Smart Morphing and Sensing for Aeronautical Configurations: Prototypes, Experimental and Numerical Findings from the H2020 N° 723402 SMS EU Project 3031225791, 9783031225796

Citation preview

Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153

Marianna Braza · Jean-François Rouchon · George Tzabiras · Franco Auteri · Pawel Flaszynski   Editors

Smart Morphing and Sensing for Aeronautical Configurations Prototypes, Experimental and Numerical Findings from the H2020 N° 723402 SMS EU Project

Notes on Numerical Fluid Mechanics and Multidisciplinary Design Founding Editor Ernst Heinrich Hirschel

Volume 153

Series Editor Wolfgang Schröder, Aerodynamisches Institut, RWTH Aachen University, Aachen, Germany Editorial Board Bendiks Jan Boersma, Delft University of Technology, Delft, The Netherlands Kozo Fujii, Institute of Space and Astronautical Science (ISAS), Sagamihara, Kanagawa, Japan Werner Haase, Neubiberg, Bayern, Germany Michael A. Leschziner, Department of Aeronautics, Imperial College, London, UK Jacques Periaux, Paris, France Sergio Pirozzoli, Department of Mechanical and Aerospace Engineering, University of Rome ‘La Sapienza’, Roma, Italy Arthur Rizzi, Department of Aeronautics, KTH Royal Institute of Technology, Stockholm, Sweden Bernard Roux, Ecole Supérieure d’Ingénieurs de Marseille, Marseille CX 20, France Yurii I. Shokin, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia Managing Editor Esther Mäteling, RWTH Aachen University, Aachen, Germany

Notes on Numerical Fluid Mechanics and Multidisciplinary Design publishes state-of-art methods (including high performance methods) for numerical fluid mechanics, numerical simulation and multidisciplinary design optimization. The series includes proceedings of specialized conferences and workshops, as well as relevant project reports and monographs. Indexed by SCOPUS, zbMATH, SCImago. All books published in the series are submitted for consideration in Web of Science.

Marianna Braza · Jean-François Rouchon · George Tzabiras · Franco Auteri · Pawel Flaszynski Editors

Smart Morphing and Sensing for Aeronautical Configurations Prototypes, Experimental and Numerical Findings from the H2020 N° 723402 SMS EU Project

Editors Marianna Braza CNRS Institut de Mécanique des Fluides de Toulouse Toulouse, France

Jean-François Rouchon Laboratoire Plasma et Conversion d’Energie Institut National Polytechnique de Toulouse Toulouse, France

George Tzabiras School of Naval Architecture and Marine Engineering National Technical University of Athens Zografou, Greece

Franco Auteri Dipartimento di Scienze e Tecnologie Aerospaziali Politecnico di Milano Milan, Italy

Pawel Flaszynski Institute of Fluid-Flow Machinery Polish Academy of Sciences, IMP PAN Gda´nsk, Poland

ISSN 1612-2909 ISSN 1860-0824 (electronic) Notes on Numerical Fluid Mechanics and Multidisciplinary Design ISBN 978-3-031-22579-6 ISBN 978-3-031-22580-2 (eBook) https://doi.org/10.1007/978-3-031-22580-2 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 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

Preface

The present book contains a detailed description and analysis of the results including new findings obtained from the H2020 N° 723402 European research project SMS, Smart Morphing and Sensing for Aeronautical Configurations, https://cordis.europa. eu/project/id/723402 and http://www.smartwing.org/SMS/EU.

In the recent decades, a considerable effort has been devoted to improve the aerodynamic performance and to reduce noise by means of different methods. Most of them involve vortex generators and riblets enabling drag reduction, as well as hydromechanical actuators and microelectromechanical systems, among other. The majority of these devices are heavy and characterized by a rather slow response. Few attempts had been made to employ electrical actuators, able to deform specific parts of the wings. Furthermore, there do not exist to our knowledge approaches permitting a simultaneous reduction of the noise sources, together with a considerable aerodynamic performance increase. Besides, the majority of existing concepts were focusing on actuation of upstream parts of the wings, to obtain laminarisation. They do not take benefits that would be obtained by feedback effects through modification of the downstream wing’s part and its surrounding turbulent vortex structures. v

vi

Preface

This book presents innovative and highly efficient Morphing concepts for an optimal and smooth modification of the wing’s shape and its vibratory character, operating at different time and length scales, according to the turbulence nature surrounding the lifting body. The topics of the book present therefore the ways of “Smart wing design through turbulence control”, enabled by “hybrid electoractivemorphing.” This operates simultaneously high deformations in low frequencies and slight deformations in higher frequencies. It creates an interaction with the turbulence vortex structures that in turn modify the structural properties, thus composing an efficient fluid-structure interaction system. Its high efficiency in lift increase, drag reduction and noise sources reduction has been demonstrated by the SMS project in laboratory scale, as well as near “scale one”. Advanced wing prototypes have been built on this purpose and presented in detail in this book. They embedded different classes of electrical actuators under the “skin” of the lifting surface controlled by an appropriate multi-point pressure system that measures the unsteady pressure on strategic areas of the wing surface. These areas, together with optimal wing shapes have been identified by adjoint-based sensitivity matrix evaluation, among other optimisation approaches in the project. The present electrically based morphing leads to much lighter and efficient wing design than other approaches in the state of the art. It is in-line with the priorities fixed by the aeronautics industry toward “a More Electric Aircraft”, MEA. This disruptive wing design is partly bio-inspired, regarding the different scales of large—span hunting bird wings that operate high cambering of the main wing’s part and simultaneously actuate small deformations and higher frequency vibrations of their ailerons and feathers. These actuations are guided from the pressure sensing of the bird that captures the aerodynamic pressure distribution. This enables the bird to optimally actuate all this arsenal of multiple-scale structures. It will be remembered the ability of the owl to simultaneously increase its aerodynamic performance and practically suppress noise when flying toward its prey. However, the electroactive morphing concepts studied in the SMS project are only partially bio-inspired because they have been adapted in realistic aircraft speeds that never these birds reach. The efficiency in aerodynamic performance increase is demonstrated in all flight phases, take-off, landing and cruise, by means of refined wind tunnel experiments, Hi-Fi numerical simulations and modeling. In many cases of the studies presented in this book, the simulations dictated the optimal parametric ranges followed by the experiments. An appropriate controller’s design studied by ONERA—Toulouse SMS partner under the responsibility of Dr. Carsten Döll— enabled the application of the optimal actuations on the prototypes to reach these performances. Thanks to the obtained performances, the SMS project prepares future wing design for aeronautics industrial applications aiming at saving energy and at reducing the pollution through these new multiscale morphing concepts. These open new ways in the design enable a considerable reduction of emissions, meeting the targets fixed by

Preface

vii

the European Commission, DG MOVE/DG RTD, Flightpath 2050: Europe’s Vision for Aviation: Maintaining global leadership and serving society’s needs. Toulouse, France Toulouse, France Zografou, Greece Milan, Italy Gda´nsk, Poland

Marianna Braza Jean-François Rouchon George Tzabiras Franco Auteri Pawel Flaszynski

Acknowledgements

The editors and all the SMS partners are grateful to the European Commission for the funding of this project under the “MOBILITY FOR GROWTH” call identifier: H2020-MG-2016-2017/MG-1.1-2016: Societal Challenges: “Smart, green and integrated transport”, topic: “Reducing energy consumption and environmental impact of aviation”, Grant number GA 723402. They also express their gratitude to AIRBUS “Emerging Technologies and Concepts Toulouse”—ETCT, endorsers of the project for the steering of its activities all over the project’s duration, under the guidance of the successive ETCT directors, Alain Fontaine and Denis Descheemaeker. The coordinator of SMS, Dr. Marianna Braza and all the partners acknowledge with warm thanks the contribution of INPT—“Institut National Polytechnique de Toulouse” administrative and accounting services under its two successive Presidents, Professor Olivier Simonin until December 2018 and Dr. Catherine Xuereb until the end of the project in May 2020, who ensured a high quality of the execution of the project under the INPT staff. Specific thanks are addressed to the INPT—SAIC, “Service des Activités Industrielles et Commerciales” under the Direction of Marion Coureau and the contribution of Isabelle Yu Wai Man for the financial accompany of the project, as well as Delphine Dubs who prepared the GA phase of the project in 2017. The coordinator expresses her thanks to the administrative services of the “Institut de Mécanique des Fluides de Toulouse”—IMFT, under the Direction of Professor Eric Climent, for having made possible the scientific coordination of the SMS project in very good conditions, as well as to Denis Bourrel, “Secrétaire Général, Responsable Financier”, Florence Colombiès and Nadine Mandement for their administrative contribution. Deep thanks are also expressed to the Workshop services of IMFT under the responsibility of Rudy Soeparno, for having built the “Large Scale”, (LS) prototype of the SMS project, as well as to the Service of Image and Signal Processing of IMFT under the responsibility of Sébastien Cazin, for their essential contribution to the Time Resolved and Tomo—PIV experiments. Warm thanks are addressed to the PRACE “Partnership for Advanced Computing in Europe” for the attribution of a considerable CPU allocation under the grant N° 2017174208—FWING ix

x

Acknowledgements

“Future Smart Wing design”, thus contributing to a significant part of the Hi-Fi simulations of the SMS project. Their dedicated two “Success Stories” articles to our project, https://prace-ri.eu/future-aircraft-wings-will-be-able-to-adapt-theirshape-mid-flight/ and https://prace-ri.eu/news-media/publications/prace-fact-sheets/ success-stories-in-engineering/ are highly acknowledged. A deep acknowledgement is also addressed to the French Supercomputing Centres CINES, TGCC and CALMIP for the substantial CPU allocation that made possible part of the Hi-Fi numerical simulations of the coordinator’s Institute in the SMS project. Moreover, warm thanks are expressed to CALMIP under the Direction of Jean-Luc Estivalezes, for having launched the Data Management Plan and the data access of the SMS project respecting the FAIR principles fixed by the European Commission, by means of the specific dataverse platform “CALLISTO”—“CALmip Launches an Interface for Semantic Toolbox Online”, developed by Thierry Louge. Thanks to this platform, data access, exchange, interoperability and reuse has been made possible for the SMS partners thanks to development of a specific ontology and workflows, described in Chap. 5 of this book. The SMS coordinator, Marianna Braza expresses a most sincere gratitude to Dr. Corinne Joffre, responsible of the “Cellule Europe” of the University of Toulouse, Dr. Johannes Scheller, Post-Doctorate at the IMFT—“Institut de Mécanique des Fluides de Toulouse” and LAPLACE—“Laboratoire Plasma et Conversion d’Energie” Laboratories of INPT—CNRS—University of Toulouse, as well as to Dr. Delphine Dubs at INPT in the period 2016–2017, who intensely worked with the coordinator for the successful submission of the SMS project and the following phase of the GA preparation. The coordinator of SMS, Marianna Braza and all the partners express our warmest thanks to the Project Officer, Miguel Marti Vidal, who has provided considerable guidance along the route to make the SMS project a success. Warm thanks are also expressed to our initial Project Officer, Daniele Violato who prepared with us the Grant Agreement. Many thanks to both of them for having invited the SMS project at important events organized by INEA—Innovations and Network Executive Agency of the European Commission, to present our developments on specific conferences on disruptive aircraft configurations and to dispose of stand areas displaying the SMS prototypes and videos. Among these invitations are highly acknowledged the SMS participation in the airshow ILA, “Innovation Leadership Aerospace” in Berlin, in May 2018, in the AERODAYS organized by the European Commission in Bucharest, May 2019 and in the “Science is Wonderful” Exhibition in Brussels, in September 2019, among other. A detailed view of the SMS participation in these events as well as the overall dissemination activities are listed at the end of Chap. 6 of the present book and can be also found in http://www.smartwing.org/SMS/EU. The publication of the book is a result of the collective effort by the contributors and authors of the chapters. The editors are grateful to Prof. Wolfgang Schröder, general editor of the Springer Series Notes on Numerical Fluid Mechanics and Multidisciplinary Design, for having made possible the publication of this book in the present series.

Acknowledgements

xi

Last but not least, warm thanks are expressed to Dr. Marouf Abderahmane at IMFT and at the ICUBE Laboratory of the University of Strasbourg, for having created and maintained the web site of the project and for the Web site of its international IUTAM Symposium, http://www.smartwing.org/iutam in 2018, as well as to his valuable contribution to the editing of this book. Many thanks are expressed to Dr. Johannes Scheller for having hosted the SMS website in the platform http://www.sma rtwing.org. Warmest thanks are addressed to Dr. Rajaa El Akoury, Post-Doctorate at IMFT for her essential contribution for the final editing of this book.

Picture from the 30th month SMS meeting at Politecnico di Milano, Aerodynamics Laboratory, in front of the Large Scale prototype of the project. Last project’s meeting in presence before COVID restrictions

Toulouse, France Toulouse, France Zografou, Greece Milan, Italy Gda´nsk, Poland July 2022

Marianna Braza Jean-François Rouchon George Tzabiras Franco Auteri Pawel Flaszynski

Contents

1 The SMS Project: Introduction and Overview . . . . . . . . . . . . . . . . . . . . . Marianna Braza 1.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 The SMS Structure and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Affiliation List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Summary of the Project Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Reduced Scale Prototype Morphing Achievements in Subsonic and Transonic Regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Jodin, M. Carvalho, C. Raibaudo, C. Döll, P. Mouyon, P. Doerffer, P. Flaszynski, P. Kaczynski, M. Piotrowicz, K. Doerffer, M. Marchal, S. Cazin, D. Harribey, J. Scheller, C. Nadal, G. Harran, M. Braza, and J. F. Rouchon 2.1 Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Higher Frequency-Lower Amplitude Trailing Edge Vibrations . . . . 2.3 RS Prototype B: Transonic Reduced Scale (tRS) Design . . . . . . . . . 2.3.1 Final Geometry of the Test Section . . . . . . . . . . . . . . . . . . . . . 2.3.2 Experimental Verification and Proof of the Morphing/Sensing Efficiency-sRS . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Large Scale Morphing Prototype: Design and Experiments . . . . . . . . . A. Giraud, B. Nogarede, Y. Bmegaptche-Tekap, M. Carvalho, C. Korbuly, A. Kitouni, J. B. Paris, V. Lamour, A. Marouf, J. B. Tô, A. Polo-Dominguez, M. Scheller, J. Scheller, D. Harribey, M. Braza, and J. F. Rouchon 3.1 Optimal Design of the Electromechanical Actuators (EMA) System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Electromechanical Actuator System . . . . . . . . . . . . . . . . . . . . 3.1.2 Electromechanical Actuator Design . . . . . . . . . . . . . . . . . . . . . 3.2 Sensing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 2 5 10 11 11 13

14 15 16 17 23 44 47

48 48 49 54 xiii

xiv

Contents

3.2.1 3.2.2 3.2.3 3.2.4 3.2.5 3.2.6 3.2.7

Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fiber Bragg Grating (FBG) Sensor Technology . . . . . . . . . . . Experimental Set-Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reynolds Number Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . Angle of Attack Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison of the Results with the Numerical Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.8 Multi-point Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.9 Conclusion—Sensing System . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Controller Hardware Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Large-Scale—Controller Interface (iF) . . . . . . . . . . . . . . . . . . 3.3.2 Conclusion—Hardware Controller and Interface . . . . . . . . . . 3.4 Experimental Verification—INPT/IMFT-LAPLACE . . . . . . . . . . . . . 3.4.1 Mean Pressure Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Unsteady Pressure Measurements . . . . . . . . . . . . . . . . . . . . . . 3.4.3 Conclusion—Experimental Verification . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

54 56 57 59 61 61

4 High-Fidelity Numerical Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Marouf, N. Simiriotis, J. B. Tô, Y. Hoarau, J. B. Vos, D. Charbonnier, A. Gehri, R. El Akoury, Y. Hoarau, F. Kramer, F. Thiele, K. Diakakis, M. Fragiadakis, J. L. Farges, T. Chaboud, G. Tzabiras, J. F. Rouchon, and M. Braza 4.1 Morphing sRS Prototype—Numerical Simulations . . . . . . . . . . . . . . 4.2 Three-Dimensional Morphing Effects . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Performances of the Hybrid Morphing: Cambering + Actuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Structural Modelling of the sRS with Embedded SMA (Shape Memory Alloys) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Structural Control of the Wing Equipped with SMA Actuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 MDO (Multi-disciplinary Design Optimisation) Results Incorporating Experimental Results and High-Fidelity CFD Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.2 Scientific and Technological Background . . . . . . . . . . . . . . . . 4.5.3 Investigation of the Reduced Scale (RS) Prototype—RSP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.4 Investigation of the Large-Scale (LS) Prototype—LSP . . . . . 4.5.5 Conclusions—MDO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Hi-Fi Simulations on the tRS Prototype (INPT) . . . . . . . . . . . . . . . . . 4.7 Hi-Fi Simulations on the Large Scale (LS) Prototype . . . . . . . . . . . . 4.7.1 Take-off Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.2 Landing Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

89

61 62 62 65 66 80 81 82 83 88 88

90 92 94 96 97

99 99 100 102 113 120 123 129 129 134

Contents

4.7.3 Aerodynamic Performance by Hi-Fi Simulations Around the Full A320 Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.4 Hybrid Morphing—Full Aircraft, Landing—CFSE . . . . . . . . 4.8 Aircraft Trajectories, Polars and Fuel Consumption . . . . . . . . . . . . . . 4.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.2 Defining Relevant Conditions for Fluid Dynamics Computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.3 Identifying Polar Models from Results of Fluid Dynamics Computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.4 Extrapolating to Aircraft Performance . . . . . . . . . . . . . . . . . . . 4.8.5 Computing Benefit in Fuel Consumption . . . . . . . . . . . . . . . . 4.8.6 Conclusion on Fuel Consumption Evaluation Through Aircraft Trajectories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Aerodynamic Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Auteri, P. Flaszynski, A. Savino, A. Zanotti, G. Gibertini, D. Zagaglia, Y. Bmegaptche-Tekap, D. Harribey, J. F. Rouchon, P. Kaczynski, P. Doerffer, M. Piotrowicz, R. Szwaba, J. Telega, T. Louge, J. B. Tô, C. Jimenez-Navarro, A. Marouf, and M. Braza 5.1 TRS Prototype Aerodynamic Evaluation . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Transonic Reduced Scale Profile—Reference Case . . . . . . . . 5.1.2 Transonic Reduced Scale Profile with Flapping Trailing Edge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.3 Conclusions for the Transonic Prototype tRS . . . . . . . . . . . . . 5.2 Large Scale Cambered Prototype Design and Experimental Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Overview and Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Design and Construction of the Wing Model . . . . . . . . . . . . . 5.3 GVPM Wind Tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 CAD Drawings for the High-Lift Flap . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.1 Morphing Wing Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Interface Definition and Camber Control of the High-Lift Flap . . . . 5.5.1 Modelling of the System and Feedback Control Design . . . . 5.5.2 Camber Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Model Specifications and Preliminary Investigation in a Scaled Wind Tunnel (1:9 Scale) . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.1 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.2 Test-Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.3 Test Rig, Instrumentation and Sensor Layout . . . . . . . . . . . . . 5.6.4 PIV Set-Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.5 Shape-Measurement Set-Up . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xv

142 146 147 148 149 150 151 151 152 152 153 155

156 156 167 173 174 174 174 175 175 177 177 189 189 192 208 210 217 220 223 225 227

xvi

Contents

5.6.7 PIV Campaign Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.8 Shape-Measurement Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Data Sharing of SMS and Workflows Through Ontology-Based Data Access in the Platform CALMIP/CALLISTO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.1 The Sharing of Data for SMS . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.2 Ontology-Based Data Description and Workflow Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.3 Practical Use of CALLISTO for SMS and Data Reusability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.4 Conclusions on the Data Sharing Access . . . . . . . . . . . . . . . . 5.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

245 253

256 258 259 264 266 267 268

6 General Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271 Marianna Braza 6.1 Obtained Benefits and Conclusion Summary . . . . . . . . . . . . . . . . . . . 272 6.2 Impact: Communication and Dissemination of the Results . . . . . . . . 273

Chapter 1

The SMS Project: Introduction and Overview Marianna Braza

Abstract Project’s General Context and Description: The Present Book Presents the Results and New Findings Obtained in the European Research Project H2020 N° 723402 SMS, “SMArt Morphing & Sensing for Aeronautical Configurations”, www.smartwing.org/SMS/EU and https://cordis.europa.eu/project/id/723402.

Work Package 1—WP1

This project has been a multi-disciplinary upstream project under the “MOBILITY FOR GROWTH” call identifier: H2020-MG-2016–2017/MG-1.1-2016: Societal Challenges “Smart, green and integrated transport”, topic: “Reducing energy consumption and environmental impact of aviation”, started on 1st May 2017 and ended on 30th April 2020. The project has been coordinated by the Institut National Polytechnique de Toulouse (INPT), involving its two laboratories: The Institut de Mécanique des Fluides de Toulouse—IMFT, “Unité Mixte de Recherche” UMR N° 5502 CNRSINPT-UT3 and the Laboratoire Plasma et Conversion d’Energie”—LAPLACE, UMR 5213. The project has been carried out under the coordination of Dr. Marianna Braza (IMFT–CNRS).

M. Braza (B) INPT—Toulouse Institut National Polytechnique, CNRS Centre National de Recherche Scientifique/IMFT—Institut de Mécanique des Fluides de Toulouse, UMR 5502, 2 Allée du Professeur Camille Soula, 31400 Toulouse, France e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. Braza et al. (eds.), Smart Morphing and Sensing for Aeronautical Configurations, Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153, https://doi.org/10.1007/978-3-031-22580-2_1

1

2

M. Braza

Table 1.1 The partnership of the SMS project No.

Participant name

1 (Coord) Institut National Polytechnique Toulouse

Acronym

Country

INPT

FR

2

Office National d’Etudes et de Recherches Aerospatiales

ONERA

FR

3

Politecnico di Milano

POLIMI

IT

4

National Technical University of Athens

NTUA

GR

5

Instytut Maszyn Przeplywowych im Roberta Szewalskiego IMP-PAN Polskiej Akademii Nauk

6

Cementys

PL

CEMENTYS FR

7

Novatem

NOVATEM

FR

8

CFD Software Entwicklungs- und Forschungsgesellschaft mBH

CFDB

DE

9

CFS Engineering SA

CFSE

CH

10

Scheller Technology GmbH

STN-SST

DE

The SMS project involved nine more European partners as presented in Table 1.1, including five SME (Small-Medium Enterprises): CEMENTYS, NOVATEM, CFSE, CFDB, STN and 4 Research Organisations/Universities: ONERA, IMPPAN/POLIMI, NTUA.

1.1 Objectives The Smart Morphing & Sensing (SMS) project is a multi-disciplinary upstream project that aims at providing novel wing design enabled by strong coupling between efficient sensing and electroactive morphing concepts, in order to simultaneously reduce drag and noise and to increase lift, by achieving optimal shapes and vibrational behaviour. These novel wing designs employ intelligent electro-active actuators to modify the lifting structure of an aircraft to obtain the optimum shape with respect to the aerodynamic performance (lift-to-drag ratio) and with respect to the noise levels and vibrational behaviour. These intelligent electro-active actuators permit to smartly manipulate turbulent eddy structures that develop around the lifting surface and in the wake, as well as the shock—vortices interaction, resulting in a breakdown of harmful vortex structures and an enhancement of beneficial ones. In this way, harmful instabilities can be attenuated or even suppressed, and a control of the shock position and of the near-wake’s width is obtained, both directly linked to drag reduction and lift increase. These objectives have been achieved by the new morphing devices operating in dynamic regimes and enabled by novel sensing devices. This smart morphingsensing system is actuated by an efficient controller design achieved by a suitable interface of control commands, in order to ensure real-time morphing.

1 The SMS Project: Introduction and Overview

3

For its sensing system, the SMS project investigates, besides existing highaccuracy unsteady pressure transducers, a new generation of fiber-optics based sensors allowing simultaneous distributed pressure measurements without needs of modifying the lifting surface and thus ensuring in-situ real-time optimisation of the aerodynamic characteristics. This new system allows for attenuation of flow separation and nuisance instabilities and also reduce trailing-edge noise and other vibration sources in flight that might come from interactions between wing and fuselage and engine or from critical meteorological phenomena as gusts, that have major impact on safety. The SMS project associates the following specific objectives and key strategies that are coupled in a multi-disciplinary environment: 1. Advanced integrated design using High-Fidelity CFDSM (Computational Fluid Dynamics-Structural Mechanics) by introducing the properties of the new generation of intelligent materials (i.e. Shape Memory Alloys (SMA)) and distributed small piezoelectric actuators (PA), as well as novel Electromagnetic Actuators (EMA) in a hybrid concept. The numerical simulations analysed the physical phenomena related to the aerodynamic performance modification and dictated optimal actuation parameters (vibrations, amplitudes) to be used in the physical experiments of the project. 2. Advanced distributed sensing using a new generation of high-fidelity fiber optics sensors based on fiber Bragg grating [1]. This allowed a simultaneous multi-point fluctuating pressure measurement on the wing’s structure, without needing modification of the structural surface. This sensing provides real-time high-accuracy information of the current flight situation enabling a direct link to the electrical actuators. 3. Advanced experimental techniques to generate data together with the Hi-Fi simulations for the iterative feedback of the controller design used for the optimisation of the morphing A320 type wing and of the high-lift flap of a two-element wingflap of a morphing A320 configuration. These experimental techniques have also been used as a basis for the validation of both the novel actuation and sensing systems via Wind Tunnel (WT) tests at subsonic speed (for optimisation of liftto-drag and noise reduction during take-off and landing phases) and transonic speeds (for drag reduction and increase of lift/drag ratio in cruise phases). 4. Controller Design by appropriate Flight Control Commands (FCC), to actuate the electro-active materials combining the actuation capabilities with the distributed pressure sensors to enable a real-time in-situ optimisation of the final prototypes in Reduced Scale (RS) and Large Scale (LS). The SMS project thus establishes a closed loop of theoretical and experimental activities, cross-comparing results and achievements—with the overall objective to improve substantially the aircraft aerodynamic performance and therefore reduce the environmental impact. This overall objective is translated into the following technical objectives: 1. Definition and selection of suitable smart actuatorsand selection of suitable smart actuators.

4

M. Braza

2. For the aerodynamic performance improvement over different time-scales of a large-scale A320 type wing with morphing flap operating in dynamic regimes and taking into account maintainability, safety and robustness. 3. Design of a distributed FBG based sensing system that permits to collect realtime information of the aerodynamic performance improvement and optimising the sensor network for reliability. 4. Provide a High-Fidelity CFDSM kernel coupled with MDO—multidisciplinary optimisation and with the physical experiments of the project in order to define optimal actuations, realized by the final experiments on the morphing prototypes. 5. Design of a novel control system combining the information of the sensing with the actuator performance indicators and enabling a real-time in-situ optimisation of the aerodynamic quantities corresponding to the current flight situation over the different time-scales of actuation. 6. Evaluation of the effective gains on the large-scale A320 wing prototype through experiments and Hi-Fi simulations, as well as on a whole A320 aircraft by Hi-Fi simulations. The SMS project combined the most relevant technologies and approaches to demonstrate how new aircraft designs benefit from electro-active morphing (EM) and distributed sensing based on the fiber Bragg grating (FBG) [1]. This project has offered the application of new methods and technologies that can be used for multifunctional intelligent large-scale deformations (cambering) and smallerscale higher-frequency trailing-edge deformation and vibration concepts actuated in multiple frequency and length scales thus enabling active loads control. Indeed, by the use of Shape Memory Alloys (SMA) or ElectroMechanical Actuators (EMA), embedded in a significant rear part of the wing’s surface along an order of 30% of the chord from the trailing-edge, a low frequency (order of 1Hz) high deformation of the rear wing’s part can be obtained. This is beneficial for the lift increase and manoeuvrability during take-off and landing. By using another class of electroactive morphing through specific mini-piezo-actuators distributed along the span in the trailing-edge region, a high frequency (in the order of 500 Hz) vibration with low deformation (in the order of mm) is obtained. This results in suitable breakdowns of the coherent vortices governing the turbulent flow in this region, leading to a reduction of the wake’s width and to enhancement of specific feedback effects towards the upstream wall pressure distribution. As demonstrated in this book, this action produces a further increase in lift and provides drag reduction especially in cruise as well as a reduction of the noise sources in all flight phases. Moreover, this higher frequency low amplitude morphing has proven efficient to refrain the drag increase (an unavoidable effect due to the cambering), thus providing a considerable lift-to-drag increase. The combination of these different classes of actuators provides the so-called “Hybrid Electroactive morphing”, a copyright term © from the publications by the coordinator’s Institutes IMFT and LAPLACE (see Refs. [2, 3]) and fully deployed in the SMS project. This association precisely acts on the multiple frequencies and length scales of the turbulence spectrum and for this reason it is able to

1 The SMS Project: Introduction and Overview

5

provide the increase of the performances. The hybrid electroactive morphing is partly bio-inspired as the functioning of the large span hunting birds where the high cambering/twist at low frequencies is ensured by their large wings and the smaller amplitude higher frequency vibrations are obtained through their feathers. The term bioinspired is used rather than biomimetic because the birds do not fly at the aircraft’s speed. The SMS project, through the results described in this book, applied the above morphing concepts in real aircraft’s flight context with emphasis to application on a real A320 airplane’s configuration. The reader is invited to vision the film devoted by the Journal of CNRS https://lejournal.cnrs.fr/videos/the-wings-of-the-future to the studies of the multidisciplinary research team of IMFT-LAPLACE Laboratories that initiated these ideas and realised the first prototypes making the basis of the SMS project. By using improved modelling capabilities, advanced prediction tools and physical experiments, SMS has led to enhanced new design processes with reduced design cycles, lower costs and improved time-to-market strategies. Moreover, the SMS project contributed to advance, maintain and foster the competitiveness of the European aeronautics industry. As pointed out by the endorser AIRBUS–ETCT “Emerging Technologies and Concepts—Toulouse”, member of the IAC-Industrial Advisory Committee of the project, the new morphing concepts presented in this book open new ways for future wing design and after proof of their efficiency, it might be envisaged to be used in future real flight testing. Indeed, as proven by the results of the SMS project, the disruptive design of the morphing wings provided are able to give an increased ambition in this sense.

1.2 The SMS Structure and Methodology In the SMS project three wing prototypes have been considered: Two reduced-scale prototypes (A and B, see Fig. 1.1), corresponding to low subsonic and transonic speeds, referenced also as sRS (or simply RS) prototype and tRS prototype, as well as one Large Scale (LS) prototype C, referenced also as LS, composed of two elements: the main wing and its high-lift flap, for low subsonic speeds. This last one corresponds to full-scale of a modified A320 flap considering a constant median section of chord 2.40 m, a dimension taken from an almost middle span section of the swept A320 wing, where the chord of the high-lift flap was of 1m. In the SMS project, a constant spanwise section has been considered. For the LS prototype, the dimensions have been chosen in collaboration with Airbus—“Emerging Technologies and Concepts— Toulouse”—ETCT, endorser of the SMS project. – The sRS prototype has a chord of 70 cm and its static configuration (e.g. without morphing) is the A320 wing. The part of the wing beyond 60% of the chord is morphed according to different time scales: a low time scale in the order of 1 Hz ensured by Shape Memory Alloy actuators and a more rapid time scale in the order of 300–500 Hz ensured by specific piezoactuators.

6

M. Braza

– The tRS prototype has a 15 cm chord and was designed by INPT/LAPLACE in accordance to the wind tunnel dimensions of partner IMP-PAN at Gdansk, for experiments in the transonic regime that corresponds to cruise speed. – The LS prototype is a two-element wing having a chord of 2.40 m and a high-lift flap chord of 1m. The high-lift flap has been morphed using two kinds of actuators, SMA (INPT/LAPLACE) and Electromechanical Actuators (NOVATEM). – Furthermore, the above morphing concepts have proven their efficiency on a full-scale A320 aircraft by means of Hi-Fi simulations. A schematic representation of the SMS prototypes and virtual wind-tunnel is presented in Fig. 1.1. Figure 1.2 presents the structure of the project through its Work Packages (WP) and sub-WorkPackages, as well as the role of the partners. The governing structure of the SMS project is presented in Fig. 1.3. The Industrial Advisory Board is composed of OEM (Original Equipment Manufacturers) and scientists external to the project. Airbus – ETCT (Emerging Technologies and Concepts Toulouse), and UCL (University College London) representative Prof. Julian Hunt have attended the official meetings of the SMS project. Dr. Alain Fontaine, former Director of Airbus ETCT and currently Director of Pégase CoFrance, has attended all the meetings and provided continuous interaction and advice during the project’s three years studies. Prof. J. Hunt (UCL) provided significant advice concerning the Turbulent-Non-Turbulent interfaces (TNT) manipulated by the morphing and producing beneficial “eddy blocking effects”, as explained in the

Fig. 1.1 Schematic representation of the prototypes design

1 The SMS Project: Introduction and Overview

7

Fig. 1.2 Workpackages and tasks of the SMS project

following chapters. Furthermore, the SMS project has received considerable advice by renown scientists worldwide, specialists in the project’s fields who attended the two international symposia chaired by the SMS coordinator, www.smartwing.org/ iutam on the topic: “Critical flow dynamics involving moving/deformable structures with design applications” in 2018, as well as the 5th FSSIC “Flow Structure Sound Interactions and Control”, www.smartwing.org/FSSIC2019. According to the above mentioned elements, the methodology and structure of the SMS project are described in more details as follows: 1. Design of an efficient actuator system based on smart electroactive devices able to simultaneously operate in low frequency ranges/high deformations of the rear part of the wing, corresponding to the high-lift flap and enabling higherfrequency vibrations/small deformation of the trailing-edge region thanks to smart hybridization of the morphing concepts, among the different time-scales. This design has been studied in low-subsonic and transonic speeds, corresponding

8

M. Braza

Fig. 1.3 SMS governance and WP leaders

to the take-off/landing and cruise phase requirements respectively. The target focus in the take-off/landing phases is the increase of lift and more specifically the lift-to-drag ratio increase (aerodynamic efficiency), as well as a reduction in noise. At cruise speed, the more lasting phase, a crucial issue is the drag reduction, in addition to the previous requirements. This has been obtained through optimal near-trailing edge shape deformations and vibrations, ensuring an optimal shockboundary layer interaction (SWBLI), able to suppress transonic buffet at its “early birth”. This eliminates the high-amplitude drag increase and dangerous risks of aeroelastic flutter triggering, both showing up in the Mach number range of Ma = 0.7–0.8 for civil aircraft. 2. Design of an efficient sensing system, based on a non-intrusive fiber-optics technique of Bragg Grating. This system has been able to ensure multi-point distributed pressure fluctuations measurements and to capture the instabilities onset and harmful vibrational modes. Within a unique fiber-optics cable, one fiber is devoted to a simultaneous measurement of the deformation and pressure, where the other one is devoted to temperature measurement. This simultaneous multipoint information is essential in order to test the performance of the morphing concepts, as well as to capture the effect of spanwise loads distribution.In contrast, conventional mechanical systems, as for example transducers, permit only onepoint measurements. The novel fiber-optics based sensing system has been applied on the “skin” of the lifting structure, thus avoiding complex instrumentation of the mock-up with cavities to embed conventional pressure transducers, as

1 The SMS Project: Introduction and Overview

3.

4.

5.

6.

9

needed in standard systems. The sensors have been located at specific positions, optimised through the project according to the indications provided by the MDO methodology, based on Hi-Fi simulations and experiments. The sensing system, including also high sampling rate conventional sensors measuring the unsteady pressure, permitted elaboration of an efficient closed-loop controller for real-time operation, a major achievement in the state of the art. Design of control laws to achieve the optimal morphing concepts. This task has been based on the pressure and forces fluctuations obtained from detailed experiments of the morphing prototypes, enabling the morphing real-time operation. The present design has been therefore based on highly Multi-Input/Multi-Output (MIMO) control laws for the operation of the sensors-actuators system. By this, it has been able to handle model uncertainties and parameter variations (like airspeed) thanks to closed-loop feedback control laws allowing self-adaptation in real time to the actual experimental conditions and morphing solicitations. Experimental verification and proof of the morphing/sensing efficiency. This task aimed at demonstrating the benefits of the concepts studied in the SMS project. It has been first studied usinng the Reduced Scale prototype (RS) in subsonic (sRS) and transonic (tRS) regimes, to provide a very detailed set of measurements. For this, highly advanced optical techniques as the TimeResolved PIV have been used. These velocity measurements, coupled with pressure sensing and forces measurements enabled studying in detail the modification and manipulation of the turbulence structure in the wall-near region, in order to extract important control mechanisms due to the morphing and to put ahead a critical assessment of the morphing performances. These elements prepared the morphing of the Large Scale A320 prototype (LS) wing with high-lift flap near scale 1. Optimisation of the morphing behaviour using High-Fidelity numerical simulations, employing the most advanced modelling approaches (CFDSMComputational Fluid Dynamics Structural Mechanics). This objective helped enriching the experimentally obtained results (data base) in order to considerably reduce the design cycles and the final prototype experiments, by allowing optimised morphing shapes. By means of Adjoint-Based approaches, these optimal shapes concerning the cambering have been evaluated and contributed to the high efficiency obtained through the prototypes. Provision of the final design of the prototypes enabled by the optimised sensing/morphing concepts and by the efficient control commands and controller Interface (IF). This task proved the target aerodynamic efficiency improvements in real-time operation and in respect of the above-mentioned benefits. It has been achieved by close interaction and synergy among the simulation, multidisciplinary optimisation (MDO) and experimental tasks.

In conclusion, the SMS project produced highly innovative, disruptive aircraft wing configurations having benefits in terms of aerodynamic efficiency and noise levels that go beyond current limits, as summarized below and in the general conclusions of this book.

10

M. Braza

1.3 Affiliation List Address

Organisation INPT/IMFT

Institut National Polytechnique de Toulouse/Institut de Mécanique des Fluides de Toulouse

2 Allée du Professeur Camille Soula, 31400 Toulouse, France

INPT/LAPLACE

Institut National Polytechnique de Toulouse/Laboratoire Plasma et Conversion d’Energie

Site of ENSEEIHT, 2, rue Charles Camichel, 31071 Toulouse, France

IMP-PAN

The Szewalski Institute of Fluid-Flow Machinery, Polish Academy of Sciences

Fiszera 14st., 80-231 Gdansk, Poland

ICUBE

Université de Strasbourg, Laboratoire des sciences de l’Ingénieur, de l’Informatique et de l’Imagerie

4 Rue Blaise Pascal, 90032, F-67081 Strasbourg, France

CFSE

Computational Fluids and Structures EPFL Innovation Park—Building Engineering A, 1015 Lausanne, Switzerland

STN Gmbh

Scheller Technology GmbH

Poeler Str. 85 a, 23970 Wismar, Germany

CFDB

CFD Software Entwicklungs- und Forschungsgesellschaft mBH, CFD Berlin

Wolzogenstra β e 4, 14163 Berlin, Germany

NTUA

National Technical University of Athens

Zografou Campus, 9, Iroon Polytechniou Str, 15780 Zografou, Greece

POLIMI

Politecnico di Milano

Piazza Leonardo da Vinci, 32, 20133 Milano, Italy

CALMIP

Calcul en Midi-Pyrénées, Institut Clément Ader

Espace Clément Ader, 3 Rue Caroline Aigle, 31400 Toulouse, France

GVPM

Politecnico di Milano Wind Tunnel Laboratory

Via La Masa, 34—20156 Milano, Italy

ONERA

Centre Français de recherche Aérospatiale

2 Av. Edouard Belin, 31000 Toulouse, France

CEMENTYS

CEMENTYS/SOCOTEC Monitoring

9 rue Léon Blum, 91120 Palaiseau, France

NOVATEM

NOVATEM, Mechatronics for the future

29 Av. Didier Daurat, 31400 Toulouse, France

PRISME

Laboratoire Pluridisciplinaire de Recherche, Ingénieurie des Systèmes, Mécanique, Energétique, Université d’Orléans

8 Rue Léonard de Vinci, 45100 Orléans

1 The SMS Project: Introduction and Overview

11

1.4 Summary of the Project Results This section provides a summary of the main outcomes of the SMS project, that are described in detail in the following chapters. The quantification of the benefits percentages is presented in the general conclusions (Chap. 6) at the end of the book. The SMS project provided the following major benefits and innovative results and led to disruptive designs for the wings of the future: • • • •

Increased aerodynamic performance in all flight phases: take-off, landing, cruise. Reduced noise levels (take-off and landing phases). Higher performance than conventional high-lift configurations near scale 1. Realisation of cambered shapes achieving simultaneous lift-to drag increase and drag reduction in landing thanks to optimal cambering shapes and to the hybrid morphing. • Simultaneous noise sources decrease. • Considerable drag decrease in cruise, that is important for the greening of aircraft transport, given that the cruise phase is the most lasing flight’s phase. • Confirmation of the high performances of the hybrid morphing concerning the whole A320 aircraft by Hi-Fi numerical simulations.

In the following, the research carried out in the SMS project is presented in five chapters and in the general conclusion. In each chapter, the corresponding author, is the Work Package leader. indicated by the symbol

References 1. A. Othonos, Fiber Bragg gratings. Rev. Sci. Instrum. 68, 4309 (1997) 2. J. Scheller et al., A combined smart-materials approach for next-generation airfoils. Solid State Phenom. 251, 106–112 (2016) 3. G. Jodin, V. Motta, J. Scheller, E. Duhayon, C. Döll, J.F Rouchon, M. Braza, Dynamics of a hybrid morphing wing with active open loop vibrating trailing edge by time-resolved PIV and force measures. J. Fluids Struct. 74, 263–290 (2017). https://hal.archives-ouvertes.fr/hal-016 38290.

Chapter 2

Reduced Scale Prototype Morphing Achievements in Subsonic and Transonic Regimes G. Jodin, M. Carvalho, C. Raibaudo, C. Döll, P. Mouyon, P. Doerffer, P. Flaszynski, P. Kaczynski, M. Piotrowicz, K. Doerffer, M. Marchal, S. Cazin, D. Harribey, J. Scheller, C. Nadal, G. Harran, M. Braza, and J. F. Rouchon Abstract The present chapter describes the design of the RS (“Reduced Scale”) prototype of the SMS project as well as the experiments carried out to quantify the gain in the aerodynamic performances thanks to the morphing. The sRS prototype has been actuated according Shape Memory Alloys (SMA) ensuring large cambering deformation (up to 15% of the chord) in low frequency (in the order of 1 Hz) and higher-frequency trailing—edge actuation (order of several hundreds of Hz) with low amplitude deformation (in the order of 1–5 mm). The combination of both is the so-called hybrid electroactive morphing, term and processing defined by INPT in the literature, Fig. 2.1.

