Research Methods in Building Science and Technology: Field-Based Analysis and Simulation [1 ed.] 3030736911, 9783030736910

Citation preview

Rahman Azari Hazem Rashed-Ali   Editors

Research Methods in Building Science and Technology

Research Methods in Building Science and Technology

Rahman Azari • Hazem Rashed-Ali Editors

Research Methods in Building Science and Technology

Editors Rahman Azari The Pennsylvania State University University Park, PA, USA

Hazem Rashed-Ali Texas Tech University Lubbock, TX, USA

ISBN 978-3-030-73691-0    ISBN 978-3-030-73692-7 (eBook) https://doi.org/10.1007/978-3-030-73692-7 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 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 Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Introduction

The development of consistent, reliable, and valid research methodologies is a key component of the development of a strong research culture in any discipline. Designing effective research studies that produce reliable and impactful results also requires a comprehensive understanding of the range of available research methodologies, the strengths and weaknesses of each of those methodologies, and the types of research questions and problems each of them is best suited for. Groat and Wang (2013) argue that research methodology is a study of processes, and distinguish between systems of inquiry, which they define as “broad epistemological perspectives such as positivism, structuralism, or post structuralism,” and methods and methodologies, which they define as mid-range processes that are common across the entire range of research in a discipline. They further distinguish between strategies or methods, as the appropriate term for those mid-range processes, and between tactics or techniques as the more appropriate terms to describe specific research activities conducted within a certain methodology. While using slightly different terms, Creswell (2014) offers a relatively similar categorization. Creswell, however, identifies three “Research Approaches,” which he defines as quantitative, qualitative, and mixed methods research. Creswell contends that those research approaches represent: “the plan or proposal to conduct research [which] involves the intersection of philosophy, research designs, and specific methods.” He then provides a framework for the interaction of these three components which he describes as: philosophical worldviews, research designs, and research methods. These three components correlate well with the three levels of worldview, strategies/methods, and tactics/techniques offered by Groat and Wang. In their book about behavioral research, however, Kerlinger and Lee (1999) define four general ways of knowing: the method of tenacity, the method of authority, the a priori method or the method of intuition, and the method of science. They then argue that research is linked to the method of science, or the scientific approach, and offer a stricter definition of scientific research as: “systematic, controlled, empirical, amoral, public and critical investigation of natural phenomena. It is guided by theory and hypothesis about the presumed relations between such phenomena.” Kerlinger and Lee, however, go on to identify four “types of research”: (1) quasi-experimental and N  =  1, (2) v

vi

Introduction

nonexperimental, (3) laboratory experiments, field experiments, and field studies, and (4) survey research. Those categories still loosely correlate with Groat and Wang’s mid-range methods and Creswell’s research designs. While there exists a strong body of literature covering both universal and discipline-­specific research methods, the literature on architectural research methods is very limited. Groat and Wang’s seminal book, Architectural research Methods, is arguably the only book aiming to provide a comprehensive taxonomy of research methods in architecture. Other books which have addressed specific types of architectural research include Brenda Laurel’s (2003) Design Research: Methods and Perspectives, which focuses on the relationship between research and design in the discipline, and Knight and Ruddick’s (2008) Advanced Research Methods in the Built Environment, which offers a general overview of research methods in the built environment. The limited discipline-specific research methods literature in architecture has forced architectural researchers to draw from both universal literature on research methods (e.g., Creswell, 2014; McCuen, 1996; Patten & Newhart, 2018) and the research method literature specific to other disciplines. Examples of discipline-­specific literature include those in social and behavioral science (e.g., Kerlinger & Lee, 1999; Sommer & Sommer, 2002; Teddlie & Tashakkori, 2008), urban and regional planning (Ewing & Park, 2020a, b; MacCallum, 2019; Silva et  al., 2015), engineering (Tang, 2021; Thiel, 2014), psychology (e.g., Morling, 2018; White, 2019), health (e.g., Jacobson, 2017), and education (e.g., Hoy & Adams, 2016). Architectural researchers also rely on research method–specific literature which focuses on qualitative, quantitative, or mixed-methods research (e.g., Creswell & Clark, 2018; Hancock et  al., 2019; Hennink & Hutter, 2010; Patten, 2015; Tracy, 2013), as well as literature on specific data collections and analysis methods and tools such as surveys, statistical analysis, and GIS modeling, among others (e.g., Fowler, 2014; Kiess & Green, 2019; Steinberg, 2015; Wang, 2006). While it could be argued that architectural research does draw from the research traditions of other disciplines, and that embracing and adopting research methods from these disciplines can raise the stature of architectural research, Groat and Wang (2013) rightly argue for the need of architecture- specific approaches to research that are more sensitive to the discipline’s context, dynamics, and constraints. This need is even more evident with regard to research in building science and technology. While research in these areas does draw from related disciplines such as the sciences and engineering, there is a lack of literature specific to how these types of research methods can be applied within the context of buildings and the built environment. The fast pace of technological change in these areas presents an additional challenge in which the available literature does not capture the fast-­ evolving state of the art in the methods, tools, and techniques used in these types of studies. This book aims to cover the range of methodological approaches, methods, and tools currently used in various areas of building science and technology research and addresses the current lack of research methods literature in this field. The book covers the use of measurement-based methods in which data is collected by measuring the properties and their variations in “actual” physical systems,

Introduction

vii

simulation-based methods, which work with “models” of systems or processes to describe, examine, and analyze their behaviors, performances and operations, and data-driven methodologies in which data is collected via measurement or simulation to identify and examine the associations and patterns and predict the future in a targeted system. The book is based on the premise that identifying and applying the appropriate methodology for a research problem requires the researcher to understand the strengths, weaknesses, limitations, and uncertainties associated with each methodology. The book, therefore, aims to present a survey of key methodologies in various specialized areas of building science and technology research including window systems, building enclosure, energy performance, lighting and daylighting, computational fluid dynamics, indoor and outdoor thermal comfort, and life cycle environmental impacts. The key objective of this book is to provide advanced research methodological insight to graduate students and beginning researchers in building science and technology areas. The book is also intended to be used as a teaching resource in graduate-­ level research methods courses in architecture as well as other disciplines typically involved in building science and technology research such as architectural engineering, construction science, and related disciplines. By focusing on specialized areas of research within the larger context of building science and technology, the book aims to provide a deeper and more focused knowledge and understanding of the current state of research in each of these specializations, as well as more helpful insight into how to select the optimum research method for specific research problems and questions. More specifically, the book achieves the following objectives: • Provides a resource that addresses the current lack of specialized research methods books in building science and technology, thus filling an important gap in the current literature • Provides a comprehensive resource combining key methodological issues in each of the major areas of current research within building science and technology • Draws from the expertise of faculty and researchers with extensive experiences in their respective specializations Contributors to this book present an overview of key methodologies, methods, techniques, and tools used in their specialized area of building science and technology research. They will also provide insight on the applications of different methodologies in their research area, and the strengths, weaknesses, opportunities, and limitations offered by each, as well as any challenges or uncertainties that may be faced using them. Contributors will also provide examples from their own research to illustrate how those research methodologies can be effectively applied. The following paragraphs provide a brief overview of the different chapters of the book: In Chapter “Research Methods for Assessing the Thermal and Optical Performance of Building Windows”, the group of authors led by Julian Wang and Yuan Zhao, provide a comprehensive review of research methods to assess thermal and optical performance of window systems in buildings. The chapter starts with a brief history of energy-efficient building window systems and explores how the oil crises of the 1970s contributed to development of these systems. The chapter then

