brick, slag (i.e., from steel and blast furnace), and quarry byproducts (i.e., marble waste, and stone dust). Markedly, the type, size, and quality of aggregate are the predominant factor in the decisionmaking of layer-specific (i.e., wearing course, base course, or sub-base course) use of aggregate in a pavement system. Importantly, for a given aggregate layer the quality of selected coarse and fine aggregate plays an important role in the overall durability of the pavement system. Moreover, pavement durability is a cumulative sum of individual layer performance. For example, the poor quality of aggregates in the base and sub-base layers leads to rutting, depressions, corrugations, fatigue cracking, longitudinal cracking, and frost heave in the flexible pavement, whereas it contributes to faulting, pumping, settlement, and corner breaking in rigid pavements. Likewise, the aggregate in the top bituminous layer plays a critical role in offering a rut-, crack-, and moistureresistive layer. Therefore, considering aggregate’s dominating proportion and significant effects on attaining desired layer properties, the highway agencies around the globe have diverse characterization methods and respective limiting criteria. The B. Rajan · D. Singh (B) Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_2

11

12

B. Rajan and D. Singh

present chapter first provides a brief introduction to different sources/types of aggregates available for pavement construction. Further, the chapter is primarily categorized into two broad sections. The first section offers an extensive discussion on past and presently available conventional aggregate characterization practices, typical limiting values, and significance. The second section focuses on the application of digital image technique in aggregate characterization and its futuristic importance for the production of quality aggregates. It is expected that the detailed information on aggregate quality and characterization methods may lead to better quality control and durable pavement construction.

2.2 Aggregate Source and Type Based on source and type, the aggregates are primarily classified into four different categories, namely stone deposit, sand and gravel deposit, industrial byproducts, and manufactured aggregates (Table 2.1). Generally, the stone deposits are the prime source of aggregate for pavement construction, however, there is limited availability of natural aggregates in some regions (i.e., areas like river basins), which either requires transportation of ‘good quality’ aggregates or gravel aggregates. The gravel aggregates are natural material deposited in rivers, glacial, and mines, and have comparable mechanical properties as of crushed stones (Nikolaides 2014; Singh et al. 2019). In addition, nowadays the pavement industry is looking forward to using slag and recycled aggregate as partial/full replacement in pavement layers. Generally, considering the transportation associated cost, the use of byproducts is encouraged in the proximity of plant/industry. Notably, the location and source of aggregate play a major role in cost-effective pavement construction. The artificial aggregates from plastic, crumb rubber, and any other form of lightweight aggregates (LWA) are commonly used in concrete and as a part of partial replacement in slope stabilization. Table 2.1 Source/type-based classification of aggregates Aggregate source/type

Remarks

Stone deposit (originated from rocks)

Igneous (granite, basalt, rhyolite, etc.); Sedimentary (limestone, sandstone, and dolomite); Metamorphic (quartzite, marble, slate, and gneiss) rocks

Sand and gravel deposit

Historic/active river valley; Glacial deposits

Byproducts

Overburnt bricks; Blast furnace slag (BFS) and Steel slag (SS)

Artificial/manufactured

Plastic aggregate; Crumb rubber aggregates; Lightweight aggregates (LWA)

2 Development in Aggregate Quality Characterization …

13

2.3 Conventional Aggregate Characterization Approaches The section provides detailed information on conventional aggregate properties, their role in pavement systems, and available past and present characterization approaches and respective typical limiting values. Conventionally, the aggregate properties are categorized in two ways: (i) Source properties, i.e., basic, toughness and strength, hardness, soundness, and deleterious material; (ii) Consensus or Shape, i.e., Flatness and Elongation (F&E), Fine Aggregate Angularity (FAA), Coarse Aggregate Angularity (CAA) and sand equivalent test ([Mix Design Methods for Asphalt Concrete 2014]; Fig. 2.1).

2.3.1 Source Properties The source properties are source-specific and primarily change with rock source and geological composition. Moreover, considering the source-specific dependence, no global consensus on the critical values is drawn. Thus, the specified values are more or less drawn by local authorities/agencies as per requirement. The source properties play a significant role in asphalt mix design, therefore, the source properties are considered as a primary and key level filter in the selection of aggregate source. The study categorizes source properties into five different sub-categories, namely basic, toughness and strength, hardness, soundness, and deleterious material.

2.3.1.1

Basic Characteristics

The basic properties include specific gravity, water absorption, and Atterberg limits of aggregates. Notably, the basic properties are directly related to volumetric and

Fig. 2.1 Conventional characterization of aggregates

14

B. Rajan and D. Singh

design parameters of asphalt and concrete mixes. For example, in asphalt mixes the water absorption and apparent specific gravity of aggregates are directly related to understanding the amount of effective binder available in asphalt mixes. Likewise, the absorption also affects the W/C ratio and optimum water content in cement and degree of compaction for stabilized mixes, respectively. In general, the aggregates with a water absorption limit of 2% and 1% are requisites for use in surface courses of road and airport pavements, respectively (MoRTH 2013; IRC 2019). The Atterberg limits control compressibility, permeability, and strength behavior of soil or fine particles ( 50 and SE > 40–50, respectively). Notably, the European specification enlists the SE value for base and sub-base layers (i.e., SE > 45), whereas interestingly the quality of fine aggregate in US and Indian standards for these layers are controlled by Atterberg limit requirements (i.e., LL > 25 and PI < 4–6; [MoRTH 2013; Saeed et al. 2001]). Also, few practices exempt slag aggregates from SE requirements, whereas lower the SE values for limestone aggregates. It is worthy to note that most conventional shape characteristics are measured for the blend of aggregates, and measurement approaches are empirical in nature. Additionally, the tests with void-based measurements cumulatively capture the effects of angularity, texture, and dimension distribution. Thus, in order to understand each individual parameter separately, the industry has developed a number of imagebased methods. Notably, the image-based methods offer an unparallel advantage over conventional methods. The detailed comparison in conventional and image-based aggregate characterization is presented in Sect. 2.5.

2.4 Image-Based Advance Aggregate Characterization The application of digital image techniques for aggregate characterization provides a number of new shape/consensus properties like angularity, texture, sphericity, flatness and elongation, and form2D. These shape indices are characterized on three different scales, namely texture, angularity, and dimensional parameters (i.e., sphericity, flatness, elongation, flatness, and elongation). Figure 2.2 shows different aggregate shape characteristics and scale of measurement. Texture is the measure of aggregate surface roughness. The aggregate surface roughness has an important role in aggregate–asphalt bond and directly affects the moisture resistance characteristics (i.e., raveling, bond strength) and fatigue characteristics of asphalt mixes. Mishra and Singh ( 2021) evaluated the effects of aggregate surface roughness on bitumen bond strength using the PAATI test and concluded that

XAxis

Scale of Measurement

Angularity

Length

Sphericity

Y-Axis

Texture

Sphericity F&E Form2D Angularity

Texture Width Fig. 2.2 Aggregate shape indices in digital image technique and scale of measurement

2 Development in Aggregate Quality Characterization …

21

bond strength increases with an increase in surface roughness. Additionally, aggregate surface texture has a significant effect on the frictional performance of pavement surfaces (Sengoz et al. 2014; Greer and Heitzman 2017). Angularity quantifies the sharpness of corners. The angularity of a given aggregate affects its interlocking with surrounding aggregate in the asphalt/cement mix matrix. An aggregate system with better interlocked aggregates not only affects its performance against dynamic loading but also plays an important role in the grain-to-grain load transfer mechanism associated with the flexible pavement. The angularity is directly associated with rut and shear-resistant behavior aggregates and asphalt mixes (Rajan and Singh 2017; Su and Yan 2018). In the case of fresh cement concrete, the angularity of aggregate particles affects the paste rheology, workability, and pumpability. The aggregate with high angularity requires more W/C ratio to provide uniform fresh concrete properties compared to low angular particles (Neville 1995; Jamkar and Rao 2004). The dimension parameters (i.e., sphericity, flatness, elongation, flatness, and elongation) are derived parameters from aggregate dimensions. Sphericity/Cubicity are three-dimensional characteristics of aggregate and are measured by capturing the mass distribution uniformity along three axis (i.e., X, Y, and Z). The aggregates with high cubicity not only provide enhanced load-bearing capacity but also offer more resistance against dynamic loading. Moreover, the flatness and elongation (F&E) is characterized using aggregate dimension ratios (width, height, and length). The flaky and elongated particles are undesirable because of generating anisotropic characteristics and their higher breakage potential (refer to Sect. 3.2.3). It is important to note that the development of anisotropic characteristics and breakage of flaky and elongated particles increase the rutting and moisture damage potential of asphalt mixes (Little et al. 2018). The sphericity and F&E characteristics are vice versa to each other (Rajan and Singh 2017; Rajan and Singh 2020). Form2D is applicable for fine aggregates only. It quantifies the variation in the radial distribution of fine aggregates. Alike angularity, form2D also affects the workability and fresh paste properties of asphalt and cement concrete mixes, respectively. Over the years, with the advancement in material characterization, a number of different computer vision-based aggregate shape characterization techniques are developed. These vision systems use mathematical algorithms to calculate different numerical shape indices. For example, the methods like Videographer40 (VDG-40), University of Illinois Aggregate Image Analyzer (UIAIA), Aggregate Image Measurement System (AIMS), Laser-based Aggregate Analysis System (LAAS), and CT-Scan are a few most common methods used in aggregate shape characterization.

2.4.1 Videographer-40 (VDG-40) In 1994, Laboratoire Central des Ponts et Chaussees (LCPC) developed a VDG40 videograder to rapidly capture the aggregate gradation for a large quantity of aggregate samples. The method is applicable for particle sizes greater than 1 mm.

22

B. Rajan and D. Singh 0.25m

Aggregate Feed Bin Controlling Wheel

Front Camera

Z-axis

Plan View

Side Camera

X-axis

Top Camera

Light Source Linear CCD Camera

Aggregate Feed Bin Side View

Z-axis Y-axis

(a)

Conveyer belt = 8cm/sec

(b)

Fig. 2.3 Schematics of a VDG-40 and b UIAIA setup

The device is optoelectronic, computer-controlled, and semiportable, which was primarily developed for gradation analysis; however, it can calculate only dimensional characteristics such as Slenderness Ratio (SR) (aggregate’s length to width ratio) and Flatness Factor (FF) (aggregate’s width to thickness ratio). The aggregate is placed in a hopper and moves using a conveyor belt and rotating wheel. The wheel is used to control the particle rate of drop (Fig. 2.3a). The system uses a line scan charge-coupled device to capture the black particle image using backlit. In the next stage, an ellipse having a similar length and width is fitted to the particle, and SR is calculated. The FF is the property of a group of aggregate samples tested. The multiplication of SR and FF results in a ratio of maximum to minimum dimension, which is also known as F&E or aspect ratio. Prowell and Weingart (1999) showed that VDG-40 has a good correlation with manually measured flatness and elongation of particles.

2.4.2 University of Illinois Aggregate Image Analyzer (UIAIA) The UIAIA captures individual particles with three orthogonal progressive scan digital cameras. The UIAIA has a video camera along all three axes and captures particle silhouettes from all three directions (Fig. 2.3b). These cameras capture monochrome images which are further converted into binary images. The binary images are used for extracting pixel coordination, which helps in defining the mathematical-based shape indices and quantification of morphological (i.e., shape) parameters through Angularity Index (AI), Surface Texture Index, and F&E ratio. The weighted average of all three indices is presented as a final shape index for a given aggregate particle (Rao et al. 2001).

2 Development in Aggregate Quality Characterization …

23

2.4.3 Aggregate Image Measurement System (AIMS) The recent AIMS is developed in two stages, namely AIMS 1 and AIMS 2. The AIMS was developed during the initial stage of research, where the possibility of digital image technique in aggregate shape characterization was ascertained. In AIMS 1, the aggregates are arranged on a backlit viewing table and digital images are acquired with a vertically attached camera. Masad (2003) provided the basic design and Masad (Masad and Fletcher 2005) explained the associated mathematical details. Although AIMS 1 proved the DIT concept is feasible for aggregate shape characterization, it was cumbersome and expensive (Gates 2010). AIMS 2 uses image acquisition hardware consisting of a single video camera along the vertical axis, lightning (upper lamp and bottom lamp), microscope, and motion control image analysis software (Fig. 2.4a). AIMS 2 has two different advanced perspectives compared with AIMS 1. 1.

2.

Enclosed Testing Chamber: The AIMS 2 has an enclosed chamber, which provides better control of external lightning conditions on grayscale images (used for texture and particle height calculation) compared with AIMS 1. Circular Loading Tray: The addition of a circular loading tray not only provides better control over individual particle positioning but also offer size-specific grove for different sizes of aggregates.

AIMS measures shape parameters for a wide range of aggregate sizes (passing from 37.5 mm to retaining on 0.075 mm). The coarse aggregates (>4.75 mm) are scanned for angularity, texture, and other particle dimensional characteristics (i.e., F&E and sphericity), while fine aggregates are for form2D and angularity. In general, three consecutive scans are conducted on coarse aggregates to capture silhouette Motion and Other Control Line

Microscope

Laser Scanner

Bottom Lamp

Laser Plane

Top Light Loading Tray

Aggregates Aggregate Positioning Groove

Black Particle Image

(a)

Gray Scale Image

Scanning Platform

Support Frame

(b)

Fig. 2.4 Measurement of aggregate shape in a AIMS 2 and b LASS setup

Data Acquisition and Control

24

B. Rajan and D. Singh

(black particle image: Fig. 2.4a), texture, and height, whereas the fine aggregates are subjected to only black particle image scans. The black particle image (silhouette) is used for angularity, form2D, and aggregate dimension distribution analysis, and a grayscale image is used for texture and height determination (Fig. 2.4a). Table 2.3 shows the typical algorithm and defined AIMS ranges for different shape characteristics. The detailed information on specific algorithms can be found with Rajan and Singh (2018, 2019, 2020).

2.4.4 Laser-Based Aggregate Scanning System In a Laser-based Aggregate Scanning System (LASS), a laser scanner is fitted vertically on a scanning platform and a laser scanner slides along the scanning platform using a linear motion slider (Fig. 2.4b). LASS performs the data acquisition for aggregate size ranging from 1 to 100 mm with 25 scan/s. The laser scanner projects a strip of 120 mm width on the scanning platform, and the reflection of the laser is captured via a CCD camera. With the help of laser source location, the three-dimensional coordinates of the particle surface can be calculated. The LASS can be used to capture aggregate angularity, texture, and grading (Kim et al. 2001). A similar laser-based aggregate morphological characterization was performed by Komba et al. (2013) and Ge et al. (2018) using a portable three-dimensional laser scanner. It is worthy to note that the laser-based methods are quick and less costly compared with identical 3D-based CT scan measurements.

2.4.5 X-Ray CT Scan The CT scan approach scans the aggregate in various parallel two-dimensional layers and reconstructs it in 3D. Typically, a CT scan consists of an X-ray generator, full digital X-ray array detector, collimator, and bearing turntable. In general, the scanning interval of 0.1 mm in the vertical direction and pixel size of 0.13 × 0.13 mm are adopted. Further, these sliced two-dimensional images are processes with binarization, and each pixel position is captured in reference to the aggregate area (i.e., inside area or in air/outside) (Yang et al. 2017). Markedly, due to their exact reconstruction of aggregate, the 3D-based methods are considered as most accurate to characterize aggregate shape indices.

Fine Aggregate

Coarse Aggregate

Form2D

F&E

Sphericity

Texture

2100

5400

High

3975

Moderate

Extreme

10000

200

Moderate 500

High 750

1000

Extreme

0.5 0.6

Moderate High

0.8

Extreme

1.0

dL dS

0

Low

6.5 8

High 10

Moderate

Extreme

20

• Relative form of particle compared with circle (in 2D)

• F&E =

• Ratio of longest dimension to shortest dimension

0

Low

• The distribution of particle mass along three axes (i.e., x, y, and z)

0

Low

• Roughness of aggregate surface ( 1.9 m: subject vehicle drives at the maximum expected speed without any influence from the adjacent vehicle. Taking into account the lateral separation characteristics on a single lane, Jin et al. (2012) considered time to collision (T.T.C.) into the optimal velocity model. At a later date, He et al. (2015) extended the optimal velocity model by considering overtaking

4 Driving Behavior Modeling in Mixed Traffic Conditions …

51

expectation through the escape corridor (Gunay 2007). The overall simulation results demonstrated that the proposed model performed better than the Optimal Velocity model and Intelligent Driver Model. Using the potential attention field of drivers, Tao et al. (2014b) established a driving behavior model by reflecting the driver’s perception of the surrounding traffic environment with the changing field intensity value. The field intensity is a function of the driver’s characteristics and state of motion of its vehicle. Numerical results indicated that the model could simulate traffic flow operations quite realistically. The lateral effect was further embodied in several works in which the lanes were classified into ‘sublanes’ or ‘strips’. Mathew et al. (2015) proposed a stripbased modeling approach to model the mixed traffic environments using trajectory data collected in Mumbai, in which a lane is divided into multiple strips. Depending on the width of the vehicle, vehicles can occupy multiple strips. Such modeling approach, however, fails to capture the lateral movements of Motorized Two-Wheelers (MTWS). Although the lateral gap maintaining behavior between different vehicle combinations has been investigated in several studies empirically (Mallikarjuna et al. 2013; Pal and Chunchu 2019; Budhkar and Maurya 2017; Siddique 2013), they do not deal with any driver behavior model. The variable lateral clearance maintaining behavior on mixed bicycle dynamics was introduced in cellular automata models (Feng et al. 2015) with an occupancy rule involving lateral gap information. A similar concept was also utilized in Luo et al. (2015) work considering interactions between bicycles and cars. Several researchers have proposed modifications in the conventional car-following models by incorporating the effects of lateral separation. Yet, these models still lack validation against a real-world database. Conversely, the strip-based or sub-lanebased approaches could capture heterogeneity in vehicles quite precisely (Mathew et al. 2015; Luo et al. 2015), but deciding a strip size or a cell size is more crucial. More precisely, there has been growing attention towards consideration of lateral separation effects in recent years. Anand et al. (2019) showed that the car-following model parameters vary not only by subject vehicle type but also by leader–follower pairs. In addition, there is a significant effect of factors such as the lateral position of vehicles and types of following behaviors. Later, Kashyap et al. (2020) used vehicle trajectory data to identify vehicle pairs in the following regime and the regime duration, classify pairs as strict and staggered following, and investigate the factors influencing the following vehicle’s speed under different regimes. However, the interdependence between longitudinal and lateral separations in a staggeredfollowing scenario and the applicability of the modified conventional models in such disordered systems remain unexplored.

52

S. Das et al.

4.2.2 Handling Multiple Leaders in the Vehicle-Following Scenario Although many advancements have been made in the existing conventional C.F. models to model staggered-following behavior, a series of models have now emerged to incorporate more than one leader case, mainly in the context of behavioral modeling of non-lane-based systems. This section, therefore, provides a systematic review of ongoing research work conducted for modeling multiple-leaders following behavior in non-lane-based disordered systems. Early research on multiple-leaders following dates back to the early 2000s. Cho and Wu (2004) developed a motorcycle traffic flow model based on the concept of thrust and repulsion. The desired speed of the following vehicle gives the thrust to proceed while the leading vehicle gives repulsion to the subject vehicle. It is based on the assumption that only the nearest left-side leader and right-side leader offer repulsion to the subject vehicle. A weight function is accordingly used to capture the lateral separation effects in the model. This model can handle single-leader as well as two-leaders following behavior in the longitudinal movement models. However, calibration and validation of the model have not yet been reported. At a later date, Nguyen (2012) proposed a safety space model to describe the dynamic behavior of MTWS in the presence of surrounding vehicles. The vehicle is considered the most influential if the subject vehicle responds to it with the maximum magnitude of acceleration. According to Mathew et al. (2015), the leader is identified as the vehicle among all possible leaders closest in the distance to the subject vehicle. If it is a tie, the one that has been identified first is considered the leader. But in such situations, both the leaders may have a similar impact on the subject vehicle and may be considered critical leaders. Choudhury and Islam (2016) argued that in the presence of multiple leaders (left-front, direct-front, and right-front) in mixed traffic streams, the driver in the following vehicle is subjected to multiple sources of stimulus for acceleration and responds to the stimulus of the influential leader. The G.H.R. modeling framework was used for the acceleration component of the leader, and calibration of the model was being done using trajectory data of Dhaka. They found that the probability of a given front vehicle being the governing leader depends on space headway, amount of lateral overlap, type of subject vehicle, and relative speed. On the other hand, Papathanasopoulou and Antoniou (2017) argued that the closest vehicle in the direction of motion is chosen as the most critical leader. Asaithambi et al. (2018) defined the governing leader as the vehicle that overlaps laterally with the subject vehicle within the look-ahead distance. Similar to the concept of Cho and Wu (2004), a data-driven locally weighted regression model was developed by Papathanasopoulou and Antoniou (2018) to estimate vehicle speeds for mixed traffic conditions. In a recent study by Raju et al. (2018), based on visual assessment, they first assumed leader–follower pairs as the ones that are separated laterally by 1.5 m from the edge of the leader. Subsequently, based on relative distance versus relative speed plots of the assumed pairs, those vehicle combinations are identified as the true

4 Driving Behavior Modeling in Mixed Traffic Conditions …

53

leader–follower pairs that exhibit hysteresis phenomenon (a common characteristic of a vehicle following process) among different combinations of vehicles. In the context of handling multiple leaders, the optimal velocity model proved to be the most prominent one among all. Li et al. (2015) proposed a full velocity difference model considering the effects of two-sided lateral gaps on the behavior of the following vehicle. The proposed model considers three leaders: left front, direct front, and right front vehicles. Accordingly, two-sided lateral gaps (one with the L.F. vehicle and the other with the R.F. vehicle) are only considered in the model. Later, Li et al. (2016) proposed a new car-following model considering the effect of visual angle and a lateral gap with the left-front and right-front vehicles. Numerical simulation results verified the impacts of a lateral gap on traffic flow with respect to smoothness and stability. The two-sided lateral gap model proposed by Li et al. (2015) was further extended by considering the effect of a gradient in addition to the lateral gaps (Li et al. 2018). Xu et al. (2018) addressed the lateral gaps with the left-front, right-front, and direct front leaders and proposed a new car-following model. Linear stability analysis illustrated that the proposed model had an enlarged stable region compared to the two-sided lateral gap car-following model. Recently, Das (2020) developed an artificial network model to predict responses of cars interacting with two leading vehicles simultaneously. Utilizing trajectory data extracted from an instrumented vehicle study, a two-leader car-following model was proposed considering longitudinal and lateral spacing with both the vehicles in-front, relative speed, subject vehicle speed as input and acceleration of the following vehicle as the response of the driver. The developed neural network model could illustrate realistic human-like following behavior of drivers, much better than the tested classical optimal velocity-based models in terms of trajectory reproducing accuracy. The proposed model could also explain the closing-in, shying-away behavior, and local stability properties. A synthesis of the literature illustrates that most of the research has focused on identifying a single governing leader among multiple vehicles ahead; lateral overlap and the longitudinal gap being considered the influential criterion in most of the work. Yet, the presence of more than a single leader may have a combined behavioral effect on the following process of vehicles. Indeed, this aspect has been acknowledged in the series of modifications made in the optimal velocity models, in which the lateral gaps with each possible group of leaders were incorporated in the models. However, empirical calibration of the developed models has not yet been explicitly undertaken, which still requires validation against the real database.

4.2.3 Comprehensive Models Modeling driver behavior in disorderly systems requires a framework for modeling the two-dimensional movement of vehicles, where vehicles search for possible opportunities to proceed through available gaps while maintaining a safe distance with the surrounding vehicles. Chakroborty et al. (2004) proposed a comprehensive modeling

54

S. Das et al.

framework based on the assumption that a driver’s behavior can be specified by choice of steering angles at small time intervals and acceleration/deceleration rates offered by the driver. Using the concepts of potential field theory, the potential emanated from roadway and traffic features is incorporated in the model to denote the attractive and repulsive potentials. The total potential at a point is assumed to represent the driver’s threat to safety. The developed model could perform satisfactorily corresponding to free-flow and two-way vehicular movements on undivided roads. However, calibration of the parameters becomes computationally intensive and could not be validated against a real database (field data). Later, Maurya (2007) developed a comprehensive two-dimensional model for heterogeneous uninterrupted traffic streams. The model consists of two modules— one for the lateral movement based on steering angle and another for longitudinal movement in terms of acceleration/deceleration rate. The lateral control module defines the best path based on several goodness values, including maximum available distance headway, obstacles, steering angle requirements, etc. Contrarily, the longitudinal movement module determines the driver’s actions in terms of anticipated relative speed, relative acceleration, and deviation of available distance headway from the stable distance headway (distance headway at which the drivers feel safe at a given speed). Although this model can replicate the disorderly traffic flow phenomena quite well, it lacks calibration of the parameters against real-world traffic databases. Nguyen and Hanaoka (2011) put forward the concept of social forces to describe the integrated movement of motorized two-wheelers (MTWS). These social forces are the forces that make the subject vehicle change direction and speed in order to arrive at specific targets or to avoid collisions with other vehicles or objects using which the velocity of each vehicle can be calculated. The social force model is defined by an acceleration force leading the MTWS to run in the desired direction towards the destination, a repulsive force from the surrounding vehicles, and a repulsive force from infrastructural boundaries. This model was further extended by Huang et al. (2012), considering the mechanical dynamic attributes of MTWS. Huynh et al. (2013) proposed the social force model considering the attractive force representing the formation of groups at signalized intersections. Another study in a mid-block section in uncongested situations was carried out by Mullakkal et al. (2015), in which they introduced the concept of perception lines. Vehicles within the perception lines of the subject vehicle are considered to influence the motorcycle movement. Using the trajectory data set from Eastern Expressway, Mumbai, the proposed model was validated and then compared to the simulated trajectory. Though the social force model can well describe the two-dimensional movement of vehicles, the complexity arises in the calibration of many parameters associated with the components of each force. The number of parameters increases when vehicletype dependent driver behavior modeling is addressed. Recently, Sharath and Velaga (2021) extended the Intelligent Driver Model (IDM) with extra parameters to model the two-dimensional (lateral and longitudinal) motion of vehicles in mixed traffic. The results showed that the proposed model better captured the longitudinal and lateral gaps in mixed traffic conditions.

4 Driving Behavior Modeling in Mixed Traffic Conditions …

55

The concept of safe space was defined by Nguyen et al. (2012) to describe the uncontrolled movements of motorcycle-only traffic under congested conditions. Safety space for a motorcycle has the form of an approximate half-ellipse. The boundaries of which are equipotential lines, meaning that all vehicles on the line have the same level of safety as perceived by the subject vehicle. The safety space model can explain the acceleration, oblique following and swerving maneuvers of the MTWSs in a congested traffic stream. However, such models cannot reproduce other behaviors of MTWS traffic, such as filtering and grouping behavior.

4.3 Macroscopic Modeling Approaches Macroscopic approaches to traffic flow modeling of mixed traffic conditions are recently gaining attention due to its wide range of corridor-level and network-level analyses, management, and control applications. However, the traditional lane-based definitions of traffic flow variables such as density, flow, and speed are inadequate to characterize the mixed traffic conditions (Chari and Badarinath (1983; Khan and Maini 1999). To overcome these limitations, these definitions were extended for area definitions (Chari and Badarinath 1983; Mallikarjuna and Rao 2006). These definitions would mean that there are longitudinal and lateral components for speed and flow and area measurement for density (Chaturvedi et al. 2021). However, such extended definitions are only meaningful with the associated traffic composition definition. To overcome this limitation, the concept of area occupancy was proposed (Bham and Benekohal 2004; Mallikarjuna and Rao 2006). It is shown that area occupancy better represents the bivariate relationships or fundamental diagrams in mixed traffic conditions (Arasan and Dhivya 2008). Another approach attempted in the literature to overcome the challenges of mixed traffic was to group multiple vehicles into a homogeneous group using passenger car units (PCU). Several studies estimated PCU values for heterogeneous traffic conditions [e.g., Chandra and Kumar (2003), Mallikarjuna and Rao (2006), Arkatkar and Arasan (2010), Mehar et al. (2014), Nokandeh et al. (2016)]. The studies showed that the PCU Values are not static and are dependent on several factors such as traffic, geometric, and control conditions. Basu et al. (2006) and Nokandeh et al. (2016) showed that traffic volume and traffic composition are the two prominent factors that can govern the PCU values. Chandra and Kumar (2003), Arasan and Arkatkar (2010), Gautam et al. (2018) showed that geometric features, such as carriageway width and grade, also affect the PCU values. The fundamental diagrams are the steady-state relationships between the traffic flow variables. In the 1930s, Greenshields (1934) developed the first unimodal relationship between traffic flow and density on links, as shown in Fig. 4.2, commonly referred to as the fundamental diagram (F.D.). The F.D. clearly distinguishes between free-flow traffic (the increasing portion of the curve) and congested traffic (the decreasing portion of the curve). However, Greenshields’s model and similar disaggregate link or vehicle-based modeling efforts were derived from data measured for

56

S. Das et al.

Fig. 4.2 Fundamental diagram

a single link. They were not sufficient to describe the entire complexity of traffic dynamics on urban road networks. These diagrams and their parameters, such as saturation flow, jam density, wave speed, free-flow speed, etc., are widely used in traffic analyses. Some of these variables are extensively studied for mixed traffic conditions. Earlier methods [e.g., Sarna and Malhotra (1967), Bhattacharya and Bhattacharya (1982)] estimated saturation flow (in terms of PCU) as a function of road width. Arasan and Jagadeesh (1995) adopted a probabilistic approach by incorporating the interaction between different types of vehicles through their intercorrelations to estimate the saturation flow. Some of the later studies attempted to include the impact of vehicle composition and vehicle classes such as motorcycles and trucks on the saturation flow [e.g., Minh and Sano (2003), Lewis and Benekohal (2007)]. Recently, Biswas et al. (2016) incorporated PCU in the estimation of the saturation flow. One of the challenges of the saturation flow measurement in mixed traffic is estimating the duration of the saturation flow. Towards this, Das et al. (2020) investigated queue discharge characteristics like discharge headway using cumulative count curves to estimate the saturation flow duration accurately. Recently a significant amount of effort has been invested in estimating saturation flow and capacity of different transportation facilities from India (Indo-HCM 2018). Regarding other fundamental diagrams, Bhavathrathan and Mallikarjuna (2012) proposed a diminishing density concept to estimate jam density in mixed conditions. Similarly, Vinaya and Chilukuri (2019) developed a regression model for wave speed at the signalized intersection using vehicle composition. Balakrishnan and Sivanandan (2015) studied the variation in free-flow speed across different vehicle subclasses and found that the vehicle’s lane position affects it in mixed traffic. These models for fundamental diagram parameters can be used in the newly developed link cost and capacity functions for mixed traffic conditions (Das and Chilukuri 2020). Fundamental diagrams in mixed traffic conditions are lately gaining attention. Thankappan and Vanajakshi (2015a, b) evaluated various homogeneous fundamental diagram functions for mixed traffic conditions and proposed a two-regime model for mixed traffic conditions. Gaddam and Rao (2019a, b) proposed new speed–density functional forms for the mixed traffic conditions based on empirically observed saturation flow, jam density, and wave speeds. Recently, Mohan (2021) investigated

4 Driving Behavior Modeling in Mixed Traffic Conditions …

57

the speed–concentration curves and found that the perceived area occupancy better captures the speed variation observed in mixed traffic simulations. Gore et al. (2020) developed an arterial FD for an urban corridor using data from Wi-Fi detections and videography for mixed traffic conditions present in India. The authors developed arterial FD for separate days to understand the variation in flow and speed for a given density value. This variation was used to analyze and quantify the stability of traffic flow. Gupta and Katiyar (2006) proposed a continuum model with a speed gradient term and extended it for heterogeneous vehicle classes. The results showed that the model can explain the platoon dispersion phenomenon in mixed traffic. Later, Mohan and Ramadurai (2017) developed a second-order macroscopic continuum model using area occupancy and extended it for multiple vehicle classes. Numerical simulations showed that the model can explain the heterogeneous traffic features such as platoon dispersion. Sreekumar and Mathew (2020) proposed traversable distance to model the passing of smaller vehicles through the gaps, as observed in the multi-class disordered traffic streams. Gaddam and Rao (2020) recently developed a two-sided lateral gap continuum model consisting of a disturbance propagation speed, viscosity term, and a frictional clearance term to describe the complex vehicular interactions in traffic streams. While these models capture some of the important features of mixed traffic, several features unique to mixed traffic, such as multiple leaders, vehicular interactions, group movement, etc., need further analysis and modeling. Godfrey (1969) was the first to propose a unimodal relationship between average vehicle speed and average vehicle flow, to describe the traffic conditions measured across an entire network using the same basic shape as Fundamental Diagram. The concept of the macroscopic fundamental diagram (MFD) was revisited by Daganzo (2007). The MFD represents a consistent and reliable relationship between the average flow and density within a defined network system. Daganzo (2007) provided the theoretical existence of a well-defined relationship between urban traffic variables. The study postulated that a well-defined unimodal curve relates to average flow and density. Geroliminis and Daganzo (2008) verified this conjecture by using empirical data from floating vehicle probes and 500 urban fixed detectors in Yokohama, Japan, as shown in Fig. 4.3. Notice that the combined flow-occupancy diagram created using local data from two separate detectors for an entire weekday shows great disorder, which would typically persist on the link level when examining local behaviours. However, this disorder disappears when flow-density relationships are constructed with the aggregate data obtained from all of the detectors in the network (latter diagram). Interestingly, the data observed is obtained from very distinct periods (denoted as A1…D2), including morning and evening rush hours and weekdays and weekends. Wu et al. (2011) and Dakic and Stevanovic (2018) also applied the concept of a fundamental diagram to urban arterials and termed them as arterial fundamental diagrams (AFD). The studies reported a huge variability in the AFD attributed to the cyclic nature of signalized intersections (Wu et al. 2011; Dakic and Stevanovic 2018). These studies argued that a stable relationship between flow and density could also be obtained for AFD by employing suitable data filtering methods.

58

S. Das et al.

Fig. 4.3 Macroscopic Fundamental diagram (Geroliminis and Daganzo 2008)

4.4 Recommendations for Future Research Although there have been recent advancements in the state-of-the-art traffic flow models, much more remains to be learned, tested, and explored in detail for disorderly traffic environments. Some of the future research recommendations to support a more realistic representation of disorderly traffic flows are illustrated below.

4.4.1 Fundamental Diagram and Macroscopic Fundamental Diagram The MFDs have profound applications in terms of evaluating the effectiveness of different network-wide control strategies (Mahmassani et al. 1987; Daganzo 2007; Geroliminis and Levinson 2009; Gerolominis and Sun 2011a; Daganzo et al. 2012; Gayah and Daganzo 2012; Haddad and Geroliminis 2012; Zheng et al. 2012a, b; Keyvan-Ekbatani et al. 2012; Haddad et al. 2013; Keyvan Ekbatani et al. 2013; Mahmassani et al. 2013; Keyvan-Ekbatani et al. 2015; Leclercq et al. 2015; Du et al. 2015; Haddad and Mirkin 2017; Lu et al. 2020). However, the application of MFDs under the Indian context is lacking. Hysteresis in MFD Past studies argue that the network-wide relation between flow and density is not always well defined, and a significant amount of scatter in MFD exists, which results in multiple values of flow for a given value of density (Mazloumian et al. 2010; Daganzo et al. 2011; Gayah and Daganzo 2011; Gayah et al. 2014; Shim et al. 2019). Hysteresis patterns can also be observed, highlighting that the flow under congestion onset is significantly different from the congestion offset (Buisson and

4 Driving Behavior Modeling in Mixed Traffic Conditions …

59

Ladier 2009; Geroliminis and Sun 2011b; Saberi and Mahmassani 2013). Geroliminis and Sun (2011b) investigated the causes of hysteresis in macroscopic fundamental diagrams (MFD). Higher network flows are observed for the same density during congestion onset and lower flow during congestion offset. They reported that distributions of individual occupancy measurements (congestion distribution) and synchronized occurrence of transient states and capacity drop at individual detectors are the two major reasons which cause hysteresis in MFD. Similarly, Buisson and Ladder (2009) and Saberi and Mahmassani (2013), using empirical data, identified hysteresis loops in the freeway network. They explored the relationship between the size of the hysteresis and the inhomogeneity of congestion distribution. Saberi and Mahmassani (2013) reported heterogeneous distribution of congestion because of traffic flow instabilities. Based on the empirical observations, Saberi and Mahmassani (2013) reported two shapes of hysteresis loops, namely, H1 and H2. These shapes exhibit the difference in the recovery path of the flow-density relation. For the H1 type of loops, the flow tends to decrease constantly with an increase in density. In the case of the H2 type of loops, the flow remains roughly unchanged with an increase in density, and as a result, a larger size of hysteresis can be noted. The H1 and H2 type of hysteresis are illustrated in Fig. 4.4. Saberi and Mahmassani (2013) reported that its width and height can quantify the size of the hysteresis, and the area covered by the loop, as shown in Fig. 4.5. The size of the hysteresis can be expressed as an ordered paired of its width and height as SH = (k, Q)

(4.1)

The area of the hysteresis can be computed as AH = (k ∗ Q)

Fig. 4.4 Shapes of hysteresis a H1 b H2

(4.2)

60

S. Das et al.

Fig. 4.5 Flow-density relation and size of a hysteresis loop

The hysteresis in MFD or FD in case of mixed traffic conditions seems unexplored and unreported. Further, implications of hysteresis on the stability of traffic flow, travel time reliability, and distribution of congestion under mixed traffic conditions are also not explored and need further study. Capacity drop Capacity drop in individual freeway sections is a well-known and well-reported phenomenon. A capacity drop in freeway networks is a result of network traffic instability. The heterogeneous distribution of traffic flow in the network for the same values of density (higher flow in loading period than the recovery periods) also implies a capacity drop (Saberi et al. 2013). Saberi et al. (2013) reported two types of capacity drop phenomenon in freeway networks. Capacity drop Type-I is associated with the inability of the freeway network to maintain throughput at its maximum for a relatively long time. This implies that capacity drops still the demand is high, and the network is loading. Capacity drop Type-II is associated with traffic instability when the network undergoes reloading during the afternoon peak period after an incomplete recovery from morning peak periods. This reloading results in lesser capacity in afternoon periods compared to morning periods. For mixed traffic conditions in India, future research efforts should be made for. • Developing AFDs and MFDs for urban networks. Disentangling the temporal variation in MFDs to quantify the changes in traffic patterns. • Analyzing the implications of capacity drop and hysteresis phenomenon on traffic flow stability, travel time reliability, and distribution of congestion. • Incorporating travel time uncertainties in analyzing traffic flow stability and modeling traffic flow. • Developing multimodal MFDs • Extending MFDs for developing perimeter control strategies and quantifying the effect of different congestion mitigation strategies.

4 Driving Behavior Modeling in Mixed Traffic Conditions …

61

4.4.2 Modeling Unique Behaviors of Motorized Two-Wheelers The ownership levels of motorized two-wheelers (MTWSSs) have seen a meteoric rise in the developing nations over the years, mainly because of its increased affordability, convenience, flexibility in parking, and faster operation than other vehicles. This steady rise has resulted in complex interactions with the surrounding traffic, higher risks associated with the riders, and transportation monitoring and management challenges. Because of compact sizes and high power to weight ratio, MTWSs exhibit complex maneuvers, and their interactions with other vehicles make the traffic system even more complex. As summarized by Lee (2008), some of the characteristic maneuvers of MTWSs that are commonly observed include “travelling alongside another vehicle in the same lane, oblique following, moving to the head of a queue, filtering, swerving or weaving, tailgating, maintaining a shorter headway when aligning to the lateral edge of the preceding vehicle, travelling according to the virtual lanes formed dynamically by the vehicles in surroundings, and self-organization phenomena”. The widespread adoption of MTWSs and the risks involved with them have spurred a growing interest in transport modelers to model their maneuvers. A previous study by Nikias et al. (2012) highlighted that filtering and overtaking are frequently observed in urban arterials. Later, Vlagonianni (2014) indicated that both these maneuvering patterns are different in terms of kinematic characteristics and existing traffic conditions. In particular, relative speed, spacing between vehicles in an adjacent lane, and a platoon of MTWSs are the common parameters that significantly affect the likelihood of accepting critical lateral distance during overtaking and filtering. In addition to relative speed, pore size and history of lateral movements were also found to influence rider’s filtering choice (Das and Maurya 2019). On modeling lateral interactions of MTWSs in disorderly traffic, lateral clearance between vehicles, speed of a subject vehicle, relative speed, presence of a constraining vehicle, and involvement of car in the direct front are found as the contributing factors (Asaithambi and Joseph 2018; Kotagi et al. 2019; Munigety 2018; Dong et al. 2019; Das and Maurya 2020). Concerning risks involved with MTWSs, Rogé et al. (2010) highlighted that one of the primary reasons for MTWS-involved accidents is the lack of attention or perception of car drivers to detect the MTWS in time. According to MORTH (2018), MTWSs accounted for 35.2% and 31.4% of the road accidents and road fatalities on Indian roads. Several risk factors that affect the safety of MTWS can be found in Damani and Vedagiri’s (2021) work. Although understanding the complex maneuvers of MTWSs has received considerable attention across the globe, some of the future research directions which can be explored are listed below. • Human factors such as reaction time, driver’s workload, drowsiness, distraction, and alcohol use can be integrated to model the maneuvering patterns of MTWSs. Such integration can provide a realistic representation of MTWS dynamic maneuvers and contribute to improving MTWS safety.

62

S. Das et al.

• Although several attempts have been made to understand lateral interactions, oblique following, filtering, and overtaking maneuvers, understanding the grouping behavior of MTWS (which is commonly observed in urban arterials) remains a matter of further exploration. • Suitable information on socio-demographics such as age, gender and driving experience, and other factors such as weather conditions, visibility, and roadside infrastructure can further be considered in MTWS. modeling research. • Using an online questionnaire survey, the emotional factors of MTWS such as riding comfort, joy in driving, and level of risks involved while interacting with bigger vehicles can be quantified. Such information can be integrated with the existing traffic scenario to improve MTWS safety and understand the need to implement several policy measures.

4.4.3 Behavior on Two-Lane Roads Two-lane roads are classically studied differently compared to multi-lane roads. This is primarily due to the unique set of variables that affect the driving behavior on these roads. Asaithambi and Shravani (2017) observed that 62% of drivers performed flying overtaking, and the rest performed accelerated overtaking, signifying that most vehicles travel with their current speed during overtaking. Chandra and Shukla (2012) studied the acceleration and overtaking characteristics of different vehicle types using the floating car method. It was observed that the shoulder condition influences the acceleration behavior during overtaking. Also, an inverse correlation is observed between the overtaking speed and the acceleration rate in the mixed traffic conditions. Recently, Choudhari et al. (2020) developed overtaking distance and overtaking time models for passenger cars on two-lane undivided rural highways. Based on data obtained from G.P.S. receivers, results indicated the speed of the overtaken vehicle, the relative speed between overtaking and overtaken vehicles, and longitudinal gap before and after overtaking as the contributing factors. Relative speed and longitudinal gap obtained from their work were found comparatively lower than that reported in the literature. Unlike the typical microscopic lane-changing models developed for multi-lane highways, the overtaking process involves additional factors such as speed of the opposing vehicle, class of the subject and overtaken vehicles, waiting duration, geometric characteristics, etc. Moreover, the overtaking decision process and the microscopic overtaking process are not well understood and need further study under mixed traffic conditions. More specifically, future studies can be directed to understanding the effects of diverse overtaking and overtaken vehicle types, overall traffic flow, presence of an oncoming vehicle, geometric elements (such as horizontal and vertical alignment, road width, roadside features, etc.), time of the day and weather conditions, on the overtaking decision processes of drivers.

4 Driving Behavior Modeling in Mixed Traffic Conditions …

63

4.4.4 Intra and Inter-Vehicular Interactions in Non-Lane-Based Conditions The haphazard spatial arrangement of vehicles in disorderly traffic makes the traffic flow phenomena more complex. Interactions among vehicles in different traffic flow regimes are characterized by longitudinal and lateral distance-keeping behavior, longitudinal and lateral speeds maintained by the subject vehicle, lateral positions of vehicles, etc. Over the years, several efforts were devoted to comprehending these interactions across diverse vehicle types. Many researchers for modeling disorderly flows have underlined consideration of vehicle type specific interactions. Drivers’ preferences in maintaining longitudinal gaps vary with the type of leading vehicle and vehicles’ operational characteristics. Vehicles with poor operating capabilities maintain larger longitudinal gaps, and even the gaps tend to increase as the dimensions of the interacting leading vehicles increase (Kanagaraj et al. 2011; Maurya et al. 2015; Das and Maurya 2017). On the understanding of lateral gaps maintained by the interacting vehicles, Budhkar and Maurya (2017) highlighted that lateral gap maintained with the same vehicle type is usually lower than that maintained with different vehicle types at similar speed levels. In parallel, Raju et al. (2018) estimated lateral amplitude (difference in minimum and maximum lateral positions) of vehicles corresponding to different traffic flow states. MTWS were found to have larger lateral amplitudes than other vehicle types, which is also consistent with the findings of Kanagaraj et al. (2015). Concerning preferred lateral positions of vehicles in disorderly traffic, Raju et al. (2018) indicated that the positional arrangement depends on traffic flow states, the extent of disordered movements becomes more prevalent as traffic flow increases (see Fig. 4.6).

Fig. 4.6 Lateral placement of vehicles at different traffic flow levels (Raju et al. 2018)

64

S. Das et al.

Recent literature has underlined the importance of integrating both the longitudinal and lateral interactions of vehicles in the car-following scenario of disorderly traffic. Using copula models, Das and Maurya (2018a, b) categorized vehicle type specific interactions considering time headway and centerline separation as the longitudinal and lateral descriptors of vehicle interactions simultaneously. Such integrations can provide a better representation of disorderly traffic and can be considered as inputs in the development of micro-simulation traffic flow models and their consequent sub-models (for instance, the car-following model). Based on the available literature on the preferred spatial arrangement of vehicles, future research efforts can be directed to. • Development of an integrated approach or a parameter that can define the twodimensional preferences of the subject vehicle for any interacting vehicle pair. Similar to the concept of copula models, some integrated mechanisms can be proposed to represent the preferred driving behavior and subsequently, this can be used for a comprehensive traffic flow model development. • Development of a lateral movement model with a detailed understanding of the amount/number of lateral shifts exhibited by a vehicle type, preferred lateral positions of the vehicle (towards the left or right side of the interacting vehicle), speed differences, current and expected spacing, and so on. • Consideration of traffic flow levels for understanding the differences in preferred positions of vehicles. Special emphasis can also be laid on determining risk levels associated with the drivers based on the preferred positional arrangement of the vehicles. In this context, a safety indicator can be proposed (similar to Raju et al. 2019, Das and Maurya 2020) to capture the two-dimensional movement of vehicles. • Understanding of vehicle-type specific interactions according to different roadway facilities, geometric conditions, and the differences in behavioral patterns at midblock sections and the approach of signalized/un-signalized intersections. The stability of C.F. models is long recognized as an important phenomenon. Traditionally, two types of stabilities are examined: local stability and asymptotic stability. The former deals with the response of the following vehicle to changes in the leader’s behavior and the latter deals with how the disturbances caused by a leading vehicle propagate in a platoon. In this context, the range of C.F. parameters used for the models should be consistent with the stability features observed in the field. Gazis et al. (1959) argued that traffic stream stability may have an impact on the safety on the road; lower symptomatic stability may lead to a higher number of rear-end accidents. Herman et al. (1959) argued that more information from the downstream vehicles in the car-following models will lead to better stability in the car-following models, as observed in the field. Therefore, new algorithms for emerging technologies must consider factors related to downstream and surrounding vehicles for improved traffic stability.

4 Driving Behavior Modeling in Mixed Traffic Conditions …

65

4.4.5 Calibration and Validation of Simulation Models Traffic simulation models (T.S.M.) have proven beneficial in analyzing the complex traffic situation beyond the scope of traditional analytical methods. Traffic simulation models have also found application in the transportation planning process due to their flexibility and feasibility in testing different alternatives that do not currently exist in the real world. Generally, traffic simulation is an imitation of behavior, characteristics, and relationships of distinct transportation system elements. The efficiency of T.S.M. lies in its ability to replicate the field conditions. No single simulation model can be expected to replicate the range of traffic conditions in the real world. A poorly calibrated model is a prime reason for skepticism among policymakers. With this as a major impetus, researchers worldwide have realized the need for proper calibration of a simulation model. Without calibration, a traffic simulation model cannot provide accurate predictions of effectiveness (MOE), which are predominant in the transportation decision-making process. Significant contributions have been made in the recent past for T.S.M. calibration for both homogeneous (Rakha et al. 1996; Fellendorf and Vortisch 2001; Park and Qi 2005; Kim et al. 2020; Chitturi and Benekohal 2008) and heterogeneous (Mathew and Radhakrishnan 2010; Manjunatha et al. 2013; Mishra et al. 2017; Bhattacharyya et al. 2020; Raju et al. 2020a, b) traffic conditions. The review on different T.S.M. calibration methods highlighted a few important issues. It recommended the following steps that included (a) formulation of calibration process (b) identification of important parameters (c) selection of goodness-of-fit measure (d) model validation. A properly calibrated T.S.M. can give helpful information on drivers ‘responses to changing traffic and geometric conditions’. In India, traffic consists of an indiscriminate mix of vehicles with varying static (length and width) and dynamic (acceleration, deceleration) properties. Further, heterogeneity also exists in road users (like pedestrians, non-motorized traffic, motorized traffic, and public transportation). Moreover, poor lane discipline, non-adherence to traffic rules form the peculiarity of such traffic conditions. Therefore, it is imperative to calibrate the simulation parameters for specific mode or vehicle types to replicate such traffic conditions. Calibration and Validation The pre-calibration step involves collecting relevant field inputs and coding the road network in the T.S.M. based on the field data. Any T.S.M. requires a wide range of input parameters. These are classified as controllable and uncontrollable parameters. Uncontrollable parameters include road geometry, traffic control characteristics, signal program. These parameters are needed to be inputted once and don’t require any modification. Controllable parameters like desired speed distribution and driving behavior (following behavior, lane changing, and lateral movement) parameters need to be calibrated to improve the simulation model’s performance. For comparing the performance of the simulation model, it is essential to identify a measure of effectiveness (M.O.E.), the value of which can be compared with the field inputs to

66

S. Das et al.

conclude whether the T.S.M. is calibrated or not. Different macroscopic parameters such as speed, travel time, traffic volume, and delay and microscopic parameters like headway, acceleration/deceleration were used as performance measures for calibrating the controllable parameters of T.S.M. Among different controllable parameters, driving behavior parameters are calibrated to replicate the local field conditions. These parameters can be calibrated either through optimization or can be empirically derived using trajectory datasets. Calibration using optimization technique Driving behavior parameters for three different driving regimes, namely, following behavior, lane changing behavior, and lateral placement, are optimized. Past studies have adopted different techniques such as manual search, mathematical optimization, and heuristic search for calibrating different driving behavior models. Recently, heuristic-based algorithms, namely genetic algorithm (G.A.) (Mathew and Radhakrishnan 2010; Bhattachharyya et al. 2020; Raju et al. 2020a) is widely used to optimize different parameters of driving behavior models. The objective function is defined as shown in Eq. (4.3): min Z = [obser vedr esponse − modelledr esponse]2

(4.3)

The modeled response is generally the measure of effectiveness. Raju et al. (2020a) considered speed as the measure of effectiveness. On the other hand, Bhattacharyya et al. (2020) considered travel time a suitable measure of effectiveness, whereas delay was considered by Mathew and Radhakrishnan (2010) for calibrating driving behavior parameters. Mishra et al. (2017) considered headway, speed, and traffic volume as M.O.E. The objective function is optimized by minimizing the error between the observed and simulated values. Bhattacharyya et al. (2020) proposed that after optimizing the driving behavior parameters, it is important to perform nonparametric and visual checks before concluding that the T.S.M well-calibrated can be extended for analyzing different traffic scenarios. Recently, Gore et al. (2020) calibrated the simulation model for an urban network (6intersection and 6links, network length 6.0 km) by optimizing the driving behavior parameters using a genetic algorithm by considering travel time as a M.O.E. The calibrated T.S.M. was validated externally at three different resolutions (a) link volumes, (b) turning proportions at intersections, (c) delay at intersections. Recently, Chaudhari et al. (2021) proposed a methodology based on optimizing performance measures at the microscopic level (acceleration, speed, and trajectory profiles) for calibrating the Wiedemann-99 model parameters for mixed traffic conditions using trajectory data. The results show that the optimized parameter values and consequently, the thresholds that delineate closing, following, emergency braking, and opening regimes vary between two-wheelers and cars. This indicated that the mixed traffic models should allow calibration of parameters for each vehicle type.

4 Driving Behavior Modeling in Mixed Traffic Conditions …

67

Empirical derivation of driving behavior using trajectory data Raju et al. (2020b) and Paul et al. (2020) used vehicle trajectory data for deriving the values of driving behavior parameters empirically. The derived parameters were inputted into the simulation model, and the simulation model represented the field condition well. The calibrated simulation model was validated at macroscopic (speed, flow, and fundamental diagram) and microscopic levels (hysteresis patterns). The calibrated and validation micro-simulation model can then be used to comprehend different traffic control strategies on the traffic flow and its properties at both microscopic and macroscopic levels. Recently, Chaudhari et al. (2021) showed that different traffic flow model parameters can lead to similar macroscopic behavior, highlighting the importance of calibration and validation at the microscopic level. Some of the future directions are highlighted as follows: • The simulation models developed for links or networks can be extended to study the impact of traffic control strategies (actuated traffic control, coordinated traffic controls, adaptive controls, and coordinated adaptive control), vehicle segregation policies (segregating lighter vehicles), and bus priority control on throughput, stability, and safety of traffic flow. • The impact of connected vehicles and connected and autonomous vehicles under Indian traffic conditions in terms of safety, throughput, and traffic flow stability can be studied. • The effect of different perimeter control strategies, such as cordon pricing, and congestion pricing, can also be extensively studied using simulation models.

4.4.6 Naturalistic Driving Studies and Models A naturalistic driving dataset is a repository of high-fidelity data collected from multiple unbiased drivers recorded continuously over several months from various parts of the U.S.A (Dingus et al. 2015). The data set includes approximately 2,000,000 vehicle miles, almost 43,000 h of data, 241 primary and secondary drivers, 12 to 13 months of data collected for each vehicle, and data from 5 channels of video and many vehicle states and kinematic sensors. The repository contains information on vehicle travel, driver behavior, and exposure. The data is collected at a level of detail as in the safety-related events, such as crashes and near-crashes. It contains many extreme driving behavior and performance cases, including severe drowsiness, impairment, judgment error, risk-taking, willingness to engage in secondary tasks, aggressive driving, and traffic violations. Parameters such as vehicle speed, vehicle headway, time-to-collision, and driver reaction time are also recorded. Such a study and dataset in the mixed traffic will help better understand the driver behavior in mixed traffic. Towards this end, Raju et al. (2019) collected microscopic data and used hysteresis plots to identify and quantify the chances of rear-end crashes using the follower’s Instantaneous perception time (IPT). In parallel, Das and Maurya (2020) defined minimum TTC thresholds according to different CS and leading vehicle types

68

S. Das et al.

in mixed traffic environments. Such studies help better understand the dynamics of crashes to reduce traffic injuries and fatalities by preventing collisions or reducing their severity. Moreover, such a database can be used to integrate the interrelationship of the driver with vehicle, roadway, and environmental factors into the traffic flow models and algorithms for improved realism.

4.4.7 Driving Simulator Systematically validated driving simulator experiments are well accepted to understand driving behavior and evaluate the strategies under different scenarios. These experiments provide rich traffic data, otherwise difficult to achieve through traditional means. Traffic simulators are gaining popularity for understanding driver behavior in mixed traffic. They have been widely used for understanding the impact of alcoholimpaired driving and sleep deprivation on crash potential, phone, eating, and other distraction on driving decisions, etc. (Yadav and Velaga 2020a, b, c; Yadav et al. 2020; Choudhary and Velaga 2019a, b; Choudhary and Velaga 2017a, b). While traffic simulators are more popular for safety studies (Dedes et al. 2011; Lee et al. 2018), they should be explored for improved fuel and travel time efficiency strategies at both driver level and network level. Further EEGs, eye trackers, and heart rate sensors can be used to provide physiological parameters during different driving maneuvers and under various interventions to understand driver health impacts better.

4.4.8 Connected, Autonomous, and Emerging Vehicle Technologies In the exciting era of emerging transportation solutions, the significant research focuses on connected, automated, and autonomous vehicle systems, electrification, and shared mobility. While the fully automated vehicle technologies may be several years away, some of these technologies may be realized on Indian roads in the shorter term. These technologies spanning multiple modes have the potential to transform the transportation domain in the near future. Towards this end, Chaturvedi et al. (2021) proposed a methodology for traffic state estimation using V2V and V2I communication for a quasi-connected environment. There are exciting research opportunities in the areas of in-vehicle sensing technology, communication, data analytics, driving and driver behavior models, traffic management algorithms, and mobility applications leveraging these technologies. Moreover, high-fidelity vehicle dynamics-related information now available through the connected vehicle technologies provides an opportunity to update the existing macroscopic and microscopic models for a more realistic representation of traffic behavior for applications in safety, sustainability, and efficiency.

4 Driving Behavior Modeling in Mixed Traffic Conditions …

69

4.4.9 Data-Driven and Integrated Models Considering the complexities of modeling disorderly traffic, data-driven models can be advocated as an essential tool to efficiently represent the traffic flow phenomena with good predictions and accuracy. Data-driven models are considered more powerful than other statistical models because of their modeling flexibility, adaptability, learning and generalization ability, and good predictive ability (Bingham 2001; Srinivasan et al. 2006). Particularly, car-following literature has been mainly oriented into the development of artificial neural network (ANN) models. Hongfei et al. (2003) applied a backpropagation neural network model to predict the following vehicle’s acceleration, considering one input layer, two hidden layers, and one output layer. Panwai and Dia (2007) developed a neural network model using a single hidden layer and compared it with Gipps model. Results indicated the improved performance of neural networks than conventional car-following models. Khodayari et al. (2012) evaluated the performance of neural networks to predict follower’s accelerations using instantaneous reaction delays. Later, Zheng et al. (2013) redefined the concept of driver-vehicle reaction delay (Khodayari et al. 2012) and incorporated it into a neural network-based car-following model. Colombaroni and Fusco (2014) applied a feed-forward neural network model to predict the acceleration of the following vehicle, trained by a swarm stochastic evolutionary algorithm. Wang et al. (2017) used a deep neural network to describe complicated human actions in car-following behavior. In a recent study, Zhu et al. (2018) illustrated that deep reinforcement learning neural networks could reproduce human-like car-following behavior with higher accuracy than traditional car-following models. In the context of disorderly traffic flows, Mathew and Ravishankar (2012) used a neural network approach to model vehicle-type dependent car-following behavior. They found that in addition to leading vehicle’s speed and space headway, interacting vehicle type needs to be considered to predict follower’s speed in a single-leader car-following regime. The model comparison demonstrated improved performance of neural networks than Gipps’ model. Although several studies have supported the development of neural network models for car-following scenarios, there is still a paucity of research concerning the applicability of data-driven models to modeling disorderly flows. In this context, future research can be conducted on several areas, which are listed below. • Consideration of diverse vehicle types, determination of regime boundaries (such as staggered-following, oblique following, multiple-leader following, filtering, swerving) and the existing traffic conditions in the existing models remain a matter of further exploration. • An integrated two-dimensional data-driven model can be developed for predicting drivers’ responses in different driving regimes by considering both the longitudinal and lateral descriptors of vehicle interactions. The combined effect of surrounding vehicles on the dynamics of the subject driver can be explored.

70

S. Das et al.

• The credibility of the developed data-driven models can be validated with the existing traffic flow models, with a suitable emphasis on the sensitivity analysis and generalization capability of the data-driven model. A synthesis of the literature indicates that there have been gradual advancements in the existing microscopic and macroscopic models to embrace a more realistic representation of traffic flow characteristics in mixed traffic environments. Several enhancements at the disaggregated levels, such as incorporating the lateral dimension of traffic, vehicle-type specific interactions in different driving regimes, the importance of human factors in microscopic decision-making strategies, still need further exploration. Even the transferability of the developed models describing different traffic states at the microscopic and macroscopic levels, with suitable calibration and validation against real-world traffic databases, entails further assessment. Although recent studies have incorporated lateral dimension in the existing car-following models, what remains unresolved is developing a full-scale twodimensional driving behavior model considering the integrated longitudinal and lateral interactions of diverse vehicle types. Suppose such integrated two-dimensional traffic flow models are developed for disorderly traffic environments. In that case, it is believed that there would be a substantial improvement in the accuracy of traffic flow representation. This, in turn, would contribute towards the development of realistic traffic simulation models, smarter and user-friendly autonomous cruise control systems, with applications in intelligent transportation systems, traffic safety evaluation, traffic operation studies, congestion mitigation, and so forth. This calls for holistic and integrated efforts required to develop traffic flow models of disorderly systems for providing a safe and efficient transportation system for road users.