Work Package 2—WP2

G. Jodin · M. Carvalho · D. Harribey · J. Scheller · C. Nadal · J. F. Rouchon (B) INPT-Institut National Polytechnique de Toulouse/ LAPLACE-Laboratoire Plasma Et Conversion d’Energie, Site of ENSEEIHT, 2, Rue Charles Camichel, 31071 Toulouse, France e-mail: [email protected] G. Jodin · M. Carvalho · M. Marchal · S. Cazin · G. Harran · M. Braza INPT-Institut National Polytechnique de Toulouse/ IMFT- Institut de Mécanique Des Fluides de Toulouse, 2 Allée du Professeur Camille Soula, 31400 Toulouse, France C. Raibaudo · C. Döll · P. Mouyon ONERA-Centre Français de Recherche Aérospatiale, 2 Avenue Edouard Belin, 31000 Toulouse, France P. Doerffer · P. Flaszynski · P. Kaczynski · M. Piotrowicz · K. Doerffer IMP-PAN-The Szewalski Institute of Fluid-Flow Machinery, Polish Academy of Sciences, Fiszera 14St., 80-231 Gdansk, Poland © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. Braza et al. (eds.), Smart Morphing and Sensing for Aeronautical Configurations, Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153, https://doi.org/10.1007/978-3-031-22580-2_2

13

14

G. Jodin et al.

Fig. 2.1 Concept of the hybrid morphing using electroactive materials

2.1 Model Specifications G. Jodin, D. Harribey, J. Scheller, C. Nadal and J. F. Rouchon The camber can tailor the lift or the lift over drag ratio to the best value at a given flight condition. Trailing edge vibrations have the potential to enhance the noise and/or the lift and drag at different flight phases. The studied Higher Frequency Vibrating Trailing Edge (HFVTE) actuators consist in vibrating bending beams within the wing wake, past the trailing edge. Then the airfoil profile is modified by the actuators. The specifications of the new wing model’s HFVTE are the following ones: • • • • •

Quasi-static maximum displacement in two ways: +/− 10% of the actuated chord First resonance mode above 100 Hz The trailing edge must respect the airfoil profile when no actuated The trailing edge deformations must be smooth The HFVTE actuated length is fixed to 5% of the wing chord. The specifications of the new wing model’s camber control actuator follow:

• • • •

Quasi-static maximum peak to peak displacement: 1 mm Actuated chord corresponds to a A320 lap length: ≈ 30% of the wing chord Stiffness: maximum 4% chord tip displacement under transonic case loading The trailing edge deformations must be smooth and airtight.

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

15

Fig. 2.2 Illustration of the SMA actuation past 60% of the chord of the A320 RS prototype

The prototype also embedded specific sensors: pressure transducers on the rear part of the suction side for investigation of low dynamics, strain gauges and temperature sensors for actuator control, aerodynamic balance to measure lift and drag. The geometry of these actuator positions is visible on Fig. 2.2. The proposed actuator deforms the last 30% of wing profile chord length. Aerodynamic forces are considered. The working principle relies on distributed and structure embedded actuation: SMA wires are spread under the upper and lower aluminum skins of the wing. These wires are encapsulated inside silicone tubes. These tubes wind through the anchors and pulleys that are glued under the skins. The actuation of the upper wires (suction side) causes bending of the trailing edge towards higher cambered shapes. Antagonistically, the actuation of SMA wires under the pressure side skin causes a decrease in camber. SMA wires properties are activated by a change in temperature. They are heated thanks to electric current through themselves, and they are cooled down by forced air in the silicone tubes. Figure 2.2 presents a side view of the actuated trailing edge, including the antagonist SMA actuators. Six independent SMA actuators are integrated with sensors and a specific control algorithm has been developed to control the trailing edge deformation.

2.2 Higher Frequency-Lower Amplitude Trailing Edge Vibrations The vibrating actuator is composed of a metallic substrate sandwiched between two “Macro Fiber Composite” (MFC) piezoelectric patches, as shown in Fig. 2.3b. The whole is covered by a flexible molded silicon that gives the trailing edge shape. The MFC patches are LZT piezoelectric fibers and electrode networks encapsulated within epoxy. On the wing prototype, the actuator’s active length is 35 mm. Figure 2.3a presents a small prototype of a composite piezoelectric beam inside its silicone cover (the picture prototype is 60 mm long). In the following, the sizing is

16

G. Jodin et al.

Fig. 2.3 a Actuator prototype, b Diagram of vibrating piezoelectric composite beam

decomposed twofold: the piezoelectric vibrating beam is firstly design without the cover. Then the silicone cover is designed in order to limit its impact on the actuator performances.

2.3 RS Prototype B: Transonic Reduced Scale (tRS) Design P. Flaszynski, P. Doerffer, P. Kaczynski, M. Piotrowicz and K. Doerffer Dedicated to experiments in the transonic wind tunnel, two versions of prototypes have been made. The first one is a static wing, build to generate a database of aerodynamic results. The second one is the topic of this section. It consists in an actuated wing with a vibrating trailing edge. This wing’s final construction is summarized as follows. Based on CFD results together with dimensionless analysis, the actuator should be able to vibrate the trailing edge at different frequencies up to 400 Hz. The amplitude is 0.26 mm for this small wing having a chord of 150 mm. The trailing edge should also be able to be moved with quasi-static deformations. After a couple of iterations, the final design consists of a steel wing made of two parts that embed a pre-stressed stack of piezoelectric ceramics. The parts are machined using a high precision Electrical Discharge Machining (EDM). Figure 2.4 presents the design. The wing is held by two clamps provided by IMP-PAN, the SMS partner in charge of the wind tunnel. The working principle of the actuator relies on an amplification mechanism and a transmission rod. Presented in Fig. 2.5, the actuator (in blue) push on the amplification rod (red) when supplied by a voltage. The rotation of the amplification rod pushes on the amplification mechanism (green part on the figure). This last rod pushes on the trailing edge that deforms the wing tip upward, as illustrated by the structural simulation on Fig. 2.6.

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

17

Fig. 2.4 CAD view of the tRS prototype: a side view, b 3D isometric view. The piezo-actuator is presented in blue, at the center of the wing

Concerning the tRS prototype, the design of the wind tunnel at IMP- PAN in Gdansk has been assisted by Hi-Fi simulations as shown in the next section. The experimental construction of the fixed wing prototype for static measurements, followed by the morphing wing design (INPT/LAPLACE) have been assisted by HiFi simulations (INPT/IMFT) providing optimal ranges of the trailing edge actuation frequencies and amplitudes, as presented in a following section.

2.3.1 Final Geometry of the Test Section The experimental wind tunnel design has been realized by IMP-PAN as follows, concerning the test section concept and final configuration. The analysis of the flow structure on the profile in cruise condition enables to define criteria for a design of a wind tunnel section. The configuration of the test section reproduces the flow structure existing on the real profile in flight. Consequently, according to numerical simulations results for the Airbus A320 profile, the test section is designed to

18

G. Jodin et al.

Fig. 2.5 a Magnification of the trailing edge linkages. b Displacement field from a static simulation of the actuated trailing edge

achieve upstream Mach number and pressure distribution on the profile to obtain the appropriate shock wave location and strength. Thus, the wind tunnel walls have been shaped appropriately in order to reduce its influence on the investigated zone on suction side of the profile (Fig. 2.6).

Fig. 2.6 Sketch of reduced scale test section for the tRS

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

19

Numerical simulations for the test section design have been carried out by IMPPAN, using FINE/Turbo Numeca using the two turbulence models: Spalart-Allmaras and Explicit Algebraic Reynolds Stress Model (EARSM). As the first step, two dimensional (2D) simulations have been carried out for the profile at free- stream conditions. The computational domain size is defined to keep a distance of 40 times the chord length, (40c), from the profile in each direction. The boundary conditions are defined for a free-stream Mach number 0.78 and angle of attack 1.8º, according to the data provided by AIRBUS ETCT (“Emerging Technologies and Concepts Toulouse”), regarding the cruise flight phase. This configuration is considered as the design one, so the test section should be designed to maintain similar conditions upstream of the profile. The size of the profile’s chord in the wind tunnel is limited by the wind tunnel dimensions and it has to be appropriately adjusted to avoid strong effect of the walls located above and below the investigated profile. The curvature of the test section upper and lower wall follows the streamlines from simulations for free stream configuration. A set of streamlines located at different distances from the airfoil has been selected from these simulations and defined as limiting walls. Above the profile, streamlines spacing 1c and 1.5c from the profile chord line have been chosen for study. Below the profile, due to the lack of shock wave on the pressure side and weaker interaction with the wall, a lower distance than 1c has been considered. The extracted streamlines have been consequently used to define the geometry of the upper and lower walls of the transonic wind tunnel of IMP-PAN. The computational domain was limited to the space between both lines, as seen in Fig. 2.7. Simulations for the modified computational domain with upper and lower walls, inlet and outlet are done for 2D model in order to compare with free stream results. At the inlet ambient conditions (as in the wind tunnel) are applied: total pressure 101 kPa and total temperature 293 K, while at the outlet, the static pressure has been adjusted to reproduce the shock location predicted for the free-stream configuration. Due to the presence of walls and increased dissipation effects in the channel, the outlet static pressure is varying with the chosen wall distance (y/c). The wind tunnel walls configuration has been defined thanks to detailed CFD simulations in order to respect the streamline forms and avoid confinement and parasitic effects. Before 3D simulations were started, an initial 2D simulation has been performed for the designed test section. At the inlet the total pressure has been imposed, while at the outlet, the static pressure (Ps) has been adjusted to match the free-stream location of the shock wave at cruise conditions. As initial configuration the static pressure at outlet location has been set equal to 67 kPa, value which provided similar shock location than the found for freestream configuration for the 2D wind tunnel without inlet region previously analysed. Figure 2.8 compares the sonic lines for free stream and full tunnel configurations. It displays also Mach contour maps around the profile. It is seen that there are important pressure losses due to the presence of new adaptation region. As a consequence, the pressure at the outlet should be further

20

G. Jodin et al.

Fig. 2.7 Mach contour maps together with sonic line for half-span cut of wind tunnel with static pressure at outlet (Ps) equal to 62 kPa

decreased. As the lateral walls add important pressure losses as seen in previous section, a further study of best fitting outlet pressure is done for the 3D case. As next step for the full wind tunnel design, the effect of proposed adaptation region is analyzed using a full 3-D approach and the EARSM model. As seen in the 2-D analysis, the incorporation of the adaptation region leads to further pressure losses than in the case without inlet region, studied in previous section. The Fig. 2.9 displays the effect of variation of the static pressure coefficient at the outlet on the pressure, Cp, the skin friction coefficient, Cf and the isentropic Mach (Misen ) number distributions along the profile chord for the full wind tunnel configuration. Additionally, the previous profiles are compared to profiles obtained for the freestream and wind tunnel without inlet configurations (at Ps = 62 kPa; the value of outlet pressure which provided the most similar profile to the free-stream configuration). One can see that a value of Ps = 61.2 kPa provides the most similar to freestream distributions of the Mach isentropic (Misen) , Cp, pressure and Cf. Two lines are plotted for a Ps value of 61.2 kPa, this is due to the fact that results, where calculated using two different grids. The Grid No 2 was a modification of the grid employed for the rest of cases and differs from the previous grid in number of employed

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

21

Fig. 2.8 Contour map of Mach together with sonic line for 2D full tunnel configuration with Ps = 67 kPa

cells and cell spacing. These modifications led to a decrease of the aspect ratio to a maximum value of AR ~3000. This modification was performed in order to minimize the convergence issues observed in previous calculations for the full tunnel. Although some convergence difficulties were still present, the profiles obtained for the Grid No 2 show a much similar behaviour compared to the configuration without adaptation region. It was needed to remark that these differences were not originated by a grid sensitivity, but due to the presence of oscillations observed for the calculations previously performed. A more detailed view of the flow structure around the profile can be seen in Figs. 2.10, 2.11, 2.12 and 2.13. These figures display contour maps of Mach around the profile at middle span. One can observe that as previously indicated the sonic line for 3D test section simulations with Ps = 61.2 kPa and Grid No 2 fits the best the sonic line obtained for the free-stream configuration. Figure 2.14 compares the pressure coefficient, Cp, the skin friction coefficient, Cf, and the isentropic Mach (Misen ) number distributions in the cases of free-stream, tunnel without inlet and full tunnel with the static pressure at outlet of 61.2 kPa.

22

G. Jodin et al.

Fig. 2.9 Comparison of isentropic Mach (Misen), Cp, pressure and skin friction (Cf) distributions around half-span profile between free-stream, tunnel without inlet (Original) and full wind tunnel with diverse values for static pressure at outlet

The experimental investigations of the tRS delivered by INPT/LAPLACE at IMPPAN in January 2020 have focused on the shock wave unsteadiness, drag measurements and drag reduction by means of trailing edge vibrations according to defined parameters, as described in the paper. Optical access (windows at both sides) allowed for investigations of flow structure downstream of the shock and in the wake. Shock wave unsteadiness is dependent on the inlet Mach number and angle of attack, which is why a rotation of the profile together with window is included as degree of freedom for the test section design. It enables a modification of inlet angle and influence on Mach number upstream the shock and the shock wave location on the profile. The final detailed experiments are presented in Chap. 5 related to the WP5 of the project. It is worth mentioning that the numerical simulations of Chap. 4 concerning the WP4 specified optimal morphing vibration ranges that have been applied in the experiments. The experiments for the tRS are provided in respect of the final achieved performances in Chap. 5.

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

23

Fig. 2.10 Contour map of Mach together with sonic line for half-span section of the 3D full tunnel configuration with static pressure at outlet (Ps) equal to 61.5 kPa and total pressure at inlet (P0) equal to 101 kPa

2.3.2 Experimental Verification and Proof of the Morphing/Sensing Efficiency-sRS G. Jodin, M. Carvalho, C. Raibaudo, C. Döll, P. Mouyon and M. Braza The objective of this task aims at demonstrating the benefits obtained by the SMS project concepts. It is first studied in Reduced Scale prototype (RS) in subsonic and transonic speeds, in order to provide a very detailed set of measurements. For this, highly advanced optical techniques as the High-Frequency (order of 6–10 kHz) TimeResolved PIV is used in a reduced area around the trailing-edge and the near wake. These velocity measurements, coupled with pressure sensing and forces measurements will enable studying in detail the modification and manipulation of the turbulence structure in the wall-near region, in order to extract important control mechanisms due to the morphing and to put ahead a critical assessment of the morphing performances. It is secondly studied for the sensing/morphing phase of a Large Scale prototype (LS) wing with high-lift flap.

24

G. Jodin et al.

Fig. 2.11 Contour map of Mach together with sonic line for half-span section of the 3D full tunnel configuration with Ps = 61.2 kPa

Concerning the sRS prototype (Fig. 2.17), the experimental design in the S4 wind tunnel of IMFT has been carried out by INPT (LAPLACE-IMFT), as follows:

2.3.2.1

Sensor Implementation and Experimental Setup—sRS

This section describes the implementation of the sensing and controlling devices. Then the experimental setup in wind tunnel is described. Figure 2.15 presents the devices and interface that are used to calibrate and control the wing before and during the wind tunnel experiments. Note that the current version relies on commercially available interfaces. Attention has been paid to have analog interfaces that will be adapted to the IF (Interface) system from partner STN, as mentioned in Chap. 5. Figure 2.16 presents the experimental wind tunnel set up. An aerodynamic balance based on strain gauges has been designed to measure lift and drag force on the wing. Precautions in design and mounting have been taken to ensure the quality of the measures despite the vibrations and the wires coming from the embedded sensors and actuators. Three dynamic pressure transducers (MEGGIT 8507C-1) are implemented on the suction surface (upper skin) of the prototype. They are lined up one behind the

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

25

Fig. 2.12 Contour map of Mach together with sonic line for half-span section of the 3D full tunnel configuration with Ps = 61 kPa

other. The respective locations of the sensor P1, P2 and P3 are closed to the trailing edge, i.e. respectively at 80%, 86% and 93% of the wing chord. The wind tunnel also includes the High Speed Time Resolved Particle Image Velocimetry (TRPIV) measurements. These measurements permit an accurate and local observation of the flow and the vortex dynamics. Figure 2.18 presents the wind tunnel with the PIV setup. Calibrated smoke particles are introduced in the airflow for this purpose. The depth of the field was focused on the 2.5 mm thick stream-wise laser light sheet—in Fig. 2.18 the green area corresponds to the laser sheet within the camera range. The laser pulsations are generated by a two cavity Nd:YLF (527 nm) laser (Photonics Industries International Inc. DS-527–60). Using mirrors, the laser lights up the wing’s wake from the bottom and the laser sheet is reflected from up to bottom to light up the upper side of the wings trailing edge. This arrangement allows for novel PIV measures of the flow over a wing with a vibrating morphing trailing edge. Particle images are recorded during the experiment using the digital highspeed camera Phantom V1210. Each image is divided into interrogation windows. The interrogation window size is 16 × 16 px2 (px being Pixel), which corresponds to 3.4 × 3.4 mm2 , with an overlap of 75%. The computation of the velocity fields

26

G. Jodin et al.

Fig. 2.13 Contour map of Mach together with sonic line for half-span section of the 3D full tunnel configuration with Ps = 61.2 kPa and Grid No. 2

from the images was done using an open source MPI software CPIV-IMFT that runs on regional clusters to compute a set of more than 50,000 images in a short time (e.g. 2 h 41 min on 15 nodes of 20 CPU each). Figure 2.19 shows an example of a very detailed experimental data base and the benefits from the morphing effects after actuation near the trailing-edge at optimal frequencies. The experiments for the sRS have been carried out in S4 wind-tunnel at the IMFT facilities. The test section is 592 mm width per 712 mm high. The wing is mounted at an incidence of 10°. As a result, a blockage ratio of 18% is obtained, which is considered acceptable in these experiments. All the results presented in the next sections were obtained at a Reynolds number of 1 million. The prototype was designed by INPT/LAPLACE and it corresponds to a reduced model of the Airbus A320 wing. The chord is 700 mm and the span is 590 mm. Both actuation systems are embedded in the last 30% of the chord, following real flap configuration in commercial aircrafts. The camber control actuation system is made of six SMA (Shape Memory Alloys) wires pre strained to +3% with respect to their initial length. At room temperature, SMA present martensite crystalline structure, but when it is heated by Joule effect, the crystalline structure becomes austenite. This modification in the material internal structure generates a change from plastic to elastic regime, so the SMA tend to

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

27

Fig. 2.14 Comparison of isentropic Mach (Misen), Cp, pressure and skin friction (Cf) distributions around half-span profile between free-stream, tunnel without inlet (Original) and full wind tunnel with static pressure at outlet (Ps = 61.2 kPa)

Fig. 2.15 General view of the test section for the tRs at IMP-PAN

28

G. Jodin et al.

Fig. 2.16 Sensing and controlling interface of the sRS prototype

Fig. 2.17 Schematic representation of the morphing on the sRS

recover the initial length once actuation is interrupted. The stress created during this transformation generates a bending moment capable of modifying the camber of the prototype.

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

29

Fig. 2.18 Wind tunnel test section drawing with hybrid morphing wing model. PIV setup is represented, the PIV plane is located mid-span

Fig. 2.19 TRPIV in static (left) and morphing (right) configuration, showing the reduction of the wake’s width, the breakdown of large coherent vortices and the reduction of the separation region. sRS prototype, Reynolds number 1 M, a = 10°, A320 wing in clean configuration

The piezoelectric actuators, also called Higher Frequency Vibrating Trailing Edge (HFVTE) is designed to reach vibration amplitudes up to 2 mm depending on the actuation frequency. The HFVTE is formed by two piezoelectric MFC patches glued on both sides of a metallic substrate. In order to obtain maximum displacement on both sides, the MFC patches are activated alternatively.

30

G. Jodin et al.

Fig. 2.20 PSD of lift signals: comparison of the spectral density reduction (in dB)

2.3.2.2

PSD of Lift Signals

Force measurements were made using an aerodynamic balance designed and fabricated at IMFT by the laboratory’s support team. For the acquisition, a Yokogawa DL850 Oscilloscope was used. It allows us to make the zero adjustment with no need for signal conditioners and the sampling frequency goes up to 5 kHz, which is the rate used in this campaign. Three dynamic pressure transducers MEGGITT 8507C-1 are placed at 60%, 80% and 85% of the chord and lined up at 56% of the span. The actuation signal is supplied through a dSPACE MicroLabBox, as well as the signal acquisition which is made at a rate of 6 kHz. Cut-off frequency corresponds to 1 kHz since we expect to identify coherent structures at maximum frequency of 400 Hz. Signal length is 300 s, more than enough to reach the convergence of mean and RMS values. Figures 2.20 and 2.21 show the PSD of lift and pressure signals showing attenuation of energy between 10 and 50 Hz, especially when we approach the trailing edge. This frequency range corresponds to coherent structures such as the Von-Karman vortices. The reduction of energy levels in this region of the spectrum may indicate that the vibration of the trailing edge acts positively on these structures. A Power Spectrum Density (PSD) analysis was made to investigate the effects of HFVTE actuation on the stabilization of the flow. Monochromatic actuation at 8, 12, 30, 98, 120, 170, 250, 300 and 320 Hz were compared to the reference case, where there is no actuation. As one can see in Figs. 2.20 and 2.21, for lower frequencies there is no relevant influence of morphing on the flow. But from 98 Hz an energy reduction of the signal is observed, similar behavior is also seen in cases of actuation at 120 and 170 Hz. At higher frequencies, this effect is less pronounced.

2.3.2.3

Mean and RMS Values of Lift

The mean and RMS values of the lift were computed for the most relevant actuation cases. The variation of the lift signal in relation to the reference case is shown is

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

31

Fig. 2.21 PSD of the pressure signals attenuation of the spectral density level (dB) in the intermediate frequency range

this section. Firstly, one can observe in Fig. 2.22, the increase in lift with increasing actuation frequency. This result is consistent with previous measurements made for different Reynolds Numbers. Even if the curve does not show a gain in lift, we can expect that it will continue to rise. The same analysis for lower Reynolds numbers shows that lift enhancement depends on the inlet velocity. The higher is the velocity, the higher is the frequency required to obtain a gain in lift. Additionnal lift increase is obtained by the hybrid electroactive morphing as shown in Fig. 2.23. The variation of RMS values also indicates a rise as the frequency increases. This can be explained by the signature of the actuation. The vibration induced on the trailing edge propagates through the prototype and is captured by the balance. The PSD analysis is more appropriate to investigate an attenuation of main instabilities and of the fluctuating flow behaviour, by quantifying the amplitudes of the whole frequency range governing the turbulent flow under the morphing action.

2.3.2.4

Particle Image Velocimetry (PIV)

M. Carvalho, S. Cazin, M. Marchal and M. Braza The goal of the PIV campaign is to investigate the effects of dynamic cambering on the flow. In other to image both upper and lower surfaces of the wing, a 400 mm

32

G. Jodin et al.

Fig. 2.22 Lift increase as a function of the actuation frequency

Fig. 2.23 By the hybrid electroactive morphing (simultaneous cambering and near-trailing-edge vibration): the fact of actuating the vibration provides 2% more lift increase

Nikon lens was chosen. It was then necessary to use a flexible mask to avoid direct laser reflection from the prototype surface to the camera sensor. Furthermore, images were recorded at 5 kHz using a digital high-speed camera LaVision Phanton V2012. The laser sheet is generated by a Photonics DM60–527-DH laser. The diameter of the particles is 0.5 µm, they were employed to obtain an investigation window size of 170 per 260 mm. The commercial software DaVis 10 from LaVision and the software CPIV, developed by IMFT, were used for post-processing.

2 Reduced Scale Prototype Morphing Achievements in Subsonic … Fig. 2.24 Snapshots of SMA actuation. a No camber and b full SMA actuation affecting the shear layer vortices

(a)

33

(b)

(b)

Cambering takes place from neutral position to full camber in 1.5 s, the tip displacement corresponds to 25 mm. Snapshots of the acquisition sequence show two different solution times of the PIV (Fig. 2.24). As expected, one can see the increase of the recirculation zone in the wake once full cambering is reached by means of SMA actuation. The Proper Orthogonal Decomposition (POD) was employed to characterize the coherent structures in the flow. It is a mathematical method used to detect the coherent structures featured by the flow based on their wake number and frequency. The POD allows us to understand the flow behavior by finding the so-called POD modes that form the dynamic system. This approach is largely used in the analysis of experimental data. In order to identify the high energy structures such as the Von-Karman vortices, the velocity field was reconstructed using the most energetic modes, as the energy distribution of the POD shows in Fig. 2.25. The first 9 modes were chosen and reconstruction was made using the modes by pairs, knowing that the mode 1 corresponds to the average inlet velocity. Figure 2.26 shows the reconstruction of the velocity field by pairs of modes. The vertical component of the velocity is displayed. The structures featured are dependent on geometry, presenting only an alternation between the positives and negatives cores seen bellow. Once they are added, we are able to reconstruct the main coherent structures of the flow, as one can see in Fig. 2.27. The figure shows three snapshots of the reconstructed velocity field sequence. Instead of having geometrically static

34

G. Jodin et al.

Fig. 2.25 Energy distribution of POD modes

structures, we observe moving vortices that propagate in the flow direction. By adding more POD modes to the reconstruction, we can isolate the most energetic structures and identify their frequency. To do so, monitor points, seen in Fig. 2.28, can be used in strategic positions of the velocity field. The PSD of the signal shows that the structures formed by the most energetic POD modes can be found in the low frequency range of the spectrum.

2.3.2.5

Feedback Control of the Reduced Scale Prototype

C. Raibaudo, C. Döll, P. Mouyon and M. Carvalho The main objective of the present work is to perform experimental feedback control for the reduced prototype in order to increase the aerodynamical performances. The overall methodology to achieve the implementation of feedback controllers for the morphing wing is presented in Fig. 2.29. From the morphing wing with target flow conditions, experimental data are acquired from the flow without and with control in open-loop. From these informations, a dynamical oscillator was chosen and designed in order to fit with the spectral properties of the natural flow. Feedback controllers are developped and first tested on this dynamical oscillator through Matlab/Simulink, then implemented in a dSPACE MicroLabBox unit for the experimental feedback control of the prototype.

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

35

Fig. 2.26 Reconstruction of the flow using reduced number of POD modes. a Modes 2 and 3; b modes 4 and 5; c modes 6 and 7; d modes 8 and 9

Fig. 2.27 Sequence of snapshots from POD reconstruction using modes 2–9

2.3.2.6

Identification of the Plant Dynamics Using Open-Loop Control

A characterization of the flow properties is necessary to build a corresponding model. The more representative the model is, the more confident we are to succeed for the practical implementation. Power Spectral Density Function (PSDF) of the baseflow

36

G. Jodin et al.

Fig. 2.28 PSD of monitor point P1 at reconstructed flow using modes 2 to 9. The point is located at (x/c = 1.07, y/c = −0.06)

Figure 2.29 Schematic representation of the feedback control: methodology implemented for the sRS prototype

at different freestream velocities are presented in Fig. 2.30. Freestreams velocities U∞ = 10.5 m/s (Fig. 2.30a), 15 m/s (Fig. 2.30b), 21.5 m/s (Fig. 2.30c), correspond to approximate Reynolds numbers of 0.5, 0.7 and 1 million respectively. Strong peaks around 12 Hz corresponds to fluid-structures interactions, between shedding dynamics and airfoil structural modes. Secondary peaks are also observed around 30 Hz, corresponding to the main dynamics observed in first POD modes spectra.

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

37

Fig. 2.30 Power spectral density function (PSDF) of the baseflow without control for U∞ = a 10.5 m/s, b 15 m/s, c 21.5 m/s (ONERA—INPT/IMFT)

Fig. 2.31 Power spectral density function (PSDF) of the flow under modulated control from 1 to 80 Hz at U∞ = 21.5 m/s for pressure sensors at x/c = a 0.6, b 0.8 and c 0.85

38

G. Jodin et al.

Modulation of the actuation frequency is performed for the plant identification. Linear modulation of the frequency has a flat spectral response, if the duration of modulation is long enough. Results of one wobulation at U∞ = 21.5 m/s for low frequency range (between 1 and 80 Hz) are presented in Fig. 2.31, for the three pressure transducers. Peaks at low frequency (around 8 and 12 Hz) are not removed, but slightly reduced. For intermediate actuation amplitudes (A = 7 V), reduction of almost a half of the peak amplitude at 12 Hz can be achieved. Natural phenomenon between 25 and 30 Hz is also slightly reduced with the control.

2.3.2.7

Feedback Control of the Morphing Wing

C. Raibaudo, C. Döll and P. Mouyon Design of Controllers Using Dynamical Models For the configuration considered, significant frequencies have been found in the baseflow spectral analysis and in the previous studies in the PSD of temporal coefficients modes of the POD corresponding to different physical phenomena in the airfoil wake [1]. In order to design closed loop control laws, a model that approximated the behavior of the flow along with its actuators and sensors is used. The model must not be too much complex so that it can be used easily for control design purpose. Nevertheless it must capture the main characteristics of the phenomenon to be control. In the present case the control is intended to reduce intrinsic vibratory phenomenons that are present in the flow prototype. A careful consideration of the flow frequencies in this low-order model used for the simulation is then crucial. Nonlinear oscillators such as the Van der Pol (VdP) oscillator are therefore an satisfying choice. Van der Pol non-linear oscillator is chosen from previous works as being able to represent flow stability and dynamics [2]. This model has been used to reproduce and control the dynamics of simplified bluff bodies [3] or airfoils [4] for example. The difficulty we encountered in order to work in 3D complex flows comes in part from the fact that the spectrum of the measures in 3D is broadband and no longer narrow band as in 2D. In order to model such a behavior we introduce a nonlinear chaotic oscillator that is known to deliver a broadband signal still in an intrinsic fashion. The state space representation for the controlled Van der Pol model is expressed as: ⎧ ⎨ x˙1 = x2 + Gu   x˙ = −ω02 x1 + 2ξ ω0 − 3K s x12 (x2 + Gu) ⎩ 2 y = x1 with x1 , x2 the states of the system, y the measured output, and u the control input. Constants are tuned based on the desired permanent regime response [2]. Figure 2.32a

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

39

Fig. 2.32 PSDF of the VDP oscillator output (a) without and (b) with random fluctuations around the main frequency—ONERA

shows the Power Spectrum Density Function (PSDF) of the VdP oscillator. Enlargement of the spectrum band around the main around the main pulsation is also considered in Fig. 2.32b in order to represent more properly the physical phenomenons dynamics of the studied flow. Static gain controller is first considered. It corresponds by fixing the gain K and delay τ manually of the output feedback control: u(t) = K y(t − τ ). Scans of gain and delay are performed in order to find the optimal values for which the fluctuations of output signal y are completely reduced. Results of the feedback control for both unperturbed and perturbed oscillators are presented in Fig. 2.33. For the unperturbed VdP oscillator, the feedback controller succeeds to reduce the instantaneous standard deviation of the signal y. However, for the perturbed VdP, by construction more default to stabilize, periods of significant destabilization are observed. To improve the control, adaptive approaches are considered here. A cost function J is defined to evaluate the controller performances, here minimizing the fluctuations 2 , with σ yy the variance of y. The feedback controller of the oscillator output J = σ yy law u(t) = K y(t − τ ) with adaptation of the gain K based on the minimization of J is implemented. Complementary to the gain adaptation, a lattice filter is included in the controller. Lattice form is another way to implement linear dynamic filters. The adaptation parameter θ . is here the poles and zeros of the filter θ = {ki , υi }. Poles location derives from the {ki }, while {υi } fixes the filter zeros. The adaptation of the lattice filter is carried out by an approximated gradient descent. Stability is easily ensured by maintaining ki  ≤ 1. Furthermore, the gradient may be approximately evaluated by an auxiliary lattice filtering.

40

G. Jodin et al.

Fig. 2.33 Results of the closed-loop control of the unperturbed (top) and perturbed (bottom) Van der Pol model for constant gain and delay—ONERA

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

41

Results of the feedback control for both unperturbed and perturbed oscillators using adaptations of gain and lattice filter are presented in Fig. 2.34. Control signal u and output signal y times series are presented in left (Fig. 2.34a–c), when the controller gain K are shown in right (Fig. 2.34b–d). For both models, even the perturbed oscillator, adaptation of the lattice form and gain is therefore more efficient for the complete stabilization of the plant.

Implementation of Controllers on the Experimental Set-Up C. Raibaudo, M. Carvalho, C. Döll and P. Mouyon Feedback controllers were therefore implemented for the reduced scale prototype. The main loop for the experimental feedback control is presented in Fig. 2.35. Pressure fluctuations of the three transducers at the pressure side are acquired in real-time using a dSPACE MicroLabBox unit. These signals are also used in the adaptation loop to calculate the cost function J , in order to evaluate the controller performances: J =

Ns 

 2 sj σ yy

j=1

 with Ns = 3 the number of sensors, and σ yy s j the variance of sensor j. Miniminization of this cost function is used to adapt the parameter θ , here the gain K or the delay τ of the control law: u(t) = K y(t − τ ). Current cost function J is compared with the value J0 for the natural flow without control. Control signal is therefore increased using power amplifiers to correspond to the piezoelectric actuators requirements. Similarly to the VdP oscillator control, scans of gain and delay of the control law explicated previously are performed in Fig. 2.36. Adaptation parameter is maintained for 30 s to ensure a proper convergence of the statistics. Cost function beneficit J /J0 = (J − J0 )/J0 is computed in percentage, for evaluation of the optimal set of values {K ∗ , τ ∗ }. The search of optimal values is done iteratively, by scanning the delay by fixing the gain from previous scan, then by scanning the gain using the optimal delay found. Figure 2.38 shows the final steps of this optimization, leading to optimal values of {K ∗ , τ ∗ } = {1.2, 0.04 s}. It should be noticed that higher values of gain, therefore higher actuation amplitude, tends to deteriorate the control efficiency, which has been found in previous experimental flow control studies ([5–7]). The optimal set of parameters {K ∗ , τ ∗ } = {1.2, 0.04 s} is then tested alone to ensure their performances. Results of constant gain and delay parameters feedback control are presented in Fig. 2.37. The actuation is first off, then starts after 120 s. Cost function with optimal parameters is reduced in mean by −6.5%, and in maximum by −10.5%. The gain is now adapted using a gradient-like approach. A search grid is generated with different gains to be tested, defined by a mean value and a range. After all gains of this grid are tested experimentally, the grid range is reduced in order to

42

G. Jodin et al.

Fig. 2.34 Results of the closed-loop control of the (a–b) unperturbed and (c–d) perturbed Van der Pol for adaptations of gain and lattice filter. Left: control and output signals, right: gain

Fig. 2.35 Main loop for the implementation of adaptative feedback controllers for the plant

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

43

Fig. 2.36 Evolution of the cost function benefit J /J0 = (J − J0 )/J0 (top) and adapted parameter (bottom) w.r.t the time. a Scan of delay τ with K = 1.33, then b scan of gain K with τ = 0.04 s

Fig. 2.37 Evolution of the cost function benefit J /J0 = (J − J0 )/J0 with constant gain and delay parameters {K ∗ , τ ∗ } = {1.2, 0.04 s} for the feedback control

44

G. Jodin et al.

Fig. 2.38 Adaptation of the gain using gradient-like approach with τ ∗ = 0.04s. a J /J0 , b Jopt /J0 , c gain K , d optimal gain K opt

converge to an optimal gain K ∗ . Compared to a strict gradient descent technique, this approach allows to find ranges where the control is efficient, and is more fitted with the current experimental set-up. Results of the approach for the experimental set-up are presented in Fig. 2.38. The optimal cost function Jopt /J0 is reduced by −4.4%, with an optimal gain K ∗ = 1.2 5. These results are consistent with the scans of gain presented in the previous section, suggesting an intermediate value where the gain is sufficiently high enough to have authority on the flow, but without detiorating the control at higher actuation amplitude.

References 1. G. Jodin, V. Motta, J. Scheller, E. Duhayon, C. Döll, J.F. Rouchon, M. Braza, Dynamics of a hybrid morphing wing with active open loop vibrating trailing edge by time-resolved PIV and force measures. J. Fluids Struct. 74, 263–290 (2017). https://hal.archives-ouvertes.fr/hal-016 38290 2. V. Motta, P. Mouyon, C. Döll, Discrete time open-loop and closed-loop control based on Van der Pol modeling. in 8th AIAA Flow Control Conference (2016)

2 Reduced Scale Prototype Morphing Achievements in Subsonic …

45

3. M. Provensal, C. Mathis, L. Boyer, Bénard-von Kármán instability: transient and forced regimes. J. Fluid Mech. 182, 1–22 (1987) 4. M. Khalid, I. Akhtar, Modeling the aerodynamic lift produced by oscillating airfoils at low Reynolds number. Eng. Phys. (2014) 5. O. Lögdberg, Turbulent Boundary Layer Separation and Control (Technical Report from Royal Institute of Technology KTH Mechanics, Sweden, 2008) 6. J. Bons, R. Sondergaard, R. Rivir, Turbine separation control using pulsed vortex generator jets. Trans. ASME 123, 198–206 (2001) 7. J. Ortmanns, M. Bitter, C. Kahler, Visualisation and analysis of dynamic vortex structures for flow control applications by means of 3c2d-piv. In: 12th International Symposium on Flow Visualization (2006)

Chapter 3

Large Scale Morphing Prototype: Design and Experiments A. Giraud, B. Nogarede, Y. Bmegaptche-Tekap, M. Carvalho, C. Korbuly, A. Kitouni, J. B. Paris, V. Lamour, A. Marouf, J. B. Tô, A. Polo-Dominguez, M. Scheller, J. Scheller, D. Harribey, M. Braza, and J. F. Rouchon Abstract The present chapter has as a main objective the design of the LS prototype by INPT/IMFT-INPT/LAPLACE, equipped with the novel Bragg grating sensing (CEMENTYS) and with the Electromechanical (EMA) actuators by NOVATEM, able to apply optimal camber control according to the shapes dictated from Hi-Fi CFD numerical simulations (next chapter). The cambering control by ONERA is described in Chap. 5 together with the experimental results. The computer architecture allowing application of the control commands has been realised by partner STN and is described in the last section of chapter.

Work Package 3—WP3

A. Giraud · B. Nogarede NOVATEM, Mechatronics for the Future, 29 Av. Didier Daurat, 31400 Toulouse, France Y. Bmegaptche-Tekap · M. Carvalho · C. Korbuly · A. Marouf · J. B. Tô · A. Polo-Dominguez · M. Braza INPT-Institut National Polytechnique de Toulouse/IMFT-Institut de Mécanique Des Fluides de Toulouse, Allée du Professeur Camille Soula, 31400 Toulouse, France M. Carvalho · D. Harribey · J. F. Rouchon (B) Laboratoire Plasma Et Conversion d’Energie, Site of ENSEEIHT, INPT-Institut National Polytechnique de Toulouse/LAPLACE, 2, Rue Charles Camichel, 31071 Toulouse, France e-mail: [email protected] A. Kitouni · J. B. Paris · V. Lamour CEMENTYS/SOCOTEC Monitoring, 9 Rue Léon Blum, 91120 Palaiseau, France A. Marouf Laboratoire Des Sciences de L’Ingénieur, de L’Informatique Et de L’Imagerie, ICUBE, Université de Strasbourg, 4 Rue Blaise Pascal, 90032, 67081 Strasbourg, France M. Scheller · J. Scheller STN Gmbh, Scheller Technology GmbH, Poeler Str. 85 a, 23970 Wismar, Germany © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. Braza et al. (eds.), Smart Morphing and Sensing for Aeronautical Configurations, Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153, https://doi.org/10.1007/978-3-031-22580-2_3

47

48

A. Giraud et al.

3.1 Optimal Design of the Electromechanical Actuators (EMA) System 3.1.1 Electromechanical Actuator System A. Giraud and B. Nogarede Electromechanical actuators or EMA are a large family of actuators and refer to different technologies, depending on their use. For the LS prototype, they could be used on the rotation axes of each articulation as a direct rotating motor. However, due to the very high torque needed to control the camber in the specified conditions, it has been preferable to use the leverage around hinges. In this way, the rotation is obtained with a linear force applied on the parts around the hinges. EMA are composed of an electrical motor, a gearbox, a screw-nut system and frames and bearings. The rotating motion is provided by the motor, then adapted thanks to the gearbox and transformed into a linear motion through the screw-nut system. The flap profile has to be controlled to reach the shapes between the two positions, Low Cambered Shape and High Cambered Shape, from a reference profile. In order to control the profile, a specific articulated wing has been designed. The wing is divided in six parts linked by five hinges, such as a dorsal vertebra. The corresponding dimensions shown in Figs. 3.1 and 3.2 are summarised in Table 3.1. Table 3.2 presents the different torques applied on each hinge and its limit angles of opening. A positive angle represents the maximum angle of opening over the hinge where the negative angle represents the maximum angle of opening under the hinge. The torque specifications are made for a macro-actuator, which is defined as a group of actuators designed produce the torques on hinges of Table 3.2. Once equipped with four macro-actuators, the morphing flap will fit in the real flight conditions. Figure 3.2 represents the concept of a four macro-actuators flap.

Table 3.1 Hinge location Distance

(m)

C

1

TE spar

0.85

LE spar

0.15

3 Large Scale Morphing Prototype: Design and Experiments Table 3.2 Torque and angle specifications

Hinges

Torque (N m)

49 Angle > 0

Angle < 0

a1

446

4.8

-2.5

a2

140

1.5

-5.3

a3

88

1.5

0.1

a4

62

2.2

0.4

a5

28

3.4

-4.1

Fig. 3.1 Articulated structure detailed

Fig. 3.2 Four macro-actuators represented in the morphing flap

3.1.2 Electromechanical Actuator Design A. Giraud, B. Nogarede and Y. Bmegaptche-Tekap 3.1.2.1

Actuation Principle

The solution presented here uses Electromechanical Actuators (EMA. EMA may be used directly on the rotation axes of each articulation. However, due to the very high torque needed to control the camber in the specified conditions of the project, we decided to use the leverage available above and below the articulation (y-axis on Fig. 3.1). In this way, the rotation is obtained with a linear force applied on the x-axis (Fig. 3.1) with a linear NOVATEM’s EMA. A more detailed scheme of the first articulation is presented in Fig. 3.3. Thus, the position of the NOVATEM’s EMA can be determined to apply the needed force to articulate the hinge a1. The same

50

A. Giraud et al.

Fig. 3.3 Detailed sketch of the leverage using above the first hinge from its centre

Table 3.3 Mechanical and geometrical specifications of all hinges Hinges

Torque (N m)

Leverage (m)

Force (kN)

Stroke (mm)

a1

446

+ 0.05

9

7

a2

140

+ 0.04

3.5

6

a3

88

− 0.03

3

1.5

a4

62

− 0.03

2

1.5

a5

28

+ 0.012

2.35

2

process is used for all hinges. Table 3.3 sums up the different torques, leverages and force for each hinge. Depending on the leverage and the angles, the corresponding force is also given in Table 3.3.

3.1.2.2

Actuator Description

The synchronous permanent magnet motor (Fig. 3.4) is designed to get the maximum possible specific torque and efficiency, but for the highest compacity. It is thus designed around a high saturation iron cobalt stator yoke with a high-density winding allowing for high linear current density. Because of the relatively high reduction ratio of the EMA assembly, the motor has a relatively high rotational speed, thus inducing iron losses. These losses are attenuated by the low magnet polarization and large airgap. A maximum efficiency of 94% was reached thanks to the implemented topology and a specific torque of almost 0.9 Nm/kg. These two combined figures make of this motor ideal for the current EMA application. The motor general parameters are gathered in Table 3.4. The gear of our EMA must have a high reduction ratio, to get the highest torque possible on the nut. The parameters of the chosen gear are gathered in Table 3.5.

3 Large Scale Morphing Prototype: Design and Experiments

51

Table 3.4 Permanent magnet motor parameters Outer diameter (mm)

Outer shaft diameter (mm)

Overall length (mm)

Max. Torque (mN m)

Max. Speed (rpm)

18.5

5

42.5

35

10k

Fig. 3.4 Photo of the NOVATEM motor

The efficiency ηg = 64% of the gearbox depends of the power flow direction. The gearbox transmits the motor torque to a screw-nut, up to 12.2 Nm. The screw-nut transmission allows the conversion of the rotational motion provided by the motor through the gearbox into a linear motion, apply a force on a leverage. The output torque of the gearbox TG is linked to the linear force FS provided by the skrew with the following equation: TG = p.