viii

Introduction

covers in situ measurement methods and scalar experiments to assess window properties such as U-factor, solar heat gain coefficient (SHGC), and visible transmittance (VT). It then covers the numerical and computational analysis of windows based on the finite element method and reviews the governing equations and some of the main tools used for window behavior analysis. The chapter proceeds with presenting computational design and optimization techniques that are used to optimize the properties of windows to achieve daylighting, thermal, and energy objectives. It also reviews how windows are used in the human-factor experimental research that evaluate thermal and visual perception. Chapter “Research Methods for Assessing the Thermal and Optical Performance of Building Windows” ends with a review of research methodological trends in building window research such as development of “perfect window.” In Chapter “Research Approaches for Building Enclosure Studies”, Elizabeth Grant reviews current research approaches for building enclosure studies. She starts the chapter by defining the overall intent and types of research questions addressed typically by building enclosure studies, then moves on to provide an overview of the range of research methodologies typically used in these studies. Building on the classification developed by McCuen (1996), Grant identifies four possible methodological approaches for building enclosure studies: laboratory experiments, simulation, field experiments, and field studies, and provides a thorough discussion of the advantage, disadvantage, affordances, and limitations of each. She then presents an interesting approach, based on the work of Suhr (1999), for selecting the optimum methodology for a specific study by comparing the relative advantages of each and recasting the disadvantages of one or more approaches as advantages for the others, thus avoiding the double-counting that can be associated with the more traditional pros-and-cons approach. Grant then offers two case studies from her work on building enclosures as examples of the different methodological approaches she identifies. The first is a field experiment study that aimed to evaluate the storm water runoff reduction potential of varying depths of modular green roof assemblies. The second case study was an example of the field study approach in which Grant investigated the influence of roof reflectivity on the temperatures of air and surfaces in the vicinity of the roof of an academic building. After describing the two studies, Grant offers an in-depth discussion of how the research methodology used in each was selected using the comparison approaches discussed earlier in the chapter. She also provides alternative study designs using the other three methodological approaches and discusses the impact of these alternative designs on the project’s feasibility and quality of outcomes. This comparison allows young researchers to better understand the advantages and disadvantages associated with each of the possible methodological approaches available for their projects. In Chapter “Building Energy Performance Research – Current Approaches and Future Trends”, Hazem Rashed-Ali focuses on research methods for building energy performance. The author starts by making the case for the continued significance of and need for energy performance research within the larger and more expanded context of building performance research. The chapter then provides an overview of current methodological approaches in this area that draws from the

Introduction

ix

classifications developed by Groat and Wang (2013) for architectural research methods and by Kerlinger and Lee (1999) for behavioral research. Based on these, and similar to the conclusions reached by Grant in Chapter “Research Approaches for Building Enclosure Studies”, Rashed-Ali identifies four approaches: simulation, laboratory experiments, field experiments, and field studies. He then introduces four important sub-areas of research within the larger context of building performance research, two of which, analysis of building typologies and post occupancy evaluation, have a considerably long history, while the other two, user behavior research and integrating the building and urban scale, have recently gained more attention in the research community. The chapter then offers a brief overview of existing tools used in energy performance research studies including leading performance simulation software, as well as equipment used in laboratory experiments and collecting field data. Finally, the chapter offers three examples of research studies conducted by the author that illustrate the different approaches available for energy performance research. The first relies primarily on performance simulation to develop performance data for a single-family house that were then integrated into an augmented reality prototype used to teach energy efficiency consents to students at different levels. The second case study relies on collecting field data to assess the impact of replacing the roofs of low-income homes with high-solar-reflectivity (cool) roofs. The third case study combines the use of simulation with the collection of field energy use data to calibrate the simulation model. The calibrated models were then used to assess the potential of selected energy efficiency retrofits for historic homes. The advantages and disadvantages of each approach are also discussed. In Chapter “Research Methods in Daylighting and Electric Lighting”, Mehlika Inanici provides a comprehensive review of research methods in daylighting and electrical lighting research. The chapter starts by the definition of the key goals and objectives of lighting in buildings and presents to the reader how the human ocular system responds to light. The author argues that estimation of quantity and distribution of lighting in architectural space is the foundation of all lighting research and reviews the methods for physical measurement of lighting quantities including photopic and colorimetric measurement, circadian light measurement, and field studies. It then presents computational methods to assess photopic and colorimetric lighting quantity using photopic illuminances, luminances, and tristimulus color space. Computational simulation of circadian light is then presented as a recent development and the author provides a brief survey of chronological developments in this field. While simulation is an important research method in the lighting research, the author highlights some key methodological limitations in using this type of technique. The chapter then demonstrates the advantages of combining physical measurement and computational simulation for calibration and validation and highlights the applications of measured and computed data. The author then presents subjective and objective psychophysical measurements to quantify the relationships between physical stimuli and human sensation and show the applications of computational models and metrics for visual comfort assessment. The chapter ends with a review of perceptual measures and metrics.

x

Introduction

In Chapter “Research Methods in Computational Fluid Dynamics”, Michael Kuenstle and Hazem Rashed-Ali focus on the foundations and applications of computational fluid dynamics (CFD) simulation in the built environment research. The chapter highlights three numerical methods in the CFD codes, including finite element method, finite difference method, and finite volume method, and reviews the extant literature, providing examples that use CFD to address research problems in indoor and urban environments. In some examples, wind tunnel and CFD simulation are used as complementary research methods. The chapter then presents a survey of the CFD tools currently used including dedicated CFD tools and those integrated within whole building simulation packages. The chapter then offers two detailed case studies that use CFD for building design research including structural design. The first investigates the integration of CFD simulation modeling into a wind mitigation design for building structures located in wind hazard prone geographies, while the second uses CFD simulation integrated with digital fabrication techniques to develop and test designs for a storm shutter panel prototype. The two case studies offer good examples for how to integrate this emerging field in research and design activities in architecture. In Chapter “Advancements in Thermal Comfort Modeling Using Modern Sensing and Computational Technologies”, Joon-Ho Choi presents the advancements in indoor thermal comfort modeling using sensing and computational technologies. While the foundations of thermal comfort assessment are comprehensively reviewed in the next chapter (Chapter “Outdoor Thermal Comfort & Human Behavior Factors, Models, and Methodologies”), this chapter highlights the significance of indoor thermal comfort assessment and reviews the limitations of the current models of environmental comfort. The chapter then shows the potentials of the skin temperature as an indicator of thermal comfort that have been confirmed by the studies that use computational algorithms and machine learning techniques. Therefore, physiological signal-based thermal comfort modeling is introduced as a promising method for thermal comfort assessment. Because using sensors worn on users’ bodies is an intrusive method of data collection, non-contact measurement of human physiological signals has emerged as a data collection technique. The chapter also concludes with highlighting the potentials of physiological signals in other parts of human body as well as other advanced techniques for thermal comfort assessment. In Chapter “Outdoor Thermal Comfort & Human Behavior Factors, Models, and Methodologies”, Zahida Khan and Rahman Azari provide a comprehensive review of outdoor thermal comfort assessment factors, models and indices. The chapter presents a survey of environmental, physiological, psychological, and behavioral factors that affect thermal comfort, followed by thermal, empirical, and linear equation-­based indices to capture thermal comfort. It then covers some of the common models including predicted mean vote (PMV), predicted percentage of dissatisfied (PPD), adaptive thermal comfort model, Klima-Michel model (KMM), standard effective temperature (SET), physiological equivalent temperature (PET), and universal thermal comfort index (UTCI). The chapter then presents a survey of empirical and numerical research methods in outdoor thermal comfort assessment

Introduction

xi

and concludes by demonstrating their applications in the extant literature. An important part of this chapter is the review of the literature with regard to their approaches in measuring human behavior in outdoor thermal comfort which is illustrated in Fig. 5 in this chapter. The research techniques for human behavior assessment range from survey-based methods to video camera recording, GIS mapping, and agent-­ based modeling (ABM). Finally, in Chapter “Life Cycle Assessment as a Research Methodology for Estimating the Environmental Impacts of Buildings”, Rahman Azari and Negar Badri present research methods in the field of life cycle assessment (LCA), as the methodology to quantify the environmental impacts of built environments. The chapter starts with a comprehensive compilation of key environmental flows of the building stock in the United States, including operational energy, operational carbon, consumption of materials (such as steel, cement, glass, and timber), and solid waste generation. The chapter then presents the principles and the workflow of the LCA methodology and demonstrates its application in a case-study research that aims to isolate the environmental impacts of building skins. The LCA results of this case-study are presented and compared with the available benchmarks. The chapter then reviews some of the main limitations of LCA studies that lead to result discrepancy and uncertainty and presents a framework to standardize the reporting of LCA results, as summarized in the form in Table 7.