References Anand PA et al (2019) Calibration of vehicle-following model parameters using mixed traffic trajectory data. Transp Dev Econ 5(2):1–11 Arasan VT, Dhivya G (2008) Measuring heterogeneous traffic density. Proceedings of international conference on sustainable Urbn transport and enviroment Arkatkar SS, Thamizh Arasan V (2010) Effect of gradient and its length on performance of vehicles under heterogeneous traffic conditions. J Transp Eng 136(12):1120–1136 Asaithambi G, Joseph J (2018) Modeling duration of lateral shifts in mixed traffic conditions. J Transp Eng Part A: Syst 144(9):04018055 Asaithambi G, Kanagaraj V, Srinivasan KK, Sivanandan R (2018) Study of traffic flow characteristics using different vehicle-following models under mixed traffic conditions. Transp Lett 10(2):92–103 Asaithambi G, Shravani G (2017) Overtaking behaviour of vehicles on undivided roads in non-lane based mixed traffic conditions. J Traffic Transp Eng (English Edition) 4(3):252–261 Balakrishnan S, Sivanandan R (2015) Influence of lane and vehicle subclass on free-flow speeds for urban roads in heterogeneous traffic. Transp Res Procedia 10(:166–175 Basu D, Maitra SR, Maitra B (2006) Modelling passenger car equivalency at an urban midblock using stream speed as measure of equivalence

4 Driving Behavior Modeling in Mixed Traffic Conditions …

71

Bham GH, Benekohal RF (2004) A high fidelity traffic simulation model based on cellular automata and car-following concepts. Transp Res Part C: Emerg Technol 12(1):1–32 Bhattacharya PG, Bhattacharya AK (1982) Observation and analysis of saturation flow through signalised intersections in Calcutta. Indian Highways 10(4):11–33 Bhavathrathan B, Mallikarjuna C (2012) Evolution of macroscopic models for modeling the heterogeneous traffic: an Indian perspective. Transp Lett 4(1):29–39 Bhattacharyya K, Maitra B, Boltze M (2020) Calibration of micro-simulation model parameters for heterogeneous traffic using mode-specific performance measure. Transport Res Rec 2674(1):135– 147 Bingham E (2001) Reinforcement learning in neurofuzzy traffic signal control. Eur J Oper Res 131(2):232–241 Biswas S et al (2018) Saturation flow model for signalized intersection under mixed traffic condition. Transp Res Record 2672(15): 55–65 Budhkar AK, Maurya AK (2017) Characteristics of lateral vehicular interactions in heterogeneous traffic with weak lane discipline. J Modern Transp 25(2):74–89 Buisson C, Ladier C (2009) Exploring the impact of homogeneity of traffic measurements on the existence of macroscopic fundamental diagrams. Transp Res Record: J Transp Res Board 2124:127–136 Chakroborty P, Agrawal S, Vasishtha K (2004) Microscopic modeling of driver behaviour in uninterrupted traffic flow. J Transp Eng ASCE 130(4):438–451 Chandra S, Shukla S (2012) Overtaking behavior on divided highways under mixed traffic conditions. Procedia Soc Behav Sci 43:313–322 Chandra S, Kumar U (2003) Effect of lane width on capacity under mixed traffic conditions in India. J Transp Eng 129(2):155–160 Chaturvedi S, Ashok A, Chilukuri BR (2021) Traffic state estimation using DSRC-enabled probe vehicles. 2021 International conference on COMmunication systems & NETworkS (COMSNETS). IEEE Chaudhari AA et al (2021) Optimization of wiedemann-99 model parameters for mixed traffic using vehicular trajectory data. No. TRBAM-21–04411 Chin et al (2018) Unlocking cities The impact of Ridesharing across India, The Boston Consulting Group (2018). https://image-src.bcg.com/BCG-Unlocking-Cities-Ridesharing-India_tcm21-185 213.pdf Chitturi MV, Benekohal RF (2008) Calibration of VISSIM for freeways. Transportation Research Board 87th Annual Meeting Transportation Research Board 08–2317 Cho HJ, Wu YT (2004) Modeling and simulation of motorcycle traffic flow. In: 2004 IEEE international conference on systems, man and cybernetics 7:6262–6267 Choudhari T, Budhkar A, Maji A (2020) Modeling overtaking distance and time along two-lane undivided rural highways in mixed traffic condition. Transp Lett, 1–9 Choudhary P, Velaga NR (2019a) A comparative analysis of risk associated with eating, drinking and texting during driving at unsignalised intersections. Transport Res F: Traffic Psychol Behav 63:295–308 Choudhary P, Velaga NR (2019b) Effects of phone use on driving performance: a comparative analysis of young and professional drivers. Saf Sci 111:179–187 Choudhary P, Velaga NR (2017a) Modelling driver distraction effects due to mobile phone use on reaction time. Transp Res Part C: Emerg Technol 77:351–365 Choudhary P, Velaga NR (2017b) Mobile phone use during driving: Effects on speed and effectiveness of driver compensatory behaviour. Accid Anal Prev 106:370–378 Choudhury CF, Islam MM (2016) Modelling acceleration decisions in traffic streams with weak lane discipline: a latent leader approach. Transp Res Part C: Emerg Technol 67:214–226 Colombaroni C, Fusco G (2014) Artificial neural network models for car following: experimental analysis and calibration issues. J Intell Transp Syst 18(1):5–16 Daganzo C, Gayah V, Gonzales E (2011) Macroscopic relations of urban traffic variables: bifurcations, multivaluedness and instability. Transp Res Part B-Methodol 41(1):278–288

72

S. Das et al.

Daganzo C, Gayah V, Gonzales E (2012) The potential of parsimonious models for understanding large scale transportation systems and answering big picture questions. EURO J Transp Logist 1(1–2):47–65 Daganzo C (2007) Urban gridlock: Macroscopic modelling and mitigation approaches. Transp Res Part B-Methodol 40:396–403 Dakic I, Stevanovic A (2018) On development of arterial fundamental diagrams based on surrogate density measures from adaptive traffic control systems utilizing stop-line detection. Transp Res Part C 94(2018):133–150. https://doi.org/10.1016/j.trc.2017.08.013 Damani J, Vedagiri P (2021) Safety of motorised two wheelers in mixed traffic conditions: literature review of risk factors. J Traffic Transp Eng (English edition) Das AK, Chilukuri BR (2020) Link cost function and link capacity for mixed traffic networks. Transp Res Record 2674(9): 38–50 Das S, Maurya AK (2018b) Bivariate modeling of time headways in mixed traffic streams: a copula approach. Transp Lett, 1–11 Das S, Maurya AK (2020) Defining time-to-collision thresholds by the type of lead vehicle in non-lane-based traffic environments. IEEE Trans Intell Transp Syst 21(12):4972–4982. https:// doi.org/10.1109/TITS.2019.2946001 Das S, Maurya AK (2019a) Modeling maneuverability of motorized two-wheelers during filtering in urban roads. Transp Res Rec 2673(5):637–647 Das S, Maurya AK (2018) Multivariate analysis of microscopic traffic variables using copulas in staggered car-following conditions. Transportmetrica A: Transport Sci 14(10):829–854 Das S, Maurya AK (2017) Time headway analysis for four-lane and two-lane roads. Transp Dev Econ 3(1):9 Das S, Raju N, Maurya AK, Arkatkar S (2020) Evaluating lateral interactions of motorized twowheelers using multi-gene symbolic genetic programming. Transp Res Rec 2674(9):1120–1135 Das S (2020) Driver behaviour modelling in disordered traffic systems: vehicle-following and filtering scenario. Doctoral dissertation. Department of Civil Engineering, Indian Institute of Technology Guwahati, India Das S, Maurya AK (2019) Defining time-to-collision thresholds by the type of lead vehicle in non-lane-based traffic environments. IEEE Trans Intell Transp Syst 21(12): 4972–4982 Dedes G, Wolfe S, Guenther D, Park BB, So JJ, Mouskos K, Heydinger G (2011) A simulation design of an integrated GNSS/INU, vehicle dynamics, and microscopic traffic flow simulator for automotive safety. Adv Transp Studies Dingus TA et al (2015) Naturalistic driving study: technical coordination and quality control. No. SHRP 2 Report S2-S06-RW-1 Dong H, Chen YY, Cirillo C, Wong KI (2019) Lateral movement decision model for powered two-wheelers in Taiwan. Transp Res Rec 2673(2):686–697 Du J, Rakha H, Gayah VV (2015) Deriving macroscopic fundamental diagrams from probe data: Issues and proposed solutions. Transp Res Part C. https://doi.org/10.1016/j.trc.2015.08.015 Feng X, Wang XF, Xie DF (2016) Lateral drift behavior analysis in mixed bicycle traffic: a cellular automaton model approach. Math Problems Eng, 1–10. https://doi.org/10.1155/2016/7962171 Fellendorf M, Vortisch P (2001) Validation of the microscopic traffic flow model VISSIM in different real-world situations. In: 80th Annual Meeting of the Transportation Research Board, Washington, DC Gaddam HK, Rao KR (2019a) Speed–density functional relationship for heterogeneous traffic data: a statistical and theoretical investigation. J Modern Transp 27(1):61–74 Gaddam HK, Ramachandra Rao K (2020) A two-sided lateral gap continuum model and its numerical simulation for non-lane based heterogeneous traffic environment. J Intell Transp Syst 24(6):635–653 Gaddam HK, Ramachandra Rao K (2019b) Speed–density functional relationship for heterogeneous traffic data: a statistical and theoretical investigation. J Modern Transp 27(1): 61–74 Gautam A et al (2018) Estimation of PCE values for hill roads in heterogeneous traffic conditions. Transp Lett 10(2): 83–91

4 Driving Behavior Modeling in Mixed Traffic Conditions …

73

Gayah VV, Gao X, Nagle AS (2014) On the impacts of locally adaptive signal control on urban network stability and the macroscopic fundamental diagram. Transp Res Part B 70(2014):255–268 Gayah V, Daganzo CF (2011) Clockwise hysteresis loops in the macroscopic fundamental diagram: an effect of network instability. Transp Res Part B-Methodol 45(4):643–655 Gayah V, Daganzo C (2012) Analytical capacity comparison of one-way and two-way signalized street network. Transp Res Record: J Transp Res Board 2301:76–85 Gazis DC, Herman R, Potts RB (1959) Car-following theory of steady-state traffic flow. Oper Res 7(4):499–505 Geroliminis N, Daganzo C (2008) Existence of urban scale macroscopic fundamental diagram: some experimental findings. Transp Res Part B 42:759–770 Geroliminis N, Levinson DM (2009) Cordon pricing consistent with the physics of overcrowding. Transp Traffic Theory. Golden Jubilee. Springer 2009:219–240 Geroliminis N, Sun J (2011a) Hysteresis phenomena of a macroscopic fundamental diagram in freeway networks. Transp Res Part A: Policy Pract 45(9):966–979 Geroliminis N, Sun J (2011b) Properties of a well defined macroscopic fundamental diagram for urban traffic. Transp Res Part B: Methodol 45(3):605–617 Godfrey JW (1996) The mechanism of a road network. Traffic Eng Control 11:323–327 Gore N, Arkatkar S, Joshi G, Antoniou C (2020) Modifying bureau of public road link functions using travel time uncertainty captured using WI-FI sensors. Presented at 100th Annual meeting of Transportation Research Board, Washington D.C., U.S.A Greenshields BD (1934) A study of traffic capacity. Proc Highway Res Board 14:448–477 Gunay B (2007) Car following theory with lateral discomfort. Transp Res Part B: Methodol 41(7):722–735 Gupta AK, Katiyar VK (2007) A new multi-class continuum model for traffic flow. Transportmetrica 3(1): 73–85 Haddad J, Geroliminis N (2012) On the stability of traffic perimeter control in two-region urban cities. Transp Res Part B: Methodol 46(9):1159–1176 Haddad J, Mirkin B (2017) Coordinated distributed adaptive perimeter control for large-scale urban road networks. Transp Res Part C: Emerg Technol 77:495–515. https://doi.org/10.1016/j.trc.2016. 12.002 Haddad J, Ramezani M, Geroliminis N (2013) Cooperative traffic control of a mixed network with two urban regions and a freeway. Transp Res Part B 54:17–36 He ZC, Wang YM, Sun WB, Huang PY, Zhang LC, Zhong RX (2015) Modelling car-following behaviour with lateral separation and overtaking expectation. Transportmetrica B: Transport Dyn. https://doi.org/10.1080/21680566.2015.1083911 Herman R et al (1959) Traffic dynamics: analysis of stability in car following. Operat Res 7(1): 86–106 Hongfei J, Zhicai J, Anning N (2003, October) Develop a car-following model using data collected by “fivewheel system”. In: Proceedings of the 2003 IEEE International Conference on Intelligent Transportation Systems (Vol 1, pp 346–351). IEEE Huang W, Fellendorf M, Schoenauer R (2012) Social force based vehicle model for two-dimensional spaces. Transportation research board, 91st annual meeting, Washington Huynh DN, Boltze M, Vu AT (2013) Modelling mixed traffic flow at signalized intersections using social force model. J Eastern Asia Soc Transp Stud 10:1734–1749 Jin S, Huang ZY, Tao PF, Wang DH (2011) Car-following theory of steady-state traffic flow using time-to-collision. J Zhejiang Univ Science A 12(8):645–654 Jin S, Wang D, Xu C, Huang Z (2012) Staggered car-following induced by lateral separation effects in traffic flow. Phys Lett A 376:153–157 Kanagaraj V, Asaithambi G, Srinivasan K, Sivanandan R.: Vehicle classwise analysis of time gaps and headways under heterogeneous traffic. In: Transp Res Board 90th annual meeting 4249 (2011). Kanagaraj V, Asaithambi G, Toledo T, Lee TC (2015) Trajectory data and flow characteristics of mixed traffic. Transp Res Record: J Transp Res Board 2491(1):1–11

74

S. Das et al.

Kashyap NR, Madhuri, et al (2020) Analysis of vehicle-following behavior in mixed traffic conditions using vehicle trajectory data. Transp Res Record 2674(11):842–855 Keyvan-Ekbatani M, Kouvelas A, Papamichail I, Papageorgiou M (2012) Exploiting the fundamental diagram of urban networks for feedback-based gating. Transp Res Part B 46:1393–1403 Keyvan-Ekbatani M, Papageorgiou M, Papamichail I (2013) Urban congestion gating control based on reduced operational network fundamental diagrams. Transp Res Part C 33:74–87 Keyvan-Ekbatani M, Yildirimoglu M, Geroliminis N, Papageorgiou M (2015) Multiple concentric gating traffic control in large-scale urban networks. IEEE Trans Intell Transp Syst 16(4):2141– 2154 Khan SI, Maini P (1999) Modeling heterogeneous traffic flow. Transp Res Rec 1678(1):234–241 Khodayari A, Ghaffari A, Kazemi R, Braunstingl R (2012) A modified car-following model based on a neural network model of the human driver effects. IEEE Trans Syst Man Cybern 42:1440–1449 Kotagi PB, Raj P, Asaithambi G (2020) Modeling lateral placement and movement of vehicles on urban undivided roads in mixed traffic: a case study of India. J Traffic Transp Eng (English Edition) 7(6):860–873 Kim J, Kim JH, Lee G, Shin H, Park JH (2020) Microscopic traffic simulation calibration level for reliable estimation of vehicle emissions. J Adv Transport 13 Leclercq L, Parzani C, Knoop VL, Amourette J, Hoogendoorn SP (2015) Macroscopic traffic dynamics with heterogeneous route patterns. Transp Res Procedia 7:631–650. https://doi.org/10. 1016/j.trpro.2015.06.033 Lee C, So J, Ma J (2018) Evaluation of countermeasures for red light running by traffic simulator– based surrogate safety measures. Traffic Inj Prev 19(1):1–8 Lee TC, Polak JW, Bell MG (2009) New approach to modeling mixed traffic containing motorcycles in urban areas. Transp Res Record: J Transp Lee T-C (2008) An agent-based model to simulate motorcycle behavior in mixed traffic flow. Doctoral Dissertation, Imperial College London, United Kingdom. Retrieved (2008), www.cts. cv.ic.ac.uk/documents/theses/LeePhD.pdf Lewis EE, Benekohal RF (2007) Saturation flow rate study at signalized intersections in Panama. No. 07-3464 Li Y, Zhang L, Peeta S, Pan H, Zheng T, Li Y, He X (2015) Non-lane-discipline-based car-following model considering the effects of two-sided lateral gaps. Nonlinear Dyn 80(1–2):227–238 Li Y, Zhang L, Zhang B, Zheng T, Feng H, Li Y (2016) Non-lane-discipline-based car-following model considering the effect of visual angle. Nonlinear Dyn 85(3):1901–1912 Li Y, Zhao H, Zhang L, Zhang C (2018) An extended car-following model incorporating the effects of lateral gap and gradient. Physica A 503:177–189 Lu W, Liu J, Mao J, Hu G, Gao C, Liu L (2020) Macroscopic Fundamental diagram approach to evaluating the performance of regional traffic controls. Transp Res Record: J Transp Res Board 2674(7):420–430. https://doi.org/10.1177/0361198120923359 Luo Y, Jia B, Liu J, Lam WH, Li X, Gao Z (2015) Modeling the interactions between car and bicycle in heterogeneous traffic. J Adv Transp 49(1):29–47 Mahapatra G, Maurya AK, Chakroborty P (2018) Parametric study of microscopic two-dimensional traffic flow models: a literature review. Can J Civ Eng 45(11):909–921 Mahmassani H, Williams J, Herman R (1987) Performance of urban traffic networks. In: Proceedings of the 10th international symposium on transportation and traffic theory, pp 1–20 Mahmassani HS, Saberi M, Zockaie A (2013) Urban network gridlock: Theory, Characteristics, and dynamics. Transp Res Part C 36(2013):480–497 Manjunatha P, Vortisch P, Mathew TV (2013) Methodology for the calibration of VISSIM in mixed traffic. In: 92nd Annual Meeting of the Transportation Research Board Mallikarjuna C, Tharun B, Pal D (2013) Analysis of the lateral gap maintaining behavior of vehicles in heterogeneous traffic stream. Procedia Soc Behav Sci 104:370–379 Mallikarjuna Ch, Ramachandra Rao K (2006) Area occupancy characteristics of heterogeneous traffic. Transportmetrica 2(3): 223–236

4 Driving Behavior Modeling in Mixed Traffic Conditions …

75

Mathew TV, Ravishankar KVR (2012) Neural network based vehicle-following model for mixed traffic conditions. Europ Trans 52:1–15 Mathew TV, Radhakrishnan P (2010) Calibration of microsimulation models for nonlane-based heterogeneous traffic at signalized intersections. J Urban Plann Dev 136(1):59–66 Mathew TV, Munigety CR, Bajpai A (2015) Strip-based approach for the simulation of mixed traffic conditions. J Comput Civil Eng 29(5):04014069 Maurya AK, Dey S, Das S (2015) Speed and time headway distribution under mixed traffic condition. J Eastern Asia Soc Transp Stud 11:1774–1792 Maurya AK (2007) Development of a comprehensive microscopic model for simulation of large uninterrupted traffic streams without lane discipline. Doctoral. Dissertation, Indian Institute of Technology Kanpur, India Mazloumian A, Geroliminis N, Helbing D (2010) The spatial variability of vehicle densities as determinant of urban network capacity. Philos Trans R Soc 368:4627–4647 Mehar A, Chandra S, Velmurugan S (2014) Passenger car units at different levels of service for capacity analysis of multilane interurban highways in India. J Transp Eng 140(1):81–88 Minh CC, Sano K (2003) Analysis of motorcycle effects to saturation flow rate at signalized intersection in developing countries. J Eastern Asia Soc Transp Stud 5:1211–1222 Mishra A, Chepuri A, Arkatkar S, Maji A (2017) Safety evaluation of un-signalized intersection using hybrid approach involving empirical and simulation data sources. In: 97th annual meeting of Transportation Research Board, Washington, USA Mohan R, Ramadurai G (2017) Heterogeneous traffic flow modelling using second-order macroscopic continuum model. Phys Lett A 381(3):115–123 Mohan R (2021) On the modelling of speed–concentration curves for multi-class traffic lacking lane discipline using area occupancy. Transp Lett, 1–17 MoRTH: Road accidents in India (2017). http://www.indiaenvironmentportal.org.in/files/file/road% 20accidents%20in%20India%202017.pdf Mullakkal FAB, Vortisch P, Mathew TV (2015) Modelling of motorcycle movements in mixed traffic conditions. In: Proceedings of TRB conference, United States Munigety CR (2018) Modelling behavioural interactions of drivers’ in mixed traffic conditions. J Traffic Transp Eng (English Edition) 5(4):284–295 Nguyen LX, Hanaoka S (2011) An application of social force approach for motorcycle dynamics. Proc Eastern Asia Soc Transp Stud 8:1–8 Nguyen LX, Hanaoka S, Kawasaki T (2012) Describing non-lane-based motorcycle movements in motorcycle-only traffic flow. Transp Res Record: J Transp Res Board 2281:76–82 Nikias VA, Vlahogianni EI, Lee TC, Golias JC (2012) Determinants of powered two-wheelers virtual lane width in urban arterials. In: 15th International IEEE conference on intelligent transportation systems, pp 1205–1210 Nokandeh MM, Ghosh I, Chandra S (2016) Determination of passenger-car units on two-lane intercity highways under heterogeneous traffic conditions. J Transp Eng 142(2):04015040 Pal D, Chunchu M (2019) Modeling of lateral gap maintaining behavior of vehicles in heterogeneous traffic stream. Transp Lett 11(7):373–381 Panwai S, Dia H (2007) Neural agent car-following models. IEEE Trans Intell Transp Syst 8:60–70 Papathanasopoulou V, Antoniou C (2018) Flexible car–following models for mixed traffic and weak lane–discipline conditions. Eur Transp Res Rev 10(2):62 Park B, Qi H (2005) Development and evaluation of a procedure for the calibration of simulation models. Transport Res Rec 1934(1):208–217 Papathanaspoulou V, Antoniou C (2019) Towards an integrated longitudinal and lateral movement data-driven model for mixed traffic. Transp Res Procedia 37:489–496 Prakash Vi, Neeraj K et al (2019) Some observations of traffic flow in mixed traffic conditions. Transp Res Board Raghava CS, Badarinath KM (1983) Study of mixed traffic stream parameters through time lapse photography. Highway Res Bulletin (New Delhi) 20:57–83

76

S. Das et al.

Raju N, Arkatkar S, Joshi G (2018) Study of vehicle-following behavior under heterogeneous traffic conditions. In: International conference on traffic and granular flow, Springer, Cham, pp 87–95 Raju N et al (2019) Determining risk-based safety thresholds through naturalistic driving patterns using trajectory data on expressways. Safety Sci 119:117–125 Raju N, Chepuri A, Arkatkar S, Joshi G (2020a) A simulation study for improving the traffic flow efficiency of an intersection coupled with BRT. Eur Transport/Trasporti Europei 75(1):1–20 Raju N, Arkatkar S, Joshi G (2020b) Evaluating performance of selected vehicle following models using trajectory data under mixed traffic conditions. J Intell Transport Sys 24(6):617–634 Rakha N, Hellinga B, Aerde MV, Perez W (1996) Systematic verification, validation and calibration of traffic simulation models. In: 75th Annual Meeting of the Transportation Research Board, Washington, DC Rogé J, Ferretti J, Devreux G (2010) Sensory conspicuousness of powered two-wheelers during filtering Manœuvre, according to the age of the car driver. Le Travail Humain 73:7–30 Saberi M, Mahmassani H (2013) Hysteresis and capacity drop phenomena in freeway networks: empirical characterization and interpretation. Transp Res Record: J Transp Res Board 2391:44– 55. https://doi.org/10.3141/2391-05 Sarna AC, Malhotra SK (1967). Study of saturation flow at traffic light controlled intersections. Central Road Res Inst Sharath MN, Velaga NR (2020) Enhanced intelligent driver model for two-dimensional motion planning in mixed traffic. Transp Res Part C: Emerg Technol 120:102780 Shim J, Yeo J, Lee S, Hamdar S, Jang K (2019) Empirical evaluation of influential factors on bifurcation in macroscopic fundamental diagram. Transp Res Part C 102:509–520 Siddique M (2013) Modeling drivers’ lateral movement Behaviour under weak-lane-disciplined traffic conditions. Master’s Thesis, Bangladesh University of Engineering and Technology, Bangladesh Sreekumar M, Mathew TV (2020) Modelling multi-class disordered traffic streams using traversable distance: a concept analogous to fluid permeability. Transportmetrica A: Transp Sci 16(3): 1531– 1551 Srinivasan D, Choy MC, Cheu RL (2006) Neural networks for real-time traffic signal control. IEEE Trans Intell Transp Syst 7(3):261–272 Tao P, Hu H, Gao Z, Liu X, Song X, Xing Y, Wei F, Wei F (2014) The research of the driver attention field modeling. Discrete Dyn Nat Soc, 1–9. doi: https://doi.org/10.1155/2014/270616 Thamizh AV, Jagadeesh K (1995) Effect of heterogeneity of traffic on delay at signalized intersections. J Transp Eng 121(5): 397–404 Thankappan A, Vanajakshi L (2015) Development and application of a traffic stream model under heterogeneous traffic conditions. J Inst Eng (India): Series A 96(4):267–275 Thankappan A, Vanajakshi L (2015) Development and application of a traffic stream model under heterogeneous traffic conditions. J Inst Eng (India): Series A 96(4): 267–275 Vlahogianni EI (2014) Powered-two-wheelers kinematic characteristics and interactions during filtering and overtaking in urban arterials. Transp Res F: Traffic Psychol Behav 24:133–145 World Health Organization (2018). https://www.who.int/violence_injury_prevention/road_safety_ status/2018/en/ Wu X, Liu HX, Geroliminis N (2011) An empirical analysis on the arterial fundamental diagram. Transp Res Part B 45:255–266 Xu LH, Hu SE, Luo Q, Zhang LY (2015) Research on car-following model considering lateral offset. Int J Eng Res Africa 13:71–80. Trans Tech Publications Ltd Xu L, Li Y, Feng H (2018) Non-lane-discipline-based car-following model considering the effects of full lateral gaps. In: Chinese automation congress (CAC), pp 1146–1150 Yadav AK, Khanuja RK, Velaga NR (2020) Gender differences in driving control of young alcoholimpaired drivers. Drug Alcohol Dependence 213:108075 Yadav AK, Velaga NR (2020a) Alcohol-impaired driving in rural and urban road environments: effect on speeding behaviour and crash probabilities. Accident Anal Prev 140:105512

4 Driving Behavior Modeling in Mixed Traffic Conditions …

77

Yadav AK, Velaga NR (2020b) Alcohol-impaired driving in rural and urban road environments: effect on speeding behaviour and crash probabilities. Accident Anal Prevention 140:105512 Yadav AK, Velaga NR (2020c) An investigation on the risk factors associated with driving errors under the influence of alcohol using structural equation modeling. Traffic Injury Prevention 21(4): 288–294 Zheng J, Suzuki K, Fujita M (2013) Car-following behavior with instantaneous driver–vehicle reaction delay: a neural-network-based methodology. Transp Res Part C: Emerg Technol 36:339– 351 Zheng L, Zhong S, Jin PJ, Ma S (2012a) Influence of lateral discomfort on the stability of traffic flow based on visual angle car-following model. Physica A 391(23):5948–5959 Zheng N, Waraich RA, Axhausen KW, Geroliminis N (2012b) A dynamic cordon pricings scheme combining macroscopic fundamental diagram and an agent-based traffic model. Transp Res Part A 46:1291–1303

Chapter 5

Pedestrian Flow Characteristics Over Different Facilities: Findings and Way Forward Arunabha Banerjee and Akhilesh Kumar Maurya

5.1 Introduction Walking or traveling on foot is the most efficient and effective mode of transportation. In India, as per IRC: 103 (Guidelines for Pedestrian Facilities 2012), short trips of 2 km are covered on foot daily. The ‘pedestrians’ or ‘persons who travel on foot’ form the most vulnerable road user group. To travel from one location to the other, pedestrians use different facilities like sidewalks, walkways, crosswalks, stairways, underpasses, overpasses, etc. These facilities are there to minimize the interaction between the pedestrians and the motorized traffic, and thus provide a safe and comfortable travel on foot. However, in India, post the new millennium, significant importance was given to vehicular traffic and thus curbing of pedestrian infrastructures began. This impels pedestrians to start using the carriageways and thus come in direct contact with vehicular traffic. Moreover, even the locations where proper sidewalks are available, they are either occupied by unauthorized vendors or parked vehicles, which again forces pedestrians to use the main carriageway. A common method of segregating pedestrian-vehicular movement is through atgrade facilities (in the form of signalized/unsignalized crosswalks) or grade-separated facilities (in the form of overpasses/underpasses). However, past studies (Herms 1972; Koepsell et al. 2002) have shown that providing signalized crosswalks may sometimes be hazardous, as vehicles tend to drive even during their red phase. Provision of grade-separated facilities even does not ensure proper utilization of the facilities, due to poor maintenance, absence of proper security measures, and elevators/escalators. These situations lead to pedestrians crossing at grade using illegal median openings and thus coming in direct contact with the motorized traffic (Golakiya et al. 2019).

A. Banerjee · A. K. Maurya (B) Department of Civil Engineering, IIT Guwahati, Guwahati, Assam 781039, India © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_5

79

80

A. Banerjee and A. K. Maurya

The movement of the pedestrians is defined by different pedestrian flow characteristics like macroscopic/fundamental (speed, flow, density, and area module/space) or microscopic (demographic, cultural, and physique) characteristics. The relationships between these macroscopic parameters help in predicting the free-flow speed, jam density, and maximum flow rate/capacity. The development of fundamental diagrams form the basis for developing level of service (LOS) models for different pedestrian facilities and thus the prediction of the capacity of the facilities. Predicting the LOS helps in understanding the existing condition of the facilities, and thus improving them to enhance the walking comfort to the pedestrians. As the pedestrian flow characteristics vary across different countries, it is of utmost importance to conduct a detailed study for setting up the guidelines under the Indian context. The flow characteristics can vary across different regions or countries, as well as across different pedestrian facilities. Therefore, it is extremely important to understand that the walking behavior of pedestrians is different across at-grade facilities (sidewalk, walkway, and crosswalk) and grade-separated facilities (overpass, underpasses, and stairways). US-HCM (2010) and Indo-HCM (2018) define the pedestrian facilities into nine different types: sidewalks, walkways, pedestrian zones, queuing area, crosswalk (marked and unmarked), underpasses (subways), overpasses (foot over bridges: FOBs and skywalks), stairways, and shared pedestrian-bicycle paths (refer to Fig. 5.1). Before any data collection and analysis, one needs to identify the type of pedestrian facility which is appropriate for the required study. Post identification of the facility, it is essential to collect authentic data to understand the behavior of pedestrians, and thus develop macroscopic/microscopic models as well as predict the LOS. Across the globe, different researchers used various data collection techniques (videography, questionnaire, controlled experiments, and manual counting) to come up with practical and feasible solutions based on the actual field conditions. It is crucial to understand the importance and necessity of collecting data using the appropriate data collection technique. As different data collection techniques have their own set of challenges and limitations, selection of the appropriate technique needs to be identified first before proceeding with it. Once the data has been collected from the field, the subsequent important part is the extraction of data using manual, semi-automatic, or fully automatic techniques. Among the different techniques, the manual technique is quite time-consuming, however, it captures most of the essential microscopic pedestrian flow characteristics. On the other hand, to extract pedestrian trajectory data, the fully automatic technique can be used. While using the appropriate data extraction technique, one needs to carefully extract the essential information, which can help in the development of macroscopic and microscopic models for the facility. Dynamic modeling (Henderson 1971; Helbing 1991; Pauls 1995; Hughes 2002; Colombo et al. 2012), continuous modeling (Helbing and Molnár 1995; Lakoba et al. 2005; Parisi and Dorso 2007; Zhou et al. 2020), and discrete modeling (Blue et al. 1997; Fukui and Ishibashi 1999; Lämmel and Flötteröd 2015; Burstedde et al. 2001; Guo and Huang 2008; Ruiz and Hernández 2018) form the three primary modeling approaches for estimating pedestrian walking behavior over different facilities and across different geometries.

5 Pedestrian Flow Characteristics Over Different Facilities …

81

Fig. 5.1 Types of pedestrian facilities (Source Exhibit 3.14, US-HCM (2010))

The dynamic modeling approach is more of a macroscopic modeling approach, whereas continuous and discrete modeling approaches are microscopic in nature. The current chapter presents the types of pedestrian facilities, different pedestrian flow characteristics, methods of field data collection, microscopic/macroscopic modeling approaches, pedestrian level of service, application of soft computing approaches in pedestrian studies, and the way forward.

5.2 Types of Pedestrian Facilities (At-Grade and Grade-Separated) In India, the most common type of pedestrian facilities are at-grade (sidewalk, walkway, and crosswalk) and grade-separated (FOB, skywalk, underpass, and stairway). Figure 5.1 shows the different pedestrian facilities as per US-HCM (2010). Table 5.1 shows the description of the different at-grade or grade-separated pedestrian facilities available for either crossing or through movement of pedestrians.

82

A. Banerjee and A. K. Maurya

Table 5.1 Definition of different pedestrian facilities under Indian scenario (as per Indo-HCM (2018)) Facility

Grade

Important descriptions

Sidewalk/footpath

At-grade

• Raised path along side of road • Separated from motorized vehicles by curbs and guardrails • Accommodate the highest volume of pedestrians • Width varies in the range 1.5–5 m

Walkway

• Located far away from vicinity of motorized traffic • It is unraised and wider than sidewalk

Crosswalk

• Provided for safe and easy crossing • Can be provided at intersections (signalized/unsignalized) or mid-blocks

Overpass (FOB and skywalk) Grade-separated • Elevated facilities which either allow easy and continuous crossing access (FOB) or connect one strategic location to another (skywalk) • Should be provided with ramps/elevators/escalators along with stairways Underpass (subway)

• Underground facilities which allow safe dispersal from one side of the road to another

Stairway/escalators elevator

• These are facilities which provide vertical connectivity between grade-separated and at-grade facilities

5.3 Parameters Related to Pedestrian Flow Characteristics and Infrastructure Geometry 5.3.1 Parameters Related to Pedestrian Flow Characteristics The pedestrian walking behavior over various facilities is dependent on different macroscopic and microscopic factors. The macroscopic parameters are related to the speed, flow, density, space, and the relationship among these parameters. The macroscopic parameters are extracted using the videography data collection technique. The developed relationships between the different macroscopic factors help in developing fundamental diagrams (FD)/fundamental relationships between the parameters which can be useful for predicting the free-flow speed (FFS), jam density (Kj), maximum flow rate (capacity), and level of service estimation of the facility. The different macroscopic factors along with the fundamental diagrams are discussed as follows: Pedestrian flow: This represents the number of pedestrians crossing a particular section of a facility over a specified time (ped/min). However, while analyzing one

5 Pedestrian Flow Characteristics Over Different Facilities …

83

generally uses the term flow rate, i.e., the flow over the effective width considered (ped/min/m) as one generally tends to compare or combine different locations together. Pedestrian speed: The walking speed of the pedestrian refers to the time taken to cross a particular trap length (i.e., distance between two sections) over a specified time. It is represented as m/s. However, while developing fundamental diagrams, m/s is converted to m/min. Pedestrian density: The density refers to the average number of pedestrians over a defined walkway or queuing area. The factor is generally represented as ped/m2 . The minimum density is generally observed at free-flow speed, i.e., when pedestrians are able to choose their walking speed over different facilities. However, at jam density condition the walking speed becomes zero, i.e., pedestrians come to standstill. Pedestrian space: The area module or walking space is defined as the average area required by the pedestrian to pass through a particular facility. Space is considered as the inverse of density, i.e., in m2 /ped. Indo-HCM (2018) suggests a minimum area of 0.18m2 and 0.26m2 for pedestrians without and with luggage, respectively. The space is an important parameter for defining LOS. Fundamental diagrams: The fundamental diagrams represent the relationships between different macroscopic pedestrian flow characteristics. The development of such diagrams help in estimating the capacity and the LOS of the facility. Across different studies, the authors preferred to use either linear or exponential speed– density relationships. Table 5.2 shows the different studies conducted over various facilities and the observed/adopted speed–density relationships. Figure 5.2a–c shows the different speed–density relationships established for sidewalks, walkways, and stairways. Apart from the macroscopic factors, understanding the microscopic factors helps in the development of models for the prediction of walking speeds over different facilities. These microscopic factors help in the identification of the relevant factors which affect the walking behavior over the different facilities. Some of the definitions of such factors are discussed as follows: Age: The age of the pedestrian plays a significant role in determining the walking speed over different facilities. In general, the age is divided into 3 categories: Table 5.2 Developed relationships across different facilities Facility

Authors

Speed–density relationship

Sidewalk/walkway

Oeding (1963), Older (n.d), Navin and Wheeler (n.d.), Bargegol and Gilani (2015), Rungta and Sharma (2016)

Linear/exponential

Stairway/FOB

Fruin (1971), Daly et al. (1991), Weidmann (1993), Shah et al. (2017a), Indo-HCM (2018)

Linear

84

A. Banerjee and A. K. Maurya

Fig. 5.2 Speed–density relationships for sidewalk, walkway, and stairway facilities (Source Banerjee et al. (2018))

55 years. However, dividing the age into 4 or 5 categories gives a better representation of the actual age category which affects the speed. Studies across different facilities (sidewalks, crosswalks, and stairways) reported that young pedestrians (55 years) pedestrians (Fruin 1987; Montufar et al. 2007; Patra et al. 2017; Shah et al. 2017b). Gender: Gender plays a crucial role while defining the walking speed of the pedestrians over different facilities. In general, male pedestrians are observed to walk at a higher speed than females by 7.7 m/min, 7 m/min, 6.6 m/min, 5.6 m/min, and 7.6 m/min across sidewalk, walkway, crosswalk, ascending stairway, and descending stairway (Polus et al. 1983; Chandra et al. 2014; Siddharth and Vedagiri 2018; Gore et al. 2020). Luggage: The size and weight of the luggage plays an important role in the walking speed, especially when pedestrians are traversing slopes/stairways. The luggage generally considered across different studies are trolleys, side bags, backpacks, carrying children, etc. Previous studies reported that depending on the type and grade of the facility, walking with luggage reduced the walking speed by 3–15% (Shah et al. 2017b; Morrall et al. 1991; Laxman et al. 2010).

5 Pedestrian Flow Characteristics Over Different Facilities …

85

Use of handheld devices: The use of mobile phone/handheld devices tends to distract the pedestrians and thus influence their walking speeds. The reduction of walking speed while using the phone, texting, and calling were 3 m/min, 1.2 m/min, and 15.6 m/min (New York Pedestrian Study 2006; Krasovsky et al. 2017; Reynolds 1999). Group size/group formation: The size and formation of groups tend to impact the walking speed of the pedestrians while using different pedestrian facilities. When a pedestrian walks alone in comparison to walking in a group of 2 or more pedestrians, their walking speed is impacted. Previous studies reported that while walking in groups, pedestrians tend to engage in talking or observe the surroundings more. This leads to a reduction in walking speed by 12–20% (Laxman et al. 2010; New York Pedestrian Study 2006; Reynolds 1999; Al-Masaeid et al. 1993; Rastogi et al. 2011; Vanumu et al. 2017). Disability: In general, while studying pedestrian walking behavior, one may have the tendency to overlook specially abled/disabled pedestrians. However, they form an integral part of the pedestrian network and their inclusion/exclusion can affect the overall walking speed of the system. Previous researchers (Clark-Carter et al. 1986; Wrigbt et al. 1999; Arango and Montufar 2008; Sharifi et al. 2016) reported that under complex environments, the walking ability of pedestrians with special-ability decreased. Direction of movement: The direction of movement (i.e., whether a pedestrian is ascending/descending stairways or walking in a major/minor direction of flow) impacts the overall walking speed. The studies conducted by previous researchers (Daly et al. 1991; Lee and Lam 2006; Alhajyaseen et al. 2011) reported that if a pedestrian was walking in the minor flow direction or was ascending stairways, then their walking speeds significantly decreased. The speed and flow rate were the most common macroscopic factors, while gender, age, and luggage were the commonly used microscopic factors considered across different studies in the past. However, in order to estimate the pedestrian flow characteristics, it is essential to estimate the density (macroscopic factor) along with the mobile use, group size, movement direction, and disability (microscopic factors) as these factors significantly affect the walking behavior of the pedestrians.

5.3.2 Parameters Related to Infrastructure/Geometry Apart from different pedestrian flow characteristics, there are infrastructure/geometry-related factors, which can influence the walking behavior of the pedestrians. These factors are related to the physical characteristics as well as user characteristics [IRC-103 (Guidelines for Pedestrian Facilities 2012)]. Physical characteristics are the factors which the pedestrians observe while using the facility and which can affect their walking behavior/choice. On the other hand, the user

86

A. Banerjee and A. K. Maurya

Table 5.3 Description of different physical and user characteristics which influence pedestrian behavior Characteristics

Description

Physical characteristics Surface

It indicates whether there are any cracks or bumps to comfortable walking

Width

It defines the walkable width of the facility which can accommodate enough pedestrians for comfortable travel

Obstructions

The presence of garbage bins, trees, electric poles, and hoardings can obstruct the travel path of the pedestrians

Encroachments

The encroachments are generally in the form of vendors/beggars/standing pedestrians which reduce the overall walkable width of the facility

Potential for vehicular conflict To segregate the pedestrians (using the at-grade) facilities from the vehicular traffic, guardrails and raised facilities can be helpful Continuity

For disabled and old pedestrians, the continuity of a particular walkway helps in easy traversing

User characteristics Safety and Security

Safety is generally ensured by segregating pedestrians from vehicular traffic by guard rails/raised walkways. The security is ensured through proper street lighting, police patrolling, and installation of CCTV cameras

Comfort

Comfort is generally measured through the proper shade, provision of chairs/benches, and washrooms

Walk Environment

The surroundings (i.e., whether clean and free from stinks) provide a pleasant travel experience

characteristics are the factors which the pedestrians perceive while using a particular facility, and they can influence their choice toward using a particular facility. Table 5.3 shows the different physical and user characteristics which impact walking behavior. The past studies reported that among different physical and user characteristics, the most common factors used by different researchers were width, surface, safety and security, and comfort level of the facility (Banerjee et al. 2018). However, considering the other factors mentioned in Table 5.3 would be beneficial to the development of better infrastructures for pedestrians.

5.4 Data Collection Techniques Researchers across the globe use quantitative (videography technique), qualitative (questionnaire/perception survey), controlled (experimental), manual counting, and

5 Pedestrian Flow Characteristics Over Different Facilities …

87

semi-manual data collection techniques. The initial studies on pedestrian behavior used manual counters and tally mark sheets to calculate the pedestrian flow. However, the major issue with such a technique was that it is quite labor-intensive which limits the extracted amount of data. Subsequently, majority of the past studies used either videography or questionnaire techniques for data collection, and some even used both the techniques. Table 5.4 discusses the pros and cons of the different data collection techniques for pedestrian walking behavior study. Table 5.4 shows that across different grade and at-grade facilities, commonly extracted parameters for qualitative survey were demographics (age, gender, and luggage), safety, and width (Parida and Parida 2008; Bivina et al. 2018; Rankavat and Tiwari 2020; Banerjee et al. 2020; Banerjee and Maurya 2020). Similarly, in the case of videography data collection, apart from demographics other common parameters extracted were flow rate, speed, density, and space (Shah et al. 2017b; Marisamynathan and Lakshmi 2016; Sahani and Bhuyan 2017; Patra et al. 2020). Controlled experiments were also set up to study pedestrian behavior under different geometric conditions. Mostly, these studies were conducted to understand the walking behavior of pedestrians under different types of geometries like bottlenecks, corridors, etc. These studies explored to understand the impact of flow direction, pedestrian–pedestrian interactions, and flow characteristics (Daamen and Hoogendoorn 2003; Helbing et al. 2007; Seyfried et al. 2009; Chattaraj et al. 2009). Apart from the videography/questionnaire/controlled data collection techniques, recently a semi-manual data collection technique has been developed by Banerjee et al. (Banerjee et al. 2021). To study the pedestrian path/trajectory, top-down videos are essential, however, getting top-down videos (to capture trajectory data over facilities) is difficult where shade is available at low heights. In the semi-manual method, the camera can be placed on a tripod stand with a frontal camera angle (similar to the CCTV), and data can be extracted using a vanishing point method [as developed by Fung et al. (2003)]. Even from high vantage points, it could be difficult to mark the foot of each pedestrian (especially at moderate to high densities) to get the position coordinates of the pedestrians. Thus, this method tries to mark the head of the pedestrians (considering the appropriate elevation corrections) and converts the video image coordinates to real-world coordinates. Apart from pedestrian coordinates, this technique can also provide body dimensions, trajectories, gaps maintained by pedestrians, and other microscopic features. Figure 5.3 shows the snapshot of the semi-manual technique.

5.5 Level of Service (LOS) The level of service is defined as the “quantitative stratification of a performance measure that represents the quality of service, measured on an A-F scale, with LOS A representing the best-operating conditions from the traveler’s perspective and LOS F the worst (US-HCM 2010)”. The performance of a facility under the existing

Description

Questionnaire/qualitative/perception In survey interviewer-administered questionnaire survey, the respondents are randomly selected and those willing to undergo entire survey are interviewed

Type of data collection

1. The sample size per location needs to be estimated 2. Hoardings should be set up to capture the view of the pedestrians and significance of the study

Special cautions to be considered

Table 5.4 Data collection technique and crucial parameters considered

1. Respondent rate is low (especially during peak hours) 2. Enumerators should be well trained

Limitations

Facility

Grade-separated (overpass/underpass)

1. Perception At-grade under current (sidewalk/walkway/crosswalk) existing conditions can be reported 2. Future required improvements can be detected

Advantages

(continued)

Demographics, height of facility, length of facility, walkable width, safety, security, illumination, comfort, walk environment, slope, and surface

Demographics (age, gender, luggage), surface, comfort, accessibility, safety, security, obstruction, width, and actual and perceived risks

Parameters considered for study

88 A. Banerjee and A. K. Maurya

Description

It is the most primitive method of data collection where pedestrian movement is captured using cameras set up at high vantage points

Type of data collection

Videography/quantitative survey

Table 5.4 (continued)

1. Camera needs to be set up at high vantage point 2. The trap length and total width need to be measured 3. Duration of data collection should be well planned in advance

Special cautions to be considered 1. Positioning of camera can impact the data collected 2. The enumerators need to be trained in the functioning of the camera

Limitations

Facility

Grade-separated (overpass/underpass)

1. Permanent At-grade record (sidewalk/walkway/crosswalk) available 2. Macroscopic and microscopic factors can be extracted

Advantages

(continued)

Demographics, group size, use of handheld devices, travel direction, speed, flow rate, density, space, geometric dimensions, direction of movement, and travel time

Demographics, flow rate, speed, density, area module/space, width, vehicle volume, obstruction, v/c ratio, pedestrian volume, number of lanes, and delay

Parameters considered for study

5 Pedestrian Flow Characteristics Over Different Facilities … 89

Description

Locations where field data collection is difficult, researchers develop experimental set up to study pedestrian walking behavior

Type of data collection

Controlled experiments

Table 5.4 (continued)

1. The sample size needs to be predefined 2. The pedestrians participating in the experiment need to be well aware of the study

Special cautions to be considered 1. Actual walking behavior available from the field may not be captured 2. The sample size may be restricted

Limitations

Facility

1. With a Bottleneck (wide, narrow), small sample corridor (straight, ring shaped, size of circular, angled) 20–100 pedestrians, the study can be conducted 2. Different geometries (bottlenecks and corridors) can be set up

Advantages

Flow under normal and panic situations, step length and frequency, speed, and characteristics

Parameters considered for study

90 A. Banerjee and A. K. Maurya

5 Pedestrian Flow Characteristics Over Different Facilities …

91

Fig. 5.3 Snapshot of developed semi-manual technique (Source Banerjee et al. (2021))

conditions and subsequent need for its redesigning depends on the LOS of the facility. LOS of a facility can be defined by different ranges (4–6) for either qualitative method (using perception) or quantitative method (by videography) or a combination of both. The development of LOS helps in understanding the current situation and thus improving the facilities in the near future. Table 5.5 shows the different studies related to LOS which have been conducted using both techniques. Table 5.5 Studies related to quantitative and qualitative LOS development Data collection technique Facility studied adopted Qualitative/Perception survey

Measures of effectiveness

Sidewalk/Walkway/Stairway Sarkar (2003), Perception score Marisamynathan and Lakshmi (2018), Parvathi (2018), Bivina and Parida (2019)

Quantitative/videography

Quantitative and Qualitative

Author

Sidewalk/Stairways/FOB

Fruin (Fruin 1971), Polus et al. (1983), Sahani and Bhuyan (2017), Rastogi et al. (2014)

Flow rate, density, speed, space, v/c ratio

IRC-103 (2012), US-HCM (2010), Indo-HCM (2018)

Quantitative: Flow rate, space Qualitative: Perception score

92

A. Banerjee and A. K. Maurya

5.6 Modeling Approaches Related to Pedestrian Studies The movement of pedestrians is generally modeled using a dynamic approach, continuous approach, or discrete approach. The dynamic approach deals with pedestrian behavior at the macroscopic level, while continuous and discrete models deal with pedestrian motion at the microscopic level.

5.6.1 Dynamic Approach It is a macroscopic modeling approach based on Navier–Stokes or Boltzmann equations, and which considers the pedestrians as gas or fluid. Henderson (1971) used the Navier–Stokes equation to predict the motion of pedestrians. Thereafter, Helbing (1991) improved Henderson’s equation to develop a fluid dynamic model based on the gas kinetic theory of the Boltzmann constant. Further, Pauls (1995), Hughes (2002), and Colombo et al. (2012) used fluid dynamics to predict the walking behavior of pedestrians. The main drawback of dynamic modeling is that pedestrian interaction at the individual level is not available.

5.6.2 Discrete Approach The discrete approach is portrayed by the Cellular Automata (CA) model where the behavior of individuals is defined using a discrete particle-based hopping model using a certain set of local rules. The CA approach was used by Nagel and Schreckenberg (1992) on vehicular traffic which was later adopted by Blue et al. (1997) for pedestrian motion. The developed approach could imitate the goals of the pedestrians in reality, such as avoiding conflicts and minimizing traveled distance. As per the CA model, the subject pedestrian finds its way from origin cell to destination cell using a certain set of rules. The model was further modified by Blue and Alder (1998, 1999, 2000, 2001) under normal situations for different directional flows (uni-, bi-, and multi-) using various cell sizes. Figure 5.4 shows the pedestrian movement captured using the CA model. Later on, using the developed CA concept, researchers tried to model the pedestrians’ movement for normal situations (Fukui and Ishibashi 1999; Lämmel and Flötteröd 2015) and evacuation scenarios (Burstedde et al. 2001; Guo and Huang 2008; Ruiz and Hernández 2018). The different interesting features observed at normal/evacuation scenarios were back-stepping, collective phenomenon, and oscillation at doors.

5 Pedestrian Flow Characteristics Over Different Facilities …

93

Fig. 5.4 Illustration of pedestrian movements (Blue et al. 1997)

5.6.3 Continuous Approach Based on the social fields’ concept of Lewin (1951), the social force model (a continuous modeling approach) was developed to model the pedestrian movement. The first social force model was developed by Hirai and Tarui (1975) inspired by a model developed by Suzuki and Sakai (1973) based on the motion of shoals. Later, Okazaki (1979) used the magnetic force to develop a social force model. However, in order to overcome the limited computational power of the developed models, Helbing and Molnár (1995) proposed the social force model for normal situations. The model was based on three formulations: a. b. c.

Reaching destination using the shortest path. A repulsive effect from other pedestrians and objects (walls). An attractive effect from other pedestrians and objects (window displays).

The developed model was further modified (by adding new parameters or modifying existing parameters) by various researchers (Lakoba et al. 2005; Parisi and Dorso 2007; Zhou et al. 2020) for normal and emergency/evacuation situations. Different parameters (like memory effect, respect area, panic, self-stopping mechanism, and collision avoidance) were introduced to analyze pedestrian interactions at microscopic levels. Features such as reduction of pushing behavior, self-slow mechanism, clogging, and lane formation were observed under normal/evacuation scenarios. Some major benefits of the social force model over the cellular automata model are: • Gives a realistic description of pedestrian movement at a microscopic level considering pedestrian intentions, desired velocities, and pair interactions. • Self-organizing phenomena are well portrayed.

94

A. Banerjee and A. K. Maurya

• Flexible while considering geometry. • Commercially available software Viswalk (2013) which has a better graphical user interface used a social force model to predict the pedestrians’ movement behavior. Moreover, the developed model can also be calibrated using a genetic algorithm through the COM interface in MATLAB. In general, the Mean Absolute Percentage Error (MAPE) between the observed and simulated velocities of the pedestrian are calculated to validate the performance of the model.

5.7 Application of Soft Computing Tools in Pedestrian-Related Studies Soft computing has become an integral part of the transportation sector. Different soft computing approaches such as tree-based methods (random forest, gradient boosting machine, decision trees, etc.) and neural networks (artificial neural network and deep neural network) have found significant importance in addition to traditional methods such as regression modeling. These methods use different hyper-parameters to tune the best fit model to predict the training and testing data. The evaluation metrics commonly used are mean absolute percentage error (MAPE), root mean squared error (RMSE), and mean absolute error (MAE). Table 5.6 shows the different studies which have been conducted using soft computing approaches. Although the application of soft computing in the field of vehicular traffic is widely explored, however, its application in pedestrian-based research is limited. Table 5.6 Application of soft computing approaches in pedestrian-related studies Soft computing technique used

Author

Type of study

Evaluation metric used

ANN

Yuen et al. (2014)

Intelligent based route – choice modeling

ANN

Das et al. (2015)

Interrelationships between flow parameters

MAE, RMSE

CNN

Dong et al. (2016)

Automatic age estimation

–

ANN

Cohen and Dalyot (2019)

Pedestrian traffic flow prediction

–

ANN

Chakraborty et al. (2019)

Crash prediction modeling

MAE, RMSE, R2

GLM, GBM, RF

Banerjee et al. (2020)

Overpass utilization modeling

AUC, MSE, Accuracy

Note ANN—Artificial Neural Network, CNN—Convolution Neural Network, GLM—Generalized Linear Model, GBM—Gradient Boosting Machine, RF—Random Forest, MAE—Mean Absolute Error, RMSE—Root Mean Squared Error, R2—Coefficient of Determination, AUC—Area Under Curve, and MSE—Mean Squared Error

5 Pedestrian Flow Characteristics Over Different Facilities …

95

5.8 Concluding Remarks and Way Forward The current study tried to present an overview of pedestrian facilities, flow characteristics, techniques for data collection, different modeling approaches, level of service, and application of soft computing techniques. The major findings of the study are as follows: • Across the globe, researchers majorly used qualitative or quantitative data collection techniques. Wherever data collection from the field was cumbersome, the researchers used controlled experiments across different geometries. • Majority of the studies developed linear/exponential relationships between speed and density for different at-grade and grade-separated facilities. • The most common parameters related to pedestrian flow characteristics which were considered across different studies, were macroscopic: speed and flow rate; and microscopic: gender, age, and luggage condition. • In the case of parameters related to pedestrian infrastructure, the most common factors considered were width, surface, safety–security, and comfort. • Majority of the studies either used perception survey or videography survey techniques to develop LOS standards. • Macroscopic parameters (flow rate, density, speed, and space) and perception score were important measures of effectiveness while developing LOS using perception and videography techniques, respectively. • Dynamic modeling only represents the pedestrians’ interaction at the macroscopic level. However, continuous (social force model) and discrete (cellular automate) models could replicate the microscopic interactions. • Self-organizing phenomena such as back-stepping, collective phenomenon, pushing behavior, clogging, and lane formation could be observed under normal/evacuation scenarios. • Soft computing techniques have not been well explored for pedestrian-related studies. The majority of the studies used ANN to predict the flow characteristics and crash prediction. Some important challenges for the future research direction are as follows: • Applying both qualitative and quantitative techniques for data collection, which might provide information on both the actual ground condition as well as the perception of the pedestrians. This can be useful for the development of LOS standards for different facilities as well. • Apart from linear/exponential speed–density models, other relationships can also be explored. • While studying microscopic parameters, the size of the group, use of handheld devices, direction of movement, and disability need to be considered to have a better understanding. • Grade-separated facilities need to be studied in more detail under the Indian scenario.

96

A. Banerjee and A. K. Maurya

• Model development should focus on studying pedestrian interactions at microscopic levels to predict the self-organizing phenomenon. • Application of well-calibrated and validated simulation models can be explored to evaluate the crowd management strategies during large gatherings. • The development and application of soft computing approaches should be encouraged for different pedestrian studies.

References Alhajyaseen WK, Nakamura H, Asano M (2011) Effects of bi-directional pedestrian flow characteristics upon the capacity of signalized crosswalks. Procedia Soc Behav Sci 16:526–535 Al-Masaeid HR, Al-Suleiman TI, Nelson DC (1993) Pedestrian speed-flow relationship for central business district areas in developing countries. Transp Res Rec 1396:69–74 Arango J, Montufar J (2008) Walking speed of older pedestrians who use canes or walkers for mobility. Transp Res Rec 2073(1):79–85 Banerjee A, Raoniar R, Maurya AK (2020) Pedestrian overpass utilization modeling based on mobility friction, safety and security, and connectivity using machine learning techniques. J Soft Comput Springer 24(22):17467–17493 Banerjee A, Maurya AK (2020) Planning for better skywalks systems using perception of pedestrians: a case study of Mumbai, India. Journal of urban planning and development, American society of civil engineers (ASCE), Volume 146, Issue 2 Banerjee A, Budhkar AK, Maurya AK (2021) Development of a semi manual approach for extraction of inter-pedestrian interactions at an overpass facility. In the 100th transportation research board annual meeting, Washington D.C Bargegol I, Gilani VNM (2015) The effect of rainy weather on walking speed of pedestrians on sidewalks. Buletin Teknol. Tanaman 12:217–222 Banerjee A, Maurya AK, Lämmel G (2018) A review of pedestrian flow characteristics and level of service over different pedestrian facilities. Collective Dyn 3:1–52 Bivina GR, Parida P, Advani M, Parida M (2018) Pedestrian level of service model for evaluating and improving sidewalks from various land uses. Europ Transp Trasporti Europei 67(2) Bivina GR, Parida M (2019) Modelling perceived pedestrian level of service of sidewalks: a structural equation approach. Transport 34(3):339–350 Blue VJ, Embrechts MJ, Adler JL (1997) Cellular automata modeling of pedestrian movements. In: 1997 IEEE international conference on systems, man, and cybernetics. computational cybernetics and simulation 3:2320–2323. IEEE Blue VJ, Adler JL (1998) Emergent fundamental pedestrian flows from cellular automata microsimulation. Transp Res Rec 1644(1):29–36 Blue VJ, Adler JL (1999) Cellular automata microsimulation of bidirectional pedestrian flows. Transp Res Rec 1678(1):135–141 Blue VJ, Adler JL (2000) Modeling four-directional pedestrian flows. Transp Res Rec 1710(1):20– 27 Blue VJ, Adler JL (2001) Cellular automata microsimulation for modeling bi-directional pedestrian walkways. Transp Res Part B: Methodol 35(3):293–312 Burstedde C, Klauck K, Schadschneider A, Zittartz J (2001) Simulation of pedestrian dynamics using a two-dimensional cellular automaton. Physica A 295(3–4):507–525 Chakraborty A, Mukherjee D, Mitra S (2019) Development of pedestrian crash prediction model for a developing country using artificial neural network. Int J Inj Contr Saf Promot 26(3):283–293 Chattaraj U, Seyfried A, Chakroborty P (2009) Comparison of pedestrian fundamental diagram across cultures. Adv Complex Syst 12(03):393–405

5 Pedestrian Flow Characteristics Over Different Facilities …

97

Chandra S, Rastogi R, Das VR, Ilango T (2014) Pedestrian behaviour under varied traffic and spatial conditions. Trasporti Europei (56) Clark-Carter DD, Heyes AD, Howarth CI (1986) The efficiency and walking speed of visually impaired people. Ergonomics 29(6):779–789 Colombo RM, Garavello M, Lécureux-Mercier M (2012) A class of nonlocal models for pedestrian traffic. Math Models Methods Appl Sci 22(04):1150023 Cohen A, Dalyot S (2019) Pedestrian Traffic flow prediction based on ANN model and OSM data. In: Proceedings of the ICA (Vol. 2, pp. NA-NA). Copernicus GmbH Daamen W, Hoogendoorn SP (2003) Controlled experiments to derive walking behaviour. Eur J Transp Infrastruct Res 3(1):39–59 Das P, Parida M, Katiyar VK (2015) Analysis of interrelationship between pedestrian flow parameters using artificial neural network. J Modern Transp 23(4):298–309 Daly PN, McGrath F, Annesley TJ (1991) Pedestrian speed/flow relationships for underground stations. Traffic Eng Control 32(2):75–78 Dong Y, Liu Y, Lian S (2016) Automatic age estimation based on deep learning algorithm. Neurocomputing 187:4–10 Fruin JJ (1971) Pedestrian planning and design. Metropolitan association of urban designers and environmental planners, New York, pp 2–6 Fruin JJ (1987) Pedestrian planning and design, revised. Elevator World Inc., Mobile, AL Fukui M, Ishibashi Y (1999) Self-organized phase transitions in cellular automaton models for pedestrians. J Phys Soc Jpn 68(8):2861–2863 Fung GS, Yung NH, Pang GK, Lai AHS (2003) Camera calibration from road lane markings. Opt Eng Bellingham Int Soc Opt Eng 42(10): 2967–2977 Golakiya HD, Patkar M, Dhamaniya A (2019) Impact of midblock pedestrian crossing on speed characteristics and capacity of urban arterials. Arab J Sci Eng 44(10):8675–8689 Gore N, Dave S, Shah J, Jain M, Rathva D, Garg V (2020) Comparative analysis of pedestrian walking speed on sidewalk and carriageway. In: Transportation research (pp 65–76). Springer, Singapore Guo RY, Huang HJ (2008) A modified floor field cellular automata model for pedestrian evacuation simulation. J Phys A: Math Theoret 41(38):385104 Herms BF (1972) Pedestrian crosswalk study: accidents in painted and unpainted crosswalks. Highway Res Rec 406:1–13 Hirai K, Tarui K (1975) A simulation of the behavior of a crowd in panic. In Proceedings of the 1975 international conference on cybernetics and society (pp 409–411) Helbing D, Molnár P (1995) Social force model for pedestrian dynamics. Phys Rev E 51(5):4282 Helbing D, Johansson A, Al-Abideen HZ (2007) Dynamics of crowd disasters: an empirical study. Phys Rev E 75(4):046109 Henderson LF (1971) The statistics of crowd fluids. Nature 229:381–383 Helbing D (1991) A mathematical model for the behaviour of pedestrians. Behavioural Science 36:298–310 Hughes RL (2002) A continuum theory for the flow of pedestrians. Transp Res Part B: Methodol 36(6):507–535 Indian Roads Congress (2012) Guidelines for pedestrian facilities, IRC: 103–2012. First revision Koepsell T, McCloskey L, Wolf M, Moudon AV, Buchner D, Kraus J, Patterson M (2002) Crosswalk markings and the risk of pedestrian–motor vehicle collisions in older pedestrians. JAMA 288(17):2136–2143 Krasovsky T, Weiss PL, Kizony R (2017) A narrative review of texting as a visually-dependent cognitive-motor secondary task during locomotion. Gait Posture 52:354–362 Lakoba TI, Kaup DJ, Finkelstein NM (2005) Modifications of the Helbing-Molnar-Farkas-Vicsek social force model for pedestrian evolution. Simulation 81(5):339–352 Lee JY, Lam WH (2006) Variation of walking speeds on a unidirectional walkway and on a bidirectional stairway. Transp Res Rec 1982(1):122–131

98

A. Banerjee and A. K. Maurya

Lämmel G, Flötteröd G (2015) A CA model for bidirectional pedestrian streams. Procedia Comput Sci 52:950–955 Lewin K (1951) Field theory in social science: selected theoretical papers (Edited by Dorwin Cartwright.) Laxman KK, Rastogi R, Chandra S (2010) Pedestrian flow characteristics in mixed traffic conditions. J Urban Plann Dev 136(1):23–33 Manual IHC, (Indo-HCM) (2018) Council of scientific and industrial research. New Delhi, India Manual HC, (US-HCM) (2010) Transportation research board. National Research Council, Washington, DC Marisamynathan S, Lakshmi S (2016) Performance analysis of signalized intersection at metropolitan area. J Adv Res Appl Sci Eng Tech 2(1):19–29 Marisamynathan S, Lakshmi S (2018) Method to determine pedestrian level of service for sidewalks in Indian context. Transp Lett 10(5):294–301 Montufar J, Arango J, Porter M, Nakagawa S (2007) Pedestrians’ normal walking speed and speed when crossing a street. Transp Res Rec 2002(1):90–97 Morrall JF, Ratnayake LL, Seneviratne PN (1991) Comparison of central business district pedestrian characteristics in Canada and Sri Lanka. Transp Res Record (1294) Nagel K, Schreckenberg M (1992) A cellular automaton model for freeway traffic. J Phys I 2(12):2221–2229 Navin FP, Wheeler RJ (1969) Pedestrian flow characteristics. Traffic Eng Inst Traffic Engr, 39 New York Pedestrian Study, New York City, 2006. Pedestrian Level of Service, Phase I, Dept. of City Planning, Transportation Division, New York Oeding D (1963) Verkehrsbelastung und Dimensionierung von Gehwegen und anderen Anlagen des FuÞga«ngerverkehrs” [Traffic volume and dimensioning of footways and other facilities of pedestrian traffic], StraÞenbau und StraÞenverkehrstechnik series number 22, Ministry of Traffic, Bonn Okazaki S (1979) A study of pedestrian movement in architectural space, part 1: Pedestrian movement by the application on of magnetic models. Trans AIJ 283:111–119 Older SJ (1968) Movement of pedestrians on footways in shopping streets. Traffic Eng Control 10(4) Patra M, Sala E, Ravishankar KVR (2017) Evaluation of pedestrian flow characteristics across different facilities inside a railway station. Transp Res Procedia 25:4763–4770 Polus A, Schofer JL, Ushpiz A (1983) Pedestrian flow and level of service. J Transp Eng 109(1):46– 56 Parida P, Parida M (2008) Qualitative level of service for sidewalks in Delhi. In: Proceedings of international conference on the best practices to relieve congestion on mixed-traffic urban streets in developing countries, IIT Madras, Chennai (pp 295–304) Patra M, Perumal V, Rao KK (2020) Modelling the effects of risk factor and time savings on pedestrians’ choice of crossing facilities at signalised intersections. Case Studies Trans Policy 8(2):460–470 Parvathi MS (2018) The analysis of factors affecting the pedestrian level of service on footpaths in Uppal X road at Hyderabad. Int J Scientific Res Rev 7(9):738–754 Parisi DR, Dorso CO (2007) Why “faster is slower” in evacuation process. In Pedestrian and evacuation dynamics 2005 (pp 341–346). Springer, Berlin, Heidelberg PTV AG (2013) Behaviour parameter file: Parameters Vissim 6.00 Pauls J (1995) Movement of people. Din Nenno, Washington Rankavat S, Tiwari G (2020) Influence of actual and perceived risks in selecting crossing facilities by pedestrians. Travel Behaviour Soc 21:1–9 Rastogi R, Chandra S, Mohan M (2014) Development of level of service criteria for pedestrians. J Indian Roads Congress 75(1): 61–70 Rastogi R, Thaniarasu I, Chandra S (2011) Design implications of walking speed for pedestrian facilities. J Transp Eng 137(10):687–696

5 Pedestrian Flow Characteristics Over Different Facilities …

99

Reynolds CW (1999) Steering behaviors for autonomous characters. In: Game developers conference (Vol 1999, pp 763–782) Rungta A, Sharma S (2016) Effects of various parameters on pedestrian characteristics in hilly urban area. J Adv Res Civil Environ Eng 3(2):7–23 Ruiz S, Hernández B (2018) A Hybrid reinforcement learning and cellular automata model for crowd simulation on the GPU. In: Latin American high performance computing conference (pp 59–74). Springer, Cham Sahani R, Bhuyan PK (2017) Pedestrian level of service criteria for urban off-street facilities in mid-sized cities. Transport 32(2):221–232 Sarkar S (2003) Qualitative evaluation of comfort needs in urban walkways in major activity centers. Transp Q 57(4):39–59 Seyfried A, Passon O, Steffen B, Boltes M, Rupprecht T, Klingsch W (2009) New insights into pedestrian flow through bottlenecks. Transp Sci 43(3):395–406 Shah J, Joshi GJ, Parida P, Arkatkar SS (2017a) Effect of directional distribution on stairway capacity at a suburban railway station. Transp Lett 9(2):70–80 Shah JH, Joshi GJ, Arkatkar SS, Parida M (2017) Impact of human factors and functional characteristics of location on walking speed at stairway facility. In: 96th Annual meeting transportation research board, No. 17–06476 Siddharth SMP, Vedagiri P (2018) Modeling the gender effects of pedestrians and calibration of the modified social force model. Transp Res Rec 2672(31):1–9 Suzuki R, Sakai S (1973) Movement of a group of animals. Biophysics 13(281–282) Sharifi MS, Stuart D, Christensen K, Chen A, Kim YS, Chen Y (2016) Analysis of walking speeds involving individuals with disabilities in different indoor walking environments. J Urban Plann Dev 142(1):04015010 Tanaboriboon Y, Hwa SS, Chor CH (1986) Pedestrian characteristics study in Singapore. J Transp Eng 112(3):229–235 Vanumu LD, Rao KR, Tiwari G (2017) Analysis of pedestrian group behaviour. In: Proceedings of the transportation research board 96th annual meeting (No. 17–04866) Weidmann U (1993) Transport technique of pedestrian. Schriftenreihe Ivt-Berichte, 90 Wrigbt MS, Cook GK, Webber GMB (1999) Emergency lighting and wayfinding provision systems for visually impaired people: Phase of a study. Int J Light Res Technol 31(2):35–42 Yuen JKK, Lee EWM, Lam WWH (2014) An intelligence-based route choice model for pedestrian flow in a transportation station. Appl Soft Comput 24:31–39 Zhou J, Li S, Nie G, Fan X, Xia C (2020) Developing a revised social force model for pedestrians’ earthquake emergency evacuation. Geomat Nat Haz Risk 11(1):335–356

Chapter 6

Emerging Traffic Data Collection Practices Under Mixed Traffic Conditions: Challenges and Solutions Anuj Kishor Budhkar, Gowri Asaithambi, Akhilesh Kumar Maurya, and Shriniwas S. Arkatkar

6.1 Introduction and Background The traffic in developing countries including India, Nepal, Vietnam, Bangladesh, etc., consists of higher proportion of vehicles other than cars, moving in a disorderly manner as compared to conventional traffic streams, and is commonly termed as ‘mixed traffic’. This results in a unique traffic movement, wherein drivers interact not just with the front vehicles, but also with vehicles present on their sides. Drivers in such streams maintain certain lateral gaps and move about creating their own channels, tailgate other vehicles in a staggered manner, swerve and maintain higher acceleration/deceleration which generally lead to a chaotic and complex disorderly vehicular movement. In order to plan, design and manage such roadway facilities operating under these traffic conditions, an extensive understanding and analysis of the traffic A. K. Budhkar (B) Department of Civil Engineering, Indian Institute of Engineering Science and Technology Shibpur, Howrah 711103, West Bengal, India e-mail: [email protected] G. Asaithambi · A. K. Maurya Department of Civil and Environmental Engineering, Indian Institute of Technology Tirupati, Setttipalli 517506, Andhra Pradesh, India e-mail: [email protected] A. K. Maurya e-mail: [email protected] G. Asaithambi Department of Civil Engineering, Indian Institute of Technology Guwahati, Assam 781039, India S. S. Arkatkar Department of Civil Engineering, National Institute of Technology Surat, Ichchanath, Surat 395007, Gujarat, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_6

101

102

A. K. Budhkar et al.

stream behavior is essential. To this end, the researchers need to define exclusive parameters, which describe the uniqueness of such traffic. This also warrants the use of the state-of-the-art equipment and suitable techniques for more insightful and reliable traffic data collection. The extraction of these unique parameters from such traffic streams must be thoroughly attempted with least magnitude of errors. This in turn potentially offer a better understanding of the intricacies involved in the disorderly traffic movement as a part of traffic flow modeling process. In view of this, the chapter discusses various methods of data collection and extraction techniques, development of traffic datasets followed by applications, and future directions on data collection practices that may be useful for the researchers to comprehend challenges of traffic in developing economies.