FS 2π ηs

where p is the skrew thread and ηs is the efficiency of the skrew-nut, given by the following equations, respectively for direct and inverse transmission: ηs =

Table 3.5 Gearbox parameters

1 1+

dμ p

= 0.88 ηsinv = 2 −

1 = 0.87 ηs

Outer diameter (mm)

Length (mm)

Reduction ratio

42

84

546

52

A. Giraud et al.

Table 3.6 Screw-nut parameters Screw diameter (mm)

Nut diameter (mm)

Nut length (mm)

Screw length (mm)

Thread (mm)

16

32.5

30

90

5

The constant μ = 0.010 depends on the helix angle α = 4.3◦ of the skrew whose diameter is d. Therefor, with 12.2N.m on the nut, the skrew can apply a force up to 13.37, kN. Table 3.6 presents the parameters of the screw and the nut.

3.1.2.3

Actuator Integration

First, the EMA itself must be integrated. It has been done in NOVATEM Facilities. All parts are connected thanks to a specific frame. This frame consists in a carter, a nut-carrier and bearings. The carter brings a mechanical cohesion to the EMA and permits its integration to the flap ribs. The nut-carrier allows the nut rotation around the screw, transforming the rotation (and so torque) from the motor (though the gearbox) in a linear movement (and so force). Carter and screw are connected to the two different ribs around a hinge, allowing its opening and closing. The bearings have a transmission and a protection role: they transmit the force from the screw to the carter, and from the carter to the screw. The gearbox is inserted inside a dedicated frame with an integrated inverter. A LVDT sensor is added to measure the stroke of the screw, between the screw head and the nut-carrier. This inverter adapts the power to the motor and allows its control. In this way, power supply and control are localized on the EMA improving its compactness. Figure 3.5 presents a detailed CAD view of the EMA and Figure 3.6 presents the manufactured parts of the EMA. Second, each EMA is connected to the flap in order to actuate each hinge, forming a macro-actuator of five EMA. Each EMA is located between two rib chain. A rib chain consists in the six ribs along the profile chord, from leading edge to trailing edge. In this way, each EMA has a dedicated space around its location. The mechanical structure which forms the rib chain and so the macro-actuator structure has been designed in a highly collaborative work between INPT and NOVATEM. The interface between the EMA and the flap are determinant: the forces provided by the EMA have to be perfectly transmitted to its structure. The whole structure is manufactured in the INPT workshop, providing a high-quality result. The macro-actuator is 500 mm span long and the expected four macro-actuators fit in the flap span. A CAD sketch of one macro-actuator with five EMA is presented in Fig. 3.7. Each EMA is connected to the flap in order to actuate each hinge, forming a macro-actuator of five EMA, Figs. 3.8 and 3.9. Each EMA is located between two rib chain. A rib chain consists of six ribs along the profile chord, from leading edge to trailing edge. In this way, each EMA has a dedicated space around its location. The

3 Large Scale Morphing Prototype: Design and Experiments

53

Fig. 3.5 Detailed CAD view of the EMA

Fig. 3.6 Photo of the EMA manufactured parts

mechanical structure which forms the rib chain and so the macro-actuator structure has been designed in a highly collaborative work between INPT and NOVATEM. The interface between the EMA and the flap are determining: the forces provided by the EMA have to be perfectly transmitted to its structure. The whole structure

54

A. Giraud et al.

Fig. 3.7 Macro-actuator with five EMA

Fig. 3.8 Morphing flap with EMAs

is manufactured in the INPT workshop, providing a high-quality result. The macroactuator is 500 mm span long and the expected four macro-actuators fit in the flap span.

3.2 Sensing System A. Kitouni, J. B. Paris and V. Lamour

3.2.1 Theory The system consists of a Fiber Bragg Grating (FBG) pressure sensor (shown in Fig. 3.10) followed by an optical interrogator and a treatment unit. The fiber Bragg grating is glued on a membrane which is also glued on a baseplate, see Fig. 3.11.

3 Large Scale Morphing Prototype: Design and Experiments

55

Fig. 3.9 Morphing flap CAD with embedded EMAs Fig. 3.10 Fiber Bragg Grating (FBG) pressure sensor

Optical Fiber outlet

Pressure FBG Fig. 3.11 Schematic side-view of the sensor

Wing

56

A. Giraud et al.

Fig. 3.12 Schematic representation of the Fiber Bragg grating

3.2.2 Fiber Bragg Grating (FBG) Sensor Technology A FBG behaves as an optical filter that selects a narrow spectrum around the Bragg wavelength. Indeed, it is a periodic modulation of the core refractive index of an optical fiber which creates a resonant structure. The peak wavelength of the narrowband spectral component reflected by the FBG is given by [1], see Fig. 3.12: λ B = 2n e f f Λ

(3.1)

With λ B the Bragg wavelength, n e f f the effective refractive index of the grating and Λ the period of the grating. Any perturbation that can change the refractive index or the periodicity of the grating area will result in a shift in Bragg wavelength. The relative shift in the Bragg wavelength is given by [1] as follows: Δλ B λB

=

(

1 ∂Λ Λ ∂ε

+

1 ∂n e f f n e f f ∂ε

) ( ε + Λ1 ∂Λ + ∂T

1 ∂n e f f ne f f ∂ T

) ΔT

(3.2)

With ε the strain applied on the grating area, ΔT the change in temperature. The change in pressure is neglected compare to strain and temperature variations and thus, not considered in Eq. 3.2. Indeed, as the FBG sensor is more sensitive while measuring strain constraints (compare to pressure), isoradial-placed FBG sensors are fixed on the membrane. Consequently, pressure applied on the membrane will be seen as a strain on the FBG sensor. If the strain is the only perturbation to be considered, the relative shift becomes [2]: Δλ B λB

= Kl ε

(3.3)

) ( 1 ∂n e f f with K l = Λ1 ∂Λ + the strain sensitivity. For optical fibers, K l is about ∂ε n e f f ∂ε 0.78. Consequently, at 1550nm, the sensitivity coefficient is about 1.2pm/με for the FBG. As represented in Fig. 3.13, the membrane can be modelled to study the mechanical performances of the selected design.

3 Large Scale Morphing Prototype: Design and Experiments

57

Fig. 3.13 Mechanical modeling of the sensor’s membrane

The pressure P applied on the membrane is derived as follows [3]: P=

8εr Et 2 3(1−υ 2 )(r 2 −3x 2 )

(3.4)

With εr the radial component of the applied strain and x the position of the FBG on the membrane.

3.2.3 Experimental Set-Up Y. Bmegaptche-Tekap, C. Korbuly and A. Kitouni The measurements were made in the S1 wind tunnel at INPT/IMFT on the Large Scale (LS) prototype in high-lift take-off configuration, see Fig. 3.14. The two following flow parameters have been varied: (1) The Reynolds number;

Fig. 3.14 Large scale prototype in S1 wind tunnel at the IMFT: flap detached from the wing translated with 8–10° of incidence

58

A. Giraud et al.

(2) The angle of attack. Two sensors have been adapted on the wing surface at 59.6 and 72% of the cord, see Fig. 3.15. The turbulent vortices related to the pressure fluctuations are mainly located downstream of the trailing edge, see Fig. 3.16. They entail turbulences downstream of the tailing edge in the wake and around the rear part of the wing.

Fig. 3.15 View of the Sensors adapted on the wing’s surface. It is worth noticing that no cavities are needed to locate the sensors as in conventional systems, thus highly facilitating the design

59.6% of the cord 72% of the cord

Fig. 3.16 View of the turbulence structure around the LS prototype after the High-Fidelity numerical simulations of INPT/IMFT presented in Chap. 4, [4]

3 Large Scale Morphing Prototype: Design and Experiments

59

3.2.4 Reynolds Number Variation A. Kitouni, Y. Bmegaptche-Tekap, M. Carvalho, A. Polo-Dominguez and A. Marouf The sensors have been able to measure the pressure fluctuations (rms signals) from which the Power Spectral Density (PSD) has been evaluated versus the significant frequency range. In the following, the angle of attack has been constant and the Reynolds number varied. The Von Kármán instability mode associated with the alternating vortices in the wake (see following Fig. 3.18) yields a first predominant frequency peak f 1 evaluated at 16 Hz, see Figs. 3.17 and 3.19. The results shown in Fig. 3.20 have been obtained by the numerical simulations (Chap. 4) at the same Reynolds number and angle of attack and clearly show the red and blue shear layers formed around and downstream of the high-lift flap, which are subjected to undulations due the to the shear-layer KelvinHelmholtz instability. The sensing system captured these shear layer frequencies f 2 , f 3 and f. The existence of these predominant frequency peaks whose ratios are incommensurate—e.g. not an entire number, produce a multitude of new frequencies in the spectra, being linear combinations of the mentioned one and translating the nonlinear interaction among all these instability modes. Therefore, the present sensing system has proven quite sensitive in capturing these important features which will be drastically modified and attenuated by the morphing as shown in Chaps. 4 and 5. An important aspect is that the numerical simulations allowed understanding of the physical phenomena governing the aerodynamic performances, prior to the experiments and therefore guided the experiments in strong synergy in the SMS project. It is therefore emphasized that for all the work-packages there is a strong synergy among them and that most often the experiments were driven by the simulations thus achieving the optimal aerodynamic performances.

Fig. 3.17 Spectra at 59.6% of the cord at an angle of attack of 4° for a Reynolds number of 2.2 and 3 million

60

A. Giraud et al.

Fig. 3.18 Shear layers structure and undulation forming the coherent vortices after KelvinHelmholtz instability, leading farther downstream to the von Karman vortices in the wake and relation with the captured predominant frequencies. Results from the numerical simulations by INPT/IMFT (Chap. 5), [4]

Therefore, the developed innovative sensing system presented here was able to capture the organised frequency modes due to the coherent vortices shedding (Fig. 3.19). Each of these vortex rows (see Fig. 3.20 for their 3D representation) are shed downstream and create by feedback regular pressure fluctuations that are captured by the present sensors system.

Fig. 3.19 Spectrum at 72% of the cord at an angle of attack of 4° for a Reynolds of 3 million

3 Large Scale Morphing Prototype: Design and Experiments

61

Fig. 3.20 3D dynamics of the turbulence structure and of the coherent vortex rows around the LS prototype, Results from High-Fidelity numerical simulations by INPT/IMFT (Chap. 4), [4]

3.2.5 Angle of Attack Variation Figures 3.21 and 3.22 present the PSD for the angles of incidence 0, 4 and 8°. The increase of the predominant frequency peaks amplitude is obtained with the increase of the incidence, as expected and shown in these figures.

3.2.6 Synthesis The predominant frequency values are presented in Tables 3.7 and 3.8 at 59.6% of the cord and 72% of the cord respectively. At 72% of the cord: Thus, when the angle of attack and the Reynolds number are increased, more predominant frequencies are distinguished in the spectrum. It is especially relevant in Tables 3.7 and 3.8 for an angle of attack of 8° and a Reynolds number of 3 million.

3.2.7 Comparison of the Results with the Numerical Simulations Figure 3.23 compares signals and PSD of pressure obtained from numerical simulations (Chap. 4) and experimental results. A good agreement in capturing the predominant frequencies is obtained.

62

A. Giraud et al.

Fig. 3.21 Spectra at 59.6% of the cord for a Reynolds number of 3 million at angles of attack of 0, 4 and 8°

3.2.8 Multi-point Sensing The present sensing system allows simultaneous measurement of the pressure in multiple points, see Fig. 3.24. As previously, the predominant frequencies are well captured as previously (Fig. 3.25). The simultaneous multi-point measurements enable the control loop with higher efficiency and robustness. Furthermore, the CEMENTYS system is not intrusive and can be applied on the lifting surface without the need to drill small cavities to embed the conventional pressure transducers. This highly facilitates the design and construction of the prototypes and creates a powerful sensing system.

3.2.9 Conclusion—Sensing System • CEMENTYS designed optical pressure sensors based on Fiber Bragg Gratings. • Successful measurements have been realised in the S1 wind tunnel of INPT/IMFT with analysis of the turbulence spectrum for:

3 Large Scale Morphing Prototype: Design and Experiments

63

Fig. 3.22 Spectra at 72% of the cord for a Reynolds number of 3 million at angles of attack of 0, 4 and 8° Table 3.7 Frequency values at 59.6% of the cord Angle of attack (◦ )

4

8

2.2 × 106

( f1 , f2 , f3 ) = (16 Hz, 35 Hz, 50 Hz)

( f1 , f2 , f3 , f4 ) = (16 Hz, 35 Hz, 50 Hz, 84 Hz)

3 × 106

( f1 , f2 , f3 ) = (16 Hz, 35 Hz, 50 Hz)

( f1 , f2 , f3 ) = (15 Hz, 30 Hz, 120 Hz)

Reynolds number

Table 3.8 Frequency values at 72% of the cord Angle of attack (◦ )

4

8

2.2 × 106

/

( f 1 , f 2 ) = (14 Hz, 35 Hz)

3 × 106

f 1 = 31 Hz

( f 1 , f 2 , f 3 , f 4 , f 5 ) = (14 Hz, 31 Hz, 54 Hz, 72 Hz, 131 Hz)

Reynolds number

64

A. Giraud et al.

Fig. 3.23 Left: Numerical simulations in Chap. 4 (WP4) by INPT/IMFT. Right: Experimental results obtained by the present pressure sensors

Fig. 3.24 Sensors Positions over the wing and flap—multi-point sensing

– Reynolds number variation – Angle of attack variation • With the present sensing system, consistent results have been obtained with the numerical simulations of INPT/IMFT carried out in Chap. 4 (WP4). • Multi-point sensing on the wing allowed monitoring of the turbulence special variation.

3 Large Scale Morphing Prototype: Design and Experiments

65

Fig. 3.25 Spectrum at 77.66% of the cord (left) and at 97.6% (right)

3.3 Controller Hardware Construction M. Scheller and J. Scheller In order to reliably guarantee the timings a proprietary high-frequency controller solution was considered as the basis for the acquisition of the high-frequency pressure and force signals as well as the control of the high-frequency actuators. Three main high-frequency controllers were evaluated after a detailed investigation of properties in a large variety of equipments (see detailed description in: http://smartwing.org/ SMS/EU/DOCUMENTS/STN-SMS-Controller-Hardware-and-Interface.pdf): • Opal-RTs OP4510 platform • DSpace Microlabbox • DSpace Scalexio A summary of the performance of these controllers i.e. the number of IOs and their respective velocity is provided in Table 3.9. Wherein all three controllers have the capacity to fulfil the high-frequency timing requirements they are mainly differentiated by the total number of available IOs (16 Analog in Opal RT, 32 Analog ins Microlabbox, 25 Analog Ins Scalexio), the timings and the resolution. The larger number of analog inputs, the capacity to achieve the desired timings as well as the availability finally made the Microlabbox the final choice. A closer look at the available interfaces shows that for interactions with the environment the Microlabbox allows for interfacing via CANbus, Ethernet, RS422 or RS232. Hence, in order to account for the large number of actuators and sensors in the low-frequency controller section it was proposed to use the Microlabbox as the main HF-controller and interface a number of n× LF-controllers using a standardized bus interface.

66

A. Giraud et al.

Table 3.9 Summary of the controller hardware characteristics Analog in

OPAL OP4510

Dspace Microlabbox

DSpace ScalexIO

16 channels, 16 bits, 2.5 microsecond conversion time for all channels simultaneously, ± 10 V true differential input

8 14-bit channels, 10 Msps, differential; functionality: free running mode, ± 10V 24 16-bit channels, 1 Msps, differentia;

25 14-bit channels, 4 Msps, ± 5V or ± 30 V

Analog out 16 channels, 16 bits, 1.0 microsecond update time for all channels simultaneously, ± 10 V, 15 mA Digital in

Digital out

16 16-bit channels, 1 25 14-bit channels, 7.8 Msps, settling time: 1 µs, Msps, ± 10 V ± 10V

32 channels, 4 V to 50 V, 3.5 mA min, 110 ns typical propagation delay, galvanic isolation with fast Opto-couplers

48 bidirectional channels, 2.5/3.3/5 V (single-ended); functionality: bit I/O, PWM generation and 32 channels, push-pull, 65 measurement (10 ns resolution), pulse ns typical propagation generation and delay, 5V to 30 V adjustable by an external measurement (10 ns voltage supplied by users resolution), 4 x SPI Master 12 bidirectional channels (RS422/485 type) to connect sensors with differential interfaces

50 bidirectional channels, output voltage either 3.3 V or 5.0 V, input voltage selectable from 1 V to 7.5 V

Hence, in summary, the conceptual structure both for the controller interface of the subsonic reduced scale (sRS) and the large-scale prototype (LS) relative to Chaps. 2 and 3 was developed. A proprietary high-frequency controller was selected in order to account for the elevated timing constraints of the high-frequency actuators and sensors. Since, the number of IOs of the Microlabbox is sufficient to support the reduced scale prototype the focus was put on how the controller IF could support the elevated number of IOs of the large-scale prototype. In order to account for this requirement a structure as in Figs. 3.26 and 3.27 was envisioned with a main HF-controller which interfaced with a number of n× LF-controllers.

3.3.1 Large-Scale—Controller Interface (iF) Based on the initial specification of the controller interface developed in accordance with the partners, the focus of this work was in the development of the LF-controller modules which are to interface with the HF-controller. The main tasks carried out can be summarized as follows:

3 Large Scale Morphing Prototype: Design and Experiments

67

Fig. 3.26 Unit tests

• WP3.4.1: Structural design • WP3.4.2: Unit tests for temperature acquisition & Unit tests for CAN-BUS single node and multi node communication • WP3.4.3: Unit Integration • WP3.4.4: Test In the following parts a short overview of the different sub-tasks will be provided.

3.3.1.1

Structural Design of the Controller

The general structure of the LF-controller modules is illustrated in Fig. 3.30. The structure was driven by the partner’s input and output specification for the LS prototype which is for completeness provided in Tables 3.10 and 3.11: As can be seen a total of 180 sensors is to be used. A total of maximum 60 outputs are needed as per the output specification of the LS-prototype given in Table 3.10. In order to meet these needs an ATmega2560 microcontroller was selected as a basis of on LF-controller cell as is shown in Fig. 3.28. Each LF-controller cell is provided with 8 temperature inputs, 8 general purpose analog inputs and 8 PWM outputs. Furthermore, in order to interface the individual controller cells a CAN-Bus interface is provided. The temperature inputs and the CAN bus are provided using SPI attached MAX6675 K-Thermocouple to-Digital Converter and an MCP2515 Stand-Alone CAN Controller. The selected structure permits a maximum of flexibility allowing to add inputs and outputs on the fly by adding further LF-controller modules to the CAN-Bus.

68

Fig. 3.27 Structural design of the HF Controller hardware

A. Giraud et al.

3 Large Scale Morphing Prototype: Design and Experiments

69

Table 3.10 Input specification LS prototype Level

Flexibility

Sampling frequency

10S/s

To be defined

Position feedback (angular)

20 sensors (1 per articulation per macro-actuator)

Specification Actuator feedback (per actuator)

Specific to SMA flap Temperature sensors

40 to 80 (1 or 2 per actuator)

Force sensor

40 to 80 (1 or 2 per actuator)

Specific to NOVATEM solution –

Only position feedback needed

Table 3.11 Output specification LS prototype Specification

Level

Flexibility

Emergency stop

Logic output used to cut the power by software

Mandatory

Sampling frequency

10S/s

To be defined

Position reference (angular)

20 outputs

Actuator control (per actuator)

Specific to SMA flap Heating orders

40 (1actuator)

Specific to NOVATEM solution Position references or current references

20 analog outputs

Fig. 3.28 LF controller module structure

70

3.3.1.2

A. Giraud et al.

Unit Tests for Temperature Acquisition & Unit Tests for CAN-BUS Single Node and Multi Node Communication

In order to verify the interaction between different LF controller modules using the provided CAN-Bus interface and in order to verify the attached MAX6675 temperature sensors unit-tests were conducted. The structure of these tests was relatively simple in order to verify the designed code. In Fig. 3.29 we can see the single-unit test setup for temperature acquisition. The left side shows the test setup while the right side of Fig. 3.29 shows the result of the temperature acquisition. In a first step a single thermocouple was used for the temperature acquisition and in a second step the capacity for acquiring the data from multiple thermocouples was evaluated. Since both unit tests were successful in a second step the CAN Bus data transfer was evaluated. The focus of the unit test was to setup and evaluate CAN communication as such in a first step a single unit test was conducted and once data transfer via CAN Bus was achieved a multi node test was conducted in order to verify data integrity with multiple senders and receivers at varying transfer speeds. The main structure of this multi-node tests whose test setup can be seen in Fig. 3.30 was to use senders and receivers in interrupt mode. In other words, an interrupt was triggered every time data was received. One master sender was used with a node ID of 0x00 and every receiving node (3 in total) was forwarding the message from the master sender. As such each receiving node received data from at least two nodes (0x00 0x1n where n is the previous node ID). The messages were verified by checking the IDs of the sender with the expected number of messages from this sender. Different transfer rates up to a maximum transfer rate of 125kbps were tested without data loss.

Fig. 3.29 K-type thermocouple single-unit test

3 Large Scale Morphing Prototype: Design and Experiments

71

Fig. 3.30 CAN-Bus multi node test

3.3.1.3

Unit Integration

The verified modules were integrated into a single LF-controller unit shown in Fig. 3.31. In order to handle the data transfer and between the HF-controller and the LFcontroller nodes a simple state machine was developed which runs on every LFcontroller node. This state machine is shown in Fig. 3.32 and as can be seen contains a total of four states. Independent of the state temperature, position and force data are acquired at every iteration of the program. Furthermore, at every iteration it is checked whether new CAN messages have been received. Upon power-on of the nodes each node will send a can message to 0x02 with its own node id i.e. OWN_ID which can be defined as shown in the parameters Sect. 3.4.1.3.1. The node will then jump to the init state.

Fig. 3.31 LF-controller unit

72

A. Giraud et al.

READ temp1..N READ pos1…N READ force1…N CAN:check:NewMessage CAN:get:STA1 CAN:send:STA1

CAN:send:ID

init

OUTPUT PWM1…N = 0

ID+0x80

setup

idle

OUTPUT PWM1…N = 0 CONTROL LOOP if CAN:get:TEMP||AIN1||AIN2(#CHAN) CAN:send:TEMP||AIN1||AIN2 ID+0xn0 (#CHAN) ID+0xn0 If CAN:get:PWM0(#CHAN)(#VAL) OUTPUT PWM(#CHAN) = #VAL CAN:get:PWM0(#CHAN)(#VAL) ID+0x70

WA CA TCH N: DO ER G: R0 tim eo ut ||

loop

OUTPUT PWM1…N = 0 CAN:send:error

Fig. 3.32 LF-Controller state machine

The default state is init. If an error has occurred, or the LF-controllers have just been powered on in order to provide input/output control the system first has to transferred into the loop state from the init state by sending the “STA1” command to the nodes to be activated. In the loop state input output control can be performed by sending appropriate CAN messages to the individual nodes. In short, the temperature, force and position data acquired by the nodes can be transmitted via CAN bus and similarly the PWM outputs of the nodes can be set via CAN. A more detailed overview of the individual CAN messages can be found in the Sect. 3.4.1.3.3. The loop state is either exited when error message is received via CAN (ERR0) or if the software watchdog times out. This will automatically also send an error message via CAN and reset the PWM outputs to 0. A couple of parameters are defined in order to control the functionality of the program which will briefly be discussed in the following section. Followed by this overview of the system parameters, the function of the software watchdog is briefly explained and finally the CAN messages which can be send to the individual nodes.

Parameters The following parameters are specified:

#define USE_TEMP #define USE_POSITION

3 Large Scale Morphing Prototype: Design and Experiments

73

#define USE_FORCE #define USE_PWM The USE_TEMP, USE_POSITION, USE_FORCE and USE_PWM parameters essentially activate and deactivate the acquisition of temperature and analog input (position and force) data as well as the PWM output. Commenting any of these lines will result in the deactivation of the corresponding acquisition or output. #define WATCHDOG_TIMEOUT 60 The WATCHDOG_TIMEOUT parameter is the amount of time in seconds before an error is triggered because no CAN messages have been received in the loop state.

#define OWN_ID 0x11 The OWN_ID is the id of the node which will be used when this specific node is addressed by the master and also to differentiate between the nodes when receiving position, force or temperature data. The default ID of the master is 0x02, command messages to be received by everyone have the ID of 0x01 and as will be explained below for error messages the ID is 0x00 having thereby the highest priority. Every node only receives messages from three ids (0x00, 0x01 and OWN_ID) thereby compartmentalizing commands and providing efficient filtering.

#define NUM_CHANNELS 8 The NUM_CHANNELS parameter sets the number of channels i.e. the number of inputs (temperature, position and force) and outputs (PWM) used. The maximum number of channels is 8 and minimum of 1 channel should be selected.

#define BUFFER_SIZE 8 #define BUFFER_SIZE_SEND 8 × 4 The BUFFER_SIZE and the BUFFER_SIZE_SEND parameters define the byte size of the send and receive buffer. As the biggest data type to be send is the temperature value stored in a double (4 Byte) the send buffer consequently has a maximum size of 32 Byte (i.e. for a maximum of 8 channels).

74

A. Giraud et al.

Watchdog As previously explained the software watchdog is only active in the loop state as it is the only state where the PWM output can be set. The function of the watchdog is simple every time a CAN message is received an internal timer gets the current time and stores it in an internal variable (time_last_message) and at every iteration this internal variable is compared to the current time and if the difference is larger than the WATCHDOG_TIMEOUT parameter the system switches from the loop state to the error state. Evidently, every time a CAN message is received the internal variable (time_last_message) is updated.

CAN Messages In order to enable communication between the host and the LF-Controller nodes a simple can messaging protocol has been implemented. In short, the LF-Controller nodes are able to differentiate between 6 different message types: ‘ERR0’, ‘STA1’, ‘PWM0’, ‘TEMP’, ‘AIN1’, ‘AIN2’. A short summary of the function of the different message types is provided in Table 3.12. A more detailed description will follow in the subsequent sections. Depending on the type of message, either two, one or no arguments need to be transmitted with the command. Below are the explanations of the abbreviated messages: ERRO This message is either send when encountering an error such as the software timeout running out or received when another node or the master encountered an error. The ID of this message is always 0x00 as it thereby has the highest priority. Receiving this message will directly move the receiving node from the loop to the error (idle) state and subsequently into the start (init) state and setting all PWM outputs to 0 thereby ensuring safety in case of fault. STA1 This message is send by the master in order to indicate the start of the processing i.e. the move from the startup state (init) to the loop state it has to be send after power on or after recovery from an error state. No parameters are needed. Upon receiving the STA1 command the node will mirror the command send as a response to Can ID: OWN_ID + 0x80 i.e. their own ID offset by 0x80 thereby allowing the differentiation between different nodes and verification which nodes are active. PWM0 In order to set the PWM outputs a CAN message with the PWM0 command has to be send. Following the command two parameters are needed in the same message,

3 Large Scale Morphing Prototype: Design and Experiments

75

Table 3.12 CAN messages summary Message Argument

Description

ERR0

–

Error encountered by one of the nodes or the host

STA1

–

Start signal to initiate processing i.e. jump from state init to setup and subsequently loop

PWM0

ARG0: #Chan Set the PWM output value via VAL from 0 to 255 corresponding to the duty cycle increments. The values can be set for one channel by ARG1: Val specifying the channel number (0–7) or for all channels by sending (0xF) for the first argument of the message

TEMP

ARG0: #Chan Get the measured temperatures for either one channel by specifying the channel number (0–7) or for all channels by sending (0xF) for ARG1: – the first argument of the message Please note that the response of the message can be a multi-frame message (details below)

AIN1

ARG0: #Chan Get the measured analog input values for the first 8 analog input channels for either one channel by specifying the channel number ARG1: – (0–7) or for all channels by sending (0xF) for the first argument of the message Please note that the response of the message can be a multi-frame message (details below)

AIN2

ARG0: #Chan Get the measured analog input values for the second 8 analog input channels for either one channel by specifying the channel number ARG1: – (0–7) or for all channels by sending (0xF) for the first argument of the message Please note that the response of the message can be a multi-frame message (details below)

Table 3.13 PWM parameters # Parameter Values 1 Channel

0x0 – 0x7 or 0xF 0x0 – 0x7 correspond to channels 0–7 and 0xF sets the PWM output for all channels

2 Value

0–255

255 corresponds to a 100% duty cycle and 0 corresponds to a 0% duty cyle

the channel number to be set and the value the channel is to be set to. The parameters are once again shown in Table 3.13 along with the expected values and a brief explanation. Upon receiving the PWM0 command and setting the PWM outputs the node will mirror the command send as a response to Can ID: OWN_ID + 0x70 i.e. their own ID offset by 0x80 thereby allowing the differentiation between different nodes. TEMP In order to acquire the temperature measurements a CAN message with the TEMP command has to be send. Following the command one parameter is needed in the

76

A. Giraud et al.

Table 3.14 TEMP parameters #

Parameter

Values

1

Channel

0x0 – 0x7 or 0xF

0x0 – 0x7 correspond to channels 0–7 and 0xF acquires the TEMP input for all channels

Fig. 3.33 Illustration of a response send by a LF-Controller node upon receipt of a TEMP command

same message, the channel number to be acquired. The table here below quickly summarizes the parameter and its expected values (Table 3.14). Upon receiving the TEMP command for a single node or for all nodes the node/s will respond by sending a message containing the requested temperature values. Each value has a length of four bytes and as such a total of 32 bytes is send via CAN if the temperature value is requested for all channels (maximum 8) of a node. Since CAN messages contain a maximum of 8 bytes the transmitted data if exceeding this limit is split in multiple messages. In order to reassemble the complete data message the first byte of the data package is used in order to indicate the amount of messages still to be received i.e. if the last message has been received this byte is going to be 0, if 1 message is remaining the byte is going to be 1, if 2 messages are remaining the byte is going to be set to 2 and so on. To facilitate differentiation between the nodes if temperature values a requested for multiple nodes the nodes send the response message to the ID being OWN_ID + 0x10 i.e. their own ID offset by 0x10 thereby allowing the differentiation between different nodes. To better illustrate this concept the following Fig. 3.33 illustrates CAN data frame send by the nodes upon receipt of a TEMP command via CAN. Please note that only the identifier and the data section of the data frame are manipulated the remaining fields correspond to standard CAN data frames. AIN1 In order to acquire the analog input measurements of the first 8 analog input channels (A0-A7) a CAN message with the AIN1 command has to be send. Following the command one parameter is needed in the same message, the channel number to be

3 Large Scale Morphing Prototype: Design and Experiments

77

Table 3.15 AIN1 parameters #

Parameter

Values

1

Channel

0x0 – 0x7 or 0xF

0x0 – 0x7 correspond to channels 0–7 and 0xF acquires the TEMP input for all channels

Fig. 3.34 Illustration of a response send by a LF-Controller node upon receipt of a AIN1 command

acquired. The table here below quickly summarizes the parameter and its expected values (Table 3.15). Upon receiving the AIN1 command for a single node or for all nodes the node/s will respond by sending a message containing the requested analog input values. Each value has a length of two bytes and as such a total of 16 bytes is send via CAN if the analog input value is requested for all channels (maximum 8) of a node. Since CAN messages contain a maximum of 8 bytes the transmitted data if exceeding this limit is split in multiple messages. In order to reassemble the complete data message the first byte of the data package is used in order to indicate the amount of messages still to be received i.e. if the last message has been received this byte is going to be 0, if 1 message is remaining the byte is going to be 1, if 2 messages are remaining the byte is going to be set to 2 and so on. To facilitate differentiation between the nodes if temperature values a requested for multiple nodes the nodes send the response message to the ID being OWN_ID + 0x20 i.e. their own ID offset by 0x20 thereby allowing the differentiation between different nodes. To better illustrate this concept the following Fig. 3.34 illustrates CAN data frame send by the nodes upon receipt of a AIN1 command via CAN. Please note that only the identifier and the data section of the data frame are manipulated the remaining fields correspond to standard CAN data frames. AIN2 In order to acquire the analog input measurements of the second 8 analog input channels (A8-A15) a CAN message with the AIN2 command has to be send. Following the command one parameter is needed in the same message, the channel number to

78

A. Giraud et al.

Fig. 3.35 Illustration of a response send by a LF-Controller node upon receipt of a TEMP command

be acquired. The table here below quickly summarizes the parameter and its expected values. # Parameter Values 1 Channel

0x0 − 0x7 or 0xF 0x0 – 0x7 correspond to channels 0–7 and 0xF acquires the TEMP input for all channels

Upon receiving the AIN2 command for a single node or for all nodes the node/s will respond by sending a message containing the requested analog input values. Each value has a length of two bytes and as such a total of 16 bytes is send via CAN if the analog input value is requested for all channels (maximum 8) of a node. Since CAN messages contain a maximum of 8 bytes the transmitted data if exceeding this limit is split in multiple messages. In order to reassemble the complete data message the first byte of the data package is used in order to indicate the amount of messages still to be received i.e. if the last message has been received this byte is going to be 0, if 1 message is remaining the byte is going to be 1, if 2 messages are remaining the byte is going to be set to 2 and so on. To facilitate differentiation between the nodes if temperature values a requested for multiple nodes the nodes send the response message to the ID being OWN_ID + 0x30 i.e. their own ID offset by 0x30 thereby allowing the differentiation between different nodes. To better illustrate this concept the following Fig. 3.35 illustrates CAN data frame send by the nodes upon receipt of a AIN2 command via CAN. Please note that only the identifier and the data section of the data frame are manipulated the remaining fields correspond to standard CAN data frames.

3.3.1.4

Example of a Workflow

In the following Fig. 3.36 an example workflow is illustrated for the acquisition of the analog input values of the first 8 analog inputs (A0-A7) of a LF-controller

3 Large Scale Morphing Prototype: Design and Experiments

79

Fig. 3.36 Example workflow illustration for acquiring analog input data from a LF-controller node

node by the master in the form of the messages exchange between the master and a LF-controller node. As can be seen in the illustration the master requests the analog input data of node 0x11 by sending a message to ID 0x11 whose data package contains the AIN1 command as well as 0xF in order to indicate to the node that the analog input data of all channels is requested. Hence, the total length of the data package send to the node is 5 bytes (byte 0: A, byte 1: I, byte 2: N, byte 3: 1, byte 4: F). Upon receipt of the command the node will respond in the present case by sending the latest acquired temperature data via CAN in the form of a split message wherein the first byte is always the number of messages remaining to be send.

3.3.1.5

Test of Robustness and Efficiency

In order to evaluate the performance of the above described controllers a test-setup was developed which allowed to test individual controller nodes as well as the ensemble of controllers. The test setup which is schematically shown in Fig. 3.37 consisted of a group of N-nodes (four of them are shown schematically in the illustration) commanded via a master controller in the form of a Ubuntu 16.04 LTS running PC interfacing

80

A. Giraud et al.

Fig. 3.37 LF-controller CAN network overview

with the CAN Network using a USBtin (https://www.fischl.de/usbtin/) USB to CAN interface. A dedicated test interface using Python which permits to send and receive the above described CAN packets. A screenshot of the main interface can be seen in Fig. 3.38a. Using this test-program the individual nodes were verified including their capacity to provide temperature acquisition and analog data acquisition. Furthermore, the commanding of the individual and global PWM commands were tested. A maximum speed command speed of 125 kbps was tested as this was deemed sufficient in line with the input and output specifications shown in Tables 3.2 and 3.3. Further tests included the verification of the watchdog on the individual nodes and error handling mechanisms previously described. All tests showed that no data loss occurred and that error handling and watchdog mechanisms performed as desired. A detailed documentation of the computer system handling the controller is available by STN in: http://smartwing.org/SMS/EU/DOCUMENTS/STN.SMS.LF-controller-doc umentation.pdf.

3.3.2 Conclusion—Hardware Controller and Interface During the course of this project a controller HW solution was developed which allowed at the same time to deterministically control a limited number of highfrequency actuators while allowing the versatility to at the same time control a large number of lower frequency inputs and outputs. This was achieved by a hybrid approach between a DSpace high-frequency controller interfaced using CAN with a group of individual lower frequency controllers which were based on the ATMega

3 Large Scale Morphing Prototype: Design and Experiments

81

Fig. 3.38 Controller network test program (a) and extension of the architecture to the full wing’global integration

2560 and each ATMega 2560 did control a group of 8 temperature measurement nodes, 8 analog inputs and 8 PWM outputs. An interface protocol between different nodes of the bus was developed and the bus system was tested using a custom-built test program which allowed to interface a Linux PC with the CAN network. Tests showed the performance of the network as well as its capacity to provide data acquisition and control of the outputs at bus speeds up to 125 kbps. This solution hence allows interfacing both to adhere to the strict high-frequency requirements provided by the different partners while allowing to individually control a large number of lower frequency inputs and outputs. The selected CAN bus approach provides significant flexibility as it allows the simple addition and removal of controllers and thereby a simple subdivision of the to be controller LS prototype.

3.4 Experimental Verification—INPT/IMFT-LAPLACE Y. Bmegaptche-Tekap, D. Harribey, M. Carvalho, A. Polo-Dominguez and A. Marouf The LS prototype’s high lift flap is presented in Figs. 3.39 and 3.40. It has 2 m span and 1m chord length. It has been adapted in the fixed wing’s configuration of 2m span detailed in the RP1 report. The overall configuration of wing + high lift flap has been installed in the S1 wind tunnel of INPT/IMFT, whose test section is of 2.40 m (Fig. 3.41). The results in this section concern the take-off configuration. The wing and the flap have been equipped with a considerable number of pressure tabs for the mean pressure measurements. The system is shown in Fig. 3.42.

82

A. Giraud et al.

Fig. 3.39 Schematic view of the high-lift flap LS prototype structure

Fig. 3.40 Photo of one section of the high lift flap under construction in the Workshop of INPT/IMFT, in collaboration IMFT-LAPLACE

3.4.1 Mean Pressure Measurements The mean wall pressure around the flap has been measured by pressure taps at Reynolds numbers 2.2M and 2.7M. Results of pressure coefficient measurements are shown in Figs. 3.43 and 3.44. There is a reasonable agreement between the

3 Large Scale Morphing Prototype: Design and Experiments

83

Fig. 3.41 View of the fixed wing (left) and of the overall wing + flap installed in the S1 wind tunnel of INPT/IMFT

Fig. 3.42 Mean pressure measurements acquisition system

experiments and the simulations carried out in WP4-Chap. 4, given the fact that this part of the simulations have been done in 2D.

3.4.2 Unsteady Pressure Measurements The turbulent flow presents significant variations in pressure. As a consequence, it is possible to characterise the nature of the turbulence from a pressure signal that contains the information of these fluctuations. For this purpose, a pressure sensor has been tested in the facilities of INPT/IMFT, Fig. 3.45. The selected sensor was a piezoresistive pressure transducer, the model 8507C-1 by Endevco. This sensor was placed at 72% of the flap chord (x/c = 0.72), close to the trailing edge where a turbulent flow is expected. Several tests have been carried out in

84

A. Giraud et al.

Fig. 3.43 Mean pressure coefficient, clean configuration at Re = 1M at 10° angle of attack

Fig. 3.44 Pressure coefficients at Re 2.2M, take-off configuration at different angles of attack of the upstream fixed wing part

3 Large Scale Morphing Prototype: Design and Experiments

85

Fig. 3.45 Meggitt pressure sensor (left) and overview of the measuring section (right)

the wind-tunnel S1 of the IMFT, always with acquisition frequencies up to 20,000 Hz, far enough from the resonance frequency of the transducer (55,000 Hz). This open return wind-tunnel has an open test section with a circular diameter of 2.40 m, with a velocity range of 1–38 m/s. Measurements are carried out at ambient temperatures (22 °C). Previous studies ([5, 6]) as shown the importance of studying the power spectral density of the pressure to search the gains of the morphing wing. The method used to obtain the spectra is the Welch method, carried out by dividing the time signal into successive blocks, forming the periodogram for each block, and averaging. Different parameters could be changed in this method, like the number of points of each window or block, the overlap between consecutive windows, or the number of points used to calculate each periodogram. In the current spectra, Fig. 3.46, the tests have a duration of 10 min with a frequency acquisition of 10,000 Hz (6 Million samples) and a cut-off frequency of 3000 Hz according to the Nyquist-Shannon criterion. The configuration of interest is the take-off, that involves a flap deflection angle of 10° and an angle of attack (AoA) of 8° with Reynolds number of 2.2 Million. The blue spectra represent the take-off configuration at a Reynolds number of 2.2M. The orange-colour spectrum corresponds to the take-off configuration at Reynolds of 2.7M. Both spectra have a similar behaviour, but also present differences. The energy of the signal is higher for the case at Reynolds number of 2.7M. The higher level of the spectrum corresponding to Reynolds number 2.7M was expected because of the turbulence level increase as Reynolds number increases. It is illustrated that both spectra display similar predominant frequency bumps. This is due to the flow instabilities being of an absolute character in the present low subsonic Reynolds number range. Among the principal instabilities, the Von Kármán mode characterises the vortex dynamics in the wake and it is measured by the present spectra. It corresponds to the frequency bump indicated in Fig. 3.47b the fact that this mode appears as a bump and not as a distinct predominant peak is due to the chaotic turbulent eddies which create a smearing of the coherent alternating Von Kármán eddies. The predominant frequency at maximum amplitude in this bump is of order 17–19 Hz. This corresponds to a Strouhal number of 3.8–4.3. Furthermore, previous numerical

86

A. Giraud et al.

Fig. 3.46 a Pressure spectrum at Re 2.2M (blue) and Re 2.7M (orange); b Peaks of predominant frequencies shown on the blue spectrum (as in left figure), at Re 2.2M (fVK = 17–19 Hz, f2 = 30 Hz, f3 = 110 Hz)

and experimental studies ([5, 6]) has been proved the existence of the Von Kármán instability at low frequencies. A comparison between experimental spectra and numerical spectra has been realised. Numerical simulations were carried out with the Navier-Stokes Multi-Block (NSMB) code, which solves the compressible form of the Navier-Stokes equations. A two-dimensional Multi-Block structured grid was built, presented in Fig. 3.47a, c. The grid captures quite well the real physics around the wing, the flap and the wake, see Fig. 3.47b where the Mach field contours are shown. The numerical signal has a duration of 2 s, obtained with a numerical sampling rate of 100,000 Hz (200,000 samples) at a Reynolds Number of 2.2 M. The same incidence angles of the take-off configuration from the experiments has been adopted. Figure 3.48 shows the numerical spectrum at 72% of the flap’s chord (x/c = 0.72), with the previous experimental spectra at Reynolds number of 2.2 M. Both spectra are obtained from pressure signals and show notable similarities (Fig. 3.49). The von Kármán instability bump is visible at 17–22 Hz in a close position as in the experiment. It is recalled that the simulations have been carried out under a 2D approximation, whereas the physics in the experimental study are 3D. It is worth

(a)

(b)

(c)

Fig. 3.47 Multi-block structured grid for the two-element wing-flap large scale prototype

3 Large Scale Morphing Prototype: Design and Experiments

87

Fig. 3.48 Numerical and Experimental pressure spectrum at Re 2.2M (fVK = 17–22 Hz, f2 = 30 Hz, f3 = 110 Hz)

Fig. 3.49 Left: Simulations, right: experiments. f 1 : von Karman frequency f 2 , f 3 : shear layer frequencies. Sampling rate with Meggitt transducer: 10 kHz

88

A. Giraud et al.

mentioning that despite these differences in the approaches, the spectra show these similarities. The slope on the right part of the numerical spectrum is quite decreased than the experimental one because of the turbulence modelling assumptions and the 2D approximation. For the same reasons, the overall level of the spectral amplitudes in the numerical results are lower than in the experiment. However, the predominant frequency peaks and bumps due to the coherent vortices clearly appear in the same frequency ranges as in the experiments.