References Creswell, J. (2014). Research design, qualitative quantitative and mixed methods approaches (4th ed.). Sage. Creswell, J., & Clark, V. (2018). Designing and conducting mixed methods research (3rd ed.). Thousand oaks: Sage. Ewing, R., & Park, K. (2020a). Basic quantitative research methods for urban planners (APA planning essentials). New York: Routledge. Ewing, R., & Park, K. (2020b). Advanced quantitative research methods for urban planners. New York: Routledge. Fowler, F. (2014). Survey research methods (5th ed.). Thousand Oaks: Sage. Groat, L., & Wang, D. (2013). Architectural research methods (2nd ed.). Hoboken: Wiley. Hancock, G., Stapleton, L., & Mueller, R. (2019). The reviewer’s guide to quantitative methods in the social sciences (2nd ed.). New York: Routledge. Hennink, M., & Hutter, I. (2010). Qualitative research methods. Thousand Oaks: Sage. Hoy, W., & Adams, C. (2016). Quantitative research in education: A primer (2nd ed.). Thousand Oaks: Sage. Jacobson, K. (2017). Introduction to health research methods (2nd ed.). Burlington: Jones and Bartlett Learning. Kerlinger, F., & Lee, H. (1999). Foundations of behavioral research (4th ed.). Wadsworth Publishing. Kiess, H., & Green, B. (2019). Statistical concepts for the behavioral sciences (4th ed.). New York: Cambridge University Press. Knight, A., & Ruddock, L. (Eds.). (2008). Advanced research methods in the built environment. Chichester: Wiley-Blackwell.

xii

Introduction

Laurel, B. (Ed.). (2003). Design research, methods and perspectives. Cambridge, MA: The MIT Press. MacCallum, D. (2019). Doing research in urban and regional planning (Natural and built environment series). New York: Routledge. McCuen, R. H. (1996). The elements of academic research. New York: ASCE Press. Morling, B. (2018). Research methods in psychology: Evaluating a world of information (3rd ed.). W.W. Norton and Company. Patten, M. (2015). Qualitative research & evaluation methods: Integrating theory and practice (4th ed.). Thousand Oaks: Sage. Patten, M., & Newhart, M. (2018). Understanding research methods: An overview of the essentials (10th ed.). New York: Routledge. Silva, E., Healey, P., Harris, N., & Van der Broeck, P. (2015). The Routledge handbook of planning research methods (Routledge handbooks). New York: Routledge. Sommer, R., & Sommer, B. (2002). A practical guide to behavioral research, tools and technique (5th ed.). Oxford: Oxford University Press Steinberg, S. (2015). GIS Research methods, incorporating spatial perspectives. New  York: ESRI Press. Suhr, J. (1999). The choosing by advantages decisionmaking system. Westport: Quorum. Tang, H. (2021). Engineering research: Design, methods, and publication. Hoboken: Wiley. Teddlie, C., & Tashakkori, A. (2008). Foundations of mixed methods research: Integrating quantitative and qualitative approaches in the social and behavioral sciences. Thousand Oaks: Sage. Theil, D. (2014). Research methods for engineers. Cambridge: Cambridge University Press. Tracy, S. (2013). Qualitative research methods: Collecting evidence, crafting analysis, communicating impact. Hoboken: Wiley. Wang, F. (2006). Quantitative methods and applications in GIS. Boca Raton: CRC Press. White, T. (2019). Research methods (9th ed.). Wadsworth: Cengage learning.

Contents

 Research Methods for Assessing the Thermal and Optical Performance of Building Windows����������������������������������������������������������������    1 Yuan Zhao, Yanxiao Feng, Qiuhua Duan, Nan Wang, Laura E. Hinkle, Enhe Zhang, Nathan Brown, and Julian Wang  Research Approaches for Building Enclosure Studies����������������������������������   33 Elizabeth J. Grant  Building Energy Performance Research – Current Approaches and Future Trends��������������������������������������������������������������������������������������������   51 Hazem Rashed-Ali  Research Methods in Daylighting and Electric Lighting ����������������������������   71 Mehlika Inanici  Research Methods in Computational Fluid Dynamics��������������������������������   95 Michael Kuenstle and Hazem Rashed-Ali  Advancements in Thermal Comfort Modeling Using Modern Sensing and Computational Technologies ����������������������������������������������������  115 Joon-Ho Choi  Outdoor Thermal Comfort & Human Behavior Factors, Models, and Methodologies ������������������������������������������������������������������������������������������  131 Zahida Khan and Rahman Azari  Life Cycle Assessment as a Research Methodology for Estimating the Environmental Impacts of Buildings ������������������������������������������������������  151 Rahman Azari and Negar Badri Index������������������������������������������������������������������������������������������������������������������  175

xiii

Research Methods for Assessing the Thermal and Optical Performance of Building Windows Yuan Zhao, Yanxiao Feng, Qiuhua Duan, Nan Wang, Laura E. Hinkle, Enhe Zhang, Nathan Brown, and Julian Wang

Outline In this chapter, we systematically introduce research methods for assessing the thermal and optical performance of building windows and demonstrate the application of select methods via examples from previous research projects. The chapter consists of six sections, including: 1. The history of and rationale for energy-efficient building windows. This section explains the evolution of energy-efficient building windows and the concepts and rationales supporting this development procedure. 2. Experimental methods for building windows. This section introduces typical experimental methods, including scaled and full-scale models for testing window performance, in situ measurement methods, and tools developed and used in labs and/or industry. 3. Numerical methods for building windows. This section discusses conventional and emerging simulation methods for use in analyzing windows’ thermal and optical performance, including regular window analysis programs, whole building energy simulation methods, and COMSOL Multiphysics software for transient analysis. 4. Computational design and optimization techniques for building windows. This section provides examples of state-of-the-art computational design technologies and optimization methods used for building windows. In a parametric coding environment, intelligent, multi-objective optimization algorithms can be applied to identify the optimal balance between competing characteristics related to building windows.

Y. Zhao · Y. Feng · Q. Duan · N. Wang · L. E. Hinkle · E. Zhang · N. Brown · J. Wang (*) Department of Architectural Engineering, Pennsylvania State University, University Park, PA, USA e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 R. Azari, H. Rashed-Ali (eds.), Research Methods in Building Science and Technology, https://doi.org/10.1007/978-3-030-73692-7_1

1

2

Y. Zhao et al.

5. Subjective study for building windows. This section describes the general procedure and methods in human-factor experiments evaluating subjective thermal and visual perceptions across different windows and spaces. 6. Research trends and other issues. This section discusses the gaps, trends, and new topics in building window research.

1  H  istory of and Rationale for Energy-Efficient Building Windows When we build or renovate a house, it is commonly encouraged to use energy-­ efficient windows to reduce future utility bills. However, the history of the energy-­ efficient building window is surprisingly short, only about 40  years. In the late 1970s and early 80s, energy prices increased ten-fold due to the oil crisis. The high cost of energy promoted the investigation of energy-efficient windows since windows are a key source of heat loss due to their poor insulating properties. At that time, single-pane windows were the primary choice in most residential buildings. Although the concept of the double-pane window first emerged in the 1860s, no one took the notion seriously until the late 1970s (Jones, 2020). It was at that time that double-pane windows truly became acceptable for residential construction. Inspired by storm windows, double-pane windows include space between two layers of glass. This space is filled with insulating gases such as argon or air to improve thermal insulation. Next, triple- or multi-pane windows were developed to encourage even better insulation. Researchers also soon learned that window frames were significant sources of heat loss; the most popular frame material, aluminum, is an extremely poor insulator. Vinyl and composite frames were then produced to replace aluminum frames, soon sweeping the market (The evolution of your double pane windows starts with ancient history, 2020; From energy sink to energy efficient: A walk through window technologies, 2020; The history of energy efficient windows, 2020). At the same time in the 1970s, low-emissivity (low-e) coatings for window glass were developed and quickly boomed in the market. The Pilkington and Flachglas Group created the first commercially available low-e coating using thin layers of gold (Gläser, 2008). However, the coating was green in color. The German glass manufacturer Interpane developed the first colorless low-e coating in 1981; it was made of silver (AGC Interpane, 2020). Due to the Department of Energy’s substantial support for research related to this topic, by 1988, 20% of windows sold in the US had low-e coatings (Matulka, 2020). Today, more than 80% of residential windows and 50% of commercial windows have low-e coatings, saving consumers billions of dollars in energy costs (Matulka, 2020). In the 1990s and 2000s, electrochromic and thermochromic glazing materials were investigated, respectively. Both techniques are based on a dynamic system that allows coating materials to darken the window in certain circumstances: when

Research Methods for Assessing the Thermal and Optical Performance of Building…

3

low-­voltage electricity is actively applied, and passively by sun heating, respectively. The tinted glass limits sunlight entering the building, reducing solar heat gain, especially in tropical regions and hot climate zones. In the past decade, thin-film technologies have made possible two new benefits to energy-efficient windows: photothermal effects and photovoltaics. In 2016, the Department of Energy’s Advanced Research Projects Agency-Energy promoted a program called Single-Pane Highly Insulating Efficient Lucid Designs. The goal was to reduce heat loss by developing innovative transparent and insulating materials for retrofitting existing single-pane windows. Various spectrally selective coatings or films have been developed since then. One of the characteristic technologies is electrochromic window films that have been developed to address this critical issue in the past several decades. A new approach based on plasmonic electrochromism has recently been proposed and demonstrated by DeForest et al. (2017). One of the unique features of EC materials is the ability to selectively control NIR absorption via modulated localized surface plasmon resonance (LSPR). Another representative emerging spectrally selective technology is based on the recently discovered photothermal effect of metallic nanoparticles. Such effects have been mainly studied for biomedical applications, while just recently been investigated for energy-efficient windows. Researchers have claimed that by applying spectrally selective photothermal films, non-visible light, especially in the NIR region, can be utilized and converted to heat for self-heating of windows, improving their insulating properties (Zhao et al., 2017). Another similar direction reflects the increasing improvement of solar panel technology. Researchers have investigated thin-film solar panels with transparent solar energy absorbers, integrating the technology into windows. These new designs not only save building energy but also harvest solar energy for other uses (Lunt & Bulovic, 2011).