6.1.1 Characteristics of Mixed Traffic Heterogeneous nature of vehicles: A large number of vehicle types varying significantly in their sizes maneuverability are generally observed on roads in developing countries. It is not convenient to maintain lanes and drivers adopt to move in available empty spaces as per the vehicle characteristics. Various vehicle classes such as motorized two-wheelers (M2W), motorized three-wheelers (M3W), Light commercial vehicles (LCV) consist of a significant percentage of the traffic stream to cause disorderliness and deviation from conventional streams. Due to variability in their sizes and maneuvering capabilities, the microscopic properties need to be assessed as per individual vehicle class. Weak or no lane discipline: Vehicles interact in two dimensions—lateral and longitudinal, with their neighboring vehicles, due to which they do not follow demarcated paths (or lanes). A driver’s reaction is not just braking or acceleration but also simultaneous veering. Thus, the drivers may occupy any available space across the road width, and need to constantly perceive the scenario in front of the leading vehicles obstructing their view. Thus, staggered car-following is common. Vehicles may overtake based on available lateral spaces between leaders, resulting in sidewise interactions. Apart from these two primary characteristics, it is also observed that drivers maintain certain lateral clearance between vehicles for their safety while veering, which depends on speed and vehicle type. Smaller size vehicles are observed to move parallel to each other, and create situations when there are multiple leader/follower vehicles for a single follower/leader vehicle. Moreover, the overtaking decision is a complex set of dependent variables consisting of available spaces, speeds, and staggering. Due to this behavior, various phenomena are observed such as staggered following, attempting to overtake through the diagonal spaces available between vehicles, moving abreast, and traveling as per the dynamic lanes of different widths formed by the vehicles. This is described in Munigety et al. (2014). Thus, unique parameters may be essential for analysis of these traffic streams.

6 Emerging Traffic Data Collection Practices …

103

6.1.2 Unique Parameters of Mixed Traffic The simultaneous lateral and longitudinal movement, and the unique disorderliness compels researchers to identify essential parameters describing scenarios/behavior of mixed traffic. Some unique parameters identified by researchers include maintaining lateral gaps, staggering, decision to overtake/follow, overtaking, etc. (i)

(ii)

(iii)

(iv)

Lateral gaps—Since drivers interact laterally while overtaking, they maintain a certain safe lateral clearance between them, termed as ‘lateral gap’, and studied by various researchers (Budhkar and Maurya 2017a; Gunay 2007). Each driver maybe comfortable maintaining a certain gap between his/her vehicle and other interacting vehicles, which may depend on vehicle speeds, vehicle classes, and driving experience. This parameter is a combined effect of comfort levels of both the drivers contributing this gap, and describes the lateral distance-keeping characteristic of drivers. It is important to study this unique parameter, since it largely affects vehicle occupancy at various stream speed levels and also may contribute to define a appropriate surrogate safety term while assessing microscopic safe movements of vehicles in such stream. Longitudinal staggering—Since simultaneous veering and acceleration is observed in mixed traffic, drivers are often observed to follow other vehicles in a staggered manner, maintaining lateral separation between the centers of leaders and followers. This parameter termed as centerline separation is a measure of the amount of staggering a driver may maintain while following. Apart from other parameters affecting the safe car-following headways (such as speeds, vehicle types, driver variability, etc.), it may largely affect car-following behavior (Das and Maurya 2018) Clearance available for overtaking—Since drivers tend to overtake by judging the available lateral space between two vehicles (or between one vehicle and road edge/median), the clearance available for overtaking also constitutes an important parameter for extraction from mixed traffic. The decisions to overtake or follow—In homogeneous traffic conditions, overtaking comprises lane-changing decisions. However, in mixed traffic, since the concept of lanes ceases to exist, overtaking decisions need to be separately analyzed for simulating a driver’s decision to travel laterally or continue following the leader in a staggered manner.

These parameters become essential component in modeling the vehicular movement in mixed traffic. Therefore, for better understanding of these novel parameters an exhaustive and accurate dataset is required which can be created by using modern and accurate data collection techniques. Due to complex interaction existing in mixed traffic, data collection is a challenging task. The next subsection highlights the traditional data collection techniques adopted to extract the mixed traffic parameters while other advanced data collection techniques and extraction methodologies are dealt in the next sections.

104

A. K. Budhkar et al.

6.1.3 The Data Collection Efforts in Mixed Traffic Conditions The research world has contributed largely for data collection in mixed traffic. Several parameters including the space mean speed, throughput, time headways can be extracted as per their basic definitions. A recorded traffic video has assisted extractors to calculate these parameters with better accuracy. Radar guns are effective for extracting spot speeds for low volume straight roads. Vehicle heterogeneity while measuring throughput is conveniently addressed using Passenger Car Units (PCU)—a term used to replace one unit of a particular vehicle type with equivalent number of cars so as to produce the same traffic characteristics as earlier. The Indian Roads Congress (IRC 106:1990) mentions the PCUs for various vehicle types on mid-block urban arterials. Several such studies have been conducted on intersections (Budhkar et al. 2012; Saha et al. 2009), gradients (Arkatkar 2011), or roads with mid-block streams (Tiwari et al. 2007; Mallikarjuna and Rao 2011; and a number of other researches). The parameter, which can estimate crowding of vehicles, traffic density, cannot be directly estimated from the field as the ideal conditions like similar vehicles types, speeds, and maintaining lanes are not generally observed in mixed traffic. So as an alternative measure, Mallikarjuna and Rao (2006) proposed an area occupancy method, which considers vehicle areas, and the area of the trap section to represent density of the section. Measurement of the area occupancy is not accurate unless image-based methods estimate the vehicle sizes accurately. Edie’s method of measuring density from space–time diagrams or trajectories (Edie 1961) is proven to represent the density values in the field accurately (Bharadwaj et al. 2016). Therefore, for an accurate estimation of traffic parameters under heterogeneous traffic conditions, collection of microscopic parameters such as distance gaps and developing vehicle trajectory data from the real-world mixed traffic stream becomes essential. The inter-vehicular gaps in mixed traffic conditions consist of staggered headways, diagonal headways, and lateral gaps apart from the longitudinal headways. The longitudinal headways in mixed traffic are studied by various researchers (like Maurya et al. 2015; Das et al. 2017). Lateral gaps and the vehicle spread across road width in a mixed traffic were attempted by different researchers like Nagaraj et al. (1992), Dey et al. (2006a, 2006b), and Balaji et al. (2013). The process of physically marking strips of specific width on the roadway and estimation of lateral gaps, or vehicle’s lateral position on the road is quite cumbersome and laborious. Though these approaches lack in accuracy, they were novel attempts to determine vehicles spread across the roadway in mixed conditions. This approach was further improved by developing an accurate grid over the video images and manually noting the lateral position, as developed in Kotagi et al. (2020). They calculated lateral, longitudinal gaps, staggering, etc., at a fair level of accuracy. However, automation is much needed in such data collection/extraction practices to minimize the errors due to human limitations. Accuracy as well as quantum of datasets generated are always questionable in a manual approach. Therefore, efforts are underway by many researchers to automate the data collection/extraction practices.

6 Emerging Traffic Data Collection Practices …

105

6.1.4 Traffic Data Collection Techniques for Lane-Based Homogeneous Traffic Vehicular movement in homogeneous traffic conditions can be quantified primarily by means of two important phenomena—(i) Car-following and (ii) Lane-changing. Researchers have contributed significantly to a body of literature describing both these phenomena and driver behavior on a microscopic and a macroscopic scale. Various car-following (reviewed by Brackstone and McDonald 1999) and lanechanging models (described in detail by Ahmed 1999) are devised separately, which can be valid for vehicles maintaining strict lane discipline. Traffic flow count can be considered a lane-wise exercise, since vehicles follow strict lane discipline. Therefore, employment of automatic counters detecting lane-wise vehicle presence is a good indicator of flow. These counters include inductive loop detectors (notably, research in Gajda et al. 2001), piezoelectric sensors (reviewed by Bajwa et al. 2011), and magnetometers (used since Scarzello et al. 1978, the notable research by Bajwa et al. 2011). Other parameters such as speeds, headways, and the car-following parameters too can be extracted lane-wise. Therefore, the data collection of a multi-lane carriageway can be completed by merely conducting lane-wise analysis multiple times. Thus, it can be observed that majority of traffic analysis is fulfilled by lanewise data collection efforts, which largely does not exist for mixed traffic. However, few researchers (May 1990; Gunay 2003) have shown that vehicles interact laterally in lane-based traffic too, due to the discomfort traveling parallel to sidewise vehicles. This discomfort increases as vehicles drive closer sidewise, and parameters pertaining to quantifying lateral distance-keeping become prominent. Therefore, there is a need for robust data collection techniques, which can extract the parameters such as lateral clearance, centerline separation, the vehicle positions (trajectories), and describe driver behaviors during overtaking and following vehicles in mixed traffic conditions.

6.2 Advanced Data Collection Techniques This section describes some robust and accurate data collection techniques which can be used in quantifying driver behavior in mixed traffic conditions. The different traffic data collection practices used globally are presented in Fig. 6.1.

6.2.1 Video-Based Data Collection The technique of video-based data collection is substantially used by researchers to track and identify vehicles from a pre-recorded traffic video captured using a camera

106

A. K. Budhkar et al.

Fig. 6.1 Overview of traffic data collection techniques

fixed at a proper vantage point. Vehicles can be identified manually, tracked semimanually, or automatically using software and requisite microscopic and macroscopic parameters can be extracted. Parameters such as speed, flow, time headway between vehicles can be extracted manually from a video without the use of sophisticated data extraction techniques. However, the microscopic parameters describing mixed traffic need precarious extraction from a video due to their scale. Therefore, ideally an accurate video data extraction covering all aspects of mixed traffic over a larger study section is the need of the hour. Various video-based data collection techniques can boost the collection of microscopic and macroscopic traffic parameters accurately, such as camera calibration, video data collection of larger stretches using Unmanned Aerial Vehicles (UAVs), and stitching of video data described in this subsection.

6.2.1.1

The Theory of Camera Calibration

Camera calibration is a promising data extraction technique used for extracting vehicle locations (or trajectories) from a set of images, typically videos. Efforts have been undertaken in the field of optics since the last six decades (notably from Bas and Crisman 1997; Chou and Tsai 1986; Wang and Tsai 1991) for calculating coordinates of vehicles across and along the length of a road. The pioneering research for traffic data collection using camera calibration is conducted by Fung et al. (2003), wherein the camera’s orientation is required to calculate field coordinates (X and Y, along and across the road length respectively) from the corresponding image coordinates (x and y on the screen), and the camera orientation. Camera orientation is

6 Emerging Traffic Data Collection Practices …

107

in the form of tilt (t), swing (s), and pan (p) angles, whereas camera focal length (f) and distance of camera from road surface (l) is also necessary as scaling parameters. These five unknown camera parameters (s, t, p, f, and l) can be calculated (calibrated) if image coordinates of four points forming a perfect rectangle in the field with a given width are accurately known. Camera orientation may depend on the type of data required. The accuracy of data extraction by camera calibration depends on precision of marking a vehicle edge for tracking its trajectory, explained in detail in Budhkar and Maurya (2017b). Therefore, accuracy increases with higher zoom, however, higher zoomed sections may largely restrict trap length of the section.

6.2.1.2

Applications and Limitations of Camera Calibration Techniques

The camera calibration technique can be suitably adopted for a straight road with no vertical curves, and can be used for marking trajectories of points touching the road surface, such as vehicle edges either manually or with the help of automation. The semi-manual method adopted in Budhkar and Maurya (2016) employs mouse clicking the vehicle edges at certain frames, so as to obtain image coordinates which can be later converted to field coordinates. On the other hand, automated extraction methods are developed by researchers (since Jung and Ho 1999), which employ edge detection and other machine learning techniques to identify and track a vehicle. The limitations of this method are explained in Budhkar et al. (2018). The criteria include all straight roads without vertical gradient, (ii) marking of vehicle edges touching the road surface (and not the overhangs), and (iii) consistency of camera orientation throughout video recording. Any road surfaces falling within this criterion have been suitably adopted for trajectory data collection and extraction for mixed traffic conditions as well. This method however is not tested on horizontal curves, in which the trajectory of both—the road edges (curved) and vehicle trajectories need to be noted to evaluate vehicle performances. An extension of camera calibration technique has been used for dynamic data collection of vehicle-vehicle interactions. In this setup (Choudhari et al. 2020), the (imaginary) calibration rectangle travels with the probe vehicle, and its image coordinates remain consistent irrespective of the probe vehicle’s movement across the road. Figure 6.2 presents the setup used by Choudhari et al. (2020) for the measurement of longitudinal gap as well as lateral position from the road edge. This setup consisted of simultaneous front (Camera-1) and rear (Camera-3) view calibration before commencing the data collection. One of the corner points of the calibration was considered as the reference point for measuring distances on the real-world scale, from which the lateral and longitudinal distances were measured. The distances from road edge and centreline could also be calculated, since distance of road edge and probe vehicle’s edges from the calibrated rectangle (at the beginning of data collection) is known.

108

A. K. Budhkar et al.

Fig. 6.2 Camera calibration setup in a dynamic mode (Choudhari et al. 2020)

6.2.1.3

Data Collection Using Unmanned Aerial Vehicles (UAVS)

The data collected from video cameras placed at vantage points have the problem of occlusion, and is only measured at a particular point (Puri 2005). Data from advanced ITS technologies (vehicle-to-infrastructure (V2I), probe vehicles with GPS, and other smartphone sensor technologies) are not always easily converted to useful traffic data (Vlahogianni 2015). Moreover, the use of GPS technology is not so correct for analyzing the driver behavior since the drivers are aware that they are being monitored (Salvo et al. 2017; Barmpounakis et al. 2016). Recently, unmanned aerial systems commonly also known as drones are being used in traffic monitoring, management, and control. Due to technical advancement (gimbal stabilizer for cameras with UAVs and HD videos) and easy availability of UAVs, capturing high quality and stabilized videos covering this distance from a bird eye view is technically achievable. Also, UAVs have many advantages compared to manned air vehicles including low management and operation costs. The UAVs are equipped with high-resolution cameras and can capture the traffic from top view from which the lateral and longitudinal position of the vehicles can be extracted. Due to the high-resolution images obtained, the trajectory data extracted have a higher accuracy compared to the other methods. There were few attempts made to create a trajectory data set using the data collected from UAVs at highways and intersections. A study conducted by Barmpounakis and Geroliminis (2020) reported the collection of 500,000 trajectories using 10 drones covering an urban road network of 1.3 square km in Athens, Greece,

6 Emerging Traffic Data Collection Practices …

109

during morning peak hours to investigate different transportation phenomena occurring on the urban roads. Krajewski et al. (2018) collected the vehicle trajectory data using videos from UAVs on German highways (at six different locations) known as HighD dataset with high stability during non-windy periods. Similar data collection was conducted by Khan et al. (2017) and Robicquet et al. (2016) for the study of intersections and pedestrians respectively. When the traffic data is needed for a longer road section, say 1 km, a single UAV will not be able to capture it due to the limitation of minimum altitude at which the drone has to be flown in order to achieve the stability of the drone. When the drone is flying above 100 m altitude, due to winds, stability could not be achieved. In this case, more than one drone will be needed to cover the larger road stretch with 30–40% image overlap between the drones (to facilitate stitching the vehicle trajectories) as shown in Fig. 6.3. The following points are important while planning to collect traffic data using UAVs: • Scheduling the flight of drones considering the weather conditions, technical instrumental problems, physical obstacles, regulations for flying the drone, permission from the respective authorities, etc. • Stabilization of camera and video in order to capture the necessary data accurately with maximum visibility. • Geo-registration: Ground Control Points (GCPs) should be obtained in each video covering region to know the ground coordinates. • Image overlap between the drones (when more than one drone is used) for synchronizing the videos recorded to identify the vehicles moving from an area covered by one drone to the other.

Fig. 6.3 Alignment of drones covering the road section with image overlap

110

A. K. Budhkar et al.

• Identifying suitable locations for the safe take-off and landing of the drones. • Setting the time of drones to a predefined time in order to obtain the synchronization in the recorded videos. • If the drones couldn’t be kept flying for more than the survey period, it is necessary to keep a very small time interval in between the flights for the replacement of batteries. 6.2.1.4

Video Data Extraction from Sequential Videos

The UAVs have limited times of flight, thereby continuous video recording is not possible. A high-quality video data over a longer stretch of road is required in order to understand fully the driving behavior and particular maneuvers. Keeping this in view, the United States Federal Highway Administration (FHWA) developed a trajectory dataset using image processing, for 400–600 m trap lengths with varying locations and roadway facilities, as a part of the Next-Generation SIMulation (NGSIM) project (FHWA 2007). This dataset primarily helps understand driver behavior in lane-based traffic. A similar attempt was made by Raju et al. (2021) in mixed traffic conditions, where the authors have combined semi-manually extracted trajectory data from consecutive and connected roadway sections using videos from multiple camera sources. A matching algorithm assists in connecting timestamp and vehicle ID of videos with consecutive traffic sections. Thus, an extended trajectory dataset can be developed by ‘stitching’ vehicle trajectories from four consecutive segments.

6.2.2 Instrumented Vehicles Instrumented vehicles can be defined as those vehicles, which enable the realtime recording driver’s physiological characteristics, driver reactions (in the form of vehicle responses and positions, distance-keeping), and traffic events (reactions of neighboring vehicles, their characteristics), similar to an observer in that vehicle (Helander and Hogwell 1976). The vehicle can thus be equipped with one or more ‘On-Board Units’ (OBU) which can simultaneously record various vehicle parameters dynamically, as visible to an observer. The essential functionality of an instrumented vehicle is synchronization of data obtained from various OBUs, and correlate accuracy of data from these instruments. On-board units include the following-

6.2.2.1

GPS Device

A Global Positioning System (GPS) tracker records the global coordinates of an observer. An accurate GPS tracker logging at a high frequency, fitted in a moving vehicle can obtain the vehicle’s trajectory as it maneuvers through traffic. This system

6 Emerging Traffic Data Collection Practices …

111

can provide the dynamic parameters of vehicles, such as operating speeds, longitudinal acceleration and deceleration, lateral acceleration and deceleration by means of position data. Mahapatra and Maurya (2018) conducted a study to evaluate the parameters like lateral accelerations, longitudinal speeds for different vehicle types using a GPS equipped with video cameras. Similar GPS trackers are also used to measure travel time (Amita et al. 2016; Chepuri et al. 2019; Sekhar et al. 2013), to calculate acceleration and deceleration characteristics of various vehicle types in mixed traffic stream (Bokare and Maurya 2017), and other vehicle-related parameters (Mahapatra and Maurya 2018). Speed profiles of multiple vehicles can be assessed simultaneously for calculation of parameters describing corridor traffic, such as level of traffic congestion, speed variations, as determined by Ko et al. (2006).

6.2.2.2

Sensors to Collect Traffic Data

Electromagnetic sensors are widely used in the traffic safety world. They can transmit waves, which when reflected against an obstruction, can be detected, thereby indicating the presence and distance of an obstruction from the instrumented vehicle. This concept is popularly used in automatic collision avoidance systems in vehicles, but it can be modified to obtain the distances between a vehicle and its neighboring entities, such as other vehicles or road edges. One of the tried and tested methods is distance recording using ultrasonic sensors. Wong and Qidwai (2004) have devised collision avoidance of a car by means of attaching a system consisting of ultrasonic sensors and actuators. All the processes were processed using a vehicle-electronic control unit. A similar approach was also used by Venter and Knoetze (2013) for measuring lateral distances between bikes and other vehicles in order to predict safe width of bike lanes. This approach was also used by Budhkar and Maurya (2017) to estimate lateral distance-keeping behavior of various vehicle types in mixed traffic streams. The authors have used ultrasonic sensors to the range of 4 m (the assumed extent of lateral interaction). These sensors can trigger ultrasonic pulses and calculate distance between themselves and the required object based on the time of flight of transmitted and the received echo pulse. Six sensors are placed, three on each side of the instrumented vehicle. If a vehicle overtakes/gets overtaken by the test vehicle, lateral distances are recorded at a constant frequency of 10 Hz. Multiple sensors on one side of the vehicle, and multiple detections at different time stamps enable the observer to calculate relative speed of the interacting vehicle (as illustrated in Fig. 6.4). The points A-B-C, L-M-N, and P-Q-R of the interacting vehicle in Fig. 6.4a are reflected in the readings in Fig. 6.4b. Therefore, the assembly of ultrasonic sensors is used to determine the (i) presence, (ii) distance, and (iii) relative speed of the interacting vehicle. Similar to ultrasonic sensors, video image or vision-based sensors have been used in the traffic detection world, reviewed by Sun et al. (2004). Moving vehicle detection and classification systems are popular, with video image or vision-based sensors being used on a large scale. Therefore, dynamic detection and calculation is possible with the help of these sensors.

112

A. K. Budhkar et al.

Fig. 6.4 Ultrasonic sensor assembly to measure lateral vehicle interactions (Budhkar and Maurya 2017)

6.2.2.3

Data Collection Using Vehicles Equipped with LIDAR Sensors

One of the recent advances in traffic data collection methods is the usage of instrumented vehicles equipped with sensor devices, cameras, microphones, etc. Light Detection and Ranging (LIDAR) is one of the sensor devices which works on the principle of sending laser beams to the target object (e.g., vehicle) and measures distance by the returned beams. The data obtained from LIDAR sensors are mainly useful for detection and tracking of surrounding vehicles to obtain the trajectory of the vehicle, gap with the surrounding vehicles, etc. Few researchers made an attempt to collect data using LIDAR sensors mounted on instrumented or probe vehicles. Dutta and Vasudevan (2020) and Soni et al. (2020) made an attempt to understand driving behavior during overtaking on Indian highways in disorderly traffic using naturalistic driving data from an instrumented vehicle equipped with different sensors such as VLP-16 LiDAR, four video cameras, IMU-GPS unit, on board diagnostic (OBD) scanner, and steering angle sensor. Dutta and Vasudevan (2021) also studied the effects of road geometry, roadside infrastructure, and static objects near the road on driver behavior using an instrumented vehicle. Zhao et al. (2016) used the instrumented vehicle with four 2D LIDARS mounted on it in order to obtain the trajectories of surrounding vehicles on the ring roads of Beijing. Vasconcelos et al. (2014) used the trajectory data collected from instrumented vehicle with LIDAR sensors for the purpose of calibration of GIPPS’ car-following model. Two instrumented vehicles (one as a leader and other one as a follower) with data logger and GPS (gives positions of vehicles) were used for the purpose of data collection. The LIDAR rangefinder was connected to the data logger present in the

6 Emerging Traffic Data Collection Practices …

113

follower vehicle. The distance to the leader with respect to the follower was obtained using this sensor, from which space headways were extracted in combination with the LIDAR data. The advantage of obtaining data from LIDAR sensors is that they can be used to extract both the macroscopic and microscopic traffic flow parameters (Cetin et al. 2017). The trajectories obtained from LIDAR data can be extracted for long time periods. It also offers omni-detection covering all the interacting vehicles in the surroundings along with consistency and continuity along time and space during data collection. Use of multiple LIDARS can maximize the coverage of sensing (Zhao et al. 2016) but the limitation of this method is that the mounting of the sensor has to be done properly to avoid more occlusion with the body of the vehicle. The range of detection is also limited in the range of 20 to 100 m. The sensor being expensive cannot be used for many of the applications, which is one of the major limitations (Zhao et al. 2016). The primary limitation in using the instrumented vehicles is that the obtained parameters describe the driving characteristics of a particular driver (who drives the instrumented vehicles) and the vehicle at the given conditions. Therefore, diverse driver population and vehicles need to be instrumented to analyze the behavioral aspects at a particular corridor. Further, vehicle-to-vehicle interactions recorded by two instrumented vehicles by deliberate interaction to measure the gap-maintaining behavior and safety parameters deviate from naturalistic driving promise.

6.2.3 Data Collection Using Road Side Units (RSUs) A roadside unit (RSU) collects traffic data either from a static sensing area or instrumented vehicles for transmitting or storage of requisite traffic-related information. The passage of a vehicle can trigger detection information in the unit. Several such units are attached at various intervals along a thoroughfare can help identifying parameters such as vehicle flow, speeds, and travel times. They can be classified as intrusive or non-intrusive units, based on the temporary damage caused to the pavement during their installation or functioning. The intrusive techniques mainly magnetometers, pneumatic tubes, or inductive detection loops are essentially lane-based on wider roads. Therefore, the passage of multiple vehicles in the same lane may prove inefficiency in their application for mixed traffic conditions. The non-intrusive techniques are based on transmission or sensing of electromagnetic waves from the unit. They include Bluetooth, Wireless Fidelity (Wi-Fi), License plate recognition systems, and Radio Frequency identification (RFID) scanners. Table 6.1 summarizes a comparative report of different roadside units for traffic data collection based on their accuracy, cost, and privacy concerns to the drivers. It can be noted that bluetooth and Wi-Fi-based media scanners can yield data with higher precision data in addition to being cost-effective. In recent years, Wi-Fi/Bluetooth-based travel time estimation system has received wide attention and researches are being undertaken in this field due to its non-invasiveness, cost-effectiveness, and ease of installation. Hence, this

114

A. K. Budhkar et al.

Table 6.1 Comparative report on different RSU for data collection (Abbott-Jard et al. 2013) Technologies

Accuracy

Cost

Privacy

Bluetooth media scanners

High

Low

Moderate

Wi-Fi media scanners

High

Low

Moderate

Loop detectors

Low

High

Low

Magnetic sensors

Low

Moderate

Low

License plate recognition systems

High

High

Moderate

RFID

High

Low

Moderate

technology was found suitable to be adopted. The application of this technology is simple. The detailed explanation of Bluetooth/Wi-Fi technology is given in the chapter, named, Automated Sensors for Indian Traffic: Challenges and Solutions.

6.2.4 Driving Simulators A driving simulator is an open cockpit of a vehicle with simulated screens and instruments that record and test driver’s movements in a closed environment. They are used for studying driver behavior under controlled conditions. During the last few decades, there have been parallel research streams to describe the driving behavior(i) using field data, which is more realistic but cannot explicitly describe the effects of certain factors on driving; and (ii) using driving simulators, which may lack realism but can quantify driving behavior as a causation of several other factors that cannot be experimented in the field. Thus, driving simulator datasets can assist researchers to generate specific scenario and study the driver behavior. Two types of vehicles are generated by driving simulation software—(i) Ambient vehicles which do not react to any scenarios or movement of the test vehicle, and (ii) Scenario-based vehicles which can be programmed to follow a particular path to generate a specific scenario. Ambient vehicles in driving simulators are governed by specific car-following rules, which by far the knowledge of the authors are lane-based. Effective study of vehicle interactions in simulators requires generation of scenario-based vehicles triggered at different cases based on the test vehicle’s movement. A large number of scenarios may be generated to test the effect of a particular driving maneuver on the drivers. This has been tried and tested on individual vehicles (Choudhary and Velaga 2017; Mahajan et al. 2019; Verma et al. 2016; Choudhari and Maji 2019, etc.). However, to study driver maneuvers in mixed traffic requires generation of a large quantum of such scenario-based vehicles, which is a cumbersome effort. A study of driver behavior in mixed conditions using driving simulators is promising prospective research, but a compatibility check is required in a similar field section to ensure transferability of developed simulated scenes, such as that by Papadimitriou and Choudhury (2017).

6 Emerging Traffic Data Collection Practices …

115

6.3 Development of Comprehensive Datasets 6.3.1 Development of Trajectory Datasets Obtainable measurements of the trajectory of a vehicle (by any one of the methods as discussed in previous section) consist of discrete observations of its position, equally spaced in time, that is the series of vehicle coordinates in the two- or threedimensional space (coordinates refer to a specific point of the vehicle, e.g., the front bumper centre). From trajectories, many traffic characteristics can be derived, e.g., speed and acceleration functions as well as gaps, and time to collision.

6.3.1.1

Semi-manual Techniques

Vehicle positions at various time stamps can be obtained using a semi-manual technique and camera calibration, as per the procedures adopted by various researchers including (Munigety et al. 2014; Das and Maurya 2018, etc.). This is possible by developing a customized video player wherein users can pause the video at a particular frame, click and record the image coordinates, and record the supplementary information such as vehicle types and dimensions. After necessary data processing, this information can be used to obtain individual trajectory datasets.

6.3.1.2

Extended Trajectory Database

A micro-level high-quality vehicle trajectory dataset obtained for a longer road stretch will be useful to understand the driving behavior including lane-changing behavior (Leclercq et al. 2007) and car-following behavior (Hao et al. 2016). In order to develop such a trajectory dataset for a longer road section, collecting data from a single camera or a UAV is not feasible due to many factors as discussed in the previous section. In that case, multiple cameras or UAVs need to be employed at different continuous road sections to cover a larger road stretch with 30–40% overlap between the videos. To create an extended trajectory database, a stitching algorithm needs to be applied to stitch the trajectories obtained from the videos of multiple road sections. Raju et al. (2021) developed such a trajectory database using vehicle trajectories recorded from four continuous road segments to cover an extended study section of 535 m in Surat city, India. The extracted vehicle trajectories had issues—for instance, vehicles had different ID numbers in the videos from two segments. A set of thresholds were specified for different sequential runs to synchronize the vehicle category, timestamp, time difference, longitudinal and lateral positions of individual vehicle trajectories from two consecutive segments. This was validated with GPS-based vehicle runs and effectively smoothened. Figure 6.5 showcases the data collection section and pre-stitched trajectory data plots obtained by Raju et al. (2021).

116

A. K. Budhkar et al.

Fig. 6.5 a Road segments surveyed by four cameras, and b Time–space diagram for vehicular movement over the study section (Raju et al. 2021)

6.3.1.3

Smoothening and Removal of Errors

The extracted real-world trajectory data may contain measurement errors and can be interpreted as random or correlated noise on the positional location of vehicles, errors that may be further amplified in the differentiation process when speeds and acceleration values are calculated. Once the position data are extracted, they need to be smoothed to overcome missing observations which are caused by occlusions, reduce measurement errors, and calculate other variables, such as speeds and accelerations, etc. Different smoothing techniques like moving average techniques (Ossen and Hoogendoorn 2008; Papailias and Thomakos 2015; Raju et al. 2017), locally weighted regression (Toledo et al. 2007; Kanagaraj et al. 2015a), and filtering (Montanino and Punzo 2013) will be applied to remove noise from the data so as to reconstruct the trajectory, subject to constraints on physical vehicle kinematics (i.e., acceleration) and consistent traffic flow dynamics (i.e., internal and platoon consistency). A better smoothening is observed with a higher degree of moving average, however, there is a chance that the fluctuations observed in the field conditions will be obliterated (Raju et al. 2017). In the local regression method, the accuracy of the result depends on the window size. Duell et al. (2014) mentioned that the large window size required for higher order polynomials might not be suitable for widely displaced datasets. Punzo et al. (2011) analyzed the trajectory and accuracy of speed data using the jerk, consistency, and spectral analysis. Yuan (2009) applied empirical mode decomposition (EMD)-based trajectory smoothing algorithm on both x and y coordinates. The author concluded that for smoothing the raw trajectory data, EMD is a more convenient method compared to the wavelet-based techniques. Two problems in using EMD approach are (1) correction of the end points, (2) mode mixing (Wu and Huang 2009). The addition of white Gaussian noise solves the mode-mixing problem, but the reconstructed

6 Emerging Traffic Data Collection Practices …

117

signal includes residual noise and different realizations of signal plus noise may produce a different number of modes (Torres et al. 2011). To avoid such problems, Pal and Chunchu (2018) used an approach based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) to smooth the trajectory data. Venthuruthiyil and Chunchu (2020) highlighted the need for separate smoothing and proposed a trajectory reconstruction technique to address the following issues: (1) noise in the vehicle path corresponding to different vehicles and directions shows unique characteristics, (2) the existing practices of vehicle trajectory reconstruction have used invariant smoothing parameters across different vehicle trajectories, and the lateral component of the path. There are not many studies available in evaluating the effectiveness of smoothing techniques, and there is a need to come up with a new approach to handle the vehicle trajectory smoothing operations. The obtained trajectory datasets may be employed to study individual driver behavior as well as interaction of a driver with the environment, which includes vehicle-to-vehicle and vehicle-to-infrastructure interactions. The datasets for individual and interactive driver behavior are discussed in subsequent paragraphs.

6.3.1.4

Automation in Trajectory Data Extraction

Many deep learning techniques are being used for extracting trajectory data from video images. The steps involved are (1) Vehicle detection (2) vehicle classification (3) vehicle tracking, and (4) conversion of image coordinates to ground coordinates. Most of the studies used YOLOv algorithm (Sudha and Priyadarshini 2020) and RCNN algorithm (Seong et al. 2019; Clausse et al. 2019) for vehicle detection from images recorded by videos. Chen et al. (2019) used a Canny-based ensemble detector for vehicle detection based on the selection of ROI (Region of Interest). Abdeljaber et al. (2020) used MATLAB computer vision system toolbox, the “trainRCNNObjectDetector” for vehicle detection. Kalman filtering technique was used for vehicle tracking in most of the studies (Sudha and Priyadarshini 2020; Seong et al. 2019). Chen et al. (2019) used the Kernelized Correlation Filter (KCF) algorithm for tracking the detected vehicle positions accurately with a lesser computational time. Clausse et al. (2019) were tracked the detected vehicles using Intersection Over Union (IOU) tracker. The detailed procedure about data extraction using artificial intelligence is explained in the chapter “Automated Sensors for Indian Traffic: Challenges and Solutions”. For a better and quick mitigation of traffic incidences, traffic congestion and to enable faster operations and safety, it would be encouraging if researchers may develop algorithms for real-time traffic data extraction from live video streaming.

118

A. K. Budhkar et al.

6.3.2 Naturalistic Driving Behavior It is important to describe a naturalistic driving behavior by designating parameters which inform the driver’s reaction, measure the comfort levels and may specify surrogate safety. A Global Positioning System (GPS)—enabled performance box can be used to establish high accuracy trajectory data. This can be further used to establish parameters such as speed, acceleration, and jerk in lateral or longitudinal direction. However, parameters pertaining to a vehicle’s rotation (roll, pitch, and yaw) cannot be extracted by means of trajectory data. Vehicle’s yaw remains an important parameter since it largely affects passenger comfort and vection. A larger yaw rate may cause discomfort, and rapid yaw rate changes may cause motion sickness, as studied under simulated conditions in Nooij et al. (2017). These vehicle rotationbased parameters can be extracted by means of an inertial measurement unit (IMU) sensor, similar to Mahapatra and Maurya (2013). The authors could calculate rate of change of vehicle heading angle and compare it with longitudinal speeds and lateral accelerations Here, the authors had synched IMU sensors with GPS-based vehicle position data obtained from GPS-based data loggers. Stand-alone IMU sensors can be also used for data collection, however various errors (mostly accumulative in nature) need to be eliminated as addressed in Mousa et al. (2017).

6.3.3 Interaction of a Vehicle With Its Environment Apart from individual vehicle characteristics, a vehicle can interact with its surroundings such as other vehicles or road edges/medians. Vehicle-to-vehicle interactions can be captured either with the help of a probe vehicle or by static means, i.e., by videographic means or roadside units. For the analysis of car-following in mixed conditions, two interacting vehicles (say, a LV and FV for following behavior) need to be identified by the data extractor by observing their proximity and movements, as in Budhkar and Maurya 2017b. In this research, vehicle types, their sizes, speeds, and lateral/longitudinal distances (Centerline separation and longitudinal gap) were extracted from a calibrated traffic video. The vehicle probes have the issue that driver behavior of the probe vehicle occupies substantial data, therefore driver (and perhaps vehicle) variability is required for a larger dataset. Further, drivers may not drive naturally if they notice that they are being monitored. The parameters for vehicle-to-vehicle interactions may involve distances, speeds, and accelerations. Using a probe vehicle, inter-vehicular distances can be calculated as explained in Budhkar and Maurya (2017). This probe vehicle was instrumented with multiple instruments—sensors and GPS devices. The GPS device was able to calculate vehicle speed at particular time steps. Multiple sensors made multiple detections at different time stamps, thereby, allowing the authors to calculate lateral interaction times (Budhkar and Maurya 2017d) and thus relative

6 Emerging Traffic Data Collection Practices …

119

speeds. However in this setup too, full automation regarding vehicle classification was not possible, which had to be conducted manually. The vision-based techniques have an advantage of identifying vehicles through automation, though their accuracy needs consideration. Researchers (Budhkar et al. 2018; Choudhari et al. 2020) have employed camera calibration to traffic videos obtained from a camera fixed to a probe vehicle’s dashboard. Image coordinates of interacting vehicle’s corner points with supplementary data were extracted manually, to observe car-following or overtaking (Choudhari et al. 2020). Obtained parameters included centerline separation, longitudinal gap, time required to complete overtaking, lateral position of vehicles, and speeds of all the interacting vehicles. Capturing realistic interactions is difficult for an external observer of the data (i.e., it may appear that vehicles are interacting, but in reality they may not). Authors have usually captured the driver’s reaction (acceleration, brake, or steering) in order to recognize the state of interactions—following, overtaking, etc. For example, Budhkar and Maurya (2017c) assumed sufficient veering as the decision to overtake in mixed traffic streams, and Choudhury and Islam (2016) used this approach to assume the latent leader of the following vehicle. These approaches have been used for manual data extraction, and maybe suitably adopted in the automated data extraction tools. Furthermore, a vehicle may dangerously travel towards the edge of a road while overtaking. Vehicle clearance from road edges is a sensitive parameter, and a slight change in lateral distance compromises the safety of vehicles traveling along the edges. Mahapatra and Maurya (2015) investigated the lane-specific gap maintenance of vehicles with the shoulder and median. The authors have fixed two synchronized, ultrasonic sensors on the road edge separated 10 m from each other, to simultaneously detect the presence, distance maintenance and speed of vehicles from the median or edge. The difference between the time stamps generated by vehicle passage provided vehicular speed, whereas the readings from sensors provided accurate distances to the accuracy of 1 cm. This setup can be further improved to cover larger distances using wireless techniques to remove the distance limitation, and can be used for travel time and delay studies. The sensors can also be bluetooth, radio, video, or laser-based.

6.3.4 Travel Time Data Traditional travel time data collection methods consist of direct measurement techniques and estimation methods. Floating car technique is one of the most popular direct measurement methods, which uses a probe vehicle traveling with the traffic flow to record travel time and location information. Since the 1990s, this technique has been combined with global positioning system (GPS) devices to provide more comprehensive information in the form of vehicle trajectory along with time stamps, thereby providing frequent sampling along the route. This can implicitly avoid human error associated with traditional travel time and location recording technique (Quiroga and Bullock 1998). However, as the floating car technique only

120

A. K. Budhkar et al.

provides travel time information for the probe vehicle, it becomes extremely difficult to collect a large data set for spatial–temporal analysis of travel time variability and reliability. To address the aforementioned limitation, direct measurement techniques were replaced by passive data collection technologies which can be broadly classified as (a) fixed sensors (such as loop detector, Bluetooth/Wi-Fi, Automatic Number Plate Recognition System (ANPR), and radio frequency identification (RFID)) that provide traffic information at the location where the sensors are installed and (b) mobile sensors (such as GPS equipped vehicles and Automatic Vehicle Location (AVL)), which provide data for the entire journey of the vehicle equipped with such sensors. The passive technologies can often be implemented more economically and faster than the traditional methods. Further, richness in the resultant data set is maintained as the observed behavior is captured using these passive data collection techniques as opposed to the stated behavior. The Bluetooth/Wi-Fi sensors work on the principle of recurring detections of MAC ids for the devices and their analogous time stamps for the duration they remain within the detection range. Thus, re-identification of the same moving vehicle along various network points will help to determine the travel time and the uniqueness of the ids makes it conceivable to track the id in upstream and downstream ends. The time difference between the two perceptions can be utilized to appraise travel time. Considering that the MAC address in the detention zone is detected multiple times, four components, mainly, first in first out (FIFO), last in last out (LILO), first in last out (FILO), and last in first out (LIFO) can be deduced for deriving travel time between two locations. Figure 6.6 illustrates the conceptual diagram for estimating travel time. Among the four components, i.e., FIFO, FILO, LILO, LIFO, the component which can be used to estimate travel time will depend on the objective of the study. However, before extending the application of travel time derived using BT/Wi-Fi sensors, it is important to study the reliability of the data. A few studies have analyzed the

Fig. 6.6 Deriving travel time

6 Emerging Traffic Data Collection Practices …

121

reliability of the Wi-Fi/BT sensor-based travel time data for application in transportation engineering. Erkan and Hastemoglu (2016) investigated the applicability of Bluetooth sensors for travel time estimation in heterogeneous traffic. The results obtained were promising and showed that Bluetooth is a cost-effective technology for estimating travel time for heterogeneous traffic conditions. Araghi et al. (2015) performed a controlled field experiment to collect travel time data using Bluetooth and a global positioning system (GPS). Authors found that, on average, Bluetoothenabled devices will be detected 80% of the time while passing a sensor location. Wang et al. (2011) and Mei et al. (2012) compared the travel time data from BMS data with that from video cameras for motorways, and arterial and promising results were reported. Haghani and Aliari (2012) reported that travel time obtained from the traditional matching of BMS data could be considered ground truth travel time. Moghaddam and Hellinga (2013) examined the magnitude of errors in detection time and travel time measurement. The results showed that the mean travel time error was essentially zero for all traffic conditions. Through multiple regression, the standard deviation of the travel time measurement error is modeled. It shows that under some conditions, the 95% confidence interval of this error may reach 25% of the true mean travel time.

6.3.5 Traffic State Information from Application Interfaces Application performance interfaces (API) can provide real-time information about the traffic state such as estimated travel time, density levels, and transit information. The commonly used APIs for this purpose include Google Maps, TomTom (Abadi et al. 2016), etc. Further, the General Transit Feed Specification (GTFS) has emerged as a worldwide popular open-source industry data standard to describe and publish fixed-route transit operations. The researchers may explore sourcing information including travel time, density levels, corridor reliability, etc., from these sources. Table 6.2 summarizes the applicability (or suitability) of various data collection techniques studied in this chapter (mentioned as columns), for the generation of different traffic datasets (mentioned as rows).

6.4 Application and Future Research Directions 6.4.1 Applications in Traffic Operations and Safety Rapid, comprehensive, and accurate traffic data from multiple sources is important for planning, monitoring, and operations of traffic facilities. Capturing the data from fields is now confidently conducted using semi-automated and automated techniques, thereby enabling an increase in quantum and accuracy of datasets for modeling.

122

A. K. Budhkar et al.

Table 6.2 Applicability of data collection techniques Technique → Video-based Dataset ↓

Instrumented Vehicles

Road-side units

Web-based Driving datasets simulators

Travel time data

Synchronized video Comprehensive Modern and datasets at two locations time-based accurate maybe explored information means of can be obtained obtaining for individual corridor-level vehicles travel time data

Robust Not datasets applicable can be obtained, validation is required

Individual vehicle studies

Cannot cover entire maneuvers

Most suitable for obtaining time-based parameters, vehicle rotations

Only vehicle Not speeds can be applicable obtained

May be applied, validation is required

Vehicle interactions

Comprehensive V2V information can be obtained, multiple sites need to be chosen for V2I information.*

Sensors can assist evaluating gap maintenance with vehicles and road edges

Not suitable

Not applicable

Controlled scenarios can be generated to evaluate interactions

Trajectory datasets

Large-scale trajectories can be developed using semi-manual/automated techniques with smoothing

Suitable for obtaining individual vehicle trajectories

Not suitable

Not applicable

Limited application

Driver behavior

Studies can be conducted only based on extracted vehicle trajectories

Accurate driver behavior information based on driver psychology

Limited, clearance from road edges

Not applicable

Driver perception and responses can be evaluated

There are manifold applications of generated datasets, including safety, behavioral aspects, and operational aspects of planning. The inter-vehicular gaps and driver’s reactions can provide information about surrogate safety and investigate the causes of road crashes. Requisite information about traffic parameters including desired speed levels, speed distribution, lateral clearances, and headways can simulate mixed traffic accurately. The data collection and extraction methods can also be extended for analysis of pedestrian movements across dedicated facilities. There is a demand for real-time decisions to improve traffic safety and operations issues, therefore the automated data collection efforts can copiously assist the planners and engineers.

6 Emerging Traffic Data Collection Practices …

123

6.4.2 Beyond Mid-Block Data Collection A thorough investigation of traffic characteristics need not be restricted only to mid-block sections, where generally no external factors disturb the traffic stream. Conflicting traffic streams in mixed traffic cause unique disruptions, which may require novel data collection approaches for investigations. Limited studies on conflicting streams in mixed traffic such as merging areas (Kumar et al. 2017; Budhkar et al. 2020) or intersections (Kanagaraj et al. 2015b) have considered broadly the macroscopic aspects, and did not evaluate gap-maintaining, following or overtaking behaviors in these conflicting streams. A large gap exists in microscopic analysis of conflicting streams in these traffic conditions. Extended trajectory datasets (similar to Raju et al. 2021) with vehicle-wise and stream-wise information can fulfill gaps related to microscopic analysis of conflicting mixed traffic streams. Furthermore, a gap exists in the study of mixed traffic movement in diverging roadways. It is also interesting to study the lane choice behavior of drivers near toll plaza (Parmar et al. 2013) where lane discipline is not observed in mixed traffic stream. A further study may be required for better analysis of upstream traffic movement at a toll plaza. Although trajectory datasets can provide information about individual vehicle movement, it is also important to study the driver capabilities such as assigning a specific leader to a following vehicle, or their reaction (braking, steering or acceleration) after perturbation from conflicting stream. Therefore, instrumented vehicles in the future may also need to collect sufficient driver-related information, such as brake intensity, rate of steering, etc. (Singh and Kathuria 2021). Under mixed traffic scenario, driver’s psychological conditions such as gaze and eye tracking (Khan and Lee 2019; Xu et al. 2018), galvanic responses (Healey et al. 1999), Electrocardiograms, etc., have not been explored till date. Further, correlation of driver’s responses with their psychological conditions assists researchers to comprehend the behavioral aspects of driving in more detail. Even in case of mid-block streams with geometric features such as horizontal or vertical curves, there is a large scope for performance evaluation of vehicles with changing geometric features, or vehicle-to-vehicle interactions. Prominent research work has primarily evaluated the performance of curves (i.e., conformance to driver’s expectancy), which may include speed distribution of vehicles (Sil et al. 2019) on the sites. However, missing gaps exist relating to the vehicle interactions on curves, or performance evaluation of vertical or compound curves. Thus, it can be observed that inspite of presence of a large number of novel data collection techniques to quantify and model the mixed traffic, there exist some severe gaps for data collection in few arenas of transportation world.

124

A. K. Budhkar et al.

6.4.3 Future Research Directions Researchers are contributing towards numerous data collection methods with noteworthy technological advances that also work for heterogeneous traffic conditions to better capture traffic flow characteristics. There is now sensible confidence in implementing automated or semi-automated systems capable of capturing data directly from actual field conditions. Due to the ingress of different Intelligent Transportation System (ITS)-based techniques, the ability to collect a vast amount of quality data from the real field conditions has become a reality. The researchers’ primary challenge is to handle and manage this archived data for extracting essential traffic flow characteristics. Potentially, there are numerous uses of monitoring traffic data in an expeditious and comprehensive manner and responding to the real traffic situations to improve efficiency, safety, and mobility. The approaches of traffic data collection, extraction, and analysis significantly vary under heterogeneous traffic conditions compared to homogenous traffic. In this regard, research attempts, addressing the above-mentioned issues related to collection, handling, and analysis of quality traffic data, exhibiting certain pragmatic applications are encouraged. Some of the areas which need future research are—Innovative traffic data collection Instruments and Techniques, Vehicle detection technologies under varying roadway and traffic conditions, Visualization tools for detecting traffic volume and vehicle classification data, Evaluating driver behavior at different roadway elements using naturalistic driving data, Use of driving simulators for developing driver profiles, Case studies that demonstrate the need for high-quality traffic data, Application of traffic data to support public agencies’ operation and management needs, Advancements in Crash risk estimation, Big data in intelligent mobility, and Human factors and implications in the design.

References Abadi M et al (2016) Tensorflow: a system for large-scale machine learning. In: 12th {USENIX} symposium on operating systems design and implementation (OSDI 16) Abbott-Jard M, Shah H, Bhaskar A (2013) Empirical evaluation of Bluetooth and Wifi scanning for road transport. In: Australasian Transport Research Forum (ATRF), 36th Abdeljaber O, Younis A, Alhajyaseen W (2020) Extraction of vehicle turning trajectories at signalized intersections using convolutional neural networks. Arab J Sci Eng 45(10):8011–8025 Ahmed KI (1999) Modeling drivers’ acceleration and lane changing behavior. Diss. Massachusetts Institute of Technology Amita J, Jain SS, Garg PK (2016) Prediction of bus travel time using ANN: a case study in Delhi. Transp Res Procedia 17:263–272 Araghi BN et al (2015) Reliability of bluetooth technology for travel time estimation. J Intell Transp Syst 19(3):240–255 Arkatkar SS (2011) Effect of intercity road geometry on capacity under heterogeneous traffic conditions using microscopic simulation technique. Int J Earth Sci Eng 4(6):375–380

6 Emerging Traffic Data Collection Practices …

125

Bajwa R et al (2011) In-pavement wireless sensor network for vehicle classification. In: Proceedings of the 10th ACM/IEEE international conference on information processing in sensor networks. IEEE Balaji K, Bharadwaj MRK, Dey PP (2013) A study on lateral placement and speed of vehicles on two-lane roads. Indian Highw 41(9):57–61 Barmpounakis EN, Vlahogianni EI, Golias JC (2016) Extracting kinematic characteristics from unmanned aerial vehicles, no 16–3429 Barmpounakis E, Geroliminis N (2020) On the new era of urban traffic monitoring with massive drone data: the PNEUMA large-scale field experiment. Transp. Res Part C Emerg Technol 111:50– 71 Bharadwaj N et al (2016) Traffic data analysis using image processing technique on Delhi–Gurgaon expressway. Curr Sci 808–822 Bokare PS, Maurya AK (2017) Acceleration-deceleration behaviour of various vehicle types. Transp Res Procedia 25:4733–4749 Brackstone M, McDonald M (1999) Car-following: a historical review. Transp Res Part F: Traffic Psychol Behav 2(4):181–196. https://doi.org/10.1016/S1369-8478(00)00005-X Budhkar AK, Maurya AK, Maji A (2018) Dynamic data collection of following and merging behavior in mixed traffic. In: ISTS & IWTDCS 2018 Matsuyama, Japan Budhkar AK, Maurya AK (2016) Modeling gap-maintenance in heterogeneous no lane-discipline traffic. DEStech Trans Eng Technol Res ICTIM Budhkar AK, Maurya AK (2017a) Characteristics of lateral vehicular interactions in heterogeneous traffic with weak lane discipline. J Modern Transp 25(2):74–89 Budhkar AK, Maurya AK (2017b) Multiple-leader vehicle-following behavior in heterogeneous, weak lane discipline traffic. J Transp Dev Econ (TiDE) 3:1–17 Budhkar AK, Maurya AK (2017c) Analysis of lateral interaction time in mixed traffic conditions. In: 4th conference of transportation research group of India (4th CTRG) Budhkar AK, Maurya AK (2017d) Overtaking decision modeling in heterogeneous, weak lane discipline traffic. J Eastern Asia Soc Transp Stud 12:1740–1754 Budhkar AK, Maji A, Gandge S (2020) Speed reduction on merging sections in mixed traffic: a case study. Transp Res Procedia 48(2):850–859 Budhkar AK et al (2012) Passenger car equivalence (PCE) of vehicles at intersections on narrow roads in hilly urban areas. In: Transportation Planning and Methodologies in Developing Countries (TPMDC 2012) Cetin M, Sazara C, Nezafat RV (2017) Exploring the use of LIDAR data from autonomous cars for estimating traffic flow parameters and vehicle trajectories Chen X et al (2019) Extracting and denoising vehicle trajectory automatically from aerial roadway surveillance videos. In: 98th annual meeting of transportation research board, no 19–03147. Chepuri A et al (2019) Travel time reliability analysis on selected bus route of mysore using GPS data. Transp Dev Econ 5(2):1–15 Chou H-L, Tsai W-H (1986) A new approach to robot location by house corners. Pattern Recogn 19(6):439–451 Choudhari T, Maji A (2019) Effect of horizontal curve geometry on the maximum speed reduction: a driving simulator-based study. Transp Dev Econ 5(2):1–8 Choudhari T, Budhkar A, Maji A (2020) Modeling overtaking distance and time along two-lane undivided rural highways in mixed traffic condition. Transp Lett 1–9 Choudhary P, Velaga NR (2017) Mobile phone use during driving: effects on speed and effectiveness of driver compensatory behaviour. Accid Anal Prev 106:370–378 Choudhury CF, Islam MM (2016) Modelling acceleration decisions in traffic streams with weak lane discipline: a latent leader approach. Transp Res Part C Emerg Technol 67:214–226 Clausse A, Benslimane S, de La Fortelle A (2019) Large-Scale extraction of accurate vehicle trajectories for driving behavior learning. In: 2019 IEEE intelligent vehicles symposium (IV). IEEE.

126

A. K. Budhkar et al.

Das S, Maurya AK (2018) Multivariate analysis of microscopic traffic variables using copulas in staggered car-following conditions. Transp Transp Sci 14(10):829–854 Dey PP, Chandra S, Gangopadhaya S (2006a) Lateral distribution of mixed traffic on two-lane roads. J Transp Eng 132(7):597–600 Dey PP, Chandra S, Gangopadhaya S (2006b) Lateral placement of vehicles under mixed traffic conditions. Indian Highw 34(9):9–17 Duell M, Levin M, Waller ST (2014) On estimating vehicle energy consumption using dynamic traffic assignment vehicle trajectories. In: Proceedings of the 5th International Symposium on Dynamic Traffic Assignment, 17–19 June 2014, Salerno, Italy Dutta B, Vasudevan V (2021) An empirical study on the influence of on-road static obstacles on driver behaviour. Curr Sci 120(4): 00113891 Dutta B, Vasudevan V (2020) Insight into driver behavior during overtaking maneuvers in disorderly traffic: an instrumented vehicle study. Transp Res Procedia 48:719–733 Edie LC (1961) Car-following and steady-state theory for noncongested traffic. Oper Res 9(1):66–76 Erkan I, Hastemoglu H (2016) Bluetooth as a traffic sensor for stream travel time estimation under Bogazici Bosporus conditions in Turkey. J Modern Transp 24(3):207–214 FHWA (Federal Highway Administration) (2007) Next generation simulation (NGSIM) overview. The United States: weblink of FHWA. https://ops.fhwa.dot.gov/trafficanalysistools/ngsim.htm Fung GSK, Yung NHC, Pang GKH (2003) Camera calibration from road lane markings. Opt Eng 42(10):2967–2977 Gajda J et al (2001) A vehicle classification based on inductive loop detectors. In: IMTC 2001. Proceedings of the 18th IEEE instrumentation and measurement technology conference. Rediscovering measurement in the age of informatics (Cat No 01CH 37188), vol 1. IEEE Gunay B (2007) Car following theory with lateral discomfort. Transp Res Part B Methodol 41(7):722–735 Gunay B (2003) Methods to quantify the discipline of lane-based-driving. Traffic Eng Control 44(1):22–27 Haghani A, Aliari Y (2012) Using Bluetooth sensor data for ground-truth testing of reported travel times. In: Transportation research board 91st annual meeting, Washington DC, USA Hao H, Ma W, Xu H (2016) A fuzzy logic-based multi-agent car-following model. Transp Res Part C Emerg Technol 69:477–496 Healey J, Seger J, Picard R (1999) Quantifying driver stress: developing a system for collecting and processing bio-metric signals in natural situations. Biomed Sci Instrum 35:193–198 Helander M, Hagvall B (1976) An instrumented vehicle for studies of driver behaviour. Accid Anal Prev 8(4):271–277 IRC, Indian Roads Congress (1990) 106. guidelines for capacity of urban roads in plain areas. Journal of Indian Roads Congress, New-Delhi Jung Y-K, Ho Y-S (1999) Traffic parameter extraction using video-based vehicle tracking. In: Proceedings 199 IEEE/IEEJ/JSAI international conference on intelligent transportation systems (Cat No 99TH8383), pp 764–769 Kanagaraj V et al (2015a) Trajectory data and flow characteristics of mixed traffic. Transp Res Rec 2491(1): 1–11 Kanagaraj V et al (2015b) Study of unique merging behavior under mixed traffic conditions. Transp Res Part F Traffic Psychol Behav 29:98–112 Khan MA, Ectors W, Bellemans T, Janssens D, Wets G (2017) Unmanned aerial vehicle–based traffic analysis: methodological framework for automated multivehiclet trajectory extraction. Transp Res Rec: J Transp Res Board 2626(1):25–33. https://doi.org/10.3141/2626-04 Khan MQ, Lee S (2019) Gaze and eye tracking: techniques and applications in ADAS. Sensors 19(24):5540 Kivanc Bas, E, Crisman JD (1997) An easy to install camera calibration for traffic monitoring. In: IEEE conference on intelligent transportation systems (1997: Boston Mass) Ko J et al (2006) Instrumented vehicle measured speed variation and freeway traffic congestion. Appl Adv Technol Transp 356–361

6 Emerging Traffic Data Collection Practices …

127

Kotagi PB, Raj P, Asaithambi G (2020) Modeling lateral placement and movement of vehicles on urban undivided roads in mixed traffic: a case study of India. J Traffic Transp Eng (English Edition) 7(6):860–873 Krajewski R et al (2018) The highd dataset: a drone dataset of naturalistic vehicle trajectories on German highways for validation of highly automated driving systems. In: 2018 21st international conference on intelligent transportation systems (ITSC). IEEE Kumar P et al (2017) New methodology for estimating PCU on multi lane urban roads under mixed traffic scenario based on area occupancy, No 17-03248 Leclercq L et al (2007) Relaxation phenomenon after lane changing: experimental validation with NGSIM data set. Transp Res Rec 1999(1):79–85 Mahajan K et al (2019) Effects of driver work-rest patterns, lifestyle and payment incentives on long-haul truck driver sleepiness. Transp Res Part F Traffic Psychol Behav 60:366–382 Mahapatra G, Maurya AK (2018) Dynamic parameters of vehicles under heterogeneous traffic stream with non-lane discipline: an experimental study. J Traffic Transp Eng (English Edition) 5(5):386–405 Mahapatra G, Maurya AK (2013) Study of vehicles lateral movement in non-lane discipline traffic stream on a straight road. Procedia-Soc Behav Sci 104:352–359 Mahapatra G, Maurya AK (2015) Study on lateral placement and speed of vehicles under mixed traffic condition. In: Eastern Asia Society for transportation studies conference, Cebu City, Philippines Mallikarjuna C, Ramachandra Rao K (2006) Area occupancy characteristics of heterogeneous traffic. Transportmetrica 2(3):223–236 Mallikarjuna C, Ramachandra Rao K (2011) Heterogeneous traffic flow modelling: a complete methodology. Transportmetrica 7(5):321–345 Maurya AK, Dey S, Das S (2015) Speed and time headway distribution under mixed traffic condition. J Eastern Asia Soc Transp Stud 11:1774–1792 May AD (1990) Traffic flow fundamentals. Library of Congress, Prentice-Hall Inc. Mei Z, Wang D, Chen J (2012) Investigation with Bluetooth sensors of bicycle travel time estimation on a short corridor. Int J Distrib Sensor Netw 8(1):303521 Moghaddam SS, Hellinga B (2013) Quantifying measurement error in arterial travel times measured by bluetooth detectors. Transp Res Record 2395(1):111–122 Montanino M, Punzo V (2013) Making NGSIM data usable for studies on traffic flow theory: multistep method for vehicle trajectory reconstruction. Transp Res Rec 2390(1):99–111 Mousa M, Sharma K, Claudel CG (2017) Inertial measurement units-based probe vehicles: automatic calibration, trajectory estimation, and context detection. IEEE Trans Intell Transp Syst 19(10):3133–3143 Munigety CR, Vicraman V, Mathew TV (2014) Semiautomated tool for extraction of microlevel traffic data from videographic survey. Transp Res Rec 2443(1):88–95 Nagaraj BN, George KJ, John PK (1992) A study on linear and lateral placement of vehicles in mixed traffic environment through video-recording. Highw Res Bull Indian Road Congr 42:105–136 Nooij SAE et al (2017) Vection is the main contributor to motion sickness induced by visual yaw rotation: implications for conflict and eye movement theories. PloS One 12(4):e0175305 Ossen S, Hoogendoorn SP (2008) Validity of trajectory-based calibration approach of car-following models in presence of measurement errors. Transp Res Rec 2088(1):117–125 Pal D, Chunchu M (2018) Smoothing of vehicular trajectories under heterogeneous traffic conditions to extract microscopic data. Can J Civ Eng 45(6):435–445 Papadimitriou S, Choudhury CF (2017) Transferability of car-following models between driving simulator and field traffic. Transp Res Rec 2623(1):60–72 Papailias F, Thomakos DD (2015) An improved moving average technical trading rule. Physica A 428:458–469 Parmar H, Chakroborty P, Kundu D (2013) Modelling automobile drivers’ toll-lane choice behaviour at a toll plaza using mixed logit model. Procedia Soc Behav Sci 104:593–600

128

A. K. Budhkar et al.

Punzo V, Borzacchiello MT, Ciuffo B (2011) On the assessment of vehicle trajectory data accuracy and application to the Next Generation SIMulation (NGSIM) program data. Transp Res Part C Emerg Technol 19(6):1243–1262 Puri A (2005) A survey of unmanned aerial vehicles (UAV) for traffic surveillance. Department of computer science and engineering, University of South Florida, pp 1–29 Quiroga CA, Bullock D (1998) Travel time studies with global positioning and geographic information systems: an integrated methodology. Transp Res Part C Emerg Technol 6(1–2):101–127 Raju N et al (2021) Developing extended trajectory database for heterogeneous traffic like NGSIM database. Transp Lett 1–10 Raju N et al (2017) Examining smoothening techniques for developing vehicular trajectory data under heterogeneous conditions. J Eastern Asia Soc Transp Stud 12:1549–1568 Robicquet A et al (2016) Learning social etiquette: human trajectory understanding in crowded scenes. In: European conference on computer vision. Springer, Cham Saha P et al (2009) Passenger car equivalent (PCE) of through vehicles at signalized intersections in Dhaka Metropolitan City, Bangladesh. IATSS Res 33(2):99–104 Salvo G et al (2017) Traffic data acquirement by unmanned aerial vehicle. Eur J Remote Sens 50(1):343–351 Das S, Maurya AK, Budhkar AK (2017) Determinants of time headway in staggered car-following conditions. Transp Lett Scarzello JF et al (1978) SPVD: a magnetic vehicle detection system using a low power magnetometer. IEEE Trans Magn 14–5(1976):574–576 Sekhar CR et al (2013) Analysis of travel time reliability of an urban corridor using micro simulation techniques. Curr Sci 319–329 Seong S et al (2019) Determination of vehicle trajectory through optimization of vehicle bounding boxes using a convolutional neural network. Sensors 19(19):4263 Sil G et al (2019) Operating speed prediction model as a tool for consistency based geometric design of four-lane divided highways. Transport 34(4):425–436 Singh H, Kathuria A (2021) Analyzing driver behavior under naturalistic driving conditions: a review. Accid Anal Prev 150:105908 Soni R, Vasudevan V, Dutta B (2020) Analysis of overtaking patterns of Indian drivers with data collected using a LiDAR. Transp Res F Traffic Psychol Behav 74:139–150 Sudha D, Priyadarshini J (2020) An intelligent multiple vehicle detection and tracking using modified vibe algorithm and deep learning algorithm. Soft Comput 24:17417–17429 Tiwari G, Fazio J, Gaurav S (2007) Traffic planning for non-homogeneous traffic. Sadhana 32(4):309–328 Toledo T, Koutsopoulos HN, Ahmed KI (2007) Estimation of vehicle trajectories with locally weighted regression. Transp Res Rec 1999(1):161–169 Torres ME et al (2011) A complete ensemble empirical mode decomposition with adaptive noise. In: 2011 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE Vasconcelos L et al (2014) Calibration of the Gipps car-following model using trajectory data. Transp Res Procedia 3:952–961 Venter CJ, Knoetze H (2013) Lateral clearance between vehicles and bicycles on urban roads. In: SATC 2013 Venthuruthiyil SP, Chunchu M (2020) Vehicle path reconstruction using Recursively Ensembled Low-pass filter (RELP) and adaptive tri-cubic kernel smoother. Transp Res Part C Emerg Technol 120:102847 Verma A et al (2016) Assessment of driver vision functions in relation to their crash involvement in India. Curr Sci 110(6):1063 Vlahogianni EI (2015) Computational intelligence and optimization for transportation big data: challenges and opportunities. In: Engineering and applied sciences optimization. Springer, Cham, pp 107–128

6 Emerging Traffic Data Collection Practices …

129

Wang L-L, Tsai W-H (1991) Camera calibration by vanishing lines for 3-D computer vision. IEEE Trans Pattern Anal Mach Intell 13(4):370–376 Wang Y, Vrancken JLM, Seidel P (2011) Measure travel time by using Bluetooth detectors on freeway. In: Proceedings of the ITS world congress Wong CY, Qidwai U (2004) Vehicle collision avoidance system [VCAS]. In: SENSORS, 2004 IEEE. IEEE Wu Z, Huang NE (2009) Ensemble empirical mode decomposition: a noise-assisted data analysis method. Adv Adapt Data Anal 1(01):1–41 Xu J, Min J, Hu J (2018) Real-time eye tracking for the assessment of driver fatigue. Healthc Technol Lett 5(2):54–58 Yuan H (2009) A novel trajectory smoothing algorithm based on empirical mode decomposition. In: 2009 fifth international conference on image and graphics. IEEE Zhao H et al (2016) On-road vehicle trajectory collection and scene-based lane change analysis: Part I. IEEE Trans Intell Transp Syst 18.1:192–205

Chapter 7

Planning for Suitable Walk-Access Infrastructure Components in Various Classes of Urban Bus Stop Catchment Area Subhojit Roy and Debasis Basu

7.1 Introduction In recent years, urban areas in India have been experiencing a significant increase in travel demand in spite of augmenting the adequate capacity of road infrastructure. This has also resulted in a severe impairment of urban mobility. In order to improve travel demand management, recent studies (Maitra and Sadhukhan 2013; Cheranchery and Maitra 2018) emphasized the need to promote city bus services in urban areas in India. In order to improve the overall quality (Maitra and Sadhukhan 2013; Roy and Basu 2019) of city bus service, it is not only required to improve its service quality; but also its accessibility by walk-mode. Recently, a few studies (Rastogi and Rao 2003; Rastogi 2010; Rahul and Verma 2014; Sadhukhan et al. 2015) were carried out in the context of urban public transport, where improvement of walk-accessibility around service access points was of the primary focus. Some other studies (Zhu and Liu 2004; Martin et al. 2008) also deliberated on the improvement of operational characteristics of public transport as a part of overall improvement. In a similar line, Roy and Basu (2020a) observed that the propensity to access urban bus service by walk mode is bi-directional elastic—i.e., the walk-access infrastructure and the operational characteristics of bus service. This fact was also realized by transport planners and professionals in India and several initiatives of the Government of India, such as the Atal Mission for Rejuvenation and Urban Transformation (AMRUT) mission (MoUD 2015a), SMART CITY mission (MoUD 2015b), etc., emphasized on the need for integration between the walk-access and the city bus service. The SMART City mission explicitly pays importance on the “first-and-last mile” connectivity (the need for the same was discussed in the workshop on “Smart Mobility Solutions in Bhubaneswar” in 2017, S. Roy · D. Basu (B) School of Infrastructure, Indian Institute of Technology Bhubaneswar, 752050, Jatani, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_7

131

132

S. Roy and D. Basu

organized by Bhubaneswar Development Authority) as a part of the overall improvement of transit-oriented development (TOD) with convenient accessibility. Various civic authorities in developed countries (AASHTO 2004; KMOCT 2001; Department of Transportation, State of Hawaii 2013; DoIT, Taipei City Government 2012; Department of Transport and Australia 2016) have a set of guidelines for the desired engineering features of walk-access facility as a part of achieving the “first-andlast mile” connectivity. But such attempts in the context of urban India have only been limited to the proposition of generic walk facility guidelines. The contemporary design guideline of best practices (Traffic et al. 2009; IRC103:2012 2012; Directorate of Urban Land Transport (DULT) 2014) for walk facility in urban India does not specifically incorporate planning and integration of walk-access infrastructure with urban bus corridors. In light of such requirement, the current investigation scientifically identifies a plausible combination of various walk-access infrastructure components in urban bus stop catchment areas by empirically evaluating the choice perspective of walk-accessed bus users. The empirical study is carried out by considering various types of bus stop catchment areas on a selected urban bus service corridor in Bhubaneswar, India. As a part of empirical study, a choice model is developed and economic values are estimated of various functional attributes describing infrastructure components of walk-access facility. Several hypothetical door-to-door travel scenarios of walk-access bus users are generated considering various types of bus stop catchment on the study corridor, availability of road space on access roads, and their other physical constraints in the catchment area. In this regard, a classification of bus stop catchment areas developed by Roy and Basu (2020c) in the context of an urban area has been considered as a guidance for generating various hypothetical but feasible travel scenarios of walk-accessed bus users. With due consideration of all the above, a door-to-door (d2d) generalized cost (GC) function is framed for walk-accessed bus users. It could be mentioned that a d2d-trip for a walk-accessed bus user consists of three trip-legs, the access leg, the bus journey leg, and the egress leg. Therefore, the d2d-GC of walk-accessed bus users is likely to be composed of the GCs of all these three legs of a trip. The saving in the value of d2d-GC under various hypothetical travel scenarios is then considered as the measure of effectiveness (MoE). Using such MoE, a suitable combination of walk-access infrastructure components are determined for various types of bus stop catchment areas based on near-minimal generalized cost incurred by walk-accessed bus users across various bus stop catchment areas. The remainder of the chapter is organized as follows. In the subsequent section, a discussion on the bus stop catchment area on either side of the study corridor is made. Following that, the choice preference study on various walk-infrastructure components is presented in the next section. Once the choice-based study is carried out and the generalized-cost function for d2d travel is constructed, the selection of suitable walk-infrastructure components is discussed in the next section considering saving in the values of generalized cost under various improvement scenarios. In this section, a table is presented, which suggests the possible implementation plan of various walk-infrastructure components with reference to a specific class of bus stop catchment area.