3.4.3 Conclusion—Experimental Verification In this section, wind tunnel evaluation results for the novel LS morphing high-lift wing-flap have been presented in static take-off configuration and validation with the numerical simulation results obtained through the NSMB code. The experimental results for the non-actuated configuration show a quite good agreement with the simulation. These results have proven the ability of the overall wing + high lift flap configuration to bear the aerodynamic loads under these dimensions near scale 1 and thus allow further experiments detailed in Chap. 5. The evaluation of the aerodynamic performances between the actuated (cambered) and non-actuated configuration are provided in the Chap. 5 concerning WP5.

References 1. M.M. Werneck, R.C. Allil, B.A. Ribeiro, F.V. de Nazaré, in A guide to Fiber Bragg grating sensors, Current Trends in Short- and Long-period Fiber Gratings (IntechOpen, 2013) 2. Z. Zhou, J. Ou, in Techniques of temperature compensation for FBG strain sensors used in longterm structural monitoring, Fundamental Problems of Optoelectronics and Microelectronics II (Khabrovsk, Russian Federation, 2004) 3. J.-L. Le Goër, J. Avril, Capteurs à jauges extensométriques. Techniques de l’ingénieur Mesures de longueurs et d’angles (1992) 4. A. Marouf, Analyse physique de concepts du morphing électroactif pour accroître les performances aérodynamiques des ailes du futur par simulation numérique de Haute Fidélité et modélisation de la Turbulence à nombre de Reynolds élevé, Ph.D. Thesis, Université de Strasbourg (2020) 5. G. Jodin, V. Motta, J. Scheller, E. Duhayon, C. Döll, J.F Rouchon, M. Braza, Dynamics of a hybrid morphing wing with active open loop vibrating trailing edge by time-resolved PIV and force measures. J. Fluids Struct. 74, 263–290 (2017). https://hal.archives-ouvertes.fr/hal-016 38290 6. N. Simiriotis, G. Jodin, A. Marouf, P. Elyakime, Y. Hoarau, J.C. Hunt, J.F. Rouchon, M. Braza, Morphing of a supercritical wing by means of trailing edge deformation and vibration at high Reynolds numbers: Experimental and numerical investigation. Fluids Struct. 91 (2019). https:// doi.org/10.1016/j.jfluidstructs.2019.06.016

Chapter 4

High-Fidelity Numerical Simulations A. Marouf, N. Simiriotis, J. B. Tô, Y. Hoarau, J. B. Vos, D. Charbonnier, A. Gehri, R. El Akoury, Y. Hoarau, F. Kramer, F. Thiele, K. Diakakis, M. Fragiadakis, J. L. Farges, T. Chaboud, G. Tzabiras, J. F. Rouchon, and M. Braza

Abstract In this chapter, the physical analysis and optimisation of the morphing behaviour have been performed by means of numerical High-Fidelity (Hi-Fi) fluid-structure interaction simulations. The most advanced turbulence modelling approaches in the context of CFDSM (Computational Fluid Dynamics Structural Mechanics) have been used. This effort had as main objective to analyse the optimal actuations and to explain why they provide a high aerodynamic performance. This work provided the optimal wing shapes as well as the optimal ranges of frequency vibrations, amplitudes, and actuators positions and accompanied the experiments in the context of a synergistic database among the numerical (Hi-Fi) and experimental results. This considerably reduced the design cycles and the final prototype experiments, by allowing targeted optimised sensing/morphing configurations. By means A. Marouf · N. Simiriotis · J. B. Tô · R. El Akoury · M. Braza INPT—Institut National Polytechnique de Toulouse, IMFT—Institut de Mécanique des Fluides de Toulouse, 2 Allée du Professeur Camille Soula, 31400 Toulouse, France A. Marouf · Y. Hoarau · Y. Hoarau Laboratoire Des Sciences de L’Ingénieur, ICUBE—Université de Strasbourg, de l’Informatique et de l’Imagerie, 4 Rue Blaise Pascal, 90032, 67081 Strasbourg, France J. B. Vos · D. Charbonnier · A. Gehri CFSE—Computational Fluids and Structures Engineering, EPFL Innovation Park—Building A, 1015 Lausanne, Switzerland F. Kramer · F. Thiele CFDB—Software Entwicklungs- und Forschungsgesellschaft mBH, CFD Berlin, Wolzogenstra β e 4, 14163 Berlin, Germany K. Diakakis · M. Fragiadakis · G. Tzabiras (B) NTUA—National Technical University of Athens, Zografou Campus, 9, Iroon Polytechniou Str, 15780 Zografou, Greece e-mail: [email protected] J. L. Farges · T. Chaboud ONERA—Centre Français de recherche Aérospatiale, 2 Avenue Edouard Belin, 31000 Toulouse, France J. F. Rouchon LAPLACE—Laboratoire Plasma et Conversion d’Energie, INPT—Institut National Polytechnique de Toulouse, Site of ENSEEIHT, 2, Rue Charles Camichel, 31071 Toulouse, France © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. Braza et al. (eds.), Smart Morphing and Sensing for Aeronautical Configurations, Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153, https://doi.org/10.1007/978-3-031-22580-2_4

89

90

A. Marouf et al.

of Adjoint-Based approaches, the optimal wing shapes have been derived, therefore ensuring a high efficiency of the final morphing prototypes, beyond the initially expected orders of magnitude. The main achievements are presented in four parts: sRS, tRS and LS prototypes, as well as for the full A320 aircraft with morphing wings. This last is an important added value, not initially previewed in the DoA, that proved the efficiency of the morphing wing concepts on the full aircraft.

Work Package 4—WP4

4.1 Morphing sRS Prototype—Numerical Simulations Concerning the sRS prototype, (take-off and landing phases), the principal optimal ranges concerning higher-frequency/small deformation of the near-trailing-edge region of the A320 wing (RS prototype) in low-subsonic speeds is provided below, in Fig. 4.1. In all flight phases, the aerodynamic performance increase, lift-todrag ratio has been ensured. Concerning the landing phase, essentially its last approaching phase, we need to increase the drag in order to reduce the airplane’s velocity. Concerning the take-off, we need to ensure a lift-to-drag increase and a drag reduction. The simulations by INPT/IMFT have been accomplished by means of the NSMB code and a quite efficient turbulence modelling approach, the Organised Eddy Simulation, OES [1–3], allowing the physical development of the instabilities and of the coherent vortices, important for their manipulation through the morphing. The simulations by NTUA have been carried out by the k-ω turbulence model [4], also quite efficient in capturing the vortex flow dynamics. A significant lift-to-drag increase has

Fig. 4.1 View of the piezo-actuators near the trailing edge along the span (left) and of the rear part deformation and vibration of this area (right) applied in the numerical simulations

4 High-Fidelity Numerical Simulations

91

Fig. 4.2 Significant aerodynamic performance increase (lift-to_drag ratio) thanks to the optimal vibration and slight deformation near the trailing edge region. NSMB solver results, INPT/IMFT

been obtained by only the piezoactuators vibration, leading to lift increase and simultaneous drag reduction (Fig. 4.2), in the actuation frequency ranges around 60 Hz and (250–300) Hz. Both are characterised by simultaneous mean drag reduction and rms reduction The second range is wider and is adopted for the experiments with the sRS prototype carried out in WP2-Chap. 2, thus pointing out the synergy between the simulations and the experiments and the fact that the simulations contributed to the efficiency of the final evaluation of the morphing performances in the SMS project. The rms of the forces versus the amplitude of the piezoactuators has been investigated in detail by means of the NSMB code (Fig. 4.3) and the structural solver MaPFlow of NTUA (Fig. 4.4). The range 100–300 Hz has been proven advantageous and have shown that there is no need to increase the vibration amplitude beyond a threshold, because this would increase the drag.

92

A. Marouf et al.

Fig. 4.3 Significant aerodynamic performance increase (lift-to_drag ratio) thanks to the optimal vibration and slight deformation near the trailing edge region. NSMB solver results, INPT/IMFT

Fig. 4.4 RMS of lift and drag versus amplitude of the piezoactuators—MaPFlow solver—NTUA, incidence 10°, Reynolds number 1 Million, sRS

4.2 Three-Dimensional Morphing Effects The Hi-Fi numerical simulations using the NSMB code and the efficient OES turbulence modelling allowed analysis of three-dimensional morphing effects on the coherent vortices issued from the main flow instabilities, the von Karman and shear layer modes as well as their spanwise predominant undulation according to well

4 High-Fidelity Numerical Simulations

93

Fig. 4.5 Schematic deformation and vibration of the trailing edge region, median spanwise section of the sRS prototype, spanwise length of 70 cm, the same as in the experiments of Chap. 2, chord of 70 cm, incidence 10°, Reynolds number 1 Million

distinct wavelengths. The deformation and motion of the rear part of the A320 wing (sRS prototype) is shown in Fig. 4.5. For the 3D simulations, the angle of attack is also 10° and Re = 1 Million, as for the 2D simulations and in the experiments presented in Chap. 2. The same spanwise length as in the experiments has been used for the 3D simulations. After quite detailed 3D simulations on parallel MPI supercomputing architectures using an order of 2000 parallel processors at the French Supercomputing Centres CINES, IDRIS and CALMIP, the dynamics of these coherent structures have been analysed. A significant modification of the 3D secondary instability has been obtained through an optimal actuation and slight deformation of the trailing edge region (Fig. 4.6). This morphing has been able to modify the axes ratio of the ellipses formed by the 2D cutoff section of the three-dimensional vortex rows and diminished this ration to under-critical thresholds. Based on the theory of elliptic instability [5–7], it is known that the secondary instability is amplified when this ratio becomes higher than a critical value. In the present case, by creating smaller-scale vortices thanks to the vibration, the eddy-blocking effect has been enhanced between the upper and lower shear layers past the trailing edge and upstream of it thanks to feedback effects. This eddy blocking has been able to constrict the shear layers through shear-sheltering and to diminish the wake’s width, and therefore the drag, as explained in the INPT/IMFT publications [8, 9], with simultaneous creation of lift. Thanks to these effects, the schematic ellipsis ratio of the von Karman vortices became under-critical and the secondary instability has been suppressed (Fig. 4.6). This morphing effect is quite significant because it suppresses the rms of the forces fluctuations: not only the mean values are interesting for the performances but essentially the rms too. The decrease of rms is enhanced by suppression of this three-dimensionality and especially of a main 3D vortex pattern, the vortex dislocations, [10], highly associated to the forces fluctuations. The vortex dislocations consist of a break of the “spinal column” of the 3D undulated coherent vortex rows. As seen in Fig. 4.6, the vortex dislocations have

94

A. Marouf et al.

VL

Fig. 4.6 Three-dimensional morphing effect on the coherent vortex rows: Figures on the left: no morphing. Figures on the right: morphing with vibration of 300 Hz: suppression of their 3D undulation and of the vortex dislocations (VL)

been suppressed by the 3D morphing, where all the patches vibrate with the same frequency and amplitude along the span of the wing. These effects have been consolidated and confirmed by carrying out simulations with a finer grid of 50 Million cells and using the hybrid turbulence modelling approach DDES-OES (Detached Eddy Simulation with embedded Organised Eddy Simulation), able to provide a more rich statistical content in respect of a variety of smaller-scale eddies as shown in the next paragraph for the LS prototype.

4.3 Performances of the Hybrid Morphing: Cambering + Actuation The cambering of the sRS has been obtained by the SMA actuation at low frequency (order of 1 Hz) and high deformation (order of 10–30% of the chord) as illustrated

4 High-Fidelity Numerical Simulations

95

in the experimental TRPIV fields presented in Chap. 2. Simultaneously, higher frequency vibrations of the near trailing edge region with slight deformation as presented in the previous figure act as the flight of large-span hunting birds which in addition to the large cambering of the main wing, they actuate their feathers according to their sensing system able to capture the distribution of the aerodynamic pressures. Thanks to this multi-scale actuation, they simultaneously obtain high lift/drag and reduction of noise, especially when they hasten to their pray. Therefore, in the SMS project, the hybrid electroactive morphing is partly bio-inspired. The term “partly” is used because the speeds of the wing on take-off and landing are higher than the ones developed by the birds. It is worth noticing that the optimal actuation frequency and amplitude values have been derived by the Hi-Fi simulations prior to the experiments. Therefore, it is emphasised that a strong synergy operated among the different tasks in the SMS project ensuring the best results. This synergy will be mentioned in all cases in the document. The physical parameters for the hybrid morphing are shown in the following Fig. 4.7. The flapping at 300 Hz in the hybrid morphing provides more energy to the shear layer than in case of SMA actuation only, thus leading to longer shear layers and to delay of the von Karman mode formation (Fig. 4.8) with high benefits in lift-to-drag ratio. The hybrid morphing for the sRS with actuation of 300% has led to a 15% lift-to drag ratio (aerodynamic performance) in the intermediate wing cambering phase of 10% of the chord and provided a 2% more lift than the only SMA actuation.

Fig. 4.7 Physical parameters for the cambering + flapping = hybrid morphing

96

A. Marouf et al.

Fig. 4.8 Benefits from the hybrid morphing (SMA + flapping) comparing to SMA actuation only (INPT and NTUA)

Fig. 4.9 PSD (Power Spectral Density) of the pressure coefficient according to cambering and hybrid morphing (cambering + piezos vibration), showing the drastic decrease of the noise sources

Moreover, the hybrid electroactive morphing has led to a drastic reduction of the predominant frequency peak of the shear layer vortices and of the overall spectral level, thus considerably reducing the aerodynamic noise sources generated from the trailing edge (Fig. 4.9).

4.4 Structural Modelling of the sRS with Embedded SMA (Shape Memory Alloys) N. Simiriotis, M. Fragiadakis, J. F. Rouchon and M. Braza

A novel robust algorithm, developed for predicting the nonlinear response of the SMA-structure interaction problem, is first presented. Then, the coupling of the analysis algorithm with an optimization method is proposed in order to predict the optimal structural and operational parameters. The proposed design methodology

4 High-Fidelity Numerical Simulations

97

Fig. 4.10 Stress and strain versus Temperature during the martensitic and austenitic phases of the Shape Memory Alloys, implemented in the model, INPT/IMFT

selects the design parameters of the problem at hand, i.e. the location of the actuators and the operating temperature, for given loading conditions. The methodology presented is validated and demonstrated with three case studies of the sRS, including the design of a realistic morphing concerning the sRs with targeted cambering. The algorithm has coded the electro-structural properties of the actuators (Shape memory Alloys in the structural model (CSM) through the behavior law relating stress and strain, given the martensitic and austenitic phases of the SMA material as a function of the Temperature. This model is described as follows in Fig. 4.10.

4.4.1 Structural Control of the Wing Equipped with SMA Actuators The shape control of an aeronautical configuration refers to the identification of equilibrium points between the deformation capacity of the structure and the working range of the SMA actuator. The working range of the actuators depends on the thermomechanical properties of the material. The solution requires the coupling of the nonlinear thermomechanical behavior law of the SMA with the structure

98

A. Marouf et al.

Fig. 4.11 Schematic representation of the sRS equipped with SMAs

response to any temperature change of the actuators. The structural model and the optimal shapes design are shown from Figs. 4.11, 4.12, 4.13 and 4.14. The coupling is accomplished through an iterative procedure since the complete system is material nonlinear due to the presence of SMA actuators and geometrically nonlinear due to the large displacements that the morphing structure typically undergoes. It is important for a shape control algorithm to accurately predict the working points of the whole controlled configuration. The solution of the design problem follows a two-step procedure. The first step requires the accurate prediction of the structural response under the control of SMAs. The second step refers to the coupling of the shape-control solver to an optimization tool in order to predict the design that best produces the desired shape. A detailed documentation of this methodology can be found in the PhD Thesis of Simiriotis [11]. The structural solver has been coupled with a stochastic optimisation solver. The result is an optimal design needing less attachments and less actuation temperature. Therefore, the present approach leads to greener architectures.

Fig. 4.12 Structural model of the morphing sRS equipped with SMAs and 2D mid-span section

4 High-Fidelity Numerical Simulations

99

Fig. 4.13 Realisation of the target shapes—A320 sRS prototype

Fig. 4.14 Optimal shapes with minimised temperature needed to actuate the SMAs. “ORG”: Original design. “OPT”: Optimal design. On the left: Temperature is a design variable. One joint is removed. The deflection is well captured at a lower temperature. On the right: Temperature is fixed (not a design variable). One joint is removed. The deflection is well captured at the fixed lower temperature

4.5 MDO (Multi-disciplinary Design Optimisation) Results Incorporating Experimental Results and High-Fidelity CFD Simulations F. Kramer and F. Thiele

4.5.1 Introduction The intended prototype design of the SMS project combines shape memory alloys and Electromechanical Actuators (EMA) for low frequency—high deformations as well as distributed piezoelectric actuators for high frequency low-deformation actuations [9, 12]. This is applied to two prototypes of increasing complexity. Amongst other investigations, Multi-Disciplinary Optimization (MDO) and Adjoint sensitivity analysis was applied to assist and improve the actuator placement of the two prototypes. The coupling of the fluid solver (CFD) with the structural solver (CSM) has already been reported. Therefore, the present work focuses on the adjoint multi-disciplinary optimization.

100

A. Marouf et al.

The main contribution of the present work is to numerically identify the most efficient placement for distributed actuators at different frequency regimes. When dealing with a distributed actuator system, a sensitivity-based approach using an adjoint solver has the inherent advantage that the costs for the sensitivity evaluation are independent of the number of control faces, rendering it very efficient for distributed controls [13, 14]. In the present work, such an adjoint solver is applied a realistic commercial airfoils closely matching the experimental campaign [9]. The resulting sensitivity maps identify efficient locations for the low (LF) and the high frequency (HF) actuation in order to increase the lift and performance. Being a design tool, it is important to quickly assess the different designs of complex configurations. Therefore, the presented approach focuses on methods with rapid turnaround times. The MDO optimization was performed on two prototypes and for multiple objectives like the lift and the lift-to-drag ratio. Two optimization strategies were employed to show also the potential of non-restrictive optimizations. The low frequency deformation of the wing was investigated by a quasi-steady approach whereas the high frequency part was modelled by mode decomposition and sensitivity windowing. Finally, the noise emission was evaluated for the optimized shapes by high-fidelity simulations.

4.5.2 Scientific and Technological Background The use of sensitivity maps from an adjoint flow solver is a very effective approach to reduce the overall costs of an optimization with many parameters. Such an adjoint solver can be created from a primal solver in many ways. Here, it was derived by discrete adjoints from an existing primal solver which itself was specialized to the specific demands of this project. The flow solver is a block-structured Finite Volume code solving the NavierStokes equations on a co-located grid and with second order accuracy in time and space. It uses a fully-implicit scheme and a pressure-based algorithm with a numerically advanced Rhie and Chow interpolation [15], suitable for sensitive applications such as aeroacoustics. A hybrid wall boundary condition blends smoothly between a low and a high Reynolds boundary condition, relaxing common constraints on the near wall mesh. However, the mesh in the present high-Reynolds number investigation is designed to resolve the boundary layer with an adapted low-Reynolds damping of the turbulence towards the wall, in the context of the boundary condition. The adjoint solver is generated by discrete adjoint code transformation using the software Tapenade [16] and by a manifold of manual code optimizations afterwards. All parts of the primal solver including the turbulence model and all limiters are incorporated in the derivation process. The accuracy of the adjoint solver was verified by comparison with Finite Differences at selected locations. Common excessive memory demands of the adjoint solver being caused by storing flow solutions forward in time are avoided by using a two-level check-pointing approach which carefully balances disk and memory usage in a hybrid storage scheme.

4 High-Fidelity Numerical Simulations

101

The sensitivity map represents the local gradient of the objective with respect to the parameters. In this case, the parameters correspond to each coordinate of each grid node on the wall i.e. the shape of the airfoil. Being a gradient, the sensitivity map is valid for an infinitesimal step size only. For typical gradient based algorithms, however, the gradient must be prolongated to achieve a meaningful optimization with reasonable step sizes. Higher order methods like BFGS approximate the second derivative to estimate the local step size but have shown to be extremely unstable because of the noisy response of unsteady simulations. Therefore, we chose the commonly used gradient descent method where the step size is chosen by a linesearch algorithm which can result in very large steps. In the present context, large steps have some disadvantages: 1. Large shape updates produce long transients which increase the computational cost of unsteady adjoint simulations tremendously. 2. Together with the free-node parametrization, the grid quality may suffer. Therefore, only small step sizes were applied which allowed an iterative process of combined smoothing and limiting to keep the face ratio and the angles between adjacent faces in a predefined range. Special care had also to be taken of the trailing edge corners. These modifications limit the design space but are required to achieve meaningful optimizations of unsteady flows, and will be referred to as realizability constraints. Another set of constraints stems from the posed problem and is stronger linked to physical interpretations and constraints: 1. Only the rear part of the airfoil shall be deformed. 2. Is the trailing edge allowed to move changing the effective angle of attack or the chord length? 3. How small are features on the airfoil allowed to develop? Bullet points 2 and 3 were collected into two morphing strategies: 1. A “Restricted” optimization to stay closely to the intended design: a. Fixing the trailing edge. b. Allowing only large features on the airfoil. 2. A “Free” optimization to show the potential of the concept: a. The trailing edge may rotate around a virtual centre located in the airfoil keeping only the chord length fixed. b. Small features are not limited beyond the above mentioned realizability constraints. The computational meshes were generated and morphed using the software ANSA from BETA CAE Systems.

102

A. Marouf et al.

4.5.3 Investigation of the Reduced Scale (RS) Prototype—RSP 4.5.3.1

Setup Details

The Reduced Scale Prototype (RSP) is presented in Fig. 4.15, having a chord length c = 0.7 m with the dimensions of the computational domain L x = 11c and L y = 1.02c. The computational mesh in Fig. 4.16 consists of 196,000 cells preserving y + < 1 on most of the airfoil surface as shown in Fig. 4.17. In accordance with the experiments, the Reynolds number is Re = cu in /ν = 106 and the Mach number Ma = u in /a = 0.065 where u in = 21.5m/s is the inflow velocity and a the speed of sound. For the k − ω SST-sust turbulence model, a turbulent intensity of T U = 0.01 and a viscosity ratio of ν/νt = 1 are applied as inflow conditions. All simulations are compressible, two dimensional and solved using the unsteady RANS equations. As agreed amongst the numerical partners, the tunnel walls are not fully resolved but modelled as symmetry walls.

Fig. 4.15 Computational domain of the RSP

Fig. 4.16 Computational mesh at the trailing edge

4 High-Fidelity Numerical Simulations

103

Fig. 4.17 Contours of the cell height y+

4.5.3.2

Experimental Comparison of Base Flow

An example of the primal simulation of the base flow is shown in Fig. 4.18. Close to the trailing edge, the flow detaches slightly but stronger than in the experiments as visible on the right side. The stronger wake in the simulations can originate from various differences between the physical modelling: Compared to the no slip walls in the experiment, the symmetry walls have no displacement thickness which reduces the flow speed between the airfoil and the tunnel wall in the simulations. The experiments are naturally 3D. The inflow conditions are chosen close to the experimental conditions, but in practice it is quite difficult to map accurately wind tunnel inflow conditions in the simulations. Especially the boundary values for the turbulence models are impacted by this problem. However, for the present investigations, the results are more than satisfactory. Even if the absolute values of the experiments are not exactly met, the optimization is improving the objective relative to its initial value. It is assumed that an improvement in the simulations produces a similar improvement in the experiments.

Fig. 4.18 Time average of streamwise velocity: compared with PIV from experiments (IMFT)

104

4.5.3.3

A. Marouf et al.

Sensitivity Analysis: Low-frequency Investigation

Database of Sensitivity Maps for Reference Deformations The partner INPT/IMFT delivered a series of reference deformations resembling classical flap movements but realized and measured with their pre-project feasibility assembly. These deformations were used here to generate a database of sensitivity maps with the above portrayed adjoint flow solver. The deformation states are characterized by the displacement of the trailing edge (TE) whereof only the baseline case, 20 mm upward and 25 mm downward displacement are discussed here. The baseline case with no additional bending of the TE in Fig. 4.19 (top) reveals a negative sensitivity in the leading edge (LE) region and positive values in the rest of the upper nose region. This corresponds to a flattening of the leading edge and an upward deformation of the upper nose region whereas the lower part is comparatively neutral. The trend is therefore to reduce the curvature of the LE and increase the thickness. A similar effect is commonly achieved by slats on multi-element airfoils. The outward deformation on the mid-section of the upper wing surface increases the camber of the airfoil. The sensitivity in the TE region is clearly dominated by the inward deformation of the lower surface which increases the circulation (Kutta condition). This is reinforced by the local maximum of the upper surface slightly upstream of the TE. The upwards and downwards deformed reference cases (only one representative shown for each) share the same characteristics of the baseline case. The maximum downward deformed reference case produces the maximum lift of the reference cases, and illustrates that better optimized shapes tend to show weaker sensitivities. The mid-section as well as the TE region feature significantly reduced sensitivities, with only the nose and lower TE regions retaining strong sensitivity. The negative local minimum on the rear part of the airfoil might indicate that a reduction of the excessive local curvature is favorable. The database was forwarded to the related partners according to the DMP of the SMS project.

4.5.3.4

Optimization for Lift

The optimization was performed for two objectives and two strategies as introduced in Sect. 4.5.2. Each optimization cycle consists of one adjoint simulation to generate the sensitivity map, one primal simulation to evaluate the current value of the objective and one line-search which itself may include more than one primal simulation if the initial step size of the gradient’s prolongation was too large. In practice, the line search terminated successfully on the first iteration with the initial step size. Shape updates are performed only during the line-search. The basic time interval of the adjoint and primal simulations was chosen to be large enough to allow the flow to adapt sufficiently to the shape update. “Sufficiently”

4 High-Fidelity Numerical Simulations

105

Fig. 4.19 Sensitivity of the lift with respect to wall normal outward deformation characterized by the displacement of the trailing edge (top to bottom: baseline, 20 mm, −25 mm) [17]

depends in this context strongly on the case and the objective. For the lift optimizations of the RSP, a dimensionless time interval of 1.23 demonstrated a good compromise by incorporating most of the shape update’s effect at reasonable computational time for a single cycle. The results from the restricted optimization and its progress are shown in Fig. 4.20.The upper left graph shows the portion of the airfoil which may be deformed (x > 0.5 m). The black shape represents the baseline shape. The lower graphs are enlarged views to improve the visibility of important features. The upper right graph shows a nearly monotonically increasing lift coefficient with increasing optimization cycles. Two example shapes are high-lighted and plotted in red and green. The green shape is the final shape when the grid quality was no longer preserved as visible at the TE. It achieved an increase of the lift of 4.4% by moving the airfoil mainly upwards except at the TE which was fixed. The outcome is a Gurney-flap type deformation in the vicinity of the TE. The red shape illustrates an intermediate step (30th cycle) with 2.6% lift increase. As the basic idea of the restricted optimization strategy is to stay close to the experimental design, the red shape was designated as final shape of this strategy. Besides the realisability constraints, the free optimization poses no restrictions except that it forces the TE to stay on a circle around a virtual centre to keep the chord length fixed. In Fig. 4.21, the improved deformations have similar features as

Fig. 4.20 Shapes and progress of the restricted RSP optimization for lift

106 A. Marouf et al.

Fig. 4.21 Shapes and progress of the free RSP optimization for lift

4 High-Fidelity Numerical Simulations 107

108

A. Marouf et al.

Fig. 4.22 Noise integration for multiple observers through DDES simulation—16.6 M cells

in the restricted optimizations like the upward motion as well as the Gurney flap type deformation but to stronger extents. However, on the lower airfoil between 0.5 < x/m < 0.6, a small counter movement is visible. The red shape with 10.6% lift increase is again designated as final “free” optimized shape being a good compromise between engineering perspectives and a strong increase of lift.

4.5.3.5

Optimization for the Lift-to-Drag Ratio

The setup of the RSP enforces a strong angle of attack of 10° which fits well to a landing scenario where a lift increase is the main interest. Nevertheless, optimizing for the lift-to-drag ratio allows to demonstrate the features of the technology as well as some generalizations for non-slotted airfoils. The optimization for the lift-to-drag ratio is much more challenging because the link from a shape update to a response of the objective is much weaker i.e. less direct. Combined with the unsteadiness of the flow and the related statistical uncertainty, this poses a huge problem for the optimization and required a strongly adapted methodology. Still, the optimization eventually stalls which is probably because the realizability constraints are too aggressive and modify the effective gradient such that it is no longer a descent direction. Figure 4.23 shows the result of the optimization until it stalls. The final shape achieves 0.1% increase of the lift-to-drag ratio by mainly thickening the airfoil. DDES (Detached Eddy Simulations) have been accomplished (Fig. 4.22) to evaluate the base flow and the variance, associated with the Adjoints Optimisation approach. A remarkably stronger effect is achieved by the free optimization in Fig. 4.24. Both, the deformation at the TE which reduces the effective angle of attack, and a much thicker airfoil produce an increase of the lift-to-drag ratio of 0.9%.

4.5.3.6

Noise Impact of Optimized Shapes

Compared to common acoustic evaluations of airfoils, the boundary conditions of the wind tunnel are very close to the airfoil producing massive reflections. Far-field prediction methods are generally not applicable in such setups. However, in order to approximate sound radiation into the far-field, we use the loading noise source term

Fig. 4.23 Shapes and progress of the restricted RSP optimization for the lift-to-drag ratio

4 High-Fidelity Numerical Simulations 109

Fig. 4.24 Shapes and progress of the free RSP optimization for the lift-to-drag ratio

110 A. Marouf et al.

4 High-Fidelity Numerical Simulations

111

of Curle’s equation. Applying far-field assumptions and neglecting retarded time and rigid body motion leads to Eq. (4.4), where a0 is the ambient speed of sound, xi the observer position, yi the position of the emitting surface element, n i its unit normal vector, R the distance between observer and surface element, and S the rigid surface. For low Mach and Strouhal numbers, this simplified equation produces reasonably good results compared to a full-featured FWH solution. ' pload ∼ =

1 4πa0

{

(xi − yi )n i dp dS R2 dt

(4.4)

The noise investigation is performed using high-fidelity DDES on an extruded mesh with L z = 0.4 m using 80 cells in the spanwise direction (total mesh size is 15.6 million cells). The spanwise domain extent was chosen to allow decorrelation between the spanwise periodic boundary conditions. However, the flow still features correlated turbulent structures in the wake of the airfoil. The narrow-band spectrum of an observer vertically below the airfoil at a distance of 100c is shown in Fig. 4.25. The baseline airfoil shape produces a strong peak around 150 Hz and two harmonic peaks around 450 and 900 Hz. The shape optimized for lift (L-optimized) is approximately 8 dB louder at the first two peaks but 11 dB quieter at the third peak. The shape optimized for lift-to-drag ratio (L/D-optimized) is very close to the base shape but produces more noise at higher frequencies. The overall sound pressure level (OASPL) for one observer can be calculated by integrating its spectrum. The directivity of the OASPL in Fig. 4.26 shows that the

Fig. 4.25 Narrow-band spectra of the base and the optimized shapes for an observer of the RSP at a distance of 100c below the airfoil (ϑ = 90°)

112

A. Marouf et al.

Fig. 4.26 Directivity for the RSP’s fly-over line at a distance of 100c. ϑ = 0° is parallel to the mean flow [17]

L-optimized shape is around 4 dB louder than the base shape over the whole fly-over line whereas the L/D-optimized shape is only around 1.5 dB louder. It is unclear how much the channel walls influence the results in this study but a similar result in freestream is very likely. This stresses the need to incorporate noise constraints directly into the optimization.

4.5.3.7

High-Frequency

The high-frequency deformation is modelled by a scale decomposition approach using temporal windowing on a static airfoil. This approximation allows to visualize short term effects of local deformation on the objective. The time integration interval T for which the sensitivity is calculated plays an important role for interpreting sensitivity results. A long interval like in the previous section filters out most of the unsteady effects such that the sensitivity will not change much with even longer periods. With shorter intervals, a favourable deformation along the sensitivity map might have an adverse effect outside of the used interval. However, this commonly unintended trade-off between intervals can be used to derive conclusions for high-frequency deformations. The difference to a long interval is stronger the shorter the chosen interval, as can be seen comparing Fig. 4.27a, g

4 High-Fidelity Numerical Simulations

113

where the former is the same as in the quasi-steady investigation, and the latter is a factor 100 shorter. The sensitivity can be split into three contributions: • Long-term: Sensitivity map converges with larger T (quasi steady). • Short-term: A strong contribution which is illustrated by the features that change only mildly with increasing T in Fig. 4.27 namely e.g. the favour for pitching up ceases with larger T. Fast gust events are an example for this class. • Unsteadiness: A weaker contribution linked to local unsteady structures. The long-term contribution is independent of the phase, angle (i.e. starting point of the interval). This allows the contributions to be separated by shifting the phase angle in a stepwise manner and decomposing the result into an ensemble average and fluctuating part. For sufficiently small intervals T, the average represents the short-term contribution and the fluctuating part the unsteadiness. For the short-term part, Fig. 4.27 also indicates that it could be interesting to use higher-frequency actuation on the whole airfoil such that a pitching up moment is quickly followed by a pitching down moment. As an example, a shifting interval was used in Fig. 4.28. For a single point on the upper surface about 0.04c upstream of the TE. The x-axis denotes the shifted starting point of each interval. The sensitivities oscillate at a frequency of approximately 250 Hz for all the intervals shown. In Fig. 4.29, the modal content was integrated for the band 200–300 Hz and plotted for the whole airfoil. It motivates that local high frequency actuations are preferably performed at the lower TE and on top of the main airfoil. This is to act on the local unsteady flow features. Figure 4.30 shows the band of the next harmonic (400– 500 Hz) where only the lower TE remains as very effective spot for high frequency actuation.

4.5.4 Investigation of the Large-Scale (LS) Prototype—LSP The LSP is a two-element airfoil in take-off condition. It consists of a main airfoil and an extended flap as shown in Fig. 4.31. The chord length in clean configuration (flap retracted) is c = 2.2 m. The airfoil is placed in a computational domain with far-field boundaries at 60 m as illustrated in Fig. 4.32. In accordance with the experiments, the Reynolds number is Re = cu in /ν = 2.25 · 106 and the Mach number Ma = u in /a = 0.032 where u in = 10.6 m/s is the inflow velocity at an angle of attack of 8.8°. A turbulent intensity of T U = 0.01 and viscosity ratio of ν/νt = 1 are applied as inflow conditions. All simulations are unless otherwise stated compressible, two dimensional and solved using the unsteady RANS equations. The computational mesh consists of 787,000 cells.

114

A. Marouf et al.

Fig. 4.27 Influence of integration interval on to sensitivity of the lift with respect to wall normal outward deformation of the baseline case at AoA = 10. Data is based on single simulations from a common starting point. The contour scalings of a, b deviate from the remaining plots [17]

4 High-Fidelity Numerical Simulations

115

Fig. 4.28 Oscillation of sensitivities at a fixed point on the airfoil evaluated on short intervals T for increasing starting times

Fig. 4.29 Magnitude of the 200–300 Hz band of the sensitivity fluctuation at T = 0.002 s

Fig. 4.30 Magnitude of the 400–500 Hz band of the sensitivity fluctuation at T = 0.002 s

116

A. Marouf et al.

Fig. 4.31 Shape of LSP

Fig. 4.32 Computational domain and mesh of the LSP

4.5.4.1

Optimization for Lift

In accordance with the experimental design, only the rear part of the flap x > 2.15 m may change its shape. Otherwise, the methodology is the same as for the RSP. The restricted optimization for lift (with a fixed TE and aggressive smoothing) in Fig. 4.33 produced a maximum of 2.2% lift increase at the 86th optimization cycle. The corresponding airfoil shape features a downward deformation in the region 2.2 m < x < 2.6 m which can be interpreted as a reduction of the maximum curvature of the surface on both sides. The TE is slightly thinner and shows a small outward bump resembling roughly a smooth but restrained Gurney-flap. The free optimization in Fig. 4.34 achieves a maximum lift increase of 2.8% with a similar shape but with a more pronounced Gurney-flap type deformation and a strong downward shift very close to the TE.

4.5.4.2

Optimization for the Lift-to-Drag Ratio

The restricted optimization for the aerodynamic performance is demonstrated in Fig. 4.35. The optimized shape achieves 1.3% improvement of the lift-to-drag ratio by showing a slightly downward bending between 2.2 m < x < 2.6 m and a thickened

Fig. 4.33 Progress of the restricted LSP optimization for lift and the most effective shape

4 High-Fidelity Numerical Simulations 117

Fig. 4.34 Progress of the free LSP optimization for lift and the most effective shape

118 A. Marouf et al.

Fig. 4.35 Progress of the restricted LSP optimization for the lift-to-drag ratio and the most effective shape

4 High-Fidelity Numerical Simulations 119

120

A. Marouf et al.

part closer to the TE. In Fig. 4.36, the restrictions are slightly weakened allowing more local curvature and producing a lift-to-drag ratio increase of 1.8%. The free optimization achieves an increase of the lift-to-drag ratio by 2.6% as depicted Fig. 4.37. The most distinct feature is the dent on the upper TE. However, a separate simulation without the dent had shown that the dent is not responsible for the increase which indicates that the upward motion of the TE is the essential feature.

4.5.4.3

Noise Impact of Optimized Shapes

Instead of the Curle’s equation, the noise evaluation of the optimized LSP shapes was performed by FWH integration of permeable surfaces which enclose all noise sources. The resulting spectrum for one single observer at 32c distance at 70° on the flyover line is shown in Fig. 4.38. The three shapes displayed are the base shape, the L-optimized shape from Fig. 4.34, and the L/D-optimized shape from Fig. 4.36. The spectra are close but the L-optimized shape is visibly louder up to 10 Hz. The directivity in Fig. 4.39 reveals that the L-optimized shape is up to 1 dB louder than the base case depending on the observer position. However, the L/D-optimized shape is 1 dB quieter than the base line case over a wide range of observer positions along the fly-over line.

4.5.5 Conclusions—MDO For the low frequency investigation of the RSP, a database of sensitivity maps for a multitude of reference deformations was generated and forwarded to the partners. It indicates where such deformations may be beneficial or may have an adverse effect. Additionally, an MDO shape optimization was performed achieving a lift increase of 2.6 and 10.6% depending on the level of freedom that is granted to the admissible shapes. This influences the actuator placement by demonstrating effective shapes that could be realized by the actuator design. Similarly, an increase of 0.1 and 0.9% of the lift-to-drag ratio was demonstrated. Comparison with experimental results showed sufficient agreement. A high-fidelity noise evaluation of these improved shapes revealed that the optimized shapes produced up to 4 dB more noise than the base shape. The high-frequency investigation isolated two regions of major influence. On the lower airfoil side very close to the lower TE, and the mid-upper side of the airfoil. The most effective frequency for these actuations was found to be close to 250 Hz being similar to the shedding frequency of the wake. The LSP optimization of a two element airfoil during take-off conditions produced shapes that increased the lift by 2.2 and 2.8% as well as the lift-to-drag ratio by 1.3 and 2.6% depending on the restrictions imposed on the surfaces. The high-fidelity

Fig. 4.36 Progress of a less restrictive LSP optimization for the lift-to-drag ratio and the most effective shape

4 High-Fidelity Numerical Simulations 121

Fig. 4.37 Progress of the free LSP optimization for the lift-to-drag ratio and the most effective shape

122 A. Marouf et al.

4 High-Fidelity Numerical Simulations

123

Fig. 4.38 Narrow-band of the base and the optimized shapes for an observer of the LSP at a distance of 32c below the airfoil (ϑ = 70°)

noise evaluation, however, shows a noise increase for the lift enhancing case but a noise reduction for the lift-to-drag ratio improving the design’s setup. Results and shapes were forwarded to the partners in WP2 and WP3, drafting the RSP and the LSP to be considered for the final design. Additionally, basic characteristics were derived, discussed and made available for the controller design in WP3.3 and WP3.4.

4.6 Hi-Fi Simulations on the tRS Prototype (INPT) A. Marouf, Y. Hoarau, M. Braza, J. B. Vos, D. Charbonnier, K. Diakakis and G. Tzabiras

The tRS prototype is a 15 cm chord and span wing, whose experiments have been carried out at the IMP-PAN transonic wind tunnel, detailed in Chap. 5. This prototype is equipped with a piezoactuator able to apply vibration and slight deformation of the near trailing-edge region in a similar way as detailed previously. The design of this prototype has been presented in Chap. 2. The Numerical simulations preceded the experimental evaluation (Chap. 5) and provided a detailed investigation of the optimal

124

A. Marouf et al.

Fig. 4.39 Directivity for the LSP’s fly-over line at a distance of 32c. ϑ = 0° is parallel to the mean flow

frequency and amplitude ranges for the vibration, in respect of the aerodynamic performance. The optimal ranges, as well as a thorough physical analysis of the mechanisms leading to the benefits have been accomplished and have been in strong synergy with the experiments (Chap. 5). The transonic regime corresponds to the cruise phase, for which a typical nominal Mach number for the A320 is 0.78 and angle of attack of 1.8°, thanks to interaction within the SMS project with AIRBUS “Emerging Technologies and Concepts Toulouse”, ETCT, endorser and part of the advisors of the project. Concerning the cruise phase, which is the most lasting phase of flight, the crucial issue is the drag reduction, together with the lift-to-drag increase. Therefore, these criteria have been targeted in the present investigation, as well as in the experimental evaluation (Chap. 5). The detailed investigation of the tRS can be found in the PhD thesis of J.B. Tô [18]. The mesh view and the rear part actuation are presented in Fig. 4.40. The actuation occurs at the rear region. As has been analysed, strong feedback effects occur towards the SWBLI (Shock Wave Boundary layer interaction) region. Through the morphing concepts, manipulation of the coherent and turbulent vortices near the trailing edge and in the wake have proven creating a strong impact even upstream of the SWBLI. Therefore, these effects have been suitable used to improve

4 High-Fidelity Numerical Simulations

125

Fig. 4.40 Mesh around the tRS prototype, Re = 2.06 Million

the aerodynamic performance and particularly to decrease drag and noise sources. Three kinds of actuation have been operated as presented in the following Fig. 4.41. The compromise has been to reduce drag and to keep up the lift-to-drag ratio, even if several cases produce less drag but reduce also the lift. The effects of a slight static upward deflection had been analysed in RP1 and precisely found interesting in decreasing the drag but simultaneously it decreased lift. Therefore, in RP2 we have focused on the benefits taken from optimal frequency/amplitude vibration as well as hybridisation of this with a slight static upwards deflection. A thorough investigation has been carried out which permitted to define the optimal ranges of the vibration frequency and amplitude. The optimal vibration

Fig. 4.41 Schematic view of the actuation cases

126

A. Marouf et al.

Fig. 4.42 Optimal vibration ranges

frequency has been proven in the range (300–350) Hz, providing simultaneously drag reduction and lift increase (Fig. 4.42). Furthermore, the amplitude of the vibrations is of 0.13 mm and the vibration length of the piezo-region 7.5 mm (lengths to consider in analogy to the chord’s length of the tRS of 15 cm being less than the chord length of the sRS of 70 cm). Figure 4.43 shows the Power Spectral Density of the lift coefficient for the static (non-morphing) case, compared to the morphing case with 350 Hz vibration of the near-trailing edge region. It is remarkable that a significant decrease of the spectral amplitude is obtained through this kind of actuation (blue curve). Furthermore, the predominant frequency peaks in the low frequency range that correspond to the buffet mode in the static case, have been shifted towards the actuation frequency and its harmonics in the morphing case. These facts indicate a strong lock-in effect of the buffet and therefore its control through the actuation frequency. The explanation for these benefits related to the optimal range is provided here: In the present cruise phase, an important instability occurs associated to the shock wave motion back and forth along the extrados, known as transonic buffet. This phenomenon creates a strong increase of the rms of the forces and an increase of the mean drag. Furthermore, it can be a triggering factor for the onset of a very dangerous structural coupling instability, the “dip-flutter”. It is therefore worthwhile to attenuate/suppress it by the morphing.