2  Experimental Methods for Building Windows In this chapter section, we introduce the methodology employed by energy-efficient windows research, including in situ measurement methods, small- and large-scale experiments, human-factor trials, computational simulations, and design optimization techniques.

2.1  In Situ Measurement Methods There are two main types of window property measuring systems currently in use. The National Fenestration Rating Council (NFRC) performance rating is mainly used in North America. The International Organization for Standardization (ISO) rating is primarily used in Europe. They not only define performance rating levels for window products but also set standards for measuring the properties of glass and

4

Y. Zhao et al.

whole windows. Although the two rating standards measure almost the same parameters, they do have some differences in terms of measuring algorithms, such as in window sample selection, boundary conditions settings, solar radiation received, etc. 2.1.1  Key Measurement Standards for Window Properties The NFRC rating system is a reliable and widely used energy performance rating system in North America. It evaluates window performance based on the following properties: insulating factor (U-factor), solar heat gain coefficient (SHGC), visible transmittance (VT), air leakage, ventilation rating, and condensation resistance. The first three are the most discussed and mandatory for NFRC-certified fenestration products. The ANSI/NFRC 100 and 200 explain the procedures for determining the U-factor and SHGC/VT, respectively, for fenestration products (ANSI/NFRC 100-2017, 2017; ANSI/NFRC 200-2017, 2017). The procedures defined therein all comply with Standard ISO 15099 (2003) and have been referenced in building codes such as ASHRAE Standard 90.1 and the International Energy Conservation Code. Standard ISO 15099 can be used via a computer simulation to calculate the total fenestration product U-factor, total solar energy transmittance, and VT. Both Standards ISO 10077 and 15099 can be used to calculate the total U-factor and solar energy transmittance for windows products. Apart from the computer simulation processes mentioned above, some standards include in situ experimentation methods to obtain windows’ thermal properties. Standard ISO 9869 (2014) describes experimental procedures for measuring the thermal transmittance of glass and Standard ISO 19467 (2017) provides a standardized method for measuring the SHGC values of complete windows. 2.1.2  U-Factor Measurement NFRC 100 uses two different test procedures to determine the total fenestration U-factor and glazing component U-factor, including the center-of-glazing U-factor. Computer simulations using the latest approved software are required to calculate this property value, except when the area-weighted method outlined in Section 4.1.3 of ISO 15009 is applied. This is the only method for overall U-factor calculation (Section 4.3.1 NFRC 100). THERM and WINDOW are two approved software packages for 2D vertical simulation of components’ transferred conducted heat. Physical testing is an alternative method for products that cannot be simulated. The NFRC 100 process complies with the method defined in Standard ISO 15099. It is used to calculate the total U-factor when boundary conditions differ from one another. Standard ISO 10077, widely used in Europe, differs not only in terms of the boundary conditions setting but also in the calculation method employed. The whole product U-factor is calculated according to ISO 10077-1, and the U-factor calculation for the frame profiles (considering linear thermal transmittance) is specified in ISO 10077-2 (ISO 10077-1, 2017; ISO 10077-2, 2017). For the

Research Methods for Assessing the Thermal and Optical Performance of Building…

5

Table 1  Boundary Condition Comparison of NFRC and ISO Standards for U-factor Calculations U-factor Rating Standard Boundary condition Interior ambient temperature Tin (°C) Exterior ambient temperature Tin (°C) Exterior wind speed (m/s) Interior coefficient boundary (W/m2K)

Exterior coefficient boundary (W/m2K)

NFRC 100 21.0

ISO 10077 20.0

EN673 ISO 15099 20.0 20.0

−18.0

0

0

0

4 7.69

4 3.6

25.0

20.0

5.5 Aluminum frame Thermally broken frame Thermally improved frame Wood/Vinyl frame 26.0

4 3.29 7.69 3.00 3.12 2.44 25.0

glazing U-factor, Standard EN 673 from the European Committee for Standardization can be used for flat and parallel surfaces. It is also referenced in ISO 10077-1 for center-of-glass U-factor calculations. A comparison of experimental boundary conditions for the NFRC and ISO standards is shown in Table 1. In particular, Standard 9869 develops an in situ heat flow meter (HFM) method for calculating the U-factor. It follows the definition in ISO 7345, which states that thermal transmittance equals the heat flow rate in the unit area, with one degree of the temperature difference between both sides of the system in a steady state (ISO 7345, 2018). The quasi-steady state is assumed to be a good approximation of the steady state because in practice it cannot be reached on site. The environment to environment U-factor can be determined via Eq. 1 (Section 3.1 ISO 9869). U=

φ q = Ti − Te A ( Ti − Te )

(1)

where q = density of heat flow rate (W/m2), ϕ = heat flow rate (W) Ti = interior temperature (°C or K), Te = exterior temperature (°C or K) An HFM is used to measure the heat flow transmitted through glass. The procedure is shown in Fig. 1. Several surface temperature sensors are used to measure the surface temperature on the same side as the HFM. An environmental temperature sensor is also needed to measure the other side’s temperatures, such as the air temperature. Sensor selection significantly influences measurement accuracy. The heat flux sensor should have low thermal resistance and a high level of sensitivity to manifest signals for low heat flow rates. Thin thermocouple and flat resistance thermometers are usually selected to measure surface temperature, due to their high level of accuracy. These sensors must avoid direct solar radiation during measurement. Before starting a measurement, the heat flux sensor should be calibrated using

6

Y. Zhao et al.

Fig. 1  Measurement procedure for window U-factor

an absolute test method such as the guarded hot plate apparatus (ISO 8302) or an HFM apparatus (ISO 8301) on different materials and under various environment temperatures and heat flow rates (ISO 8301, 1991; ISO 8302, 1991). The measurement is finished when the result variation across three nights is within 5%. Following the HFM method, a representative in situ U-factor measurement tool was developed by the greenTEG company (gSKIN, 2020). This measurement instrument uses a gSKIN heat flux sensor and two temperature sensors. Since the sensors are designed to be attached to the surface of the glass, this design measures the center-of-glazing U-factor. The measurement should begin at night when there is no direct solar irradiation. The greenTEG software processes the measured data and evaluates the measurement reliability. For a group of measurements taken from the same object, a quasi-steady state is expected to be reached in a few hours. If the maximum difference in the results does not exceed 5%, the value satisfies the criteria described in ISO 9869 and the instrument is considered to have a reliable level of accuracy. Daytime measurements feature significant deviation and should be avoided. 2.1.3  SHGC Measurement Standard NFRC 200 specifies that the component SHGC values for center-of-­ glazing and edge-of-glazing can be determined using approved software packages, two of which are WINDOW and THERM. In accordance with ISO 15099, Standard NFRC 200 considers the combined effects of heat transfer resulting from conduction, convection, and radiation by using a numerical heat balance equation. Standard ISO 10077 and EN 673 describe a simplified version of this method. ISO 9050 is referenced by ISO 15099. ISO 9050 and EN 410 are considered virtually identical by the LBNL program Window Optics (ISO 9050, 2003; EN 410, 2011). EN 410 references EN 673 and can be used only for center-of-glazing solar energy transfer calculations. The different standards define various boundary conditions for the simulation. A comparison of the boundary conditions for the solar energy transfer

7

Research Methods for Assessing the Thermal and Optical Performance of Building…

Table 2  Boundary Condition Comparison for Standard NFRC and ISO Values for Solar Energy Gain Calculations Boundary Condition Interior ambient temperature (°C) Exterior ambient temperature (°C) Exterior wind speed (m/s) Incident solar irradiance (W/m2) Solar spectrum

Solar Energy Gain Rating Standard NFRC 100 & 200 ISO 10077 EN673 24.0 25.0 25.0 32.0 30.0 30.0 2.75 1 1 783 500 500 NFRC 300 EN 410 EN 410