7 Planning for Suitable Walk-Access Infrastructure Components …

133

7.2 Description of Study Area As mentioned, the study area is taken from Bhubaneswar city. A bus service corridor is selected (as shown in Fig. 7.1), which is passing through major residential areas and various types of economic activity areas of the city such as KIIT Square, Patia, Chandrasekharpur, Jaydev Vihar, Vani Vihar, and Master Canteen areas. Major bus boarding and alighting activities are usually found to happen on this study corridor, and bus passengers are mostly found as walk-accessed bus users. Around any bus stop location on the selected study corridor, a bus stop catchment area is demarcated (TRB 2003) by the geographical boundary representing the maximum spatial extent from where walk-accessed bus users are usually found to access bus service corridor by walk-mode. As per the work of Southworth (2005), this geographical boundary could be approximated by a buffer area having a suitably assumed Euclidean radius. In line with this, a study was conducted by IIT Bhubaneswar (IIT Bhubaneswar unpublished report, 2014), where a walk-access length of 700 m was taken as the equivalent Euclidian radius for defining a typical bus stop catchment area in Bhubaneswar. The current study considers about 16 of such bus stop catchment areas around 16 bus stop locations on the study corridor (Fig. 7.1). It was observed during the field study that each catchment consisted of a small network of access roads. Such access roads within a catchment were found to have varying geometric features along its running length. Figure 7.1 shows the typical network of access roads within a typical bus stop catchment area. The figure also highlights various types of road-space constraints usually observed on the access roads such as the absence of pedestrian

Bhubaneswar, India

Typical Bus Stop Catchment Area No pedestrian RoW

KIIT Square Patia Square

Chandrashekharpur

Unpaved shoulder as RoW

Vani Vihar Square

Jaydev Vihar

Inadequate RoW

Master Canteen

Pedestrian RoW on one side only

Study Corridor

Fig. 7.1 Illustration of typical bus stop catchment area around urban bus stops on the study corridor

134

S. Roy and D. Basu

right-of-way (RoW), inadequate pedestrian RoW, availability of pedestrian RoW but only at one side of the road, unpaved shoulder space, etc.

7.3 Preference Study on Walk-Access Infrastructure Components Though many of the previous studies on the types of walk-access facility with reference to the developed countries (Woldeamanuel and Cyganski 2011; Jiang et al. 2012; Tilahun and Li 2015) were found in contemporary literature, only limited studies (Roy and Basu 2019; Sadhukhan et al. 2015) were found in the context of developing countries like India. Besides, most of these studies were carried out focusing on the user-perceived data collected on an ordinal scale. None of these studies highlighted the need of the users’ choice-perspective evaluation for planning of suitable access facility. In addition, there isn’t enough evidence found in established literature highlighting engineering specifications or types of walk-infrastructure components in line with the expectation of road users. Although the existing guidelines (AASHTO 2004; KMOCT 2001; IRC103:2012 2012; CoLAPW 2010) of best practices suggest various types of walk infrastructure, these guidelines do not provide any implementation plan of walk-access infrastructure components with their specifications as per users’ choice preferences. In absence of such guidelines, the current work takes an attempt to identify and then to select a suitable combination of various walk-infrastructure components of access roads by paying due importance to the choice preferences of users. As a part of this primary focus area, a preliminary study (Roy and Basu 2017, 2020a, b) was carried out to select the intervention areas by considering two sets of attributes: one set of attributes pertaining to the walk-access infrastructure, and another set of attributes pertaining to the operational characteristics of bus service. This study revealed that walk-access attributes such as the presence of dedicated sidewalk, adequate width of sidewalk, quality of road surface on sidewalk, presence of crosswalk, presence of street lighting and road surveillance on sidewalk, and operational characteristic attributes of bus service such as service headway, journey speed, and fare of journey as intervention areas for overall improvement of d2d travel condition of walk-accessed bus users. These intervention areas are then re-expressed by a set of functional attributes for capturing their choice preferences in a choice-based analysis. Tables 7.1 and 7.2 show the set of functional attributes of walk-access infrastructure and operational characteristics of bus service, respectively, where their levels of functional attributes stand for their specifications.

7 Planning for Suitable Walk-Access Infrastructure Components …

135

Table 7.1 Functional attributes and their levels for walk-access infrastructure Sl# Functional attributes

Level (code)

1

Level 1 (base) (SW0) Unmarked and unpaved shoulder adjacent to the vehicular RoW

2

3

4

Types of sidewalk

Description of infrastructure components

Level 2 (SW1)

Marked pedestrian RoW, which is at-grade with the carriageway, and surfacing of RoW is color-coded (i.e. green marked) and having illuminated (i.e. cat’s eye) corridor

Level 3 (SW2)

Raised pedestrian sidewalk having paved surface but without buffer zone from the vehicular traffic

Level 4 (SW3)

Raised pedestrian sidewalk having paved surface but having 0.5 m buffer from the vehicular traffic

Length of walk-access Level 1

15% of current walk length

Level 2

30% of current walk length

Level 3

45% of current walk length

Lighting and security

Types of crosswalk

Level 4

60% of current walk length

Level 1 (base) (LS0)

Street light is present at every 20–30 m. with light pole-mounted at 6–8 m height but having relatively low lighting intensity, but no electronic security and surveillance system

Level 2 (LS1)

Street light is present in the form of bollard light at every 5–10 m and having medium to high lighting intensity, but no electronic security and surveillance system

Level 3 (LS2)

In addition to Level 2, CCTV camera is installed at every 25 m along the access road and at all major intersections within of the catchment area

Level 1 (base) (CW0) Crosswalk is provided only at the intersection of access road and main bus corridor Level 2 (CW1)

Crosswalk is provided at the intersection of access road and main bus corridor, and also at every 100–200 m of distance along the access road

Level 3 (CW2)

Table-top crosswalk along with the rumble surface acting as traffic-calming measures and in-pavement flashers acting as warning signal is provided at the intersection of access road and main bus corridor, and also at every 100–200 m of distance along the access road

136 Table 7.2 Functional attributes and their levels for operational characteristics of bus service

S. Roy and D. Basu Sl#

Functional attributes

Level

Operational levels

1

Journey speed

Level 1 (base)

20 kmph

Level 2

22 kmph

2

3

Headway

Fare

Level 3

25 kmph

Level 1 (base)

15 min

Level 2

10 min

Level 3

5 min

Level 1

30% more than the current fare

Level 2

20% more than the current fare

Level 3

10% more than the current fare

Level 4 (base)

Current fare

7.3.1 Elicitation of Preference Data Using the functional attributes and their corresponding levels (as mentioned in Tables 7.1 and 7.2), a total of 36 choice scenarios is developed for d2d travel condition. In those d2d travel scenarios, access-leg and egress leg consist of walkmode only. These choice scenarios are developed using orthogonal optimal design as demonstrated in Choice Metrics (2018). While carrying out data collection on access roads of a given bus stop catchment area, initially any pedestrian was randomly approached and was enquired with whether the intercepted pedestrian was accessing to (or egressing from) the bus stop by walk-mode. Once, the randomly intercepted pedestrian was confirmed as a walk-accessed bus user, then she or he was requested to participate in the survey. During the choice-experimentation study in the field, each survey respondent was requested to randomly pick up any one of the choice cards from the pool of 36 choice cards such that the probability of picking up any choice card remains equally likely. This process of experimentation was repeated for four consecutive times for all survey respondents, thus resulting in obtaining four choice preference sample observations from each of the survey respondents. In this way, a total of 5092 choice preference samples was finally used for developing a choice-based model.

7 Planning for Suitable Walk-Access Infrastructure Components …

137

7.3.2 Valuing of Functional Attributes Using Choice-Based Model and Construction of GC Function In wake of valuing of attributes, the multinomial logit (MNL) model (as described in Train (2009)) was initially attempted as a base model in order to diagnose various estimated coefficients of functional attributes and their levels with appropriate signs and t-statistics. When a MNL model with statistically significant estimated coefficients was developed, then the Random Parameter Logit (RPL) model as described in Hensher et al. (2015) was estimated. While developing of RPL model, the levels of various non-cost functional attributes were assumed to follow the constrainedtriangular distribution, where the mean of the distribution equals the spread of the random parameter. Such RPL model accounted for the unobserved taste heterogeneity across sample datasets through estimation of mean and spread of the assumed distribution function. The application of such constrained-triangular distribution for capturing random taste heterogeneity on various non-cost travel attributes was also reported in the work of Sadhukhan et al. (2016) and Dandapat and Maitra (2020). During the development of a choice model, the levels of qualitative attributes such as “types of sidewalk”, “lighting and security”, and “types of crosswalk” were entered into the model by dummy coding. On the other hand, the quantitative attributes such as “length of walk-access”, “headway”, “journey speed”, and “bus fare” were entered into the model in cardinal linear form. It was also mentioned that the journeyspeed data was converted into the equivalent journey time before being entered into the model estimation. The journey time was calculated by dividing the de-coded journey-distance value between boarding and alighting points. Needless to mention, the consideration of journey time instead of journey speed during logit model estimation introduced a large variation on journey time datasets due to the large variation of observed journey lengths in datasets. Once the RPL model was developed, the values of all the non-cost attributes were estimated, and these values were later considered for the construction of d2d-GC functions for walk-accessed bus users. Reporting of the coefficient estimates from MNL model is not the scope of the current chapter. The following section reports only the estimated coefficients and economic values of functional attributes as emanating from the RPL model. Table 7.3 shows the coefficient estimates and also the model performances such as the loglikelihood values and the adjusted ρ 2 values. The same table also shows that all estimated coefficients are statistically and significantly different from zero at 99% confidence level. However, the values of the estimated coefficients are primarily useful for checking the statistical significance of the functional attributes. The interpretation of the coefficient estimates of dummy-coded functional attributes, and thereby their values in GC function usually requires a substitution by relevant values while expressing the generalized cost function. This substitution needs to be carried out by rescaling the values of dummy coded attributes. The values of the dummy coded attributes, and their levels are re-scaled (as mentioned in Hensher et al. (2015)) in such a manner that the value of the attribute level corresponding to the base condition represents the maximum cost, thereby

1.23

0.63

0.33

–

0.89

0.39

–

1.18

0.52

–

SW3

SW2

SW1

SW0

LS2

LS1

LS0

CW2

CW1

CW0

–

0.06

0.08

–

0.06

0.07

–

0.09

0.09

0.12

Std error

–

8.60

13.99

–

6.30

12.22

–

3.88

6.65

10.18

t-stat

0.250

Adjusted ρ 2

t-stat −16.69 −17.75

−9.27

–

0.0009

0.0011

Value of attribute

Re-scaled values of attribute-levels

Rs./min/metre walk length

Units

Rs./metre walk length

Rs./metre walk length

Rs./metre walk length

Units

All coefficient estimates are statistically significant (i.e. significantly different from zero at 99% confidence level with t-stat value >2.576) (1 USD ≈ Rupees 70.41 as on 2019)

5092

8.79

−156.08

Bus fare

−2293.16

0.02

−0.17

Journey time

Log-likelihood

0.01

−0.15

Headway

No. of obs

Std error

Coeff

Attributes

0.0076

0.0043

−0.0033 –

0.0057 0.0000

–

0.0032

−0.0025 −0.0076

0.0076 0.0000

0.0055

−0.0021 –

0.0038

−0.0038

−0.0057

0.0000

−0.0076

Value of attribute-levels

Functional attributes relating to operational characteristics of bus service

Coeff

Codes for attribute-levels

Functional attributes relating to walk-access facility

Table 7.3 Coefficient estimates from RPL model and Values of functional attributes and their various levels

138 S. Roy and D. Basu

7 Planning for Suitable Walk-Access Infrastructure Components …

139

drawing of the minimum utility. The rescaling of the value of any given level is carried out by fixing the value-cost of the most-desired level as zero. The rescaled values of all dummy-coded levels of all attributes, as well as for cardinal-coded attributes, are reported in Table 7.3. It can be interpreted from Table 7.3 that a walkaccessed bus user currently incurs about Rs. 0.0076/- for every metre of a walk length on a sidewalk facility of type SW0, i.e., the base condition. This cost reduces by Rs. 0.002/-, if a sidewalk facility of type SW1 is introduced in comparison to the base level. The reduction in cost for every metre of walk on a sidewalk of type SW2 from SW1 remains almost the same (about Rs. 0.017/-) as observed above. But, the reduction in cost incurred for level change from type SW2 to type SW3 relatively becomes more (i.e., Rs. 0.0038/-). This implies that the introduction of safety-feature such as providing of buffer zone on the sidewalk facility (as mentioned in SW3 level) would be perceived as a significant benefit toward users. In case of “lighting and security”, a walk-accessed bus user currently incurs about Rs. 0.0057/- for every metre of walk on access facility having LS0 type of lighting and security, i.e., the base level. It is observed that the reduction in cost incurred from the level change of type LS0 to type LS1 is about Rs. 0.0025/-, whereas from the level change of type LS1 to type LS2 is about Rs. 0.0032/- for every metre of walk length. This observation implies that the enhancement of security features by the addition of just an electronic surveillance system is more appreciated by users in comparison to the bollard light (i.e., LS1) being introduced instead of the pole-mounted street light (i.e., base level LS0). In the case of “types of crosswalk”, currently, a walk-accessed bus user incurs a cost of Rs. 0.0076/- for every metre of walk while experiencing the CW0 type of crosswalk, i.e., the base condition. The reduction in cost from the level change of type CW0 to type CW1 is about Rs. 0.0033/-, whereas the same level change of type CW1 to type CW2 is about Rs. 0.0043/- for every metre of walk length. This observation implies that the safety feature by introducing traffic-calming and warning measures is more appreciated by the walk-accessed bus users. Using the values of functional attributes as presented in Table 7.3, the generalized cost function of d2d travel of any given walk-accessed bus user is constructed. The GC function consists of three legs such as the access leg to bus stop, the bus journey leg, and the egress leg from the bus stop. Therefore, the d2d-GC (i.e., door-to-door generalized cost) of walk-accessed bus users is likely to be composed of the GCs of all of these three legs of a trip. In line with this philosophy, the total GC of the walk-accessed bus users is calculated in accordance with the following Eq. (7.1) (d2d − GC)/tri p =GCaccess−leg + GCbus−leg + GCegr ess−leg   α H × H × (L a + L e ) ={swa L a + lsa L a + cwa L a } + +α J T × J T × (L a + L e ) + F + {swe L e + lse L e + cwe L e } =L a {swa + lsa + cwa } + (L a + L e ){α H × H + α J T × J T } + F + L e {swe + lse + cwe }

(7.1)

140

S. Roy and D. Basu

where wa = value of the level for “types of side-walk” on access-leg we = value of the level for “types of side-walk” on egress leg ls a = value of the level for “lighting and security” on access-leg ls e = value of the level for “lighting and security” on egress leg cwa = value of the level for “types of crosswalk” on access-leg cw e = value of the level for “types of crosswalk” on egress leg α H = value of the time headway (H) α J T = value of the journey time (JT) F = bus fare L a = walk-length on access-leg L e = walk-length on egress leg Usually, in the study area, it was observed that a typical walk-accessed bus user experienced about 20 min journey time with a headway of about 15 min of bus service and an average fare of Rs. 10/- per trip. Besides, it was also observed that walk-accessed bus users on an average walked about 470 m on access leg and about 420 m on egress leg. Considering this information, it may be mentioned that the d2d-GC of any typical walk-accessed bus user usually equals about Rs. 60/- trip (as per Eq. 7.2). d2d − GC = L a {swa + lsa + cwa } + (L a + L e ){α H × H + α J T × J T } + F + L e   0.0011 × 15 = 470{0.0076 + 0.0057 + 0.0076} + (470 + 420) +0.0009 × 20 + 10 + 420{0.0076 + 0.0057 + 0.0076} = Rs. 59.3/ ≈ Rs. 60/ − per d2d trip

(7.2)

7.4 Selection of Suitable Walk-Access Infrastructure Components As observed, the generalized-cost function of walk-accessed bus users is greatly influenced by the types of walk-infrastructure components experienced on access leg as well as on egress leg. But, the provision of implementation of an improved version of infrastructure component with the best specification may not always be possible because of various types of constraints usually observed on access roads in

7 Planning for Suitable Walk-Access Infrastructure Components …

141

urban areas of developing countries. As discussed, these road-space constraints arise due to the absence of pedestrian RoW, inadequate pedestrian RoW, and availability of pedestrian RoW but only at one side of the road unpaved shoulder space, etc. Therefore, appropriate types of walk infrastructure needs to be implemented in specific types of catchment area, where road-space constraints are of varying nature. With regard to the road-space constraints, Roy and Basu (2020a) developed a classification for bus stop catchment areas in accordance with the access roads usually observed in urban areas, and then suggested the implementation plan of the appropriate sidewalk facility for each class of bus stop catchment area. The following Table 7.5 is adapted from the work of Roy and Basu (2020a), which suggests the implementation strategy of the sidewalk facility. Though the above table shows a total of 6 classes of bus stop catchment area, most of the observed walk-accessed bus users in the study area belong to using of the catchment classes such as B, C, and D. In view of the implementation strategy of sidewalk facility, a set of hypothetical improvement scenarios of walk-infrastructure are generated and evaluated primarily for the above three types of catchment class. Table 7.5 Classification of Bus Stop catchment areas as adapted from the work of Roy and Basu (2020a) Catchment class

Availability of pedestrian right-of-way (PRoW)

Implementation strategy Types of sidewalk

Alignment of sidewalk

A

PRoW < 1.4 m and available on one side of access road

Marked sidewalk only

One side of walk-access road

B

1.4 m < PRoW < 2.0 m and Raised sidewalk, if available on one side of 1.50 m ≤ PRoW < access road 2.00 m Marked sidewalk, if 1.40 m ≤ PRoW < 1.50 m

One side of walk-access road

C

PRoW < 2.0 m and available on both sides of access road

Both sides of walk-access road

D

2.0 m < PRoW < 4.0 m and Raised sidewalk available on both side of access road

Both sides of walk-access road

E

PRoW < 3.0 m and available on both sides of access road

Raised sidewalk, if 1.50 m ≤ PRoW < 3.00 m

Both sides of walk-access road

F

3.0 m ≤ PRoW < TRoW and available on both sides of access road

Raised sidewalk

Both sides of walk-access road

Raised sidewalk, if 1.50 m ≤ PRoW < 2.00 m Marked sidewalk, if PRoW < 1.50 m

142

S. Roy and D. Basu

7.4.1 Generation of Plausible Improvement Scenarios of Walk-Access Infrastructure With regard to the study area, the d2d trips of walk-accessed bus users are primarily found to occur between B-C, C-D, and B-D combination of catchment classes. Before proceeding to generate improvement scenarios of these catchment classes, it is essential to take a closer look at plausible types of alternative infrastructure components that could be introduced in the above catchments. It is evident from Table 7.5 that in catchment class B, only marked sidewalk (SW1) can be implemented as an improvement measure wherever the PRoW is less than 1.50 m. This is because the minimum width of a raised sidewalk (as defined by levels SW2 and SW3) is usually specified to be 1.50 m as per Indian standards (Traffic et al. 2009; IRC103:2012 2012). In case of catchment class B, another case could arise where the PRoW lies between 1.50 and 2.0 m. In such a case, SW2 type sidewalk (i.e., raised sidewalk without buffer zone) could be provided, but not SW3 due to lateral-space constraints. However, in all the above cases, the sidewalks could be implemented on one side of the access road only. Considering the above mentioned feasible options for types of sidewalk facility, the crosswalks of type CW0 or CW1 would be appropriate but not CW2. This is because table top crosswalk would not be feasible to be implemented in association with a marked pedestrian facility or even raised sidewalk developed on one side of the access road. However, in the catchment class C, raised sidewalk could be considered on both sides of the access road, but the option would be limited to SW2 level, but not SW3 level due to lateral-space constraints. In such cases, however, table top crosswalk (CW2) could be thought of as a plausible option. In catchment class D, there are no practical restrictions on implementing of raised sidewalk with additional buffer zone (i.e., SW3), and therefore, CW2 type crosswalk could also be introduced in such cases. Based on all these considerations, a total of 10 feasible improvement scenarios are developed for each of the catchment area combinations B-C (shown in Table 7.6), C-D (shown in Table 7.7), and B-D (shown in Table 7.8). While considering d2d trips between the above classes of the catchment, the operational characteristics of the city bus service are considered to remain the same as that of the prevailing condition.

7.4.2 A Comparison of d2d-GC Savings Among Various Improvement Scenarios of Walk-Access Infrastructure The saving in the values of d2d-GC in comparison to the current condition is evaluated for each of the improvement scenarios of walk-access infrastructure as mentioned in Tables 7.6, 7.7 and 7.8. The bar charts in Figs. 7.2, 7.3 and 7.4 show the amount of saving in d2d-GC for d2d-trips between catchment combinations B-C, C-D, and B-D respectively. The average amount of d2d-GC saving is observed to be the least (about 7.81%) for the catchment combination of B-C, whereas the same is observed to be

7 Planning for Suitable Walk-Access Infrastructure Components …

143

Table 7.6 Improvement scenarios of walk-access infrastructure, when d2d-trips occur between B-C catchment classes Catchment class B

Catchment class C

Codes of improvement scenarios

Type of sidewalk

Type of lighting and security

Type of crosswalk

Type of sidewalk

Type of lighting and security

Type of crosswalk

B-C ImprSc1

SW1

LS1

CW0

SW1

LS1

CW0

B-C ImprSc2

SW1

LS1

CW1

SW2

LS1

CW1

B-C ImprSc3

SW1

LS2

CW1

SW2

LS1

CW0

B-C ImprSc4

SW0

LS1

CW1

SW2

LS0

CW2

B-C ImprSc5

SW1

LS1

CW0

SW2

LS1

CW1

B-C ImprSc6

SW2

LS1

CW0

SW1

LS0

CW1

B-C ImprSc7

SW1

LS2

CW1

SW2

LS1

CW1

B-C ImprSc8

SW2

LS1

CW0

SW1

LS0

CW1

B-C ImprSc9

SW1

LS1

CW1

SW2

LS0

CW1

B-C ImprSc10

SW1

LS2

CW1

SW2

LS0

CW2

Table 7.7 Improvement scenarios of walk-access infrastructure, when d2d-trips occur between C-D catchment classes Catchment class C

Catchment class D

Codes of improvement scenarios

Type of sidewalk

Type of lighting and security

Type of crosswalk

Type of sidewalk

Type of lighting and security

Type of crosswalk

C-D ImprSc1

SW2

LS1

CW1

SW3

LS2

CW2

C-D ImprSc2

SW0

LS1

CW1

SW2

LS1

CW0

C-D ImprSc3

SW1

LS2

CW1

SW2

LS1

CW1

C-D ImprSc4

SW0

LS1

CW1

SW3

LS2

CW2

C-D ImprSc5

SW0

LS1

CW0

SW2

LS2

CW1

C-D ImprSc6

SW2

LS1

CW0

SW0

LS1

CW1

C-D ImprSc7

SW2

LS1

CW1

SW2

LS1

CW1

C-D ImprSc8

SW0

LS1

CW1

SW1

LS0

CW2

C-D ImprSc9

SW1

LS1

CW1

SW2

LS1

CW1

C-D ImprSc10

SW3

LS2

CW1

SW0

LS0

CW2

the highest (about 10.08%) for the catchment combination C-D. The average amount of d2d-GC saving for the catchment combination B-D is observed to lie between the above two values (i.e., about 9.78%). The above happens due to the physical constraints usually observed on access roads of catchment class B, where the provision of SW0 and SW1 types of sidewalk

144

S. Roy and D. Basu

Table 7.8 Improvement scenarios of walk-access infrastructure, when d2d-trips occur between B-D catchment classes Catchment class B

Catchment class D

Codes of improvement scenarios

Type of sidewalk

Type of lighting and security

Type of crosswalk

Type of sidewalk

Type of lighting and security

Type of crosswalk

B-D ImprSc1

SW1

LS1

CW0

SW2

LS1

CW0

B-D ImprSc2

SW1

LS1

CW1

SW2

LS1

CW0

B-D ImprSc3

SW1

LS1

CW1

SW3

LS1

CW1

B-D ImprSc4

SW0

LS1

CW1

SW3

LS0

CW2

B-D ImprSc5

SW0

LS1

CW0

SW2

LS2

CW1

B-D ImprSc6

SW2

LS1

CW0

SW1

LS1

CW1

B-D ImprSc7

SW1

LS2

CW1

SW3

LS1

CW1

B-D ImprSc8

SW0

LS1

CW1

SW1

LS0

CW2

B-D ImprSc9

SW1

LS1

CW1

SW2

LS1

CW1

B-D ImprSc10

SW1

LS1

CW1

SW1

LS0

CW2

Fig. 7.2 Saving in d2d-GC under various improvement scenarios, when d2d-trips occur between B-C catchment classes

and CW0 and CW1 types of crosswalk are only considered. These types of infrastructure components incur higher generalized cost associated with them in comparison to the SW3 and CW2 types of infrastructure components. It can also be noticed (in Figs. 7.2, 7.3 and 7.4) that, a total of 11 scenarios (3 scenarios for B-D, 4 scenarios for C-D, and 4 scenarios for B-D) appear across all the catchment area combinations, where the savings in d2d-GC are greater than their respective average savings. In

7 Planning for Suitable Walk-Access Infrastructure Components … 14%

13.57% 12.87% 12.24% 11.54%

12%

Amount of % savings in d2d-GC

145

9.56%

Average % savings in d2d-GC=10.08 %

9.87%

10%

8.50% 8%

7.08%

7.57%

8.05%

6% 4% 2% 0% C-D C-D C-D C-D C-D C-D C-D C-D C-D C-D ImprSc-1 ImprSc-2 ImprSc-3 ImprSc-4 ImprSc-5 ImprSc-6 ImprSc-7 ImprSc-8 ImprSc-9 ImprSc-10

Fig. 7.3 Saving in d2d-GC under various improvement scenarios when d2d-trips occur between C-D catchment classes 14%

11.57%

Amount of % saving in d2d-GC

12%

Average % savings in d2d-GC=9.78 % 11.89%

11.35% 10.32%

9.78%

10% 8.54%

8.81% 8.00%

9.10% 8.45%

8%

6%

4%

2%

0% B-D B-D B-D B-D B-D B-D B-D B-D B-D B-D ImprSc-1 ImprSc-2 ImprSc-3 ImprSc-4 ImprSc-5 ImprSc-6 ImprSc-7 ImprSc-8 ImprSc-9 ImprSc-10

Fig. 7.4 Saving in d2d-GC under various improvement scenarios when d2d-trips occur between B-D catchment classes

146

S. Roy and D. Basu

other words, these 11 improvement scenarios would likely result in the near-optimal d2d-GC savings for walk-accessed bus users across B, C, and D catchment classes. Intuitively, it may be said that the suitable infrastructure components across all three classes of catchment areas could be introduced from the infrastructure components appearing in the above mentioned 11 scenarios. The following section describes the selection process of the appropriate walk-infrastructure components for the three classes of catchment areas.

7.4.3 Toward Selection of Suitable Components of Walk-Infrastructure Based on Their Frequent Occurrence in Various Catchment Classes In order to select the appropriate walk-infrastructure components for the catchment class of type B, it is required to investigate those improvement scenarios involving catchment class of type B and also reports higher saving in d2d-GC. Among those identified 11 scenarios, it is observed that a total of 7 scenarios is present in the aforementioned criteria. These scenarios are B-C ImprSc-2, B-C ImprSc-5, B-C ImprSc9, B-D ImprSc-3, B-D ImprSc-7, B-D ImprSc-8, and B-D ImprSc-10. Similarly, in case of catchment class of type C, a total of 7 scenarios such as B-C ImprSc-2, B-C ImprSc-5, B-C ImprSc-9, C-D ImprSc-1, C-D ImprSc-3, C-D ImprSc-4, and C-D ImprSc-7 could be introduced. In a similar manner, a total of 8 scenarios such as B-D ImprSc-3, B-D ImprSc-7, B-D ImprSc-8, and B-D ImprSc-10, C-D ImprSc-1, C-D ImprSc-3, C-D ImprSc-4, and C-D ImprSc-7 could be introduced for catchment class of type D. The d2d-GC saving for all of these identified 11 scenarios is compared and the various combination of walk-infrastructure components present in the case of each catchment class and their number of occurrences are shown in Fig. 7.5. It is evident from Fig. 7.5 that in case of catchment class B, the components of walk infrastructure consisting of SW1 type of sidewalk, LS1 type of lighting and security, and CW1 type of crosswalk are mostly found to exist across all scenarios and generate higher saving in d2d-GC. Similarly, in the case of catchment class C, the components of walk infrastructure consisting of SW2 type of sidewalk, LS1 type of lighting and security, and CW1 type of crosswalk are mostly found to exist across all scenarios and generate higher savings in d2d-GC. Finally, in case of catchment class D, the components of walk infrastructure consisting of SW3 type sidewalk, LS2 type lighting and security, and CW2 type crosswalk are mostly found to exist among all scenarios and generate higher savings in d2d-GC. Thereby, it may be mentioned that the above three components of walk infrastructure for each class of B, C, and D could be termed as a suitable combination, which induces relatively high d2d-GC saving for walk-accessed bus users. Thereby, the above combination of walk-access infrastructure is expected to bring near-minimal d2d-GCs for walk-accessed bus users in the study area. Table 7.9 summarizes the suitable combination of various walkaccess infrastructure components for B, C, and D types of catchment class based

7 Planning for Suitable Walk-Access Infrastructure Components …

147

60%

Frequency of Occurrence

50%

40%

Catchment class B Catchment class C Catchment class D

30%

20%

10%

0%

Fig. 7.5 Frequency of occurrence of various infrastructure-component combinations appearing in all improvement scenarios showing relatively higher savings in d2d-GCs

on near-minimal d2d-GCs. As mentioned earlier in Table 7.5, the marked sidewalk facility can only be introduced in catchment types of class A due to the lateral-space constraints. All other infrastructure components of walk-access facility other than the marked sidewalk, as mentioned for catchment types of Class B could be considered for catchment type of class A. But in case of the catchment types of classes E and F, the options for improvement of sidewalk facility remain un-restricted on lateral sides. In these classes of catchment, the best possible sidewalk facility and its other associated components of walk infrastructure could be introduced. But feasibility for inclusions of any improvement measures under such cases needs to be examined by considering the pedestrian demand including the walk-accessed bus users using the facility.

7.5 Closure The propensity to access bus service by walk mode is realized to be bi-directional elastic—i.e., to both the walk-access infrastructure and the operational characteristics of bus service. In this regard, several initiatives of the Government of India, such as the Atal Mission for Rejuvenation and Urban Transformation (AMRUT) mission, SMART CITY mission, etc., emphasized on the need for integration between the walk-access mode and the urban public transport service such as city bus service. The SMART City mission even pays importance on the “first-and-last mile” connectivity

148

S. Roy and D. Basu

Table 7.9 Implementation Plan of Walk-access Infrastructure Components in Bus Stop Catchment Areas Class of catchment area Functional attribute

Infrastructure components

B

Marked pedestrian right-of-way (PRoW), which is at-grade with the carriageway, and surfacing of PRoW is color-coded (i.e., green marked) and having illuminated (i.e. cat’s eye) corridor

Type of sidewalk

Lighting and security Street light is present in the form of bollard light at every 5–10 m and has medium to high lighting intensity, but no electronic security and surveillance system

C

Type of crosswalk

Crosswalk is provided at the intersection of the access road and main bus corridor, and also at every 100–200 m of distance along the access road

Type of sidewalk

Raised pedestrian sidewalk having paved surface but without buffer zone from the vehicular traffic

Lighting and security Street light is present in the form of bollard light at every 5–10 m and has medium to high lighting intensity, but no electronic security and surveillance system

D

Type of crosswalk

Crosswalk is provided at the intersection of the access road and main bus corridor, and also at every 100–200 m of distance along the access road

Type of sidewalk

Raised pedestrian sidewalk having paved surface but having 0.5 m buffer from the vehicular traffic

Lighting and security Street light is present in form of bollard light at every 5–10 m and has medium to high lighting intensity and a CCTV camera is installed at every 25 m along the access road and at all major intersections within of the catchment area Type of crosswalk

Table top crosswalk along with the rumble surface acting as traffic-calming measures and in-pavement flashers acting as a warning signal is provided at the intersection of the access road and main bus corridor, and also at every 100–200 m of distance along the access road

7 Planning for Suitable Walk-Access Infrastructure Components …

149

as a part of the overall improvement of transit oriented development (TOD) with convenient accessibility. Many civic authorities in developed countries have a set of guidelines for achieving the “first-and-last mile” connectivity, such attempt in the context of urban India has only been limited to the proposition of generic walk facility guidelines. The contemporary design guideline of best practices in India does not specifically incorporate planning and integration of walk-access infrastructure with the urban bus corridor. In light of such requirements, the article presents guidance by suggesting a set of plausible combinations of walk-access infrastructure components in urban bus stop catchment areas. While presenting such guidance, the work not only overcomes the shortcomings usually noticed in the existing design guidelines, but also suggests different combinations of walk-access infrastructure components for different classes of bus stop catchment areas. The investigation takes into account the choice perspective of the walk-accessed bus users and the saving in door-to-door (d2d) generalized-cost (GC) incurred by walk-accessed bus users is considered as the measure of effectiveness (MoE) for selection of suitable combination of walk-infrastructure components. The guidance suggested in this article is developed based on the bus stop catchment classification as proposed in the work of Roy and Basu (2020a). As a future direction of research, a cost–benefit analysis of the selected walkinfrastructure components could be carried out under various demands of pedestrians in order to ascertain their economic as well as their financial feasibility.

References AASHTO (2004) Guide for the planning, design, and operation of pedestrian facilities. National cooperative highway research program (NCHRP), Transportation research board, USA Cheranchery MF, Maitra B (2018) Investigating perception of captive and choice riders for formulating service standards of ordinary and premium buses in Indian cities. Transp Policy 72:89–96 ChoiceMetrics (2018) Ngene 1.2 user manual & reference guide: the cutting edge in experimental design, Choice Metrics Pty Ltd. http://www.choice-metrics.com/download.html#manual. Accessed 20 Dec 2018 CoLAPW (2010) LA Lights Strategic Plan, City of LA, Public Works, LA, USA. http://bsl.lacity. org/downloads/strategic_plan.pdf. Accessed 5 Jan 21 Dandapat S, Maitra B (2020) Preference heterogeneity in trip makers’ perception and policy issues: a study with reference to bus services in Kolkata. Case Stud Transp Policy 8(4):1504–1517 Department of Transportation, State of Hawaii (2013) Hawaii pedestrian toolbox: a guide for planning, design, operations and education to enhance pedestrian travel in Hawaii Department of Transport, Western Australia (2016) Planning and designing for pedestrians: guidelines, Department of Planning, Public Transport Authority, WA, Australia Directorate of Urban Land Transport (DULT) (2014) Guidelines for planning & implementation of pedestrian infrastructure, ver 1.0, Govt. of Karnataka DoIT, Taipei City Government (2012) Holistic, multifaceted, and citizen-centric smart taipei strategies, handbook of smart cities. Springer Hensher DA, Rose JM, Rose JM, Greene WH (2015) Applied choice analysis: a primer, 2nd edn. Cambridge University Press, Cambridge, UK

150

S. Roy and D. Basu

IRC103:2012 (2012) Guidelines for pedestrian facilities, Indian roads congress Jiang Y, Zegras P, Mehndiratta S (2012) Walk the line: station context, corridor type and bus rapid transit walk access in Jinan, China. J Transp Geogr 20(1):1–14 KMOCT (2001) Korean highway capacity manual, Korea Ministry of Construction and Transportation, Gwacheoun-si Maitra B, Sadhukhan S (2013) Urban public transportation system in the context of climate change mitigation: emerging Issues and research Needs in India. In: Khare A, Beckman T (eds) Mitigating climate change. Springer environmental science and engineering, pp 75–91 Martin D, Jordan H, Roderick H (2008) Taking the bus: incorporating public transport timetable data into health care accessibility modelling. Environ Plan A 40:2510–2525 MoUD (2015a) Atal mission for rejuvenation and urban transformation: mission statement and guidelines, ministry of urban development (currently, Ministry of Housing and Urban Affairs), government of India MoUD (2015b) Smart city mission transform nation: mission statement and guidelines, ministry of urban development (currently, Ministry of Housing and Urban Affairs), government of India. Rahul TM, Verma A (2014) A study of acceptable trip distances using walking and cycling in Bangalore. J Transp Geogr 38:106–113 Rastogi R, Rao KVK (2003) Travel characteristics of commuters accessing transit: case study. J Transp Eng 129(6):684–694 Rastogi R (2010) Willingness to shift to walking or bicycling to access suburban rail: case study of Mumbai, India. J Urban Plan Dev. 3–10 https://doi.org/10.1061/(ASCE)0733-9488(2010)136: 1(3) Roy S, Basu D (2017) An approach of examining service condition of sidewalk facility in urban area. In: Proceedings of the Eastern Asia society of transportation studies (EASTS), vol 11. ISSN 1881-1132 Roy S, Basu D (2019) Ranking urban catchment areas as per service condition of walk-environment. J Transp Eng Part A Syst 145(4):04019005–1 to 04019005–14 Roy S, Basu D (2020a) Selection of intervention areas for improving travel condition of walkaccessed bus users with a focus on their accessibility: an experience in Bhubaneswar. Transp Policy 96:29–39 Roy S, Basu D (2020b) An evaluation of in-service infrastructural facilities of walk-access feeder paths to urban local bus stops. Transp Res Procedia 48:3824–3831 Roy S, Basu D (2020c) Classification of urban bus stop catchments for selecting appropriate sidewalk facility on access roads. Curr Sci 119(2):364–373 Sadhukhan S, Banerjee UK, Maitra B (2015) Commuters’ perception towards transfer facility attributes in and around metro stations: experience in Kolkata. J Urban Plan Dev 141(4):04014038 Sadhukhan S, Banerjee UK, Maitra B (2016) Commuters’ willingness-to-pay for improvement of transfer facilities in and around metro stations–a case study in Kolkata. Transp Res Part A Policy Pract 92:43–58 Southworth M (2005) Designing the walkable city. J Urban Plan Dev 131(4):246–257 TRB TC (2003) Quality of service manual: TCRP Report 100. Transportation Research Board, Washington, DC Tilahun N, Li M (2015) Walking access to transit stations: evaluating barriers using stated preference. In: TRB 94th annual meeting, 15–5667 Train KE (2009) Discrete choice methods with simulation. Cambridge University Press, Cambridge, UK Unified Traffic and Transportation Infrastructure (Planning & Engineering) Centre (UTTIPEC), Pedestrian Design Guidelines: “Don’t drive…walk”, 2009, Delhi Development Authority, New Delhi Woldeamanuel M, Cyganski R (2011) Factors affecting travellers’ satisfaction with accessibility to public transportation, association for european transport and contributors, Glasgow, Scotland Zhu X, Liu S (2004) Analysis of the impact of the MRT system on accessibility in Singapore using an integrated GIS tool. J Transp Geogr 12(2):89–101

Chapter 8

Crowd Management Guidelines for Mass Religious Gatherings Ashish Verma, Harihara Subramanian Gayathri, P. S. Karthika, Nipun Choubey, and Tarun Khandelwal

8.1 Introduction With increased mobility and willingness to travel, people’s knowledge and expectations from religious trips are changing. Not only from a religious viewpoint, but many people also participate in such gatherings to explore the events, which leads to an increase in the number of participants every year across the globe. Five million estimated people participated in Allahabad Kumbh 1954, whereas 75 million estimated people participated in Ujjain Kumbh 2016. The participants in Ijtema, annual Islamic gathering has also increased from a few thousand in Bhopal 1949 gathering to above 4 million in Aurangabad 2018 gathering. Similarly, the number of participants in the Hajj pilgrimage has increased from 1.8 million in 1996 to 2.5 million in 2019. Due to the increasing numbers, the host cities of such gatherings are under tremendous pressure to manage the influx of millions of people. Also, mass religious gatherings vary from other types of gatherings for the following reasons: (a) the crowd is made up of pilgrims and devotees from both rural and urban areas with some proportion of tourists, thereby constituting a heterogeneous mix (b) participants in religious gatherings are highly spiritual and are emotionally driven, which can get aggravated due to situational influences, leading to disrupted movements and chaotic situations. Inappropriate or poorly managed crowd control measures for mass religious gatherings may promote risky situations rather than preventing them. This requires a renewed understanding of crowd behavior in mass religious gatherings and framing guidelines for event planners and managers for better management of such religious events in the future.

A. Verma (B) · H. S. Gayathri · P. S. Karthika · N. Choubey · T. Khandelwal Transportation Systems Engineering (TSE), Department of Civil Engineering, Indian Institute of Science (IISc), Bengaluru 560012, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_8

151

152

A. Verma et al.

India, the land of festivals, hosts many mass religious gatherings, pilgrimages, and processions such as Kumbh Mela, Durga puja, Sabarimala, etc. Nearly 79% of the stampedes in India are from various religious gatherings and pilgrimages only (Illiyas et al. 2013). There have been a number of stampede incidents at the same venues, indicating that proper crowd risk management measures should be implemented to avoid potential incidents. Between 2001 and 2014, the country experienced 3,126 stampedes, with 2,421 people killed. (“Not Just Varanasi, Stampede Deaths in India Etch a Sorry Tale”, NCRB, 18 October 2016). However, the scientific reasons behind these incidents are not clear. The Kumbh Mela is the world’s largest Hindu pilgrimage, with millions of people gathering to worship and bathe in a holy river. The dip, people say, cleanses their sins and leads to redemption. The Kumbh Mela is held in four locations throughout India: Haridwar, Allahabad, Nashik, and Ujjain, in order of rotation, with the dip taking place in the Ganges (the confluence of the Ganges, Yamuna, and mythical Saraswati; Godavari; and Kshipra, respectively. Kumbh Melas usually occur once every twelve years in any given place. The Kumbh Mela attracts visitors from all over India as well as the world. They take a bath in the river, worship in temples, and perform several religious and non-religious activities. The event is unique because of the heterogeneous crowd—holy men (more assertive and aggressive), rural (might be illiterate and may not be able to read and understand signs and warnings), urban (may have adequate information about the places), and foreign population (adventure-seeking participants). Also, the participants can be goal-directed, goal-seeking, or goal-identifying. The Kumbh Mela includes a variety of events (such as visiting camps, engaging in spiritual discourses, participating in processions, visiting temples, and taking a dip in the river). Besides, many devotees from rural areas participate in Panchkroshi Yatra (religious walkathon) during Ujjain Kumbh Mela. Within five days, they walk from their villages to Ujjain, taking a 118 km circuitous route around the district, visiting numerous temples, and taking a dip in the Kshipra river. These unique aspects of Kumbh Mela are discussed in detail in Gayathri et al. (2017).

8.1.1 Need for Crowd Management Guidelines A large quantum of work has been done to understand and model the various facets of crowd behavior. These models take inspiration from theories in different fields such as physics, sociology, psychology, applied mathematics, computer science, ecological sciences, etc. Despite the popularity of these crowd models, they are yet to be widely used in practice to manage large-scale events. Broadly, crowd modeling can be done at different levels: microscopic modeling, mesoscopic modeling, and macroscopic modeling. Microscopic models focus on the individual pedestrian and capture pedestrian– pedestrian interaction and pedestrian–environment interaction. At every instant, a pedestrian’s movement is defined by microscopic parameters such as position, velocity, and acceleration. One of the most widely used microscopic models is the

8 Crowd Management Guidelines for Mass Religious Gatherings

153

social-force model (Helbing and Molnar 1995). It models the movement of a pedestrian as governed by a mixture of physical and socio-psychological factors. Three kinds of social forces are considered—attractive force toward destination, repulsive force due to other pedestrians and obstacles/boundaries, and attractive forces toward other objects/pedestrians. Another popular approach is based on cellular automata (CA) (Blue and Adler 2000, Burstedde et al. 2001). In a CA model, the entire space is discretized into a lattice, where each cell can be occupied by at most one pedestrian. A uniform set of rules applied to a cell computes the state of that cell as a function of its previous state and the state of its adjacent cells. However, the CA models often become complex, with many rules deciding how the movement should be performed. In addition to these models, optimal steps model, walking speed-based models, benefit–cost model, magnetic force model, and swarm models have also been proposed. Macroscopic models treat the pedestrian flow without distinguishing between individual pedestrians and their interactions, considering the crowd as a continuum medium. These models aim to capture the pedestrian flow in terms of aggregate measures such as density (ped/sq.m), flux (ped/hr), and average velocity (m/s) of pedestrians. Fundamental flow diagrams representing the relationship between flux and density are an important component of these models. Hydrodynamic models consider pedestrian flow as analogous to fluid flow and describe pedestrian dynamics using speed distribution (Hoogendoorn and Bovy 2002). Helbing’s fluid dynamics model (Helbing, 1992) was developed to study the collective movement of pedestrians under low densities, explaining the phenomena of walking behavior and propagation of waves. Daamen-Hoogendoorn-Bovy’s first-order flow model is based on flow-density diagram (Daamen et al. 2005). Colombo-Rosini model is an evacuation model developed based on Lighthill–Whitham–Richards (LWR) model of vehicular traffic (Colombo et al. 2010). In addition to these models, Bruno’s physical model, Hughes’s model, and several other models have been proposed. Mesoscopic models consider pedestrians as individuals, but aggregate relationships are used to characterize their propagation. A network-based pedestrian flow model was proposed by Lovas (1994), where the propagation of walking pedestrians in a network of walking sections is described using an empirical density-speed relationship. In an extended static floor field model, the movement of individuals is extended to the movement of pedestrian flow between cells, and the probabilistic approach used in the CA models for the movement of each pedestrian to its neighboring cells is replaced by probability distribution in this model (Shi et al. 2018). Apart from these models, modified Hughes’ continuum model, Kinetic theory-based models, and a few other models have been proposed. These different crowd models have their advantages and disadvantages, and the application of these models should take into account the limitations of the adopted model. The need of the hour is to bridge the gap between the research on crowd behavior and its practical application in crowd management. Crowd management for mass gatherings should contain a continuous evaluation process of facilities, collecting information from various sources and disseminating correct information, assessing the crowd behavior, continuously monitoring

154

A. Verma et al.

the crowd for any disturbances, and reaching out to the people at the right time. Berlonghi et al. (1995) defines it as the process of facilitating all activities and people so that they are being entertained or able to celebrate some occasion. Crowd monitoring can be made easier by using visual sensors such as cameras, and other data sources, apart from control personnel. The data obtained from these sensors can then be used to extract crowd features such as crowd density using object detection and tracking. However, crowd behavior is governed by the psychological and emotional state of the crowd as well. Wijermans et al. (2016) gives a detailed review of how events are organized and executed currently in two phases—the event preparation phase, and the event execution phase. Perez and Zeadally (2019) classify the various crowd management approaches into three categories—modeling and simulationbased, infrastructure-based, and crowd-based approaches. As mentioned before, it is possible to model the possible scenarios and plan strategies using the first approach. However, the issues in validation of the models might lead to unrealistic results. Computer vision tools and image processing techniques might be used in the second approach to extract crowd features to monitor the crowd. The third approach includes the use of mobile sensing systems such as smartphones. To manage large crowds, it is essential to understand the type of the event, activities, heterogeneity in participants, weather conditions, event schedules, attributes, etc. It is also crucial to comprehend the crowd’s emotions (Zeith et al. 2009). Managing the crowd should be done in three phases: before the event (planning phase), during the event (built-up phase), and after the event (break-up phase). Many studies have suggested different approaches to manage large crowds (Franke et al. 2015; Taneja and Bolia et al. 2018; Zhao et al. 2020). However, as suggested by the National Disaster Management Authority (NDMA), an integrated and structured approach to crowd management at mass gatherings is essential to prevent man-made disasters (NDMA 2014). Therefore, the objective of this white paper is to assess the effectiveness of the crowd management measures taken by the event organizers presently and formulate crowd management guidelines for mass religious gatherings based on the empirical studies conducted on the crowd dynamics at Kumbh Mela 2016, Ujjain, India. The crowd management guidelines are suggested threefold: from the perspective of event planning, facility planning, and crowd control and safety, before and during the event. It is to be noted that prior to the event (planning phase) and during the event, dissemination of proper information is important. From the study conducted on service quality dimensions, it was found that service quality is an antecedent of satisfaction. Hence, the focus should be on providing and improving the services that include information on public amenities, safety of personal belongings, and local facilities like public transport, shelter to rest, etc. In addition, partial information about activities and facilities would adversely affect the crowd psychology and anxiety levels. This influence can be negated through efficient communication of instructions to the pilgrims and prevent crowd risk situations. The results from the empirical studies and analysis and the observations made in the field during Kumbh Mela 2016 are used as guiding principles to formulate the crowd management guidelines for mass religious gatherings. This can guide the

8 Crowd Management Guidelines for Mass Religious Gatherings

155

event organizers and administrators of any such religious events to manage such large crowds in the future.

8.2 Crowd Management Guidelines Crowd dynamics in Kumbh Mela were studied in different levels of detail from macroscopic, microscopic, and activity-based travel behavior, and video processing perspectives. Figure 8.1 summarizes the various studies done under each level. It is suggested to follow crowd management guidelines from the perspective of crowd control and safety, facility planning, and event planning. These guidelines are presented in the following sections.

8.2.1 Crowd Control and Safety General guidelines 1. 2.

To maintain crowd safety anywhere in the event, the critical densities should not exceed 6–7 persons per square meter. Barricades should be robust. As shown in Fig. 8.2, local turbulence because of side breaching and side friction induced by counter-movements was observed in Ujjain Kumbh Mela. These potentially risky situations arose due to improper barricading.

Fig. 8.1 Studies conducted on crowd dynamics

156

A. Verma et al.

a. Local turbulence created by side breaching

Fig. 8.2 Undesirable situations observed in Kumbh Mela 2016, Ujjain

8 Crowd Management Guidelines for Mass Religious Gatherings

157

Fig. 8.3 Schematic of funnel-shaped bottleneck

3.

4.

5. 6.

7.

8.

Pedestrian inflow is affected by holy dip duration, which in turn influences the satisfaction of pilgrims. The inflow into the ghat should be regulated by the authorities such that the satisfaction level of pilgrims increases and the critical density is not exceeded. The movement and behavior of the crowd should be constantly monitored. To connect with the crowd, loudspeakers or some other type of public announcement device can be used. CCTV surveillance of the crowd along procession routes should be carried out in order to detect potential hazards and take corrective measures. It is recommended to place an obstacle/column near an exit. This induces selforganizing behavior within the crowd and eases its exit movement (Helbing and Molnar 1998). Stopping and releasing pilgrims at suitable intervals can regulate the flow downstream. Simultaneously, proper control measures should be in place as the people start moving at higher speeds when the queue is released suddenly. An irregular geometric section can lead to bottleneck conditions. However, it is suggested to improve the flow at a bottleneck by providing a funnel-shaped section instead of a sudden change in the section width, as illustrated in Fig. 8.3. Event-specific guidelines Procession. The following points should be taken care of during the procession.

1. 2.

3.

4.

The Akharas should notify the authorities of the anticipated number of people who will be participating in processions on their behalf. With these figures in mind, the flow should be handled so that there is no chaos. For, e.g., near the ghat entrance in the 2016 Kumbh Mela, the saints were allowed first and pilgrims were denied entry. As a result, there was an unstable equilibrium and circumstances where people were at risk of being trampled. The organizers should provide participants with a thorough briefing on the route, the ascetic order that they must obey (ash-smeared sadhus, saints in tractordrawn chariots, pilgrims), and the order in which the camps enter the procession well before the processions begin. Soft boundaries or temporary barricades should be avoided on procession routes because they can lead to dangerous situations. Re-routing or other methods should be used to keep the pilgrims going as much as possible so that the traffic does not stop. This will assist in preventing circumstances that could result in a crowd crush.

158

5.

6.

7.

8.

A. Verma et al.

Only akhara members should be permitted to participate in processions. One idea is to give the participants tickets or badges so that others do not want to join the crowd. Others should not be able to enter the procession or cross over it. It is possible to prevent the creation of serpentine chains as a result of physical and psychological forces in this way. The organizers must ensure that the processions begin promptly. In the Kumbh Mela procession-2016, people began acting violently as a result of the delayed start, attempting to drive each other out of their respective camps. If a delay is inevitable, people should be given accurate information. The flow of processions should be regulated locally by security personnel. With the assistance of security forces, scenarios where people from another camp attempt to break the flow and overtake should not be permitted. In order to prevent the rise in crowd anxiety and impatience among the waiting pilgrims leading to a possible situation of crowd risk, they could be engaged through spiritual music, chanting, and hymns.

Holy dip. To organize the flow of crowd in the Ghat region, it is recommended to lead the crowd to the Ghat through parallel narrow and zig-zag channels instead of a single free passage. It is observed from Kumbh Mela that free passages lead to the movement of pilgrims in random directions, causing chaos and inefficient use of the facility. Figure 8.4 illustrates two possible barricading arrangements for a Ghat.

8.2.2 Event Planning 1.

2.

3.

4.

The expectations for hospitality are not high in mass religious gatherings. Improving the quality of services slightly will lead to better pilgrim satisfaction and perceived overall experience of the event. Our studies reveal that service quality is an antecedent of satisfaction. Hence, the focus should be on providing and improving the services that include information on public amenities, personal belongings, and local facilities like public transport, shelter to rest, etc. While emphasizing on level of satisfaction, event organizers must investigate the experience of pilgrims as well as ensure arrangements are made for all the devotees to take a holy dip and have proper worship. Ensuring and enhancing these discussed parameters will lead to better management and smooth functioning from a management point of view. Event organizers should keep in mind that the perception of a positive crowd experience can go a long way in managing unsocial behavior within the crowd. This can be achieved by augmenting the participant’s experience through various measures such as safer pilgrimage, comfortable resting places, adhering to schedules, etc., thereby, increasing the overall satisfaction. All the necessary information should be readily accessible to the pilgrims. Organizers should use different media for information dissemination. Information on personal security and safety, popular destinations, transportation facilities,

8 Crowd Management Guidelines for Mass Religious Gatherings

159

a. Separate entry and exit points for pilgrims

b. Same entry and exit point for pilgrims Fig. 8.4 Possible barricading arrangements at Ghat

5.

lockers and cloakrooms, and basic necessities would improve the service quality at the event and the experience and satisfaction of pilgrims. Figure 8.5 indicates various types of information and sources where they could be made available. Cumulative effects of the stochastic nature of barricading, partial information about activities and facilities, time constraints, etc. adversely impact crowd

160

A. Verma et al.

Fig. 8.5 The figure indicates various types of information and information sources. The types of information provided by a source are marked by respective numbers

6.

7.

8.

psychology and anxiety levels. This influence can be negated through the efficient dissemination of information among the pilgrims. In this way, crowd risk situations can be prevented. Surveillance systems including CCTV’s and drones used in Kumbh Mela should be positioned ideally to ensure that automated crowd counting, and crowd risk situations at macroscopic and microscopic levels can be understood better. This would enable not only to get real-time information with clarity, but also to further analyze such crowd situations. The quality of video data recorded through CCTV or drones should be checked through pre-event trials. The video should be of consistent quality across time of the day. The altitude of operation of a drone should also be considered before its selection. Before employing a drone or CCTV cameras for surveillance and video data collection, quality checks should be carried out. Figure 8.6 shows an example of the difference in the quality of video data obtained during Kumbh Mela.

8.2.3 Facility Planning 1.

Facilities that are strategically placed can help minimize the discomfort level and improve satisfaction. Activity peak hour times, duration, and locations should be

8 Crowd Management Guidelines for Mass Religious Gatherings

161

Fig. 8.6 The difference in quality of video snapshots from CCTV and GoPro for the same location. The top image has been captured from CCTV footage, while the bottom image is from GoPro videos

2.

identified to employ resources optimally. Emergency services like fire service, hospitals, disaster management teams should be deputed at such times, for the appropriate duration at these locations. These facilities should be resilient to situations of extreme crowd presence. Walking is observed to be the predominant mode of travel for the pilgrims in Kumbh Mela. Therefore, it is necessary to consider the accessibility of various facilities by walk-in such mass gatherings, to maximize the efficiency of facilities. These facilities include drinking water stalls, toilets, recreation centers, etc. Due to the spread of the Mela region over a large area, it is important that the facilities be optimally located. The total accessibility of the entire area should be maximized by distributing the facilities such that they are spread over the entire region rather than being in clusters. Other factors such as the importance

162

3.

4.

A. Verma et al.

of a location and certain social constraints may also influence the location of facilities. Organizers should concentrate on the participants’ specific interests and needs, such as cleanliness and hygiene, sleeping arrangements, and escape from hot weather. Our studies also reveal that the experience of pilgrims at halts influenced their overall experience. Thus, setting up halts with the expected level of service at regular intervals during the journey of pedestrians is important. Following are some recommendations with respect to acceptable trip distance (by walk) based on the analysis of Ujjain Kumbh Mela data: a.

b.

c.

d.

e.

Food stalls may be set up within a 450 m walking distance of primary activities such as riverbanks (for a “holy dip”) and temples (for a “visiting temple” activity). Arranging food stalls at an acceptable distance, as determined by the study, would greatly reduce the amount of effort required. Drinking water is usually given free of charge to pilgrims via governmentrun stalls. These drinking water facilities could be established within a 650 m walking distance of the primary activity sites. The participants were willing to walk 900 m to access points for the primary activity “holy dip.“ As a result, it would be preferable if public transportation was available up to that point on the periphery. This will significantly reduce pilgrims’ effort. Fig. 8.7 depicts the acceptable trip distance for various facilities and important locations. Facility planning should be done keeping these distances in consideration. Beyond the acceptable trip distances by walk, alternate arrangements for travel such as mass transport facilities should be provided.