4 High-Fidelity Numerical Simulations

127

Fig. 4.43 Power Spectral Density (PSD) of the lift coefficient showing the lock-in mechanism of the buffet frequency (fb ) to the actuation frequency (fac ) as well as the drastic decrease of the spectral amplitude level thanks to the optimal actuation of 350 Hz [18]

Once again, this can be achieved thanks to the feedback effects on the shock-vortex interaction. In previous studies by INPT/IMFT, it was shown that this instability is highly depending on the von Karman mode developed in the wake [3, 19] and that suppression of the von Kármán mode leads to suppression of the buffet. In http:// smartwing.org/SMS/EU/DOCUMENTS/Antonio_JIMENEZ_GARCIA.pdf [20], it had been shown that suppression of the von Karman mode by introducing a splitter plate of an optimal length led to complete suppression of the buffet. In the present investigation, the actuation is able to considerably attenuate the von Karman mode and to “push” it quite far downstream as shown in Fig. 4.44, where comparison of the vortex dynamics of the non-optimal frequencies 200 and 400 Hz with the optimal one of 350 Hz has been shown. The aerodynamic performances are shown in the following Table 4.1. A second optimal range around vibration frequencies of 700 Hz has been also determined. However, this range has not been recommended for the experiments in WP5, because it would demand a much higher energy for the piezo-actuation and not robust vibration regimes. Therefore, the optimal range selected is 300–350 Hz, leading to a considerable liftto-drag performance in cruise from 2.5 to 5.5% and simultaneously drag reduction from 3.5 to 8.9%, a strongly impacting benefit in cruise. This range has been adopted for the experiments presented in Chap. 5 and as will be seen, they permitted a very significant drag reduction too. These facts demonstrate once again the strong synergy among the simulations and the experiments on the morphing prototypes in the SMS project.

128

A. Marouf et al.

200 Hz

400 Hz

350 Hz

Fig. 4.44 Vortex dynamics through streaklines visualisation concerning non-optimal vibrations (200 and 400 Hz) versus the optimal range (350 Hz), suppressing the von Karman mode and the buffet instability

4 High-Fidelity Numerical Simulations

129

Table 4.1 Lift-to drag performance and drag variation versus actuation frequency thus defining the optimal actuation ranges (INPT/IMFT), whose first one (italic) has been adopted in the final experimental evaluation (Chap. 5), by IMP-PAN Frequency (Hz)

300

350

400

450

500

700

+ 2.4%

+ 5.5%

− 2.5%

− 5.5%

− 5.5%

+ 5.9%

720

750

800

1000

1500

+ 1.6%

+ 6.1%

+ 6%

+ 3.5%

+ 5.8%

Frequency (Hz)

300

350

400

450

500

⟨Cd ⟩−⟨Cd ⟩static ⟨Cd ⟩static

− 3.5%

− 8.9%

+ 6.1%

+ 12.3%

+ 11.9% − 9.1%

Frequency (Hz)

720

750

800

1000

1500

⟨Cd ⟩−⟨Cd ⟩static ⟨Cd ⟩static

− 1.2%

− 9.8%

− 9.6%

− 4.1%

− 9.1%

⟨Cl /Cd ⟩−⟨Cl /Cd ⟩static ⟨Cl /Cd ⟩static

× 100

Frequency (Hz) ⟨Cl /Cd ⟩−⟨Cl /Cd ⟩static ⟨Cl /Cd ⟩static

× 100

× 100 × 100

700

4.7 Hi-Fi Simulations on the Large Scale (LS) Prototype A. Marouf, Y. Hoarau, M. Braza, J. B. Vos, D. Charbonnier, K. Diakakis and G. Tzabiras The LS prototype is a wing equipped with high-lift flap of constant section. The total chord is of 2.40 m in clean configuration and reaches 2.70 m in take-off configuration. The flap’s chord is of 1 m, comparable to the scale 1 chord of the A320 high—lift flap in mid span approximately (Fig. 4.45). The objective of this work is to prove that the morphing concepts considered in SMS are applicable in real scale and this study has been strongly linked with the final experimental study with continuous synergy and interaction. The present numerical study provided a detailed set of Hi-Fi simulations which permitted detection of the optimal frequencies and amplitudes of the piezoactuators disposed along the span. Furthermore, a detailed numerical study of the cambering effect allowed specification of the four cambering shapes used for the experiments in Chap. 5.

4.7.1 Take-off Configuration 4.7.1.1

Effect of the Flapping Through Piezo-Actuators on the LS Prototype

Table 4.2 shows that the range near 300 Hz is once again an optimal one, as in case of the sRS prototype. This can be explained by the fact that the flap’s chord is 1 m, comparable to the sRS wing’s chord of 70 cm.

130

A. Marouf et al.

Fig. 4.45 View of the take-off configuration in real flight

Table 4.2 Lift coefficient relative variation for different actuation frequencies comparing to the static case—INPT and NTUA Frequency of actuation (Hz)

60

100

200

300

400

(C L −C L static ) C L static

− 0.2918

+ 0.5797

+ 0.3465

+ 0.5529

+ 0.2557

× 100

Figure 4.46 shows the flow structure around the morphing flap of the A320 LS wing with selected monitor points for which the Power Spectral Density (PSD) has been plotted. A drastic decrease of the spectral energy and of the peak of the von Karman vortices has been obtained through the 300 Hz optimal actuation. By means of DDES-OES simulations through the NSMB code (Fig. 4.47), the suppression of the vortex dislocations pattern and bidimensionalisation of the wake is obtained (Fig. 4.9) thanks to the optimal actuation frequency of 300 Hz and low amplitude vibration of 3–5 mm. It is worthwhile mentioning that other morphing concepts using hydromechanical actuations, (MEMS), as well as control strategies using AJVG (Active Jets Vortex Generators) in previous EU projects where able to reduce the drag but not achieving simultaneously an increase of lift and decrease of noise. They rather decreased lift and increased the noise. It is pointed out in the SMS project that by enhancing the beneficial vortices responsible to increase the circulation around the wing’s rear part and by destroying large coherent vortices responsible for the drag, a simultaneous drag reduction and lift increase has been obtained, together with a drastic reduction of the noise sources (Figs. 4.46, 4.48, and Table 4.3).

4 High-Fidelity Numerical Simulations

131

Fig. 4.46 Iso-velocity field and spectrum in the wake showing the drastic reduction of the spectral energy and of the predominant peak corresponding to the von Karman instability mode thanks to the morphing by 300 Hz—INPT

By means of the physical analysis and explanations given in the above paragraph and based on the obtained performances, the SMS project demonstrated a simultaneous benefit on these three aspects: lift, drag and noise.

4.7.1.2

Cambering Effects on the LS Prototype

The cambering of the LS prototype followed optimised shapes provided by adjoint optimisation methodology as well as through the optimal shapes obtained by the structural modelling presented in the previous section. The Hi-Fi simulations using the NSMB code have been carried out with the following cambering positions (Fig. 4.49) concerning the take-off configuration. The aerodynamic performances are shown in Fig. 4.50 for two Reynolds numbers and angle of attack of 4°. The position P1 shows a high efficiency in respect of lift-to

132

A. Marouf et al.

Fig. 4.47 Turbulence structure around the LS prototype, Re = 2.2 Million. HI-Fi simulations by means of DDES-OES modelling, grid of 50 Million, NSMB code, INPT

Fig. 4.48 3D vortex structure past the A320 sRS prototype in static (no morphing) configuration (left). Suppression of the Vortex dislocations (red circle regions) and bidimensionalisation of the von Kármán vortex rows with morphing at 300 Hz (right)

4 High-Fidelity Numerical Simulations Table 4.3 Aerodynamic performances, mesh of 10 Million with the optimal actuation frequency of 300 Hz

−static static

133 ×100

−static static

+ 4.2870 × 100

+ 1.6182

Fig. 4.49 Mesh around the flap of the LS wing and successive cambered positions

drag ratio, much higher than the conventional configuration (Reference case) of the wing/flap without cambering, where the flap comes out in the same way as in the real flight. The position 1 shows high performance and the same tendency is found in the experiments. The performances obtained for different angles of incidence are presented in Table 4.4.

Fig. 4.50 Aerodynamic performance versus time: left: Re = 2.2 M, right: Re = 7 M. INPT

134

A. Marouf et al.

Table 4.4 Lift-to-drag ratio for different angles of attack and two Reynolds numbers Re = 2.25Million (C L/C D−C L r e f /C Dr e f ) × 100 C L r e f /C Dr e f

Re = 7Million (C L/C D−C L r e f /C Dr e f ) × 100 C L r e f /C Dr e f

AoA

Position 1

Position 2

Position 3

Position 4

0°

+ 12.7942

+ 20.0007

+ 24.5079

+ 24.2512

2°

+ 5.7920

+ 8.0998

+ 8.0339

+ 5.0211

4°

+ 0.8836

− 2.4606

− 6.0129

− 9.3351

AoA

Position 1

Position 2

Position 3

Position 4

0°

+ 13.76

+ 20.67

+ 22.67

+ 22.74

2°

+ 7.97

+ 10.35

+ 10.79

+ 10.41

4°

+ 3.26

+ 2.63

+ 1.21

+ 0.21

Take-off configuration

Fig. 4.51 Aerodynamic performance of the cambered take-off configuration at angle of incidence of 4° versus performance of the reference configuration at incidence angle of 8°. Re = 7 Million

Figure 4.51 shows that thanks to the cambering (position 1), a higher performance than the reference case (conventional system without cambering at maximum incidence angle of 8°) is obtained, even by using a lower angle of attack of 4°.

4.7.2 Landing Configuration The landing configuration concerns higher angles of incidence. This study also has been in strong synergy with the experiments of this configuration carried out in POLIMI and presented in Chap. 5.

4 High-Fidelity Numerical Simulations

AoA = 0° - Flap deflection = 10°

135

AoA = 8° - Flap deflection = 40°

Fig. 4.52 Computational domain of the LS prototype in the wind tunnel of POLIMI

The sliding mesh approach (colored in blue) is used to change the angle of attack of the main wing, while the chimera approach (colored in red) is used for the different flap settings (angle of deflection). The solid walls of the wind tunnel are depicted in grey. Figure 4.52 shows two configurations, one with an angle of attack 0° and a flap deflection 10° (left) and the other one with an angle of attack 8°and a flap deflection 40° (right). A zoom of the computational grid (median section) is provided in Fig. 4.53.

Fig. 4.53 Structured 2D grid of large-scale prototype in the wind tunnel of POLIMI, Landing conditions with AoA = 2° and Flap deflection = 25°

136

A. Marouf et al.

K. Diakakis and G. Tzabiras (NTUA) studied the landing configuration for different angles of incidence for the flap. The performances are presented in the following Fig. 4.54, through the solver MaPFlow (Fig. 4.55). Velocities 20, 30, 40, and 50 m/s were initially tested without morphing. For the chord of 2.72 m, these velocities correspond to Re = 3.6, 5.4, 7.2, and 9.1e06. Flow angles 4, 6, and 8 were considered. The flap deflection angle varied between 25 and 40°. Increasing the flap angle will lead to flow separation. This is clearly pronounced for the case of 25° and 20 m/s flow velocity. This case has significantly higher drag increase between 4 and 6° because trailing edge separation appears on the flap suction side. This paired with the present Reynolds number range, raises the drag considerably. For higher velocities with this configuration, the drag is not as high even though the flow separates. These results indicated that it is suitable to

a) Landing configuration, 20o flap, Cl mean

c) Landing configuration, 20o flap, Cd mean

b) Landing configuration, 25o flap, Cl mean

d) Landing configuration, 25o flap, Cd mean

Fig. 4.54 Performances for the landing configuration and different angles of the high-lift flap— NTUA, MaPFlow solver

4 High-Fidelity Numerical Simulations

137

(a) Landing configuration, 35o flap, Cl mean

(b) Landing configuration, 35o flap, Cd mean

(c) Landing configuration, 40 o flap, Cl mean

(d) Landing configuration, 40 o flap, Cd mean

Fig. 4.55 As in previous figure, NTUA

avoid the very high angles of incidence of the flap for the high velocity range because they create stall conditions. Therefore, these ranges have not been investigated by the experiments presented in Chap. 5. For the same reason, in the following, the focus is given to the lower deflection of the flap. The following results have been obtained by CFSE through the NSMB code. The CFSE simulation results refer to the 4 cambering positions and shapes of cambering specified by INPT and detailed in the previous section. These positions have been used for the experiments by POLIMI and are designated by 4 different colours as in WP5 and in Fig. 4.56. This figure shows the aerodynamic performances obtained through these 4 positions concerning the landing configuration. These shapes are presented in the following figure.

138

A. Marouf et al.

Fig. 4.56 Morphing shapes of the high-lift flap. Green curve: position “0%” of camber: static position, non-cambered: reference case. Violet curve: position “33%” of camber: trailing edge at 3.33 cm lower than in the reference case. Brown curve: position”66%”: trailing edge at 6.66 cm lower than in the reference case. Blue-green curve: position “100%”: trailing edge at 10 cm lower than in the ref. Case

Figure 4.57 shows the performances according to these shapes and positions, as well as the main benefits. The colors for the selected cambered shapes are slightly modified in the following. The following figures (Figs. 4.58, 4.59, and 4.60) show the performances according to the flapping with three vibration frequencies, 100, 200 and 300 Hz and using the same patches length and amplitude as in the take-off investigation. It has been shown that the actuation by 300 Hz provides the largest increase in lift and a slightly increased drag, which is useful for the landing phase near the approach as mentioned at the beginning of this book, because it helps to reduce the speed. The lift-to drag ratio is slightly lower with this actuation frequency but the difference is quite low. Therefore, this actuation frequency is an optimal compromise. Concerning the influence of the flapping on the rms of the forces, NTUA performed a detailed investigation as follows: Investigation of the flapping characteristic length of the piezo-patches has been carried out by NTUA using their unstructured solver MaPFlow. Comparison of the flapping effect regarding the static cases are presented in Fig. 4.61. For moderate angles of flap’s incidence (20°), no significant effect has been obtained comparing the static configuration (Figs. 4.61 and 4.62). However, for higher angles of the flap’s incidence, the wake’s width has been reduced. A significant decrease of the rms values of the forces have been obtained for optimal ranges of the piezoactuator amplitudes concerning the flapping motion at 40° of flap’s incidence. A detailed investigation through numerical simulations has been accomplished for the landing configuration concerning the hybrid morphing effects. The performances are presented in Fig. 4.63 and Table 4.5. A significant benefit is obtained with this hybrid morphing concerning cambering shapes 1 and 2, where an increase of the aerodynamic efficiency (lift-to-drag) of 6% has been obtained, thanks to simultaneous drag decrease of 5.2% comparing to cambering only (Table 4.5). It is worth noticing that the overall high-lift systems (conventional and morphing systems through other concepts, mainly mechanical and not electroactive) lead certainly to an increase in lift but at the same time

4 High-Fidelity Numerical Simulations

139

a) Drag coefficient CD at M=0.15 – Cambering effect

b) Lift coefficient CL at M=0.15 – Cambering effect

c) Lift-to-drag ratio CL/CD at M=0.15 – Cambering effect Fig. 4.57 Aerodynamic performance for the landing configuration according to the cambering shapes

140

A. Marouf et al.

Fig. 4.58 Drag coefficient CD at M = 0.15—flapping effect

Fig. 4.59 Lift coefficient CL at M = 0.15—flapping effect

provide a more “bluff-body” shaped flap which consequently increases the drag. With the present hybrid electroactive morphing, a simultaneous increase of lift and decrease of drag are obtained. This is a major outcome of the SMS project distinguishing its innovation far beyond current limits.

4 High-Fidelity Numerical Simulations

141

Fig. 4.60 Lift-to-drag ratio CL/CD at M = 0.15—flapping effect

(a) Static configuration - 20° of flap’s incidence

(c) 20° flap incidence, 200 Hz actuation

(b) Static, 40° of flap incidence

(d) 40° flap incidence, 200 Hz actuation

Fig. 4.61 Influence of the flap’s incidence on the flow structure for 200 Hz actuation—NTUA

142

A. Marouf et al.

Fig. 4.62 rms of the forces versus amplitude of the piezo-patches indicating decrease of the fluctuations, MaPFlow solver—NTUA

4.7.3 Aerodynamic Performance by Hi-Fi Simulations Around the Full A320 Aircraft A. Marouf, J. B. Vos, A. Gehri, Y. Hoarau and M. Braza

To evaluate the effects of the flap cambering on the whole aircraft, a full-scale A320 airplane (Fig. 4.64) is considered during a real take-off configuration. This allows obtaining more representative gain percentages on the aerodynamic performance liftto-drag ratio. The simulations have been done with the NSMB code (Navier Stokes MultiBlock) [21]. Figure 4.65 shows the deformation of the inner and outer flaps with a camber control using the EMA actuators (WP3 and WP5) in the order of 10 cm downwards concerning the trailing edge and following the same cambering as for two-dimensional simulations previously presented for the LS prototype [22]. Based on different tests conducted for the two-dimensional simulations, a defined cambered position P1 was computed for the inner and the outer flaps regarding optimal two-dimensional results presented in the previous section, at a medium chosen angle of attack of 4°. This is compared first with the conventional configuration at the same angle, as presented in Fig. 4.66. This figure shows that the cambering forces the pressure difference P-P0 to increase over the flaps upper surfaces with a simultaneous feedback on the pressure distribution over the wing and up to its leadingedge. The same effects have been observed in the two-dimensional simulations (Fig. 4.67) where the CP was presented. In addition, the pressure is continuously increased over a considerable part of the fuselage-wing junction (Fig. 4.65). This increase allows reaching a better aerodynamic efficiency than the standard configuration and the 2D simulations.

4 High-Fidelity Numerical Simulations

a) Drag coefficient CD at M=0.1 – Flapping and cambering effects

b) Lift coefficient CL at M=0.1

c) Lift-to-drag ratio CL/CD at M=0.1

Fig. 4.63 Benefits according to the hybrid morphing: cambering + flapping: landing

143

144

A. Marouf et al.

Table 4.5 Aerodynamic performances comparing the cambering only to the hybrid morphing: cambering + flapping—CFSE Mach

AoA [°]

Cambering [% max = 0.1 m]

Flapping [Hz]

CD

CL

CL/CD

0.1

0

0

0

0.0400

5.5003

137.46

0.1

2

0

0

0.0479

6.1364

128.15

0.1

4

0

0

0.0569

6.7730

119.05

0.1

6

0

0

0.0697

7.4050

106.23

0.1

0

33

0

0.0516

5.8885

114.12

0.1

2

33

0

0.0617

6.5384

105.89

0.1

4

33

0

0.0726

7.1743

98.87

0.1

0

66

0

0.0683

6.3550

93.06

0.1

2

66

0

0.0792

6.9716

87.99

0.1

0

0

300

0.0415

5.6859

136.90

0.1

2

0

300

0.0495

6.3374

128.02

0.1

4

0

300

0.0687

8.0992

117.83

0.1

0

33

300

0.0490

5.9333

121.20

0.1

2

33

300

0.0587

6.5408

111.40

0.1

4

33

300

0.0689

7.1653

103.94

0.1

0

66

300

0.0662

6.3567

96.04

0.1

2

66

300

0.0766

6.9566

90.81

5.2%

Camber Only

Camber + Flapping

The relative gain in lift-to-drag ratio of + 0.81% is presented in Fig. 4.66, compared to the reference case at the same angle of attack. The cambered flap increases the lift-to-drag of the whole airplane. This is shown through the pressure modification in the wing and the fuselage. In addition, Fig. 4.67 shows comparison of one cambered position at 4° with the reference case at standard take-off conditions of 8°. An increase of + 2.24% lift-to-drag ratio is obtained for the full aircraft regarding the position P1 [22], but just morphing of the two high-lift flaps.

4 High-Fidelity Numerical Simulations

145

Fig. 4.64 Grid of the take-off configuration and Cp iso-surfaces around the whole A320 airplane, Re = 11 Million, INPT & CFSE

146

A. Marouf et al.

Fig. 4.65 Iso-surfaces of the dimensional pressure difference of the full Airbus A320 airplane angle of attack 4°. a: reference baseline, b: cambered position P1

It is worth mentioning that this order of magnitude of overall performance around the whole aircraft is very significant, given that only the flaps have been subjected to morphing, whereas all the rest of the airplane’s surfaces are kept the same as conventional.

4.7.4 Hybrid Morphing—Full Aircraft, Landing—CFSE The hybrid morphing has been finally applied numerically for the landing configuration around the whole aircraft for flapping frequencies 250 and 500 Hz. It has been found that an increase of the vibration frequency to 500 Hz provided an order of 1.0% of increase in the aerodynamic efficiency (Fig. 4.68) for the full aircraft, which is once again a strongly impacting result concerning the real configuration not foreseen in the DoA.

4 High-Fidelity Numerical Simulations

147

Fig. 4.66 Gain in performance at Re = 2.2 Million—whole A320 aircraft

Fig. 4.67 Gain in performance at Re = 7 Million—whole A320 aircraft

4.8 Aircraft Trajectories, Polars and Fuel Consumption J. L. Farges and T. Chaboud

This paper proposes a methodology integrating Computational Fluid Dynamics (CFD) in a wide airfoil technology assessment framework. Aircraft trajectories simulation allows a choice of relevant conditions for CFD computations. Foil models are identified by least squares from CFD results and aircraft models, that can be used for flight simulation, are extrapolated from foil models. Fuel consumption impact is then derived from equilibrium of forces and engine model.

148

A. Marouf et al.

Fig. 4.68 Aerodynamic efficiency on the whole aircraft with the hybrid morphing—CFSE

4.8.1 Introduction In the field of aerodynamics, numerical analysis performed by Computational Fluid Dynamics (CFD) provide flow speed, pressure and density in the neighborhood of airfoils. Integrating the contribution of pressure and viscous forces in each direction leads to lift and drag forces. Then, considering the mass density and the speed of the fluid and the reference surface of the airfoil, lift and drag coefficients are obtained. For given computational conditions, consideration in the numerical analysis of advanced airfoil technologies, for instance morphing or boundary layer aspiration, produce changes of the values for the resulting lift and drag coefficients. Generally, a technology is considered successful when the lift to drag ratio is higher for the computation with the technology than for the computation without the technology. The purpose of this chapter is to present a methodology that provides a wider assessment than the classical comparison of lift to drag ratios. In order to do so, several problems shall be addressed. First, relevant computational conditions for CFD have to be determined. This is the purpose of the next section. Then, Sect. 4.8.3 provides a method to aggregate the results obtained by CFD for a set of conditions into a simple model of the airfoil. The question of extending the differences between airfoils models with and without a given advanced technology to potential differences between models of aircraft using this technology or not is addressed in Sect. 4.8.4. Finally, using models of aircraft with and without the technology, a benefit in terms of fuel consumption can be computed. This last step of the methodology is presented in Sect. 4.8.5.

4 High-Fidelity Numerical Simulations

149

4.8.2 Defining Relevant Conditions for Fluid Dynamics Computations Finding relevant conditions for CFD analysis is based on aircraft trajectories. Those trajectories concern existing aircraft not using the advanced airfoil technology and performing typical missions they are designed for. It may be trajectories collected using Automatic Dependent Surveillance–Broadcast (ADS-B) receivers [23] or simulated by a computer program. Typically those simulation tools take as input origin and destination airports and a flight plan. They emulate the operational procedures and the behavior of the aircraft flight management system to provide a trajectory. The resulting trajectory is a sequence of data records, each data record including: • • • • • • • • • • • • • • • • • • •

the date of the record, the name of the geographical position currently used for navigation, the type of the current significant point, the time elapsed since the beginning of the flight, the position of the aircraft in terms of latitude, longitude and altitude, the maximum possible altitude for the current state of the aircraft, the air speed, the calibrated air speed, the Mach number, the heading of the aircraft, the ground speed of the aircraft, the distance to the next waypoint of the flight plan, the distance flown since the beginning of the flight, the components of the wind vector, the direction of the route followed by the aircraft, the current mass of the aircraft, the fuel burnt since previous record, the fuel consumed since the beginning of the flight, the current flight mode. Those records are used for defining relevant conditions for CFD computation:

• The ratio between air speed and calibrated airspeed is used to set the air density. • The Mach number is directly used. • The airspeed, the components of the wind vector, the heading of the aircraft and the differences of time and altitude between two successive records is used to set the incidence of the airfoil. • The current flight mode is used to set the airfoil configuration in terms of flaps and slats.

150

A. Marouf et al.

4.8.3 Identifying Polar Models from Results of Fluid Dynamics Computations By varying the incidence CFD provides, for identical conditions, series of N couples of aerodynamic coefficients. Each couple n includes: • a lift coefficient C L ,n and • a drag coefficient C D,n . A polar model describes the relation between the lift and drag coefficient. A classical polar model, illustrated by Fig. 4.69, is: C D = C D,0 + K C L 2

(4.1)

where C D,0 the zero-lift drag coefficient and K the sensitivity of the drag coefficient to the square of the lift are parameters to be identified. A least square identification of the two parameters leads to: ∑N C D,0 =

n=1

C L4 ,n

∑N n=1

4

∑N

C D,n −

n=1

∑N

C L4 ,n −

n=1

C L2 ,n

(∑ N

n=1

∑N

C L2 ,n

n=1

)2

C D,n C L2 ,n

(4.2)

and K =

4

∑N

∑N ∑N C D,n C L2 ,n − n=1 C D,n n=1 C L2 ,n ) ( 2 ∑N ∑N 2 4 n=1 C L4 ,n − n=1 C L ,n

n=1

(4.3)

Among identified conditions there is the presence or absence of an advanced f,t airfoil technology to be assessed for a foil f . Thus the identification leads to C D 0 Fig. 4.69 Aircraft lift-drag polar

4 High-Fidelity Numerical Simulations

151 f,¬t

and K f,t for a series of data with advanced technology and C D 0 series of data corresponding to the absence of this technology.

and K f,¬t for a

4.8.4 Extrapolating to Aircraft Performance For most aircraft, published data present full aircraft aerodynamic performance in and K a,¬t are available. the form of Eq. (4.1) [24]. Thus, for the aircraft a, C D a,¬t 0 Usually, CFD provides results for a foil that is implemented on the two sides of the aircraft. Thus, assuming absence of interaction with the other parts of the aircraft leads to the polar model of a with technology t: ) ( f,t f,¬t a,¬t C D a,t + 2 C D0 − C D0 0 = C D0

(4.4)

) ( K a,t = K a,¬t + 2 K f,t − K f,¬t

(4.5)

4.8.5 Computing Benefit in Fuel Consumption Assuming a steady state flight, the lift force shall compensate the weight of the aircraft. Thus the lift coefficient shall be given by: CL =

2mg0 ρV 2 S

(4.6)

where • • • • •

m is the current mass of the aircraft, g0 is the gravity constant, that is 9.80665, ρ is the air density where the aircraft is, V is the airspeed of the aircraft, and S is the reference surface of the aircraft.

Note that the value of C L is independent of the use of the advanced technology. a,¬t a,t and C D , when using respecNevertheless, Eq. (1) leads to two different values, C D a,¬t a,¬t tively for C D0 and K , on the one hand, C D 0 and K and, on the other hand, a,t C D a,t . 0 and K Then assuming that the thrust of engines shall compensate the drag force leads to: T a,¬t =

1 a,¬t ρ SV 2 C D 2

(4.7)

152

A. Marouf et al.

T a,t =

1 a,t ρ SV 2 C D 2

(4.8)

where T a,¬t and T a,t are the needed thrust for the aircraft respectively without and with advanced technology. Using an engine model [25] allows deducing fuel flows without and with airfoil advanced technology, respectively F a,¬t and F a,t : F

a,¬t

F a,t

( ) V =T C f,cr C f,1 1 + C f,2 ( ) V = T a,t C f,cr C f,1 1 + C f,2 a,¬t

(4.9) (4.10)

where C f,cr , C f,1 and C f,2 are parameters of the engine model. From F a,¬t and F a,t an absolute or relative benefit of the advanced airfoil technology in terms of fuel consumption can be easily derived.

4.8.6 Conclusion on Fuel Consumption Evaluation Through Aircraft Trajectories A simulation model can be used for choosing relevant conditions for CFD computations. Analytical solution to least squares problem is computationally efficient for deriving aircraft models that can be used in simulation. Those results open the way to future works. Indeed, using again the simulation model with polar parameters of aircraft with and without a technology to be assessed, a benefit in terms of fuel consumption could be computed. Therefore, this methodology allows evaluation of the fuel’s consumption reduction given the performances obtained through the SMS project in ongoing and near-future studies.

4.9 Conclusions Work Package 4—WP4 The following gains in the aerodynamic efficiency (lift-to-drag ration) have been obtained by the Hi-Fi simulations. They have been confirmed and consolidated by the final experiments (Chap. 5) and the overall final performances are presented in the general conclusions of the SMS project, at the end of the document. Thanks to this study, the optimal ranges of vibration frequencies and amplitudes realized by the piezoactuators near the trailing edge have been determined through

4 High-Fidelity Numerical Simulations

153

detailed Hi-Fi numerical simulations. These elements accompanied in synergy the experiments discussed in Chaps. 2 and 5. By means of Hi-Fi numerical simulations, an aerodynamic efficiency increase in the order of 5% has been obtained through the vibration and slight deformation near the trailing edge for the sRS prototype. In cruise conditions, by means of piezoactuation near the trailing edge applying a (300–350) Hz vibration on the tRS prototype, an 8.9% of drag reduction has been obtained and an aerodynamic efficiency increase (lift-to-drag ratio) of 5,5%. These performances are beyond current limits in the state of the art and quite important for fuel consumption and greening because the cruise phase is the most lasting phase of the flight. For the high-lift wing-flap configuration (LS prototype), an increase of the liftto-drag ratio of 10% in take-off has been obtained in high Reynolds number (7 Millions). For the landing high-lift configuration, Camber 1 has proven ability to simultaneously decrease the drag by 5% and increase the lift, yielding a lift-to-drag increase in the order of 6%. It is worth noticing that in other current and previous studies, increase of camber is accompanied by increase of drag, an issue removed in the SMS project thanks to the optimal shape of the flap’s camber 1. The hybrid morphing shows an increase in the aerodynamic efficiency by 5% comparing to only cambering. A significant reduction of the noise sources in the order of 10% has been obtained, obtained through a reduction of the predominant instabilities and the overall spectral energy level. The benefits on the whole A320 aircraft concerning the optimal cambering are of 2.5% and those by the hybrid morphing have provided a further increase by 1% by adding the flapping motion.

References 1. M. Braza, R. Perrin, Y. Hoarau, Turbulence properties in the cylinder wake at high Reynolds number. J. Fluids Struct. 22, 757–771 (2006). https://doi.org/10.1016/j.jfluidstructs.2006. 04.021 2. R. Bourguet, M. Braza, G. Harran, R. El Akoury, Anisotropic organised eddy simulation for the prediction of non-equilibrium turbulent flows around bodies. J. Fluids Struct. 24, 1240–1251 (2008) 3. D. Szubert, F. Grossi, A. Jimenez-Garcia, Y. Hoarau, J.C. Hunt, M. Braza, Shock-vortex shearlayer interaction in the transonic flow around a supercritical airfoil at high Reynolds number in buffet conditions. J. Fluids Struct. 55, 276–302 (2015). https://doi.org/10.1016/j.jfluidstructs. 2015.03.005 4. F. Menter, Zonal two-equation k-w turbulence models for aerodynamic flows. AIAA 2906 (1993) 5. F. Walaffe, The 3D Instability of a Strained Vortex and Its Relation to Turbulence. (Cambridge, M.A., 1989) 6. B.J. Bavly, Three-dimensional instability of elliptical flow. Phys. Rev. Lett. 57, 2160–2163 (1986)

154

A. Marouf et al.

7. P. Saffman, Vortex Dynamics (C. U. Press, Ed., 1992) 8. N. Simiriotis, G. Jodin, A. Marouf, P. Elyakime, Y. Hoarau, J.C. Hunt, J.F. Rouchon, M. Braza, Morphing of a supercritical wing by means of trailing edge deformation and vibration at high Reynolds numbers: experimental and numerical investigation. J. Fluids Struct. 91 (2019). https://doi.org/10.1016/j.jfluidstructs.2019.06.016 9. G. Jodin, V. Motta, J. Scheller, E. Duhayon, C. Döll, J.F Rouchon, M. Braza, Dynamics of a hybrid morphing wing with active open loop vibrating trailing edge by time-resolved PIV and force measures. J. Fluids Struct. 74, 263–290 (2017). https://hal.archives-ouvertes.fr/hal-016 38290 10. M. Braza, D. Faghani, H. Persillon, Successive stages and the role of natural vortex dislocations in three-dimensional wake transition. J. Fluid Mech. 439, 1–41 (2001) 11. N. Simiriotis, Numerical Study and Physical Analysis of Electroactive Morphing Wings and Hydrodynamic Profiles at High Reynolds Number Turbulent Flows. PhD Thesis (Toulouse INP, 2020) 12. J. Scheller et al., A combined smart-materials approach for next-generation airfoils. Solid State Phenom. 251, 106–112 (2016) 13. Z. Lyu, J.R. Martins, Aerodynamics shape optimisation of an adaptive morphing trailing edge wing, in 15th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Atlanta, GA (2014) 14. A. Nemili, E. Özkaya, N. Gauger, F. Kramer, F. Thiele, Discrete adjoint based optimal active control of separation on a realistic high-lift configuration, in New Results in Numerical and Experimental Fluid Mechanics X, ed. by A. Dillmann, G. Heller, E. Kramer, C. Wagner, C. Breitsamter (Springer International Publishing, 2016), pp. 237–246 15. W. Shen, J. Michelsen, J. Sorensen, Improved Rhie-Chow interpolation of unsteady flow computations. AIAA J. (2012) 16. L. Hascoet, V. Pascual, The Tapenade automatic differentiation tool: principles, model and specification. ACM Trans. Math. Softw. 39(3) (2013) 17. F. Kramer, M. Fuchs, T. Knacke, C. Mockett, E. Özkaya, N. Gauger, F. Thiele, Fast sensitivity analysis for the design of morphing airfoils at different frequency regimes, Critical flow dynamics involving moving/deformable structures with design applications, in Notes on Numerical Fluid Mechanics and Multidisciplinary Design, ed. by M. Braza, K. Hourigan, M. Triantafyllou (Springer, 2021), vol. 147, pp. 505–516. 18. J. Tô, Etude numérique et analyse physique du morphing de profils d’aile de type Airbus A3XX en régime transsonique par l’approche de modélisation de la turbulence "Organised Eddy Simulation" à nombre de Reynolds élevé, PhD Thesis (Toulouse, France, 2021) 19. A. Bouhadji, M. Braza, Organised modes and shock-vortex interaction in unsteady viscous transonic flows around an aerofoil, Part I: Mach number effect. Comput. Fluids 32(9), 1233– 1260 (2003) 20. A. Jimenez-Garcia, Etude du phénomène de tremblement (Toulouse, France, 2012) 21. Y. Hoarau, D. Pena, J. Vos, D. Charbonier, A. Gehri, M. Braza, T. Deloze, E. Laurendeau, Recent developments of the Navier-Stokes Multi Block (NSMB) CFD solver, in 54th AIAA Aerospace Sciences Meeting (American Institute of Aeronautics and Astronautics, San Diego, 2016) 22. A. Marouf, Analyse physique de concepts du morphing électroactif pour accroître les performances aérodynamiques des ailes du futur par simulation numérique de Haute Fidélité et modélisation de la Turbulence à nombre de Reynolds élevé, PhD Thesis (Université de Strasbourg, 2020) 23. M. Schafer, M. Strohmeier, V. Lenders, I. Martinovic, M. Wilhelm, Bringing up OpenSky: A large-scale ADS-B sensor network for research, in 13th International Symposium on Information Processing in Sensor Networks (2014) 24. J. Sun, J. Hoekstra, J. Ellerboek, Aircraft drag polar estimation based on a stochastic hierarchical model, in Eighth SESAR Innovation Days (University of Salzburg, Austria, 2018) 25. Eurocontrol: User Manual for the Base of Aircraft Data (BADA) [Online]. Available: https:// www.eurocontrol.int/sites/default/files/library/022_BADA_User_Manual.pdf. Accessed 2004

Chapter 5

Aerodynamic Evaluation F. Auteri, P. Flaszynski, A. Savino, A. Zanotti, G. Gibertini, D. Zagaglia, Y. Bmegaptche-Tekap, D. Harribey, J. F. Rouchon, P. Kaczynski, P. Doerffer, M. Piotrowicz, R. Szwaba, J. Telega, T. Louge, J. B. Tô, C. Jimenez-Navarro, A. Marouf, and M. Braza

Abstract This chapter is devoted to the aerodynamic evaluation to experimentally investigate first the flow structure and the morphing effects on the tRS (transonic Reduced Scale) prototype of 15 cm chord, and secondly to evaluate the impact of the optimal morphing cambering on the LS (Large Scale) prototype in respect of the aerodynamic performance increase. The first part concerning the transonic regimes involves morphing through optimal trailing edge region vibration performed in the wind tunnel of IMP-PAN concerning cruise phase conditions and the second investigates the aerodynamic perfomance through cambering of the high lift flap of 1 m chord and 2 m span, emebedded in the fixed wing configuration of 4 m span in the POLIMI’s wind tunnel in take-off and landing conditions.

F. Auteri (B) · A. Savino · A. Zanotti · G. Gibertini · D. Zagaglia POLIMI, Politecnico di Milano, Piazza Leonardo da Vinci, 32, 20133 Milano, Italy e-mail: [email protected] P. Flaszynski · P. Kaczynski · P. Doerffer · M. Piotrowicz · R. Szwaba · J. Telega IMP-PAN—The Szewalski Institute of Fluid-Flow Machinery, Polish Academy of Sciences, Fiszera 14st., 80-231 Gdansk, Poland Y. Bmegaptche-Tekap · J. B. Tô · C. Jimenez-Navarro · A. Marouf · M. Braza INPT—Institut National Polytechnique de Toulouse/IMFT—Institut de Mécanique des Fluides de Toulouse, 2 Allée du Professeur Camille Soula, 31400 Toulouse, France D. Harribey · J. F. Rouchon LAPLACE—Laboratoire Plasma et Conversion d’Energie, INPT—Institut National Polytechnique de Toulouse, Site of ENSEEIHT, 2, Rue Charles Camichel, 31071 Toulouse, France T. Louge CALMIP—Calcul en Midi Pyrénées, Espace Clément Ader, 3 Rue Caroline Aigle, 31400 Toulouse, France A. Marouf Laboratoire des Sciences de l’Ingénieur, de l’Informatique et de l’Imagerie, ICUBE, Université de Strasbourg, 4 Rue Blaise Pascal, 90032, 67081 Strasbourg, France

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. Braza et al. (eds.), Smart Morphing and Sensing for Aeronautical Configurations, Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153, https://doi.org/10.1007/978-3-031-22580-2_5

155

156

F. Auteri et al.

Work Package 5—WP5

5.1 TRS Prototype Aerodynamic Evaluation P. Flaszynski, P. Kaczynski, P. Doerffer, M. Piotrowicz, R. Szwaba and J. Telega

5.1.1 Transonic Reduced Scale Profile—Reference Case 5.1.1.1

Transonic Wind Tunnel

The scheme of wind tunnel is shown in Fig. 5.1. The tunnel is a blow down type with ambient conditions at the inlet. The tunnel working scheme is as follows: vacuum pumps evacuate the air from two vacuum tanks (total volume 120 m3 ) down to 95% vacuum. After reaching the desired pressure, the tunnel is ready for blow down. The test is realized by opening a valve (electrically controlled), thus connecting the tanks with the ambient air. The air, driven by the pressure difference, is being sucked from the atmosphere into the tanks flowing through the test section. Such operation secures constant conditions during the test. The air flowing into the tunnel is being driven through a layer of silica gel to extract humidity.

Fig. 5.1 Transonic wind tunnel scheme

5 Aerodynamic Evaluation

5.1.1.2

157

Experimental Setup

In order to investigate the 2D flow structure on the aircraft wing profile in transonic regime, a reduced scale model of chosen cross section of the Airbus A320 wing was designed and assembled in the wind tunnel. The model can reproduce the flow structure, pressure distribution and boundary layer development similar to the obtained on a reference free stream model. The test section (Fig. 5.2) shows a wing profile model (which shape was delivered by IMFT Toulouse and Airbus, Fig. 5.3) placed in the appropriate location in channel flow. The test section consists of profile and limiting upper and lower walls. The shape of the limiting walls and also the position of the wing was designed by means of CFD for the provided specific inflow conditions. The lower and upper wall profiles are extracted streamlines from free stream configuration in order to keep a similar flow structure as in the reference flow. The main objective of the investigation is to study details of flow structure, shock wave oscillations and wake profile at cruise conditions with the focus on future optimization of lift-to-drag ratio and drag reduction. The results presented comprise the two steps: the first one is focused on the measurements for the reference profile without flapping trailing edge, and the second one—measurements for profile with flapping trailing edge (delivered by IMFT). The inlet Mach number 0.78 corresponds to the cruise speed of the A320. Angle of attack was modified as follows: α = 1.8, 2.0, 2.2 and 2.4°. The Reynolds number for this flow conditions is 2 × 106 based on the profile chord which is c = 150 mm. The Mach number upstream of the shock wave was ~ 1.2. As the experiments in next step will concern the investigations with flapping trailing edge, steady and unsteady measurement techniques were applied. For the Fig. 5.2 View of the reduced scale test section

158

F. Auteri et al.

Fig. 5.3 View of the wing profile details

static pressure measurements (steady) a pressure scanner PSI9010 was used, but for unsteady pressure Kulite transducers were used. The velocity distribution in the wake and its unsteadiness were measured by means of LDA (Laser Dopler Anemometry). The optical instruments (details provided in next section) in form of schlieren system to register the shock wave unsteadiness and its topology was used. Summarizing, the following measurements were carried out to investigate some aspects of the flow around A320 wing profile: • • • • • • • •

Inlet stagnation parameters (pressure and temperature) Schlieren visualization Shock oscillations measurements Static pressure at inlet on lower wall Static pressure on suction and pressure sides of the wing Unsteady static pressure on suction side of the wing Inlet and wake velocity measurements by means of LDA Lift and drag of the profile by means of two aerodynamic balances in sidewalls.

Figure 5.4 shows the results of the numerical simulations and velocity measurements upstream of the tRS wing profile. The traverse is located 25 mm upstream of the airfoil leading edge and coordinate Y = 0 is placed in the leading edge centre. It can be seen that the upstream influence of the profile predicted by numerical simulation agrees very well with experimental data.

5.1.1.3

Pressure Coefficient and Pressure Unsteadiness Near the Shockwave

The distribution of the measured pressure coefficient Cp and isentropic Mach number for the reference case (AoA = 1.8°) together with numerical results are shown in Figs. 5.5 and 5.6. The pressure taps location corresponds to the markers on the plot.

5 Aerodynamic Evaluation

159

Fig. 5.4 Velocity distribution upstream of the tRS wing profile

The location of shock wave on the profile is indicated by Cp jump or Mach number drop on suction side at chord x/c = 0.67–0.7. One can notice that CFD simulations agree very well with experimental data. Small difference in the area of the shock wave (approx. x/c = 0.7) results from its oscillations noticed during experimental investigations. This affects the smoothing of pressure measurement in the shock vicinity, while steady simulations are carried out. Figures 5.7, 5.8 and 5.9 show the pressure coefficient distribution for the higher angles of attack, namely for AoA = 2.0, 2.2 and 2.4°. A comparison of the numerical simulations with experimental data for particular cases is shown on each plot. As one can notice, a very good agreement between the CFD and experiment is noticed, except for the position of shock wave, due to the confined experimental environment. In the experiment the shock wave is located more downstream than predicted by the

Fig. 5.5 Pressure coefficient along the wing profile, AoA = 1.8°

160

F. Auteri et al. 1.4

M [-]

1.2

1.0

CFD, AoA - 1.8

0.8

EXP, AoA - 1.8 0.6 0.0

0.2

0.4

0.6

0.8

1.0

x/c [-] Fig. 5.6 Mach number distribution along the wing profile, AoA = 1.8°

Fig. 5.7 Pressure coefficient along the wing profile, AoA = 2.0°

simulations. At the highest angle of attack, larger discrepancies are noticed between these two methods. Moreover it can be observed the characteristic pressure smoothing at shock wave vicinity as it was in previous flow case, i.e. for AoA = 1.8°.