ISO 15099 25.0 30.0 1 500 ISO 9050

calculation is shown in Table  2. The nomenclature for solar energy transfer in Europe is the g-value, while SHGC is used by the NFRC. In addition to commonly used calculation methods, a calorimetry hot box process using solar calorimeters is described in NFRC 201 as an interim method for measuring the SHGC value of non-standard glazing when the computer simulation process described in NFRC 200 cannot be employed (ANSI/NFRC, 2017). The solar calorimeter can be installed indoors using an artificial light source to imitate direct beam solar radiation, or outdoors at a nonvertical orientation to minimize the influence of incident angle variation. In this method, several instruments such as pyranometers, temperature sensors, and heat flow meters are needed to measure parameters such as incident solar irradiation, fluid temperature, fluid flow rate, etc. The SHGC is expressed in Eq. 2 (Eq. 3-1 in Section 3.2 of the NFRC 201):

SHGC = τ S + Niα s

(2)

where τs = solar transmittance of the window system Ni = inward release fraction of the absorbed solar radiation αs = absorptance of a single-pane window system Another experimental measurement is described in Standard 19467. Calculation of the SHGC value follows a principle similar to what is outlined in ISO 15099 and considers all of the impacts of incident solar radiation. The SHGC calculation, which is defined as gm, is shown in Eq. 3 (Formula (1) in Section 5.1 of ISO 19467):



gm =

qin − qin ( qsolar = O ) qsolar



(3)

where qsolar = the net heat flux of incident radiation, in W/m2 qin = the net heat flux through the test specimen with irradiance, in W/m2 qin (qsolar = 0) = the net heat flux through the test specimen due to thermal transmission without irradiance when the temperature difference between the internal and external sides is (𝜃𝑛𝑒 − 𝜃𝑛𝑖), in W/m2 𝜃𝑛𝑒 = the external environment temperature with irradiance, in °C 𝜃𝑛𝑖 = the internal environment temperature with irradiance, in °C

8

Y. Zhao et al.

This measurement method has two stages. The first determines the net heat flux through a window with irradiance; the measured value includes the effects of solar heat gain and thermal transmission. A radiometer must be put on the climatic chamber’s side in front of the test sample to measure the heat flux of incident radiation. The next stage is to measure the net heat flux without irradiance, which means the heat flux resulting from the thermal transmission. This requires a solar simulator, climatic chamber, and metering box (Section 6 ISO 19467) to set up the test apparatus. The solar simulator should be in a steady state, and the non-uniformity and temporal instability cannot exceed 5%. The climatic chamber simulates the external environment and is equipped with a transparent aperture, airflow generator, and surrounding panel aperture (Section 6 ISO 19467). A metering box is used to provide the internal environment and measure the heat flux. The environmental conditions for measurement can be adjusted in accordance with local standards and regulations, while the climatic chamber and metering box need a low level of relative humidity to avoid possible condensation or other factors that might affect the measurement. To ensure reliability, a stable state for heat flow should be verified before the measurement starts. Currently, only rarely are new instruments developed for field measurement of SHGC values, due to the complexity of necessary measurement tools and conditions. A tool developed by EDTM can measure the approximate SHGC value of a window sample. It has recently been improved to conduct measurements of installed windows. 2.1.4  VT Measurement VT measurement is relatively simple compared to U-factor and SHGC factor calculations. Standard NFRC 200 and ISO 15099 define the computer simulation method used to calculate visible light transmittance. VT can also refer to the total fenestration VT and component VT, such as center-of-glazing VT and edge-of-glazing VT. The measurement environment follows Standard NFRC 200. The purpose is to calculate the ratio of visible lighting passing through the window product to the incident radiation. The spectral distribution of solar energy can vary based on location, season, and time of the day, so it is recommended that early morning and late afternoon be avoided to reduce the influence of significant variations in solar spectra. A minimum solar altitude angle of 20 degrees is required. For most uncoated and tinted window systems, the measurement error with normal incidence is small. The main equipment needed for laboratory measurement includes a photometric sensor, large-diameter integrating sphere, and tubular daylighting device. There have been only a few in situ VT measurement tools developed that enable simplified field measurements. The simplified algorithm is expressed in Eq. 4.



VT =

E Ein

(4)

Research Methods for Assessing the Thermal and Optical Performance of Building…

9

where E = daylight passing through the window product, in lux (lm/m2) Ein = total incident daylight landing on the window product, in lux (lm/m2) For example, EDTM developed a window energy profiler kit that measures visible light transmittance. A group of glass samples with known properties come with the tool, so users can practice evaluating the accuracy of the measurements. The NFRC and ISO standards have outlined computer simulation and experimentation methods for use in labs, and defined standard procedures and environmental conditions that guarantee accuracy and reliability. Following these standards, more in-field measurement tools can be developed that enable engineers, professionals, and even homeowners to approximately measure a window system’s thermal properties since in field tests, environments and conditions cannot be controlled. Also, measurements mostly focus on center-of-glazing properties rather than those of the whole fenestration. For instance, a portable and easy-to-use in situ measuring system for building windows using the Arduino platform and low-cost sensors have been studied, fabricated, and then examined in the (Feng & Wang, 2020; Feng et al., 2020a). It is designed specifically to in situ measure the glazing properties, including Center-of-glass U-factor, Solar Transmittance (τs), and VT.  We devised the measurement system and associated sensors based on thermodynamic equations and intended to simplify the measuring procedures. For general use by homeowners, this device enables a simple, quick, and reliable in situ approximation of glazing properties, with about 97.2%, 93.3%, and 92.1% accuracy for VT, τs, and Center-­ of-­glass U-factor, respectively.

2.2  S  mall-Scale Experiments Methodology for Thermal and Optical Window Performance In the early stages of investigating new materials and designs for a window system, small-scale experiments are normally required prior to larger-scale testing. This saves material, time, labor, effort, and many other costs. Importantly, experiments on the nano- and/or micro-scales enable researchers to discover phenomena from the material perspective, such as photoexcitation effects, layerwise thermal-­ mechanical behaviors, etc. Such nano- and micro-scale properties can later be incorporated into large-scale experiments and simulation work. In some situations, existing full window analytical models and methods are not appropriate and cannot accommodate new physical relationships on the nano- or micro-scales, so new analytical models from the glazing material perspective can be produced and validated from these small-scale experiments. Small-scale testing is often conducted collaboratively with materials scientists, concentrating on one particular material or composite exhibiting special effects under certain conditions, which can then be engineered into glazing or coating materials. Large-scale or full-scale experiments focus on overall window structures, including multiple-layer materials and window frames. Even in a small-scale

10

Y. Zhao et al.

experiment, multiple factors should be considered, including glazing, coating or film materials, lighting, and particular circumstances. Experiment design and setup and sample preparation, preservation, and characterization are also crucial to the testing process. Figure 2 presents the general workflow for small-scale experiments analyzing a glazing’s optical and thermal properties. Most building windows are made of ordinary glass, but they can be comprised of other materials such as acrylic, tempered glass, or other functional glass. Carefully selecting samples for coating or glazing substrates is important, since different materials vary in their surface properties, which may affect the coating process.

Fig. 2  Flowchart of small-scale testing for glazing material thermal and optical performance

Research Methods for Assessing the Thermal and Optical Performance of Building…

11

Before coating or any other use of a glass sample, a standard cleaning procedure is highly recommended: • Step 1. Place the glass substrate samples in a container (with lid) filled with acetone. • Step 2. Ultrasonicate the container and samples for 10 minutes. • Step 3. Rinse the samples with methanol and DI water, then blow dry with nitrogen. • Step 4. Place glass substrate samples in a container (with lid) filled with isopropyl alcohol. • Step 5. Ultrasonicate the container and samples for 10 min. • Step 6. Blow dry with nitrogen or in a vacuum oven. There can be many different methods for the coating process, based on deposition technology. For liquid deposition methods, the surface property of the glass substrate may have to be modified. Pure water and perfectly clean glass have a very small contact angle, meaning that clean glass is hydrophilic (i.e., easier to wet with water or aqueous solutions). If a hydrophobic solution is used during processing, surface modification of the glass substrates should be conducted. Some of the coating or glazing may need preservation to avoid corrosion by ultraviolet (UV) light, oxygen, moisture, or other environmental factors, as well as physical damage from rubbing, scraping, striking, etc. If the glass samples are kept in a refrigerator or freezer, they should be sealed in containers with desiccants to avoid condensation. Lighting conditions are very important in small-scale experiments determining glazing samples’ thermal and optical performance, and even in thermal property tests. The light source can be natural or artificial. For natural lighting, if the solar light is directly irradiated onto the samples, possible UV corrosion of the coating samples should be considered and measured. Solar light may also heat a sample, in some circumstances up to 65 °C or higher (How hot do solar panels get? Effect of temperature on solar performance, 2020). For artificial lighting, if lighting performance is to be measured, the spectra and intensity of the light source should also be measured, especially since some coatings may absorb or reflect certain wavelengths. Irradiation angle and light spot size and shape should also be controlled. For both natural and artificial lighting, if the glazing sample’s temperature is recorded, the photothermal effect of the coating and glazing sample itself should be considered. Figure 3 shows an exemplary setup for measuring a coating material’s photothermal effect under a solar simulator in a lab environment. Fig. 3  Sketch of a photothermal effect experiment setup under a solar simulator (Zhao et al., 2017)