8.3 Conclusion and Future Scope This white paper presents key findings of the research on crowd dynamics in Kumbh Mela held at Ujjain in 2016. The crowd management guidelines are formulated based on the results of empirical studies and the observations made on the field. The guidelines are suggested in threefold: from the perspective of event planning, facility planning, and crowd control and safety, before and during the event. Information dissemination is critical both before (during the planning phase) and during the event. It was noted that the pilgrims experiences at halts influenced their overall experience. It was also observed from the study on service quality dimension that service quality is a predator of satisfaction and so the focus should be on enhancing the services such as information on public amenities, safety of personal belongings, and local facilities like public transport, shelter to rest, etc. Considerable thought must go into the planning of various facilities such as drinking water stalls, resting places, ATMs, etc., to ensure that these facilities are readily accessible by the pilgrims. To ensure crowd safety at bottleneck sections, a funnel-shaped

8 Crowd Management Guidelines for Mass Religious Gatherings

163

Fig. 8.7 Acceptable trip distances

geometry is suggested instead of a sudden change in width by improving the pedestrian flow. Furthermore, to ensure emergency services like fire service, hospitals, the disaster management teams should be deputed during activity peak hour times for an appropriate duration at these locations. It is to be understood that every mass religious gathering will have unique features specific to the gathering in terms of location, activities, and crowd characteristics. Depending on the crowd numbers in the previous years, the authorities plan facilities to manage the crowd efficiently. However, during the event, they monitor the crowd, assess the crowd situation, and on-the-spot decisions are made to ensure crowd safety. These might include regulating the crowd flow using barricades, ensuring minimal waiting times, and maintaining order by offering advice and direction to people. Nevertheless, the effectiveness of these strategies is not evaluated post-event. The field observations and empirical studies on crowd flow and the analysis of the crowd’s perception based on questionnaire surveys are used to arrive at the crowd management guidelines specific to mass religious gatherings. The results of this paper can be used by event planners as a decision-making method when planning, managing, and improving crowd safety at future mass religious gatherings.

164

A. Verma et al.

Though deploying emergency services like fire services, ambulance, etc., have been discussed to some extent, further studies need to be conducted to come up with emergency preparedness and evacuation plans. Crowd dynamics are influenced significantly by the psychology of the participants of such mass religious gatherings. To further understand this perception, the qualitative and quantitative effects of such “spiritual motivation” should be explored in depth by conducting a comparative study of mass religious gatherings and any other mass events that are not religious in nature. Conducting a questionnaire survey supplemented by video evidence, which can help capture the motivation aspect, could be a future objective. The guidelines proposed in this paper are specific to mass religious gatherings based on the inferences from Ujjain Kumbh Mela 2016. However, these guidelines have to be implemented keeping in mind the stochasticity of the situations in similar gatherings. Acknowledgements The work reported in this paper is part of the project titled The Kumbh Mela Experiment: Measuring and Understanding the Dynamics of Mankind’s largest crowd, funded by the Ministry of Electronics and IT, Government of India (MITO-0105), Netherlands Organization for Scientific Research, NWO (Project no. 629.002.202), and Robert Bosch Center for Cyber Physical Systems, Indian Institute of Science, Bangalore (Grant No. RBCO001). The authors also express their gratitude toward the Kumbh Mela administration and government of Madhya Pradesh, India, for providing constant support and official permissions to carry out research work and establish Indo-Dutch collaborative research camp at Kumbh Mela 2016.

References Berlonghi AE (1995) Understanding and planning for different spectator crowds. Safety Sci 18(4):239–247. Elsevier Blue VJ, Adler JL (2000) Modeling four-directional pedestrian flows. Transp Res Record 1710(1): 20–7. Sage Journals Burstedde C, Klauck K, Schadschneider A, Zittartz J (2001) Simulation of pedestrian dynamics using a two-dimensional cellular automaton. Physica A: Stat Mech Appl 295(3–4):507–25. Elsevier Colombo RM, Goatin P, Maternini G, Rosini MD (2010) Macroscopic models for pedestrian flows. InBig events and transport: the transportation requirements for the management of large scale events. HAL-Inria Daamen W, Hoogendoorn SP, Bovy PH (2005) First-order pedestrian traffic flow theory. Transp Res Record 1934(1):43–52. Sage Journals Franke T, Negele S, Kampis G, Lukowicz P (2015) Leveraging human mobility in smartphone based ad-hoc information distribution in crowd management scenarios. In: 2015 2nd international conference on information and communication technologies for disaster management (ICT-DM). IEEE Xplore, pp 27–34 Gayathri H, Aparna PM, Verma A (2017) A review of studies on understanding crowd dynamics in the context of crowd safety in mass religious gatherings. Int J Disaster Risk Reduction 25:82–91. Elsevier Helbing D, Molnar P (1998) Self-organization phenomena in pedestrian crowds. arXiv preprint cond-mat/9806152. arXiv labs Helbing D, Molnar P (1995) Social force model for pedestrian dynamics. Phys Rev E 51(5). APS Physics

8 Crowd Management Guidelines for Mass Religious Gatherings

165

Hoogendoorn SP, Bovy PH.: Normative pedestrian behaviour theory and modelling. In Transportation and traffic theory in the 21st century. Emerald Group Publishing Limited (2002) Illiyas FT, Mani SK, Pradeepkumar AP, Mohan K (2013) Human stampedes during religious festivals: A comparative review of mass gathering emergencies in India. Int J Disaster Risk Reduction 5:10–8. Elsevier Løvås GG (1994) Modeling and simulation of pedestrian traffic flow. Transp Res Part B: Methodol 28(6): 429–43. Elsevier National Disaster Management Authority: Annual Report, Government of India (2014) Perez AJ, Zeadally S (2019) A communication architecture for crowd management in emergency and disruptive scenarios. IEEE Commun Mag 57(4):54–60 Shi M, Lee EW, Ma Y (2018) A novel grid-based mesoscopic model for evacuation dynamics. Physica A: Stat Mech Appl 497:198–210. Elsevier Taneja L, Bolia NB (2018) Network redesign for efficient crowd flow and evacuation. Appl Math Model 53:251–266. Elsevier Wijermans N, Conrado C, van Steen M, Martella C, Li J (2106) A landscape of crowd-management support: an integrative approach. Safety Sci 86:142–64. Elsevier Zeitz KM, Tan HM, Grief M, Couns PC, Zeitz CJ (2009) Crowd behavior at mass gatherings: a literature review. Prehospital Disaster Med 24(1): 32–8. Cambridge University Press Zhao H, Thrash T, Kapadia M, Wolff K, Hölscher C, Helbing D, Schinazi VR (2020) Assessing crowd management strategies for the 2010 Love Parade disaster using computer simulations and virtual reality. J Royal Soc Interface 17(167). The Royal Society Publishing

Chapter 9

A New Framework for Comprehensive Mobility Plans in India Vajjarapu Harsha and Ashish Verma

9.1 Introduction Cities are the top centers for economic development, and there is a massive influx of people from neighboring places (Vajjarapu and Verma 2021). Alongside economic growth, there are multiple problems associated with it. These problems include population and vehicle growth, pollution, congestion, urban sprawl, and reduced natural vegetation and water bodies. These are threatening the sustainability and liveability of the cities. Due to increased imperviousness, a small amount of rainfall leads to flooding conditions, thereby making cities less resilient (Vajjarapu and Verma 2021). The transportation system plays a significant role in our day-to-day lives. It is also one of those sectors that primarily contribute to emissions from urban areas and gets affected due to urban flooding. In India, under the National Urban Transport Policy (NUTP), cities should submit a comprehensive mobility plan (CMP) to receive grants for improving the urban transportation system. Since the CMP serves as a future guiding document for the transport sector development, it should encompass all the elements that will enhance sustainable mobility (TERI 2011). This is increasingly important as many Indian cities see increased traffic congestion, increased vehicular emissions, longer travel times, and increased personal vehicle usage with declining public transport and NMT usage. Therefore, the urban transportation system’s sustainability should meet the social, economic, and environmental goals to attain sustainable mobility. Additionally, cities are now being bombarded with sudden and heavy rainfalls disrupting the transportation system. This leads to congestion and longer travel times and distances and, V. Harsha · A. Verma (B) Department of Civil Engineering, Indian Institute of Science (IISc), Bangalore, India e-mail: [email protected] V. Harsha e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_9

167

168

V. Harsha and A. Verma

therefore, requires the transportation system to be resilient to these effects. However, this issue is not addressed by the CMPs. Consequently, it is critical to incorporate these measures into the CMP framework. The future CMPs should consist of a comprehensive set of Sustainable Urban Transport Measures (SUTMs). In addition to congestion, the SUTMs’ main aim should be to reduce emissions from the urban transportation sector and improve the urban transportation sector’s resiliency to urban floods. For this study, CMPs from various cities across India, such as Bengaluru, Chennai, Delhi, Mumbai, Kolkata, Kochi, Jaipur, Pune, and Surat, are reviewed. It is observed that most of the CMPs focus mainly on improving the challenges associated with accessibility, but less importance is given to emissions reduction and resiliency to urban floods. Also, the policies discussed in the CMP’s are not comprehensive to reduce the urban traffic demand. Besides improving the public transportation system, the usage of alternate fuels also plays an essential role. But, most of the CMPs hardly discuss the use of alternate fuels to reduce vehicular emissions. The very few CMPs such as Chennai addressed vehicular emissions but considered only CO2 pollutant. Besides, multiple pollutants are harmful to the human body, which are neglected. The policy packages mentioned in CMP will have a disproportionate effect on various income–gender groups. The policy packages proposed in the CMPs do not highlight this equity aspect, thereby putting an additional burden on the disadvantaged groups. This study proposes an integrated framework to evaluate multiple sustainable urban transport policy bundles. These policy bundles aim to reduce congestion and vehicular exhaust emissions and improve the urban transportation system’s resiliency to urban flooding. This framework is tested on the Bangalore Metropolitan Region (BMR). The sustainable urban transport policies (SUTPs) are segregated into two groups; emissions reduction and improving resiliency. The SUTPs focused on emissions reduction are combined and formed into four policy bundles. The SUTPs to improve resiliency are made into three policy bundles. With this background, the objectives of this study are as follows. 1. 2. 3.

To propose a new framework that can be incorporated in the future CMPs for cities. To present a step-by-step guidelines on how to evaluate the proposed framework. Evaluating the proposed framework to understand its impact on congestion, vehicular emissions, and urban transportation system’s resiliency to urban flooding.

9.2 Methodology The methodological framework that can be adopted by the future CMPs in evaluating sustainable urban transport measures is shown in Fig. 9.1.

9 A New Framework for Comprehensive Mobility Plans …

169

Sustainable Urban Transport Measures

Urban Transport Emissions

Urban Transport Resilience

Evaluating the base line conditions under BAU Base year Scenario Forecasting the BAU scenario to target years (2030 & 2050)

Formulating & evaluating policy bundles to reduce emissions

Carbon emissions intensity

Formulating & evaluating policy bundles to improve resilience

Finalizing best resilience policy bundle

Consumer surplus Finalizing best emissions policy bundle

END OF CMP FRAMEWORK

Segregating trips gender wise and income wise

Estimating avg. trip length and VKT for gender and income

Emissions reduction from income and gender groups across policy bundles & BAU

Estimating total Percapita emissions of income and gender groups across policy bundles & BAU Fig. 9.1 A proposed overall framework for the future Comprehensive Mobility Plans (CMPs)

9.2.1 Urban Transport Emissions Step 1—Evaluating the base year’s conditions under the BAU Scenario: The emissions from the urban transport sector have a long-term effect. Therefore, it is essential to anticipate future consequences and develop appropriate SUTMs to

170

V. Harsha and A. Verma

reduce emissions. It requires long-term data forecasting to evaluate future scenarios and to develop appropriate measures. A proper and reasonable long-term data forecasting should have robust baseline data. The socio-economic trends, the transportation trends, and the transportation sector emissions of the base year are used to assess the baseline condition. The traditional classic four-stage travel demand model (TDM) is considered for modeling the transportation parameters. The base year data should be validated before forecasting for future years. Step 2—Forecasting the BAU scenario to target years: The validated base year data is used to forecast the target years. The changes in land use, future transportation trends, population, and employment changes should be incorporated while forecasting. Step 3—Scenario Formulation: Scenarios play an essential role in understanding future transportation conditions and aid in proper planning. Since the SUTMs focus on improving sustainability, it considers two scenarios (a) Business as usual (BAU) scenario and (b) Sustainable urban transport scenarios. The BAU scenario forecasts the baseline conditions for the future years without any emissions reduction policy intervention. The sustainable urban transport scenarios in this section focus on developing and evaluating sustainable urban transport policies to reduce congestion and emissions. Step 4—Sustainable Urban Transport Policies to reduce emissions: The critical missing factor in all the CMPs is stakeholders’ inclusion in policy planning. Due to its inherent interdisciplinary, it is crucial to involve the public and experts from various fields relevant to transportation in policy planning. The opinions of the people and experts are sought using the Delphi technique. To start with the Delphi process, the sustainable urban transport policies that result in congestion and emissions reduction should be listed and mixed to form policy bundles. Each policy bundle should essentially aim at reducing emissions and congestion from the transportation sector. These policy bundles should be circulated among the stakeholders and asked to score. The policies with high scores will be retained, and the policies with low scores will be removed from the policy bundles. After multiple Delphi rounds, the emissions-reducing policy bundles should be finalized for evaluation. Step 5—Scenario analysis using Travel Demand Modeling: This step deals with evaluating the BAU and sustainable urban transport scenarios. The evaluation process is done in three segments (a) Emissions Estimation, (b) Carbon Emissions Intensity, and (c) Consumer Surplus. Emissions Estimation: Through travel demand modeling, vehicle hours traveled for each policy bundle and the BAU scenario are estimated. Further, the transportation emissions are calculated using the VKT obtained from the model and each mode’s emission factor value. The transportation sector emissions, in this case, are the tailpipe emissions and are estimated using Eq. 9.1 (Verma et al. 2018). Emissions (gm) = Vehicle kilometers traveled(km) × Emission Factor (gm/km) (9.1)

9 A New Framework for Comprehensive Mobility Plans …

171

Carbon Emissions Intensity: The emission intensity is the level of GHG emissions per unit of economic activity. It is estimated by calculating the volume of emissions per unit of the gross domestic product. Countries worldwide express their emissions reduction in terms of carbon emissions intensity. The carbon emission intensity is determined using Eq. 9.2 (Verma et al. 2018). Emission intensity = CO2 emmisions/GDP

(9.2)

Consumer Surplus: The consumer surplus is defined as the difference between the consumers’ willingness to pay in terms of travel cost and travel time and what they pay. In this study, consumer surplus is calculated using the Rule of Half (RoH), which estimates the change in consumer surplus for small supply changes with a constant demand curve (Ma et al. 2015). Evaluating consumer surplus will provide an insight into how much each mode users are gaining and losing economically upon implementing sustainable urban transport policy bundles. Step 6—Appraisal of strategies: Each evaluated travel parameter from the sustainable urban transport policy bundles is compared with the BAU scenario. The policy bundle, which shows more significant emissions reduction, will be the best sustainable urban transport policy bundle. Each of these policy bundles has a different level of impact on the overall transportation system. Therefore, appraisal of strategies helps the policymakers understand how these policies affect travel characteristics, the linkage between economic activity and emissions, and the financial gains and losses during commuting. This will enhance the judgment of the policymakers in modifying these policies to attain better sustainability. Step 7—Finalizing the best sustainable urban transport strategy: This is the final step in the framework. After comparing various sustainable urban transport policy bundles with the BAU scenario, the best sustainable urban transport policy bundle is the one that gives a maximum reduction in actual emissions, carbon emissions intensity, and higher economic gain. To develop appropriate equity-induced mitigation policies, there is a need for a robust underlying background on which the judgments can rely. However, there are hardly any studies on emissions contributed by different income–gender groups. Therefore, the policymakers must understand these inequalities and reframe sustainable urban transport policies. This leads to the next step, which shows how the emissions from various income–gender groups are estimated. Step 8—Income and Gender equity: The entire dataset should be segregated as per the gender groups and income levels to start with this process. The World Bank organization proposes four income groups: low-income, lower-middle, uppermiddle, and high-income groups. The available data should be segregated into these four income levels for each gender separately. The policy bundles are evaluated across each income–gender group using travel demand modeling, and VKT is estimated. Next, emissions of each income–gender group for each mode are estimated using Eq. 9.1. The total emissions are then converted to Per capita emissions to understand the per person’s emissions contribution. Finally, the Per capita emissions of all

172

V. Harsha and A. Verma

income–gender groups from the SUTPBs are compared with the BAU scenario to understand each group’s emissions reduction potential. This entire step will provide the policymakers with a clear picture of how sustainable urban transport policy bundles affect each income–gender group. Based on these observations, the policymakers can modify the policy bundles to reduce the inequalities.

9.2.2 Urban Transport Resilience One of the widely used synonyms to represent cities is ‘concrete jungle’. The concretization of massive vegetation land reduced the cities ability to infiltrate rainwater. There are many instances where sudden and heavy rainfall causes the cities to get flooded, known as urban flooding (Vajjarapu and Verma 2021). Urban flooding directly impacts the urban transportation sector. The impacts are seen in road closures, increased travel times, congestion, longer trip distances, and slower speeds. Some trips get canceled when the flood levels are significantly high. Since the transportation sector is considered the backbone of any country’s economy, it must be resilient to urban flooding. However, these critical factors are not addressed in the CMPs. Incorporating urban transport resiliency in the CMP framework will minimize the urban transportation problems during flooding. The steps involved in assessing the adaptation policy bundles are as follows. Step 1—Evaluating the base year’s conditions under the BAU Scenario: The first step in urban transport resilience is similar to that of step 1 in urban transport emissions but with an additional BAU scenario in the base year. This leads to two scenarios in the base year BAU scenario; BAU-No flooding and BAU-flooding. The BAU-No flooding scenario, in this case, is the BAU scenario of the emissions estimation process. The BAU-flooding scenario is the BAU-No flooding scenario with the flood impact. To evaluate the BAU-Flooding scenario, the output from the BAU-No flooding scenario plays an important role. The road network from the model output of BAU-No flooding scenario contains attributes like traffic volume, travel time, travel speed, link length, and link capacity. This road network is overlaid with the flood map and digital elevation map (DEM). The data on the flood level and elevation at each link are extracted. Since the road network is flooded, the network attributes such as travel time and travel speed will change based on the flood depth. Once these two attributes are updated, the model is run to estimate the traffic volume on each link on the network. This provides the BAU-flooding road network under the BAU-flooding scenario for the base year. Step 2—Forecasting the BAU scenario to target years: As mentioned in step 2 of the emissions estimation, the data is forecasted for the target years. In the target years, the flood’s impact on the urban transportation system is evaluated similarly, as mentioned in the above step. Step 3—Sustainable Urban Transport Policies to improve resilience to urban flooding: The vulnerable locations are identified as those links that are low-lying,

9 A New Framework for Comprehensive Mobility Plans …

173

high traffic volume, and high flood depth. Since the vulnerability assessment is done, the urban transport resilience policies should be framed to reduce these links’ vulnerability. Similar to emissions estimation, the Delphi study is carried out to form the urban transport policy bundles. These urban transport resilience policy bundles should not just aim to reduce the flood depth. They should also focus on reducing the impact of urban flooding on people living in low-lying areas. Step 4—Evaluating the Urban Transport Resilience Policy bundles: Since the urban transport resilience policies reduce the flood depth, it leads to new travel times, travel distances, and speeds. These attributes should be updated in the travel demand model. Further, any changes in trip production and trip attraction due to trips getting canceled or shifted should be incorporated in the model. The travel demand model simulates the new travel parameters due to these policy bundles for future years. The updated travel demand model will provide information on VKT, vehicle hours traveled (VHT), average trip length (ATL), the average speed of the vehicle (ASV), and trips canceled. The travel parameters should be compared between BAU-no flooding, BAU-flooding, and urban transport resilience policy bundles for the future years. Step 5—Finalizing the best Urban Transport Resilience Policy bundle: The travel parameters obtained from the urban transport resilience policy bundles are compared with the BAU-flooding scenario. The policy bundle that best reduces the VKT, average trip length, increased vehicle speed, and minimum trips canceled from the BAU-flooding scenario will be the policy bundle that showed the best resilience.

9.3 Case Study of Bengaluru Metropolitan Region (BMR) The frameworks shown in the earlier sections are tested on Bengaluru Metropolitan Region (BMR), which is the case study area. BMR is spread across 8005 sq. km and is rapidly urbanizing. There is a swift growth in the city’s urbanization in the past four decades (Ramachandra et al. 2014) and 90.94% of its population living in urban localities (Eswar and Roy 2018). The extraordinary growth in urban sprawl and economy increased the proportion of private vehicles leading to chaotic traffic conditions. The city’s traffic issues are varied, extending from longer travel times to increasing vehicular emissions. The travel speeds during peak hours go as low as 17 kmph (CTTS report 2010). The increased built-up area and inadequate and ill-maintained drainage system caused many urban flooding events in the city. These reasons make BMR the perfect area to implement the proposed frameworks. The base year data for this study is obtained from CTTS (2010). The base year BAU data is validated in screen line analysis, mode share, and average trip length. The validated data is then forecasted to 2030 and 2050. All the proposed and accepted government projects related to transportation are considered in modeling the transportation conditions for 2030 and 2050. For urban transport emissions policy bundles evaluation, the change in share of five pollutants such as CO2 , PM2.5 , HC, NOx, and CO from transport emissions are estimated for the BAU scenario and sustainable transport scenarios. Based on the

174 Table 9.1 Sustainable transport policy bundles to reduce emissions from urban transport in Bengaluru (Verma et al. 2018)

V. Harsha and A. Verma Policies under bundle 1 (B1) Increasing the network coverage of bus transport Providing infrastructure for active transport Levying an extra tax on the vehicle purchase Policies under bundle 2 (B2) Levying an extra tax on the vehicle purchase Improving emission standards of vehicles Hiking fuel prices Policies under bundle 3 (B3) Levying an extra tax on the vehicle purchase Implementing no car roads Pricing for causing congestion Provisions for parking and riding Providing infrastructure for active transport Providing separate lanes for high occupancy vehicles and promoting carpooling Mixed building with high density along the main corridors Policies under bundle 4 (B4) Whole bundle 3 set along with an assumption of electrifying all cars and buses

proposed framework, carbon emissions intensity and consumer surplus of different travel modes are also evaluated. Thus, the stakeholders or policymakers can have a clear insight into economically profitable sustainable transport measures. The policy bundles to reduce emissions from the urban transport sector for this study are shown in Table 9.1. The future of electric vehicles penetration and electricity generation is uncertain. To incorporate this uncertainty, electric vehicles’ emissions factors are considered based on four different energy mix scenarios (Sharma and Chandel 2020), as shown below. • Scenario 1 (S1): New policies scenario (IEA 2015)—Non-renewable and renewable sources (74%-26%). • Scenario 2 (S2): Electricity generation purely from non-renewable sources (100%). • Scenario 3 (S3): Equal share of electricity generation from renewable and nonrenewable sources (50%–50%). • Scenario 4 (S4): Electricity generation purely from renewable sources (100%).

9 A New Framework for Comprehensive Mobility Plans …

175

VKT Comparison (in million km)

B1

B2

B3 & B4

61.2

45.4

52

2050 37.8

54.9

41.2

56.5

42.1

2030

BAU

Fig. 9.2 VKT comparison across all scenarios

9.3.1 Results and Discussion Urban Transport Emissions: This study assumes that all buses and cars’ electrification will not impact the mode share. However, they differ in emissions due to different emission factors for fossil fuel-based vehicles and electric vehicles. Therefore, for B3 and B4, vehicle kilometers traveled (VKT) are shown together, and the emissions are presented separately. The four-step travel demand modeling is done to estimate the VKT for all the sustainable transport policy bundles and BAU scenarios. The comparison of the same is shown in Fig. 9.2. From Fig. 9.2, it is seen that bundle 3 is showing the maximum VKT reduction from the BAU scenario for 2030 and 2050. As shown in Table 9.1, bundle 3 is a more comprehensive set of policies that aim at encouraging sustainable transportation modes like NMT and public transportation while discouraging personal transportation. Due to the restrictions on personal vehicle usage, the mode shift occurs from private to public transportation. It should be noted that public transportation mode has high occupancy levels. Therefore, when the mode shift happens from private to public transport, it leads to a substantial reduction in VKT. Now that the VKT are estimated for all policy bundles from travel demand modeling, the next step is to assess the emissions. The study formulated four sustainable transport policy bundles and four scenarios of energy mixes of electricity generation. Therefore, combining these two will give 16 scenarios that are compared with the BAU scenario. The total vehicular emissions of pollutants CO2 , PM2.5 , HC, NOx, and CO for all the sustainable transport policy bundles and BAU scenario are estimated using Eq. 9.1. Figure 9.3 shows the total vehicular emissions of CO2 and PM2.5 . In both cases, it is evident that bundle 4, with the combination of energy mix scenario 4, is proven to show the maximum emissions reduction. This is because bundle 4 (bundle 3) has the lowest VKT and scenario 4 of the energy mix has the low emissions factors due to electricity completely generated from renewable sources. A similar analysis is done for the rest of the pollutants. In all the cases, B4-S4 shows the best emissions reduction from BAU, and the results are shown in Table 9.2.

176

V. Harsha and A. Verma

Total CO2 Emissions (Tonnes/Yr)

1702506

6964265

2050

1618627 5920412 1659844 7281310 1574998 4689560 1490151 2097807 1590860 6019250 1615732 7443445 1541995 4731149 1446967 2018849 1580296 6802334 1639648 8543017 1517469 5227988 1395290 1912955 1803847 7234387 2269955 9679332 1310446 5023054 350912 366777

2030

B1 - B1 - B1 - B1 - B2 - B2 - B2 - B2 - B3 - B3 - B3 - B3 - B4 - B4 - B4 - B4 - BAU S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4

279.52

690.00 143.00 438.00 131.00 187.00 240.78 699.86

2050

175.00

156.00 461.00 160.00 564.00 152.00 367.00 143.00 171.00 152.00 475.00 164.00 583.00 147.00 377.00 138.00 171.00 149.00 558.00

2030

881.86 198.72 531.31 83.60 184.54 158.00 560.00

Total PM2.5 Emissions (Tonnes/Yr)

B1 - B1 - B1 - B1 - B2 - B2 - B2 - B2 - B3 - B3 - B3 - B3 - B4 - B4 - B4 - B4 - BAU S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4

Fig. 9.3 Total vehicular emissions for all policy bundles

Table 9.2 Percentage reduction in emissions with B4-S4 scenario from the BAU scenario Percentage reduction in emissions with B4-S4 from BAU Year/pollutant

CO

HC

NOx

CO2

PM2.5

2030

46

48

57

80

60

2050

53

70

66

94

72

The carbon emissions intensity for all the policy bundles is estimated using Eq. 9.2 and presented in Fig. 9.4. In this case, it is seen that the B4-S4 combination is proving the best reduction when compared with the BAU scenario. Based on the framework shown in Fig. 9.1, the next step is to estimate the consumer surplus to understand the economic impact of the sustainable transport policies. This study considered the rule of half theory for calculating the consumer surplus for private and public transportation. The analysis showed that, upon implementing the sustainable transport policy bundles, users who continue to use personal vehicles and autorickshaws tend to lose money. In contrast, public transport and NMT users save money. However, the differences between these two groups still give a monetary gain to the entire transportation system. The summary of economic gains for all the policy bundles is presented in Table 9.3.

9 A New Framework for Comprehensive Mobility Plans …

177

Carbon emission Intensity (Tonnes of CO 2 /Rs. Millions) for 2030 & 2050 2050

0.46 0.63 0.47 0.78 0.45 0.50 0.42 0.22 0.45 0.64 0.46 0.80 0.44 0.51 0.41 0.22 0.45 0.73 0.46 0.91 0.43 0.56 0.39 0.20 0.51 0.77 0.64 1.04 0.37 0.54 0.10 0.04 0.47 0.74

2030

B1 - B1 - B1 - B1 - B2 - B2 - B2 - B2 - B3 - B3 - B3 - B3 - B4 - B4 - B4 - B4 - BAU S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4

Fig. 9.4 Carbon emissions intensity compared between policy bundles and BAU

Table 9.3 Monetary gains from sustainable transport policy bundles from BAU scenario

Policy Bundle

CS—Net Value—INR in Millions (Gains/losses) 2030

2050

Bundle 1

1.89

9.38

Bundle 2

1.85

10.06

Bundle 3 (Bundle 4)

2.76

12.80

The next step is to understand how these sustainable transport policies impact the income and gender groups. The income levels are categorized into four tiers based on the world bank classification combined with male and female groups giving eight income–gender groups. The emissions reduction is estimated for all the income– gender groups for all sustainable transport policies. In this chapter, only CO2 pollutant is considered to estimate the emissions reduction from the BAU scenario for 2030. Also, energy mix scenario 4 is considered for electric vehicles emissions calculation. Figure 9.5 shows the total per capita CO2 emissions of all income–gender groups for 2030. But how to understand this plot? Fig. 9.5 for the male group signifies that the high-income male group exhibits high emissions due to car dependence. The highest percentage reduction from BAU to B4 is shown by the lower-middle-income male. This is because the lower-middle-income group is sensitive to additional travel costs. The added costs through mitigation policy bundles are forcing them to shift from two-wheeler to metro, which is the best alternative. Since metro occupancy is high and runs on electricity, the VKT is reduced and also emissions. For instance, if the emissions reduction is compared between BAU and B3, high-income male shows more reduction. This is because the B3 contains policies that are restricting car users in different ways. Since high-income males depend on the car for their travel, B3 forces them to shift to the metro, thereby reducing emissions. In this way, the policymaker can understand which policy affects which group, and necessary changes can be incorporated based on the requirement.

178

V. Harsha and A. Verma

Total CO 2 Emissions Percapita (Kilotonnes/person/yr) - Male - 2030

LOW

LOWER MID

B2

B3

B4

UPPER MID

44.9 42.1 41.7 38.5 15.6

B1

41.5 40.4 39.3 38.6 15.4

12.5 10.6 9.8 9.5 8.2

36.7 35.4 34.8 33.8 12.6

BAU

HIGH

Total CO2 Emissions Percapita (Kilotonnes/person/yr) - Female - 2030

LOWER MID

B4

UPPER MID

31.4 31.2 30.4 28.0

B3

9.6

9.2

8.1 7.7 7.6 7.5 6.3 LOW

B2

35.0 32.8 31.3 30.9 11.9

B1

28.5 26.4 26.2 25.9

BAU

HIGH

Fig. 9.5 Total per capita CO2 emissions of income groups for male and female across all bundles— 2030

Urban Transport Resilience: Similar to urban transport emissions reduction policies, resilience policies are also formulated based on the Delphi study. The final policies are combined into three policy bundles and are shown in Table 9.4. These Table 9.4 Sustainable Urban Transport Policies to improve resiliency (Vajjarapu et al. 2020)

Policies under bundle 1 (B1) Replacement of impermeable road surface with permeable material in vulnerable areas Slum relocation and rehabilitation Providing proper drainage facilities at vulnerable areas Construction of redundant infrastructure Policies under bundle 2 (B2) Re-routing people during flooding Restricting development in low-lying or vulnerable areas Slum relocation and rehabilitation Policies under bundle 3 (B3) Replacement of impervious surfaces with permeable material in vulnerable areas Providing proper drainage facilities at vulnerable areas Re-routing people during flooding

9 A New Framework for Comprehensive Mobility Plans …

179

Table 9.5 Comparison of travel parameters between BAU and urban transport resilience policy bundles (Vajjarapu et al. 2020) Travel parameter

Year

BAU-no flooding

BAU-flooding

B1

B2

B3

Vehicle Kilometers Traveled (in mil kms)

2030

45.4

48.2

46

47.1

46.5

2050

61.2

64.4

62.1

63.5

62.9

Average speed of vehicle (in kmph)

2030

25.4

16.8

21.3

18.7

20.2

2050

23.6

13.2

19.1

14.3

17.6

Avg. Trip Length—Private Transport (in Km)

2030

17.4

20.8

18

19.8

19.3

2050

18.2

21.7

19.1

20.8

20.2

Avg. Trip Length—Public Transport (in Km)

2030

16.4

19.2

17

18.7

17.9

2050

17.7

20.4

18.3

19.8

19.4

Vehicle Hours Traveled (VHT) (in mil hr)

2030

9.5

11.3

10.3

11.2

10.7

2050

16.6

18.9

17.4

18.3

17.8

Percentage Trips canceled

2030

NA

3.3

0

3

1.7

2050

NA

3.9

0

3.5

1.9

policies aim to reduce the road network’s flood levels and improve urban transportation’s resiliency to flooding. Each of these policy bundles will affect the travel demand model process at various stages. The new VKT, VHT, ASV, and trips canceled are estimated based on the reduced flood depth. A study by Pregnolato et al. (2017) presented a relation between the flood depth and vehicle speeds. This study is used to estimate the new travel speeds and other parameters. The urban transport resilience policies are evaluated in VKT, VHT, ATL, ASV, and percentage trips canceled and compared with the two BAU scenarios; No flooding and flooding. The results are shown in Table 9.5. Table 9.5 shows that bundle 1 showed better resilience compared with bundles 2 and 3 in all aspects. Commuters usually take the shortest path to reach their destination. The flooding might have blocked access to these shortest paths, making commuters choose an alternative path. This path is usually longer than their usual route. The implementation of urban flood resiliency policy bundles resulted in flood level reduction and opened up some shortest paths. Hence, the reduced VKT, VHT, ATL, and the low flood levels increase ASV. Additionally, having a redundant infrastructure will aid in making a trip by providing them alternative routes. Therefore, in bundle 1, there are no trips that get canceled upon this bundle 1 implementation due to its comprehensiveness.

180

V. Harsha and A. Verma

9.4 Summary Most of the existing CMP’s focus primarily on reducing congestion, ignoring other important factors related to urban transportation such as emissions and resiliency. The measures taken for congestion reduction are also not comprehensive to solve the problem. This study presents a framework to address these existing problems in the current CMPs. The proposed framework integrates a complete set of policies that address urban transportation’s congestion, emissions, and resiliency to urban flooding. The proposed framework is implemented on Bangalore Metropolitan Region’s urban transportation system. Four sustainable urban transport policy bundles to reduce emissions and three sustainable urban transport policies to improve resilience are considered for this purpose. Among the emissions-related policy bundles, bundle 4 with a combination of energy mix scenario 4 showed the maximum emissions reduction, and bundle 3 showed maximum reduction in VKT and congestion. This is because of comprehensiveness in the policy bundles aimed at improving public transportation and discouraging private transport. Similarly, bundle 3 (bundle 4) showed the best carbon emissions intensity and consumer surplus results. The framework also allows the policymakers to understand the impact of these policy bundles on income–gender groups. The emissions reduction across eight income–gender groups due to these policy bundles is also presented. Further, the urban transportation system’s resiliency to urban flooding is evaluated in VKT, VHT, ASV, ATL, and the percentage of trips canceled. Bundle 1, a strategic combination of land use and infrastructure policies, showed a better improvement in resiliency among the policy bundles aimed to improve resiliency to urban flooding. Due to the robustness and the ability to address multiple elements of the urban transportation system at once, this framework will aid the policymakers in their decision-making and policy formulation. Incorporating this framework in the future CMPs will improve the cities’ liveability, sustainability, and resilience.

9.5 Future Scope for Research In this study, the urban transport emissions did not capture the impact of electrification of cars and buses on mode share. Understanding and quantifying this impact will aid in appropriate planning for electric vehicle penetration into the Indian market. Further, there are challenges faced by developing economies in implementing electric mobility. The primary challenges can be the affordability to own an electric vehicle and the provision of charging infrastructure. The study on electric mobility is still in the early stages in developing economies. It has a vast scope for future research. Rainfall and urban floods have an impact on the mode share. However, this study assumed that there will be no change in mode share under these circumstances. This assumption was considered based on the uncertainties of sudden rainfall events.

9 A New Framework for Comprehensive Mobility Plans …

181

Therefore, seasonal-based mode choice models can be explored to better understand the flood effect of mode share. This helps the city to respond better to urban floods and improve resiliency.

References Ma S, Kockelman KM, Fagnant DJ (2015) Welfare analysis using logsum differences versus rule of half: Series of case studies. Transp Res Record 2530(1):73–83. Sage Journals Pregnolato M, Ford A, Glenis V, Wilkinson S, Dawson R (2017) Impact of climate change on disruption to urban transport networks from pluvial flooding. J Infrast Syst 23(4):1–13. Ascelibrary Ramachandra TV, Aithal BH, Vinay S, Rao GR, Kulkarni F, Tara NM, Nagar N (2014) Trees of Bengaluru, ENVIS Technical Report 75, ENVIS, Centre for Ecological Sciences, IISc, Bangalore 560012 Sharma I, Chandel MK (2020) Will electric vehicles (EVs) be less polluting than conventional automobiles under Indian city conditions? Case Stud Transp Policy 8(4): 1489–1503. Elsevier TERI: Review of Comprehensive Mobility Plans New Delhi: The Energy and Resources Institute (2011) 240 Vajjarapu H, Verma A (2021) Composite adaptability index to evaluate climate change adaptation policies for urban transport. Int J Disaster Risk Reduction 58(102205). Elsevier Vajjarapu H, Verma A, Allirani H (2020) Evaluating climate change adaptation policies for urban transportation in India. Int J Disaster Risk Reduction 47(101528). Elsevier Verma A, Harsha V, Hemanthini AR (2018) Sustainable transport measures for liveable Bengaluru. Project Sub Report, IISc Bangalore, India Wilbur Smith Associates: Comprehensive Traffic and Transportation Study (CTTS) for Bangalore Metropolitan Region (2010)

Chapter 10

Transit-Oriented Development (TOD) as a Sustainable Transport Strategy for Metropolitan Cities Manoranjan Parida, Phani Kumar Patnala, Robert Hrelja, and Ravi Sekhar Chalumuri

10.1 Introduction Transit-oriented development (TOD) is a highly dense, diverse, pedestrian-friendly, accessible, and affordable urban structure within proximity to transit facilities. It embraces the idea that locating the right urban development around transit facilities promotes sustainable transportation, i.e., the use of transit, walking, and cycling (Arrington 2008; Litman 2019). A successful TOD at the metropolitan level is inclusive and is often an effective way to create land values at the neighborhood level. It offers increased accessibility to a greater number of amenities (Higgins and Kanaroglu 2018). TOD creates a “legible” place, evokes the spirit of enclosure, human linkage, and better quality of life. It endorses “public health”, “social equity” (Appleyard et al. 2019), and “positive” travel behavior, i.e., reduction in vehicle kilometers traveled (VKT), private motorization, and traffic congestion in urban neighborhoods. Delivering positive outcomes of TOD in metropolitan cities is not easy and necessitates well-balanced urban design decisions. This transportation planning strategy M. Parida (B) Centre for Transportation Systems, Indian Institute of Technology Roorkee, Roorkee, Uttarakhand 247667, India e-mail: [email protected] R. Hrelja Faculty of Culture and Society, Department of Urban Studies, Malmo University, Malmo 20506, Sweden P. K. Patnala Department of Civil Engineering, University of Manitoba, Winnipeg, Manitoba R3T2N2, Canada R. S. Chalumuri Transportation Planning and Environment Division, CSIR—Central Road Research Institute, New Delhi 110025, India © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_10

183

184

M. Parida et al.

embraces an enormous promise for enhancing sustainability, augmenting economic development, and developing a socially inclusive environment. If implemented properly, TOD—particularly the physical integration of urban structure and transit investments—can alleviate ecological outcomes that enhance economic sustainability and help address the most critical urban challenges in the developing world: reducing poverty and increasing livability. Strengthening the economic well-being of the urban population (rich and poor) in metropolitan cities opens the gateway to a prosperous future. Hence, this chapter focuses on what planning parameters that need special attention to alleviate TOD as a sustainable transport strategy for mitigating urban problems in metropolitan cities, by taking Delhi as a suitable case study.

10.2 TOD Definitions Since 1990s, numerous cities worldwide implemented and established TOD as a sustainable transport strategy to withstand urban problems. It portrays the next evolutionary stage of American suburbs (Carlton 2009). The concept drew on earlier planning concepts such as “Garden City” of the United Kingdom, “Finger Plan” of Copenhagen, Denmark, “Compact City” of Europe, “Rosario Concept” of Korea, but presented in a contemporary way (Yang and Pojani 2017). So far, debates surrounding the definition of TOD are extensive and multifaceted. Existing descriptions of different researchers and organizations highlight distinctive emphasis following their interests, objectives, and policies. Some of these context-sensitive descriptions of TOD are: TOD is a concept, mixed-use community, centered around a transit station that invites residents, workers, and shoppers to drive their cars less and ride mass transit more, roughly at a quarter mile from a transit station. —Cervero and Kockelman (1997). A foundational planning concept that integrates the potential of coordinating land-use planning, transportation system design and infrastructure investments, as an integrated process with people and public transit. —MoUD India (2016).

Figure 10.1 reveals the evolution of TOD concept, starting from 1947 to the present. In line with the general perception of TOD, researchers recognized that TOD can take a variety of forms (Atkinson-Palombo and Kuby 2011) and never a “one-sizefits-all” approach. The argument is still pertaining to the absence of universal definition for TOD and the concept is still reforming. However, what is clear from previous research is that TOD planning must foster TOD projects that are as best adapted to the urban context as possible (Scherrer 2019). This chapter, therefore further, specifies the adaptive elements of urban structure that may guide TOD planning and decision-making in Indian metropolitan cities.

10 Transit-Oriented Development (TOD) as a Sustainable …

185

Fig. 10.1 Timeline of TOD concept

10.3 TOD Policy for Indian Cities A growing number of policies in Indian cities like Delhi, Bengaluru, Mumbai, Naya Raipur provide guidance, direction, and technical support for creating TODs at neighborhood levels. For instance, government agencies of Delhi have proposed a TOD policy (2015) for neighborhoods closer to Metro rail transit stations (MoUD 2016). According to this policy, a TOD influence zone may engage the private sector into cross-subsidizing and providing various public amenities, higher affordable housing stock, and high-quality public transport (WRI Annual and Report 2019). This draft policy intends to develop TODs from centerline to periphery of Delhi Metro corridors in three levels: (i) 300 m, (ii) 800 m, and (iii) 2000 m as intense, standard, and transition zones, respectively. Table 10.1 illustrates the development practices for TODs in Table 10.1 Proposed FARs and density for TOD in Delhi (Source DDA 2021) Gross FAR (site)

Net FAR (block)

Minimum permissible density (with 10% variation) (DU/sq.km) High-Residential (FAR ≥ 50%)

Less residential (FAR ≤ 30%)

Below 1.0

Below 2.0

Underutilization of FAR not permitted

1.1–1.5

2.1–3.0

30,000

25,000

1.6–2.0

3.1–4.0

40,000

35,000

2.1–2.5

4.1–5.0

50,000

45,000

2.6–3.0

5.1–6.0

60,000

55,000

3.1–3.5

6.1–7.0

70,000

65,000

3.6–4.0

7.1–8.0

80,000

75,000

186

M. Parida et al.

Delhi. These influence zones have a minimum gross density of 25,000 dwelling units (DU)/sq.km constituting a minimum 30% residential, 30% commercial/institutional, and 15% retail of overall floor-area-ratio (FAR). As planned, TODs augment within intense and standard zones covering a total area of 665.1 sq.km, i.e., 41% of entire Delhi city (Suzuki et al. 2015). However, the proposed planning parameters (TOD demarcation and FARs) must be reformed through solid research support.

10.4 Demarcation of TOD Influence Zone Existing studies consider different boundaries to specify a TOD influence zone. For instance, Kamruzzaman et al. (2014) used 600 m to define TODs in Brisbane, Australia. Besides, Higgins and Kanaroglou (2018), and Chen et al. (2017) considered 800 m and 1000 m for TODs in Toronto and Shanghai cities, respectively. However, the commonly admitted distance to represent a TOD influence zone is truely context-sensitive (Singh et al. 2017), because it depends on the travel and socio-demographic characteristics of individuals (Kamruzzaman et al. 2014; Hoback and Anderson 2008). For example, Kumar et al. (2018) estimated the influence of TOD on the basis of access and egress trips of individuals using non-motorized transport modes such as walk, cycle, and cycle-rickshaw. The optimum distance of TOD influence zone was estimated using cumulative frequency curves of access and egress distances. This buffer distance was obtained at the intersection point of both distribution curves (Rastogi and Rao 2003). Based on access and egress trips of 5647 individuals in Delhi (see Fig. 10.2), Kumar et al. (2018) study suggested 1200 m as an appropriate distance around transit stations, where TODs need to be prioritized. Further, from their empirical analysis, Kumar et al. (2021) suggested TOD demarcation as 600 m for the intense zone and 1200 m for the standard zone.

10.5 TOD Typologies 10.5.1 Development of TOD Typology As mentioned earlier, researchers advocate that a “one-size-fits-all” approach is not suitable for TOD planning. TODs form in diverse ways and each neighborhood have distinct features within geographical contexts (Kumar et al. 2018). Careful understanding of existing urban structures within TOD neighborhoods will help planners and policymakers in planning, designing, and implementing TODs, effectively. Thus, neighborhoods with homogeneous urban structures are clustered into typologies. Each TOD type has specific population/employment densities, mix of land uses, walkability, street patterns, physical elements, and transit facilities. TOD typologies are helpful to government authorities creating a collective set of strategies to enhance

10 Transit-Oriented Development (TOD) as a Sustainable …

187

Fig. 10.2 Cumulative frequency curves for access and egress trips

the urban structure in neighborhoods (Zemp et al. 2011). These copious benefits of classifying neighborhoods have convinced public agencies worldwide in associating typologies in their TOD policy documents (Austin et al. 2010; FDOT 2011). TOD policy for Delhi suggests a typology of seven neighborhood types, namely intermodal gateways, employment centers, destination nodes, transit neighborhoods, urban core (CBD), infill neighborhoods, and new residential areas (MoUD 2016). However, this policy document needs to lay down appropriate levels of urban structure within TOD typology. For this purpose, Kumar et al. (2020a) identified six neighborhood types among 47 neighborhoods of Delhi in a quantitative manner. Table 10.2 illustrates the proposed TOD typology for Delhi city and their quantitative levels of urban structure. This quantitative classification will guide as a practical handbook for stakeholders while investing funds on conducive TOD types.

10.5.2 Recommended FARs for TOD Typology Among a proposed typology, application of TODs must be sensitive to existing levels of policy parameters such as floor-area-ratio (FARs), mix of uses, building heights, parking provisions, etc., in maximizing the potential of neighborhoods. Appropriate and allowable tuning of these policy parameters offers a resilient TOD planning approach. Table 10.3 provides an overview of proposed FARs for typology in Asian TOD cities. It is to be noted that the recommended FARs for Indian cities are inadequate as compared to other Asian cities (Suzuki et al. 2015). For instance, the allowable maximum FARs for TOD development in Delhi is only 4, whereas, it is

Low-dense Neighborhoods dedicated to religious, political, administrative, and industrial uses

Description

11,665–14,941

0.69–0.73

27–54

Population density (Persons/sq.km)

Entropy (0–1)

Intersection density (Nos/sq.km)

TOD profiles

Low-density

TOD typology

48–96

0.48–0.60

26,349–45,003

Large housing Neighborhoods that lie outside the city core

Urban residential

Table 10.2 Proposed TOD typology for Delhi

114–127

0.58–0.68

24,689–33,936

Posh neighborhoods with car-oriented streets

Affluent

114–142

0.64–0.68

36,104–46,884

Compact neighborhoods that serve as commercial centers inside city core

Urban commercial core

61–89

0.72–0.78

16,939–23,075

Moderate to high-dense neighborhoods that serve as economic, retail, and cultural centers inside city core

Urban mixed core

35–85 (continued)

0.63–0.76

16,559–25,791

Moderate to high-dense neighborhoods that lie close proximity to Metro

Transit

188 M. Parida et al.

Low-density

5–17

2350–2550

7193–10,825

TOD typology

Network density (km/sq.km)

Distance to transit (m)

Job access to public transport (within 45 min)

Table 10.2 (continued)

7163–9120

835–1527

14–29

Urban residential

12,750–14,206

630–1410

28–45

Affluent

20,249–24,926

477–982

5–12

Urban commercial core

11,716–14,562

785–1660

20–22

Urban mixed core

24,524–28,901

390–787

8–29

Transit

10 Transit-Oriented Development (TOD) as a Sustainable … 189

190

M. Parida et al.

Table 10.3 FARs for neighborhood types in Asian cities (Source Serge and Gerald 2017) Typology

Proposed FAR mix for Tokyo Delhi*

Singapore

Hong Kong

Seoul

Residential

55%R + 5% C + 10%Com + 30%Ret

–

1.4–11.2

–

1.0–7.0

Low-dense

–

0.5–2.0

–

0.2–3.0

1.0–7.0

Medium-dense

–

1.0–5.0

1.4–11.2

0.67–5.0

1.0–7.0

High-dense

–

1.0–5.0

1.4–11.2

6.5–10.0

1.0–7.0

Commercial

30%R + 50%C + 10%Com + 10%Ret

2–13.0

1.4–12.6

3.5–12.0

1.0–15.0

Industrial

30%R + 5%C + 10%Com + 55%I

1.0–4.0

1.0–3.5

1.0–12.0

1.0–4.0

Mixed use

30%R + 5%C + 10%Com + 55%Ret

1.0–4.0

1.4–25.0

–

1.0–7.0

Maximum

4

20

25

12

10

*R—Residential, C—Commercial, Com—Community, I—Industrial, Ret—Retail

25 in Singapore and 20 in Tokyo. However, stakeholders in Delhi still perceive that increasing FARs could worsen the already congested traffic conditions (Suzuki et al. 2013). At this moment, the Delhi government must propose new building laws that encourage multi-uses by allocating higher FARs to built-up areas, which can expedite market values and employment opportunities (Cho and Rodríguez 2014). Besides, higher FARs must pertain to ensure maximum potential from different neighborhood types in the application of TODs. Therefore, upcoming TOD projects require a stronger TOD policy that allocates appropriate FARs and land uses for typology close to transit facilities. Such a comprehensive TOD policy will act as a reference tool for stakeholders while developing TOD typologies in other metropolitan cities of India.

10.6 Measurement of TOD-Ness at Neighborhood Level TOD planning at a Greenfield space is very rare in metropolitan cities. Any neighborhood possess some elements of TOD-ness within. Literature suggests that there is a need to express existing levels of TOD-ness in terms of a “TOD index” or “score” (Renne and Bartholomew 2011). A TOD index or score is a tool that measures the degree to which an urban structure in specific neighborhood orients toward transit facilities (Xu et al. 2017). Several indicators that relate to urban structure need to be measured and aggregated into a TOD metric. This metric is relative in nature and helpful in comparing different neighborhoods based on existing conditions. Such a relative TOD measurement at the neighborhood level will identify strengths and

10 Transit-Oriented Development (TOD) as a Sustainable …

191

weaknesses in each neighborhood and help planners, policymakers, and researchers in making long-term investment plans, accordingly. The measurement of TOD-ness involves the following four stages: 1. 2. 3. 4.

Selection of potential sites/neighborhoods Establishing TOD criteria/indicators Assigning weights to TOD criteria/indicators Ranking of neighborhoods based on measurement tools

The following sub-sections provide a detailed description of each stage involved in the measurement of TOD-ness at neighborhood level.

10.6.1 Selection of Potential Neighborhoods The selection/designation of a TOD neighborhood entails more than just being in proximity to transit stations. A neighborhood is designated as a TOD based on its urban structure elements such as existing levels of density (residential or employment), distance to transit, mix of land uses, and available housing options near transit stations. There are often questions on what density, which mix of uses, and how far the distance to transit can be considered for a TOD. For instance, higher densities lead to higher transit ridership, but a minimum density threshold is required to support transit systems and allow developers to determine appropriate development in a TOD area. Stakeholders seldom debate on what density is enough for supporting TODs. A typical TOD neighborhood accomplishes at least 3000 dwellings/sq.km in residential areas and 12,000 jobs/sq.km in commercial centers closer to transit stations (Austin et al. 2010; Public Policy and Transit Oriented Development 1996). Besides, Calthorpe (1990) and Dittmar and Poticha (2004) suggest a minimum density threshold of 8100 persons/sq.km, and 15,000 persons/sq.km for TOD neighborhoods, respectively. In this line, Kumar et al. (2021) study recommends 10,000–40,000 persons/sq.km as density thresholds for establishing TODs in Delhi, India. Therefore, the selection of potential neighborhoods is a preliminary investigation of existing urban structures.

10.6.2 Establishing TOD Criteria/Indicators An extensive list of criteria is available to assess TOD-ness at the neighborhood level. Table 10.4 summarizes various criteria that researchers formulated for the measurement of TOD-ness in different case studies. These criteria are usually referred to as TOD dimensions. Some of them are grouped into six aspects, namely travel behavior, urban structure, social, environment, economic, and policy (Renne 2009). Some researchers measured TOD-ness based on “5Ds” of urban structure, as developed by Ewing et al. (2017). Some transportation agencies proposed criteria based

192

M. Parida et al.

Table 10.4 Established criteria for TOD measurement in different studies References Case study

Stakeholders

Planning criteria

Measurement

Wey et al. (2016)

Ankeng line Fuzzy-ANP of the New and GIS Taipei City Metro Transit

Methodology

Industrial expert, Government official, and researcher

Population density (0.162) Density of facilities (0.126) Pedestrian spaces (0.117) Environmental capacity (0.189) Mixed land-use (0.105) Open spaces (0.139) Capacity of floor area (0.060) Residence accessibility (0.046) Daily living safety (0.056)

Evaluating performance of metro stations using a TOD Index

Motieyan and Mesgari (2017)

Tehran, Iran Fuzzy-AHP

Regional officials

Density (0.474) Diversity (0.314) Design (0.145) Socio-economic (0.067)

TOD Index at Neighborhood level

Singh et al. 21 Train (2017) stations in the city region of Arnhem and Nijmegen, The Netherlands

Spatial Regional Multi-Criteria officials Analysis (SMCA) using GIS

Density (0.150) TOD Index at Land-use diversity the regional (0.030) level Walkability/Cyclability (0.060) Economic development (0.220) Capacity utilization (0.190) User-Friendliness (0.110) Accessibility (0.150) Parking at station (0.080)

Strong et al. (2017)

Colorado, US

AHP

Regional officials

Travel Behavior (0.15) Urban structure (0.48) Economics (0.13) Social Diversity (0.23)

TOD site selection decision framework

Sahu (2018)

5 BRTS stations of Naya Raipur, India

AHP and Genetic Algorithm (GA)

Researchers, Planners, and Engineer

Density (0.372) Diversity (0.333) Distance to transit (0.294)

Comparison of alternative TOD plans

(continued)

10 Transit-Oriented Development (TOD) as a Sustainable …

193

Table 10.4 (continued) References Case study Kumar et al. (2020b)

Methodology

Delhi, India Enhanced FAHP

Stakeholders

Planning criteria

Measurement

Planners, Policymakers, and Researchers

Balance (0.024) Design (0.176) Placemaking (0.010) Multimodality (0.308) Modal Shift (0.265) Network Connectivity (0.185) Livability (0.032)

Defining TOD for Indian context

Value in the parenthesis is the weights assigned to the criteria

on the policies of their jurisdictions (Thomas 2015). MoUD India has mentioned six criteria, namely balance, placemaking, design, modal shift, multimodality, and network connectivity, to assess TOD-ness at transit neighborhoods (MoUD 2016). Hence, the selection of suitable criteria/indicators for a TOD planning problem will be a context-sensitive and data-dependent task.

10.6.3 Assigning Weights to Planning Criteria It is evident from existing studies (see Table 10.4) that TOD planning involves criteria/indicators of multiple dimensions. Despite the existence of multi-criteria, evaluation of criteria/indicators tends to be a decision-making process among stakeholders such as planners, builders, policymakers, researchers, government officials, and the public agencies (Wey et al. 2016; Strong et al. 2017). Each decision maker carry distinct beliefs, knowledge, and ideas within a TOD project (Thomas 2015). The judgments on weighing/aggregating of planning criteria conflict within decision groups and become a critical decision-making process. Existing studies have employed different multi-criteria decision-making (MCDM) methods such as analytical hierarchy process (AHP), analytical network process (ANP), geographic information system (GIS), data envelopment analysis (DEA), and their fuzzy extensions for calculating weights to established criteria/indicators (Motieyan 2017; Kumar et al. 2020b; Evans and Pratt 2007). However, there is a need for involving multiple experts and account their complex, uncertain, vague, and fuzzy decisions in TOD planning processes.

10.6.4 Ranking of Neighborhoods In recent years, the use and popularity of TOD indices to quantify and measure TODness has gained significant attention among researchers worldwide. The neighborhoods that represent TOD nature are ranked on the basis of a TOD metric. Several

194

M. Parida et al.

indices such as TOD index by Evans and Pratt (2007), Singh et al. (2017), and Motieyan and Mesgari (2017), TOD score by Aston et al. (2016) and ITDP (2013), TOD degree by Papa and Bertolini (2015), and TOD levels by Schlossberg and Brown (2004) have been proposed. These metrics differ in terms of (i) type of data used (i.e., qualitative, quantitative or subjective); (ii) method used (e.g., spatial-based, Delphibased, and questionnaire-based); (iii) unit of analysis (neighborhoods, stations, sites, etc.); (iv) objectives framed (i.e., to evaluate Greenfield sites, brownfield sites, walkability, sustainability, etc.); and (v) criteria considered (sustainability, economic, contextual). However, there is a need to investigate the relation of TOD metrics with macro/micro policy parameters for making better decisions in TOD investments.

10.6.5 TOD-Ness in the Transit Neighborhoods of Delhi This section illustrates a decision framework for measuring TOD-ness at the neighborhood level. This methodological framework is applied to a case of 87 transit neighborhoods in Delhi. Table 10.5 presents the detailed list of established criteria/indicators. These criteria/indicators are also applicable for measuring TODness in neighborhoods of other Indian cities (Kumar et al. 2021). Weights to these criteria/indicators were estimated using Enhanced fuzzy-analytical hierarchical process (EFAHP) of 31 expert judgments. For a more detailed understanding on TOD criteria and their relative weights, see Kumar et al. (2020b). A composite TOD score was calculated for selected neighborhoods using Modelo Integrado de Valor para Estructuras Sostenibles (MIVES) methodology (2021). Similar to existing studies, these TOD scores are relative in nature and a higher TOD score indicates higher levels of TOD-ness. Figure 10.3 provides an overview of TOD scores calculated for selected neighborhoods across the Blue and Yellow line corridors of Delhi Metro. Using this measurement tool, each neighborhood was assigned a TOD score between 0 and 1. These scores illustrate how the selected transit neighborhoods can continue to be improved based on the existing TOD conditions. This study identified four TOD corridors based on clustering TOD scores in transit neighborhoods. The identified TOD corridors include:

10 Transit-Oriented Development (TOD) as a Sustainable …

195

Table 10.5 Criteria/indicators for measuring TOD-ness Criteria

Definition

Weights

Indicators

Weights

Balance

Optimized residential and employment densities closer to transit stations

0.024

Population density

0.005

Employment density

0.019

Placemaking

Mixed-use developments to create vibrant choices for a better quality of life

0.010

Entropy

0.010

Design

Quality and functionality of AT

0.176

PCA

0.035

Walk score

0.106

Intersection density

0.035

Modal shift

Living closer to transit facilities promote transit usage

0.265

Distance from transit

0.265

Multimodality

Alternative and highly accessible public modes to reach station areas

0.308

PTAL

0.277

Interconnectivity ratio

0.031

Well-connected transit stations to serve travel demand for all destination choices

0.185

Parking area

0.093

Network density

0.093

A social-friendly environment that ensures affordable housing for all

0.032

Network connectivity

Livability

* weights

1. 2. 3. 4.

HH income

0.019

Vehicle ownership

0.013

obtained from Kumar et al. (2020b)

Corridor I: INA-Saket Corridor II: Nawada-Ramesh Nagar Corridor III: Rajendra Place-Indraprastha Corridor IV: Udyog Bhawan-Kashmere Gate

Further, a correspondence analysis was conducted to illustrate the association of TOD corridors with policy parameters like FAR levels and building uses. Figure 10.4 demonstrates the complementarity among transit neighborhoods in terms of TOD scores and FARs. In a correspondence map, the origin (0, 0) is a theoretical probability, where transit neighborhoods of each TOD corridor are independent of residential FARs. If two transit neighborhoods lie closer to each other, then their FARs are similar, while farther transit stations have distinct FARs. For instance, the mean profile of 300 FAR lying close to INA, AIIMS, and Green Park indicates that these transit neighborhoods are characterized by medium to high-dense TOD corridor. Other transit neighborhoods, namely Saket and Malviya Nagar, are closer to 350

196

M. Parida et al.

Fig. 10.3 TOD scores for selected transit neighborhoods

FAR, indicating a high-dense TOD corridor. In this light, corridors II, III, and IV are characterized as high-dense, low-dense, and medium-dense TOD corridors, respectively. Similarly, Fig. 10.5 illustrates the association between transit neighborhoods of TOD corridors and different building uses. Six classes of building use, i.e., residential, commercial, public, government, vacant, and educational uses at transit neighborhoods, were considered as mean profiles in correspondence analysis. The mean profile of residential use is surrounded by Saket, Hauz Khas, INA, AIIMS, and Green Park. It indicates that these transit neighborhoods are highly residential in nature. However, the mean profile of the transit neighborhood, Hauz Khas is lying between residential, commercial, and educational uses. On the other side, INA lies between residential and government uses, whereas AIIMS lies between residential and public uses. Thus, corridor I is characterized as a mix-use TOD corridor. Similarly, corridors II, III, and IV were characterized as residential, public, and commercial TOD corridors, respectively. The correspondence analysis on FARs and building uses has characterized four TOD corridors along the blue line and yellow line corridors of Delhi Metro as: 1. 2.

Corridor I: High-dense Mix-use TOD Corridor II: High-dense Residential TOD

10 Transit-Oriented Development (TOD) as a Sustainable …

197

Fig. 10.4 Correspondence among TOD corridors and residential FARs

3. 4.

Corridor III: Low-dense Public TOD Corridor IV: Medium-dense Commercial TOD

This analysis has revealed distinct differentiation in FARs and building uses, and confirmed different roles of TOD corridors in terms of density and diversity of land uses, respectively. Therefore, within a TOD corridor, investigating complementarity among transit neighborhoods can provide insights into the supply of housing and diverse land uses.

198

M. Parida et al.

Fig. 10.5 Correspondence among TOD corridors and building use

10.7 Density and Mix-use Bonus 10.7.1 Taking Advantage of High-Density Urban Areas Most of the Indian cities commonly face problems of high density in their neighborhoods, increased demand for transportation and resources, accumulated traffic congestion, pollution, and overcrowded streets. This situation is quite opposite of what TODs accomplish. The “density” dimension is an essential criterion for the development of TODs (Kockelman 1997). Developed nations like the US, Canada, and Australia adopted TOD to achieve densities, which already exists in Indian cities. Therefore, metropolitan cities of India must consider “density bonus” in achieving

10 Transit-Oriented Development (TOD) as a Sustainable …

199

its TOD objectives. These cities must learn from successful Asian TOD cities such as Singapore, Tokyo, Seoul, and Hong Kong, whose strategies showed efficacy in balancing densities along transit corridors.

10.7.2 Modernize Existing Mixed Land Uses Indian urban areas naturally have a diverse land-use structure. These urban areas are close to ideal mixed-land-use places that developed nations are now attempting to create. For Indian cities, modernizing instead of developing land uses makes sense. To an apparent extent, city development policies and administration systems are impediments. Upcoming TOD cities need to create strategic organizational changes to adopt the TOD planning process in investments, system design and operation, and demand management. The Indian government needs to pull out laws on restricted FARs and land development waivers to attain articulated densities in TOD neighborhoods. Planners, policymakers, and researchers must address these legal deficiencies if TODs are to be successful in metropolitan cities of India.

10.8 Conclusions Rapidly-expanding metropolitan cities of India are pursuing TOD as an exciting intervention to address their challenging urban transport problems. More recently, government efforts to launch TOD have associated future developments along Metro rail transit networks. At this point, it is crucial to planners, policymakers, and government authorities of Indian Metropolitan cities to consider important lines of action that ensure TOD success and enhance transit ridership. This chapter illustrated specific strategies to extend existing TOD policies of Indian cities. Future developments may admit 1200 m as an appropriate TOD influence zone. The proposed quantitative TOD typology could be more effective to allow operations and investments in the right place. TOD scores at neighborhood level could significantly guide government bodies to formulate systematic action plans for TOD planning and implementation. Maximizing planning parameters such as FARs and building uses should be a priority in enhancing TOD policies. Further, in the case of Delhi, government authorities need to pay attention to investing more funds on identified TOD corridors as immediate action plans. Findings target not only four TOD corridors, but also an attempt to initiate TOD planning at corridor/network levels. However, the current situation of land-use policies for Delhi city at the neighborhood level is poor, in that, a compatible TOD policy is missing. Therefore, it is imperative that existing urban policies have to harmonize the study findings in future developments at neighborhood or corridor level TOD planning. Overall, this chapter intend that in Indian cities, like Delhi, TOD can be the best

200

M. Parida et al.

possible sustainable transport strategy. Therefore, by wise investments, it could be possible to bridge the gap between TOD scores among transit neighborhoods. This chapter hopefully contributes to accomplishing a set of policy parameters that help fortify TOD planning in metropolitan cities of India. However, there is a need for more research in some areas. To achieve the complete benefits of TOD, there is a need for proposing measurements tools that embrace criteria relating to economic development, equity, and socially inclusive environments. Another area that this chapter has dealt with but where more knowledge is needed is how to deliver TOD at a regional scale when promoting the formation of TOD-oriented city regions. There is some research on these areas. However, the majority of TOD research and measurements tools are mainly informed by preferences and conditions from developed nations. There is a lack of research that analyzes how TOD may be adapted and applied in TOD planning and decision-making in the developing world. This chapter has helped to fill this knowledge gap. Learning from global TOD cases, the institutional and social conditions in India can adapt TODs in the best possible way. Acknowledgements The inputs received from the Doctoral research of Dr. Patnala Phani Kumar, funded by the fellowship from the Ministry of Education (MoE), Govt. of India, India is sincerely acknowledged.