5.1.1.4

Unsteady Pressure on the Profile

Measurements of pressure fluctuations at two Kulite locations are shown through the Power Spectral Density (PSD) plot in Figs. 5.10 and 5.11. The Kulite transducers are located: number 1 just upstream of the shock wave at x/c=0.57, number 2 close to trailing edge at x/c=0.9. The unit of the PSD amplitude is [hPa2 /Hz]. Pressure fluctuations presented for angle of attack 1.8 indicate higher oscillations at near the shock wave.

5 Aerodynamic Evaluation

161

Fig. 5.8 Pressure coefficient along the wing profile, AoA = 2.2°

Fig. 5.9 Pressure coefficient along the wing profile, AoA = 2.4°

5.1.1.5

Schlieren Visualisation and Shock Unsteadiness

Schlieren visualizations for the reference flow conditions and at different angles of attack (AoA = 1.8°–2.4°) are shown in Figs. 5.12, 5.13, 5.14 and 5.15. These figures show flow visualisation above the rear part of the profile, starting from midchord (at X/C ≈ 0.5) towards the trailing edge. As one can see in Fig. 5.12 that the shock wave pattern at relative low upstream Mach number M = 1.2 is linear. The λ-foot is so small that it is not clearly visible close to the wall. Such a shock topology indicates that the shock wave boundary layer interaction is of turbulent type and the boundary layer upstream of the shock is turbulent.

162

F. Auteri et al.

Fig. 5.10 Power spectral density of pressure oscillations for Kulite-1 2.0E-06 1.8E-06 1.6E-06 1.4E-06

PSD

1.2E-06 1.0E-06 8.0E-07 6.0E-07 4.0E-07 2.0E-07 0.0E+00 1

10

100

F [Hz] Fig. 5.11 Power spectral density of pressure oscillations for Kulite-2

1000

10000

5 Aerodynamic Evaluation

163

Fig. 5.12 Schlieren visualisation, AoA = 1.8°

Fig. 5.13 Schlieren visualisation, AoA = 2.0°

When the angle of attack increases, the shock wave moves towards trailing edge and its height grows significantly, which is the effect of the extension of the local supersonic zone (see Figs. 5.13, 5.14 and 5.15).

5.1.1.6

Shock Wave Unsteadiness

The shock oscillations were investigated by the recorded schlieren visualization and later image processing. The movie was recorded using a fast CCD camera and then single frames were extracted. The shock oscillations were analysed by the tracking of the intersection point of the shock at a selected line location above the profile. The location of this line for the shock oscillations measurement is shown in Fig. 5.16.

164

F. Auteri et al.

Fig. 5.14 Schlieren visualisation, AoA = 2.2°

Fig. 5.15 Schlieren visualisation, AoA = 2.4°

The distance from the profile wall to the measurement line was about 10 mm. The sampling frequency was equal to 1.5 kHz. The result of shock oscillations measurement is shown in form of the power spectral density (PSD) plot in Fig. 5.17. The unit of the PSD amplitude is [mm2 /Hz]. One can see in Fig. 5.17 some dominant peaks in range of frequency up to 10 Hz.

5 Aerodynamic Evaluation

165

10 mm

Fig. 5.16 The location of CCD camera measurements

Based on the previous measurement experience it is usually connected with pressure pulsation at the channel outlet. In case of shock oscillations one can distinguish one characteristic frequency of about 20 Hz. Approximately above the 100 Hz, the PSD of shock wave oscillation decreases significantly.

Fig. 5.17 Power spectral density of shock oscillations

166

5.1.1.7

F. Auteri et al.

Wake—Velocity and Fluctuations

The wake measurements are compared with numerical simulations. The measurement trawers for the LDV pointwise measurements is located 7 mm downstream of the profile and the coordinate Y = 0 is fixed at the trailing edge. Figure 5.18 shows the comparison of the velocity in the wakes between the CFD and the experiment for different angles of attack. It can be said that the velocity profile in experimental investigations characterizes similar thickness as in simulations, however in the centre of the wake a lower velocity was predicted. This is an expected result, because numerical simulations are quite often characterized by a larger velocity decrease at this point. As the angle of attack increases the wake thickness is also enlarged.

Fig. 5.18 Velocity distribution across the wake

Fig. 5.19 Vertical velocity component distribution across the wake

5 Aerodynamic Evaluation

167

15 EXP, AoA - 1.8, REF

10

EXP, AoA - 2.4, REF

Y [mm]

5 0 -5 -10 -15

0

10

20

30

40

50

60

U - RMS [m/s] Fig. 5.20 Velocity fluctuations across the wake

Figure 5.19 shows the Y velocity component distribution in the wake and the velocity fluctuation in form of RMS value (u’) is presented in Fig. 5.20. The Y velocity component in the area beneath the wake shows the value close to zero, what means the flow is nearly parallel to the lower wall. On the upper side of the wake (suction side of profile) the flow is directed toward the lower wall and the Y velocity becomes negative, about 10% of the longitudinal velocity component. In terms of CFD results, some discrepancies in comparison with measurements can be observed. The Y velocity component is lower (absolute value) which means a slight difference in flow direction. However, the trends are predicted properly. The velocity fluctuations increase with higher angle of attack, as shown in Fig. 5.20. The differences are clearly visible on the suction side of the wake as the effect of a stronger shock and higher fluctuations of the shock wave.

5.1.2 Transonic Reduced Scale Profile with Flapping Trailing Edge The results presented in the previous section refer to the reference profile manufactured in the first phase of the SMS project to adjust settings of the test section and to carry out investigations of the flow structure without flow control device, i.e. without flapping trailing edge. Afterwards, the actuators were developed and the profile with control of trailing edge oscillations was manufactured by INPT/LAPLACE. After the profile was delivered to IMP-PAN, it was mounted in the transonic wind tunnel and the measurements were repeated to assess if the flow characteristics are the same as for the reference profile, without control. The results of these tests are presented below in the form of velocity distribution in the wake and schlieren visualizations. The results of IMP PAN profile are marked as REF and results of the profile with

168

F. Auteri et al.

Fig. 5.21 The location of vibration measurement

control as OTE (Oscillating Trailing Edge), according to vibration frequencies and amplitudes provided by the Hi-Fi simulations of INPT/IMFT.

5.1.2.1

Trailing Edge Vibrations

Before the measurement campaign have been started, the trailing edge oscillations were measured. The main objective was to check and confirm that the prescribed signal provided to the actuators is transformed into the required trailing edge oscillations and no damping effect appears. The vibrations were measured at two points (Fig. 5.21): at the trailing edge and on the profile to check if the vibrating trailing edge does not excite the profile and measurement system. The measurements have been done with and without flow in the test section to assess damping effect of the aerodynamic forces. The results are presented in Fig. 5.22 and in Table 5.1. The most important conclusions are that the frequency and amplitude of the trailing edge oscillations are not affected by the flow and that the wing vibrations are negligible.

5.1.2.2

Schlieren Visualisation and Shock Unsteadiness

Figures 5.23 and 5.24 show differences in the wake width at the location 7 mm downstream of the trailing edge. For the profiles with flow control, a higher velocity deficit was obtained in the centre of the wake, while their thickness is similar. This may be the result of a different shock wave pattern on the suction side of the profile due to the implemented actuators, see Fig. 5.25. Additional waves appear due to surface quality of the profile that is as the effect of manufacturing. The surface imperfections exist in the supersonic area, which is very sensitive to inaccuracies. The focus of this study is on the investigation of the influence of flow control by means of an oscillating trailing edge on the flow characteristics, in the next sections only results for the profile with flapping trailing edge will be presented.

5 Aerodynamic Evaluation

169

Fig. 5.22 Velocity of trailing edge in frequency domain

Table 5.1 Results of vibration measurements TE

Profile

Flow

Without flow

Flow

Without flow

Amplitude

[mm]

0.046

0.049

0.0012

0.0022

Peak-to-peak amplitude

[mm]

0.092

0.098

0.0024

0.0043

15 10

Y [mm]

5

EXP, AoA - 1.8, REF EXP, AoA - 2.4, REF EXP, AoA - 1.8 EXP, AoA - 2.4

0 -5

-10 -15 50

100

150

200

Ux [m/s]

300

Fig. 5.23 Comparison of the velocity component in the wake for reference profile and RS profile

5.1.2.3

Wake—Velocity and Fluctuations

The research on the effect of the morphed trailing edge on the velocity profile in the wake were carried out with different frequencies of TE flapping and the same

170

F. Auteri et al. 10 5

Y [mm]

0 -5 -10 -15 -20

EXP, AoA - 1.8, REF EXP, AoA - 2.4, REF EXP, AoA - 1.8 EXP, AoA - 2.4 -30

-20

-10

Uy[m/s]

0

10

20

Fig. 5.24 Comparison of the vertical velocity component in the wake for reference profile and RS profile

Fig. 5.25 Schlieren visualisation for angle of attack 1.80◦ and 2.40◦

amplitude for all flow cases. These measurements were performed with frequencies from 60 to 360 Hz as shown in Table 5.2. For higher angle of attack only one frequency (f = 300 Hz) was used. According to INPT/IMFT numerical simulations, this is a value corresponding to an optimal range. The vibration peak to peak amplitude as shown in Table 5.1 was around 0.1 mm.

5 Aerodynamic Evaluation Table 5.2 Measurements flow case

171 AoA 1.8°

2.4°

Flow case

f [Hz]

No oscillations

0

TE oscillations

60

TE oscillations

120

TE oscillations

200

TE oscillations

300

TE oscillations

360

No oscillations

0

TE oscillations

300

The plots presented below show the results of velocity measurement in the wake for the reference and for the flow control cases. Figures 5.26 and 5.28 present results for an angle of attack (AoA) 1.8°, Fig. 5.29 for AoA 2.4°. The comparison of distributions of controlled cases with the reference one for the longitudinal (X) velocity component is shown in Fig. 5.26. In this plot almost all distributions coincide very well except for the 300 Hz flow case which indicates decrease of wake width. For this case it can be seen that the oscillating TE causes a slight decrease of the velocity deficit area in the wake downstream of the airfoil, that leads to a drag reduction. The wake width decrease after TE oscillations with a frequency of 300 Hz are applied also affects the reduction of instabilities, which can be noticed in Fig. 5.27 by a reduced RMS in the wake for this case.

Fig. 5.26 Velocity distributions across the wake for various flow cases

172

F. Auteri et al.

Fig. 5.27 Velocity fluctuations across the wake for various flow cases

The distribution of Y velocity component (Fig. 5.28) for the flow control cases shows a similar behaviour as the reference case. The only difference can be noticed on the upper side of the wake (around Y = 2.5), where an additional peak appears measured as a sudden velocity change. The probable reason of this local maxima is the interaction of flapping trailing edge with the freestream flow. However, this effect still needs further clarification. As already mentioned, a slight reduction of the wake size was observed for the 1.8 angle of attack when the TE oscillations of 300 Hz are applied. The same effect can also be seen for the higher angle of attack AoA = 2.4° and for the same oscillating frequency of the trailing edge, Fig. 5.29.

Fig. 5.28 Vertical velocity component distributions across the wake

5 Aerodynamic Evaluation

173

Fig. 5.29 Velocity distributions across the wake for cases of AoA = 2.40◦

5.1.3 Conclusions for the Transonic Prototype tRS A new test section has been developed, designed and implemented in the transonic wind tunnel with a relative narrow measurement channel. In the new test section, a flow pattern similar to the reference two-dimensional freestream flow is reproduced, what allows for investigations of flow structure on the A320 wing profile in cruise conditions. The airfoil was manufactured and equipped with two balances and systems required for the aerodynamic measurements. The following conclusions can be drawn from the above described investigations: • The flow field contains characteristic features of the flow over the reference twodimensional airfoil in freestream, i.e. pressure distribution and shock location; • Velocity distributions both on the profile and in the wake show very good agreement between the numerical simulations and the measurements; • Surface flow structure, surface friction coefficient and velocity distribution over the airfoil surface show that flow can be considered as “two-dimensional” on the profile—very weak sidewall effects and corner vortices; • Measurements with the actuated trailing edge show the potential for aerodynamic performance improvement: the most promising effects were achieved when the trailing edge flapping at 300 Hz is applied: – Lift coefficient reduction by 1.1% for AoA 1.8° and increase by 0.9% for AoA 2.4° – Drag coefficient reduction by 4% for AoA 1.8° and 1.6% for AoA 2.4° – Lift/Drag coefficient increase by 3% for AoA 1.8° and 2.5% for AoA 2.4°

174

F. Auteri et al.

5.2 Large Scale Cambered Prototype Design and Experimental Evaluation F. Auteri, A. Savino, A. Zanotti, G. Gibertini, D. Zagaglia, Y. Bmegaptche-Tekap, D. Harribey, J. F. Rouchon and M. Braza Due to the large size of the model, having a chord of approximately 2.4 m, particular care has been put in designing the wing model that will host the flap and the procedures that will be necessary to test the model in the wind tunnel. The following study is summarized in the following points: • Design and construction of the wing hosting the LS prototype Model. In particular, design, sizing, static and dynamic structural verification exploiting aerodynamic loads derived from CFD simulations (CFSE) and a simple potential model; • Preparation of test set-up procedures, model handling and assembly in the test section; • Definition of the test matrix; • Installation of sensors and data acquisition system; • Wind-tunnel testing activity; • Post-processing of the experimental results.

5.2.1 Overview and Specifications The present study provides a description of the Large Scale prototype construction concerning the high—lift configuration measured in the wind tunnel of POLIMI for take-off and landing conditions. The CAD and structural elements of the morphing high-lift flap are provided together with the characteristics and construction of the Electromechanical actuators (EMA). The camber control and the related interface are provided in detail, as well as the aerodynamic measurements showing the behaviour of the LS design and the final benefits.

5.2.2 Problem Description The goal of the experimental evaluation of the LS prototype is to measure the performance of the wing with the installed flap in order to verify the actual improvement of the airfoil performance obtained by morphing the flap. By virtue of the large size of the test section of POLIMI large wind tunnel (GVPM) and of its quite high velocity, high Reynolds numbers can be reached during testing that can give indications on the performance of the flap in the full-scale application. Moreover, experimental measurements will be useful for comparison with simulations, in order to assess

5 Aerodynamic Evaluation

175

the capability of numerical simulations in capturing correctly the flow properties observed experimentally. The present activity has been organised in six steps: 1. Design, construction and assembly of the wing model that will host the LS prototype in the wind tunnel. This means sizing and geometrical modelling of the wing model, static and dynamic structural verification; 2. Definition of the test matrix, that includes assigning a priority to each test point; 3. Design of the testing activity, defining sensor layout and all the operational procedures needed to carry out the activity; 4. Installation and testing of the sensor on the wing model and LS prototype; 5. Wind-tunnel tests; 6. Post-processing of the results and final reporting.

5.2.3 Design and Construction of the Wing Model Particular care was needed in designing the wing model and, more generally, the wind-tunnel tests owing to the large size of the model with respect to the size of the test section, which is rather unusual. In particular, the wing is expected to produce very high loads, and since rather high angles of attack must be tested, flow separation could occur leading to strong vibrations. In this section the design activity is described in detail. We start from the description of the test facility, the GVPM wind tunnel, then the preliminary investigation activity necessary to check the compatibility of such a large wing model with the wind tunnel is described. Then, the model sizing and design is discussed with particular emphasis on the static and dynamic structural verification. Finally, the construction and assembly of the wing model is described.

5.3 GVPM Wind Tunnel GVPM is a special closed-circuit wind tunnel, arranged in a vertical layout with two test rooms located on the opposite sides of the loop (Fig. 5.30) [1]. The first one is located in the lower part of the loop and is suitable for Low Turbulence tests [2]. The second one, bigger, is located in the upper part of the loop and is intended for civil engineering testing (the Boundary Layer Test Section). Due to this unique feature, GVPM offers the widest possible range of test arrangements and alternatives. The facility is powered by a flow generator array of 14 1.8 m diameter, 100 kW, fans, for a total power of 1.4 MW. The fans are organized in two rows of seven 2 × 2 m independent cells. Independent inverters drive the fans allowing for continuous control of the rotation speed of each fan to obtain the desired wind speed in the test section. After the fans, two corners fitted with vanes drive the flow to the upper level of the facility in the opposite direction. The flow is then cooled by a heat exchanger that is placed just downstream of bend number 2 and, after a grid, enters the boundary layer test section. A second set of two corners fitted with vanes leads to the lower

176

F. Auteri et al.

Fig. 5.30 Wind tunnel layout

level where, after a 2-m-long settling chamber, it passes a honeycomb screen and a set of three different porosity wire nets to reduce axial and lateral turbulence and to promote a more uniform axial flow. A two-dimensional contraction cone with area ratio 3.46:1 reduces the duct section to fit the low turbulence test section size. Finally, a short diffuser expands the duct section back to the size of the fans array. GVPM offers different test configurations: • 4 m wide × 3.84 m high × 6 m long Low Turbulence Test Section, max wind speed 55 m/s (Fig. 5.2) • 4 m wide × 3.84 m high × 5 m long Open Jet Test Section, max wind speed 55 m/s • 13.84 m wide × 3.84 m high × 35 m long Boundary Layer Test Section, max wind speed 16 m/s The Low-Turbulence test section, (used for this specific project) is 4 m wide, 3.84 m high and 6 m long. It is possible to perform tests in a closed test section and in an open jet. The maximum wind velocity is 55 m/s and the turbulence level is less than 0.1%. It is possible to remove the whole section from the airflow circuit allowing for an off-line setting-up of the test. There are two inter-changeable test sections to prepare a new experiment while the wind tunnel is running. The optimization and check of the facility layout and components design has been achieved with the help of a 1:9 scale powered tunnel model that was modified several times up to the final design. This model is currently available and has been used for a preliminary analysis of the model through a reduced scale wing/flap model during the SMS project [3].

5 Aerodynamic Evaluation

177

Fig. 5.31 Test section dimensions

5.4 CAD Drawings for the High-Lift Flap 5.4.1 Morphing Wing Concept 5.4.1.1

Specifications

The proposed morphing concept is applied to the high-lift flap designed by INPT/IMFT–LAPLACE and NOVATEM concerning a take-off /landing configuration. The flap’s profile has been adapted from Airbus specifications. The flap has a 1 m chord and a 2 m span. Based on an Airbus commercial aircraft (A320), this section deals with the morphing flap design. The flap’s profile (Fig. 5.32) has been adapted from industrial specifications: its chord is 1 m and its span is 2 m. A chordwise loading is specified, equivalent to 1.5 t of aerodynamic upward forces. An idea of the force distribution is represented in Fig. 5.34. The morphing flap is based on articulated ribs where actuators control the rotations of the elements around the hinges. The internal structures represented by the articulated ribs consist in an engineered mechanical structure composed of ribs and spars. In Fig. 5.33, a schematic view of the morphing flap’s CAD is presented.

178

F. Auteri et al.

Fig. 5.32 Schematic illustration of the wing’s shape deformation and of the two-element high-lift wing configuration

5.4.1.2

Articulated Concept

The morphing flap is based on an articulated rib where actuators control the rotations of the elements around the hinges. The proposed concept is presented in Fig. 5.34. • Articulated ribs define the geometry and carry the other components. They have to withstand the internal and external (i.e. aerodynamic) forces, whilst being low weight.

5 Aerodynamic Evaluation

179

Fig. 5.33 Morphing flap CAD—schematic view

• Hinges allow the rotation of the articulated ribs. Parts of forces are transmitted through these components without generating much parasitic force (or torque) when rotated. • Actuators are devices that transmit mechanical energy to the structure. The actuators are responsible for the shape control and have to counteract aerodynamic forces as well as internal forces coming from the other components. • Skins or covering devices guarantee the airtightness of the wing, transmit the aerodynamic forces to the structure and ensure a smooth shape during morphing. The skin must endure deformation without unexpected displacements like bumps or wrinkles. Additionally, mechanical stops are provided to limit the rotations of the articulations, thus preventing overloads in actuators. The internal structures represented by the articulated ribs actually consist in an engineered mechanical structure composed of ribs and spars. Excluding the skin and the actuators, the inner matter is only dedicated to hold the components and transmit the forces from the skin to the spar, through the actuators and hinges. This means that we neglect dynamic and thermal inertial effects of the inner structure. Considering a given structure topology (e.g. ribs, spar stiffeners), and because strength of material relies on the sections and volume of inner matter to withstand forces (e.g. maximal stress), lowering the inner structure forces decreases the required matter so the weight. From an industrial point of view, the torsion resistance along the span direction is critical issue for commercial aircraft design. By the way, torsion resistance for flap is not a crucial function; the torsional stiffness of the wing comes from the main wing box, the flap is neglected. The flap’s main solicitations come from the resistance to spanwise bending moment. For the developed application, the articulations axes are spanwise, therefore the spanwise

180

F. Auteri et al.

Fig. 5.34 Mechanism of the articulated flap

bending stiffness is not compromise. Regarding the torsion resistance, the consider flap is cut in multiple wing-boxes linked together by the actuators and the hinges. All the torsion forces are transmitted through the actuators, and these forces are parts of the force specifications. Gliding bearings (plain bearings with steel-TEFLON contacts) have been selected for the hinge function. They are suitable for low rotation velocities. They are compact, lightweight and generate low friction torque. The proposed actuators consist in cylinder like actuators. Composed of SMA wires or EMA actuators, they are able to pull on the articulated ribs, thus imposing the rotations. Different skin technologies can potentially address the previous specifications.

5.4.1.3

Skeleton of the Structural Model

The structure consists of 14 ribs in ALUMINUM 2017A and a main beam in ALUMINUM 6060 T6. The dimensions of this flap are: 1 m of chord and about 2 m of span. The skeleton of the morphing flap is presented in Fig. 5.35.

5.4.1.4

Loading, Material and Boundary Conditions

Material The flap consists essentially of parts made of aluminum: in ALUMINUM 2017A for Ribs and 6060 T6 for main beam (Table 5.3).

5 Aerodynamic Evaluation

181

Fig. 5.35 Skeleton of the morphing frap

Table 5.3 Material properties of aluminum 6060 T6 Elasticity modulus

Poisson’s ratio

Density

Tensile yield stress

Compressive yield stress

70

0.33

2.7e+3

170

170

Loading The aerodynamic force calculations are carried out on the basis of the results of the potential method (Tables 5.4 and 5.5). Fy = 0.5ρV 2 SClmax Fx = 0.5ρV 2 SCdmax ρ = 1.2kgm−3 ; S = 2 × 1m2 The calculated force has been distributed on the structure by applying 70% on the upper surface and 30% on the lower surface (Fig. 5.36).

Boundary Conditions Regarding the boundary conditions, we have built a flush-mounting in the ends of the spar. The different ribs are connected to each other using a joint which is modeled by a pivot link. The intermediate beams are connected to the ribs in order to transmit

182

F. Auteri et al.

Table 5.4 Aerodynamic coefficients Aerodynamics coefficients AoA

Landing

Cl

Cd

Cm

8

1.67112

0.624

− 0.0552

6

1.6536

0.5568

− 0.0528

4

1.608

0.492

− 0.0504

2

1.584

0.4272

− 0.048

0

1.536

0.36

− 0.04752

Aerodynamics coefficients Take-off

AoA

Cl

Cd

Cm

8

0.9624

0.276

− 0.0528

6

0.9144

0.24

− 0.0528

4

0.8592

0.204

− 0.0528

2

0.792

0.1728

− 0.0528

0

0.7368

0.144

− 0.0504

Table 5.5 Aerodynamics Forces for different configurations Velocity (m/s)

30

AoA Fy

35 Fx

Fy

40 Fx

Fy

45 Fx

Fy

50 Fx

Fy

Fx

Landing 8°

1804.8 673.9 2456.5 917.3 3208.6 1198.1 4060.8 1516.3 5013.4 1872.0

6°

1785.9 601.3 2430.8 818.5 3174.9 1069.1 4018.2 1353.0 4960.8 1670.4

4°

1736.6 531.4 2363.8 723.2 3087.4

944.6 3907.4 1195.6 4824.0 1476.0

2°

1710.7 461.4 2328.5 628.0 3041.3

820.2 3849.1 1038.1 4752.0 1281.6

0°

1658.9 388.8 2257.9 529.2 2949.1

691.2 3732.5

874.8 4608.0 1080.0

Take-off 8°

1039.4 298.1 1414.7 405.7 1847.8

529.9 2338.6

670.7 2887.2

828.0

6°

987.6 259.2 1344.2 352.8 1755.6

460.8 2222.0

583.2 2743.2

720.0

4°

927.9 220.3 1263.0 299.9 1649.7

391.7 2087.9

495.7 2577.6

612.0

2°

855.4 186.6 1164.2 254.0 1520.6

331.8 1924.6

419.9 2376.0

518.4

0°

795.7 155.5 1083.1 211.7 1414.7

276.5 1790.4

349.9 2210.4

432.0

the mechanical forces due to the aerodynamic actions through the structure and also to ensure the continuity of the forces of the actuators.

5.4.1.5

Linear Static Analysis Results

Using the software of the ALTAIR suite, we performed a linear static analysis of all the structure in order to calculate the displacements, deformations, stresses and reaction forces under the effect of the applied loads.

5 Aerodynamic Evaluation

183

Fig. 5.36 Loads distribution

Stresses In materials science and engineering the von Mises yield criterion can also be formulated in terms of the von Mises stress or equivalent tensile stress. This is a scalar value of stress that can be computed from the Cauchy stress tensor. In this case, a material is said to start yielding when the von Mises stress reaches a value known as yield strength. The von Mises stress is used to predict yielding of materials under complex loading from the results of uniaxial tensile tests. The von Mises stress satisfies the property where two stress states with equal distortion energy have an equal von Mises stress. The analysis has shown that the distribution of the stresses on the LS prototype are lower than the threshold fixed for the material and never reach the maximum values, which are lower than the fixed tolerance (Fig. 5.37).

Displacements The displacements are shown in Fig. 5.38. All the values are acceptable by the chosen material.

184

F. Auteri et al.

Fig. 5.37 Von mises stresses distribution

Reserve Factor A measure of strength frequently used in Europe is the ‘Reserve Factor (RF)’. With the strength and applied loads expressed in the same units, the Reserve Factor is defined in one of two ways, depending on the industry: RF = Proof Strength/Proof Load RF = Ultimate Strength/Ultimate Load The applied loads have many factors, including factors of safety applied. The display of the RF clearly shows the capacity of the structure to support the aerodynamic loads imposed (Fig. 5.39).

5.4.1.6

Modal Analysis Results

The modal analysis is of primary importance to evaluate the system dynamics, particularly it gives hints on the possibility some natural vibration modes could couple with themselves and with the external aerodynamic loads. Results from a Modal Analysis give us an insight of how the structure would respond to vibration/dynamic load by

5 Aerodynamic Evaluation

Fig. 5.38 Displacements

Fig. 5.39 Reserve factor

185

186

F. Auteri et al.

Fig. 5.40 Mode 1, f = 15.75 Hz

Fig. 5.41 Mode 2, f = 37.98 Hz

identifying the natural frequencies and mode shapes of the structure. The following plots show the first four modes and her frequencies (Figs. 5.40, 5.41, 5.42 and 5.43).

5.4.1.7

Morphing Flap Manufacturing

The manufacturing of the morphing flap has been accomplished according to the afore mentioned CAD and structural analysis in the workshop laboratory of INPT/IMFT, in collaboration with INPT/LAPLACE (Figs. 5.44 and 5.45). The model has been covered by a suitable layer of aluminum sheet allowing the cambering capabilities (Fig. 5.44-right). The Electromechanical Actuator system was described in Chap. 3.

5 Aerodynamic Evaluation

187

Fig. 5.42 Mode 3, f = 44.90 Hz

Fig. 5.43 Mode 4, f = 54.32 Hz

5.4.1.8

Morphing Shapes and Selected Cambers

The morphing shapes (Fig. 5.46) correspond to optimal ones derived from the studies in Chaps. 3 and 4. Green curve: position “0%” of camber: static position, non-cambered: reference case Violet curve: position “33%” of camber: trailing edge at 3.33 cm lower than in the reference case Brown curve: position “66%”: trailing edge at 6.66 cm lower than in the reference case Blue-green curve: position “100%”: trailing edge at 10 cm lower than in the reference case.

188

F. Auteri et al.

Fig. 5.44 Morphing flap assembling structure in the workshop lab of INPT/IMFT in collaboration with INPT/LAPLACE

Fig. 5.45 Zoom of the structure of the morphing flap assembly

Fig. 5.46 Morphing shapes of the flap

5 Aerodynamic Evaluation

189

5.5 Interface Definition and Camber Control of the High-Lift Flap 5.5.1 Modelling of the System and Feedback Control Design 5.5.1.1

Global Control

The global control is made by ONERA and is detailed in the following section. From the mechanical draft, a dynamical model of the camber actuation is designed using MATLAB and Simulink. The morphing wing composed of ribs and hinges is simplified as 5 strands, connected by pivots around parallel axis. The overall model is presented in Fig. 5.47. Each strand is defined by its mass m, length l, center of gravity G, inertia J and articular parameter q (corresponding to angles expressed in the global referential). Hypotheses are chosen for this simplified model: rotations are considered two dimensional and around parallel axis, strands non-deformable and pivots perfect. The description of the dynamical equilibrium of the system is therefore performed using the Lagrange principle: ˙ q˙ = Fg (q) + Fa (q) + Γq M(q)q¨ + ℵ(q, q) with Fg (q) moment of the gravitational force, Fa (q) moment of aerodynamical forces, Γq moment applied by the actuators at the pivot, M(q) inertia matrix and ˙ q˙ gyroscopic effects, here neglected due to the slow motion of the morphing ℵ(q, q) wing. This simplified model is then used to design feedback controllers. An objective of angular morphing is fixed by the global flows and pilot demands parameters. The closed-loop control goal is to achieve this objective with precision and speed. A proportional-integral-derivative (PID) regulator is first considered as a reference closed-loop controller in order to achieve the performances previously detailed.

Fig. 5.47 Schematic illustration of the morphing flap’s segmentation

190

F. Auteri et al.

Some specificities of the system could perturb the proper control, like the hysteresis behavior of the EMA and saturations of actuators. A more sophisticated approach of the feedback control synthesis will be therefore considered in order to counter these difficulties. The global control has been developed but not integrated and tested yet in the system. It is also based on the motor close control, which has been developed and integrated to the motor. The motor close control is presented in the following section.

5.5.1.2

Control Interface

Thanks to the global control provided by ONERA, the needed moment applied by the actuators at the pivot, Γq , can be determined for a specific flap shape and aerodynamic conditions. From the moment, the torques TH on each hinge are directly determined. NOVATEM and ONERA together provided a simple interface control to link those torques on hinges to the torque provided by the motor. Figure 5.48 reminds the EMA integration around a hinge. Figure 5.49 explains the corresponding geometric issue to solve and finally, the control relation between hinge torques TH and motors torques TM is given on Fig. 5.50 as a simple diagram. The coefficients K S N and K G are directly determined for screw-nut and gearbox parameters, given above in this document. The control diagram of can be added to the global control made by ONERA, with a dedicated software. Therefore, the needed

Fig. 5.48 EMA integration around a hinge

5 Aerodynamic Evaluation

191

Fig. 5.49 Geometric relation

Fig. 5.50 Relation between hinge and motor torque diagram

torques on each motor can be adapted to the aerodynamic condition. Then, from a hardware point of view, the torque demand calculated with a software is transformed into a current demand and sent to the close control hardware integrated with the motor and the inverter, on the EMA, as explained below.

5.5.1.3

Close Control Strategy of the Motor

As there is a real space and weight constraint, a full analogic close control is kept in order to reduce the complexity of the electronics. The reference torque of each motor is calculated in a central control command strategy while each motor has a driver torque close control. The driver Torque Control strategy is described in Fig. 5.51 diagram. The current set-point Iset is provided by the central electronics and serves as a modulation for the position sensor signals (Vh1m , Vh2m and Vh3m ) generically called Vhxm . The result of the modulation is the variation in the magnitude of the Vhxm signals, thus resulting in the current reference by phase Ir e f x . A comparison to the actual value of each phase current, Ixm , is performed and corrected with an amplification gain. The resulting signals are directly used as current reference to be compared to a carrier in a full Pulse-Width Modulation (PWM) strategy. Finally, the resulting gate signals are sent to a 3-phase inverter with integrated drivers.

192

F. Auteri et al.

Fig. 5.51 Close control strategy diagram

Fig. 5.52 Representation by an open kinematic chain; each strand is characterized by its length (li ), the location (xi , yi ) of its gravity center (G i ), its mass (m i ) and its inertia around its gravity center (Ji )

5.5.2 Camber Control 5.5.2.1

Modeling of the Articulated Flap

General Architecture The mechanical structure of the flap shows six segments: the leading edge segment (fixed in our wind tunnel experiments) and 5 articulated segments. The 5 joints are operated by screws, which are themselves controlled by the EMAs. In Fig. 5.52, we have represented the flap envelope, the joint rotation axes, the EMAs with the screws. To define a model of mechanical interaction between the different segments of the flap, one strand will be associated with each articulated segment (Fig. 5.52). These strands are shown in Figs. 5.52 and 5.53. The circles still correspond to the axes of joint rotation. The gravity centers of the strands (i.e., G i ) are also indicated. In this form the flap appears as a planar open kinematic chain. We will see that we can write a dynamic behavior model quite easily, provided we know the mechanical characteristics of the strands, and the forces that apply to each of them.

5 Aerodynamic Evaluation

193

Fig. 5.53 Characteristics of the i-strand

Efforts Suffered by the Segments For the i-strand the efforts to be considered are: – The joint control torque Ci . It is induced by the motor torque. – The gravity force Fgi . It generates torques on all strands except the ith, since it applies to the gravity center G i of that strand. – The aerodynamic force Fai . This force has two components (lift and drag). It applies at the pressure center Pi of the segment. It is considered as a perturbation input. It generates torques on all the strands even on the one considered because Pi is not collocated with G i . For the control law design, the considered control inputs are the joint control torques. The disturbing inputs are the gravity forces and the aerodynamic forces.

EMA Actuation Modeling The Electro-Mechanical Actuators (EMAs) are torque controlled independently of each other. The motor torque (Cm ) drives the screw which generates a force (Fv ) and results in a control joint torque (C): C = kv Fv joint control torque, kv is the joint arm level. Fv = Cm / pv driving force transmitted to the screw, pv is the screw pitch. Cm = ηHmot Cm∗ motor torque, η is the EMA efficiency (motor, gearbox, screw). We thus have a simple linear relationship between Cm and C, but also between the control torques demands Cm∗ and C ∗ . Note that the EMA transfer function Hmot has a very large bandwidth and can be neglected for the design of the joint control law. The ratings used are as follows: θm = motor rotation angle θr = angle at the output of the gearbox r = reduction ratio θ = joint angle ω = joint rotation speed

194

F. Auteri et al.

xv = linear displacement of the screw. The elementary relationships for a given joint are as follows: θm = rθr xv = pv θr xv = kv θ w = θ˙ . Here the important points are: – The linear displacements of the screws (i.e. xvi ) and the linear velocities (x˙vi ) are measured. – The torque control of the EMAs which is described by a digital transfer function Hmot with very fast sampling (20 kHz). Hmot = Hmot (Z )with f e = 20 kHz – The efficiency model η which includes the efficiency of the motor, gearbox and screw. It is therefore expressed as a function of the linear velocity x˙v and could present a fairly high uncertainty. η = η(x˙v )uncer tain – Displacement and linear velocity measurements which are filtered and sampled at rates sufficient for control. Dynamical Behavior of the Articulated Flap Lagrange’s theory is used to develop this model. The model obtained is of the form: ( ) M(θ)θ¨ + N θ, θ˙ θ˙ = LFg (θ)Fg + LFa (θ)Fa + C Here θ is the vector of the 5 relative angles of each strand with respect to the previous one. The matrix M (dimension 5 ×5) is the inertia matrix. It varies according to the joint position. The second term is the Coriolis term. It will be neglected in our case, because the rotation speeds are a priori very small. The right member is a vector of dimension 5. Its i-th component is the sum of the 3 torques applied to the i-th joint: torque induced by the gravity forces Fg , torque induced by the aerodynamic forces Fa , and engine torque C generated by the EMA. A calculation code has been developed (using Matlab and Simulink software) to generate the matrices M(θ), LFg (θ), LFa (θ). This provides a model for representing the dynamic behaviour of the flap equipped with its actuators and sensors. The inputs of the simulation are the joint control torques demand (C∗ ), the aerodynamic forces and the gravity forces on each segment, Fig. 5.54.

5 Aerodynamic Evaluation

195

Fig. 5.54 Flap model architecture

The gravity forces are obviously constant in amplitude and direction, however their points of application (i.e. the gravity centers of the segments) are moving as the flap is articulated. This is modelled by L Fg (θ ). As regards the aerodynamic forces, the simulator must be provided with their values but also with their points of application (the aerodynamic centers of pressure).

5.5.2.2

Aerodynamic Force Modeling from Databases

For the take-off and landing configurations, multiple databases from the European project partners are used here. They include spatial coordinates of the full wing and flap, and total pressure for each point at these coordinates. For the take-off configuration, the flap coordinates of different cambers, and their corresponding pressure distribution at the geometry, are used from numerical simulations of INPT/IMFT. Same analysis is performed for the landing configuration from CFSE. In addition to pressure distribution, for the take-off configuration, CFDB performed optimization of the flap shape with respect to the lift-and-drag ratio. These optimal angles will be used as targets of the camber control.

Modeling Methodology Two databases (see Sects. 5.5.2.2.2 and 5.5.2.2.4) give the pressure distribution along the wing and flap profiles (i.e. x, y, P). Several aerodynamic conditions are considered (flow velocity U∞ , wing angle of attack α, dynamic pressure Pd ) and various shapes of the flap also. From these sets of aerodynamic data, we can compute the lift and drag forces (and the pressure center) for each of the segments. The methodology we applied is as flow. – A geometrical analysis of the shape allows first to separate the suction side (extrados) information (xe , ye , Pe ) from the pressure side (intrados) ones

196

F. Auteri et al.

Fig. 5.55 From the profile to the articulated skeleton

(xi , yi , Pi ). The mean shape profile is defined as the curve whose y-coordinate is the mean of ye and yi . – We then find the location of the hinges along this mean shape, starting from the leading edge and knowing the segment lengths. This allows us to associate a part of the mean shape to each segment (see an example on Fig. 5.55). – Next, by integrating the pressure difference between intrados and extrados along each segment we found the lift force. For this integration, a span-wise width of 1 m has been chosen. The drag force cannot be computed from pressure information alone. So we assumed a constant ratio between lift and drag. The momentum induced on the hinge is also computed by integration. – Finally, from the force and the momentum, the location of the pressure center is easily deduced. We checked that the overall lift force is similar to the one given within the databases when available. The pressure centers are almost independent of the aerodynamic conditions. We also find a good agreement of our results with those given by the thin plate theory applied to each strand of the skeleton. A third database gives optimal profiles for various criteria. Our methodology has been applied to deduce the mean shape, the hinge locations and the optimal angles of the articulated flap.

Database for the Nominal Take-Off Configuration (from INPT/IMFT) The shapes of the airfoil in take-off configuration at an angle of attack AoA = 0° are depicted in Fig. 5.56 with the lines for the skin on the suction side and the pressure

5 Aerodynamic Evaluation

197

Fig. 5.56 The shapes of the airfoil in take-off configuration at AoA = 0°. On the left: without camber, on the right: with maximal camber

side, the determined mean shape as well as the locations of the 5 articulations and the computed pressure centers, where the mean shapes of the airfoil have been determined and the pressure centers have been computed as described in Sect. 5.5.2.2.1. The 2D lift force of this airfoil in function of the normalized deformation Y/c (where c is the mean chord) is plotted on Fig. 5.57. This lift force has been obtained following the procedure of Sect. 5.5.2.2.1. As expected, the bigger the deformation, the bigger is the lift. The evolution is almost linear. The lift forces at strand 0 and 1 are much more important than on the other strands as expected. These values are quite similar to the values delivered by INPT/IMFT.

Database for the Optimized Take-Off Configuration (from CFDB) CFDB has optimized the shape of the morphing flap with respect to two different criteria: • maximization of the lift L, • or maximization of the lift over drag ratio L/D. The optimization has been realized either: • without constraints on certain airfoil parameters, the so-called free optimization, • with constraints on certain airfoil parameters, the so-called restricted optimization. CFDB has optimized the shapes of the flap for the reduced scale RS and the large scale LS demonstrator. Table 5.6 summarizes the obtained optimal articulation angles. In the following, the results for the restricted optimization of the L/D ratio for the LS demonstrator are illustrated as an example. The optimal articulation angles correspond to the last line of Table 5.6 in bold. The shape of the airfoil in take-off configuration at an angle of attack AoA = 0° for this optimal L/D ratio for the LS demonstrator is depicted on Fig. 5.58 with the

198

F. Auteri et al.

Fig. 5.57 Lift force per width unit on each strand in function of the normalized deformation Y/c at AoA = 0° in take-off configuration

lines for the skin on the suction side and the pressure side, the determined mean shape as well as the locations of the 5 articulations, where the mean shape and the optimal articulation angles have been determined as described in the following section.

Database for the Landing Configuration (from CFS) The shapes of the airfoil in landing configuration at an angle of attack AoA = 0° are shown in Fig. 5.59 with the lines for the skin on the suction side and the pressure side, the determined mean shape as well as the locations of the 5 articulations and the computed pressure centers, where the mean shapes of the airfoil have been determined and the pressure centers have been computed as described in the following section. The 2D lift force of this airfoil in function of the normalized deformation ΔY /c (where c is the mean chord) is plotted on Fig. 5.60. This lift force has been obtained following the procedure of Sect. 5.5.2.2.1. As expected, the bigger the deformation, the bigger is the lift. The lift forces at strand 0 and 1 are much more important than on the other strands as expected. The evolution is nonlinear. The values are higher than in take-off. These values are quite similar to the values delivered by CFS.