12

Y. Zhao et al.

For small-scale samples, circumstances may have a significant effect; therefore, careful control and recording are essential. In addition to well-known factors such as temperature and humidity, in some situations, exposure time and airflow rate can also affect samples. If outdoor experiments are conducted, temperature, humidity, and lighting can change with time. If the temperature is a key factor in the experiment, the airflow rates must be recorded (or avoided via a closed system), since convection may be dominant in heat transfer. Neglecting convection by avoiding the airflow rate is a simple and convenient means of conducting future calculations. A scaled physical model can have significant effects on large-scale experiments. Careful control and recording are essential. The model’s sealing and insulation prevent unnecessary heat exchange through non-window parts. The distance from the solar simulator to the physical model also needs to be considered. All settings should be carefully controlled under experimental conditions. Air temperature and humidity should be steady before the experiment starts. The validation process is indispensable in large-scale experiments, especially when so many variables may affect an experiment’s results. Multiple factor analysis must be conducted and most meaningless factors neglected by carefully controlling the experiment’s circumstances. It is worth providing an error analysis since sometimes errors are magnified by environmental factors.

2.3  Large-Scale/ Full-Scale Experiments Large-scale experiments for windows focus on testing and measuring overall structural properties, including multiple-layer materials and window frames. These experiments usually involve scaled physical models, thermal control chambers, and solar simulators. Sample windows are assembled on these scaled physical models to obtain both visual effects and experimental measurements. Thermal control chambers and solar simulators are sources for simulating the ambient environment. The goal of large-scale experiments is to develop various environmental scenarios reflecting a variety of ambient air temperatures and solar irradiance levels and then conduct experimental testing on the thermal behavior of the samples. The results can be used to validate simulation settings and other related parameter adjustments in future work. A proper physical model can be determined by its design. An example of a simple model is a well-insulated and sealed moving box with an opening in the front for a standard-sized window sample. Solar simulators need to simulate solar spectra in order to estimate the thermal behavior of windows in sunlight. For example, a xenon lamp is often used as a solar simulator. There are two types of thermal control chamber: series environmental and walk-in. Walk-in chambers are much larger than thermal chambers, while thermal chambers have a wider range of temperature control. In combined systems, the inner temperature of the physical model is maintained at a certain level (e.g., 25 °C, room temperature) via micro-heating, ventilation, or air conditioning units, while the external air temperature simulates winter or summer

Research Methods for Assessing the Thermal and Optical Performance of Building…

13

Fig. 4  Schematic diagram of the full-scale experiments in a walk-in weather chamber and a solar simulator, with red spots indicating sensors

via a thermal chamber. Solar irradiance is provided by an artificial solar simulator. The placement of the solar simulator must be adjustable to precisely reflect the solar position. A series of sensors are placed on the scaled physical model and window surfaces to record data such as surface temperature, air temperature, heat flux, and airflow (Fig. 4). Most parameters in such full-scale experiments can be easily measured by sensors. Ambient temperature and humidity are measured by sensors built into the thermal chamber, and various choices of sensors can be added to the physical model and on the window’s surface. In a typical window property testing experiment, the surface temperature of the window sample, air temperature of the model (both outside and inside), humidity, efficiency of luminance, humidity, and window transmission are the most common topics of interest. For example, a series of Campbell Scientific environmental sensors that collect information on surface temperature (110PV-L thermistor), air temperature (107 temperature probe), and directional reflectance (Emissometer DS AE1/RD1) will accomplish the goal. All data recorded can be transferred to a data-logging station.

3  Numerical Analysis Methods for Building Windows 3.1  Finite Element Method The finite element method (FEM) is a simulation approach effective for analyzing the performance of windows with different (i.e., either simple or complex) geometries, materials, and boundary conditions. The FEM divides a window system into

14

Y. Zhao et al.

small elements whose solutions can be approximated. Each element is connected by a number of points in space called nodes. Each node has a set of degrees of freedom such as temperature, displacement, etc. All individual elements in the model are combined to create a set of equations that represent the window system to be analyzed. Finally, the equations are solved to reveal useful information about the behavior of the window system. A finite element model approaches a perfect representation of the window system as the number of elements becomes infinite. In previous studies, Ye used the FEM to study convection through a window glazing system (Ye, 1998). Philips et al. investigated convective and radiative heat transfer in a window glazing system with a Venetian blind (Phillips et al., 2001). Shahid and Naylor solved the convection and radiation heat transfer problem in a window system using a two-dimensional finite volume model (Shahid & Naylor, 2005). Beck and Arasteh studied how to improve the thermal performance of commercially available vinyl profiles and glazing edge systems (Beck & Arasteh, 1992). There are a number of commercially available computer programs that can be used to study the integrity and design of structural parts and efficiency and design of thermal, fluid, magnetic, and electrical systems. Many general-purpose finite element software packages have been used by mechanical, civil, aerospace, electrical, and chemical engineers in various industries. These include COMSOL, ANSYS, ABAQUS, ADINA, and THERM. However, FEM software is just a tool. It is based on the same theory and has similar functions. In this section, we choose three of these methods and introduce their applications in the analysis of building windows. Representative projects using Autodesk CAD, COMSOL, and THERM as examples to simulate and analyze window system performance are introduced below. The general procedure of computational analysis for a window system using the FEM includes the following six steps: • • • • • •

Step 1: Create a geometric model of the window system. Step 2: Discretize the window system into a set of finite elements. Step 3: Specify the material properties of the window system. Step 4: Impose boundary, initial, and loading conditions. Step 5: Solve the differential equation for the window system. Step 6: Postprocess the solution and quantities of interest.

3.2  G  overning Equations and Models in Numerical Analysis for Windows Heat transfer through windows occurs due to the temperature difference between the indoors and outdoors. It has three basic forms: conduction, convection, and radiation. (1) Conduction heat transfer through a medium is the result of a diffusion process. According to the Fourier heat conduction law, heat flow is proportional to the temperature gradient. (2) Convection heat transfer is an energy transport affected by fluid motion. Newton’s law of cooling states that heat flow is proportional to the

Research Methods for Assessing the Thermal and Optical Performance of Building…

15

temperature difference between the two media. (3) Thermal radiation is a radiant (i.e., electromagnetic) energy emitted by a medium, and only related to the medium’s temperature. The radiant energy transmitted is proportional to the fourth power of the surface temperature, as described by the Stefan-Boltzmann law (Reddy & Gartling, 2010). The governing equations used in FEM simulations of a window system are the conservation of mass (or continuity), conservation of momentum (or Navier-­Stokes), conservation of energy, and thermal energy equations. These are shown below as Eqs. (5) to (7). The continuity of mass equation is:  ∂ρ + ∇ • ρV = 0 (5) ∂ρ   where ρ is the density of the medium, V is the velocity, and ∇ • V is the divergence of the velocity. The Navier-Strokes eqcon for incompressible flow is:

( )

      ∂V   ∂ρ ρ + V∇ • V  + V∇ •  + ∇ • ρV  = ∇ • σ + ρf (6)  ∂t   ∂t   where 𝜎 is the stress tensor and f is the body force vector on the fluid element. The energy conservation equation for incompressible flow is:



   ∂ρe + ∇ • ρVe = −∇ • q + Q + ∇ • σ • V ∂t

(

)

(7)

where 𝑒 is the internal energy, 𝑞 is the heat flux vector, and 𝑄 is the internal heat generation. Heat transfer through a window can be calculated using the following equation:

Q = UA ( Tin − Tout )

(8)

where 𝑄 is the amount of heat transfer energy from the indoors to the outdoors through a window, 𝑈 is the overall heat transfer coefficient, 𝐴 is the window area, 𝑇𝑖𝑛 is the indoor temperature, and 𝑇𝑜𝑢𝑡 is the outdoor temperature.