References Appleyard BS, Frost AR, Allen C (2019) Are all transit stations equal and equitable? Calculating sustainability, livability, health, and equity performance of smart growth and transit orienteddevelopment (TOD). J Transp Health 14:100584 Arrington GB, Cervero R (2008) TCRP Report 128: effects of TOD on housing, parking, and travel. Federal Highway Administration, Washington DC Aston L, Currie G, Pavkova K (2016) Does transit mode influence the transit-orientation of urban development?—An empirical study. J Transp Geogr 55:83–91. https://doi.org/10.1016/j.jtrangeo. 2016.07.006 Atkinson-Palombo C, Kuby MJ (2011) The geography of advance transit-oriented development in metropolitan Phoenix, Arizona, 2000–2007. J Transp Geogr 19(2):189–199. https://doi.org/10. 1016/j.jtrangeo.2010.03.014 Austin M, Belzer D, Benedict A, Esling P, Haas P, Miknaitis G, Wampler E, Wood J, Young L, Zimbabwe S (2010) Performance-based transit-oriented development typology guidebook. Center For TOD 91 Calthorpe: transit oriented development design guidelines (1990) Carlton I (2009) Histories of transit-oriented development: perspectives on the development of the TOD concept. Institute of Urban and Regional Development, 24. iurd.berkeley.edu. Cervero R, Kockelman K (1997) Travel demand and the 3Ds: density, diversity, and design. Transp Res Part D Transp Environ 2(3):199–219. https://doi.org/10.1016/S1361-9209(97)00009-6 Chen F, Wu J, Chen X, Zegras PC, Wang J (2017) Vehicle kilometers traveled reduction impacts of transit-oriented development: evidence from Shanghai City. Transp Res Part D Transp Environ 55:227–245. https://doi.org/10.1016/j.trd.2017.07.006

10 Transit-Oriented Development (TOD) as a Sustainable …

201

Cho GH, Rodríguez D (2014) The influence of residential dissonance on physical activity and walking: evidence from the Montgomery County, MD, and Twin Cities, MN, areas. J Transp Geogr 41:259–267. https://doi.org/10.1016/j.jtrangeo.2014.06.007 DDA: Master Plan of Delhi 2021: Chapter 19: Transit Oriented Development (Draft). Master Plan of Delhi 2021 (2012), pp 1–22 Dittmar H, Poticha S (2004) Defining transit-oriented development: the new regional building block. Island Press Evans JE, Pratt RH (2007) Transit oriented development: traveler response to transportation system changes. In: World transit research Ewing R, Tian G, Lyons T, Terzano K (2017) Trip and parking generation at transit-oriented developments: five US case studies. Landscape Urban Plann 160:69–78. https://doi.org/10.1016/j.lan durbplan.2016.12.002 FDOT: A framework for transit oriented development in Florida. 76 (2011) Higgins C, Kanaroglu P (2018) Rapid transit, transit-oriented development, and the contextual sensitivity of land value uplift in Toronto. Urban Stud 10(55):2197–2225 Hoback A, Anderson S, Dutta U True walking distance to transit. Transp Plann Technol 31(6):681– 692 (2008). https://doi.org/10.1080/03081060802492785 ITDP: TOD Standard v1.0. Institution for Transportation & Development Policy (2013) 65. https:// rg.smartcitiescouncil.com/system/tdf/public_resources/TODStandard.pdf?file=1&type=node& id=413&force Kamruzzaman M, Baker D, Washington S, Turrell G (2014) Advance transit oriented development typology: case study in Brisbane, Australia. J Transp Gogr 34:54–70. https://doi.org/10.1016/j. jtrangeo.2013.11.002 Kockelman K (1997) Travel behavior as function of accessibility, land-use mixing, and land-use balance: evidence from San Francisco Bay Area. Transp Res Rec 1607:116–125. https://doi.org/ 10.3141/1607-16 Kumar PP, Ravi Sekhar C, Parida M (2018) Residential dissonance in TOD neighborhoods. J Transp Geogr 72:166–177. https://doi.org/10.1016/j.jtrangeo.2018.09.005 Kumar PP, Ravi Sekhar C, Parida M (2020a) Identification of neighborhood typology for potential transit-oriented development. Transp Res Part D Transp Environ 78:102186. https://doi.org/10. 1016/j.trd.2019.11.015 Kumar PP, Ravi Sekhar C, Parida M (2020b) A decision framework for defining transit-oriented development in an Indian city. Asian Transp Stud 6:100021. https://doi.org/10.1016/j.trd.2019. 11.015 Kumar PP, Parida M, Ravi Sekhar C (2021) Exploring the potential of neighborhoods for metro rail-based transit-oriented development. PhD Thesis. Indian Institute of Technology Roorkee, Febraury Litman T (2019) Online TDM encyclopaedia, Chapter on Transit oriented development. VTPI, Victoria, Canada MoUD: Consultancy services for developing guidance documents for transit oriented development (tod), non-motorised transport (nmt) and public bicycle sharing (pbs), May (2016) Motieyan H, Mesgari M (2017) Towards sustainable urban planning through transit-oriented development (A Case Study: Tehran). ISPRS Int J Geo Inf 6(12):402. https://doi.org/10.3390/ijgi61 20402 Papa E, Bertolini L (2015) Accessibility and transit-oriented development in European metropolitan areas. J Transp Geogr 47:70–83. https://doi.org/10.1016/j.jtrangeo.2015.07.003 Parsons Brinckerhoff Quade and Douglas Inc., Cervero R, Howard/Stein-Hudson Associates Inc., & Zupan J (1996) Public policy and transit oriented development: six international case studies. TCRP PROJECT H-1 Transit and Urban Form Rastogi R, Rao KVK (2003) Travel characteristics of commuters accessing transit: case study. J Transp Eng 129(6). https://doi.org/10.1061/(ASCE)0733-947X(2003)129:6(684)

202

M. Parida et al.

Renne JL, Bartholomew K, Wontor P (2011) Transit-oriented and joint development: case studies and legal issues. In: Transit-oriented and joint development: case studies and legal issues. https:// doi.org/10.17226/14588 Renne JL (2009) Evaluating transit-oriented development using a sustainability framework: lessons from Perth’s Network City. In: Planning Sustainable Communities: Diversity of Approaches and Implementation Challenges, pp 115–148. http://www.vtpi.org/renne_tod.pdf Sahu A (2018) A methodology to modify land-uses in a transit oriented development scenario. J Environ Manage 213:467–477. https://doi.org/10.1016/j.jenvman.2017.12.004 Scherrer FP (2019) Assessing transit-oriented development implementation in Canadian cities: an urban project approach. J Plann Educ Res 39(4):469–481 Schlossberg M, Brown N (2004) Comparing transit-oriented development sites by walkability indicators. Transp Res Record 1887:34–42. https://doi.org/10.3141/1887-05 Serge S, Gerald O (2017) Transforming the urban space through transit-oriented development-the 3v approach. Shared Prosperity: Paving the Way in Europe and Central Asia. https://doi.org/10. 1596/978-1-4648-0230-0 Singh YJ, Lukman A, Flacke J, Zuidgeest M, Van Maarseveen MFA (2017) Measuring TOD around transit nodes—towards TOD policy. Transp Policy 56:96–111. https://doi.org/10.1016/j.tranpol. 2017.03.013 Strong KC, Ozbek ME, Sharma A, Akalp D (2017) Decision support framework for transit-oriented development projects. Transp Res Record 2671:51–58. https://doi.org/10.3141/2671-06 Suzuki H, Murakami J, Hong Y-H, Tamayose B (2015) A tale of two metro cities: Delhi and Hyderabad, India. In Financing transit-oriented development with land values: adapting land value capture in developing countries. https://doi.org/10.1596/978-1-4648-0149-5_ch7 Suzuki H, Cervero R, Iuchi K (2013) Transforming cities with transit: transit and land-use integration for sustainable urban development. https://doi.org/10.1596/978-0-8213-9745-9 Thomas R, Bertolini L (2015) Policy transfer among planners in transit-oriented development. Town Plann Rev 86(5):537–560. https://doi.org/10.3828/tpr.2015.32 WRI Annual Report: Bigger Problems, Better Solutions. In: WRI 2018–2019 Annual Report (2019). https://doi.org/10.1080/00325481.1983.11698343 Wey WM, Zhang H, Chang YJ (2016) Alternative transit-oriented development evaluation in sustainable built environment planning. Habitat Int 55:109–123. https://doi.org/10.1016/j.hab itatint.2016.03.003 Xu WA, Guthrie A, Fan Y, Li Y (2017) Transit-oriented development: literature review and evaluation of TOD potential across 50 Chinese cities. J Transp Land-Use 10(1):743–762. https://doi. org/10.5198/jtlu.2017.922 Yang K, Pojani D (2017) A decade of transit oriented development policies in Brisbane, Australia: development and land-use impacts. Urban Policy Res 35(3):347–362 Zemp S, Stauffacher M, Lang DJ, Scholz RW (2011) Classifying railway stations for strategic transport and land use planning: context matters! J Transp Geogr 19(4):670–679

Chapter 11

Design and Evaluation of Public and Non-motorized Transport Systems for Sustainability in Indian Cities Madhu Errampalli

11.1 Introduction 11.1.1 Concept of Sustainability Sustainable development is a holistic practice that tries to minimize adverse impact generated due to transportation system. Sustainability has the ability to manage present demand keeping the future requirements also in view. The concept of sustainability consists of three important areas: economic, environmental, and social. Sustainability encourages policymakers or urban development authorities to frame decisions for the long-term, rather than on short-term gains. The concept of sustainability in transportation system focuses social, economic, and environmental issues simultaneously as shown in Fig. 11.1.

11.1.2 Need of Sustainability for Urban Transport The cities of the world are facing significant challenges especially traffic congestion, air pollution, etc., which are generated from the transportation systems and impact on environment leading to global warming situation. About 13% of Green House Gases (GHG) and about 23% of CO2 emissions worldwide are from transport sector. The transport sector consumes about 30% of total energy and out of that road transport share is about 75%. India’s share (from all transport modes) is approximately 10% of total energy in the world. Traffic congestion during peak hours in four major cities (Delhi, Mumbai, Kolkata, and Bangalore) costs around 1.5 lakh Crores annually. M. Errampalli (B) Transportation Planning and Environment (TPE), CSIR-Central Road Research Institute (CRRI), New Delhi 110025, India © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_11

203

204

M. Errampalli

Fig. 11.1 Concept of sustainability considers social, economic, and environmental issues

Government of India has launched Smart Cities Mission recently to provide sustainable environment and ‘Smart’ systems. The important issues highlighted in a smart city are efficient urban mobility, public transport, and sustainable environment. However, there are no proper guidelines mentioned which can be considered by the urban development authorities to implement or evaluate different types of sustainable transportation systems. Sustainability performance measures or indices are also necessarily required to link actions of stakeholders (urban development authorities or transport corporations or public transport operators) to their overall mission and goals. These measures which adopt rating system can help them to monitor the outcome of the policies in terms of environmental, economic, and social performance. In view of these, there is an immediate need of a methodology that can help urban development authorities or transport corporations or public transport operators to design, implement, and evaluate sustainable transportation systems, namely, public transportation (PT) and non-motorized transport (NMT) in urban areas to promote sustainability as a primary goal.

11.2 Objective and Scope Considering the above issues and the need identified, the objective of the present study has been formulated to design sustainable public transportation (PT) system and sustainable non-motorized transport (NMT) system. For this purpose, the city of Delhi has been considered as the study area to propose procedure for development of a sustainable PT and NMT system and accordingly road network selected for detailed data collection. The present scope in the paper is mainly considered to propose a procedure to enhance quality of existing transport systems in terms of sustainability. The study proposes to identify various parameters of public transport and NMT systems to be considered to achieve sustainability. This also includes evaluation methodology to estimate the sustainability index as a whole which will

11 Design and Evaluation of Public and Non-motorized Transport Systems …

205

help in assessing the status as well as changes occurred due to policy interventions in PT and NMT systems.

11.3 Framework to Develop Sustainable Transportation Considering the objectives of the study, a framework has been formulated to design and implement sustainable transportation systems and also the methodology to evaluate sustainable transportation has been developed with a main focus on PT and NMT systems (CSIR-CRRI 2017) as shown in Fig. 11.2. This would mainly include enhancement of quality of public transport system, feeder transport system, parking facilities at public transport terminals, advanced public transport information systems using Intelligent Transport System (ITS) technologies, restrict/control usage of private vehicles, integration of mass transportation system and evaluation with sustainability index considering social, economy and environment parameters.

Fig. 11.2 Broad framework to develop sustainable transportation system

206

M. Errampalli

11.4 Sustainable Public Transportation (PT) System Careful and systematic planning for public transport is a key for successful lead to urban sustainability and overall network needs to be considered as a whole to provide convenient, multi-directional, and seamless travel. The focus needs to be made on the areas of existing public transportation systems which lead to sustainable public transportation system are explained in the next sections.

11.4.1 Quality Enhancement of Public Transport System Public transport holds center stage in the urban transport agenda. A well-functioning and sustainable city cannot be achieved without strengthening quality of public transport system. The steps involved to enhance quality of public transport systems to achieve sustainability are as follows (CSIR-CRRI 2017): • Identify the parameters for evaluation of service quality using RATER Model (typical parameters identified for bus and sub-urban trains systems are given in Table 11.1). • Using techniques such as Important Performance Analysis (IPA), Customer Satisfaction Index (CSI), etc., the parameters required for improvements can be selected. • According to the identified parameters, the improvement policies shall be formulated and implemented to enhance the quality of the public transport system. Further, the quality of public transport can be enhanced considering the following parameters to attract and retain commuters to public transport: • To attract passengers towards bus transportation system the frequency, cleanliness of buses and bus stops and management of crowd during travel should be given top priority. In case of Sub-Urban railway/MRTS, the parameters namely personal safety onboard and at station, cleanliness of trains and stations, proper access system integrated with rail system, minimum service time including interchange, etc., should be given top priority. • To retain the existing users of bus transportation, the parameters namely bus fare, good on-board facilities, staff behavior, seat priority for specific segments of population, stopping of bus at exact bus stop location and proper installation and functioning of electrical equipment’s (Fan/Light) should be given high consideration. In case of Sub-Urban railway the factors like fare, availability of seats at station and in train, electrical equipment onboard, information dissemination at station, timely information announcement, and display at station should be given due consideration to retail commuters.

11 Design and Evaluation of Public and Non-motorized Transport Systems …

207

Table 11.1 Typical quality performance indicators of public transport system Public transport system

Quality performance indicators

Bus Transit System

1. Personal Safety on Board; 2. Personal Safety at Bus Stop; 3. Cleanliness of buses; 4. Cleanliness of Bus stop; 5. Bus stop Maintenance; 6. Crowd during Travel; 7. Comfort During Travel; 8. Functionality of Electrical Equipment’s (Fan/Light); 9. Bus Fare; 10. Frequency of buses; 11. Regularity/Punctuality of buses; 12. Parking Facilities; 13. Facilities for disabled People; 14. Information dissemination at bus stop; 15. Information dissemination On-Board; 16. Availability of Schedule/Map at Bus Stop, App, Booklets; 17. Complaint Registration/Redressal Facilities; 18. Periodicity of Ticket Inspection; 19. Bus Helpline Information; 20. Behavior of Bus Conductor; 21. Seat Priority for woman, children, Senior Citizen and Disable People; 22. Bus Stopping at exact Bus Stop Location; 23. Complete Stopping at marked space of every bus Stop; 24. Utilities at bus stop (especially at major stop)—bin, water, urinal, etc.; 25. Availability of benches, shelter at Bus Stop; 26. Quality of seats, standing comfort, and accessories; 28. Panic/STOP button at every seat; 29. Bus onboard quality (low floor or high floor)

Sub-urban rail transit system/MRTS 1. Personal Safety on Board; 2. Personal Safety at Station through CCTV cameras; 3. Cleanliness of Train; 4. Cleanliness of Station; 5. Cleanliness of Toilet Facilities on Train/Station; 6. Availability of Seats at Station; 7. Crowding on Board; 8. Comfort on Board; 9. Functionality of Electrical Equipment’s (Fan/Light); 10. Train Fare; 11. Frequency of Train; 12. Regularity/Punctuality of Train; 13. Parking Facilities; 14. Facilities for Disabled Persons, children, woman and elderly; 15. Luggage Facilities on Board; 16. Information Dissemination at Station; 17. Timely Information Announcement and display at Station; 18. Complaint Registration/Redressal Facilities; 19. Periodicity of Ticket Inspection; 20. Schedule and movement App, booklets; 21. Entry/exit/interchange facilities—efficiency, optimized flow; 22. On-station Service time; 23. Behavior of staffs’ personnel

11.4.2 Sustainable Feeder Transport System A multi-modal transportation system with seamless feeder transport and planned interchange facilities is highly required for sustainability as it would ensure the use of public transport to its best potential. The following are the steps to design optimal feeder network:

208

M. Errampalli

• Demarcate influence area for non-motorized transport (NMT), intermediate public transport (IPT), and public transport (PT) in terms of catchment area with 1–3 and 5 km zones for Pedestrians, Bicyclists, IPT mode, and buses should be considered, respectively, at each metro station. • Identify feeder service routes to metro station connecting attraction point for commuters. • Develop the feeder service-based access-route maps and install at prime locations. These may also be accompanied with PT service information kiosks for readily available information on services available in the area. • Consider the plan of the existing roads for the feeder service to attract more commuters. • Estimate the number of feeder service (minibus, share auto, etc.) required for each route taking into consideration the frequency of the mass rapid transit per hour and the loading to be catered by the system. • Consider integration of transport system to achieve seamless fare collection system. • Assess the requirement of separate feeder service which should almost matches with existing public transport service in the area.

11.4.3 Advanced Parking Facilities at Public Transport Terminals Equitable provision of parking for different feeder modes and traffic management plan at and near mass public transport/metro terminals is the prerequisite of a sustainable transport system. The demand needs to be estimated from feeder transport survey and that can be utilized to estimate parking/circulation demand of various feeder modes. Further to understand the characteristics and demand for parking by private vehicle users, namely, car, two-wheeler and bicycle, the parking accumulation and parking duration surveys need to be carried out for both organized and unorganized parking areas to design parking facilities for these modes. The traffic circulation plan for especially IPT and feeder modes is also necessarily prepared and implemented considering issues as given below (CSIR-CRRI 2017): • Cycle Rickshaws may be given top priority for access to the metro station in order to encourage non-motorized/green transport by providing segregated entry lanes and designated halting place just in front of the metro station. • Second priority for access to the metro station may be given to bus feeder system in order to enhance efficiency and sustainability. Direct entry/exit to the station area and designated halting place may be provided close to the gate of the metro station/bus terminal. • Third priority regarding circulation and access to the metro station may be given to IPT modes comprising of E-Rickshaw and Gramin Sewa.

11 Design and Evaluation of Public and Non-motorized Transport Systems …

209

• Fourth priority for circulation and access to the metro station may be given to other IPT modes comprising of taxi and auto rickshaws. • Appropriate traffic management/circulation plan for each permitted IPT and feeder bus need to be prepared for each metro station and should be strictly enforced along with lane discipline. • Advance public information system and parking information system be designed and provided at the metro stations in order to provide information of the feeder transport system, and parking and traffic management system.

11.4.4 Advanced Public Transport Information Systems (APTIS) The efficient use of public transport systems depends on frequency of services, fare, seat availability, and real-time information of the public transport. Usually, frequency and fare are considered to be most significant to satisfy the people. However, the satisfaction of the user highly depends on information about the availability of seats and arrival time. In order to inform the commuters to plan their departure time to reach public transport boarding point, the real-time information is necessary to be communicated through APTIS using Intelligent Transport System (ITS) technologies. In this way, the services to public transport users can be improved by providing the dynamic information of the buses on the network and attracting the people to use the public transport system in order to reduce the private vehicle usage. However, there is always been some ambiguity on what kind of ITS technologies that a city should have based on population, technology usage by people, system implementation costs, etc., as there are varieties of technologies being developed in the domain of information technology and number of vehicle tracking systems are available over the globe. Considering this, the points to be considered to evaluate the appropriateness of suitable ITS technologies in terms of Advanced Public Transport Information System (APTIS) and policy guidelines for Indian conditions are given below (CSIR-CRRI 2017): 1.

2.

3.

The success of APTIS depends on availability of technology in the area in terms of modes of access to the information. The sample survey of the public transport users shall be carried out to know the modes of access available with passengers, capability of usage by passengers, probable percentage of people who will be using it in the event of implementation, prevailing waiting times at bus/metro terminals, etc. The specification of the system going to implement in the area in terms of budget, maintenance, accuracy of information dissemination, etc., along with share of public transport trips in total trips, size of the city, and an absolute number of trips, average income of commuters, etc., shall be obtained. The total savings of all passengers of public transport shall be calculated in terms of total savings arising out of reduction in wait time/total travel time along with other benefits.

210

4.

5. 6.

M. Errampalli

APTIS success will be dependent upon the accuracy of information and campaign of the system, including short training of the passengers at terminals, stops and at large gathering places to explain the usage. The system may not be viable or successful in small cities 2

Acceptable

1 30

SR > 20

Note SR = Speed reduction between successive sections

difference between the maximum speed at a tangent and the minimum speed at the curve, while ΔV mean is the difference between the mean speed of the whole test course and the minimum speed at the curve. Based on the 50th percentile (median) and the 85th percentile critical values of the sample distribution of ΔV max and ΔV mean , respectively, the design class of each curve element was identified. For example, the speed reduction below the 50th percentile of the ΔV max value was considered as the good, above it is considered as fair, and above 85th percentile of ΔV max value is considered as poor. Thus, Cafiso and Cava (2009) evaluated the design class of each curve of the entire study stretch. Accordingly, the theoretical design guidelines were prepared using the geometric variables and operating speed performance.

12.2.3 Safety Evaluation Studies in Indian Context So far, Jacob et al. (2013) provided the localized safety evaluation criteria for the individual horizontal curves along two-lane rural highways using the equivalent property damage only (EPDO) data. Moreover, they developed speed prediction models for various vehicle classes such as trucks, buses, passenger cars, and two wheelers. For all vehicle types, they prepared a safety level categorization (see Table 12.3) for tangent-curve and curve-curve successive sections with slightly increased values of a threshold than Lamm’s criteria II (as indicated in Table 12.1). The EPDO was calculated from the historical crash data along several horizontal curves.

12.3 Discussion The safety evaluation studies available in the literature are primarily based on speed because of their close resemblance with the design speed of the road, easier in data collection and field applications. The speed-based studies showed a significant correlation with the historical crash data and could be successfully used to evaluate the safety of rural highways. Most of these criteria are available for two-lane rural highways and need further development for multilane and divided highways or expressways. Most of the current studies from India represented speed-based prediction

12 Speed-Based Safety Evaluation …

227

models (Maji et al. 2018; Choudhari and Maji 2019a; Sil et al. 2019b). Except Jacob et al. (2013), these studies did not provide any safety evaluation criteria, possibly due to the non-availability of reliable crash data. A holistic data collection procedure is required to overcome this challenge. With the advanced data acquisition technologies, the performance measures such as the speed profile of the driver can be compared with the visual demand of the drivers or physiological variables of the drivers. The physiological measures generally include the heart rate, skin conductance, and eye-tracking. For instance, eye-tracking could indicate the amount of visual demand of the driver along a specific curve. Thus, higher demand would likely increase the mental workload of the driver. Such demand could be correlated with the driver performance measures such as speed profile to identify uncomfortable or unexpected situations for the drivers. Accordingly, the safety along the horizontal curve could be evaluated. Studies from India presented different speed prediction models for different vehicle categories, such as passenger cars, heavy commercial vehicles, light commercial vehicles, and buses. (Jacob and Anjaneyulu 2013; Maji et al. 2018; Sil et al. 2019a; Maji and Tyagi 2018; Malaghan et al. 2020). However, the developed safety evaluation criteria presented in previous sections are primarily available for passenger cars. Some studies consider commercial vehicles, but those are limited in numbers and represent different geographic locations than India (Jacob and Anjaneyulu 2013; Misaghi and Hassan 2005; Maji and Tyagi 2018; Medina and Tarko 2005). Recently, Maji et al. (Maji et al. 2018) and Maji and Tyagi (2018) proposed models for passenger cars and heavy and light commercial vehicles in India. These studies used a small sample size and a limited number of study sites. Hence, speed choice (i.e., 85th percentile speed) modeling for different vehicle categories has further scopes to overcome the existing limitations. The safety evaluation guidelines could also be explored for different vehicle types. Previous studies used only the 85th percentile value of the complete data distribution. However, researchers argue that the two different data distributions (having different mean and standard deviation) could have the same 85th percentile value (Sil et al. 2019a; Medina and Tarko 2005). Adopting a deterministic operating speed (i.e., 85th percentile speed) as a measure of geometric design consistency evaluation ignores the variability of speed distribution. In other words, it is unable to accommodate the variability in speed choice behavior among the drivers. Therefore, researchers considered variability of speed distribution (Sil et al. 2019a; Medina and Tarko 2005) and referred to it as an essential factor for evaluating the consistency and safety of horizontal curves. Indeed, some of them adopted the speed distribution for a reliable consistency and safety evaluation of horizontal curves (Medina and Tarko 2005; Jesna and Anjaneyulu 2016; Himes and Donnell 2010). The speed is one of the vehicle dynamic variables and is considered as a longitudinal performance variable. The lateral performance of the vehicle may not be captured using speed-based studies. The lateral acceleration could be explored as a suitable performance variable in the lateral direction because of its direct relationship with the centrifugal force acting on the vehicle along the curves (Wang et al. 2015; Choudhari and Maji 2021). The lateral acceleration is a ratio of the square of speed

228

T. Choudhari et al.

to the path radius of the vehicle. The speed and radius of the curve are primary inputs for the design of the horizontal curves (IRC 1990). With advancements in data collection technologies, lateral acceleration could be suitably used for the safety evaluation along horizontal curves. Moreover, for multilane highways, where the speed differential is not significant, the lateral acceleration can induce skidding or overturning risks along curves. The driver demographic and experience factors also affect the driving pattern (Choudhari and Maji 2019b). The crash rate is more related to age than the experience of a driver (Rutter and Quine 1996). Young drivers have a significantly high number of crashes than older drivers (Curry et al. 2015; Lambert-Bélanger et al. 2012; Stevenson et al. 2001). Along with age and experience, researchers used other factors such as driver type, trip frequency and qualification, and occupation to assess and identify risky drivers (Brusque and Alauzet 2008; Wu et al. 2016). Several crash-avoidance studies had successfully explained the effect of socio-demographic factors during critical events (Meng et al. 2019; Haque and Washington 2015). However, the studies on horizontal curves to assess driving behavior are extremely limited. The driving strategy is an important factor related to road safety, vehicle dynamics, and control research. It includes the effective use of vehicle control measures such as pedal forces (throttle/accelerator, brake, and clutch), steering wheel position, and manual transmission (gearbox) (Wang et al. 2018). Drivers effectively use steering and brake pedal as a compensatory measure during the accident (or collision) avoidance process (Venkatraman et al. 2016). Several models were developed to capture crash-avoidance behaviors primarily during distraction (Haque and Washington 2015), situation urgency (Wang et al. 2016), driver fatigue conditions (Meng et al. 2019), and driving support studies (Blommer et al. 2017). However, there is a lack of understanding of the perceptual processes that influence drivers’ braking or steering choices (Venkatraman et al. 2016). Currently, no studies are available to understand the effect of the horizontal curve on the accident-avoidance behavior of drivers using vehicle control measures. The vehicle dynamic measures are the manifestation of vehicle control measures, and any runoff avoidance behavior are compensated using the control measures. Hence, there is a scope to explore the vehicle control measures to minimize the road traffic accident risk along the horizontal curves. Several countries started following the iterative road design methodology for the consistent road design using operating speed criteria. The method used in the UK assumes a trial design speed for an approach section of the highway. The operating speed of the vehicle is predicted by using statistical models. The design speed and operating speed are compared with each other, and the design speed is raised to the value of operating speed. Thus, the consistency between design and operating speed is ensured. Further, in German standards, the design speed for a horizontal curve is selected first. Accordingly, the minimum radius of the curve and gradient are assigned. Then using these values, the operating speed from curvature change rate and road width is calculated, and the superelevation is applied accordingly. However, if the difference between design speed and operating speed is more than 20 km/h, the design speed is raised toward operating speed, and a similar iteration is performed once again. The French standard is similar to German practices, where

12 Speed-Based Safety Evaluation …

229

the operating speed is calculated directly. For each alignment, the sight distance is checked for the available sight distance. All these methods are for improving the consistency of a single road geometry element. Therefore, further studies have constituted more consistent measures for successive alignments. Lamm’s IInd criteria explained in Table 12.1 is considered for successive elements. Lamm et al. (Lamm et al. 1995) considered mean accident data for each successive element. Hence, the operating speed predicted for the two elements, if the difference between them is more than 20 km/h, the road elements are redesigned. In Russia, for good highway design, Garach et al. (2014) indicated that the difference between operating speeds of successive elements should be less than 15% of the initial operating speed. Currently, there is no iterative design methodology in the Indian design guidelines.

12.4 Conclusion and Future Research Direction The literature presented in this study has shown the importance of consistency in the alignment design of rural roadways. Particularly, this chapter presented several safety evaluation studies for examining the consistency of horizontal curves. Most of these studies represent strong lane disciplined traffic. Since Indian traffic exhibits weak lane discipline along with a heterogeneous traffic environment, a separate behavioral study is imperative. In India, the conventional design method is used for road geometric design which does not rely on the performance of the geometric elements. The performance parameters such as operating speed and corresponding safety evaluation criteria would help in providing consistency along rural highways. This would reduce surprises and instill safety along curves. Several Indian studies have already developed speed prediction models; however, they are yet to be used for design evaluation. The inclusion of a performance-based approach in the geometric design process would help the design engineers, transport organizations, and policyholders to assess safety, security, and pre- and post-crash management strategies from the road users’ point of view. Hence, to achieve the overall goal of safety along horizontal curves, this article recommends the following research directions: • Indian researchers face challenges in getting well-justified accident data for research purposes. A holistic crash data collection method is required for the assessment of horizontal curves. This would help in developing crash-based models and provide suitable safety evaluation criteria. In absence of crash data, the correlation between surrogate safety measures such as driver performance, vehicle stability, alignment index, and driver workload needs further exploration to find crash risk along curves. • The speed choice of drivers varies based on the efficiency of law enforcement. It motivates to develop regional speed prediction models. Moreover, considering the heterogeneity traffic condition, such models need better calibration for different vehicle types.

230

T. Choudhari et al.

• Speed-based studies are reasonably explored in India because of ease in understanding, data collection, and suitable application in the field. However, with the advancement in data collection technologies, the other performance measures such as acceleration (longitudinal and lateral) and jerk need further investigation. These measures could be compared with the thresholds available in previous studies to understand driver’s comfort and safety. • The dynamic character of the vehicle is the manifestation of vehicle control measures. The vehicle control measures provide a better understanding of the brake and steering usage of drivers. Several studies have used these measures to examine the collision-avoidance behavior of drivers. Similarly, these measures could be explored along horizontal curves to avoid the runoff or overturning risk along curves. Such a study would be helpful for developing connected vehicle technology in India. This could improve the driver–vehicle–infrastructure communications and help in reducing the human error while driving. • The driving behavior and risk-taking tendency also vary with the sociodemographic and experience factors of the driver. The crash-avoidance behaviors of several driver groups should be explored to develop a graduated training program for risky/unsafe drivers. Currently, no such training program is available in India. • Drivers’ perception of a horizontal curve can influence driving operations (e.g., speed, reaction time, and acceleration/deceleration) and in turn the safety. For example, an unsafe speed selection in horizontal curve might be attributed to the sharpness misperception. Therefore, a comprehensive study on drivers’ perception could be beneficial to improve the safety at horizontal curves.

References Anderson IB, Bauer KM, Harwood DW, City K, Trans- T (1999) Relationship to safety of geometric design consistency measures for rural two-lane highways. Transp Res Rec J Transp Res Board 1658:43–51 Blommer M, Curry R, Swaminathan R, Tijerina L, Talamonti W, Kochhar D (2017) Driver brake versus steer response to sudden forward collision scenario in manual and automated driving modes. Transp Res Part F Traffic Psychol Behav 45:93–101 Brusque C, Alauzet A (2008) Analysis of the individual factors affecting mobile phone use while driving in France: socio-demographic characteristics, car and phone use in professional and private contexts. Accid Anal Prev 40:35–44 Cafiso S, Cava G (2009) Driving performance, alignment consistency, and road safety. Transp Res Rec J Transp Res Board 2102:1–8 Choudhari T, Maji A (2017) Miscellaneous study on speed reduction along horizontal alignments of rural highways: a driving simulator based approach for developing countries like India. In: Annual meeting at transportation research board, pp 17–06699 Choudhari T, Maji A (2019a) Effect of horizontal curve geometry on the maximum speed reduction: a driving simulator-based study. Transp Dev Econ 5:14 Choudhari T, Maji A (2019b) Socio-demographic and experience factors affecting drivers’ runoff risk along horizontal curves of two-lane rural highway. J Safety Res 71:1–11

12 Speed-Based Safety Evaluation …

231

Choudhari T, Maji A (2021) Risk assessment of horizontal curves based on lateral acceleration index: a driving simulator-based study. Transp Dev Econ 7:2 Curry AE, Pfeiffer MR, Durbin DR, Elliott MR (2015) Young driver crash rates by licensing age, driving experience, and license phase. Accid Anal Prev 80:243–250 Dell’Acqua G, Russo F, Mauro R (2013) Validation procedure for predictive functions of driver behaviour on two-lane rural roads. Eur Transp - Trasp Eur 1–13 Fitzpatrick K, Anderson IB, Bauer KM, Collins JM, Elefteriadou L, Green P, Harwood DW, Irizarry N, Koppa R, Krammes RA, McFadden J, Parma KD, Passetti K, Poggioli B, Tsimhoni O, Wooldridge MD (2000) Evaluation of design consistency methods for two-lane rural highways, executive summary Fitzsimmons EJ, Kvam V, Souleyrette RR, Nambisan SS, Bonett DG (2013) Determining vehicle operating speed and lateral position along horizontal curves using linear mixed-effects models. Traffic Inj Prev 14:309–321 Garach L, Calvo F, Pasadas M, De Oña J (2014) Proposal of a new global model of consistency: application in two-lane rural highways in Spain. J Transp Eng 140:1–9 García A, Llopis-Castelló D, Camacho-Torregrosa F, Pérez-Zuriaga A (2013) New consistency index based on inertial operating speed. Transp Res Rec 105–112 Gibreel GM (1999) State of the art of highway geometric design consistency. J Transp Eng Am Soc Civ Eng 305–313 Haque MM, Washington S (2015) The impact of mobile phone distraction on the braking behaviour of young drivers: A hazard-based duration model. Transp Res Part C Emerg Technol 50:13–27 Himes SC, Donnell ET (2010) Speed prediction models for multilane highways: Simultaneous equations approach. J Transp Eng 136:855–862 IRC: 73–1980 (1990) Geometric design standards for rural (non-urban) highways. Indian Roads Congress, New Delhi Jacob A, Anjaneyulu MVLR (2013) Operating speed of different classes of vehicles at horizontal curves on two-lane rural highways. J Transp Eng 139:287–294 Jacob A, Dhanya R, Anjaneyulu MVLR (2013) Geometric design consistency of multiple horizontal curves on two-lane rural highways. Proc - Soc Behav Sci 104:1068–1077 Jesna NM, Anjaneyulu MVLR (2016) Reliability analysis of horizontal curves on two lane highways. Transp Res Proc 17:107–115 Krammes RA, Brackett RQ, Shafer MA, Ottesen JL, Anderson IB, Fink KL, Collins KM, Pendleton OJ, Messer CJ (1995) Horizontal alignment design consistency for rural two-lane highways Lambert-Bélanger A, Dubois S, Weaver B, Mullen N, Bédard M (2012) Aggressive driving behaviour in young drivers (aged 16 through 25) involved in fatal crashes. J Safety Res 43:333–338 Lamm R, Psarianos B, Choueiri ME, Soilemezoglou G (1995) A practical safety approach to highway geometric design international case studies: Germany, Greece, Lebanon, and The United States. In: The international symposium on highway geometric design practices, Boston Lin Y, Niu J (2011) Effect of curvature change rate of highway horizontal curve on the path a vehicle. In: ICTIS 2011: multimodal approach to sustained transportation system development: information, technology, implementation. pp 904–912 Maji A, Sil G, Tyagi A (2018) 85th and 98th percentile speed prediction models of car, light, and heavy commercial vehicles for four-lane divided rural highways. J Transp Eng Part A Syst 144:1–8 Maji A, Tyagi A (2018) Speed prediction models for car and sports utility vehicle at locations along four-lane median divided horizontal curves. J Mod Trans 1–8 Malaghan V, Pawar DS, Dia H (2020) Speed prediction models for heavy passenger vehicles on rural highways based on an instrumented vehicle study. Transp Lett 00:1–10 Medina AMF, Tarko AP (2005) Speed factors on two-lane rural highways in free-flow conditions. Transp Res Rec 39–46 Meng F, Wong SC, Yan W, Li YC, Yang L (2019) Temporal patterns of driving fatigue and driving performance among male taxi drivers in Hong Kong: a driving simulator approach. Accid Anal Prev 125:7–13

232

T. Choudhari et al.

Misaghi P, Hassan Y (2005) Modeling operating speed and speed differential on two-lane rural roads. J Transp Eng 131:408–418 MORTH: Road Accidents in India 2015 (2016) Transport research wing, ministry of road transport and highways, Government of India, New Delhi MORTH: Road Accidents in India 2018. Transport Research Wing, Ministry of Road Transport and Highways, Government of India, New Delhi (2019) Nama S, Maurya AK, Maji A, Edara P, Sahu PK (2016) Vehicle speed characteristics and alignment design consistency for mountainous roads. Transp Dev Econ 2:23 Pérez-Zuriaga AM, Garcia AG, Camacho-Torregrosa FJ, D’Attoma P (2010) Modeling operating speed and deceleration on two-lane rural roads with global positioning system data. Transp Res Rec J Transp Res Board 2171:11–20 Poe CM, Tarris JP, Mason JM (1996) Operating speed approach to geometric design of low-speed urban streets Polus A, Mattar-Habib C (2004) New consistency model for rural highways and its relationship to safety. J Transp Eng 130:286–293 Polus A, Fitzpatrick K, Fambro DB (2000) Predicting operating speeds on tangent sections of two-lane rural highways. Transp Res Rec J Transp Res Board 1737:50–57 Rutter DR, Quine L (1996) Age and experience in motorcycling safety. Accid Anal Prev 28:15–21 Sil G, Maji A, Nama S, Maurya AK (2019b) Operating speed prediction model as a tool for consistency based geometric design of four-lane divided highways. Transport 34:425–436 Sil G, Nama S, Maji A, Maurya AK (2020) Modeling 85th percentile speed using spatially evaluated free-flow vehicles for consistency-based geometric design. J Transp Eng Part A Syst 146:04019060 Sil G, Nama S, Maji A, Maurya AK (2019) Effect of horizontal curve geometry on vehicle speed distribution: a four-lane divided highway study. Transp Lett 1–10 Stevenson MR, Palamara P, Morrison D, Ryan GA (2001) Behavioral factors as predictors of motor vehicle crashes in young drivers. Crash Prev Inj Control 2:247–254 Venkatraman V, Lee JD, Schwarz CW (2016) Steer or brake? Transp Res Rec J Transp Res Board 2602:97–103 Wang J, Dixon K, Li H, Hunter M (1961) Operating speed models for low speed urban environment based on in-vehicle global positioning system data. Transp Res Rec J Transp Res Board 2006:24– 33 Wang X, Wang T, Tarko A, Tremont PJ (2015) The influence of combined alignments on lateral acceleration on mountainous freeways: A driving simulator study. Accid Anal Prev 76:110–117 Wang X, Zhu M, Chen M, Tremont P (2016) Drivers’ rear end collision avoidance behaviors under different levels of situational urgency. Transp Res Part C Emerg Technol 71:419–433 Wang J, Li W, Li J, Liu Y, Song B, Gao H (2018) Modeling a driver’s directional and longitudinal speed control based on racing track features. Shock Vib 2018:12 WHO: Global Status Report on Road Safety 2018 (2018) Wu J, Yan X, Radwan E (2016) Discrepancy analysis of driving performance of taxi drivers and nonprofessional drivers for red-light running violation and crash avoidance at intersections. Accid Anal Prev 91:1–9

Chapter 13

A Global Perspective of Railway Security Malavika Jayakumar, Aparna Joshi, Avijit Maji, and Prasanta K. Sahu

13.1 Introduction Rail transportation is considered the backbone of a nation for its vital role in promoting economic growth and productivity. As of the financial year (FY) 2019, the Indian rail network, the third-largest rail network globally, has a total route network of 67,415 km with 13,523 passenger trains and 9,146 freight trains. Freight remains the leading revenue-earning segment for Indian Railways, accounting for 64% of the total revenue in FY 2020 (IBEF: Railways 2021). Construction of rail facilities requires a massive amount of capital investment compared to other transportation modes. Since railway infrastructure offers mass transportation of goods and passengers, any vulnerabilities associated with its operational interruption can result in a tremendous loss to the national exchequers. Railways are increasingly at the risk of potential vulnerabilities due to their large-scale expansion in many countries, the high volume of passengers using the facility, and limited safety measures compared to other public transport modes. These attributes make it a promising target for terrorist and cyber-based attacks. The transportation sector has always been a prime target for attacks due to the extent of destruction it could inflict on civilians and a nation’s economy. The terrorist attack on the world trade center on 11 September 2001 shook the world as it questioned the safety and security of all modes of public transport, especially the airlines. Since then, the security measures implemented in airports against any potential vulnerabilities have been strengthened manifold. Airports adopted strict M. Jayakumar · A. Maji (B) Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra 400076, India A. Joshi · P. K. Sahu Department of Civil Engineering, Birla Institute of Technology and Science Pilani, Hyderabad, Telangana 500078, India © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_13

233

234

M. Jayakumar et al.

passenger identity verification, baggage screening, and strong immigration policies for increased safety of passengers, goods, and the infrastructure itself. However, similar strategies cannot be implemented for railways. Rail transportation has specific features that make it susceptible to both terrorist attacks and cyber threats such as: rail network is quite extensive and open; trains have fixed routes and scheduled stops, they have many access points for boarding and alighting of passengers and handling of goods; the significant unrestricted movement of passengers on platforms and the possibility of staying anonymous within the large number of passengers and the availability of several escape routes (Setola et al. 2015; Jenkins 2001). All these characteristics make railways worldwide prone to security breaches. Therefore, adequate security measures must be adopted to secure the rail infrastructure while ensuring that these features are not compromised. This book chapter summarizes the various security issues in the railway system and their impacts. It further reviews the security measures adopted, and current status of railway security in India and its’ relevance for the upcoming High-Speed Rail network.

13.2 Critical Infrastructure Critical infrastructure (CI) refers to a system, a resource, or a process whose destruction, disruption, or momentary unavailability can affect the efficiency and operations of a country, including its central and local administration (Aradau 2010). A country’s critical infrastructure includes transportation, telecommunication, water supply system, credit and finance, energy, and emergency services. Being a mode that allows mass transportation, the railway is considered a critical infrastructure that needs to continue its operation even in abnormal circumstances, ensuring utmost safety. The critical safety-related components within the railways are mentioned in the following subsection. According to Carlson (2003), there are several “safety–critical” systems/components within a train, like doors, brakes, propulsion, environment, power distribution, communication systems, and lighting, which can severely affect people and property if damaged. These systems play a significant role in the operation of a passenger train and, to a smaller extent, in freight trains. The braking system is one of the most critical systems within a train; any potential threat by an external entity can result in a train’s derailment and massive damage to the surrounding environment. Similarly, propulsion and the control of speed is another safety–critical system that can result in collision or derailment in case of failure. Intruders may make attempts to cause damage to the working of the door, which can cause injury to passengers. The next important aspect is the environmental system; changes in the temperature caused by any means can adversely affect transported or stored goods in freight trains. Communication system, along with the signaling system, is also critical in ensuring trains’ safe movement. The drivers and other crew members rely on it to guide the train. Incidents of hacking or cyberattack like viruses can disrupt the signals and cause malfunction, resulting in collisions and

13 A Global Perspective of Railway Security

235

accidents (Carlson et al. 2003). Since such systems are the most critical within a train, numerous attempts were made over the years to compromise these systems’ safety. Some of the past major security breaches in railways are discussed in the next section.

13.3 Security Breaches in Railways After the 9/11 attack, the security measures adopted in airlines include reinforcing and bullet-proofing cockpit doors to avoid unauthorized access, identification checks, increased security screening for passengers, checked baggage items at all airports, and installing CCTV cameras. These measures improved security in airlines, but other easily accessible transportation modes become more vulnerable. It is not always possible to protect railways by extensively using metal detectors, X-ray machines, explosives sniffers, hand searches, and armed guards. Such security protocols would considerably increase delays, lower the accessibility, convenience, and make travel in the railways more expensive (Jenkins 2001).

13.3.1 Passenger Rails Global Terrorism Database (GTD) is an open-source database on terrorist attacks that have taken place since 1970, excluding the year 1993. GTD is developed by The National Consortium for the Study of Terrorism and Responses to Terrorism (START) at the University of Maryland, U.S.A (Miller et al. 2021). The database contains more than 201,000 terrorist incidents in various sectors like maritime, transportation, business, government, telecommunication, educational institutions, religious institutions, and several other sectors from 1970 to 2019 (START: GTD Codebook 2019). 1970 to 2019, out of which 1918 incidents are some common types of attacks on rail transportation include bombing or explosion, facility or infrastructure attack, armed assault, assassination, unarmed assault, hostage-taking, and hijacking. Bombings contribute to around 81% of the attack on rail infrastructure, followed by armed assault and facility or infrastructure attack. These statistics reinforce the necessity of implementing adequate security measures to prevent terrorism-based attacks on rail infrastructure worldwide. In December 2000, a series of bombings occurred in Manila, one of which was within a light rail transit coach on Blumentritt Street, resulting in the death of 10 people. Ten bombs exploded on four trains in Madrid on 11 March 2004 (Ray 2004), resulting in many civilian casualties. A series of bombings in Mumbai’s commuter train system by a terrorist group in 2006 killed 209 people, injuring more than 700 (Economic and Times: Six Terror Attacks that Shook India 2021). The 2011 Minsk Metro bombing in Belarus had resulted in the death of 15 people and the injury of 204 (BBC News 2021a). After a twin blast in Zaventem airport, in 2016, a suicide

236

M. Jayakumar et al.

bomber detonated at the Maelbeek metro station near European Union headquarters in Brussels, resulting in the death of around 20 people (BBC News 2021b). Figure 13.1 shows the number of terrorist incidents targeting railway infrastructure in various regions worldwide from 1970 to 2019. From the figure, it can be observed that South Asia reports the most significant number of attacks, with 715 incidents, while North America, Australasia, and Oceania report the least number of incidents, three incidents for each. Figure 13.2 focuses on the percentage contribution of each type of attack on the top five regions with the highest number of incidents shown in Fig. 13.1. Armed assault, assassination, bombing/explosions, facility/infrastructure attacks are the most common types; other attack types like hijacking, unarmed assault, and hostage-taking are less in number. From Fig. 13.2, it can be observed that armed assault is most common in Sub-Saharan Africa than in other regions. At the same time, South America shows the least number of such attacks. In contrast, South America has the highest number of bombings/explosions, followed by the Middle East and Australasia & Oceania North America Cental Asia Central America & Caribbean East Asia Southeast Asia Eastern Europe Middle East & North Africa Sub-Saharan Africa South America Western Europe South Asia

3 3 23 43 81 143 144 161 174 198 230 715 0

100

200 300 400 500 600 Number of incidents (1970-2019)

700

800

Fig. 13.1 Number of incidents on railways at various geographical regions from 1970 to 2019 (Data source Global Terrorism Database (Miller et al. 2021))

Middle East & North Africa Sub-Saharan Africa South America Western Europe South Asia 0 Armed Assault Bombing/Explosion Other attacktypes

20

40 60 80 Attack types (%) Assassination Facility/Infrastructure Attack

100

Fig. 13.2 Attack types at the top five regions from 1970 to 2019 (Data source Global Terrorism Database (Miller et al. 2021))

13 A Global Perspective of Railway Security

237

North Africa, South Asia and Western Europe, and the least in Sub-Saharan Africa. It shows that the attacks in South America are more discreet than in Sub-Saharan Africa because the physical presence of armed terrorists is less. Assassination is also reported most in Sub-Saharan Africa, while a small percentage was observed in Western Europe and South Asia. Facility/Infrastructure attack was observed more in Western Europe, followed by South Asia; South America, Sub-Saharan Africa, and the Middle East, and North Africa showed around the same percentage of attacks. Western Europe and South Asia are more subjected to facility attacks which could be attributed to better infrastructure. The progression in technology has resulted in the widespread use of information and communication technology in transportation infrastructure, which increases the risk of cyberattacks. Hackers targeted rail systems in the USA, Poland, South Korea, and various other countries. In 2003, a computer virus infected the computer system at CSX Transportation in Florida, disrupting railway signals for several hours (Kour et al. 2020). In 2008, hacking of a tram system resulted in its derailment, which further collided with another tram, thereby injuring 12 people in Poland (Kour et al. 2020). In 2011, another cyberattack disrupted railway signals for two days in the Pacific Northwest. In 2015, a denial-of-service attack was carried out targeting railway infrastructure, among other sectors, to destabilize the Ukrainian government (Liveri et al. 2020). In May 2017, the German railway company Deutsche Bahn became a victim of the WannaCry ransomware, which affected passenger information systems and closed-circuit television (CCTV) networks. A distributed denial-of-service attack on the Danish state rail operator in 2018 affected Denmark’s ticketing systems. In May 2020, Swiss rail vehicle manufacturer Stadler was hit by a malware attack that may have allowed attackers to steal sensitive company data (Liveri et al. 2020). These incidents show the need to respond to and prevent further attacks in the future.

13.3.2 Freight Rails Freight trains are different from passenger trains as they do not contribute to mass casualties on their own due to the low density of passengers within them but can be dangerous in another sense. Freight trains may carry hazardous chemicals and cargo along the passenger train tracks. The alignment of these tracks can pass through highly populated areas, or pass by infrastructures like governmental buildings, religious places, industrial and commercial areas. The size and density of the population are the main factors that terrorists consider while planning an attack (Khanmohamadi et al. 2018). Therefore, any mishaps or attempts to derail a freight train can severely affect the public residing near these tracks. The fire caught in a rail car while passing through Baltimore in 2001 disrupted the rail movement and communications. The derailment of a train carrying liquid fertilizer in North Dakota, in 2002, resulted in toxic cloud formation, and residents had to be evacuated (Riley 2004). Intruders aiming to cause destruction can reroute the cargo, change freight destination, and even cause harm. Hence, similar to passenger trains, freight trains are also vulnerable

238

M. Jayakumar et al.

to malicious attacks to passenger trains, freight trains are also vulnerable to malicious attacks.

13.3.3 Characteristics of Attack Literature highlights the transnational nature of new attacks, which contrasts with the purely national motives of traditional terrorism (Kurtulus 2011). Attackers seek provocation and publicity by using weapons of mass destruction like chemical, biological, radioactive, and nuclear substances (CBRN) for indiscriminate killings. The new terrorism is different from traditional ones, with religion being one of the chief motivators for attacks. These are some features of contemporary terrorism. The security breaches in railways can be categorized into three types: those affecting passengers’ safety, those concerning railway operations, and those affecting both passengers and operations. Attacks affecting passengers’ safety include armed assault, assassination, unarmed assault, hostage-taking, and CBRN use. Attacks affecting railway operations include facility and infrastructure attacks, train schedule database breaches, cyber-attacks, and attacks on railway signaling systems. Attacks on both passengers and operations include bombing or explosion, which is the most commonly used method to cause mass destruction to people and infrastructure. Attacks Targeting Passengers. Armed and unarmed assault, assassination, and hostage-taking are major attacks directed to the passengers. The use of any destructive device or instruments such as guns, firearms, grenades, projectiles, and sharp objects such as a knife, scalpels, rocks, etc., or any other handheld objects to inflict harm or cause death to passengers present within the railway facility can be considered as armed attacks (START: GTD Codebook 2019). Unarmed attacks do not include the use of such instruments to cause physical harm to the passengers. Assassination is also a directed attack exclusively directed to one or more high profile individuals of rank or importance, where the objective is to kill them. The main objective of hostage-taking is to control hostage(s) and hold them for a short or prolonged period by disrupting usual operations to achieve a specific purpose. These attacks may occur within the train, while in motion, or within the railway station premises. CBRN can adversely affect human health, resulting in severe illness or death depending on the type of agent used and the circumstance of exposure (ICRC: Chemical 2014). The deliberate use of CBRN agents by criminals or terrorists can cause panic among the passengers; it may even result in illness or injury, death and generate fear and panic in individuals, groups, or the local population. The Sarin gas attack in the Tokyo subway system on 20 March 1995 killed 12 and injured more than 5000 people (History: Tokyo Subways are Attacked with Sarin Gas 2021). The use of weapons comprising of CBRN substances is possible in the case of contemporary terrorism. Attacks Targeting Railway Operations. Infrastructure attack aims to destroy buildings, individual trains or compartments, tracks, tunnels, bridges, or any other infrastructure related to railways, using any means of explosives. It includes sabotaging railway tracks, platforms, or other facilities, which may or may not result in

13 A Global Perspective of Railway Security

239

passenger fatality. This type of attack may even aim to damage particular installations, pipelines, etc. Attacks on communication and database system may include: tampering attack, where the signaling information from a balize (device to transmit the accurate location of a train to prevent collision of trains traveling on the same track) may be modified to inject false data; cloning attack, where the attacker may try to copy information from one balize onto another, which may result in changes in the timing or degree of breaks applied (Lim et al. 2019); tampering with the railway signaling system which may result in disruption or collision of trains; hacking into the train schedule database to gain access to passenger data and other sensitive information, and various other means making the railways vulnerable. Cyberattacks on railway infrastructure pose a threat to the safety of the infrastructure, which may incidentally affect passengers. Some of the significant types of cybersecurity challenges that threaten the safety, lead to loss of sensitive data and information include malware, DDoS, SQLi, phishing, undisclosed factors, brute force, physical access, and other such factors, the share of which is shown in Fig. 13.3 (IBM 2015). Attacks Targeting both Passengers and Operations. In bombing or explosions, the most important effects are caused by the rapid decomposition of hazardous material, releasing pressure waves, and causing damage to the surrounding environment (e.g., dirty bomb) (START: GTD Codebook 2019). Madrid and London bombings of 2004 and 2005 are examples of such attacks causing massive destruction. Figure 13.3 shows the percentage contribution of each type of attack on railway infrastructure from 1970 to 2019. It can be observed that bombing or explosion contributes to more than 80 percent of attacks on railways which can be attributed to the fact that terrorists may find this the easiest way to cause damage and mass casualty. Malware

2% 17%

Heartbleed DDoS SQLi

1% 40%

XSS

17%

Phishing Physical access Watering hole

9% 3% 4% 1% 5% 1%

Fig. 13.3 Share of cyberattack types. (Data source IBM 2015 (IBM 2015))

240

M. Jayakumar et al.

13.3.4 Countermeasures Against Cyberattacks Ensuring protection from cyberattacks on any railway network requires in-depth knowledge of its operational process and risks. Layered security should be provided to the network, data, identity, and access management to prevent attempts to access these areas by exploiting any security gaps present in the system. Railway personnel should be careful not to create any security gaps unknowingly, stay alert, and report any anomalies to the authorities to avoid potential risks. Some essential elements to ensure security include automation of security processes, end-to-end security, network segmentation and use of firewalls, security analytics using machine learning, and encryption. Ensuring high availability and stability can allow for a speedy recovery from any attacks (Nokia: Cyber Security for Railways 2017).

13.3.5 Countermeasures Against Terrorist Attacks Provision of adequate close circuit television (CCTV) cameras at critical locations and crowded areas, assigning security guards for surveillance of any suspicious behavior or activity from the side of the public, seems to be the best method. The use of sniffer dogs and device detection systems to check passengers and luggage at the platform will detect any dangerous implanted device on the attacker’s person or luggage and concealed dangerous weapons. Preventing public access to railway tracks will help avoid any chance of planting detonating devices on the tracks and train compartments. Installing smoke detectors in the atrium, platforms, waiting areas of the railway station, inside train compartments, washrooms, tunnels, and other critical areas will help detect the presence of smoke and thereby set off fire alarms in the event of an incident. Police patrolling of the areas can help identify any suspicious behavior and protect the rail facility from intruders. Yolmeh and Baykal-Gürsoy (2018) came up with a game theory model to generate schedules for patrolling using a non-cooperative simultaneous move game between a defender and an attacker. Such models will help randomize the patrolling schedules so that it becomes difficult for attackers to cause damage. Creating awareness among the public regarding security issues and creating an environment where they can readily report such suspicious instances can be quite helpful. This technique is widely implemented in countries like the United Kingdom, the United States, France, and Japan (Loukaitou-Sideris et al. 2006). The “See it. Say it. Sorted.” campaign helped to increase reporting of such incidents in the United Kingdom and Denmark (Pearce et al. 2020).

13 A Global Perspective of Railway Security

241

13.4 Security in High-Speed Railway High-Speed Rail (HSR) is currently in function in more than 20 countries worldwide, proving to be one of the most significant present age engineering accomplishments. There are some complex challenges with HSR technology due to its attractive features like open architecture, high concentration of people, rapid and easy movement of patrons. For example, an Alvia high-speed train from Madrid to Ferrol jumped the track on a curve near the railway station at Santiago de Compostela, Spain in 2013. In that crash, out of the 218 people aboard, more than 100 were injured, and 79 died. On 26 August 1995, terrorists, to derail the TGV (Train à Grande Vitesse) between Lyon and Paris, tried to plant a bomb, which coincidentally failed to explode. On 23 July 2011, two high-speed trains traveling on the Yongtai Wen railway line in China collided and derailed each other, leading to the third deadliest HSR accident in history. The mismanagement of the bullet train company led to the killing of 40 people and almost 200 injuries. On 3 June 1998, a fatal high-speed train accident occurred due to derailment in Germany. It went down into a road bridge, killing 101 people and leaving 88 injured. A fatigue fissure in one wheel led to the worst recognized high-speed rail tragedy. In another instance high-speed train ran off its rail in France on 14 November 2015, resulting in 10 deaths and 32 injuries. Excessive speed on a curve due to late braking caused the first fatal derailment in the history of the high-speed train. The 2018 Ankara train collision, in Turkey, killed at least nine people at the scene and injured 84 people. It indicates that the security concern of a railway attack is hardly new. Hence, it is crucial to ponder over the security issues and incorporate preventive measures while developing HSR systems. Terrorists are using new and inexpensive tools of destruction, such as bombcarrying drones or unmanned aerial vehicles (UAV). Developing a capable system to suppress radio channels guiding UAVs without interfering with the communication and operation signals of HSR can protect it from mass destruction (Shvetsov and Shvetsova 2017). Furthermore, controlled access to railway infrastructure, rolling stock, and critical rail systems can improve security (International Union of Railways 2016). In the UK, Eurostar trains have a separate dedicated platform, allowing a special entrance to HSR passengers after screening them and their luggage. In Spain, some HSR trains have ticket control at the entrance of the platform. Other security measures such as ID checks and passenger screening are also practiced by authorities to prevent any mishap. A cloud computing security framework was proposed for the railway environment to tackle security issues (Tan and Ai 2011). An installation of STATCOM can be seen in France’s high-speed rail system. For grids dominated by HSR traction, STATCOM can be used to improve those grids’ power quality (Grünbaum et al. 2009). To improve the security of Long Term Evolution (LTE) based future HSR, three important security schemes are adopted that are advantageous over current authentication schemes (Wang et al. 2020). There are right of way fences in France and no level road crossings in HSR infrastructure, which significantly reduce fatalities. As a result of many pedestrian accidents, Amtrak Acela in the United States installed fences and eliminated extraneous infrastructure. Similarly, China has

242

M. Jayakumar et al.

adopted a fail-safe concept in its railway’s signaling system, facilitating protection from any unprecedented event and allowing automatic guidance to safety.

13.5 Indian Railway System

Running track kilometres (in thousands)

The transportation hub in India every day serves millions of long-distance train travelers. As shown in Fig. 13.4, the running track kilometers of Indian railways increased from 59,315 to 95,981 km in the last seven decades (CEIC: India Length of Railways: Running Track Kilometres 2021). While India continues to demonstrate remarkable growth of the rail system, past attacks, such as 1991 in Punjab (Ludhiana), 1996 in Assam (Brahmaputra mail), 2002 in Uttar Pradesh (Jaunpur), 2003 in Maharashtra (Mumbai), 2014 in Tamil Nadu (Madras), and 2017 in Madhya Pradesh (Bhopal), exposes the vulnerabilities. The historical events (see Fig. 13.5) indicate that terrorists aim to attack railways, particularly the commuter rails (SATP 2001). Hence, it is critical to examine the implications of preventive measures for better security in Indian railways. The railway authorities in India had strengthened security arrangements, but some gaps still exist in the current Indian Railway Security System (IRSS). Researchers developed scenarios to minimize threats to IRSS and analyzed those scenarios using the Delphi technique and Harva method, which helped to identify an action plan for the dominant scenario using fuzzy dominance (Srivastava et al. 2017). Ad hoc Wireless Networks using WiMAX (Worldwide Interoperability for Microwave Access) can reduce the rate of accidents on railway lines (Nallakaruppan et al. 2014). The cause of derailments that occurred in the past is mainly due to cracks in rail. To detect the exact location of such cracks without any human intervention, “Advanced Niyantran” can be a cost-effective and efficient solution (Mujawar and Borkar 2017). 105 95 85 75 65 55

Year Fig. 13.4 Running track kilometers of Indian railways from 1951 to 2019

13 A Global Perspective of Railway Security

243

Fig. 13.5 Attacks on railways in India from 1996 to 2018

Similarly, the Railway Defender Kill Chain (RDKC) can be useful in minimizing cyberattacks in railways.

13.6 Policy Frameworks The European Programme for Critical Infrastructure Protection (EPCIP) recognizes terrorism as a priority and includes several programs for the protection of critical infrastructures in Europe based on an all-hazard approach (Matsika et al. 2016). On the other hand, the European Union Agency for Network and Information Security (ENISA) focuses on developing cybersecurity measures for the railway sector (Liveri et al. 2020). The European Commission co-funded SECRET project to evaluate the risk of electromagnetic attacks on the rail infrastructure and develop preventive and recovery measures for a secured rail network. Similarly, the USA established the

244

M. Jayakumar et al.

Homeland Security Presidential Directive (HSPD-7) to protect the critical infrastructures from terrorist attacks. The Department of Homeland Security, USA identified several critical infrastructures and designated a federal Sector-Specific Agency (SSA) for its protection (CISA 2021). Further, they established a Risk Analysis and Management for Critical Asset Protection (RAMCAP) framework to analyze and manage threats against critical infrastructures (Moore et al. 2007). The National Strategy for Critical Infrastructure Protection, Canada, has a framework for strengthening the critical infrastructure within the country (Matsika et al. 2016). In India, the National Critical Information Infrastructure Protection Centre (NCIIPC), created under Sect. 70A of the Information Technology Act, 2000, aims to take precautionary measures to protect critical infrastructures from unauthorized access, modification, use, and disruption. It lays guidelines for cybersecurity and helps create information security awareness among all stakeholders (NCIIP 2021). These organizations also address the various safety and security issues faced by modern railways. Some reflections on how to tackle the inherent vulnerabilities of HSR can be obtained from inter-organizational collaboration at different levels. The existing security hierarchy of Indian Railways comprises various components such as Special Intelligence Branch, Crime Intelligence Branch, Railway Protection Force (RPF), and Government Railway Police (GRP). For maintaining additional security and intelligence input within railways, organizations such as Counter-Terrorism and Counter Radicalization Division and Cyber and Information Security Division can be roped in to tackle terrorism and cybersecurity, respectively. The HSR requires a necessary shift from the policy-as-usual approach. The Counter-Terrorism and Counter Radicalization Division and Cyber and Information Security Division can be involved in the HSR security framework development.