Resulting Aerodynamic Forces The main result of this aerodynamic force modelling step is twofold:

5 Aerodynamic Evaluation

199

Fig. 5.58 The shape of the airfoil in take-off configuration at AoA = 0° for an optimized L/D ratio for the LS demonstrator Table 5.6 Optimal articulation angles θi (from CFDB) for the RS and the LS demonstrator RS demonstrator Optimization cases Free

L

Optimal articulation angles of strand i [°] Θ0

Θ1

Θ2

Θ3

Θ4

Θ5

− 7.28

− 2.03

1.42

0.42

− 0.07

− 4.25

Free

L/D

− 7.28

− 2.03

1.42

0.42

− 0.08

− 4.54

Restricted

L

− 7.28

− 2.03

1.42

0.42

− 0.08

− 4.49

Restricted

L/D

− 7.28

− 2.03

1.42

0.42

− 0.08

− 4.58

Θ4

Θ5

LS demonstrator Optimization cases

Optimal articulation angles of strand i [°] Θ0

Θ1

Θ2

Θ3

Free

L

4.25

− 7.69

− 4.58

− 1.81

− 1.97

− 2.69

Free

L/D

4.25

− 7.69

− 4.59

− 1.87

− 2.09

− 2.41

Restricted

L

4.25

− 8.20

− 5.02

− 0.95

− 1.23

− 2.72

Restricted

L/D

4.25

− 7.50

− 4.10

− 2.43

− 2.53

− 2.51

– We set a first table 𝝫 as the aerodynamic force table. It gives the aerodynamic force (lift and drag) along with its application points (pressure center) for each of the segments,

200

F. Auteri et al.

Fig. 5.59 The shapes of the airfoil in landing configuration at AoA = 0°. On the left: without camber, on the right: with maximal camber

Fig. 5.60 Lift force per width unit on each strand i in function of the normalized deformation Y/c at AoA = 0° in landing configuration

(F a , xa , ya ) = 𝝫(U∞ , α, θ ) The inputs are the flow velocity U∞ , the wing angle of attack α, and the joint angles of the articulated flap. – We set a second table ψ as the optimal angle table. It gives the optimal hinge angles of the articulated flap,

5 Aerodynamic Evaluation

201

Table 5.7 The mass m i of the ith strand Strand i

0

1

2

3

4

5

mi [kg]

1.5

1.2

1

0.3

0.2

0.1

θ ∗ = ψ (U∞ , α, crit) The inputs are the flow velocity U∞ , the wing angle of attack α, and the optimized criterion. The first table will be used in the simulation to generate the appropriate values of the aerodynamic forces. It requires an interpolation scheme to be used in our simulator, since the table is only set up for discrete values of θ . The second table will be used to generate input demands for the flap control law.

Gravity Force Modeling From the flap and actuator design for the large scale LS demonstrator realized by INPT/IMFT and NOVATEM, 6 strands and the following normalized 5 articulation positions have been identified: xCi /c = [0.234, 0.518, 0.601, 0.656, 0.755], yCi /c = [0.028, 0.058, 0.059, 0.055, 0.036], where c is the mean aerodynamic chord. The first strand starts at [x c0 ,yc0 ]/c = [0,0], the last strand ends at [x c5 ,yc5 ]/c = [1,0]. From these positions, the ith strand length L i could be derived. With the mass mi of the ith strand, its inertia J i can be determined following the relation: Ji = m i L i2 . The chosen mass values are summarized in Table 5.7. They correspond to the reduced scale RS demonstrator. The simulated gravity torques do therefore not correspond to the ones which would be observed on the LS demonstrator. As soon as the actual mass values of the LS demonstrator will be delivered by INPT/IMFT and NOVATEM, these data could be updated and the gravity torques could be resimulated.

202

5.5.2.3

F. Auteri et al.

Camber Control Law Design and Simulation

Control Law Design The methodology we applied to develop the control law is called the computed torque approach. Recall that the model has the following structure: ( ) M(θ)θ¨ + N θ, θ˙ θ˙ = LFg (θ)Fg + LFa (θ)Fa + C where Fg is the vector of the segments weights, Fa the vector of aerodynamic forces applied to the segments, and C the vector of control torques. The computed torque approach consists in an open-loop nonlinear disturbance compensation, plus a linear regulator. It writes: Δ

Δ

C = −LFg (θ)Fg − LFa (θ)Fa +M(θ)Creg The disturbance compensation uses an estimation of the disturbances and also requires to know the current geometry (vector of angles θ). It results in a closed-loop governed by: θ¨ ∼ = Creg Then the joints’ angles regulation may be achieved using a simple proportional integrate derivative (PID) control (recall that θ˙ = ω) with anti-windup: eθ = θ − θ∗ eC = C − sat (C) { Creg = −Kp eθ − Ki

(eθ − K AW eC ) − Kd ω + Kf θ∗

Here sat (C) is the value of the computed control torque C saturated by its physical boundaries. The PID gains are Kp , Ki and Kd . And K AW is the anti-windup gain. The feedforward gain Kf does not change the closed loop dynamic. But it allows to tune the shape of time responses. The computed torque approach is very efficient whenever the estimation of the disturbances is good. Its performances may however be degraded in the presence of control saturation. Indeed in such a case the equation θ¨ ∼ = Creg is not fulfilled anymore and the regulator design must involve an unknown disturbance rejection specification. The anti-windup term limits such unexpected effects. The control design specifications are the settling time (Tr ep ) and the damping (ξ ). The settling is the time delay to move from one shape to another. We chose Tr ep = 10s

5 Aerodynamic Evaluation

203

Fig. 5.61 Feedback control architecture

(rather slow), and ξ = 1 (rather high in order to prevent for unexpected oscillations). The tuning of controller gains K p and K i is deduced from these specifications. Because the flap behaves like an open kinematic chain, the control of one hinge interacts with all others. Then the achieved settling time for a shape change demand is not equal to Tr ep . An iterative tuning methodology has been proposed to adjust the torque settling time specification until the effective shape settling time specification is fulfilled. A few iterations of a dichotomy optimization procedure are enough to find the right setting.

Feedback Control Architecture The full camber control loop is presented in Fig. 5.61. Motor torques demands to achieve for the EMA Cm∗ i are computed by the feedback controller from the shape target angles θi∗ , the current angles θi and velocities ωi of each part of the flap. The objective is to regulate the angles θi to the targets θi∗ . Note that the inverse shape model has not been implemented in a dynamic fashion but is currently tabulated. For some optimal shapes computed by CFDB in previous Sect. 4.6, we deduced the associated optimal values of θi∗ . It clearly appears that the EMAs are used as actuators, since they apply torques on the flap hinges, but they are also used as sensors, since they deliver the signals (θi and ωi ) used by the camber control law. We emphasize this important point by showing in Fig. 5.62 the complete architecture of the simulation which is used to validate the camber control law. The θi angles (and the angular rates) are part of the state of the flap mechanical model. And thus the flap dynamic is well feedback controlled. Even if in practice, the value used by the camber controller is provided by the EMAs which are integrated into the flap.

204

F. Auteri et al.

Fig. 5.62 Simulation architecture

Simulation of the Camber Control Nominal Take-Off Configuration In Fig. 5.63, the evolution of the skeleton, the mean shape of the whole flap, is depicted during the cambering from the initial shape (no camber) to the final shape (maximum camber). It can be observed that the strand 0 is fixed and that the cambering is the most important for strand 1. After 5 s, strand 1 reaches its maximum deflection, the other strands continue to move. When strand 2 reaches its maximum deflection, it stops and the remaining strands continue to move, and so on. You can observe that strand 3 does almost not move with respect to strand 2. The relative strand angles θi , the control torques Ci , the gravity torques Mgi and the aerodynamic torques Mai are first simulated during the cambering of the flap from the initial shape to the final shape at an angle of attack of AoA = 4°. Their evolution is plotted on Fig. 5.64. As expected, the aerodynamic torques increase during the cambering. The gravity torques remain constant during the cambering. The control torques decrease during the cambering. All torques at the strands 0 and 1 are much bigger than the other ones as the cambering of strand 1 is the most important. The evolution of theses parameters is also simulated for other angle of attacks: AoA = 0° and AoA = 2°. The control torques and the aerodynamic torques are plotted on Fig. 5.65. As the angles of attack are smaller than in the previous case, the aerodynamic torques at strand 0 and strand 1 are slightly smaller while their control torques are slightly bigger. Take-off configuration with optimized angles In Fig. 5.66, the evolution of the skeleton is depicted during the cambering from the final shape to the optimized shape obtained by the restricted optimization of the L/D ratio of the LS demonstrator (see bold line in Table 5.6). It can be observed that strands 0 to 4 do almost not move with respect to the final shape, just strand 5 is morphed. This was expected as the optimization done by CFDB has been concentrated on the trailing flap area.

5 Aerodynamic Evaluation

205

Fig. 5.63 Evolution of the mean shape during the cambering of the flap from initial shape to final shape in the take-off configuration

Fig. 5.64 Time evolution of the deflection angles, the control torques, the gravity torques and aerodynamic torques of each strand i during the cambering of the flap from initial shape to final shape at AoA = 4°

The relative strand angles θi , the control torques Ci , the gravity torques Mgi and the aerodynamic torques Mai are simulated here during the cambering of the flap from the final shape to the optimized shape at an angle of attack of AoA = 0°. Their evolution is plotted on Fig. 5.67. As expected, the aerodynamic and control torques

206

F. Auteri et al.

Fig. 5.65 Time evolution of the control torques and aerodynamic torques of each strand i during the cambering of the flap from initial shape to final shape at AoA = 0° (left) and AoA = 2° (right)

Fig. 5.66 Evolution of the mean shape during the cambering of the flap from the final shape to the optimized shape

do not change a lot as the morphing to reach the optimized shape is quite small. The gravity torques remain again constant during the cambering.

5 Aerodynamic Evaluation

207

Fig. 5.67 Time evolution of the deflection angles, the control torques, the gravity torques and aerodynamic torques of each strand i during the cambering of the flap from final shape to optimized shape at AoA = 0°

Landing Configuration In Fig. 5.68, the evolution of the skeleton is depicted during the cambering from the initial shape (no camber) to the final shape (maximum camber) in the landing configuration. It can be observed that the strand 0 is fixed and that the cambering is the most important for strand 1 and then 5. You can observe that strand 3 and 4 do almost not move with respect to strand 2. The relative strand angles θi , the control torques Ci , the gravity torques Mgi and the aerodynamic torques Mai are simulated here during the cambering of the flap from the initial shape to the final shape at an angle of attack of AoA = 4°. Their evolution is plotted on Fig. 5.69. As expected, the aerodynamic torques increase during the cambering. The gravity torques remain constant during the cambering. The control torques decrease during the cambering. All torques at the strands 0 and 1 are much bigger than the other ones as the cambering of strand 1 is the most important. The absolute values of the torques are bigger than in the take-off configuration as the relative strand angles are higher. A detailed description including appendages of the camber control and its interface can be found in: http://smartwing.org/SMS/EU/DOCUMENTS/ONERA-Cambercontrol-and-interface.pdf.

208

F. Auteri et al.

Fig. 5.68 Evolution of the mean shape during the cambering of the flap from initial shape to final shape in the landing configuration

Fig. 5.69 Time evolution of the deflection angles, the control torques, the gravity torques and aerodynamic torques of each strand i during the cambering of the flap from initial shape to final shape at AoA = 4° in the landing configuration

5.6 Model Specifications and Preliminary Investigation in a Scaled Wind Tunnel (1:9 Scale) The wing model shown in Figs. 5.70 and 5.71 has the following specifications:

5 Aerodynamic Evaluation

209

• Full span: 4 m; • Wing chord: 2.4 m; • High lift device (flap prototype) for half span (2 m) with different positions (Landing, Take-off, Clean); • Angle of attack from 0° up to 8°. Given the considerable size of the model, especially with respect to the wind tunnel, a preliminary test has been carried out on the scaled wind tunnel (1:9) [3] to check the compatibility of the large model with the wind tunnel and exclude problems related to a possible stall of the diffuser. As a matter of fact, the wind-tunnel model has been used in the past years both to test the design choices that led to the construction of the full-scale wind tunnel and to test measurement layouts or new features, such as the devices used to improve the performance in open-test-section configuration [3].

Fig. 5.70 Rapid prototyping scaled model (1:9)

Fig. 5.71 Scaled model in scaled wind tunnel

210

F. Auteri et al.

The landing configuration has been tested for two angles of attack: 0°and 8°. Micro-tufts have been used to detect separation and stall of the axial compressors [4]. In both cases the wind tunnel operated correctly: no massive separation was detected in the diffuser nor fan stall observed. A 10% reduction in maximum achievable velocity has been measured for an angle of attack of 0°, and a 20% reduction for 8°, which is acceptable for the present project. As a result, the wind tunnel was cleared as capable of sustaining the high flow deflection induced by such a large model. Moreover, the performance degradation seemed not critical for the project, since Reynolds numbers and forces significantly higher than that obtained in IMFT wind tunnel resulted within the range of operations.

5.6.1 The Model The complete wind-tunnel model consists of two main parts: the prototype of the large-scale morphing flap and the main wing on which the flap is mounted. POLIMI is the responsible of the design and construction of the main wing, the related strategy to install it in the test section and the measurement campaign, especially for the landing configuration.

5.6.1.1

Model Wing Structure Design

The specifications driving the model design were: 1. Safety of operations: the model must sustain the high loads, a few tons, and vibrations coming from possible flow separation when operating at the maximum speed achievable by the wind tunnel with the model installed, namely 50 m/s in take-off configuration and 45 m/s in landing configuration. Moreover, operating the model, for instance changing the model angle of attack or configuration, must be feasible in total safety by the operators. 2. The wing model must be easy to operate. It must be accessible to install the measurement system; operations like changing the angle of attack or the flap position should be easy and take the minimum time possible in order to expand the test matrix. 3. The wing model must be rigid: the allowed deformation must be sufficiently small not to affect the measured quantities. 4. The wing model must be feasible and not excessively expensive. A framed structure has been selected for the wing model, with a structural skeleton and a non-collaborating skin (Fig. 5.72). The skeleton is composed of two steel spars that extend along the full spanwise and sustain the whole loads. The spars are connected to each other through ten aluminium ribs, which transfer the aerodynamic loads coming from the skin to the spars. Two force ribs, among the ten, are used to sustain the flap and transfer the loads coming from it to the spars.

5 Aerodynamic Evaluation

211

Fig. 5.72 3D model of the wing model (left: wing skeleton, right: wing with skin)

The skeleton is covered by a skin that gives the model the appropriate shape. The skin is composed by two different materials: – aluminium alloy panels in the low-curvature regions – fiberglass-reinforced composite panels for the high-curvature regions, namely the leading edge and the trailing edge in the bay hosting the flap. During the design phase of the wing model, all the necessary structural checks were carried out to obtain a model that satisfied the adequate safety requirements. In this phase, in order to evaluate the aerodynamic loads, both the results of two-dimensional potential analysis [5] and those of numerical simulations were exploited. The computational fluid dynamics (CFD) incompressible ReynoldsAveraged-Navier-Stokes simulations carried out by the partner CFS Engineering were very useful, since simulations considered the correct shape and dimension of the model and the effect of the walls of the test section. For what concerns the structural analysis, the theory of beams was used for the verification of the spars [6]. The ribs, the skin and other components such as the wing supports were verified by finite element analysis. An example of the results of the finite element analysis carried out for the force rib is reported in Fig. 5.73. In order to exclude any instability phenomena associated with the presence of periodic fluctuations in the loads, a modal analysis was carried out on the model. In this phase, the body of the flap was modelled as a rigid body having the mass and inertia properties of the real flap. The first four frequencies are reported in Table 5.8 and the corresponding four eigenmodes of the structure are depicted in Fig. 5.74. The flap is fastened to the wing in the bay between the two force ribs through bolts, see Fig. 5.75. The interface was designed in collaboration with the INPT/IMFT partner in order to obtain the best solution that would guarantee the passage of all the cables and connectors that must come out from the flap. Furthermore, the interface system must allow the flap to be moved with respect to the wing so that it can be positioned in the clean, take-off and landing configurations. The design of the interface should also allow the easiest possible operation when changing the flap position without having to disconnect and reconnect all the cablings and tubings (Fig. 5.76).

212

F. Auteri et al.

Fig. 5.73 Finite element analysis on wing force rib

5.6.1.2

The Mock-Up in the Test Section

The wing model has been installed in the test section by a system that allows the rotation of the main wing around is main spar in order to change the angle of attack of the wing. Since the main spar, namely the one located at mid-chord, is free to rotate around its mounting point for adjusting the angle of attack as shown in Fig. 5.77, it is necessary to fix the wing model to suppress this degree of freedom and correctly set the angle of attack. This is obtained by the fixing the front spar, namely the one located in the region near the leading edge of the wing, to a plate by a simple bolted joint. As a consequence, the procedure for changing the angle of attack requires one to act only on the front-spar fixing system. In this way, it is possible to obtain all the requested test configurations, 0°, 2°, 4°, 6°, 8° angle of attack (Fig. 5.78) and clean/take-off/landing (Fig. 5.79), by operating the connection between the front spar and the wind-tunnel wall and the connection between the flap and the wing (Fig. 5.79).

5.6.1.3

Construction and Assembly

The construction of the model required the use of non-standard machinery [7], in particular because of his large size. Each rib was machined as a single part from a plate of EN AW 5083 aluminium (Fig. 5.80). Since the two spars are made from welded rectangular steel tubes, the accuracy of the raw material is quite low. For this reason, to secure the proper alignment, the two spars were entirely machined in order to obtain perfectly straight beams on which to mount the ten ribs (Figs. 5.80 and 5.81).

5 Aerodynamic Evaluation

213

Fig. 5.74 Modal analysis, first four natural frequencies Table 5.8 Lowest four natural frequencies of the model

Frequency (Hz)

Figure Complete

Frame

35.3

(a)

(aa)

61.7

(b)

(bb)

122.7

(c)

(cc)

145.8

(d)

(dd)

214

F. Auteri et al.

Table 5.9 PIV take-off tested cases Take-off configuration Baseline

AoA°

Speed (m/s)

4

30 34.1

AoA°

Speed (m/s)

4

30

6

√

30

30

8

√

√ √

30

√

34.1

√

40

√

40

√

34.1

√

34.1

√

40

√

40

√

30

√ √

34.1

√

34.1 8

Camber 1

√

40 6

√

√

40

Table 5.10 PIV landing tested cases Landing configuration AoA° Speed (m/s) Baseline 0

30 34.1 40

2

30 34.1 40

4

30 34.1 40

6

30 34.1 40

AoA° Speed (m/s) √ Camber 0 √ 1 √ √ √

2

√

√ √

30 40

4

30 34.1

√ √

34.1 40 34.1

√ √

30

40 6

30 34.1 40

Fig. 5.75 Bay containing the flap and details of the force ribs

AoA° Speed (m/s)

√ Camber 0 √ 2 √ √ √

2

√

√ √

30 40

4

30 34.1

√ √

34.1 40 34.1

√ √

30

40 6

30 34.1 40

√ √ √ √ √ √ √ √ √ √ √ √

5 Aerodynamic Evaluation

215

Fig. 5.76 Wing model with flap mounted (left: upper side, right: lower side)

Fig. 5.77 Rear spar fixing system

Fig. 5.78 Variation of the angle of attack by rotation around the main spar located at 50% of the chord

Fig. 5.79 Variation of the flap configuration (left: take-off, right: landing)

216

F. Auteri et al.

Fig. 5.80 One rib of wing clean part

Fig. 5.81 Mounting phase of ribs on the spars

The alignment of the wing sections was guaranteed by carrying out the assembly on a large reference plane using a Taylor-Hobson Micro Alignment Telescope [8] and finally passing two turned, ground and polished shafts through precision-bored holes obtained on the spars. As previously mentioned, 1.5 mm thick aluminium alloy panels were used for the external surface, while in the highly curved areas the cover was made with moulded fiberglass-reinforced composite (Fig. 5.82). In order to ensure high precision of the composite parts, a dimensional analysis was carried out on the molds, as reported in Fig. 5.83, from which it is possible to

Fig. 5.82 Composite skin parts, (left: leading edge, right: trailing [9] edge of flap bay)

5 Aerodynamic Evaluation

217

Fig. 5.83 Dimensional analysis of the trailing edge composite skin, error expressed in mm

Fig. 5.84 The final model parked in vertical position (left: without flap, right: with flap)

observe that the error is well below one-half of a millimeter. The final model is shown in Figs. 5.84 and 5.85.

5.6.2 Test-Matrix The test matrix has been designed to fully investigate the performance of the wing with the installed morphing flap. The dimension of the space of test parameters is quite large. We have: 1. three different flap configurations: clean, take-off and landing; 2. four different flap geometries, each one with a different camber: baseline, 1, 2, 3; 3. five possible angles of attack, 0°, 2°, 4°, 6°, 8°;

218

F. Auteri et al.

Fig. 5.85 Model installed in the GVPM test section, (left: wing model, right: particular of the flap)

4. finally, we can set the wind speed in order to change the Reynolds number and the flap loads, and therefore its deformation. A set of three different wind speeds was considered sufficient to investigate the effect of Reynolds number and flap deformation on its performance. Therefore, ideally, a total of 180 different configurations had to be tested. In designing the test matrix, it must be taken into account that changing the flap configuration, changing the flap geometry and changing the angle of attack are all time consuming activities that require the operators to stop the wind tunnel, enter the test section and manually work on the model. Building a model capable of automatically setting of these parameters, considering its size, would have increased by an order of magnitude its cost, this option was therefore discarded in the design phase. Since the allocated time in the wind tunnel was fixed, 15 working days, we decided to assign different priorities to the entries of the test matrix. For what concerns the flap configuration, the landing configuration was assigned the highest priority, priority 1, since this measurement was possible only in POLIMI, given the size of the wind tunnel. Priority 2 was assigned to the take-off configuration and priority 3 to the clean configuration, since the last has been tested in the S1 wind tunnel at INPT/IMFT [10]. The flap geometries (Fig. 5.86) were assigned the same priority, since there is no a-priori geometry that is most interesting to test than the other ones. After measuring forces on the model, the base geometry and the most promising configurations were assigned a higher priority for the successive tests. Also, the angles of attack were assigned the same priority, but for 8° in landing configuration. This angle of attack was discarded, since the RANS simulations of the wind-tunnel test set-up showed massive separation, which could lead to strong force oscillations and strong vibrations that could jeopardise the integrity of the flap. Moreover, some of the measurement techniques were incompatible with each other and therefore they had to be taken separately. For instance, measuring the flap deformation required the introduction of a large number of cameras in the wind tunnel that could be detrimental to the quality of the fluid vein. Also, the PIV measurements were separated from the other ones for safety reasons [11]. Therefore, different

5 Aerodynamic Evaluation

219

Fig. 5.86 Flap cambering configurations, (left: Take-off configuration, right: Landing configuration)

priorities were also assigned to three different sets of measurement to be carried out on the model: 1. pressure, wake and force measurements: priority 1; 2. deformation measurements: priority 2; 3. PIV measurements: priority 3. The test matrix was then divided in three different measurement sessions: • The first session was dedicated to measuring loads and pressure distributions both on the model surface and in the wake; • The second session was dedicated to measuring the flap deformation; • The third session was dedicated to measuring the velocity field in the region near the suction side of the flap. First session: force, pressure and wake measurements Seven working days were assigned to force, pressure and wake measurements. The time needed to carry out the different activities of the test was estimated as follows: • • • •

Change of configuration: approximately 3 h; Change of flap geometry: approximately 1.5 h; Change of angle of attack: approximately 0.5 h; Each speed: 5’

From this estimate, it was clear from the beginning that 7 workdays, 8 h a day, were not sufficient to complete the whole programme. It was then decided to cut the least interesting case, namely the clean configuration that was scheduled to be tested at INPT/IMFT.

220

F. Auteri et al.

Second session: deformation measurements Three working days were assigned to the deformation measurements, owing to the lower priority, to the lower time needed to acquire each measurement point with respect to the previous case, and taking into account that the learning-curve effect will reduce the time needed to change the configuration, flap geometry and angle of attack. In order to satisfy the required test length, it was decided to test only two different speeds, 20 and 30 m/s. The maximum tested speed was selected to avoid vibrations and movements of the cameras that could significantly decrease the accuracy of the measurement and increase the test time, due to an increased time spent for system calibration. Third session: PIV measurements Five working days were assigned to the PIV measurements. The reason is that such measurements are particularly time consuming. Indeed, each change of the angle of attack requires re-adjusting the position of the two cameras, a time that must be added to the time usually spent to change the angle of attack. Moreover, in order to acquire a sufficient number of frame couples, the acquisition time is quite long, and some preliminary checks were necessary at the end of each acquisition to be sure that the acquired image pairs had the required quality. The time dedicated to PIV was, however, not enough to complete the whole test matrix employed for force and pressure measurements. Therefore, we chose to test only a selected number of promising configurations together with the baseline.

5.6.3 Test Rig, Instrumentation and Sensor Layout In this section the instrumentation set-up and all data acquisition systems are described. We start from the description of the pressure measurement on the wing. Moreover, the measurement of the stagnation pressure across the wake, using a dedicated rake purposely built, is also presented. Then the layout and instrumentation used for the particle-image-velocimetry (PIV) surveys is described. Finally, the layout and instrumentation used for shape measurements is described.

5.6.3.1

Pressure-measurements Set-Up

The pressure measurement campaign can be divided in two parts: I. the first part consists of the measurement of the pressure distribution on the surface of the model; II. the second part concerns the measurement of the total-pressure defect downstream of the model, in the wake region.

5 Aerodynamic Evaluation

221

Fig. 5.87 Layout of the pressure-tap distribution on the suction side (left) and pressure side (right) of the model surface

Static Pressure on the Model A series of 141 static pressure taps have been drilled on the surface of the model. Each pressure tap is obtained by drilling a hole of diameter 2.2 mm on the model surface. Then, a plastic tube is inserted in the hole and glued to the skin on the internal side of the model. The plastic tube is then flush cut on the outside side of the model. The internal diameter of the plastic tube, which is also the internal diameter of the pressure tap, measures 1.7 mm. The distribution of the pressure taps on the surface of the wing model is illustrated in Fig. 5.87. The pressure taps are mainly located on the symmetry axis of the wing model. But, since the dimensions of the model compared to those of the test chamber are considerable, it is possible that important three-dimensional phenomena are present. Therefore, some static pressure taps were placed along the spanwise in order to verify the flow symmetry with respect to the symmetry plane of the model. The pressure measurements were carried out by means of 5 Measurement Specialties Inc. Pressure scanners with different full-scale values (3 scanners: 1PSI F.S., 1 scanner: 2,5PSI F.S., 1 scanner: 10PSI F.S., accuracy 0.1% F.S.) installed inside the model and driven by a DTC Initium. The average pressure distribution was obtained over an acquisition time of 20 s for each configuration tested.

Measure of the Total-Pressure Defect in the Wake Starting from the pressure measured on the surface of the model, it is possible to obtain the aerodynamic coefficients due to the pressure field around the wing. While normal stresses provide the main contribution to the lift force, drag has an important contribution provided by viscous tangential stresses [12]. The contribution of the tangential stress to the drag force is larger in the regions were the thickness of the boundary layer is smaller and decreases when separation occurs, therefore it is less important in the landing configuration for high angles of attack [13]. In any case,

222

F. Auteri et al.

Fig. 5.88 Wake rake location and traversing system

Fig. 5.89 Geometric details of designed rake

to have a rough estimate of the viscous contribution to the drag force, it is possible to measure the total pressure defect in the wake of the airfoil. Indeed, the total pressure is constant for an irrotational incompressible flow, and, in the present case, the rotationality of the flow is produced by the viscous stresses. The goal of these measurements is therefore to estimate the total-pressure defect and the related coefficient. In order to do that, an opposite total-pressure rake located downstream of the wing, about 1 m from the wing trailing edge, was designed and built as shown in Fig. 5.88. Some limitations and difficulties of the method are well exposed in [13, 14]. The rake is designed in order to move in the Z direction (Fig. 5.89) through a system of linear guides and a stepper motor. Since the mean pressure over time is required, in this way it is possible to increase the actual number of pressure measurement points, allowing us to use a single pressure scanner with 32 channels but still obtaining the desired resolution. The rake is also provided of two Pitot tubes located at the two extremes and a fairing has been used in the measurement region, about 1.3 m long, with the shape of a NACA0018 wing section to reduce the perturbation on the measurements. See Fig. 5.90 for more details.

5 Aerodynamic Evaluation

223

Fig. 5.90 The designed wake rake mounted in the test section

5.6.4 PIV Set-Up PIV surveys were carried out in the region near the trailing edge of the flap, on the suction side. The aim of these measurements is to characterise the flow behaviour in this region which is particularly critical from the viewpoint of performance. Indeed, since the goal of the present experimental activity is mainly to evaluate the performance of the morphing flap in high lift configurations, landing and take-off, particular care must be put in the design of the morphed shape of the flap to avoid separation, while maximising the airfoil lift or efficiency [15]. The system was set up to measure the two velocity components on the longitudinal X–Z symmetry plane on two overlapped windows. The choice of the symmetry plane is dictated by the fact that, due to the low aspect ratio of the model dictated by the size of the LS prototype, we expect strong three-dimensional effects to be present. The symmetry plane is the measurement plane when the flow best approximates a twodimensional flow and is less influenced by the extremities. Two windows have been

224

F. Auteri et al.

Fig. 5.91 PIV measurement windows on the longitudinal plane x–z

selected to cover a large area with a resolution sufficient to capture the phenomena that characterise the boundary layer and the first part of the wake with sufficient resolution, still maintaining enough laser illumination. The area of investigation of each camera was 345 mm × 110 mm and the position of the measurement windows with respect to the flap is illustrated in Fig. 5.91. The PIV system comprised a Litron NANO-L-200-15 Nd:Yag double pulsed laser with a 200 mJ output energy and a wavelength of 532 nm, and two Imperx ICL-B1921M CCD cameras with a 12 bit, 1952 × 1112 pixel array. The layout of the PIV instrumentation is illustrated in Fig. 5.92.The laser was attached to a single axis traversing system positioned on the top of the wind-tunnel test section. The cameras were mounted on two slides attached to the side wall of the test section to move the image planes in the Z and X directions in order to follow the displacement of flap when the configuration or the AoA is changed (Fig. 5.92).

Fig. 5.92 Layout of the PIV instrumentation in the wind tunnel test section

5 Aerodynamic Evaluation

225

Each camera was equipped with a Nikkor 50 mm lens. The synchronisation of the two laser pulses with the image-pair exposure was controlled by a 6-channel Quantum Composer QC9618 pulse generator. A particle generator with Laskin atomiser nozzles was used for the seeding of the entire test-section. The particles consisted of small oil droplets with a diameter in the range of 1–2 μm. The image-pair analysis was carried out by the PIVview 3C software, developed by the PIVview 3C software, developed by PIVTEC [16]. All the resulting velocity fields presented in this work were averaged over 500 image pairs. The accuracy of the PIV measurement can be estimated considering a maximum displacement error of 0.1 px, as found in [9].

5.6.5 Shape-Measurement Set-Up The ultimate performance of a morphing flap can be influenced by its deformation. Owing to the intrinsic flexibility of every structure, and particularly in this case where the structure must possess the degrees of freedom necessary to be morphed, the deformation produced by the aerodynamic loading must be quantified for two reasons [17]: firstly, to insure that the measured performance are not biased by an excessive deformation of the flap; secondly, to better compare the measured aerodynamic performance with the that predicted numerically. For these reasons, the displacement and deformation of the flap produced by the aerodynamic loading has been measured exploiting an optical non-intrusive technique. A Qualisys system [18] has been employed. This technology, already used in the past within the European GLAMOUR project, exploits the use of multiple infrared cameras strategically positioned in order to frame the object to be studied from different point of view. The body is covered by a set of markers which are the targets that allow the software to reconstruct the displacement of the model. All images are processed, thus obtaining the relative displacements of the markers [19]. In this specific case, we used 8 Miqus M3 cameras (Fig. 5.93) providing a resolution of 1824 × 1088 pixel, with an acquisition frequency of 100Hz. Fig. 5.93 Qualisys Miqus camera, M3 models (2MP, 1824 × 1088 resolution, 340 fps)

226

F. Auteri et al.

Since this technology suffers from reflection of the infrared wavelengths, the entire flap and the force ribs were painted in matt black. The instrumentation set-up is shown in Fig. 5.94. The marker system consists of 21 adhesive-tape targets that reflect infrared wavelengths, positioned on the flap suction side along the spanwise and chordwise directions, see Fig. 5.95. The measurement procedure is preceded by a calibration phase through the use of a special target, shown in Fig. 5.96, whose dimensions are known by the system. The practical procedure consists in moving the instrument with circular movements in the measurement volume. The procedure should be carried out without the presence of the object to be measured, but in our case, due the impossibility of removing the model, all the markers were covered with a black blanket. In this way, it was possible to calibrate the system without removing the model. The calibration procedure has been repeated periodically during the tests, in particular every time the camber and the flap configuration was changed.

Fig. 5.94 Shape measurement set-up

Fig. 5.95 Adhesive reflective markers on flap

5 Aerodynamic Evaluation

227

Fig. 5.96 Instruments for calibration

A reference system was also placed on the model to refer the displacement of the points on the flap to the main wing. The reference system was made of 4 spherical markers (Fig. 5.96), placed on a single bracket, whose positions form a reference triad of axes. During the tests we placed two fixed triad, one on the wall of the test section and one on the main wing. In order to reduce the impact on the aerodynamic performance of the wing, the reference triad on the wing was made only with markers during the tests.

5.6.6 Results In this section we present the results of the experimental campaign. Nomenclature CP

Pressure coefficient

CL

Total lift coefficient

CD

Total pressure drag coefficient

CPT,wake

Coefficient computed from total pressure deficit across the wake

CL,wing

Main wing lift coefficient (continued)

228

F. Auteri et al.

(continued) Nomenclature CL,flap

Flap lift coefficient

CD,wing

Main wing pressure drag coefficient

CD,flap

Flap pressure drag coefficient

CM,wing

Main wing moment coefficient w.r.t. main wing leading edge

CM,flap

Flap moment coefficient w.r.t. flap leading edge

L2D

Bidimensional aerodynamic lift force

N/m

D2D

Bidimensional aerodynamic pressure drag force

N/m

M2D

Bidimensional aerodynamic moment

N

c

Clean wing chord (2.418 m)

m

α

Angle of attack

deg

U∞

Free-stream velocity

m/s

P

Local pressure

Pa

P∞

Free-stream static pressure

Pa

ρ

Air density

Kg/m3

PT

Total Pressure

Pa

PT,ext

Total Pressure out of the wake measured by rake Pitot probes

Pa

PIV

Particle Image Velocimetry

RMS

Root Mean Square

The definitions of the aerodynamic coefficients used below are reported: Pr essur e coe f f icient : Li f t coe f f icient :

CP = CL =

Pr essur e drag coe f f icient : Moment coe f f icient w.r.t. L E :

T otal pr essur e drag coe f f icient :

C P T,wake

P − P∞ 1 2 ρU∞ 2

L 2D 1 2 c ρU∞ 2

CD =

1 = c

(5.2)

D2D 1 2 ρU ∞c 2

C M,L E =

(5.1)

M2D,L E 1 2 c2 ρU∞ 2

(5.3) (5.4)

{l [PT,ext − PT (z)]dz (5.5) 0

5 Aerodynamic Evaluation

5.6.6.1

229

Pressure Measurement Results

During this campaign, static pressure measurements were acquired on the surface of the model, in the middle section, along the chord and along the spanwise section; in addition, some taps have been located along the chord to the right and left of the measurement section, in particular at ± 5 cm (Fig. 5.97). In this way it was possible to monitor the symmetry of the problem. Figure 5.97 shows the pressure distribution on the whole model, in the landing configuration, for the camber 3 flap geometry and 6° AoA. This condition is particularly critical, since we have maximum angle of attack and maximum camber on the flap. Notwithstanding, very good symmetry and two-dimensionality of the pressure distribution can be observed. Furthermore, the total pressure (PT ) has been acquired in the symmetry plane, in the wake of the airfoil. More precisely, each total pressure tap provides the difference between the local value of the total pressure and the free-stream static pressure (P∞ ). To evaluate the total pressure and the wind speed in the external region of the wake, the two Pitot tubes placed at the ends of the rake have been used. For this measure, the rake was moved in 4 different positions, reducing the effective space between the total pressure taps, and therefore increasing the spatial resolution. A total pressure profile across the wake is presented in Fig. 5.98 for the take-off flap configuration, camber 2 and 0° AoA, wind speed equal to 40 m/s. In this case the wake is quite thin, but the resolution of the total pressure profile seems totally adequate, since more than 20 total pressure measures are taken in the wake region, resolving quite well the high gradient at the boundary of the wake. It is interesting that two local minima of the total pressure, probably corresponding to the wake of the flap and to the wake of the main wing, can be distinguished. This observation is confirmed by PIV results, presented later.

Fig. 5.97 Pressure measured on all pressure taps on the model, example case: landing, camber 3, AoA = 6°, 40 m/s

230

F. Auteri et al.

Fig. 5.98 Total-pressure trend across the wake, example case: take-off, camber 2, AoA = 0°, 40 m/s

5.6.6.2

Effect of the Wind Speed on the Measured Aerodynamic Coefficients

The pressure distribution on the surface of the airfoil has been measured for three different speeds: 30, 34.1 and 40 m/s. Changing the wind speed mainly has an impact on three properties of the flow: the Mach number, the Reynolds number and the deformation of the flap. The variation of the Mach number is quite negligible, since this parameter is always less than 0.15, and therefore the flow can be assumed as incompressible. For what concerns the Reynolds number, considering an average wind tunnel temperature equal to 20 °C and a characteristic length equal to the clean chord (2418 m), the Reynolds numbers corresponding to the three tested speeds are 4.800.000, 5.5000.000 and 6.4000.000, respectively. Also in this case, since the order of magnitude of the Reynolds number is quite high and does not change dramatically for the three different speeds, we expect a small effect of the Reynolds number. Third, and more importantly, the increase in speed produces an increase in the acting forces which imply an increase in deformation and therefore a variation of the shape of the profile. Since the forces grow as the second power of the wind speed, we expect the impact of the flap deformation to be more important than that of the change of the Mach number and Reynolds number. The lift coefficient for the take-off configuration seems very marginally affected by changing the wind speed, as shown in Fig. 5.99. The three lines representing the three wind speeds can hardly be distinguished from each other, we therefore conclude that the flap deformation, while significant in these conditions, does not significantly affect the lift coefficient.

5 Aerodynamic Evaluation

231

Fig. 5.99 Take-off configuration, effect of the wind speed on the lift coefficient

Fig. 5.100 Landing configuration, effect of the wind speed on the lift coefficient

The pressure-drag coefficient, in contrast, depends significantly on the wind speed, Fig. 5.101. Except for a single point, whose reliability seems to be questionable and that will not be investigated further, there is a significant decrease in the pressuredrag coefficient as the wind speed is increased. This is not surprising, since both aeroelastic effects and Reynolds number effects are expected to produce a decrease of the pressure drag as the wind speed is increased, and hence both the aerodynamic loads and the Reynolds number, are increased. Indeed, increasing the aerodynamic loads will lead to a deformation decreasing the local angle of attack of the flap, as

232

F. Auteri et al.

Fig. 5.101 Take-off configuration, effect of the wind speed on the pressure-drag coefficient

Fig. 5.102 Landing configuration, effect of the wind speed on the pressure-drag coefficient

shown later on, and therefore we expect a reduced drag. Moreover, the boundary layer becomes thinner as the Reynolds number is increased, leading to a reduced pressure drag. These effects change the drag coefficient monotonically, as one would expect, and the behaviour is very similar for all cambers and angles of attack. We can therefore conclude that no strange phenomena occur in this range of angles of attack for the tested Reynolds numbers. The picture changes appreciably for the lift coefficient in landing configuration, Fig. 5.100. In this case we have that the baseline and camber 2 configurations are

5 Aerodynamic Evaluation

233

almost as insensitive to the change in the wind speed as in take-off configuration. The camber 1 configuration is slightly more sensitive, with the lift coefficient decreasing with the wind speed, probably due to the flap deformation, since we would rather expect an increase in the lift coefficient increasing both the Reynolds number and the Mach number. For camber 3, the decrease in lift coefficient wind speed becomes significant, of the order of a few percent, for the highest wind speed and angles of attack, also in this case probably due to the flap deformation. To interpret this phenomenon, the corresponding curves for the drag coefficient can be considered, Fig. 5.102. In this case, a dramatic drop in the drag coefficient is observed for the highest angles of attack. The drop is observed especially for the highest wind speed and angles of attack and is probably an effect of the deformation of the flap that causes the decrease of both the lift coefficient and the drag coefficient. Also in this case, one point of the curves for lift and drag (4° AoA, 30 m/s) can be considered an outlier, and therefore it will not be discussed further. For the other points, the picture is consistent with a monotonic decrease of lift and drag with the wind speed, due to an increase of the flap loading, and therefore a decrease of the effective angle of attack of the flap. Two considerations are in order, here. First, owing to its structural deformability, a morphing flap seems more prone to aeroelastic effects. Therefore, it is strongly advised that the investigation of the aerodynamic performance of a morphing flap take properly into account its deformation in order to correctly predict the actual performance under loading. Second, our measurements seem to indicate that, in the most promising configurations, aeroelastic effects have a stronger impact on drag (approx. 10%) than on lift (approx. 1%). This may give some space to further optimise the performance of the system by taking advantage of morphing. By investigating the same effect on the measurement of the total-pressure-defect, obtained by integrating the total-pressure deficit across the wake, we observe that measures are more scattered. This fact is probably due to an insufficient acquisition time. However, this measurement was the most time consuming for the first part of the measurement campaign, and increasing the acquisition time would have decreased significantly the number of test points in the test matrix. In any case, it is worthwhile to remember that these data, are intended to give a rough estimate of the viscous contribution to the total drag. The results for the take-off configuration are reported in Fig. 5.103. The trend is similar for the baseline, camber 1 and camber 3 geometries, with a minimum of the total-pressure defect observed between 2° and 4° AoA. For these geometries, however, we have a stronger variation of the coefficient as the camber is increased. Interesting enough, the defect is lower for the more cambered geometries for the lowest angles of attack, but it increases faster with the angle of attack as the camber is increased. For camber 2, we observe a different trend, with a monotonic increase of the defect with the angle of attack. A slight effect of the free-stream speed is observed, qualitatively similar to that observed for the pressure drag. For the landing configuration, we observe instead a monotonic increase of the total-pressure defect for all the flap geometries, as shown in Fig. 5.104. The trend is very similar for the baseline, camber 1 and camber 2 geometries, with an almost

234

F. Auteri et al.

Fig. 5.103 Take-off configuration, total-pressure defect measured for two different free-stream speed and four flap geometries

linear increase of the defect as a function of the AoA. It is interesting to note that the defect is lower for the more cambered configurations, suggesting the camber 1 and, especially, camber 2 could lead to a significantly increased efficiency with respect to the baseline. This is somewhat surprising, given the flow fields measured by PIV and discussed later on, and seems worth of further investigation. In this case the effect of the free-stream speed is less pronounced than for the take-off. This can be due to the fact that when an extended separation occurs, as in the present case, that is not affected by aeroelastic effects, the impact of the Reynolds number on the total-pressure defect is quite small, since turbulent diffusion is dominant. Quite different is the case of the most cambered flap. Here, we observe a strong increase of the total-pressure defect as we increase the angle of attack, and, for AoA = 4° and AoA = 6°, also as we increase the free-stream speed. This last effect is in contrast with what has been observed for the rest of the cases, and could be due to a Reynolds number effect or deformation effect that increase significantly the separated area.

5.6.6.3

Effect of Camber on the Aerodynamic Performance

In order to evaluate the effects that changing the camber of the flap produces on the aerodynamic performance of the wing, it is interesting to compare the pressurecoefficient distributions obtained for different cambers for the same angle of attack and flap configuration. The results are reported in from Figs. 5.105, 5.106, 5.107, 5.108, and 5.109 for the take-off configuration, and from Figs. 5.110, 5.111, 5.112, and 5.113 for the landing configuration.