3.3  N  umerical Analysis for Indoor Environments by Side Windows We created the FE model of an office in Autodesk CFD, as shown in Fig. 5. In this model, we simulated two different windows with low and high amounts of insulation. These represented the traditional single-pane window system and

16

Y. Zhao et al.

Fig. 5  Layout settings of office FE models (a) FE model of office with Layout A (b) FE model of office with Layout B (c) FE model of office with Layout C

contemporary triple-pane windows, respectively (Duan et al., 2017; Duan & Wang, 2019). We investigated an office space with three different configurations of supply inlet and return outlet. In Layout A, the air supply register was placed on the floor underneath the exterior window, and the air return grille was on the top part of the office, in the center of the wall and, near the door. In Layout B, the air supply register was above the window; the vent was on the ceiling just above the exterior window, and the air return grille was located in the same position as in Layout A. In Layout C, the air supply register was located in the center of the office ceiling and the air return grille was placed in the same position as in Layout A (see Fig. 5). Combined with two different window types with high and low levels of insulation, there were a total of six combinations of air vent placement and window type and 12 simulation tasks for the summer and winter seasons (Duan et al., 2017; Duan & Wang, 2019). Results of the vertical air temperature distribution obtained from Autodesk CFD are shown in Fig. 6. In summer (Fig. 6a), we can see that Layout B with the above-­ the-­window type vent placement, provided a uniform vertical temperature gradient (0.7 °C) when the window insulation was low. The vertical temperature difference (4.1 °C) in Layout A was over the level recommended by ASHRAE 55–2013, so it could have resulted in local thermal discomfort. When it comes to highly insulated windows, the vertical temperature differences in all three cases were reduced, and Layout C presented a similar vertical temperature gradient (0.9 °C) to that of Layout B (0.7  °C). Layout A’s vertical temperature difference was still over the recommended level. In winter (Fig.  6b), when the window had low insulation, the air temperature distribution showed that Layout A, with its under-the-window type of vent placement, provided more uniform thermal conditions (a 1.8 °C vertical temperature difference) than the other two layouts. Layout C had the highest temperature gradient (4.19 °C), and both Layouts B and C generated high vertical temperature

Research Methods for Assessing the Thermal and Optical Performance of Building…

17

Fig. 6  Different cases of vertical temperature distributions (a) Summer scenario (b) Winter scenario

differences that were over the ASHRAE 55 recommended level. Conversely, comparing the models with highly insulated windows indicated that all three layouts provided more even local thermal conditions. Also, the distinction and advantages of Layout A were attenuated. Layout C also performed well in this situation. These comparisons reveal the consistent effects on vertical temperature differences and airflow patterns that result from combining window properties and air vent layouts. With highly insulated windows, although Layouts A and B performed slightly better in winter and summer, respectively, the central-ceiling type of air vent placement would serve as an acceptable alternative solution in terms of vertical temperature variations (Duan et al., 2017; Duan & Wang, 2019). Table 3 summarizes the heat transfer through the exterior window at the eventuate state (average room temperature of approximately 25 °C) in summer and winter.

3.4  Numerical Analysis for Thermal Behaviors of Windows We simulated three building glazing models in COMSOL Multiphysics 5.4, including single-pane without low-e coating, single-pane with exterior low-e coating, and single-pane with interior low-e coating. The setpoint of indoor temperature was 21 °C, with indoor temperature ranging from −15 °C to 15 °C and indoor humidity

18

Y. Zhao et al.

Table 3  Thermal transfer through windows under different layouts and window types Window type Single pane

U-factor (W/m2K) 5.6

Triple pane

0.65

Heat Transfer (W) Summer Layout A Layout B 73.0 70.7 Summer Layout A Layout B 24.0 23.2

Layout C 63.3 Layout C 21.0

Winter Layout A 382.8 Winter Layout A 114.4

Layout B 380.6

Layout C 347.1

Layout B 108.3

Layout C 103.6

Fig. 7  Single-pane window with low-e coating (a) Finite model of a single-pane window (b) Condensation potential of a single-pane window

from 20% to 70%. Comparing the temperatures of the interior surfaces of the windows, heat flux through the windows, and winter center-of-glass U-factors, we can understand the potential impacts of using a low-e coating on the thermal performance and condensation risk of single-pane windows and find the best position for the coating on the window’s surface. Figure 7 shows the simulation results for the single-pane windows with low-e coating (Duan et al., 2020). Similar to the numerical analysis via COMSOL, the LBNL THERM program enables a more user-friendly and simplified procedure to analyze the thermal behaviors of the window systems. For instance, Fig. 8 shows the simulation results of the thermal profiles of the double-pane window system with a low-e coating in THERM. To conclude, numerical analysis programs are very useful to investigate the window’s thermal properties under various external conditions. Some tools, like CFD and COMSOL provide more sophisticated functions and analytical details for window systems consisting of multiple layers and specific spectral features, while some tools, like LBNL THERM, are more ease-of-use for researchers and designers with sufficient graphic analytical results.

Research Methods for Assessing the Thermal and Optical Performance of Building…

19

Fig. 8  Double-pane window simulation in THERM (a) Temperature distribution isotherms (b) Temperature distribution color gradient

4  C  omputational Design and Optimization Techniques for Building Windows With an increasing need to reduce building energy consumption and other performance metrics, researchers and designers in the AEC industry are beginning to adopt optimization techniques. However, many aspects of a building attempt to satisfy multiple objectives, complicating the direct application of optimization, which also drives the building windows to be more kinetic (Wang et al., 2012). For instance, building windows must satisfy optical and thermal properties in addition to meeting other qualitative goals such as occupant comfort and aesthetics. Due to the intrinsic limitations f window materials, most low-SHGC windows may decrease the building cooling loads but have the potential to increase both heating and lighting loads due to the coatings, tints, and films applied to achieve the low SHGC. Similarly, a low U-factor may reduce heating and cooling loads but deteriorate daylighting performance due to the addition of more layers or coatings, which might increase lighting loads and internal heat gains from electrical lights. As shown in Fig.  9, a statistical analysis based on a large database that covers nearly 8000 different

20

Y. Zhao et al.

Fig. 9  Visible Transmittance (VT) vs SHGC for typical window products

manufacturer product lines and more than 550,000 NFRC-certified fenestration products demonstrates that the clear correlation between SHGC and VT. In other words, optimization studies in terms of the optimal indoor comfort and minimum energy use in window studies are basically multi-objective and multi-variable problems. Such problems are often too complex to solve manually, and the architectural design process requires quick, comprehensive studies. Leveraging the computational tools to perform optimizations allows for the advancement of building design and associated technology (Attia et al., 2013). Because simulation engines can now be accessed in parametric coding environments, such as EnergyPlus’ EMS module that has been used in our previous dynamic envelope studies for parametric energy simulations (Wang & Beltran, 2016a, b), optimization techniques can easily be integrated into the existing workflow. Optimizing windows for energy, daylighting, view, or other custom functions can now be performed easily and used to suggest improvements for more advanced simulations such as those for dynamic facades. For instance, in studies (Wang et al., 2016a, b), three key factors of building windows: U-factor, VT, and SHGC were involved in a genetic algorithm optimization process to determine the best set of window properties for selected weather conditions. The database of NFRC certified products was used to generate a multivariate regression model which was subsequently used as input into GENE_ARCH for optimization (Caldas, 2008). The resulting combinations of VT, U-factor, and SHGC were compared with reference models that meet ASHRAE 90.1–2013.