13.7 Conclusions and Future Research Scope The railway infrastructure is part of a country’s critical infrastructure as it helps in the mass transportation of both passengers and goods. Thus, the open built environment of railway infrastructure makes it a prime target of various terrorism and cyberbased attacks. Over the years, railway infrastructure around the world was subject to various types of attacks: bombings, derailment, armed attacks, CBRN-based attacks, to name a few. Such attacks lead to a wide range of losses to the public, the organization, and the nation. Therefore, all measures must be taken to protect these critical infrastructures. The need of the hour is to assess the risks and vulnerabilities associated with railways and develop countermeasures against new forms of attacks. It is a complex task, which requires further studies to carefully develop new technologies and policy frameworks to ensure railway infrastructure safety. While it may not always be possible to ensure the complete safety of railways, insights can be gained from the previous incidents. It would help to formulate appropriate policies and measures for safer railway infrastructures. Such policies and measures should enhance the safety and security of railway infrastructures and passengers.

13 A Global Perspective of Railway Security

245

The security system in railways is evolving over time. So are the miscreants intending to disrupt the normal operations of railways. The concerned agencies and research communities can collaboratively work for improving the security of railways. It will require developing a systematic data collection plan for a comprehensive database on past incidences, capacity building, and a dedicated funded research program. Available security-related past incidents can be reviewed to reveal its trends, correlation with geopolitical events, effectiveness of mitigation strategies, preparedness of post-event response, etc. One of the important aspects of security is effective surveillance to identify potential threat prior to its occurrences. Researchers can focus on developing simulation models to review and rate the implementable security protocols for electronic and manual surveillance. A systematic and collaborative approach addressing the challenges related to the security of railway infrastructure can make railway transportation safe and secure.

References Aradau C (2010) Security that matters: Critical infrastructure and objects of protection. Secur Dialogue 41(5):491–514. https://doi.org/10.1177/0967010610382687 BBC News: belarus: explosion at minsk metro system. https://www.bbc.com/news/av/world-eur ope-13042262. Accessed 02 Mar 2021 BBC News: brussels attacks: zaventem and maelbeek bombs kill many. https://www.bbc.com/news/ world-europe-35869254. Accessed 02 Mar 2021 Carlson AH, Frincke D, Laude MJ (2003) Railway security issues: a survey of developing railway technology. In: Proceedings of the international conference computer communication control technology, vol. 1, pp 1–6 CEIC: India Length of Railways: Running Track Kilometres. https://www.ceicdata.com/en/india/ length-of-railways-running-track-kilometres. Accessed 18 April 2021 CISA: Homeland Security Presidential Directive 7. https://www.cisa.gov/homeland-security-presid ential-directive-7. Accessed 04 Mar 2021 The Economic Times: Six Terror Attacks that Shook India. https://economictimes.indiatimes.com/ news/defence/six-terror-attacks-that-shook-india/2006-mumbai-train-bombing/slideshow/741 46269.cms. Accessed 02 Mar 2021 Grünbaum R, Hasler JP, Larsson T, Meslay M (2009) STATCOM to enhance power quality and security of rail traction supply. In: 8th international symposium on advanced electromechanical motion systems and electric drives joint symposium ELECTROMOTION 2009, pp 1–3. https:// doi.org/10.1109/ELECTROMOTION.2009.5259136 History: Tokyo Subways are Attacked with Sarin Gas. https://www.history.com/this-day-in-history/ tokyo-subways-are-attacked-with-sarin-gas. Accessed 03 Mar 2021 IBEF: railways. https://www.ibef.org/download/Railways-January-2021.pdf (2021) (Accessed 26 Mar 2021) IBM: IBM X-Force Threat Intelligence Quarterly, 1Q 2015 (2015). https://essextec.com/wp-con tent/uploads/2015/09/IBM-X-Force-Threat-Intelligence-Quarterly-Q1_2015.pdf. Accessed 02 Mar 2021 ICRC: Chemical, Biological, Radiological and Nuclear Response Introductory guidance. In: International committe of the red cross, Geneva, Switzerland (2014). https://shop.icrc.org/chemicalbiological-radiological-and-nuclear-response-introductory-guidance-pdf-en. Accessed 10 Mar 2021

246

M. Jayakumar et al.

Jenkins BM (2001) Protecting public surface transportation against terrorism and serious crime: an executive overview. https://transweb.sjsu.edu/sites/default/files/01-14.pdf. Accessed 02 Mar 2021 Khanmohamadi M, Bagheri M, Khademi N, Ghannadpour SF (2018) A security vulnerability analysis model for dangerous goods transportation by rail – case study: chlorine transportation in texas-illinois. Saf Sci 110:230–241. https://doi.org/10.1016/j.ssci.2018.04.026 Kour R, Karim R, Thaduri A (2020) Cybersecurity for railways – a maturity model. Proc Inst Mech Eng Part F J Rail Rapid Transit 234(10), 1129–1148 (2020). https://doi.org/10.1177/095440971 9881849 Kurtulus EN (2011) The ‘New terrorism’ and its critics. Stud. Confl. Terror. 34(6):476–500 (2011). https://doi.org/10.1080/1057610X.2011.571194. Lim HW, Temple WG, Tran BAN, Chen B, Kalbarczyk Z, Zhou J (2019) Data integrity threats and countermeasures in railway spot transmission systems. ACM Trans Cyber-Phys Syst 4(1). https://doi.org/10.1145/3300179 Liveri D, Theocharidou M, Naydenov R (2020) ENISA: railway cybersecurity – security measures in the railway transport sector (2020). https://doi.org/10.2824/235164 Loukaitou-Sideris A, Taylor BD, Fink CNY (2006) Rail transit security in an international context. Urban Aff Rev 41(6):727–748. https://doi.org/10.1177/1078087406287581 Matsika E, O’Neill C, Battista U, Khosravi M, Laporte ADS, Munoz E (2016) Development of risk assessment specifications for analysing terrorist attacks vulnerability on metro and light rail systems. Transp Res Proc 14:1345–1354. https://doi.org/10.1016/j.trpro.2016.05.207 Miller E, LaFree G, Dugan L (2021) Global terrorism database (GTD). https://www.start.umd.edu/ data-tools/global-terrorism-database-gtd. Accessed 06 Mar 2021 Moore DA, Fuller B, Hazzan M, Jones JW (2007) Development of a security vulnerability assessment process for the RAMCAP chemical sector. J Hazard Mater 142(3):689–694. https://doi.org/ 10.1016/j.jhazmat.2006.06.133 Mujawar M, Borkar S (2017) Design and implementation of wireless security system for railway tracks. In: IEEE international conference on power, control, signals and instrumentation engineering, pp 772–776. Nallakaruppan MK, Senthil MK, Chandrasegar T, Suraj KA, Magesh G (2014) Accident avoidance in railway tracks using adhoc wireless networks. Int J Appl Eng Res 9(21):9551–9556 NCIIP: National Critical Information Infrastructure Protection Centre, Government of India. https:// nciipc.gov.in/. Accessed 04 Mar 2021 Nokia: Cyber Security for Railways (2017). https://www.terrapinn-cdn.com/exhibition/asia-pacificrail/Data/cyber-security-for-railways.pdf. Accessed 07 Mar 2021 International Union of Railways: Rail High Speed Network - Security Handbook (2016). https:// uic.org/IMG/pdf/2015-hs-security_handbook_public.pdf. Accessed 12 Mar 2021 Pearce JM, Parker D, Lindekilde L, Bouhana N, Rogers MB (2020) Encouraging public reporting of suspicious behaviour on rail networks. Polic Soc 30(7):835–853. https://doi.org/10.1080/104 39463.2019.1607340 Ray M (2004) Madrid train bombings of 2004, Terrorist attacks, Spain. https://www.britannica. com/event/Madrid-train-bombings-of-2004. Accessed 02 Mar 2021 Riley J (2004) Terrorism and rail security (2004). http://www.rand.org/ Accessed 04 Mar 2021 SATP: Terrorist Attacks on Railways in India. South Asia Terrorism Portal (2001). https://www. satp.org/satporgtp/countries/india/database/railwayattack.htm. Accessed 15 Feb 2021 Setola R, Sforza A, Vittorini V, Pragliola C (2015). Railway Infrastructure Security Springer. https:// doi.org/10.1007/978-3-319-04426-2 Shvetsov AV, Shvetsova SV (2017) Research of a problem of terrorist attacks in the metro (Subway, U-Bahn, Underground, MRT, Rapid Transit, Metrorail). Eur J Secur Res 2(2):131–145. https:// doi.org/10.1007/s41125-017-0019-3 Srivastava A, Gaur SK, Swami S, Banwet DK (2017) Projecting futuristic scenarios for Indian railway security system (I.R.S.S.) using fuzzy dominance and contingency planning. J Adv Manag Res (2017). https://doi.org/10.1108/JAMR-05-2017-0056

13 A Global Perspective of Railway Security

247

START: GTD Codebook: inclusion criteria and variables. University of Maryland (2019). http:// www.start.umd.edu/gtd/downloads/Codebook.pdf. Accessed 10 Mar 2021 Tan X, Ai B (2011) The issues of cloud computing security in high-speed railway. Proceedings of the 2011 International Conference Electron. Mechnical Engineerings Information Technology EMEIT 2011, vol 8, pp 4358–4363. https://doi.org/10.1109/EMEIT.2011.6023923 Wang Y, Zhang W, Wang X, Guo W, Khan MK, Fan P (2020) Improving the security of LTE-R for high-speed railway: from the access authentication view. IEEE Trans Intell Transp Syst 1–15. https://doi.org/10.1109/tits.2020.3024684 Yolmeh A, Baykal-Gürsoy M (2018) Urban rail patrolling: a game theoretic approach. J Transp Secur 11(1–2): 23–40 (2018). https://doi.org/10.1007/s12198-018-0187-z

Chapter 14

What Drives the Battery-Electric-Bus Introduction in Indian Setting: Operators Perspective and Way Forward Bandhan Bandhu Majumdar , Prasanta K. Sahu , and Dimitris Potoglou

14.1 Introduction The significant increase of the urban population coupled with parallel trends in motorization has been affecting the quality of life (QOL) in Indian cities. The public transportation (PT) system—particularly, the conventional bus system, which relies on fossil fuel as the energy source contributes to poor local air pollution. The air quality has been continuously deteriorating because diesel buses generate significantly higher levels of particulate matter (PM) than other fuels and emit a higher amount of nitrogen oxides (NOx) compared to other vehicle technologies (Shrivastava et al. 2013). These pollutants are the primary factors and actors for visibility reduction, environmental degradation, and health hazards (Re¸sito˘glu et al. 2015). At present, several Indian cities feature among the world’s 20 most polluted cities including Delhi—the capital of India (IQ Air 2021; Times of India 2021). Delhi— the capital city receives 168 µg/m3 of PM2.5 and 394 µg/m3 of PM10, whereas the WHO permissible safe limits for PM2.5 and PM10 are 10 µg/m3 and 20 µg/m3, respectively (IQ Air 2021). Motorized traffic in metropolitan cities accounts for 70% of CO, 50% of HC, 30–40% of NOx, 30% of SPM, and 10% of SO2 of the total pollution, where the significant amount is added from the diesel or CNG buses (Shrivastava et al. 2013; Sood 2012). Among the top ten world list of the most congested cities four megacities: Bangalore (world ranking 1), Mumbai, Pune, and Delhi are present, and the travel time during peak hours in these cities increases by 71%, 65%, 59%, and 56%, respectively (Times of India 2021). Additionally, bus B. B. Majumdar (B) · P. K. Sahu Birla Institute of Technology and Science Pilani, Hyderabad Campus, Hyderabad, Telangana 500078, India D. Potoglou School of Geography and Planning, Cardiff University, Wales Cardiff C10 3WA, UK e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_14

249

250

B. B. Majumdar et al.

and suburban train systems are overcrowded, unreliable, slow, inconvenient, inaccessible, and unsafe (Pucher et al. 2005; Sahu et al. 2018). The low performance of PT eventually contributes to congestion, noise, greenhouse gas (GHG) emissions, traffic accident, lower commuter productivity, and energy wastage. Past research efforts indicate that exposure to diesel emission leads to respiratory problems, lung impairment, and cancer in humans (Burr and Gregory 2011; Lewtas 2007; Lloyd and Cackette 2001; Sydbom et al. 2001; Whichmann 2006). It is evident that this conventional oil-dependent transportation system gives rise to multifaceted externalities and therefore, alternate fuel technology is central to curb the issues derived from the transportation externalities. Although Compressed Natural Gas (CNG) buses emit lower PM and hydrocarbons, the CO emission is five times more in case of diesel buses (The Hindu 2021). Nonetheless to mention that CO damages the cardiovascular system (Lee et al. 2015); about 62.5 million years of life were lost due to cardiovascular disease in India during 2016 (Prabhakaran et al. 2018). Although there are several technological advancements in the area of exhaust emission control, Battery Electric Buses (BEB) with zero tailpipe emissions could be an attractive option for PT use. Shifting from diesel buses to BEB will give rise to a number of potential benefits such as reduced local pollution and noise levels, energy savings, improved human health, QOL, and wellbeing. Some recent research efforts, Zhou et al. suggested that CO2 emissions would be reduced by 3.7 million tons if 150,000 diesel buses were replaced by BEB (Zhou et al. 2016). Padmanaban reported that the revenue from BEB operation could be 82% higher than diesel- or CNG buses, due to their relatively lower operation and maintenance costs relative to diesel/CNG fueled buses (Padmanaban 2016). Notwithstanding the several benefits derived from BEB technology, there are still barriers to its promotion and adoption. The key barriers include very high capital costs and concerns about the fire hazard of the battery in some regions with very high-temperature conditions during summer. Although, with the continuous evolution of BEB technology and successful operation across many parts of the world, it is still at a nascent stage in India. Therefore, prior to the introduction of BEB in India, a detailed investigation of its operational feasibility and operator’s acceptance is essential to understand the key motivators and deterrents of BEB operation for formulating policy guidelines and strategies (e.g., subsidized electricity, lower interest rate for fleet acquisition, capital cost subsidy, fare structure, charging infrastructure, non-pecuniary incentives to users, etc.) to promote BEB in India. To the best of our knowledge, no research effort is made to assess the operational feasibility of BEB in Indian cities or consensus on how and under what conditions a conventional bus may be replaced with a BEB. We analyzed the operational requirements to introduce BEB in cities in India. The findings of this research will be helpful toward achieving the United Nations sustainable development goals: SDG 3—Good health and wellbeing and SDG 11—Sustainable cities and communities. Against this background, the research attempted to: (i) identify and prioritize the key factors those influence the diesel- and CNG bus operation; (ii) assess the importance of operational requirements of factors specific to BEB; (iii) identify the most important motivators of BEB operation from the operator’s

14 What Drives the Battery-Electric-Bus …

251

perspective; (iv) identify the key areas, for which driver/staff training is required for BEB operation in India. The paper has six sections out of which this is the first. Section 2 depicts a short review of the existing research studies. Section 3 describes the data collection process in four smart cities in India: Jaipur, Indore, Bhubaneswar, and Goa. The data collection process is described in Sect. 3. The analytical approach used for the analysis is discussed in Sect. 4. Section 5 interprets and discusses the results. Conclusions and policy recommendations are discussed in Sect. 6.

14.2 Relevant Literature Past research efforts in Europe, USA, and elsewhere on alternative fuel vehicles (AFVs) and their techno-economic feasibility were reviewed to prepare the motivation of the present research. Fowler et al. (1995) conducted an in-depth study to assess the techno-economic feasibility of electric bus operations in Austin, Texas. The authors compared benefits and costs associated with electric, diesel, and CNG bus operations and suggested that an electric bus is a captivating option to diesel technology given requirements for improved air quality and energy sustainability. Ranta et al. (2016) found that e-bus with higher charging times was the most costefficient alternative compared to diesel buses to operate in terrains with steep heels, humidity, precipitation, and snowfall. Johansson and Olsson investigated the possibility of implementing dual-mode buses in Gothenburg’s PT system and highlighted the significant social and environmental benefit derived from such transport operation. The electric bus (e-bus) operation would be successful with the commitment and co-operation from the transit agencies. However, higher initial investment costs for fleet acquisition and staff training are major concerns for successful implementation (Johansson and Olsson 2011). Schwartz evaluated BEB operational feasibility in King County, Canada. They reported that there could be annual savings up to $240 million, $314 million, and $195 million in fuel use, maintenance, and clean fuel credit, respectively, upon shifting toward BEB (Schwartz 2018). Marcon conducted a feasibility study of e-bus in Edmonton, Canada by running field trials on selected routes to assess the impact of e-buses on external factors. He assessed bus users’ perceptions of electric buses to (a) determine how do electric bus features impact the riding quality and comfort; (b) evaluate riders demand to the government to purchase e-buses, and (c) estimate commuter’s willingness to pay for bus service improvement so that the transit agency can purchase e-buses. Marcon suggested a detailed test plan with a focus toward staff training, charging infrastructural requirements, and other facility up-gradation that is needed to be organized prior to e-bus operation implementation (Marcon. 2016). Vilppo and Markkula examined the economic feasibility of e-bus and suggested reducing battery size to lower the overall cost of operation (Vilppo and Markkula 2015).

252

B. B. Majumdar et al.

Noosa Council and Trans-Link Division assessed the techno-economic feasibility of e-bus routes in Noosa to provide adequate evidence to support the E-bus project. They conducted interviews with bus operators and transit agencies to analyze the costs, benefits, risks, opportunities, and issues associated with e-bus operation. The interviews suggested that the agencies are concerned toward the relatively new technology, uncertainty in battery life, operational performance, passenger carrying capacity, and untested design life. They developed an assessment criteria framework to help in: (i) decisions for fleet size, recharging and fleet procurement options; (ii) selecting appropriate routes to maximize coverage, operational efficiency, and minimize the network transfers (GHD 2014). The government of India has been promoting and pursuing the strategy for introducing electric vehicles in India. In September 2017, the government announced the FAME (Faster Adoption and Manufacturing of Hybrid & Electric Vehicles) subsidy program for e-buses. Among Indian cities, Shimla introduced the e-bus commercial operation in October 2017, subsequently, by BEST-Mumbai started the operation of 6 e-buses in November 2017. Ola inaugurated a pilot project in Nagpur to operate a multi-modal electric fleet operation (Indian Express 2018). In December 2017, the government selected 11 cities for a pilot project for multi-modal public transport network. Indian Prime Minister Shri Narendra Modi promised a stable policy regime to the global technology and automobile companies to make India a pioneer in electric mobility. India aims to power all new vehicles through electricity on Indian roads by 2030. Press released reports suggested that many businesses including state owned fuel retailer Indian Oil Corp. Ltd. and Power producer NTPC Ltd have ventured into setting up charging stations to capture the opportunities from electric mobility. India is initially targeting to have 15% of all vehicles be electric-driven by 2020 (Invest India 2018). This smart e-bus is likely to encourage people to shift from their cars and other private modes to make cities greener and cleaner. Despite Indian government’s announcement of several schemes and programs to move toward e-vehicles, hand count research studies are conducted on e-bus operational assessment in India. Global Green Growth Institute and Center for Study of Science, Technology, and Policy published a detailed report on different aspects of e-bus in India (Global Green Growth Institute 2015). Saini and Sarkar assessed the BEB feasibility in Delhi. They conducted both economic and financial analysis on BEB operation for Delhi. Financial analysis suggested to reduce the capital cost of e-bus for its adoption, whereas the economic analysis revealed that total net income would be negative after considering the benefits and operating cost. The negative income is attributed to the higher capital cost (Saini and Sarkar 2017). However, from the operator’s perspective, we did not find any research suggesting the priority factors which influence the BEB introduction in the transit market. We conducted an in-depth investigation of BEB operation in India from different bus-operator agencies’ perspectives. An analysis of the importance and performance of attributes related to the bus operation in the context of diesel, CNG, and BEB was conducted.

14 What Drives the Battery-Electric-Bus …

253

14.3 Data Collection 14.3.1 Questionnaire Design The use of the interview method (IM) with an easy to understand questionnaire helps in collecting reliable and usable data to achieve the research objectives (Saunders 2011). IM also enables the interviewer to direct the needed discussions with the participants while collecting the required information, particularly more complex data (Saunders 2011; Ghashat 2012; Hall and Pain 2006). We designed as an easy to understand questionnaire to collect the operator’s perception on the performance of the Diesel-, CNG bus, and BEB. The questionnaire has four parts (A-D) and the response options are on the Likert scale (5-very satisfactory, 4-satisfactory, 3-neutral, 2-unsatisfactory, and 1-very unsatisfactory). Part A collects the information related to participants (name, designation, years of experience, etc.) and their organization details (name, location, number of operational routes, fleet size, etc.). Part B, C, D collect the satisfaction level of 17 (18 for BEB) attributes related to Diesel-, CNG bus, and BEB operation. Table 14.1 presents the set of selected key attributes related to bus operation and their brief description. The attributes were identified from the past research and expert opinion to elicit the operator’s perception.

14.3.2 Participating Operators and Data Collection Table 14.2 shows the key authorities (operators) who are majorly responsible for the smart-city developmental activities of the study cities (Jaipur, Bhubaneswar, Goa, and Indore), specifically for sustainable mobility planning. These authorities were targeted for the face-to-face interview. The respondents were associated with bus-transport corporation, transport department of respective state governments as well as smart-city authorities, who have already purchased e-bus/initiated the process/interested in procurement. Prior appointments (generally off-peak business hours) were taken from the participants to minimize the respondent’s burden and participation time. Each expert respondent was initially provided a detailed background of the study, a sheet containing brief information on each attribute was then circulated and some individual information such as their role, experience on transit agency, etc., was obtained. During the interview, each part of the questionnaire and questions are explained to obtain reliable and accurate information. A response rate of 50% was achieved during the study. A total of twelve expert data were collected and analyzed.

254

B. B. Majumdar et al.

Table 14.1 List of selected attributes for evaluation and their brief description Sl. No

Factors

Definition

01

Power source

For BEB, the power source is Electricity, whereas for diesel buses, the power source is diesel; for CNG buses CNG

02

Power/mechanical energy generator

For BEB, the power source is Battery, whereas for diesel and CNG buses, the power source is IC Engine

03

Electric bus battery system

There are mainly two types of batteries, namely, Nickel (Ni)-based aqueous batteries and Lithium-ion batteries. Between them, Lithium-ion batteries are extensively used in BEB’s worldwide

04

Acquisition cost

A standard BEB bus costs around 2.6 Crore INR, whereas a diesel and CNG bus costs around INR 20–80 Lakhs

05

Fuel efficiency

Fuel efficiency for BEB: 1.5 kWh/km, whereas for a standard diesel bus 2.2–3.3 KMPL, CNG: 2–2 KMPKG

06

Fuel tariff

BEB: 6.95 INR/kWh | Diesel bus: 50 INR/L | CNG 40 INR/KG

07

Fuel cost

BEB: 10 INR/km | Diesel bus: 15–23 INR/Km; CNG: 13–20 INR/KM

08

Charging infrastructure

Charging of the electric bus can be continuous during operation, at dedicated stations or along the bus route. (NOOSA, 2013). In typical fuel stops the current Diesel and CNG buses are charged

09

Charging Time

The duration required to charge the battery or refuel the tank. The charging time ranges from 0.5 h to 4–5 h for BEB. For diesel/CNG buses, the fuel recharging time It requires 10–15 min for refill of diesel/CNG bus

10

Distance Range

The distance range of a bus is defined as the distance it can go on a single full tank of gas or charge. Electric buses have an average range of 206 km, 421 km for CNG buses; 560 km for diesel buses (Wang and González 2013)

11

Maximum operating speed

BEB has relatively low maximum speed, on an average the maximum speed is observed as 64–73.6 KMPH (Wang and González 2013), which has made it difficult to be used along freeways. A typical diesel/CNG bus can operate at around 100 KMPH (Maximum operating speed) (continued)

14 What Drives the Battery-Electric-Bus …

255

Table 14.1 (continued) Sl. No

Factors

Definition

12

Passenger capacity

The electric, CNG and diesel has been designed with varying passenger capacities and is able to meet the requirements of most transit agencies

13

Emission level

Electric buses do not generate tailpipe emissions, and thereby significantly reduce the ambient pollution load. Diesel buses generate significant tailpipe emission, CNG buses produce less than Diesel buses

14

Noise level

According to existing research, electric buses are the quietest, producing 11.9 dBA above ambient levels on average. The noisiest buses are CNG buses, which emit 34.7 dBA above ambient levels. Diesel buses emit 34.4 decibels above ambient, while diesel hybrids emit 29.9 decibels

15

Overall Journey comfort level

BEB provides a unique, intimate, quiet, clean, and safe ambiance

16

Overall Maintenance cost

Maintenance cost is one of the important considerations while purchasing a bus. For BEB, the cost is lowest

17

Post-purchase availability of spare parts

This is an important consideration before purchasing new buses. Being relatively new in the market, the availability of spare should be considered carefully

18

Staff Training

Adequate training of staffs for BEB operation is important, whereas staffs are accustomed with operation of diesel/CNG buses

14.4 Data Analysis and Results Figures 14.1, 14.2, and 14.3 indicate the list of attributes considered for performance evaluation of Diesel, CNG, and Battery Electric buses and the proportion of expert responses against each scale for the relevant attribute. Based on Figs. 14.1, 14.2, and 14.3, preliminary analysis on operator’s perception toward Diesel, CNG Bus, and BEB can be presented. It can be observed that Fuel charging time is the most satisfactory attribute with around 50% of operators being very satisfied and 41% being satisfied with it with respect to diesel bus operation. Factors such as bus running speed and fuel charging time were also perceived as satisfactorily performing with 90% and 72% of operators being either satisfied or very satisfied with their performance with respect to diesel bus operation. Significant noise and air pollution associated with diesel-operated buses were perceived as least satisfactory attributes (75% and 65% operators perceived as either unsatisfactory or very unsatisfactory respectively), indicating improvement required in such buses

256

B. B. Majumdar et al.

Table 14.2 Participating operators Operator

Location

Participant’s rank

Experience (years)

Kadamba Transport Corporation Limited, Goa

Goa

Assistant Depot 25 Manager

No. of bus routes

Total no. of buses

Participant number

196

544

1

Deputy Financial Controller

36

1

Depot Manager

25

1

Assistant Traffic 36 Superintendent

1

General Manager

36

1

Bhubaneswar Smart City Limited

Bhubaneswar

Assistant General Manager

35

19

150

1

Indore Smart City

Indore

Manager

15

32

288

1

Jaipur City Transport Service Ltd. (JCTSL)

Jaipur

Chief Manager

40

107

500

1

Depot Manager

41

1

Manager Operation

38

1

Project Manager

16

2

for better performance. For CNG-operated buses, bus running speed (100% operators were satisfied), bus operation ranges (18% very satisfied and 82% satisfied), acquisition cost (85% of operators were either very satisfied or satisfied), and CNG as a power source (80% operators found it a satisfactorily performing attribute) were found to be satisfactorily performing attributes from operator’s perspective. However, fuel cost, fuel tariff, and fuel efficiency were perceived as less satisfactorily performing determinants associated with CNG-operated buses followed by emission level, which was perceived to be performing unsatisfactorily by 34% of operators. This finding clearly indicates that fuel cost and efficiency needs to be significantly improved for better CNG bus operation. Among BEB-specific attributes, battery as the power source (90% of users perceived this attribute as very satisfactory or satisfactory), power/mechanical energy generator, and overall journey comfort (90% of users perceived as satisfactory or very satisfactory) were perceived as satisfactory attributes requiring upkeep. Higher acquisition cost, poor charging infrastructure, and higher charging time associated with BEB were perceived to be the least satisfactory attributes. This finding clearly indicates that the provision of subsidy to lower the bus acquisition cost, better charging infrastructure and lower charging duration

14 What Drives the Battery-Electric-Bus …

257

Table 14.3 RIDIT based ranking of attributes related to different bus operations Attributes

Rank based on performance w.r.t Diesel bus operation

Rank based on performance w.r.t CNG bus operation

Rank based on importance w.r.t BEB operation

Power source

6

6

8

Power/mechanical energy generator

8

2

7

Acquisition cost

1

8

18

Fuel efficiency

13

15

5

Fuel tariff

15

16

3

Fuel cost

14

13

2

Fuel Charging infrastructure

4

11

10

Fuel Charging Time

2

5

17

Bus operation Range

9

1

16

Bus running speed

3

4

13

Passenger capacity

4

9

14

Emission level

16

12

6

Noise level

17

13

1

Overall Journey comfort level

10

17

4

Overall Maintenance cost

12

10

9

Post-purchase availability of spare parts

11

3

12

Staff Training

6

7

15

Li-Ion battery for BEB

NA

NA

11

should be prioritized to make BEB an attractive public transport alternative in India. Further analysis is needed for an appropriate policy implication.

14.4.1 Analytical Approach The aim of the study is to know the key attributes related to Diesel/CNG/Battery electric bus operation based on the operator’s perspective in the Indian context. A user-perception-based ranking using mean as an indicator can be used as a tool. However, for Likert-type or ordinal data, the mean cannot be adopted as a tool. Hence, we used a widely adopted multi-attribute decision-making technique, i.e., RIDIT analysis for this research (Bross 1958). RIDIT is a basically short form for

258

B. B. Majumdar et al.

Fig. 14.1 Expert responses against the performance of attributes related to diesel bus operation

Fig. 14.2 Expert responses against the performance of attributes related to CNG bus operation

14 What Drives the Battery-Electric-Bus …

259

Fig. 14.3 Expert responses against the performance of attributes related to BEB operation

“RID— Relative to an Identified Distribution and the suffix and IT represents a type of transformation” (Bross 1958). RIDIT is an appropriate technique for comparative analysis of two sets of “ordered qualitative data”. Among both sets of data, one dataset is assumed as the reference, whereas the other set is used for comparison. Several prior studies have used RIDIT for prioritization applications. Among such studies, RIDIT technique was adopted by Chang and Chang (2005) for prioritization of factors affecting recreational bicyclists’ choice decisions. In a previous Indian application, factors associated with Nano car purchase were prioritized using RIDIT (Bikash and Pravat 2010). RIDIT technique uses a probability relative to a reference distribution as a means for identifying differences among the groups. In this technique, data are categorized into distinct classes, and then the ranking is done on the basis of scaled values of the variables. In the first step, RIDIT identifies a population that is regarded as a reference class (Sadhukhan et al. 2015). The stepwise methodology as given by Wu (2007) is followed in this study. The steps are given below. Step-1: Select a reference dataset. For a Likert scale survey, a reference dataset consists of the overall responses. Step-2: Estimate frequency f j for each class of responses, where j = 1,2, …, n Step-3: Determine the mid-point accumulated frequency F j for each class of responses. F1 =

1 f1 2

(14.1)

260

B. B. Majumdar et al.

Fj =

 j−1 1 fj + f k , W her e k = 1, 2, ..., n k=1 2

(14.2)

Step-4: RIDIT value Rj needs to be estimated for each response category. Rj =

Fj N

W her e j = 1, 2, ..., n

(14.3)

In the above formula, N is the total number of responses. Step-5: Calculate the RIDITs and mean RIDITs for comparative datasets. The frequency of responses for each category of a Likert-type scale item constitutes a comparative data set. (Wu 2007). If there are m Likert-type scale items in the response scale, there will be a total of m comparison data sets. RIDIT value ri j is calculated for every class using the following formula: ri j =

Rj N

W her e j = 1, 2, ..., n

(14.4)

π ij = Frequency of category j for the ith scale item. π i = Sum of frequencies for scale item i across all categories, i.e. πi =

n 

πik

(14.5)

k=1

The mean RIDIT ρ i for each Likert scale item is calculated using the equation below. ρi =

n 

rik

(14.6)

k=1

Subsequently, the confidence interval for the mean RIDIT is also calculated and the ranking of attributes with respect to the RIDIT score is estimated.

14.5 Results and Analysis Here, the RIDIT analysis results are presented for the prioritization of attributes related to Diesel, CNG, and BEB attributes using the RIDIT technique. The analysis is carried out for prioritizing attributes related to diesel and CNG operation with respect to their relative performance and BEB attribute’s importance from operator’s perspective. The detailed methodology as described in Sect. 4 is followed to obtain the prioritized ranking of different attributes. For the limited scope of the manuscript, the detailed analysis is not presented, only the obtained rankings are shown in a tabular manner in Table 5 and subsequently interpreted.

14 What Drives the Battery-Electric-Bus …

261

The results clearly indicate that low acquisition cost, relatively lower fuel charging time, and available charging infrastructure are the better performing attributes related to diesel bus operation. On the other hand, noise and air pollution along with poor fuel efficiency, higher operation cost are the major deterrents of diesel buses as perceived by bus operators in India. Analysis of the importance of CNG buses reveals bus operation range, mechanical or energy generators to be the positive factors of CNG bus operation, whereas journey comfort level and fuel tariff are perceived as key deterrents by major bus operators across India. Analysis of attributes specific to BEB operation clearly indicates that low noise level, less operation cost, and low pollution level are the key motivators toward BEB operation in India. However, relatively higher acquisition cost and higher charging time are the major deterrents of such operation in the Indian context.

14.6 Conclusions and Way Forward In order to understand the operational requirements related to BEB operation, this study has developed and demonstrated a two-step scientific methodology. In the first stage, a questionnaire has been prepared to elicit bus operator’s perception toward various attributes related to (a) diesel bus operation, (b) CNG bus operation and (c) BEB operation. This study has initially selected an exhaustive set of seventeen factors related to diesel and CNG bus operation and eighteen attributes specific to BEB operation in the Indian context. To collect the operator’s responses toward the perceived performance of the identified attributes related to diesel and CNG buses and the perceived importance of the attributes specific to BEB, a standard five-point Likert scale was used. RIDIT was subsequently adopted for the prioritization of the set of factors. On the basis of the interpretation of the results, the discussions can be made as follows. Firstly, it could be clearly inferred that higher air and noise pollution from diesel buses, fuel tariff, and increasing fuel cost are the key concerns for diesel buses from the operator’s perspective in the Indian context. Such findings clearly suggest improved vehicular emission technology implementation for improvement in dieseloperated bus quality. On the other hand, poor fuel efficiency, poor journey comfort level and increased price need to be seriously addressed by the policymakers to make the CNG-run buses more attractive. Secondly, it can also be concluded that reducing the fuel charging time and acquisition cost of battery run electric buses could make these buses attractive alternatives to the policymakers. Low pollution level was observed to be the key motivator related to BEB operation in the Indian context. Finally, the results and related findings could be effectively utilized for the development of planning guidelines for the introduction of BEB operation in the typical Indian setting. In India, there is still no consensus on BEB operation, the results and findings of this study could significantly augment the research in this area and provide useful input to planners. This study also could be useful for the diesel and CNG bus

262

B. B. Majumdar et al.

operators by identifying the key areas of concern, addressing which could be instrumental toward improving the quality of life of Indian commuters. Like any other research, this research has certain limitations, one of them is the lower sample size, which may significantly influence the statistical representativeness of the sample. As a further extension of the present study, more data would be collected and analyzed for more accuracy. Before closing, the author would like to state that although this study attempted to know the major drivers of introducing BEB in Indian cities, the finding needs to be generalized by obtaining relevant data from more cities and other expert groups. Nonetheless, the approach developed could be used to carry out more in-depth analysis with larger sample size. Furthermore, the survey instrument is generic and can be adopted to collect more data in order to develop guidelines for the successful implementation of BEB in other South Asian countries. Funding Acknowledgment Adoption of Battery Electric Buses (BEB) in India, Global Challenges Research Fund, UK—JA2610RA50 GCRF-HEFCW—SMALL PROJECTS 50: 2018–2019.

References Bikash RD, Pravat SK (2010) Factors influencing purchase of “NANO” the innovative car from India-an empirical study. Asian J Business Manag 2(3):48–56 Bross ID (1958) How to use ridit analysis. Biometrics 18–38 Burr M, Gregory C (2011) Vehicular exhaust. Encylopedia. Environ Health 49:645–563 Chang H-L, Chang H-W (2005) Comparison between the differences of recreational cyclists in national scenic bikeway and local bike lane. J Eastern Asia Soc Transport Stud 6:2178–2193 Fowler TM, Euritt MA, Walton CM (1995) Electric bus operations: a feasibility study. Research report (No. PB-95–240453/XAB). Texas University, Austin. Center for Transportation Research Ghashat H (2012) The governance of Libyan ports: determining a framework for successful devolution. Edinburgh Napier University GHD (2014) Prefeasibility study of electric bus routes in Noosa. Noosa Council and Trans-Link Division (2014) Global Green Growth Institute (2015) Centre for Study of Science, Technology and Policy.: Electric Buses in India: Technology, Policy and Benefits, GGGI. Seoul, Republic of Korea (2015) Hall PG, Pain K (2006) The polycentric metropolis: learning from mega-city regions in Europe. Routledge, London https://timesofindia.indiatimes.com/india/bengaluru-has-worlds-worst-traffic-congestion-mum bai-at-number-4/articleshow/73747725.cms. Accessed 17 June 2021 https://www.iqair.com/. Accessed 25 May 2021 https://www.thehindu.com/news/national/kerala/cng-could-ease-bus-operators-burden/article32 007011.ece. Accessed 12 June 2021 Indian Express (2018) Web-link- https://indianexpress.com/article/india/nagpur-becomes-firstcity-with-electric-mass-mobility-system-ola-mahindra-e-vehicle4676750/. Accessed 05 Dec 2018 Invest India (2018) Web-link- https://www.investindia.gov.in/team-india-blogs/indian-electric-veh icle-industry-raring-fly. Accessed 05 Dec 2018 Johansson M, Olsson O (2011) Feasibility study of dual-mode buses in Gothenburg’s public transport” Department of Technology Management and Economics. Chalmers University of Technology, Göteborg, Sweden, Division of operations management

14 What Drives the Battery-Electric-Bus …

263

Lee FY, Chen WK, Lin CL, Kao CH (2015) Carbon monoxide poisoning and subsequent cardiovascular disease risk: a nationwide population-based cohort study. Medicine 94(10). 10.1097/MD.0000000000000624. https://trid.trb.org/view/1566760 Lewtas J (2007) Air pollution combustion emissions: characterization of causative agents and mechanisms associated with cancer, reproductive, and cardiovascular effects. Mutat Res/rev Mutat Res 636(1–3):95–133 Lloyd AC, Cackette TA (2001) Diesel engines: environmental impact and control. J Air Waste Manag Assoc 51(6):809–847 Marcon (2016) Electric bus feasibility study for the city of Edmonton. City council Edmonton Padmanaban D (2016) Electric buses earn 82% more profit than diesel daily. http://archive.indias pend.com/cover-story/electric-buses-earn-82-more-profit-than-diesel-daily-85587. Accessed 05 June 2021 Prabhakaran D, Singh K, Roth GA, Banerjee A, Pagidipati NJ, Huffman MD (2018) Cardiovascular diseases in India compared with the United States. J Am Coll Cardiol 72(1):79–95 Pucher J, Korattyswaropam N, Mittal N, Ittyerah N (2005) Urban transport crisis in India. Transp Policy 12(3):185–198 Ranta P, Rahkola P, Mikko P, Pihlatie V, Weber C, Amundsen A, Hagman R (2016) Feasibility of electric buses in Tromsø. VTT Technical Research Centre of Finland Ltd (2016) Re¸sito˘glu ˙IA, Altini¸sik K, Keskin A (2015) The pollutant emissions from diesel-engine vehicles and exhaust aftertreatment systems. Clean Technol Envir 17:15–27. https://doi.org/10.1007/s10 098-014-0793-9 Sadhukhan S, Banerjee UK, Maitra B (2015) Commuters’ perception towards transfer facility attributes in and around metro stations: experience in Kolkata. J Urban Plann Develop 141(4):04014038 Sahu PK, Sharma G, Guharoy A (2018) Commuter travel cost estimation at different levels of crowding in a suburban rail system: a case study of Mumbai. Public Transp 10(3):379–398. https://doi.org/10.1007/s12469-018-0190-6 Saini P, Sarkar PK (2017) Feasibility of electric bus operation in urban areas, case study- Delhi. School of Planning and Architecture, New Delhi Saunders MN (2011) Research methods for business students. Pearson Education India Schwartz S (2018) Battery bus feasibility. Transportation consultants Shrivastava RK, Neeta S, Geeta G (2013) Air pollution due to road transportation in India: a review on assessment and reduction strategies. J Environ Res Develop 8(1):69 Sood PR (2012) Air pollution through vehicular emissions in urban India and preventive measures. In: International conference on environment, energy and biotechnology, vol 33. IPCBEE Sydbom A, Blomberg A, Parnia S, Stenfors N, Sandström T, Dahlen SE (2001) Health effects of diesel exhaust emissions. Eur Respir J 17(4):733–746 Vilppo O, Markkula J (2015) Feasibility of electric buses in public transport. EVS28 KINTEX, Korea (2015) Wang X, González JA (2013) Assessing feasibility of electric buses in small and medium-sized communities. Int J Sustain Transp 7(6):431–448 Whichmann HE (2006) Environmental pollutants: diesel exhaust particles. Encyclopedia Respir Med 1:96–100 Wu CH (2007) On the application of grey relational analysis and RIDIT analysis to Likert scale surveys. In International Mathematical Forum 2(14):675–687 Zhou B, Wu Y, Zhou B, Wang R, Ke W, Zhang S, Hao J (2016) Real-world performance of battery electric buses and their life-cycle benefits with respect to energy consumption and carbon dioxide emissions. Energy 96:603–613

Chapter 15

Sustainable Freight Transportation Planning and Policies for a Logistics-Driven India: Current State and Future Ahead Agnivesh Pani, Prasanta K. Sahu, and Bandhan Bandhu Majumdar

15.1 Introduction The importance of urban freight transportation in serving the industrial activity in cities and fulfilling the consumer needs for goods and services cannot be overstated. Day in and day out, trucks contribute to the tremendous fuel consumption levels generated in cities around the world by delivering all kinds of products to households, offices, distribution centers, retail stores, and warehouses (Tavasszy 2020). Emissions from freight deliveries are projected to increase by nearly a third, and freight traffic delivery vehicles in the top 100 cities around the world are expected to grow by over 36% in the next decade (Hillyer 2020). Even when the outbreak of the COVID-19 pandemic halted passenger transport to a standstill, freight transport continued to keep the economy afloat and deliver essential goods and services (Loske 2020). Physical shopping trips to brick-and-mortar stores, retail shops, city centers, and restaurants were replaced by online shopping and deliveries to our doorsteps (Harikumar 2020). The post-COVID world is unchartered territory for most of the consumer industries and service sectors, including the freight transport sector (Abu-Rayash and Dincer 2020). It is also not surprising that there are unprecedented changes in preferences and needs of consumers, around which transport planners and policymakers will have to adapt and innovate to foster a sustainable shift in the post-pandemic world. Therefore, if there are any lessons to be learned from this pandemic, among them is its power to serve as a wake-up call to focus on the threat of climate change while we still have time to stave off its most damaging effects. About 50% of a product’s total transport cost consists of last-mile delivery services. The final-mile delivery of an item accounts for about 181gCO2 /km, which A. Pani Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India P. K. Sahu (B) · B. B. Majumdar Birla Institute of Technology and Science, Pilani, Hyderabad Campus, India © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_15

265

266

A. Pani et al.

is much higher than the limits set by the EU in 2017, i.e., 0.175mgCO2 /km (Harris-Burland 2020). Therefore, the conventional form of freight delivery is unsustainable in the long run and needs to be replaced with low-carbon alternatives (Pani et al. 2019a). The COVID-19 crisis, therefore, needs to be treated as a critical chance to rectify the vulnerabilities in freight transport systems, especially in the last-mile delivery services. Therefore, adopting lowcarbon freight transport options (crowd-shipping, cargo bikes, electric trucks, autonomous robots, etc.) is a direct response to the calls for green and sustainable recovery from the COVID-19 pandemic (International Institute for Sustainable Development: Green Recovery: ADB outlines pathway to a low-carbon and resilient future, https://www.iisd.org/sustainable-recovery/news/green-recoveryadb-outlines-pathway-to-low-carbon-and-resilient-future). The subsequent sections take a closer look at the various sustainability measures implemented in freight operations in different countries and the lessons learnt from these initiatives. The public response to these innovations, the early-stage reliability of the systems, and the underlying potential for cutting GHG emissions are critical factors that need to be investigated for designing improved low-carbon logistics strategies and tackling the climate change crisis already underway. This book chapter outlines the status of freight data collection efforts and freight traffic impacts in India and then subsequently elaborates on the emerging research themes and key recommendations for planning and policy interventions. The insights provided in this chapter are expected to promote freight research and effective policy instrument design and help address the growing needs to reduce the overall logistics cost for moving goods in India. The remainder of this chapter is organized as follows. Section 15.2 provides a comprehensive overview of freight surveys and the latest data collection efforts in India. Section 15.3 describes the overview of freight traffic impacts in India, and Sect. 15.4 explains the emerging research themes and key planning interventions with some key statistics relevant to India. Section 15.5 discusses the freight policy measures summarized in three broad categories and provides the key implications related to the regulation of freight traffic in India. The final section discusses the concluding remarks.

15.2 Freight Data in India 15.2.1 Disaggregate Level Freight Surveys Freight surveys are mainly intended to address the data needs associated with modeling urban freight movement at (a) city level, (b) establishment level, and (c) supply chain level (Kriger et al. 2011). It can be seen that most of the freight surveys focus on the freight flow at the establishment level (Allen and Browne 2008). This is possibly due to the fact that establishments are the root cause of goods flow in urban

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

267

areas. Another advantage of this practice is that the collected data may help in representing commodity flow as well. Nonetheless, there are only a handful of studies, which considered vehicle trips or supply chain links at the city level as the focus of the survey. The state of practice of freight survey techniques includes several techniques, as outlined in Table 15.1. As can be seen, an establishment-based freight survey is the most commonly adopted survey technique for collecting data about urban freight movements (Pani et al. 2018; Pani and Sahu 2019a, b). The strengths, weaknesses, applications, and data obtainability of these survey techniques are summarized in this table.

15.2.2 Recent Advances in Freight Surveys in India Data and sources of information related to urban freight transportation are quite scarce in a developing country like India. The interest in data collection and modeling for urban freight has, however, increased in the recent years with notable studies emerging for cities in Kerala (Pani et al. 2018, 2020a; Pani and Sahu 2019b, c), Rajasthan (Chandra et al. 2020, 2021a; Pani et al. 2019b), Tamil Nadu (Venkadavarahan et al. 2020; Middela and Ramadurai 2020), Gujarat (Dhonde and Patel 2021), and Delhi (Malik et al. 2017, 2019). However, urban freight data collection efforts at a national level are still missing, despite the recent surveys in different parts of the country; thus, India is still lacking comprehensive freight manuals related to urban freight flow and impact assessments, such as the Cooperative Freight Research Program (NCFRP) reports in the USA (Holguín-Veras et al. 2012, 2016). Comprehensive survey programs are already available in Europe as well, with experimental research initiatives and networking activities supported by the European Union, such as BESTUFS (BEST Urban Freight Solutions) which provides harmonized freight data for researchers and practitioners (Allen et al. 2007). Much like these public– private partnerships fostering freight studies and surveys around the world, India also requires forums and Government level initiatives to encourage freight data sharing between logistics providers and transport practitioners in the interest of improving the urban environment.

15.3 Freight Traffic Impacts in India Among the costs created by freight transportation, some of them, namely negative externalities, fall on the society at large and the overall environment that has no direct relevance to the operations of transportation. These costs are not borne by those causing them (e.g., shippers, receivers, or logistics service providers) and are not reflected in any economic transaction (i.e., when the good is produced, transported,

268

A. Pani et al.

Table 15.1 Summary of Freight Survey techniques Sl. no

Freight Survey Technique

Summary and application

Type of available or collected data

1

Establishment-based Freight Survey (EBFS)

The fundamental unit of freight trip making, i.e., business units are surveyed by inquiring about delivery and collection of goods. This is deemed to be the most adequate source of information to model freight demand (Alho and de Abreu e Silva 2014)

• Freight delivery/collection • Commodity types • Establishment size variables • Trip origin and destination • Loading/unloading activity

2

Commodity flow survey Analogous to EBFS, except (CFS) that CFS is specifically intended to collect goods flow information (in tons) than focusing on truck trips. CFS data is collected quinquennially in USA for supporting public policymaking and transportation planning (Shabani and Figliozzi 2012)

• Freight delivery/collection in tons • Commodity types, Shipment value • Establishment size variables • Trip origin and destination • Mode of transport

3

Roadside or Intercept Survey

Interviews of truck drivers • Trip origin and destination along the highway with due • Goods characteristics assistance and permission of • Vehicle type traffic police or enforcement agencies (Combes and Leurent 2013)

4

Freight Operator or Carrier Survey

Used for collection of • Origin and destination of wide-ranging data about the goods flow • Transportation cost pattern of goods activity in particulars an area by surveying logistic • Loading and unloading service providers (Rowell activity et al. 2014)

5

Vehicle Trip Diary Survey

Mainly used to collect detailed whereabouts of trucks (commercial services and movement of goods) which require drivers to enumerate trip data (Kulpa 2014)

• Goods types and locations (destinations) served • Service times at locations • Arrival and departure times

(continued)

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

269

Table 15.1 (continued) Sl. no

Freight Survey Technique

Summary and application

Type of available or collected data

6

GPS Survey

The latitude and longitude of • Vehicle speed the GPS receiver attached • Locations served, Trucking on-board to the vehicle are routes • Service times at locations determined using satellite and crossing times transmissions. These surveys are similar to vehicle trip diaries, although more reliable than the latter (Allen et al. 2012)

7

License Plate Survey

Mainly used for • Trip origin and destination corridor/expressway • Vehicle type entry–exit analysis by • Entry and exit time recording vehicle’s license plate as it passes through two or more points along the link (Allen and Browne 2008)

8

Parking Survey

Surveyors positioned at parking spaces capture the information about heavy vehicle parking activity (Jaller et al. 2013)

9

Truck Traffic Count Survey

Traffic counts collected from • Truck traffic count by time selected routes or cordons of day, day of week, etc are disaggregated by vehicle • Truck traffic proportion to total traffic flow type into different types of trucks (Muñuzuri et al. 2009)

10

Administrative Survey

Conducted for • Establishment administrative purposes such characteristics, such as as financial and institutional employment, area, and year record-keeping (Kriger et al. of establishment 2011)

11

Stakeholder Survey

Conducted for soliciting qualitative comments about different facets of freight system (Kriger et al. 2011). The survey is conducted with establishments, terminals, ports, carriers, distribution centers, and infrastructure planners

• Loading/unloading time • Parking space needs of truck type or by commodity

• Qualitative data about policy impacts • Supply chain links • Logistical needs of private sector

270

A. Pani et al.

or consumed). These external costs can be broadly categorized into three (Demir et al. 2015): (i) environmental impacts, (ii) social impacts, and (iii) economic impacts, as explained below.

15.3.1 Environmental Impacts The environmental impacts from freight transport include air pollution, climate change caused by greenhouse gas emissions, noise, and water pollution. Air pollution is caused by emissions of particulate matter (i.e., microscopic solid or liquid particles in the air), carbon monoxide, ozone, and hazardous air pollutants such as benzene, which causes cancer and other serious health effects. Most trucks run on diesel, which is more polluting than petrol. Climate change is caused by GHGs. Excessive noise can negatively impact human health, disturb sleep, and cause cardiovascular and psychophysiological problems (World Health Organization: Noise 2021). Most of the external costs from trucks in Europe come from noise (Santos 2005, 2017). Water pollution can result from freight transport when there are spills, leakages, or disposal of cargo material in water bodies. Although freight traffic constitutes merely 3–15% of total traffic in urban arterials and expressways (Bharadwaj et al. 2018), it is estimated to be responsible for up to 50% of road transport emissions (Dablanc 2007). In the case of noise pollution and vibration hindrance as well, in general, road freight has a much larger impact than cars. Compounding these impacts is the fact that freight trucks used for urban deliveries are generally older and more polluting than trucks used for long-haul shipments (Coulombel et al. 2018). Finally, land-use changes associated with freight flow and transport infrastructure development are an increasing source of concern as they can cause visual intrusion on the environmental landscape, destruction of habitats, and species loss.

15.3.2 Social Impact The main negative externality from freight transportation with regard to social impacts is road crashes. A significant share of crashes can be attributed to trucks, as shown in Fig. 15.1, based on crash data published by the Transportation Research and Injury Prevention Program at the Indian Institute of Technology in Delhi (Mohan et al. 2015). The vehicle types include motorized two wheelers (MTWs), three-wheeled scooter taxi (TST), bus, cars, trucks, and others. As it can be seen in Fig. 15.1, 72 and 65% of fatal crashes in 6-lane National Highways and urban highways are associated with trucks as one of the impacting vehicles.

Proportion of Impacting Vehicle Type in Fatal Crashes (%)

15 Sustainable Freight Transportation Planning and Policies for a Logistics … 80

72%

Highways 2-lane NH 4-lane NH 6-lane NH

70 60

271

65%

50 40 30 20 10 0 Bus

Car

TSTs

MTWs

Others

Truck

Vehicle Type

Fig. 15.1 Proportion of Impacting Vehicle Type in Fatal Crashes from 2015 to 2018. Source Data from Transportation Research and Injury Prevention Report (Mohan et al. 2015)

15.3.3 Economic Impact The main externality from road freight that has economic impacts is congestion. Congestion caused by road freight has become a common problem in cities around the world (Alho et al. 2018). In Europe, most of the external cost from trucks comes from congestion (and noise) (Santos 2017). Trucks take between 2 and 4 times the road space that cars take, and their speeds also tend to be lower. In addition, due to scarcity or inadequate configuration of loading or unloading bays/zones, freight trucks often double-park during their delivery tours (Edoardo and Scaccia 2015), thereby blocking the road for other vehicles. Traffic congestion has substantial negative impacts in terms of reduced productivity and wasted fuel. A high-level estimate of the economic loss resulting from congestion in major cities in India is over 22 billion USD, annually. Two conflicting interests emerge regarding congestion—public authorities aim to reduce freight traffic to improve the attractiveness of their city to residents as well as tourists, while private companies seek to operate at the lowest cost with quick deliveries to satisfy consumers’ expectations in a highly competitive market (Schliwa et al. 2015). The regulations and restrictions brought by public authorities can cause a ‘detour’ of delivery tours through narrow streets and unsafe delivery areas with low vertical clearance, further amplifying traffic congestion (Vidal Vieira and Fransoo 2015).

272

A. Pani et al.

15.4 Emerging Research Themes and Planning Interventions Considering the enormous impact of freight traffic in India, a portfolio of planning and policy solutions needs to be proposed and implemented to address the negative externalities and inefficiencies in freight transport. While the planning solutions aim to reduce the impact of freight transport within the existing expanse of transport infrastructure, policy interventions play a critical role in terms of energy demand and economic consequences. This section outlines the key emerging research themes and recommendations for freight transportation planning in developing countries like India.

15.4.1 Freight Traffic Regulations and Truck Electrification The government departments, such as the Ministry of Shipping and Logistics, can employ a wide range of policy measures, ranging from taxation instruments (e.g., fuel taxes, excise tax, and toll) to financial incentives (e.g., tax rebates for supporting greener modes, capital grants) and regulation orders (e.g., vehicle design, entry time, emission standards). Vehicle regulations are among the most common policy response taken by public authorities when freight traffic is sharing the same right of way with passenger traffic (Fernandes et al. 2016). Implementation of these vehicle regulations, predominantly imposed as rout-specific restrictions, without the provision of urban logistics spaces or urban consolidation centers are found to have negative impacts on the regional economy. Besides, these restrictions are often found to turn counterproductive due to increased delivery activity using small vans. To overcome the need for such restrictions, significant advances in the fuel efficiency of trucks and reductions in the daily emission levels are achieved through emission filters. As a means to overcome these restrictions, electric trucks (ETs) are also becoming more and more scalable and viable alternatives to diesel-powered trucks around the world, including India (Anderhofstadt and Spinler 2019; Feng and Figliozzi 2012). Fostering the replacement of traditional truck fleets with ETs through incentive schemes and tax reductions will significantly reduce the negative externalities of freight transport. Recent policy initiatives by the Indian government, such as ‘Faster Adoption and Manufacturing of Hybrid and Electric Vehicles’ (FAME), are a valuable step toward fostering technological advancements in freight transport.

15.4.2 Freight Modal Shift to Bike-Based Alternatives As a distribution solution, non-motorized cargo bikes were conceptualized in the last decade to deliver last-mile packages on bike routes. The loading rate of cargo

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

273

bikes typically ranges up to 100 kg, and large-scale implementation at a city level is projected to decrease last-mile delivery cost by up to 45%. These are zero-emission alternatives to light goods vehicles but have several adoption challenges. There is also an increasing market for the use of electric cargo bikes in several parts of the world, especially in the European countries with dedicated bike lanes. From the research into the feasibility of adopting carbon–neutral bike deliveries of goods, the European Union concludes that about 25% of all goods can be delivered by cargo bikes in urban settings. As per an ITDP internal survey report, many cities have low but growing use of e-bikes, whereas a few cities in Brazil are reported to have higher levels of e-cargo bike delivery (Yanocha and Allan 2019). In Copacabana, Rio de Janeiro, Brazil, about 11,000 deliveries are made each day by means of carbo bikes (World Bank Blogs 2020). In a 2011 study on cargo bike deliveries in Copacabana and surrounding regions, researchers identified 372 establishments that used cargo bikes. An overview of cargo bikes and the weight of the cargo delivered by bike trips are as shown in Figs. 15.2 and 15.3 based on the Rio de Janeiro study (Hagen et al. 2013). The advent of apps and extensions such as EcoCart (https://apps.shopify.com/ ecocart) will further boost the adoption of cargo bikes because it empowers the consumers to make their deliveries carbon neutral. Once the delivery apps show the greenhouse gas (GHG) emission avoided by choosing a low or zero-carbon delivery method, it is expected that a sustainable shift in last-mile delivery choices can be fostered among the consumers. Data-driven tools for calculating the exact cost of offsetting the emissions from each package delivery will also help to inform the consumers about the impact of their internet orders and assist the global efforts to curtail GHG emissions.

Fig. 15.2 Overview of cargo bikes and percentage of bike deliveries in Copacabana, Brazil

274

A. Pani et al.

Fig. 15.3 Overview of crowd-shipping services for last-mile delivery

15.4.3 Last-Mile Delivery Management Using Crowd-Shipping Initiatives Crowd-shipping is a concept that incorporates underutilized passenger transport mode capacity and related infrastructure to deliver freight packages in the last mile (Gatta et al. 2019). Packages are delivered with the help of regular passengers (i.e., transit users, car owners) wherein they drop off packages at designated places on their way, which will, in turn, be picked up by another commuter and delivered to the customer at the destination (see Fig. 15.2). Crowd-shipping follows the principles of the sharing economy and can be a front runner for low-carbon logistics in the years to come, especially for last-mile deliveries. The products are sent along with trusted individuals who are commuting toward the direction of the customer. Unlike in the case of business-to-consumer (B2C) shipments, the products do not have individual straight travel paths, instead they undergo a series of piecewise commutes before the destination. As there is no need for dedicated delivery trips, it significantly reduces the carbon cost of deliveries. In the Philippines, senders can avail of crowd-sourced delivery services by transporters. The crowd-shipping platform called Jojo was made available across three cities and 19 municipalities of Pampanga province in Manila (Yatco 2019). This project has resulted in the monetizing of Metro Manila traffic, all the while reducing a huge share of delivery vehicles on the road (https://www.ungeek.ph/2019/03/crowds hipping-is-now-a-thing-in-the-ph-with-jojodelivery-app/). In the time of COVID-19, using already available commuters on the road to effectively deliver parcels on their way is an ingenious idea both from the perspective of enhanced delivery times and carbon emission reduction. Implemented in Manila, this is said to have created a win–win situation for both the sender and transporter (Ramos 2019). However, right now, the company has limited the transporters commuting on a motorcycle even though the plans for other modes of transporting freight parcels are underway.

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

275

Another instance of successful crowd-sourced delivery is the Dabbawala (lunchbox carrier) system in Mumbai, India. The Dabbawala system is as-yet non-digitized and collects food parcels from individual households in Mumbai and delivers them to workplaces in time for lunch. Each ‘Dabba’ eliminates a trip by the office-going person to their individual houses or to the nearest restaurant for lunch, saving both time and the environment. The Dabbawalas rely on cycling, walking, and the Mumbai sub-urban railways for their commute from households to workplaces, none of which contributes to added GHE. This business model can, therefore, be expanded to other products with perpetual demand and homogeneity (Baindur and Macário 2013). While the Dabbawala system was limited to lunch boxes for the most part, in 2015, the Indian online-delivery giant Flipkart teamed up with the venture to improve their last-mile delivery capabilities in Mumbai using the lunch carriers to deliver the online orders to the designated places along with the food packages (Chengappa 2018). In the first phase of the system, the ‘Dabbawalas’ will collect the packages from Flipkart delivery hubs in Mumbai along with the Dabbas. As of now, this mode of delivery is available only for prepaid orders, even though further expansion to cash-on-delivery is also expected soon. In the case of nearby sellers, the Dabbawalas can even assist in bypassing the seller to delivery hub trips by delivering the product directly to the customer. This venture is set to have huge impacts in bringing down the carbon footprint associated with the last-mile delivery of Flipkart products, all the while improving the speed of delivery.

15.4.4 Delivery Automation and Last-Mile Planning Autonomous delivery is one of the emerging technologies in retail and a crucial part of the supply chain ecosystem (Kapser and Abdelrahman 2020). ADRs are defined as self-driving ground vehicles, which can deliver parcels or other goods like groceries and prepared meals to the doorstep. ADRs are designed like little moving robots (Fig. 15.4a) or like a mobile parcel locker (Fig. 15.4b) and they drive at a speed of approximately 5–10 km/h on the sidewalks. The short-range local deliveries by ADRs (up to 4 miles) typically handle packages weighing up to 20 kg. With various tests underway, researchers believe that automation could revolutionize the system and reduce delivery costs by 80–90%. In light of rising online purchases (16% in the USA), Jennings and Figliozzi (2020) have shown that a sustainable solution for counteracting the increased carbon emissions from these last-mile deliveries and pickup is the deployment of ADRs. The COVID-19 pandemic has amplified the demand for contactless deliveries; the essential workers (Fig. 15.4c) and pharmacies (Fig. 15.4d) have relied upon ADRs for handling package deliveries. Although the current state of autonomous delivery has many vulnerabilities, the capabilities of the technology to revolutionize freight deliveries are considerable (Pani et al. 2020b). In the near future, multiple ADRs carried by large autonomous vans are expected to carry even heavier loads covering longer distances seamlessly (Jennings and Figliozzi 2020).

276

A. Pani et al.

Fig. 15.4 Case studies of autonomous delivery robot deployment in the USA

Starship technologies is one of the market leaders working on automated delivery systems (VentureBeat 2018). In August 2019, Starship announced its plan to deploy thousands of ADRs across university campuses in the United States. It also has a remarkable record when it comes to delivering parcels autonomously: the ADR services from Starship have been tested in about 100 cities in 20 countries. They have traveled about 563,270 kms and completed their 100,000th delivery (Hawkins 2019). This is a remarkable feat for a start-up company that came into existence only a few years back with a technology that was not tested for performance till then. It is also proof that the technology can be scaled up to neighborhoods and to cities effectively and contribute to a significant modal shift in the freight transport sector, accompanied by greater energy savings and less carbon emissions. Nuro.ai is another start-up ADR company that has partnered with Kroger, the largest supermarket chain in the US. Nuro has developed autonomous delivery vehicles capable of carrying the parcel deliveries to the end-consumer without human interaction and has started deployments in Scottsdale, Arizona, and Houston, Texas (A More Prosperous, Safe, and Sustainable Society|by Nuro Team|Nuro|Medium 2021). Nuro deployed the first crewless delivery vehicle in Scottsdale, Arizona in December 2018 (https://medium.com/nuro/launching-the-first-unmanned-delivery-service-3fa 574ca6e25). Refraction AI is another ADR developer in the USA that focuses on delivering food and grocery items to those who opt for contactless delivery from service outlets partnered with the company. The cost of the delivery is about half that of the conventional truck delivery systems and can also result in significant fuel savings (Delivery robots help Ann Arbor restaurants weather COVID|Mechanical Engineering 2020). According to Refraction AI, the pandemic has completely changed consumer behavior and the demand for ADRs delivering groceries and food has increased significantly, forcing the start-up to pace up its production of delivery

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

277

robots (Sarah Parlette 2020). While the pandemic has catalyzed the acceptance of ADRs, it is likely that the trend will continue well after the pandemic, given the technological developments in the fields of connected vehicles and 5G communication systems. The existing policy framework in India, however, does not allow testing of autonomous technology, and significant research attention is required to assess the implementation challenges in enabling ADRs in India.

15.4.5 Integrated Land Use and Logistics Planning Land-use planning needs to develop designated locations for intermodal facilities such as inland container terminals, which can reduce urban congestion and foster a shift toward smaller commercial vehicles (Chandra et al. 2020; Pani et al. 2019b). For this purpose, premium city space may need to be made available for logistical development near significant freight generating areas (Sahu and Pani 2020). Another important aspect of land-use planning is to encourage spatial clustering of manufacturing firms that can achieve economies of density (Chandra et al. 2021b); this will lower down the transport costs and improve the delivery efficiency. Due to increasing land values in city cores, logistic land uses tend to locate farther from the city center (Aljohani and Thompson 2016). The increasing logistics sprawl in Indian cities (Gupta 2017; Mohapatra et al. 2021) underlines the need for integrating land use and logistics planning. This sprawl of logistic facilities increases daily truck kilometers traveled, as well as congestion on urban arterials. By developing an efficient zoning policy for reserving suitable land uses in city centers (e.g., creation of urban logistics spaces near major retailing chains), logistics sprawl can be reversed. By bringing urban logistics spaces to city centers, urban residents can also benefit in terms of superior access to goods and services.

15.5 Recommendations for Freight Policies in India Freight transport policies have been receiving increasing attention from both decision-makers and practitioners following the growing recognition that commercial vehicles (i.e., heavy-duty trucks and tractor-trailers) are one of the largest emitters in the transport sector. This new interest has been accompanied by growing recognition of the value of urban freight transport systems for cities’ life both in economic and social terms. The freight transport policies have, thus, critical importance since the idea is to mitigate the environmentally negative externalities without hampering the role of goods movement toward the economy and fulfilling urban residents’ consumer needs. Overall, freight transport policies around the world can be generally organized into three broad categories: (i) Policies for last-mile delivery efficiency; (ii) Policies for emission mitigation, and (iii) Policies for Trade node management. In the first category, the last-mile policies aim to address the inefficiencies associated with local

278

A. Pani et al.

Fig. 15.5 Overview of sustainable freight transport policies around the world

pickups and deliveries to or from retailers or households (e.g., Business-to-consumer E-commerce deliveries). The emission mitigation policies in the second category seek to reduce truck emissions and energy consumptions by improving the fuel efficiency of commercial vehicles, shifting to low-carbon freight vehicles (e.g., autonomous delivery robots), and curbing the use of old diesel trucks from logistic operations. The last category of sustainable freight policies focuses on the freight flows associated with trade hubs and gateways, such as ports, airports, intermodal transfer points, and border crossings. An overview of specific policies implemented in three of these categories around the world is presented in Fig. 15.5. The freight policies aimed at improving the last-mile delivery efficiency have attained significant importance in the recent past since one-third of urban truck traffic is engaged in goods pickups. Despite being the shortest-link in urban freight tours, the last mile corresponds to the largest unit share of emissions since multiple deliveries are often required to satisfy the consumer needs. The focus of shipment consolidation programs, for example, is to consolidate last-mile deliveries between multiple retailers, logistic providers, and warehouses and, in turn, enable efficient routing to the end-consumer and an overall reduction in delivery tours. The policies aimed to facilitate off-hour deliveries attempt to restrict deliveries in certain routes or time periods (e.g., peak-hour traffic period) and, in turn, improve the urban quality of life and mitigate the unintended freight delivery consequences of worsening congestion. The local planning policies related to curb space and loading facilities are also critically important for enabling seamless freight traffic in the last mile. With increased land value in city cores and more frequent deliveries for each business, local planning policies are crucial for regulating the already scarce curb space and providing standards for new developments in the central business district. Finally, policymakers are increasingly deploying intelligent transportation systems

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

279

(ITS) to manage urban freight movement by providing real-time traffic and parking information, toll collection, and automated access control. The policies aimed at emission mitigation primarily focus on putting a price on the environmental burden generated by private industry and imposing substantial costs (e.g., carbon tax) that enable a sustainable shift in freight travel pattern. Truck fuel efficiency and emission standards have been one of the effective policy interventions in this category. For instance, the implementation of Bharat Stage-VI (BSVI) emission standards in 2020 is planned to restrict new registrations of large trucks after 2021 and is expected to have significant impacts on reducing the negative externalities associated with freight trips in Indian cities. The vision of low-emission zones (LEZs) in this policy category is to restrict the types of vehicles that may enter a given part of the city and has been widely implemented in European countries. Despite the effectiveness of LEZs to reduce freight emissions, they have not been implemented much in North America or Asia. The most popular policy in North America and the Asian countries has been toward promoting the usage of alternative fuel vehicles (AFV), such as electric trucks, autonomous and connected trucks, and autonomous delivery robots. The market penetration of electric trucks into the global medium and heavy-duty market is projected to be 9.4% by 2030. The third category of sustainable freight policies focuses on trade hubs, because of the intense scale and scope of freight operations taking place in these facilities. The rising land costs in most of the cities around the world have pushed freight distribution facilities toward the city periphery, and in turn, have been forcing freight deliveries from trade hubs to pass through the city core. Land-use policies around trade hubs are, therefore, imperative to reduce the environmental impacts of trade hubs. Implementing road pricing strategies (e.g., weight-distance fees) has been shown to result in significant benefits in Europe and Asia in the recent years. Accelerated truck emission reduction programs are an alternative policy for accelerating the adoption and usage of low-carbon trucks at trade nodes. In contrast with the progress made around the world in implementing these policies, freight transport policy in India is still in the nascent stage (RMI 2019). Save for the policy initiatives aimed at increasing capacity and building facilities (truck terminals, consolidation centers), freight policies in India have largely been restrictive in nature (Malik et al. 2019) and require substantial advancements using data-driven decision-making tools.

15.6 Conclusion It is quite evident that consumer behavior has shifted tremendously during the COVID-19 lockdown period. People are hesitant to go out of their homes to purchase goods in person and increasingly prefer online deliveries. This reliance on e-commerce has resulted in increased delivery trips to households and has pushed online delivery platforms in India to continue to expand their delivery network coverage. Even though the extent of the permanence of this behavioral shift cannot be determined before the lockdown restrictions are lifted, and COVID-19 induced fear

280

A. Pani et al.

has subsided, it can be assumed that a large portion of the population has warmed up to the convenience of having things delivered to their doorsteps without having to spend exorbitant amounts of time in crowded shopping malls and grocery stores. Therefore, the post-COVID world is going to see a tremendous increase in last-mile delivery trips, which can significantly contribute to carbon emissions. Therefore, a gradual shift to low-carbon freight delivery is the need of the hour, as it can reduce the carbon footprint of freight delivery services without compromising delivery efficiency. The deployment of crowd-shipping services, electric and autonomous delivery vehicles, and the adoption of cargo bikes piloted in different countries around the world are indeed the essential first step toward the low-carbon future for freight delivery services in the last mile to consumers. This book chapter outlines the progress made on these fronts by elaborating on freight traffic impacts, emerging research themes and planning interventions, and policy measures in India. The planning interventions are summarized into five categories, covering freight traffic regulations, truck electrification, freight modal shift to bike-based alternatives, crowd-shipping, delivery automation, and integrated land use and logistics planning. The policy interventions are classified into three: (i) policies for last-mile delivery efficiency; (ii) policies for emission mitigation, and (iii) policies for Trade node management. The coordination and implementation of these planning and policy actions are likely to require financial and time resources and to encounter some degree of stakeholder backlash. Nonetheless, India is in a position to leapfrog and make important advances in policy implementation, and doing so will increase the efficiency of freight transport, with consequent positive impacts on the economy.

References A More Prosperous, Safe, and Sustainable Society|by Nuro Team|Nuro|Medium (2021). https:// medium.com/nuro/a-more-prosperous-safe-and-sustainable-society-1e8741324452 Abu-Rayash A, Dincer I (2020) Analysis of mobility trends during the COVID-19 coronavirus pandemic: exploring the impacts on global aviation and travel in selected cities. Energy Res Soc Sci 68:101693. https://doi.org/10.1016/j.erss.2020.101693 Alho AR, de Abreu e Silva J (2014) The development and application of an establishment-based Freight Survey: revealing retail establishments’ characteristics, goods ordering and delivery processes for the city of Lisbon. In: Transportation Research Board (TRB) 93rd annual meeting, Washington, DC, USA. https://doi.org/10.1007/s12544-015-0163-7 Alho AR, de Abreu e Silva J, de Sousa JP, Blanco E (2018) Improving mobility by optimizing the number, location and usage of loading/unloading bays for urban freight vehicles. Transp Res Part D Transp Environ 61:3–18. https://doi.org/10.1016/j.trd.2017.05.014 Aljohani K, Thompson RG (2016) Impacts of logistics sprawl on the urban environment and logistics: taxonomy and review of literature. J Transp Geogr 57:255–263. https://doi.org/10.1016/j.jtr angeo.2016.08.009 Allen J, Browne M, Cherrett T (2012) Survey techniques in urban freight transport studies. Transp Rev 32:287–311. https://doi.org/10.1080/01441647.2012.665949 Allen J, Browne M (2008) Review of Survey Techniques used in Urban Freight Studies, London Allen J, Thorne G, Browne M (2007) Good practice guide on urban freight transport. Bestufs Adm Cent 84

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

281

Anderhofstadt B, Spinler S (2019) Factors affecting the purchasing decision and operation of alternative fuel-powered heavy-duty trucks in Germany—a Delphi study. Transp Res Part D Transp Environ 73:87–107. https://doi.org/10.1016/j.trd.2019.06.003 Baindur D, Macário RM (2013) Mumbai lunch box delivery system: a transferable benchmark in urban logistics? Res Transp Econ 38:110–121. https://doi.org/10.1016/j.retrec.2012.05.002 Bharadwaj N, Mathew S, Pani A, Arkatkar S, Joshi G, Ravinder K (2018) Effect of traffic composition and emergency lane on capacity: a case study of intercity expressway in India. Transp Lett 10:316–332. https://doi.org/10.1080/19427867.2016.1265237 Chandra A, Pani A, Sahu PK (2020) Designing zoning systems for freight transportation planning: a gis-based approach for automated zone design using public data sources. Transp Res Procedia 48:605–619. https://doi.org/10.1016/j.trpro.2020.08.063 Chandra A, Sharath MN, Pani A, Sahu PK (2021a) A multi-objective genetic algorithm approach to design optimal zoning systems for freight transportation planning. J Transp Geogr 92:103037. https://doi.org/10.1016/j.jtrangeo.2021.103037 Chandra A, Pani A, Sahu PK, Majumdar BB, Sharma S (2021) Identifying large freight traffic generators and investigating the impacts on travel pattern: a decision tree approach for lastmile delivery management. Res Transp Bus Manag 100695. https://doi.org/10.1016/j.rtbm.2021. 100695 Chengappa S. Flipkart’s Dabbawala tie-up for last-mile delivery. https://www.thehindubusinessline. com/info-tech/flipkarts-dabbawala-tieup-forlastmile-delivery/article7085866.ece Combes F, Leurent F (2013) Improving road-side surveys for a better knowledge of road freight transport. Eur Transp Res Rev 5:41–51. https://doi.org/10.1007/s12544-012-0083-8 Coulombel N, Dablanc L, Gardrat M, Koning M (2018) The environmental social cost of urban road freight: evidence from the Paris region. Transp Res Part D Transp Environ 63:514–532. https:// doi.org/10.1016/j.trd.2018.06.002 Crowdshipping is now a thing in the PH with Jojo Delivery App. https://www.ungeek.ph/2019/03/ crowdshipping-is-now-a-thing-in-the-ph-with-jojo-delivery-app/ Dablanc L (2007) Goods transport in large European cities: difficult to organize, difficult to modernize. Transp Res Part A Policy Pract 41:280–285. https://doi.org/10.1016/j.tra.2006.05.005 Delivery robots help Ann Arbor restaurants weather COVID|Mechanical Engineering (2020) Demir E, Huang Y, Scholts S, Van Woensel T (2015) A selected review on the negative externalities of the freight transportation: modeling and pricing. Transp. Res. Part E Logist. Transp. Rev 77:95–114 (2015). https://doi.org/10.1016/j.tre.2015.02.020 Dhonde BN, Patel CR (2021) Estimating urban freight trips using light commercial vehicles in the Indian textile industry. Transp Res Interdiscip Perspect 11:100411. https://doi.org/10.1016/j.trip. 2021.100411 EcoCart: EcoCart: Carbon Neutral Orders. https://apps.shopify.com/ecocart Feng W, Figliozzi MA (2012) Conventional vs electric commercial vehicle fleets: a case study of economic and technological factors affecting the competitiveness of electric commercial vehicles in the USA. Procedia—Soc Behav Sci 39:702–711. https://doi.org/10.1016/j.sbspro.2012.03.141 Fernandes P, Bandeira JM, Fontes T, Pereira SR, Schroeder BJ, Rouphail NM, Coelho MC (2016) Traffic restriction policies in an urban avenue: a methodological overview for a trade-off analysis of traffic and emission impacts using microsimulation. Int J Sustain Transp 10:201–215. https:// doi.org/10.1080/15568318.2014.885622 Gatta V, Marcucci E, Nigro M, Serafini S. (2019) Sustainable urban freight transport adopting public transport-based crowdshipping for B2C deliveries. Eur Transp Res Rev 11. https://doi.org/ 10.1186/s12544-019-0352-x Gupta S (2017) Garima: logistics sprawl in timber markets and its impact on freight distribution patterns in Metropolitan City of Delhi, India. Transp Res Procedia 25:965–977. https://doi.org/ 10.1016/j.trpro.2017.05.471 Hagen J, Lobo Z, Mendonça C (2013) The benefits of cargo bikes in Rio De Janeiro: a case study. In: 13th WCTR, pp 1–15

282

A. Pani et al.

Harikumar A (2020) Effects of COVID-19 on transportation demand. https://www.teriin.org/art icle/effects-covid-19-transportation-demand Harris-Burland H (2020) Saving the environment one robot delivery at a time. https://www.busine sschief.eu/leadership/saving-environment-one-robot-delivery-time Hawkins AJ (2019) Thousands of autonomous delivery robots are about to descend on US college campuses. https://www.theverge.com/2019/8/20/20812184/starship-delivery-robot-exp ansion-college-campus Hillyer M (2020) Urban deliveries expected to add 11 minutes to daily commute and increase carbon emissions by 30% until 2030 without effective intervention. https://www.weforum.org/ press/2020/01/urban-deliveries-expected-to-add-11-minutes-to-daily-commute-and-increasecarbon-emissions-by-30-until-2030-without-effective-intervention-e3141b32fa/ Holguín-Veras J, Jaller M, Sanchez-Diaz I, Wojtowicz J, Campbell S, Levinson H, Lawson C, Powers EL, Tavasszy L (2012) NCFRP 19: freight trip generation and land use: final report, Washington, DC, USA Holguín-Veras J, Lawson C, Wang C, Jaller M, González-Calderón C, Campbell S, Kalahashti L, Wojtowicz J, Ramirez D (2016) NCFRP 37: using commodity flow survey microdata and other establishment data to estimate the generation of freight, freight trips, and service trips. https:// doi.org/10.17226/24602 International Institute for Sustainable Development: Green Recovery: ADB outlines pathway to a low-carbon and resilient future. https://www.iisd.org/sustainable-recovery/news/green-recoveryadb-outlines-pathway-to-low-carbon-and-resilient-future/ Jaller M, Holguín-Veras J, Hodge S (2013) Parking in the city. Transp Res Rec J Transp Res Board 2379:46–56. https://doi.org/10.3141/2379-06 Jennings D, Figliozzi M (2020) Study of road autonomous delivery robots and their potential impacts on freight efficiency and travel. Transp Res Rec OnlineFirs 317–326. https://doi.org/10.1177/036 1198119849398 Kapser S, Abdelrahman M (2020) Acceptance of autonomous delivery vehicles for last-mile delivery in Germany—extending UTAUT2 with risk perceptions. Transp Res Part C Emerg Technol 111:210–225. https://doi.org/10.1016/j.trc.2019.12.016 Kriger D, McCumber M, Clavelle A, Gan B, Chow T (2011) Freight Transportation Surveys, NCHRP Synthesis 410. Transportation Research Board, Washington, DC Kulpa T (2014) Freight truck trip generation modelling at regional level. Procedia—Soc Behav Sci 111:197–202. https://doi.org/10.1016/j.sbspro.2014.01.052 Launching the first unmanned delivery service|by Dave Ferguson|Nuro|Medium (2018). https://med ium.com/nuro/launching-the-first-unmanned-delivery-service-3fa574ca6e25 Loske D (2020) The impact of COVID-19 on transport volume and freight capacity dynamics: an empirical analysis in German food retail logistics. Transp Res Interdiscip Perspect 6:100165. https://doi.org/10.1016/j.trip.2020.100165 Malik L, Sánchez-Díaz I, Tiwari G, Woxenius J (2017) Urban freight-parking practices: The cases of Gothenburg (Sweden) and Delhi (India). Res Transp Bus Manag 24:37–48. https://doi.org/10. 1016/j.rtbm.2017.05.002 Malik L, Tiwari G, Thakur S, Kumar A (2019) Assessment of freight vehicle characteristics and impact of future policy interventions on their emissions in Delhi. Transp Res Part D Transp Environ 67:610–627. https://doi.org/10.1016/j.trd.2019.01.007 Marcucci E, Gatta V, Scaccia L (2015) Urban freight, parking and pricing policies: an evaluation from a transport providers’ perspective. Transp Res Part A Policy Pract 74:239–249. https://doi. org/10.1016/j.tra.2015.02.011 Middela MS, Ramadurai G (2020) Incorporating spatial interactions in zero-inflated negative binomial models for freight trip generation. Transportation (AMST). https://doi.org/10.1007/s11116020-10132-w Mohan D, Tiwari G, Bhalla K (2015) Road Safety in India, Transportation Research and Injury Prevention Programme, Indian Institute of Technology Delhi. https://www.roadsafetynetwork.in/ wp-content/uploads/2019/01/tripp-road-safety-in-india-status-report-2015.pdf

15 Sustainable Freight Transportation Planning and Policies for a Logistics …

283

Mohapatra SS, Pani A, Sahu PK (2021) Examining the impacts of logistics sprawl on freight transportation in indian cities: implications for planning and sustainable development. J Urban Plan Dev 147:4021050. https://doi.org/10.1061/(asce)up.1943-5444.0000745 Muñuzuri J, Cortés P, Onieva L, Guadix J (2009) Modeling freight delivery flows: missing link of urban transport analysis. J Urban Plan Dev 135:91–99. https://doi.org/10.1061/(ASCE)UP.19435444.0000011 Pani A, Sahu PK (2019a) Planning, designing and conducting establishment-based freight surveys: A synthesis of the literature, case-study examples and recommendations for best practices in future surveys. Transp Policy 78:58–75. https://doi.org/10.1016/j.tranpol.2019.04.006 Pani A, Sahu PK (2019b) Modelling non-response in establishment-based freight surveys: a sampling. Transp Policy. https://doi.org/10.1016/j.tranpol.2019.10.011 Pani A, Sahu PK (2019c) Comparative assessment of industrial classification systems for modeling freight production and freight trip production. Transp Res Rec 2673:210–224. https://doi.org/10. 1177/0361198119834300 Pani A, Sahu PK, Patil GR, Sarkar AK (2018) Modelling urban freight generation: a case study of seven cities in Kerala, India. Transp Policy 69:49–64. https://doi.org/10.1016/j.tranpol.2018. 05.013 Pani A, Sahu PK, Majumdar BB (2019a) Expenditure-based segmentation of freight travel markets: identifying the determinants of freight transport expenditure for developing marketing strategies. Res Transp Bus Manag 33:100437. https://doi.org/10.1016/j.rtbm.2020.100437 Pani A, Sahu PK, Chandra A, Sarkar AK (2019b) Assessing the extent of modifiable areal unit problem in modelling freight (trip) generation: Relationship between zone design and model estimation results. J Transp Geogr 80:102524. https://doi.org/10.1016/j.jtrangeo.2019.102524 Pani A, Bhat FA, Sahu PK (2020a) Effects of business age and size on freight demand: decomposition analysis of Indian establishments. Transp Res Rec 2674:112–126. https://doi.org/10.1177/036119 8120902432 Pani A, Mishra S, Golias M, Figliozzi M (2020b) Evaluating public acceptance of autonomous delivery robots during COVID-19 pandemic. Transp Res Part D Transp Environ 89:102600. https://doi.org/10.1016/j.trd.2020.102600 Pani A, Sahu PK, Holguín-Veras J (2021) Examining the determinants of freight transport emissions using a fleet segmentation approach. Transp Res Part D Transp Environ 92. https://doi.org/10. 1016/j.trd.2021.102726 Ramos CM (2019) PH startup firm unveils “Jojo” to turn time wasted in traffic into cash. https://technology.inquirer.net/83951/ph-startup-firm-unveils-jojo-to-turn-time-wasted-intraffic-into-cash RMI (2019) Efficient urban freight policy framework. Ministry of Housing and Urban Affairs, Government of India. https://rmi.org/wp-content/uploads/2019/07/rmi-efficient-urban-freight. pdf Rowell M, Gagliano A, Goodchild A (2014) Examining carrier categorization in freight models. Res Transp Bus Manag 11:116–122. https://doi.org/10.1016/j.rtbm.2014.06.006 Sahu PK, Pani A (2020) Freight generation and geographical effects: modelling freight needs of establishments in developing economies and analyzing their geographical disparities. Transportation (AMST) 47:2873–2902. https://doi.org/10.1007/s11116-019-09995-5 Santos G (2005) Urban congestion charging: a comparison between London and Singapore. Transp Rev 25:511–534. https://doi.org/10.1080/01441640500064439 Santos G (2017) Road fuel taxes in Europe: do they internalize road transport externalities? Transp Policy 53:120–134. https://doi.org/10.1016/j.tranpol.2016.09.009 Sarah Parlette: Refraction AI launches contactless robot grocery delivery in Ann Arbor (2020). https://www.clickondetroit.com/all-about-ann-arbor/2020/06/30/refraction-ai-launchescontactless-robot-grocery-delivery-in-ann-arbor Schliwa G, Armitage R, Aziz S, Evans J, Rhoades J (2015) Sustainable city logistics—making cargo cycles viable for urban freight transport. Res Transp Bus Manag 15:50–57. https://doi.org/ 10.1016/j.rtbm.2015.02.001

284

A. Pani et al.

Shabani K, Figliozzi M (2012) A statistical study of commodity freight value/tonnage trends in the United States. In: Transportation Research Board 91st annual meeting compendium of papers, Washington, DC, USA VentureBeat (2018) Starship Technologies launches commercial package delivery service using autonomous robots Tavasszy LA (2020) Predicting the effects of logistics innovations on freight systems: directions for research. Transp Policy 86:A1–A6. https://doi.org/10.1016/j.tranpol.2019.11.004 Venkadavarahan M, Raj CT, Marisamynathan S (2020) Development of freight travel demand model with characteristics of vehicle tour activities. Transp Res Interdiscip Perspect 8:100241. https:// doi.org/10.1016/j.trip.2020.100241 Vidal Vieira JG, Fransoo JC (2015) How logistics performance of freight operators is affected by urban freight distribution issues. Transp Policy 44:37–47. https://doi.org/10.1016/j.tranpol.2015. 06.007 World Bank Blogs (2020) An old solution to new challenges: the rebirth of the cargo bike. https:// blogs.worldbank.org/transport/old-solution-new-challenges-rebirth-cargo-bike World Health Organization: Noise (2021). https://www.euro.who.int/en/health-topics/environmentand-health/noise Yanocha D, Allan M (2019) The electric assist: leveraging e-bikes and e-scooters for more livable cities Yatco R (2019) Jojo Crowdshipping App rolls out services in Pampanga starting June 10, Punto! Central Luzon

Chapter 16

Automated Sensors for Indian Traffic: Challenges and Solutions Lelitha Devi Vanajakshi and Shriniwas S. Arkatkar

16.1 Introduction Robust traffic data is a principal need for building realistic transport models, particularly toward achieving sustainable urban development in Indian cities. Growing demand for transport infrastructure also warrants reliable, accurate, yet cost-effective methods of traffic data collection. In addition, rapid, exhaustive, and accurate data acquisition is critical for real-time monitoring and strategic planning of traffic facilities. Automated traffic sensors can be classified as vehicle detectors/identifiers and vehicle tracking devices. The vehicle detectors/identifiers are mostly location-based and collect data from the entire vehicle population that crosses the location. They cannot collect spatial parameters such as travel time or density. Vehicle tracking devices, on the other hand, are usually fixed inside vehicles and can collect spatial parameters such as travel time from that individual vehicle. However, data can be obtained only from vehicles that voluntarily participate by housing the tracking device. This limits the sample size. Commonly used location-based sensors for automated traffic data collection include magnetic, radar, infrared, laser, inductive, and video sensors. Although these are proven data collection devices for traffic conditions elsewhere, they may not work for Indian traffic conditions, with its heterogeneity and lack of lane discipline. Most of the above data collection technologies are limited by the need for lane-based L. D. Vanajakshi (B) Department of Civil Engineering, Indian Institute of Technology Madras, Chennai 600036, India e-mail: [email protected] S. S. Arkatkar Department of Civil Engineering, Sardar Vallabhbhai National Institute of Technology, Surat 395007, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_16

285

286

L. Vanajakshi and S. Arkatkar

traffic and cannot classify various types of vehicles. Until now, there are no proven technologies suited specifically to Indian traffic conditions. Spatial or tracking devices commonly adopted include automatic vehicle identifiers (AVI) such as RFID tags and GPS-based Automatic Vehicle Location (AVL). The advantage of these devices is that they are mostly independent of the traffic and environmental conditions and hence can be used for varying traffic conditions. However, the sample size constraint and concerns of all vehicle classes not being represented uniformly are more under the heterogeneous traffic conditions. Presently, in India, traffic counts are obtained manually, which has a range of problems related to data accuracy, reliability, and timely output, due to the involvement of huge manpower. While this approach is immensely helpful in getting accurate data for a short period, it can be expensive to implement and not scalable. Hence, there is a need to develop automated sensors that can work under Indian traffic conditions. This chapter will discuss some of the automated sensors that are used under Indian traffic conditions. This includes a novel Inductive loop detector system, GPS, Bluetooth/Wi-Fi sensors, and RFID tags, in addition to the use of advanced methods based on crowdsourcing through social media. These sensors are discussed in the following sections.

16.2 Inductive Loop Detectors Inductive loop detectors (ILD) are one of the most extensively used sensors due to their high sensitivity and cost-effectiveness. However, the traditional loops are made for the homogeneous and lane-disciplined traffic. They cannot function well under the lane-less movement of multiple types of vehicles, as on the Indian roads. Parallel movement of different types of vehicles, within the same lane (same loop area), is the biggest challenge. In addition, those traditional loops cannot differentiate large vehicles (e.g., bus) from small vehicles like a bicycle. Hence, a new and simple inductive loop sensor structure that senses and differentiates both large and smallsized vehicles and works even for the parallel movement condition is developed (Ali et al. 2012). The sensor provides a unique output signature for each type of vehicle. Details of the new inductive loop structure are briefly discussed below. Figure 16.1 shows the shape of the new loop, with an outer loop and an inner loop formed using a continuous conductor. When current flows through this coil, a magnetic flux is generated in the outer loop aiding the flux produced by the inner loop. There will be a change in inductance when a vehicle crosses this loop, which will be more in the outer loop if it is a large vehicle and more in the smaller loop if the vehicle is small. Thus, it helps in identifying both small and large vehicles. The loop can be placed just below the road surface and the overall connections can be as shown in Fig. 16.1. To address the lane indiscipline, the placement of the loops was done as shown in Fig. 16.2. The proposed loops with the new shape are made in sizes that can fit only one two wheeler on top of it. Several of such small size loops in the proposed

16 Automated Sensors for Indian Traffic: Challenges and Solutions

287

Fig. 16.1 The newly developed inductive loop detector system

Fig. 16.2 Multiple inductive loop system placement

shapes are placed across the road width, covering the whole width of the road (Ali et al. 2013). A Data Acquisition System (DAS) is used for collecting and sending the collected data to a central processing system. This central system with suitable algorithms installed can do quality control and processing of the data to generate the number and type of vehicles, their speeds, occupancy, etc. (Ali et al. 2014; Yogesh and Vanajakshi 2018). The main objective of this process is to extract the vehicle signatures from the raw signal, in real time. A wide range of features can be obtained from the vehicle signature. The signal was first de-noised with a discrete wavelet transform (DWT) analysis to remove the high-frequency white noises followed by segmenting the vehicle signatures using moving average and moving standard deviation techniques. The approximation coefficients obtained from DWT performed on each signature are used as the features in the present study. For the classification of vehicle signatures from the derived features, supervised machine learning algorithms like random forest and multi-class SVM were used. All these methods were implemented and tested using the data collected, as detailed below. A prototype system has been built and tested for vehicle classification performance for a toll plaza application under heterogeneous Indian traffic near Perungudi toll

288

L. Vanajakshi and S. Arkatkar

plaza in Chennai, India. A total of 407 vehicle signatures were collected. A sample signature of a car and the corresponding video, generated by the experimental setup, is shown in Fig. 16.3. The obtained raw data is analyzed in three sub-phases as de-noising, signal segmentation, and vehicle classification. A Daubechies wavelet of second order (db2) with two vanishing points was chosen for de-noising. Figure 16.4 shows a sample of the de-noised signal after applying this method. Moving standard deviation-based segmentation was implemented next. From the identified signatures, features are extracted for model training. The approximation coefficients obtained from DWT-based feature extraction were used as the input for classification using the random forest method. Figure 16.5 shows the classification

Fig. 16.3 A sample view from field testing

Fig. 16.4 De-noised sample raw data

16 Automated Sensors for Indian Traffic: Challenges and Solutions

289

Fig. 16.5 Accuracy for vehicle classification

accuracy using the random forest technique which gave an accuracy of 95.77%. Linear SVM was applied next for classification and was built on the dataset with a similar 75 and 25% train and test strategy. The overall test accuracy of this model was found to be 97.03% showing this as a promising method for automated vehicle classification.

16.3 Image Processing Extracting traffic information from video camera recordings possesses its own challenges such as change of illumination due to weather conditions and time of day and occlusion of vehicles, etc. For heterogeneous and lane-less traffic, the challenges like occlusion become more serious due to the smaller vehicles becoming invisible behind or at the side of bigger vehicles. One representative study on microscopic traffic data extraction and macroscopic traffic data extraction using image processing under heterogeneous and lane-less traffic is discussed in this section.

16.3.1 Extraction of Microscopic Variables Automated trajectory extraction using image processing is discussed first (Raveendran et al. 2019). In order to develop an automated trajectory data extractor, it is

290

L. Vanajakshi and S. Arkatkar

necessary to detect the vehicles in the video frame by frame automatically as well as track them in subsequent frames. For achieving this, an object detection model was trained which was capable of detecting the vehicles in a given frame. To achieve this, the Faster RCNN inception V2 (Halawa et al. 2019) model pre-trained on the Microsoft Coco (Lin et al. 2015) dataset is adopted. For vehicle classification, vehicles were labeled in the images extracted from the video files to generate the training data for each of the study sections. Later, those labeled images are loaded to the LABELING (Python Community) utility and converted to the XML files (Lalams). Further, data in the XML files were used for training the Faster RCNN inception V2 model to record the vehicle features. After recording the vehicles’ features, the video files were given as inputs to the trained models. Initially, the video was divided into image frames, and each frame is given as input to the model. Based on the trained Faster RCNN inception V2 model, the vehicular features were searched over the whole image frame. With the matching criteria, the objects were identified, and the vehicle category was assigned to each of the detected objects in a frame. Along with the vehicle category, other features such as time, centroid longitudinal, and lateral positions were recorded for the development of the trajectory data through Euclidean Object Tracking (Wang et al. 2005). The entire process is shown in Fig. 16.6. A sample trajectory dataset extracted using the developed tool is shown in Fig. 16.7. The trajectory data is developed for the Western Expressway section for a period of 15 min. To validate, the trajectory data extracted using the automated tool was compared with a semi-automated trajectory tool. In the semi-automated tool, the vehicle position was tracked using the pointer. In balancing the trajectory data accuracy with the manual efforts, the vehicles over the section are tracked with an update interval

Fig. 16.6 Automated trajectory data development tool

16 Automated Sensors for Indian Traffic: Challenges and Solutions

291

Fig. 16.7 Space–time plot developed using automated trajectory extractor

of 0.5 s. Longitudinal speed, longitudinal acceleration, lateral speed, and lateral acceleration were compared as shown in Fig. 16.8. From the figure, it can be noted that the microscopic parameters computed using trajectory data extracted using the semi-automated tool and automated tool are in

Fig. 16.8 Comparison of different microscopic parameters

292

L. Vanajakshi and S. Arkatkar

close approximation with each other, highlighting the efficiency of the developed automated trajectory data extractor.

16.3.2 Extraction of Macroscopic Variables Extraction of macroscopic traffic variables, such as average speed and occupied area, as a surrogate of density under heterogeneous traffic conditions, is discussed next (Reddy et al. 2019). The occupied area, defined as the percentage area of a road section occupied by vehicles, is estimated first. The main steps involved are (1) background subtraction, (2) shadow elimination, and (3) foreground extraction. Each of these steps and the corresponding algorithms implemented are discussed below. Background Subtraction (BGS) is performed to separate moving vehicles (the foreground of an image) from the other objects (the background) in a scene. The main challenges in this step are varying illumination, movements in the background (trees, water), and new objects being included in the background. BGS is usually implemented by building a model for the background and then detecting the foreground by taking the difference between the current frame and the background model built. Usually, to eliminate noise, a threshold is used on this difference to build a binary foreground mask. In this study, a recursive BGS algorithm called Gaussian Mixture Model with Adaptive Number of Gaussians (AGMM) proposed by Zivkovic (2004) has been utilized. In AGMM, each channel of each pixel in the image is modeled as a Gaussian mixture model and the algorithm automatically adapts the number of Gaussians required for each pixel. A sample frame with the region of interest marked is shown in Fig. 16.9a and the output of the background modeling step is shown in Fig. 16.9b. The foreground mask obtained after the previous step is a binary mask, with 1’s representing the foreground. Then using the Sobel detector (Peli and Malah 1982), all the edges corresponding to vehicles and shadows were obtained as shown in Fig. 16.9c. The next step was shadow elimination, as shadows of vehicles move with the foreground leading to errors by distorting the shapes of vehicles and increasing the number of occlusions. The edge intensity (gradient intensity) of shadows will be less compared to vehicle bodies, and this property is utilized for identifying and removing shadows (Xu et al. 2005; Sanin et al. 2012). In the next step, the outer edges are obtained as shown in Fig. 16.9d by using the binary mask from the foreground detection step. Now, to obtain the inner edges, the outer edges are subtracted from all the edges (Fig. 16.9e). In the next, the foreground corresponding to vehicles is reconstructed using the inner edges. By connecting the inner edge pixels in each contour of the foreground mask, horizontal and vertical candidate images are obtained. In the next step, an AND operation of the horizontal and vertical candidates is used to get the final output as shown in Fig. 16.9f. To remove any noise in the data, morphological operations are performed at the end (Suzuki and Abe 1985). A sample output frame after this step is shown in Fig. 16.9g.

16 Automated Sensors for Indian Traffic: Challenges and Solutions

(a) Sample Frame with Region of interest marked

(c) Images with all the edges

(e) Image with inner edges

(b) Output of Background Subtraction

(d) Outer edges from foreground mask

(f) Inner edges with AND operation on the horizontal & vertical candidates

Fig. 16.9 Sample outputs of various steps in the image processing for the area occupied

293

294

L. Vanajakshi and S. Arkatkar

The number of white pixels in a region of interest indicates the area occupied by vehicles. The ratio between the number of white pixels to the total number of pixels present in the region of interest gives an estimate of the percentage area occupied. For speed estimation, tracking of vehicles over a period of time is needed. Featurebased tracking was used in this study to estimate the average speed of the vehicles. Corner features were identified and tracked using optical flow techniques (Lucas and Kanade 1981). Tracking these points over a sequence of frames gives the average speed. The perspective transformation was performed to obtain real-world speed values. Figure 16.10 shows the corners (features) obtained by the above-described method. Figure 16.10a shows a sample frame picked from a video and Fig. 16.10(b) is the image with detected corner points. Figure 16.11 shows sample results of the occupied area obtained using the developed algorithm for randomly selected frames with the real occupancy values. It can be seen that the estimated values closely match the real ones. The errors were quantified using Mean Absolute Percentage Error (MAPE) and were found to be 9.6% in this case. Multiple video sets from this location were ran and the results obtained

(a) Sample Frame

Fig. 16.10 Detection of corner features

Fig. 16.11 Estimated and actual area occupied (%)

b) Frame with detected corner points

16 Automated Sensors for Indian Traffic: Challenges and Solutions

295

Fig. 16.12 Estimated and observed speed values

were in the range of 7.25–12.2% showing good performance. Similarly, Fig. 16.12 shows a graph showing the estimated speed values along with the actual observations. Error analysis gave a MAPE of 6%, showing very good performance in this case too.

16.4 Wi-Fi/Bluetooth Sensors and RFID Sensors The idea of deploying Bluetooth (BT) media scanners (BMS) or Wi-Fi media scanners (WMS) can be beneficial. A BMS/WMS scanner has a communication range (say having a radius of about 100 m) defined as a detection zone. The detection zone scans the Media Access Control addresses (MAC-ID) of the ascertainable BT/Wi-Fi devices detected within the zone. The MAC addresses are characteristically 48-bit electronic identifiers for each device in the form of ‘12:34:56:78:90:ab’. The first three octets ‘12:34:56’ represent the Organizationally Unique Identifier (OUI) which provides information on the manufacturer, and the last three ‘78:90:ab’ are assigned by the manufacturer as a unique address to the device. The communication range between two BT or Wi-Fi devices depends on the power of the BT/Wi-Fi radio transmitter, the receiver’s sensitivity, and the medium’s absorption rate. These devices are classified into three classes (Class-1, Class-2, and Class-3). Patra et al. (2019, 2021) reported the details of the development and evaluation of a Wi-Fi MAC Scanner and its application in predicting stream travel time. Murphy et al. (2002) experimented with the communication range of Class-1 and Class-2 devices and reported that both these classes outperformed the minimum range specification. Class-1 and Class2 were reported to be communicating at a maximum range of 250 m and 122 m, respectively. Given the communication range up to 250 m, which is advantageous for transportation-related applications, it provides the necessary time for BMS to discover the BT devices traveling within the range. With this, any device is detected multiple-time within the plausible range. From this, by applying a MAC-ID matching algorithm, one can obtain the segment’s travel time accurately. Aggregating travel

296

L. Vanajakshi and S. Arkatkar

Fig. 16.13 Arrangements of BT/Wi-Fi sensors for travel time computation

time from different logged devices provides the average travel time of the segment under consideration. A pictorial depiction of the BT/Wi-Fi sensors-based travel time computation is shown in Fig. 16.13. There could be some concern regarding the privacy of drivers, as the sensor tracks the pathway of the detected vehicle. However, privacy is safeguarded because the MAC address is not related to any other personal data; thus, the examined data cannot be associated with particular individuals. Incidentally, RFID (Radio Frequency Identification)-based detection concept is also similar to that of the Wi-Fi/BT sensors. It works on the inductive coupling principle, based on a radio frequency or radio waves. RFIDs warrant an electromagnetic field to recognize objects or track the objects automatically. In the case of an RFID sensor, the scanner will read the unique RFID tag fixed on a given vehicle and record the time stamp when the ID is detected. In India, RFID tags are made essential to be installed in different vehicles for implementing electronic toll collection (ETC) aiding automatic tollway operations. Contemplating this, there would be a certain addition in the number of vehicles with secured RFID tags, in the near future. This, in turn, is likely to enhance the penetration rate with a greater number of RFID-based vehicle detections. Moreover, presently, the cost of installation and maintenance of RFID sensors is lesser compared to that of Wi-Fi sensors. Thus, the RFID sensors shall have more noteworthy returns over Wi-Fi/BT sensors as a probable opportunity for traffic data collection under Indian conditions.

16 Automated Sensors for Indian Traffic: Challenges and Solutions

297

16.4.1 Performance Characteristics of BT/Wi-Fi Sensor Performance of the sensor is generally gauged using penetration rate or fraction read (Porter et al. 2013; Mathew et al. 2016; Gore et al. 2019a). The penetration rate is defined as the ratio of sensor-recorded volume to the actual volume. It indicates the proportion of traffic volume penetrated by sensor detections. A higher penetration rate indicates higher efficiency of the given sensor to capture traffic variations and viceversa. The performance characteristics of the Wi-Fi/BT sensor by varying antenna size, height, and lateral position of the sensor is studied extensively, suggesting that the antenna size, height, and lateral position of the sensor significantly influence sensor efficiency in terms of penetration rate (Brennan et al. 2010; Porter et al. 2013; Abedi et al. 2015). Brennan et al. (2010) suggested that the BT sensor should be placed along the median at the height of 8feets for effective data collection. On the other hand, Gore et al. (2019a) recommend installing sensors at the ground along the median to capture both temporal and spatial variation in traffic flow effectively. Bakula et al. (2011) and Gore et al. (2019a) reported that the probability of device detection increases with a decrease in vehicular speeds. Abedi et al. (2015) reported that a 2 dBi antenna was not suitable for collecting MAC address samples from cyclists. On the other hand, the 16 dBi antenna collected more anomalies, and it had a better performance in capturing cyclists’ and runners’ Wi-Fi and Bluetooth-enabled devices. The authors concluded that antenna gain significantly affected the dataset’s accuracy in terms of pedestrians’ and cyclists’ movement tracking. Porter et al. (2013) reported that vertically polarized antenna with antenna gain between 9 and 12 dBi as potential candidates for travel time data collection. It is expected that the performance of RFID sensors would also be influenced by the speed of the upcoming vehicle and the lateral and vertical position of the RFID sensors. However, the influence of different parameters on the performance of RFID sensors is not explored much. In India, the penetration rate of Wi-Fi/BT technology is reported between 8 and 52% (Gore et al. 2019a; Sharma et al. 2020). On the other hand, the penetration rate of RFID is approximately 15%. However, because RFID tags are made mandatory by the Government of India, RFID sensors could be a better prospect for traffic data collection under the Indian context.

16.4.2 Application of Wi-Fi/BT and RFID Sensor Data: A Few Case Studies in India With the number of vehicles equipped with BT/Wi-Fi technology increasing, BT/WiFi sensor is a potential future traffic data source. The concept of re-identification facilitates estimation of (i) traffic flow, (ii) direction of travel (origin–destination), (iii) route choice, and (iv) travel time and delay. Some of the other applications

298

L. Vanajakshi and S. Arkatkar

reported using these data sources include signal state prediction (Sharma et al. 2020), travel time prediction (Thakkar et al. 2021), vehicle occupancy (Gore et al. 2019b), and OD estimation (Advani et al. 2020a, b). The use of this data for travel time reliability (TTR) analysis is discussed below. Mathew et al. (2016), Singh et al. (2019), and Adavni et al. (2020) studied travel time reliability for urban corridors using Wi-Fi/BT sensor-derived travel time data. Mathew et al. (2016) and Singh et al. (2019) studied the reliability of urban arterials in Chennai, whereas Advani et al. (2020) analyzed the travel time reliability of urban arterial in Ahmedabad. Mathew et al. (2016) studied the reliability of two routes in Chennai using travel time data derived using BT sensors. Travel time data for the study corridors were collected for a period of one month. The reliability of the corridors was studied by developing a cumulative distribution plot of travel times as shown in Fig. 16.14. The study concluded that BT technology could provide fairly accurate travel time estimates for urban arterials. Singh et al. (2019) examined the travel time reliability of an urban arterial using travel time collected using Wi-Fi data. The study analyzed the variation in TTR measures against the shape factor of GEV distribution to derive an effective measure of TTR. BTI was concluded as an appropriate measure of TTR and λvar as a suitable travel time variability index. The authors developed reliability-based level of service (LOS) criteria for urban roads. The reliability-based LOS thresholds are summarized in Table 16.1. Sakhare et al. (2018, 2019) used Bluetooth data for analyzing corridor level travel time. Jayan and Anusha (2019) analyzed the travel time reliability of an urban arterial using travel time data collected using RFID sensors. The travel time reliability was studied using cumulative percentile plots and buffer time index as a suitable travel time reliability measure.

Fig. 16.14 Cumulative percentile plots of travel time (Mathew et al. 2016)

16 Automated Sensors for Indian Traffic: Challenges and Solutions Table 16.1 Reliability-based level of service thresholds (Singh et al. 2019)

Cluster

λvar

299 BTIx100 (%)

High reliability

0.760

16.4.3 Social Media for Traffic Information Social-networking sites such as Twitter, Facebook, and Google+ have emerged as very promising sources for any information, where people share information with each other about various topics including traffic conditions/incidents. Such information can be one of the sources for traffic conditions, especially unexpected incidents, because they just not only tell about the location of the incident but also tell about the context of the incident. Twitter, which is one such media to share information, in 140 characters maximum, has become a powerful and inexpensive tool to extract such information. The availability of an application program interface (API) for Twitter makes it more attractive compared to other media. A methodology to crawl, process, and filter public tweets and to classify them into a two-class dataset and further classify the traffic-related tweets into a threeclass dataset was presented in Ahmed et al. (2019). Analysis was carried out to extract traffic information using transportation lexicon and text-mining techniques. A lexicon-based approach and supervised machine learning (ML) approach were used to perform classification of tweets as either traffic tweets or non-traffic tweets, and ML-based approach was used for further classification of traffic-related tweets. The whole methodology was tested using tweets from Chennai and Bangalore city. The overall methodology is divided into three stages. Stage 1 involves extracting the tweets using Twitter APIs based on keywords and coordinates of the place. To derive high-quality information from the raw text of collected tweets, text processing of the tweets using text-mining techniques was attempted in Stage 2. In stage 3, the high-quality information was used to generate features from these traffic-related tweets. Generated features were used to train the classifier to perform classification of tweets into the two-class dataset, i.e., ‘traffic related’ or ‘not traffic related’ and further classification of traffic-related tweets into the three-class dataset, i.e., traffic congestion due to accident, traffic congestion due to external events, and traffic congestion due to road work. For binary classification, two approaches were used: i) Lexicon-based approach and ii) Machine Learning (ML)-based approach, while for further classification of traffic-related tweets, only ML-based approach was used. Tweets were collected from two regions, Chennai and Bangalore metro areas in India. Chennai was chosen as the main study location, and tweets were acquired using both Twitter streaming API and search API. Tweets collected using Twitter APIs were unstructured, irregular, and included all tweets—with and without traffic information. Hence, to remove such errors and get meaningful information, texts of the tweets were processed. Text processing is a set of techniques used to analyze text to obtain desired information from the raw text (in this case, from tweets). It

300

L. Vanajakshi and S. Arkatkar

Fig. 16.15 Pipeline for text processing

involves pre-filtering followed by text-mining techniques. The used pipeline for text processing is shown in Fig. 16.15. There were many tweets, especially, in the case of tweets collected using Twitter streaming API, without any traffic-related words in their text, which are called as noisy tweets. These noisy tweets need to be removed to make them useful for the present study. Pre-filtering was the process used for this, where noisy tweets were removed using a transportation/traffic lexicon containing terms related to traffic and the terms that are commonly used by people to share trafficrelated information. For this purpose, a transportation/traffic lexicon was prepared by studying the existing tweets. It was assumed that tweets that did not contain any keywords were not giving any kind of traffic information and can be removed. This step helped in reducing the number of tweets to be processed in the next step of text mining. On the other hand, the Twitter search API-based data collection approach does this processing at the extraction stage itself as traffic-related keywords and coordinates with ‘OR’ operation are used for choosing the tweets to be downloaded. Texts of the pre-filtered tweets were still unstructured and irregular in nature. Hence, text-mining techniques such as tokenization, stop-word filtering, and stemming were employed to pre-process pre-filtered tweets to derive high-quality information from these texts. Tokenization was the first step in text processing in which all punctuation marks and other non-text characters were removed. It also transforms a stream of characters into a stream of processing units called tokens/words. Stop-word filtering eliminated stop-words, i.e., words that provide very little or no information to text analysis, e.g., articles, pronouns, prepositions, conjunctions, etc. The sole purpose of text mining was to keep only those words from tweets which would help in training the classifier during the classification of texts. The high-quality information obtained after text processing needs to be classified into different groups such as congestion and accidents. Text classification involves two steps: feature extraction and training of classifier using these extracted features. The texts generated after text mining were used to generate features related to different desired classes. These features were then used to train the classifier to perform classification. In the present study, two approaches were used for classification: lexicon-based approach and machine learning (ML)-based approach. A sample comparison of classification results between ML-based classification approach and Lexicon-based classification approach is shown in Fig. 16.16 for Chennai City.

16 Automated Sensors for Indian Traffic: Challenges and Solutions

301

Fig. 16.16 Comparison of classification results (Chennai city)

16.5 Future Research Directions Fusing data from multiple sources such as GPS, Wi-Fi/BT, loop detectors, image processing, and RFID sensors could be used to develop reliable traffic data. The data then can be used for some of the open research problems such as the following: • • • • • • •

Developing Network level fundamental diagram. Network partitioning for developing perimeter control strategies. Developing adaptive traffic signal control systems and bus priority signal controls. Examining network-level travel time reliability. Incorporating travel time uncertainty in traffic and travel demand models. Revisiting the traffic flow stability analysis probabilistically. Developing a probabilistic framework to identify congestion and develop level of service (LOS) criteria for urban road networks. The LOS criteria can facilitate the development of appropriate congestion mitigation measures.

References Abedi N, Bhaskar A, Chung E, Miska M (2015) Assessment of antenna characteristic effects on pedestrian and cyclist’s travel-time estimation based on Bluetooth and Wi-Fi MAC addresses. Transp Res Part C: Emerg Technol 60:124–141 Advani C, Thakkar S, Arkatkar S, Bhaskar A (2020a) Performance evaluation of urban arterial network using Wi-Fi sensors under heterogeneous traffic conditions. Transp Res Procedia 48:1022–1037 Advani C, Thakkar S, Shah S, Arkatkar S, Bhaskar A (2020b) A Wi-Fi sensor-based approach for examining travel time reliability parameters under mixed traffic conditions. Transp Dev Econ 6(1):1–9. https://doi.org/10.1007/s40890-019-0089-1 Ahmed MF, Vanajakshi V, Suriyanarayanan R (2019) Real-time traffic congestion information from tweets using supervised and unsupervised machine learning techniques. Transp Lett: Int J Transp Res 5(2) Ali SM, George B, Vanajakshi L, Jayashankar V (2012) A multiple inductive loop vehicle detection system for heterogeneous and lane-less traffic. IEEE Trans Instrum Meas 61(5):1353–1360

302

L. Vanajakshi and S. Arkatkar

Ali M, Sheik, Boby G, Lelitha V (2014) Mutually coupled multiple inductive loop system suitable for heterogeneous traffic. IET Intell Transp Syst 8(5):470–478 Ali M, Boby G, Lelitha V (2013) An efficient multiple loop sensor configuration applicable for undisciplined traffic. IEEE Trans Intell Transp Syst 14(3) Bakula C, Schneider WH IV, Roth J (2011) Probabilistic model based on the effective range and vehicle speed to determine Bluetooth MAC address matches from roadside traffic monitoring. J Transp Eng 138(1):43–49 Brennan TM, Ernst JM, Day CM, Bullock DM, Krogmeier JV, Martchouk M (2010) Influence of vertical sensor placement on data collection efficiency from bluetooth mac address collection devices. J Transp Eng 136(12):1104–1110 Gore N, Arkatkar S, Joshi G, Bhaskar A (2019a) Exploring credentials of Wi-Fi sensors as a complementary source of transport data: An Indian experience. IET Intell Transp Syst 13(12):1860–1869. https://doi.org/10.1049/iet-its.2019.0251 Gore N, Arkatkar S, Joshi G, Bhaskar A (2019b) A novel methodology to derive vehicle occupancy using Wi-Fi sensors under heterogenous traffic conditions. Presented in 98th Annual Meeting of Transportation Research Board (TRB), 13th–17th January 2019, Washington, DC, USA Halawa LJ, Wibowo A, Ernawan F (2019) Face recognition using faster R-CNN with inception-V2 architecture for CCTV camera Jayan A, Anusha SP (2019) Evaluation of effectiveness of RFID and bluetooth sensors for travel time studies under mixed traffic conditions. Transportation Research Board 98th Annual Meeting Transportation Reasearch Board (19-04397) Lin T-Y, Maire M, Belongie S, Bourdev L, Girshick R, Hays J, Perona P, Ramanan D, Zitnick CL, Dollár, P (2015) Microsoft COCO: common objects in context. In: Proceedings of the IEEE computer society conference on computer vision and pattern recognition. https://doi.org/10.1109/ CVPR.2014.471. Lucas BD, Kanade T (1981) An iterative image registration technique with an application to stereo vision. In: Proceedings of 7th international joint conference on artificial intelligence, Vancouver, Canada, vol 2, pp 674–679 Mathew JK, Devi VL, Bullock DM, Sharma A (2016) Investigation of the use of bluetooth sensors for travel time studies under Indian conditions. Transp Res Procedia 17:213–222 Murphy P, Welsh E, Frantz JP (2002) Using bluetooth for short-term ad hoc connections between moving vehicles: a feasibility study. In: IEEE 55th vehicular technology conference. VTC Spring 2002 (Cat. No. 02CH37367), vol 1, pp 414–418. https://doi.org/10.1109/VTC.2002.1002746 Patra SS, Rama BB, Vanajakshi L (2019) Development and evaluation of a low-cost wifi media access control scanner as traffic sensor. In: COMSNETS: intelligent transportation systems workshop, Bangalore, January 2019 Patra S, Muthurajan B, Vanajakshi L (2021) Development and evaluation of a Wi-Fi MAC scanner and its application in forecasting. In: Prediction intervals for stream travel time, Transportation Research Board, National Research Council, Washington, DC Peli T, Malah D (1982) A Study of edge detection algorithms. Comput Graph Image Process 20(1):1–21 Porter JD, Kim DS, Magana ME, Poocharoen P, Gutierrez Arriaga CA (2013) Antenna characterization for Bluetooth-based travel time data collection. J Intell Transp Syst: Technol Plan Oper 17(2):142–151 Python Community, D. LABELIMG Utility. Project description, p 1. https://pypi.org/project/lab elImg/ Raveendran B, Arkatkar SS, Vanajakshi LD (2019) Development of a video image processing-based micro-level data extractor for non-lane-based heterogeneous traffic conditions. Transp Dev Econ 5(2):1–10 Reddy SK, Kanda BR, O’Byrne M, Vanajakshi L, Ghosh B (2019) Alternative approach to traffic state analysis on Indian roads using image processing, ICE–Transport, vol 172, no 6, pp 336–346 Sakhare R, Vanajakshi L (2019) Reliable corridor level travel time estimation using probe vehicle data. Transp Lett: Int J Transp Res, Accepted 2019

16 Automated Sensors for Indian Traffic: Challenges and Solutions

303

Sakhare R, Mathew JK, Avr A, Hubbard SM, Devi L, Bullock DM (2018) Comparison of Bluetooth and Bus GPS data for estimating arterial travel time and trip chaining. Transportation Research Board, National Research Council, Washington, DC Sanin A, Sanderson C, Lovell BC (2012) Shadow detection: a survey and comparative evaluation of recent methods. Pattern Recogn 45(4):1684–1695 Sharma S, Maripini H, Khadhir A, Arkatkar SS, Vanajakshi L (2020) Analysis and use of Wi-Fi data for signal state identification. Transp Res Procedia 48:1008–1021 Singh V, Gore N, Chepuri A, Arkatkar S, Joshi G, Pulugurtha S (2019) Examining travel time variability and reliability on an urban arterial road using Wi-Fi detections—a case study. J Eastern Asia Soc Transp Stud 13:2390–2411. https://doi.org/10.11175/easts.13.2390 Suzuki S, Abe K (1985) Topological structural analysis of digitized binary images by border following. Comput Vis Graph Image Process 30(1):32–46 Thakkar S, Sharma S, Advani C, Arkatkar S, Bhaskar A (2021) Comparative analysis of travel time prediction algorithms for urban arterials using Wi-Fi sensor data. In: International conference on communication systems & networks (COMSNETS), Bangalore, India, pp 697–702. https://doi. org/10.1109/COMSNETS51098.2021.9352845 Wang L, Zhang Y, Feng J (2005) On the euclidean distance of images. IEEE Trans Pattern Anal Mach Intell. https://doi.org/10.1109/TPAMI.2005.165 Xu D, Li X, Liu Z, Yuan Y (2005) Cast shadow detection in video segmentation. Pattern Recogn Lett 26(1):91–99 Yogesh GKV, Yogesh V (2018) Automated tolling solution with novel inductive loop detectors using machine learning techniques. ASCE J Comput Civil Eng 32(6) Zivkovic Z (2004) Improved adaptive gaussian mixture model for background subtraction. In: Proceeding of 17th IEEE international conference on pattern recognition, Cambridge, England, UK, vol 2, pp 28–31

Chapter 17

Planning for Traditional and Special Area Requirement Travel Modes—Existing Scenario and Way Forward Udit Jain, Dharitri Kahali, Vivek R. Das, and Shriniwas S. Arkatkar

17.1 Background Globally, rapid urbanization and improved socio-economic status from the past few decades have resulted in increased vehicle ownership and more traffic on the roads. Adverse impacts are becoming more severe with multimodal traffic on urban streets (e.g., pedestrians, bicycles, buses, cars, motorcycles, and auto-rickshaws) that share the same street space creating inefficient mobility conditions. In recent years, pedestrian safety has been given greater importance because reports indicate that pedestrians are the single largest category of those injured and killed in road crashes in India. It can also be considered an excellent example of sustainable transportation mode, especially suitable for relatively shorter distances. While walking, people use the sidewalk, crosswalk, and skywalk to reach their destination, and these modes are connected to public transit. As per the annual report of road crashes in India, 15% of

U. Jain Department of Civil Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010, India e-mail: [email protected] D. Kahali Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee 247667, India e-mail: [email protected] V. R. Das Department of Civil Engineering, Dayananda Sagar College of Engineering, Bangalore 560078, India S. S. Arkatkar (B) Department of Civil Engineering, Sardar Vallabhbhai National Institute of Technology, Surat 395007, India e-mail: [email protected] © Transport Research Group of India 2022 A. K. Maurya et al. (eds.), Transportation Research in India, Springer Transactions in Civil and Environmental Engineering, https://doi.org/10.1007/978-981-16-9636-7_17

305

306

U. Jain et al.

pedestrians were killed across the country in 2018. This may be attributed to pedestrian apprehensions which are always neglected in the design process. Because of this, pedestrian behavior and other relevant engineering aspects have to be explored at different facilities such as bus and railway terminals, near educational institutions, markets, special event locations, public offices, etc. Moreover, walking must be considered an integral part of the whole transportation system; appropriate and adequate pedestrian infrastructure facilities need to be created along urban streets to increase the comfort and convenience of pedestrians and increase pedestrian safety. On the other hand, non-motorized transport (NMT) such as bicycles is recognized as a low-cost, high-efficiency mode of transport in road space, with numerous benefits in human health, environment, and improving quality of life. Over the years, it has been seen that the system is underutilized across the world and has tremendous scope in fulfilling short-distance trips. NMT requires minimal space on the roads and can be used for connecting through routes that might be difficult to reach by other modes of transport. NMT has been widely used as a feeder mode to the public transport systems, and they provide ease in traveling short distances to the public transport nodes and create a level of equity among the commuters. There is also a socio-economic bias associated with a bicycle as a mode of transportation in Indian cities. It is often perceived that bicycles are used as the primary transportation mode by the economically weaker society units. There is a strong need to change this perception for NMT modes. This is essential to ensure that NMT is a severe mode of transportation by all society units. Several issues hold back bicycle trips from becoming an essential mode of transport for short and medium-length trips in cities. These include a lack of cycling infrastructure in cities, which leads to the bicyclists sharing road space with motorized vehicles and jeopardizing their safety. A special-purpose conveyor system is a classic piece of mechanical handling equipment that moves materials from one place to another. Since 1975, cable conveyor systems are used in many industries, including mining, automotive, agricultural, computer, electronic, food processing, aerospace, print finishing, packaging, and medicine. However, a wide variety of materials can be conveyed, such as food, dairy, beverage, motor oil, laundry soap, and pet food. Conveyors are primarily helpful in applications involving the transport of heavy or bulky materials. Various types of conveying systems are available and are used according to the multiple needs of different industries. This can be classified as (a) haulage conveyor, which is a particular class of chain conveyor in which load is pushed or pulled and stationary troughs carry the weight, surfaces or rail and which can be drag chain, flight or tow with over-head, flush-floor, and under-floor configuration, (b) cable conveyor, (c) belt conveyors, which can be a flat, trough, closed, metallic, portable, or telescoping type, (d) chain conveyors, which can be apron or pan, slat, cross-bar or arm, car type, carrier chain and flat-top, trolley, power and free, and suspended tray or swing-tray type, (e) pneumatic conveying systems which can be a pipeline, airactivated gravity, or tube-type, (f) hydraulic conveying system, (g) screw conveyor, (h) bucket conveyor which can be with gravity discharge pivoted bucket, or bucket elevator type, and (i) roller conveyor which can be gravity. In this transportation system, acceleration/deceleration of vehicles between the high-speed segments and

17 Planning for Traditional and Special Area Requirement Travel Modes …

307

low-speed segments is provided by variable speed cables gripped by the vehicles and driven at gradually varying speeds. Another special-purpose mode of transport can be an escalator, a moving staircase, which facilitates the movement of pedestrians between two floors. These systems move pedestrians efficiently and have become an essential component of public building design. These can cater to the instantaneous pedestrian flows (like at metro stations) and can be used upward and downward, depending upon the requirement. The speed can be varied to adjust to the requirement of flow dispersal. Currently, escalators are designed and operated based on manuals prepared by companies or Indian Standards or European or British standards. The escalators have been studied relatively less concerning the efficiency analysis and its utilization during peak flows. Being constrained in size and moving continuously, the inexperienced users hesitate to board the step and thus cause capacity issues. The open spaces in front of these facilities have pedestrian accumulation during peak periods, and the pedestrian behavior under such a scenario needs to be studied. It is important to develop criteria for the provision of escalator facilities under the influence of cultural and regional differences. The information may be utilized to formulate design guidelines for escalating public places in developing countries. To summarize, the current chapter attempts to provide a thorough understanding of present challenges and future opportunities in developing traditional modes of transport, including pedestrian and NMT (bicycles) and special-purpose modes of transport such as conveyors and escalators.

17.2 Non-motorized Transport Facilities 17.2.1 Introduction In early 1990, with the start of motorized transportation in Indian cities, people found a faster way to travel. The ones still using non-motorized transport (NMT) modes started facing safety issues as they had to share the same road space with the fast-moving motorized vehicles. Along with the safety issues, owning a motorized vehicle became a symbol of prestige, due to which the bicycle riders started getting classified as economically weaker sections of the society. This social stigma and the safety issues due to lack of infrastructure became the primary reasons for the decline of bicycle modal share in Indian cities. In recent years, with the increase in population levels in cities and the environmental hazards of motorized vehicles, people have again started looking towards NMT as a mode of transportation for their daily travel needs. Today, the biggest hurdle faced by bicycle riders is the lack of NMT infrastructure and increasing trip lengths due to the expanding city sizes. Even though NMT is an economically feasible mode of transportation and is also eco-friendly, the share of NMT users in Indian cities has been steadily decreasing over the past three decades in Indian cities, as shown in Fig. 17.1 (Tiwari and Jain 2008).

308

U. Jain et al.

Fig. 17.1 Declining modal share of NMT in Indian cities (Source Tiwari and Jain 2008)

17.2.2 Characteristics of NMT Modes NMT modes are non-polluting and require no fossil fuels or electricity to run. These are sustainable transport modes in the true sense and have little to no carbon footprints. As the world is shifting towards electric vehicles to promote sustainable transportation, it is essential to understand that the electricity required for running electric vehicles might not always be from renewable energy sources. A significant amount of electricity generated in India is from thermal power plants that use coal and other fossil fuels to generate electricity. Thus, electric vehicles are also not truly sustainable modes of transportation. It is important to promote and support the use of NMT modes for the daily travel needs of the people. Even if the short trips are converted to NMT modes, it would considerably reduce the carbon emissions and the pollution caused by motor vehicles. Again, this can be made possible only after providing substantial NMT infrastructure in cities, making the NMT journey safe and comfortable for the users. Despite all the benefits discussed above, the number of bicycle trips in Delhi has declined from 36 to 7% between 1957 and 2010. With the rise of the automobile sector, the number of motorized vehicles on roads has increased steadily. With the Indian economy getting strengthened over the years, the purchasing power of people increased, and most households were able to afford a motorized mode of transportation. This led to the decline in the modal share of bicycle trips in cities across the country. The increase in motorized traffic started making congestion on Indian roads a common phenomenon. To make things worse, the focus of transportation authorities and professionals to alleviate congestion has usually been on providing wider

17 Planning for Traditional and Special Area Requirement Travel Modes …

309

roads and better infrastructure for motorized traffic. Once the roads are widened, more traffic is induced on the roads, and this loop continues forever. Among the widening roads and ensuring decent vehicular speeds on the roads, bicycle users are often forgotten and neglected. Other NMT modes like tricycles (or cycle rickshaws) have been around for the last three decades in Indian cities. They are still quite commonly used for short trips, access-egress trips, and traveling in old city areas where the road widths are too small for motorized cars. Animal transportation has been around both in rural and urban India for centuries in the form of horse carriages for passengers, bullock carts for passengers, and freight and the use of mules/horses for traveling in hilly terrain. In the last two decades, animal transportation has reduced considerably in Indian cities. This has been due to the difficulty in upkeep and maintenance of animals used for these modes, with the changing times tilting towards faster and cheaper alternatives for travel. However, horse carriages and bullock carts are still easily found even today in small towns and villages where people still use them as a daily mode of transportation, both for passengers and goods.

17.2.3 Review of Past Studies NMT is the mode that requires little to no operational cost, and thus, a lot of people who have low household income are dependent on these modes for their daily travel needs. NMT is often attributed to short trips made by the people for whom NMT is the only available mode of transportation. It has been found that in the informal sector workers in India, 80% of factory workers and 73% of office workers use a bicycle for their work trips. Also, the informal service sector, like painters, plumbers, electricians, gardeners, postmen, milkmen, newspapermen, etc., prefers bicycles (TRIPP 2006). It was found that the distance between the residence and the workplace also plays a significant role in choosing NMT as the preferred mode of transportation. The studies conducted in Mumbai, Pune, Delhi, and Hyderabad indicate that 80% of the trips have trip lengths