Research Methods for Assessing the Thermal and Optical Performance of Building…

21

Fig. 10  Example optimization workflow for electrochromic glazing

Research findings based on a small office model indicate apparent energy savings by the selection of optimal window properties. An overall flowchart for implementing optimization in window design is provided in Fig. 10. The “design space,” or set of all possible solutions in an optimization framework, is determined by identifying the variables and respective bounds. For example, variables for modeling the tint levels of electrochromic glazing may include the SHGC and VT, while other parameters such as the number of panes and U-value may remain constant. In this example, the variables are dictated by discrete values provided in manufacturer data, so they cannot be freely selected. Conversely, parameterizing fenestration geometry may involve varying the window width, height, and window-to-wall ratio, which are continuous variables. Many building design problems with a mixture of variable types are modeled in popular visual coding environments such as Grasshopper or Dynamo, programs capable of generating complex building geometries. In the context of building window optimization, window geometry and glazing characteristics can be explored simultaneously. While performing optimization in a research setting may grant more time and flexibility, it is necessary for designers to carefully consider which variables are most important, which regions of the design space should be examined, and if there are types or groups of designs that can help break up the design space (Brown & Mueller, 2019). The next step is to set the axes of the “objective space,” which involves selecting formal objective functions for minimization. Once a quantitative goal (or goals) is determined, an evaluation is modeled to calculate the objective(s) (see Step 2  in Fig. 10). For this simulation-based problem, objectives may include minimizing the building energy use index while maximizing spatial daylight autonomy. Simulation software such as Energy Plus and Radiance can be accessed directly through plug-­ ins in parametric design software. Although the relationship between daylight and energy is rather intuitive, many multi-objective problems yield unexpected harmonies or tradeoffs. Optimization generally seeks to find the best solution, but the resulting data can also be used to identify universal relationships among key variables. The third and final step in setting up an optimization problem is selecting and testing an algorithm based on the nature of the problem. The researcher or designer should consider the variable type (e.g., discrete, continuous, categorical, ordinal, etc.), order of the design space, and constraints. There are two main types of the algorithm implemented in building design problems: gradient-based and heuristic. Gradient-based methods seek the nearest best value based on the initial value. While

22

Y. Zhao et al.

gradient-based methods are often quicker, the risk of converging to a local minimum is higher. Conversely, heuristic methods do not leverage derivative information and are more time-consuming, but less likely to converge to a local minimum. Heuristic methods include evolutionary processes such as genetic algorithms and simulated annealing, which are popular in building design optimization. As each problem is uniquely defined, it can be difficult to identify the best algorithm; the answer often depends on whether speed or the quality of the solution is most important. In many cases, objectives can be reduced, refined, or simplified. For single-objective optimization problems, the algorithm converges to reveal the best solution. However, multi-objective optimization yields a set of non-­ dominated solutions. A common data analysis technique is to generate a Pareto front and select the “best” design based on priorities or preferences. Researcher or designer engagement is required to determine the “optimal” solution, demonstrating that computational design and optimization are intended to aid in the design process. As mentioned above, multi-objective problems are innate to building design. Applying optimization techniques to complex systems such as building windows assists with making fast, informed decisions, ultimately facilitating the advancement of emerging technologies.

5  H  uman-Factor Experiments Evaluating Subjective Thermal and Visual Perception Windows are experimental variables in human factor experiments evaluating subjective thermal and visual perception by controlling the quantity and spectral power distribution of daylight entering an experimental space. Daylight is a combination of light reflected from the ground and terrestrial objects and light transmitted through the air via diffuse sky radiation and direct sunlight. It contains the full light spectrum, including the ultraviolet, visible, and infrared parts. Daylight is often seen as a reference light source in human experiments exploring the visual effects of light sources, providing a standard rhythm for non-visual effects. Thus, daylight is considered a realistic light source and preferred in lighting design. However, more daylight is not necessarily better, due to the complexity of the spectrum’s construction and the variety of possible human responses. Windows, an indispensable part of building façades and ceilings because they provide a natural view and import daylight, can regulate the quantity and quality of daylight penetration into an experimental space. This is accomplished by their being set at different directions and heights (of façades or ceilings) and with various dimensions, structures, shading controls, glazing methods, tint colors, film types, and the number of layers between window panes (Chinazzo et al., 2018; Wang et al., 2020; Pineault & Dubois, 2008; Dubois et al., 2007; Galasiu & Veitch, 2006; Yun et al., 2014; Arsenault et al., 2012; Clear et  al., 2006). The above properties determine the quantity of the daylight imported in terms of light intensity and heat transfer amount (i.e., the VT and SHGC

Research Methods for Assessing the Thermal and Optical Performance of Building…

23

values). They can also determine the parts of the daylight spectrum to be transmitted, reflected, or absorbed by the windows, causing different spectral power distributions entering the experimental spaces and influencing human perception. Spectral regulation of the visible part of the full spectrum can cause different visual and non-visual responses. The daylight absorbed and transformed through windows, especially the infrared part, can result in different human perceptions of the thermal environment. The main components of human experiments are the design, setup, window property selection controlling transmitted radiation, objective parameter measurement, subjective evaluation, and data analysis. Figure 11 presents the general workflow for how human-factor experiments are performed to evaluate subjective thermal and visual perception across different windows. Before designing human-factor experiments, one must consider the research object. Daylight can have multiple effects on humans. Light perception, thermal perception through different pathways and under daylight, and the interaction of light and thermal perceptions may all have an effect (Feng et al., 2020b). Thus, it is important to determine the experimental object and control variables, and from those the environmental variables that relate to thermal, visual, acoustic, and air qualities (Schweiker et al., 2020). The focused wavebands of the daylight spectrum vary for different research objectives. The main effect on human visual perception from daylight is caused by the visible part of the overall spectrum. Correlated color temperature and photopic illuminance values are often used to describe properties of the visible spectrum (Chinazzo et al., 2018). Thermal perception can be influenced by the visible and infrared parts of the spectrum, which have heat effects. Daylight changes throughout the day. Thus, it is especially important to measure the quantity and quality of daylight entering the experimental space as one of the variables. A spectroradiometer is often used to measure the visible spectrum (Figueiro et  al., 2013), which directly gives the correlated color temperature and illuminance of the visible spectrum. Non-visual parameters of the visible spectrum can either be calculated according to the spectrum’s power distribution or by a Daysimeter (Figueiro et al., 2013). A spectrophotometer is often used to measure the ultraviolet and infrared portions of the spectrum.

Fig. 11  Flowchart of human-factor experiments for subjective thermal and visual perception

24

Y. Zhao et al.

A randomized control trial (RCT) experimental design is often used in human-­ factor experiments (Bhide et al., 2018). RCT is a reliable method of determining whether a cause-and-effect relationship exists between experimental and control groups. These are randomly allocated to subjects, and the differences between them are observed (such differences are likely to be due only to the intervention in the experimental group). A crossover trial experimental design is also employed in human-factor experiments. Each subject is randomly administered a sequence of treatments sequentially throughout the whole experiment and different effects observed (Evans, 2010). When the experiment is interested in multiple factors and interactions among factors, a factorial design is preferred. If there are two factors explored in an experiment and each factor has three levels, a 3x3 factorial design would be used to explore the effects of the factors and relationships among them. Psychological and physiological measurements are typically performed in human-factor experiments to evaluate human subjective visual and thermal perceptions. Questionnaires are normally employed in psychological measurements. Questions evaluating visual perception include visual comfort, sensation, relaxation, level of pleasantness, and so on. For thermal perception, there is thermal comfort and sensation, preferred temperature, dressing behavior, and others (Yang & Moon, 2018; te Kulve et al., 2016). Questions are designed to meet experimental requirements. Thermal physiological factors can be used to support or to some extent explain subjective thermal perception; thus, physiological factors such as core body temperature, skin temperature, heart rate, blood pressure, and skin conductance are often used in human-factor experiments. Commonly, core body temperature is measured by telemetric temperature pill, skin temperature by an iButton data logger, and heart rate, blood pressure, and skin conductance by an Empatica E4 wristband (Chinazzo et al., 2018; Feng et al., 2020b). Data analysis is crucial in human-factor experiments to test the significance of the effects of different elements and their interaction. Typical data analysis methods used in the human-factor experiment include analysis of variance, one- or two-tailed t-tests, post-hoc testing, and mixed linear modeling (Chinazzo et al., 2018; Yang & Moon, 2019). There are also many other statistical methods that can be used. One or multiple data analysis methods can be selected to deal with data obtained from human-factor experiments, based on the data type and research target.

6  T  rends and Other Issues of Research Methods for Building Windows The current methods in developing windows can be simply divided into two directions: structural and spectral designs. In current structural design, multi-plane glazing is the most popular approach which focuses primarily on conductive heat transfer. Such a structure normally consists of two or more glass panes separated by insulating layers of air, inert gas, vacuum cavities, phase changed materials, silica

Research Methods for Assessing the Thermal and Optical Performance of Building…

25

aerogel, or some dynamic controls, such as airflow or water flow (Wang & Beltran, 2016a; Kim et  al., 2006; Eames, 2008; Chow et  al., 2010; Baetens et  al., 2011; Chow & Lyu, 2017; Li et al., 2018; Goia et al., 2015). The materials that can serve as transparent thermal barriers are therefore sought with low thermal conductivity and good visible light transmission. Spectral design mainly deals with radiative heat transfer for energy-efficient windows, normally through thin films, tints, or coatings that respond to surrounding thermal radiations. For temperatures of practical interest in the building energy studies, thermal radiation flux occurs in four basic bands: ultraviolet radiation (100 nm