5 Aerodynamic Evaluation

235

Fig. 5.104 Landing configuration, total-pressure defect measured for two different free-stream speed and four flap geometries

Fig. 5.105 Distribution distribution of the pressure coefficient on the symmetry plane section for different cambers. Take-off configuration, AoA 0°, 34.1 m/s. The abscissas axis has been nondimensionalised by the chord of the clean configuration

Since the results are quite similar for different free-stream speeds, and for brevity, the results obtained for free-stream speed of 34.1 m/s, corresponding to Mach = 0.1, will be considered, to better compare with the numerical simulations that were carried out by the other project partners. Considering, for example, the distribution of the pressure coefficient reported in Fig. 5.112, which corresponds to the landing configuration, an AoA = 4° and

236

F. Auteri et al.

Fig. 5.106 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Take-off configuration, AoA 2°, 34.1 m/s

Fig. 5.107 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Take-off configuration, AoA 4°, 34.1 m/s

a velocity of 34.1 m/s, it is clear that the effect of changing the camber is quite small, but still well measurable. The pressure distribution on the pressure side is very similar for the four flap geometries. Here the main difference is observed in the rear region of the main wing and in the central portion of the flap, were a higher pressure is measured for more cambered geometries. The opposite phenomenon is observed on the suction side, were the pressure coefficient decreases as the camber

5 Aerodynamic Evaluation

237

Fig. 5.108 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Take-off configuration, AoA 6°, 34.1 m/s

Fig. 5.109 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Take-off configuration, AoA 8°, 34.1 m/s

is increased. The two phenomena both lead to an increase of the pressure force. To be more precise, we observe that the baseline and camber 1 flap lead to very similar pressure distributions on the main wing. Here, the camber 2 and 3 show a sensibly lower pressure on the suction side, especially on the leading edge, were the low-pressure peak is almost 10% lower. The situation is more involved on the suction side of the flap. The low-pressure peak on the flap, for instance, is lowest for

238

F. Auteri et al.

Fig. 5.110 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Landing configuration, AoA 0°, 34.1 m/s. The abscissas has been nondimensionalised by the chord of the clean configuration

Fig. 5.111 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Landing configuration, AoA 2°, 34.1 m/s

the camber 3, then we have camber 2 and the baseline. Camber 1 is the geometry leading to the smallest peak on the flap leading edge. In the central part of the flap, in contrast, the baseline has the highest pressure, camber 2 and 3 the lowest, while camber 1 is in a somewhat intermediate position. The pressure distributions of the baseline and camber 1 exchange their respective positions again in the flap rear end.

5 Aerodynamic Evaluation

239

Fig. 5.112 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Landing configuration, AoA 4°, 34.1 m/s

Fig. 5.113 Distribution of the pressure coefficient on the symmetry plane section for different cambers. Landing configuration, AoA 6°, 34.1 m/s

These observations give a detailed explanation of the relative positions of the curves in Fig. 5.113. The most visible effect on the pressure-coefficient distribution consequent to a variation of the AoA, as visible from Figs. 5.110, 5.111, 5.112, and 5.113 is the suction peak on the leading edge of the main airfoil that becomes more intense as the angle of attack is increased, as expected. In contrast, we observe that the suction peak

240

F. Auteri et al.

Fig. 5.114 Percentual variation of partial lift coefficient with respect to the baseline configuration

on the leading edge of the flap is only slightly affected by the change in the AoA, meaning that this peak is more related to the geometry of the slot between the flap and the main wing than to the incidence. Another effect of the AoA is a slow change of the amplitude of the separated region, witnessed by the increase of the portion of the flap near its trailing edge where the pressure coefficient is approximately constant. The main difference between this case and the landing configuration is the behaviour of the pressure coefficient on the flap. Here we observe a less pronounced,

Fig. 5.115 Percentual variation of partial pressure drag coefficient with respect to the baseline configuration

5 Aerodynamic Evaluation

241

more rounded suction peak on the leading edge. A behaviour similar to that observed for the landing configuration is observed with respect to the effect of the AoA. Also in this case, the main effect is the impact on the suction peak near the leading edge of the main airfoil, while the suction peak on the leading edge of the flap remains quite unaltered. It is interesting to observe the effect of the flap geometry. The baseline and camber 3 geometries present the most pronounced suction on the leading edge of the flap, while camber 1 the lowest. In general, camber 1 seems to provide the less intense suction on the upper side of the flap, an effect that is counter-balanced by an increased suction in the main airfoil and a slightly increased pressure on the bottom side of the flap. The difference between the baseline and camber 3 is particularly evident in the central region of the flap, where the more cambered geometry provides a definitely increased suction, thus leading to the increased lift coefficient. Less evident, but similar, is the behaviour for camber 2. The pressure coefficient is very similar for all geometries in the aft portion of the flap for all the angles of attack and geometries, except for camber 3, that shows a slightly higher suction in this region for AoA = 6° and 8°, probably due to incipient separation. To better understand the quantitative impact of the modification of the flap camber, we consider separately the aerodynamic coefficients of the main wing and of the flap, evaluating the percentual variation of each one with respect to the value corresponding to the base configuration (Figs. 5.114, 5.115 and 5.116). The results show that cambering has an important effect also on the performance of the main wing. As shown in Fig. 5.114, increasing the camber results in all cases in an increase of the lift and a decrease of the drag of the main wing. The lift and drag of the flap increase for the camber 2 and 3 geometries, while it decreases substantially for the camber 1. The moment coefficient, positive nose-up, decreases with increasing camber due to the shift of the pressure centre towards the trailing edge, except for camber 1. In Figs. 5.114, 5.115 and 5.116, we show the overall percentual variation produced by morphing for all tested configurations (Mach = 0.1) on lift coefficient, pressure drag and total-pressure defect in the wake, respectively. Considering the take-off configuration, the desired effect is an increase of the overall efficiency of the airfoil, especially for the highest angle of attack, that could lead to reduced runway lengths [15]. Unfortunately, in all cases the substantial increase of the lift is counter-balanced by an even more important increase of the drag, so that the most efficient configuration is still the baseline. The situation changes if we consider the landing configuration. In this case the goal of the shape optimisation is to increase the lift coefficient [15]. A decrease in the efficiency is not critical in the landing phase, in fact it can be beneficial since it leads to increased landing approach angles that are good to avoid obstacles near the runway and to improve the visibility and precision of the approach. In this case, we observe that camber 3 geometry offers the highest lift coefficients for all angles of attack, and therefore it is the geometry to be preferred.

242

F. Auteri et al.

Fig. 5.116 Percentual variation of partial moment coefficient with respect to the baseline configuration

Fig. 5.117 Percentual variation of lift coefficient with respect to the baseline configuration

It is of the utmost importance to notice that the optimal shape is different for the take-off and landing configurations, therefore perfectly justifying the need for a morphing flap, in order to exploit the best geometry in all conditions. Figure 5.120 shows the percentual variation of the ratio of the lift coefficient to the pressure drag, for the take-off (left) and landing (right) configuration. The results show that the efficiency always decreases as the camber is increased for the take-off configuration, and therefore the baseline is the most efficient geometry. It is interesting to observe, however, that the difference between the efficiency of the morphed geometries and that of the baseline decreases as the angle of attack is increased, especially for camber 1. It is therefore possible that the efficiency of the

5 Aerodynamic Evaluation

243

Fig. 5.118 Percentual variation of pressure drag coefficient with respect to the baseline configuration

Fig. 5.119 Percentual variation of wake total pressure loss coefficient with respect to the baseline configuration

Fig. 5.120 Percentual variation of the ratio between lift coefficient to the pressure-drag coefficient with respect to the baseline configuration

244

F. Auteri et al.

latter geometry reaches that of the baseline for higher angles of attack not tested in the present investigation. Concerning the landing configuration, the figure clearly shows that camber 1 gives a dramatic improvement of the efficiency of the order of 8% for all tested angles of attack, camber 2 gives a modest improvement, while camber 3 gives a penalty. In conclusion, the experimental results presented so far clearly show that the use of a morphing flap cambering is effective to increase the aerodynamic performance of the airfoil. In Fig. 5.121, the ratio of the lift coefficient to the total-pressure-defect coefficient for the take-off configuration, for three different speeds and four geometries, shows the increase of this ratio as the camber increases. This indicator has been judged interesting especially for this configuration, were we expect viscous stresses to be important for the overall efficiency. The figure shows a very similar trend for the baseline and camber 1, with a marked increase as a function of the angle of attack. The results for the three free-stream speed are very similar. This parameter is consistently higher for camber 1 than for the baseline geometry, suggesting that camber 1 is more gentle on the boundary layer. For camber 2, instead, we have always a monotone increase, but with a smaller slope, which probably means that we have a faster increase in the boundary layer thickness with the angle of attack and/or incipient separation. For camber 3, the behaviour is no longer monotone increasing, but for the highest angles of attack, the ratio is decreasing. In this case we probably have separation. Starting from these results, we have chosen to further study camber 1 together with the baseline configuration with PIV surveys as for high angles of attack, which are more interesting for the take-off phase, it proves to be the most efficient and stable. A similar representation is reported in the bottom row of Fig. 5.122. The situation here is definitely more complicated. For the baseline flap geometry, we have an almost monotonic increase of the ratio, even if the slope is smaller than for the takeoff. The same is true for camber 1 for all angles of attack, but only for the highest free-stream speed. For 30 and 34.1 m/s we observe that the maximum ratio is reached for 2° angle of attack, and then the performance deteriorates. A similar behaviour is observed also for camber 2, while camber 3 shows an almost monotonic decrease of the performance.

Fig. 5.121 Ratio between the lift coefficient and the total-pressure-defect coefficient, bottom row, for the take-off configuration, three free-stream speeds and four flap geometries

5 Aerodynamic Evaluation

245

Fig. 5.122 Ratio between lift coefficient and the pressure-drag coefficient, top row, and between the lift coefficient and the total-pressure-defect coefficient, bottom row, for the landing configuration, three free-stream speeds and four flap geometries

The same figure also reports the ratio of the lift coefficient to the pressure-drag coefficient, in the top row. In this case we have a monotonic increase of the performance for all cambers, and the maximum value seems to be attained for AoA = 6°. The best performance is obtained with camber 1, while the performance of the baseline and camber 2 geometries is very similar. Camber 3 seems to be the less performant geometry with respect to this indicator. In this case, we opted to analyse, together with the baseline geometry, the camber 1 and camber 2.

5.6.7 PIV Campaign Results Due to the available wind-tunnel time, the test matrix for the PIV measurements was reduced with respect to the pressure measurements. The configurations tested during the PIV measurement campaign are chosen through the observation of the previous results. The configurations tested in this phase are listed in Tables 5.9 and 5.10 for the take-off and landing configuration, respectively.

5.6.7.1

Take-off Configuration

For the take-off configuration, the baseline and camber 1 cases have been selected, even though camber 2 and 3 provided higher lift. The reason for this choice is

246

F. Auteri et al.

related to the fact that camber 1, despite giving almost the same lift as the baseline configuration, provides the least increase in drag, without a dramatic change in the performance that would render the comparison rather complicated. A comparison of the pressure distribution on the flap and the flow field obtained from the PIV surveys is reported in Fig. 5.123 for the baseline geometry, left, and for camber 1, right. The higher peak in the pressure coefficient for the baseline geometry seems related to the smaller camber that reduces the Theodorsen angle and therefore increases the load on the leading edge with respect to camber 1, where the load is displaced downstream. The increased camber seems to lead to an increased pressure gradient in the aft portion of the flap, producing an incipient separation, as highlighted by the PIV surveys in the bottom line of the figure. In Fig. 5.124, the distribution of the absolute value of the mean velocity is reported as a function of the angle of attack with the superposed streamlines obtained for 34.1 m/s free-stream velocity. From this image, it is possible to notice the presence, the position and the intensity of the wake of the main wing in the region of investigation, in particular in the upper left part of the window. This wake, that has been already noticed in the plot of the total-pressure defect measured downstream of the airfoil, is well separated from the boundary layer of the flap so that the interaction of the two rotational regions is negligible. Moreover, it is possible to notice that the velocity in the gap between the flap and the main airfoil increases with the angle of attack, an indication of the correct operation of the flap.

Fig. 5.123 Pressure distribution and PIV averaged velocity magnitude (case: 34.1 m/s, AoA = 4°, take-off configuration)

5 Aerodynamic Evaluation

247

Fig. 5.124 Mean velocity magnitude and streamlines, all take-off tested cases for speed 34.1 m/s

By comparing the first row of visualisations in Fig. 5.124, relative to the baseline geometry and the second row, relative to the camber 1 geometry, we observe that the latter presents a thin separation near the trailing edge, as witnessed by the higher distance between the streamlines and the low speed region, colored in blue, near the airfoil surface. This separation region produces a thickening of the boundary layer near the trailing edge, thus increasing the pressure drag. This effect can be better appreciated by looking at the bottom of Fig. 5.124, were the larger distance between the streamlines near the trailing edge can be clearly seen.

5.6.7.2

Landing Configuration

A larger number of different geometries have been tested for the landing configuration, which is more interesting for the purpose of the project, since it shows an improvement of the performance obtained by morphing the flap. Unlike the take-off case, the landing configuration has large flow separations in the trailing edge area of the flap. This fact is also visible from the distribution of the pressure coefficient obtained previously, where the pressure coefficient is almost flat in a rather large portion of the flap surface near the trailing edge. In Fig. 5.125, a comparison between the pressure measurements on the airfoil and the mean velocity field is reported, obtained using the PIV technique. As visible in the left panel of this figure, the pressure coefficient remains almost constant starting from x/c = 1.025 on the suction side of the airfoil. This corresponds almost exactly to the place where the recirculation bubble starts, as can be observed in the right panel of the figure. The same figure shows that no separation occurs on the pressure side, as expected. The presence of a strong separation is usually associated with unsteady phenomena. Therefore, it could be interesting to exploit the velocity snapshots to

248

F. Auteri et al.

Fig. 5.125 Pressure distribution and PIV mean velocity magnitude (case: 34.1 m/s, AoA = 4°, Landing configuration

compute the rms (root mean square) value of the velocity and the 2D components of the Reynolds-stress tensor, which give an indication of the turbulence level and can be useful to validate and refine numerical simulations. Let us consider, for example, the landing configuration for AoA equal to 0° and compare the results for the baseline geometry, where there is no separation, and those for the camber 2, where we observe a quite extended separation. The mean velocity field and the streamlines are reported in Fig. 5.126, while the rms values of the two velocity components are shown in Fig. 5.127. The mean-velocity field clearly shows that the effect of the camber in this condition is to cause separation in the trailing edge of the flap. The RMS information

Fig. 5.126 Mean velocity magnitude and related streamlines, comparison between two landing cases: AoA = 0°, velocity 34.1 m/s, baseline and camber 2

5 Aerodynamic Evaluation

249

Fig. 5.127 RMS of the two velocity components, comparison between two landing cases: AoA = 0°, velocity 34.1 m/s, baseline and camber 2

allows us to understand the fluctuations of the two speed components. In Fig. 5.127, it can be clearly seen that in the case of camber 2 these fluctuations are considerably higher, reaching values of about 30% of the asymptotic speed for the u component, and about 20% for the v component. The fluctuations of the longitudinal component of the velocity are particularly strong in the region across the streamline separating the recirculation region from the outer stream. The thickness of this region grows moving downstream, as the velocity fluctuations are amplified by the Kelvin-Helmholtz instability [20]. The fluctuations of the vertical velocity component are less intense, the intensity growing moving downstream. No fluctuations can be detected for the baseline configuration, confirming that the flow is attached in this case. Further information on the turbulent fluctuations can be obtained by looking at the Reynolds stresses. In Fig. 5.128, we show the three main components of the Reynolds stress tensor, through which turbulent kinetic energy and turbulent shear stresses can be calculated. Despite the normal turbulent stresses are strictly related to rms values, their plotting gives interesting pieces of information, especially for the baseline geometry. Indeed, for the baseline geometry, we observe how the turbulent fluctuations are

250

F. Auteri et al.

Fig. 5.128 Reynolds Reynolds stresses, comparison between two landing cases: AoA = 0°, velocity 34.1 m/s, baseline and camber 2

concentrated in a region which is thinner than for the camber 2 geometry. Moreover, this region is definitely closer to the flap surface. The normal Reynolds stress in the longitudinal direction, < uu >, has its maximum in a region which is farther from the flap surface than its transverse counterpart, < vv >. A local maximum can also be observed for both these stresses in the wake of the trailing edge of the flap. This is

5 Aerodynamic Evaluation

251

true for both geometries. By looking at the turbulent shear layer, < uv >, we observe that its peak is even nearer to the flap surface, in both cases. In Fig. 5.129, we report the mean velocity magnitude and the streamlines obtained for all the tested landing configurations for 30 m/s free-stream velocity. We observe a quite different behaviour for the different cambers. For AoA = 0°, the flow over the baseline geometry seems attached. For camber 1, we have a small separation bubble near the trailing edge, while for camber 2 quite extended separation, with a large recirculation bubble, is present. Increasing the angle of attack to 2° has the effect to increase the magnitude of the separated region. It is interesting to observe here that the baseline geometry shows a large recirculation bubble, actually larger than that observed for camber 1 and similar to that observed for camber 2. For higher angles of attack, all configurations appear deeply separated, with the size of the separation bubble becoming particularly extended for camber 2 and 4° and 6° AoA. For 4° AoA, camber 1 still seems to perform better that the baseline, with a slightly smaller separated region. Further investigation on how different geometries could favorably impact the performance of the flap is a direct outlook of this study. A comparison with numerical results from INPT/IMFT shows a good agreement between the experimental and numerical approaches concerning the separation area (Fig. 5.130 for the static case, and Fig. 5.131 for the morphing with camber 1).

Fig. 5.129 Averaged velocity magnitude and streamlines, all landing tested cases for speed 30 m/s

252

F. Auteri et al.

PIV Experiments - POLIMI

Simulations - NSMB code - INPT/IMFT

Fig. 5.130 Comparison between PIV experiments and numerical simulations

5 Aerodynamic Evaluation

253

Fig. 5.131 Dimensionless velocity magnitude of numerical and experimental PIV results for different angles of attack at camber position 1. Left: numerical, right: experimental, a, b: 4◦ , c, d 8◦

5.6.8 Shape-Measurement Results The measurement system generated an output composed of the x, y, z coordinates of each marker with respect to a fixed reference system, as described above. From these measurements, through a linear interpolation it is possible to build a grid surface with multiple points and evaluate the displacement of the flap surface as a function of the speed with respect to the zero-displacement condition at zero velocity. In order to have a treatment consistent with the rest of the data, an appropriate transformation was used to fix the reference system to the leading edge of the main wing, at mid-span. Figure 5.132 shows the interpolated grid built from the marker data, it is also possible to see the position of the reference fixed triad on the main wing.

254

F. Auteri et al.

Fig. 5.132 Position of the markers for measuring the displacement and interpolated grid for the take-off configuration of the flap, baseline geometry, AoA = 0°

Fig. 5.133 Displacement as a function of the wind speed for the take-off configuration, camber 3, AoA 8°

5 Aerodynamic Evaluation

255

It is important to remember that the goal of these measurements is not to provide the exact position of the flap, that will depend on several factors including the actual assembly process and camber adjustment, but to evaluate the change of the flap shape under load, assuming the case without wind as a reference. Let us consider the take-off configuration and take the higher load case, namely AoA = 8° and Camber 3. The results, reported in Fig. 5.133, confirm the good symmetry of the displacement, and therefore of the flow. The figure shows that the displacements along the y direction are negligible with respect to those in the two other directions. The components of the displacement along the x and z directions are not negligible, with values well above one millimeter. Their distribution shows that the displacement mainly concerns the trailing edge of the flap, in the central region which is most distant from the constraints applied at the extremities and therefore in the region where the structure is more deformable. The percentual displacement, with respect to the chord of the flap, is not high, order of 1 or 2%, for the tested wind speed. The variation from 20 to 30 m/s of wind speed seems consistent with a quadratic law. Because loads scale quadratically with the wind speed, we therefore expect a displacement order of 4% for the maximum speed attained in the present tests. It may be interesting to study the effect of camber on these results. Considering for example the same case of Fig. 5.134 for different cambers, we observe

Fig. 5.134 Effect of camber on flap deformation along z direction

256

F. Auteri et al.

Fig. 5.135 Deformed airfoil geometry, take-off configuration, Camber 3, AoA 8°, speed 0–20– 30 m/s

that by increasing the camber, greater deformation of the flap is obtained, particularly appreciable by observing the displacement component along z direction, see of Fig. 5.134. In order to obtain the actual shape under load of the two-dimensional airfoil geometry in the mid-span, useful for future numerical simulations, the displacements measured in the z and x directions are linearly interpolated and applied to the points of the profile. Introducing the hypothesis of extending the results found for the suction side to the pressure side, which means that the thickness of the flap does not change, the following deformed profiles are obtained as functions of velocity, cambering, AoA and configuration. For brevity, only the cases of maximum aerodynamic load for take-off and landing configurations are reported in Figs. 5.135 and 5.136. These figures show that deformations are quite small, but still visible.

5.7 Data Sharing of SMS and Workflows Through Ontology-Based Data Access in the Platform CALMIP/CALLISTO T. Louge, J. B. Tô, C. Jimenez-Navarro, A. Marouf and M. Braza The SMS project’s coordinator and the partners participate in the open science and open data initiative DATANOOS, “From data to Noosphere”, https://datanoos.univtoulouse.fr/on data sharing and interaction towards improvements of scientific and technical approaches and in view of progressive open access.

5 Aerodynamic Evaluation

257

Fig. 5.136 Deformed airfoil geometry, landing configuration, Camber 3, AoA 6°, speed 0–20– 30 m/s

In the context of the SMS project, the Supercomputing center member of INPT “Calcul en Midi-Pyrénées” CALMIP has strong links with the SMS Coordinator Institute INPT and created the repository CALLISTO, “CALmip Launches an Interface for Semantic Toolbox Online”, https://callisto.calmip.univ-toulouse.fr/. This has been used to deposit data, exchange, reuse and cross-fertilise the approaches in the sense of FAIR (“Findable, Accessible, Interoperable, Reusable”) requirements in the Open Science framework. This repository is not only an area to deposit the data but it has built specific semantic tools, “ontologies” allowing finding and cross-operating the data. In this context, the partners and the coordinator of SMS collaborate in the UseCase1 (UC1) of the platform DATANOOS, “Enhanced and intelligent aerodynamic modelling”, https://datanoos.univ-toulouse.fr/en/use-case-enhanced-and-int elligent-aerodynamic-modeling. Thanks to the morphing, the modification and optimisation of airplane’s trajectories are studied, in respect of the aerodynamic forces modification. A significant and representative ensemble of final data corresponding to the major outcomes of the project have been uploaded and suitably described in the private area of the platform CALLISTO. Three main data sets have been stored in the repository, in three separate sections containing: • the distribution of the pressure coefficients on the lifting surfaces, for the final results. In addition to the coordinates of the pressure transducers and of the numerical simulation results, wind tunnel and testing parameters are provided for the SMS prototypes; • the aerodynamic coefficients from the experimental and numerical studies of the project for all the tested cases, useful for faster consultation; • the elaborated results of the PIV measures for selected cases concerning the SMS prototypes.

258

F. Auteri et al.

The UC1 created an efficient ontology allowing the FAIR practice and facilitating the data exchange and interoperability among the SMS partners as well as to create suitable services through appropriate workflows, allowing intelligent post-treatment of the data. This ontological representation makes possible to link the data, articles and softwares used. An example of a data-paper elaborated in open access after the SMS project, Tô et al. [21] describes how the workflows developed within the CALLISTO’s ontology allow automatic, on-the-fly reprocessing of data. A subset of vocabulary elements defined inside the ontology have been defined into supplemental metadata for the project’s dataset. The ontology linked to the project and used by CALLISTO is also part of the results of the project itself and provides the formalization of the conceptualization of the data and of the scientific context summarizing the work of the project itself on its data management component.

5.7.1 The Sharing of Data for SMS The SMS project has used the CALMIP computing center for the Hi-Fi simulations using the NSMB CFD solver. Building from this collaboration, members of the SMS consortium have been involved with the CALMIP staff in the definition of a platform for collaborative science. This platform, named CALLISTO is now made available for public. SMS data are shared by the means of this tool, that ensures access to the data available in the platform. Figure gives an overview of the SMS project from the point of view of collaboration among experts with different scientific background (aeronautics, fluid mechanics, structural mechanics, control system…). Different teams from different countries worldwide made their own experiments in their competency areas, and then needed to agree on common concepts to work with one another. The software libraries used by the various teams may be shared, and projects papers shall be understood by all the project’s stakeholders. A major requirement is then to have a common metadata set that all stakeholders understand, and all future users will also understand. Meanwhile, metadata shall also be able to express very narrow elements for specialists. In order to share data among the project members (private access) or publicly (through DOI attribution and open access) when needed, CALLISTO provides a shared space with storage space and metadata browsing. This is provided by Dataverse, which is used as a metadata catalog and data registry in CALLISTO as shown in Fig. 5.138. CALLISTO embeds Dataverse in a wider environment that goes beyond metadata browsing, data storage and DOI attribution. This environment is meant to provide fine-grained data description and enhance data interoperability. Another important requirement for experiments reuse and reproducibility, is a good understanding of the analysis processes the data went through [22]. Reproducible science requires a deep understanding of the data, of the scientific claims they support and of the overall context of the experiments. In CALLISTO, fine-grained data description, interoperability and understanding of data analysis software and pipelines is ensured by the means of a dedicated ontology that

5 Aerodynamic Evaluation

259

Fig. 5.137 SMS project overview for CALLISTO

Fig. 5.138 SMS Project data storage

has been co-constructed by the SMS project members and CALMIP technical staff. This ontology and its use in the platform for SMS is depicted in the following section.

5.7.2 Ontology-Based Data Description and Workflow Execution Ontologies are at the heart of the Semantic Web (SW) for several reasons, as underlined by Tim Berners-Lee in his grounding paper [23] and its update in 2006 [24]. Ontologies model the knowledge they contain in a shared, formalized and evolving

260

F. Auteri et al.

fashion. The degree of accuracy of this modeling can be as precise as desired. Moreover, they represented the concepts and properties with a formal semantics that make it possible to assign a formal type to domain-level entities. Formal semantics also enables to define axioms and to infer new facts with the help of a logic-based reasoner. Ontologies include more explicit relationships, annotations, properties, definitions, and comments than can be indicated on the UML (Unified Modeling Language) graph of, for example, a metadata set. Their major advantage compared with keyword metadata is that each item in the ontology has a unique identifier which reduces ambiguities and multiple inconsistent definitions. Ontologies are by nature intended to be browsed and interpreted by algorithms as well by humans. For this reason, they benefit from several advanced tools allowing their edition by graphical means [25], inline browsers (http://vowl.visualdataweb. org/webvowl.html) and several query languages (SPARQL is one of the most wellknown). CALLISTO embeds a core ontology named ARCAS: “Arming CAllisto with Semantics”, which is domain-agnostic and needs refining and specialization in order to tailor specific project needs. Generically speaking, those specialized ontologies are named “ARCADIE” (for ARCAS Domain Implementation, ARCAS enriched and populated for a specific project). Taking benefit from the CALMIP staff, the SMS project members came out with a version of ARCADIE specifically adapted for SMS. A comprehensive view of SMS scientific whereabouts, tools and results is offered by the semantics through ARCAS/ARCADIE and its modules shown in Fig. 5.139 that shows how ARCAS and ARCADIE handle these elements. ARCAS embeds several ontologies in order to provide a fine-grained domain description. The bibliography for a project, written by the project’s partners or external to the project, is described using a Micropublications ontology. SWO (SoftWare Ontology) [26] describes software libraries in their non-grounding aspects (e.g. algorithms provided, languages…) and GEOS, coming from ASON [27] is used for grounding elements (e.g. Inputs and outputs combinations, URLS, parameters…). Both bibliography and software context are related to domain-specific elements either coming from external ontologies or from the description of the domain coming from the project stakeholders themselves. In any case, reusing existing ontologies or providing domain description in ARCADIE needs ontological competencies. These skills are available within the CALMIP team, which provides its users with a data steward to help them understand, build and use the ontologies related to their project.

5.7.2.1

Metadata Generation from Ontological Description of Knowledge

SMS ontology, coming from ARCAS specialization for SMS project serves two main purposes. The first goal is to provide a fine-grained description of the scientific whereabouts, methods and software used in the project. Some of these elements may be used as specific metadata for SMS datasets, allowing a more detailed definition than the native metadata sets available in Dataverse. Figure 5.140 shows a subset of

5 Aerodynamic Evaluation

261

Fig. 5.139 SMS Project and ARCAS/ARCADIE

SMS metadata deriving from SMS ontology available in the CALLISTO platform for dataset description. Deriving metadata elements from ontological description of SMS scientific context allows to modify metadata sets (removing or adding elements, make elements mandatory or optional, specify predefined values…) easily by editing the ontology and generating a new metadata set by using CALMIP software extracting the elements from the ontology. An important issue regarding data “FAIRisation” is to be able to reproduce or evaluate the content of scientific articles using either the same or comparable data. Another possibility based on the same principles is to use different data following the same methodology as a specific article, or to find related data to highlight stated arguments. As SMS uses CALLISTO, the use of Micropublications, SWO, GEOS, and disciplinary field description facilitates this work. This requires having publications described within the ontology and therefore registering them there. However, populating an ontology cannot be considered a routine task for a platform user. Therefore, CALLISTO offers a bibliography entry interface that only requires the user to know the scientific concepts expressed in the ontology and relevant for the description of the article to be added. In this interface, the user enters bibliographic items into the ontology for a project. This process is made in a paper-by-paper basis. The user extracts manually the claim(s) exposed in the paper, together with bibliographic elements (title, authors…). Relevant ontology concepts are also mandatory in this interface, as they allow the linking of paper with datasets and software elements.

262

F. Auteri et al.

Fig. 5.140 SMS—specific metadata in CALLISTO

These concepts can be identified by using either SPARQL queries or a graphical ontology browsing. Figures 5.141 and 5.142 show this bibliographic entry interface, Fig. 5.142 presenting the availability of graphical browsing (“Graphic ontology search”) or SPARQL querying (“Advanced ontology browse”).

5.7.2.2

Semantic Workflow Composition and Execution

Building from data, scientific domain and software description, CALLISTO uses reasoning capabilities offered by ARCADIE to compose pipelines, chaining data access and data analysis. Such pipelines are designated by the term “workflows”. The starting point for workflow composition and execution is always a dataset deposited in the CALLISTO Dataverse instance. When a user queries the interface and a set of data available relevant for the query is proposed, CALLISTO queries ARCADIE and presents workflows that the user may run from the data. Those workflows are not set in stone inside ARCADIE, but calculated on-the-fly following the ontology

5 Aerodynamic Evaluation

Fig. 5.141 Bibliography entry interface, part 1

Fig. 5.142 Bibliography entry interface, part 2

263

264

F. Auteri et al.

content. When a SMS member deposits data inside CALLISTO, the corresponding metadata are automatically ingested by the ontology and linked, through the description capabilities, to the other elements (e.g.: software, scientific concepts) already present in the ontology. This linking made by the ontology is at the core of the reasoning leading to workflows composition, and the process of ingesting metadata from Dataverse and translating those metadata to ARCADIE description takes place every hour. Examples of workflows are presented hereafter.

5.7.3 Practical Use of CALLISTO for SMS and Data Reusability Providing a user-friendly interface for the repository of datasets, the browsing of SMS ontology and the composition of workflows is an important point. A user-friendly interface is what will make the tool usable for SMS project members and for public use after the end of the project with public data coming from SMS results. Efforts have been made in this regard during the conception of the CALLISTO web interface (https://callisto.calmip.univ-toulouse.fr). This interface, through the menus “tools” and “Sada” (Semi-Automatic Data Analysis) allows the user to select datasets and gives an assistance during the workflows composition (e.g.: choosing the analysis methods to be used, understanding the different steps).

Fig. 5.143 Service composition in CALLISTO

5 Aerodynamic Evaluation

265

As an example, the PSD presented in Fig. 5.144 can be calculated on-the-fly using CALLISTO capabilities, as shown in Fig. 5.143. Another example, the following file (https://doi.org/10.48531/jbru.calmip/p76fwg), provided in the Callisto Dataverse repository, exposes its metadata and a scientific analysis can be produced by using a suitable workflow involving the right scripts that will extract the desired iso-contours from this file as shown in Fig. 5.145. The three-dimensional flow dynamics around the tRs prototype are shown in Fig. 5.146.

a) PSD (Power Spectral Density) diagram of the pressure signal monitored at the point indicated in figure 8b of the tRS prototype for an actuation frequency of 750 Hz

b) Position of the monitor point (red) for the pressure signal used in the workflow evaluating the PSD

Fig. 5.144 Workflow in CALLISTO for instantaneous evaluation of the a PSD (Power Sectral Density) from the data of monitor point indicated by the red circle on the profile b example: tRS prototype

266

F. Auteri et al.

Fig. 5.145 Iso-density contours, Morphing A320 wing, transonic regime, actuation frequency at 720 Hz near trailing edge, by means of piezo-actuators

Fig. 5.146 Iso-contours of the Q criterion colored by the Mach number and showing the turbulence structure and formation of secondary instability shells, [28, 29]

5.7.4 Conclusions on the Data Sharing Access By the means of CALLISTO Dataverse instance and semantic reasoning capacities, SMS public data are integrated in a series of tools for comparing and analyzing SMS datasets. Public datasets are fully accessible and may be re-processed online by the semi-automatic workflows composition proposed in the Web user interface. The

5 Aerodynamic Evaluation

267

underlying ARCAS ontology is downloadable (https://callisto.calmip.univ-toulouse. fr/ARCAS.owl). The SMS declination of ARCAS is also available online (https:// callisto.calmip.univ-toulouse.fr/SMS.owl), though its actual content may vary with ingestion of new data, services, and domain elements throughout exploitation of enriched data after the project.

5.8 Conclusions Work Package 5—WP5 The experimental activity on the large-scale prototype planned in the Work Package 5 of the SMS project and concerning the aerodynamic evaluation of the morphing Large Scale prototype flap has been carried out in the GVPM wind tunnel of Politecnico di Milano (POLIMI) for what concerns the take-off and landing configuration. The design of the high-lift LS prototype by INPT/IMFT-LAPLACE and NOVATEM has been successful in respect of adaptation in the fixed wing configuration by POLIMI and has proved ability of supporting and rendering the appropriate aerodynamic loads. The design of the camber control has been also successful. The experimental activity comprised an extensive measurement campaign to measure the pressure distribution on the surface of the airfoil and of the total pressure in its wake. For each one of the two tested configuration, four flap geometries have been evaluated, for five and four angles of attack in the take-off and landing configuration, respectively. The force coefficient obtained as the integral of the measured pressures have also been computed. These tests clearly show that morphing the flap has a measurable effect on the aerodynamic performance. Moreover, they show that the optimal geometry for the take-off configuration is different from the optimal geometry for the landing configuration. This study shows that a significant advantage, of the order of 5–8%, can be gained by the morphing flap. The overall performances are summarised in the following Fig. 5.147. The displacement of the surface of the flap under loading has also been measured. This piece of information is important to give a feedback that can help the design and optimization of the flap structure for the final application. The measured displacements show that the region most affected by the deformation is the trailing edge, as expected. Displacements can also be used to improve the correlation between experiments and numerical simulations. The PIV surveys allowed us to gain additional insight in the phenomena taking place when the geometry of the flap is modified. The first observation is that an attached flow is always observed for the baseline geometry for the take-off configuration, while in this case the camber 1 geometry leads to a small separation region near the trailing edge which is most probably responsible of the drag increase. We infer that care must be taken in morphing the flap to avoid separation for the takeoff configuration. For the landing configuration, instead, quite extensive separation

268

F. Auteri et al.

• • • • •

Main results Lift coefficient has been significantly increased by increasing camber in both TO and L: quite high gains have been obtained; Drag has been reduced for the landing configuration and camber 1; Lift-to-drag has been increased in the landing configuration for camber 1; PIV results show the physics behind the results; Aeroelastic effects negligible for lift, have a measurable impact on drag. Main figures Lift coefficient increase: >7% (TO), >4% (L) depending on a (camber 3) Drag coefficient reduction: >8% (L) depending on (camber 1) Lift-to-drag increase: >8% (L) (camber 1)

Fig. 5.147 Final performances of the morphing high-lift flap: LS prototype: TO = Take-off, L = Landing. AoA 6°, speed 0–20–30 m/s

regions are observed for all the configurations. The separation, however, does not prevent higher camber geometries from increasing the lift, which is the design goal for this flight condition. Moreover, we observed that camber 1 is also able to reduce drag by reducing the extent of the separated region. A good agreement of the PIV with the numerical simulations confirm the above conclusions.

References 1. Politecnico di Milano, GVPM Wind Tunnel [Online]. Available: http://www.windtunnel.pol imi.it/ 2. G. Campanardi, G. Gibertini, M. Pozzi, M. Quici, La camera di prova aeronautica della galleria del vento del Politecnico di Milano, in Atti del XVII Congresso nazionale AIDAA, Roma (2003) 3. G. Gibertini, D. Grassi, A. Maggioni, A. Redaelli, L’utilizzo del modello in scala della galleria del vento del Politecnico di Milano, in Atti del XVIII Congresso nazionale AIDAA, Volterra (2005) 4. D. Fisher, J. Del Frate, D. Richwine, In-flight flow visualization characteristics of the NASAF18 high alpha research vehicle at high angles of attack, in NASA TM-41 (1990) 5. J. Hess, A. Smith, Calculation of potential flow about arbitrary bodies. Progr. Aerosp. Sci. 8, 1–138 (1967) 6. M. James, J. Barry, Mechanics of materials, in Stresses in Beams, 8th edn. Stamford, Chap. 5 7. Metaltech s.r.l. [Online]. Available: https://www.metaltech.it/ 8. Rank Organisation, Optical Alignment with the Rank Taylor Hobson Micro-alignment Telescope and Its Accessories (Rank Organisation, Rank Taylor Hobson, 1965) 9. A. Zanotti, M. Ermacora, G. Campanardi, G. Gibertini, Stereo particle image velocimetry measurements of perpendicular blade–vortex interaction over an oscillating airfoil. Exp. Fluids 55(9), 1–13 (2014). https://doi.org/10.1007/s00348-014-1811-8 10. B. Chanetz, Experimental Aerodynamics: An Introductory Guide (Springer Nature Switzerland, Cham, 2020) 11. Raffel, M., Willert, C., Scarano, F., Kähler, C., Wereley, S., Kompenhans, J.: Particle Image Velocimetry (Springer International Publishing, Berlin, 2018) 12. F. Auteri, L. Quartapelle, Fluidodinamica incomprimibile. Casa Editrice ambrosiana (2013) 13. E. Houghton, P. Carpenter, Aerodynamics for Engineering Students, 5th edn. pp. 522, 468–471

5 Aerodynamic Evaluation

269

14. A. Pope, J.B. Barlow, W.H. Rae, Low Speed Wind Tunnel Testing, 3rd edn. (Wiley, 1999), pp. 176–180 15. R. Pepper, C. van Dam, Design Methodology for Multi-Element High-Lift Systems on Subsonic Civil Transport Aircraft, NASA CR-202365, Davis, CA, USA (1996) 16. PIVTEC, PIVview 2C/3C, User Manual (2010) [Online]. Available: www.pivtec.com 17. A. Burner, T. Liu, Videogrammetric model deformation measurement technique. J. Aircr. 38(4), 745–754 (2001) 18. G.S. Qualisys, https://cdn-content.qualisys.com/2020/01/PI_Miqus.pdf [Online] 19. D. Eller, M. Carlsson, An efficient aerodynamic boundary element method for aeroelastic simulations and its experimental validation. Aerosp. Sci. Technol. 7(7), 532–539 (2003) 20. P.G. Drazin, Kelvin–Helmholtz instability of finite amplitude. J. Fluid Mech. 42(2), 321–335 (1970) 21. J.-B. Tô, T. Louge, C. Rouaix, C. Jimenez-Navarro, A. Marouf, M. Braza, Datapaper—aerodynamic Performance in Transonic Regime Around an A320 Airfoil by Means of Electroactive Morphing Through Vibration and Slight Deformation of the Near-Trailing Edge Region at High Reynolds Number [Online]. Available: https://doi.org/10.48531/JBRU.CALMIP/19XGRY 22. P. Alliez et al., Attributing and referencing (research) software: best practices and outlook form INRIA. Comput. Sci. Eng. 22(1), 39–52 (2001) 23. T. Berners-Lee, J. Hendler, O. Lassila, The semantic web. Sci. Am. 284(5), 39–52 (2001) 24. N. Shadbolt, T. Berners-Lee, W. Hall, The semantic web revisited. IEEE Intell. Syst. 21(3), 96–101 (2006) 25. M. Musen, The protégé project: a look back and a look forward. AI Matters 1(4), 4–12 (2015) 26. J. Malone et al., The software ontology (swo): a resource for reproducibility in biomedical data analysis, curation and digital preservation. J. Biomed. Seman. 5(1), 1–13 (2014) 27. T. Louge, M.-H. Karray, B. Archimède, J. Knodlseder, ASON: an Owl-s based ontology for astrophysical services. Astron. Comput. 24, 1–16 (2018) 28. J. Tô, Etude numérique et analyse physique du morphing de profils d’aile de type Airbus A3XX en régime transsonique par l’approche de modélisation de la turbulence, in “Organised Eddy Simulation” à nombre de Reynolds élevé, PhD Thesis, Toulouse, France (2021) 29. C. Jimenez-Navarro, C. Rouaix, A. Marouf, A. Ninet, Y. Hoarau, M. Braza, Numerical simulation of the aerodynamic performance of a morphing wing in the transonic regime, in The 8th European Congress on Computational Methods in Applied Sciences and Engineering, ECCOMAS, Oslo, Norway, 5–9 June 2022

Chapter 6

General Conclusions Marianna Braza

The SMS project demonstrated important innovations that open new ways for the design of disruptive aircraft’s wings. Before to summarise the major achievements, it is worthwhile mentioning that these have been the fruit of a stimulating, multidisciplinary collaboration among academic and industrial (SME) partners, steered also by Airbus (endorser) and advisor of the SMS project. These achievements have been derived thanks to a thorough physical investigation of the fluid-structure interaction based on a strong synergy among theory, simulation and experiments. Thanks to this effort, the morphing concepts “played” a lot with the manipulation of the surrounding turbulence vortices which in turn modify the fluid-structure behaviour towards optimal states. The significant outcomes have been derived by manipulating through Turbulence control around the rear wing’s region and its near wake, the turbulence vortex structures, thus enhancing the “beneficial” vortices contributing to high efficiency and attenuating through suitable breakdown the “harmful” ones (Figs. 6.1 and 6.2). This manipulation took benefit of the strong feedback effects permitting modification of the upstream pressure distributions by modifying the rear part and near-wake regions through suitable electroactive morphing devices, ensuring lighter, more versatile and less consuming in energy morphing concepts than the heavy hydromechanical actuators. Therefore, the SMS project opened new ways to consider the optimisation of the performances, regarding previous and ongoing studies rather attempting modification of the upstream wing’s part (for example, laminar wing design, etc. …). The SMS morphing concepts, based on these feedback effects have been partly bio-inspired from the actuation of the large wings, ailerons and feathers of largespan hunting birds. As them, the SMS project realised through efficient prototypes M. Braza (B) INPT—Toulouse Institut National Polytechnique, CNRS Centre National de Recherche Scientifique/IMFT—Institut de Mécanique des Fluides de Toulouse, UMR 5502, 2 Allée du Professeur Camille Soula, 31400 Toulouse, France e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. Braza et al. (eds.), Smart Morphing and Sensing for Aeronautical Configurations, Notes on Numerical Fluid Mechanics and Multidisciplinary Design 153, https://doi.org/10.1007/978-3-031-22580-2_6

271

272

M. Braza

Fig. 6.1 Bio-inspiration (left), vortex structure around the RS A320 prototype (middle) and vortex breakdown (right) thanks to the morphing in SMS

Fig. 6.2 Feedback effects from the downstream of the shock coherent vortices towards upstream than the shock-boundary layer interaction region: here illustration in cruise, tRS prototype

a multi-scale morphing in terms of length and time scales, as the situation characterising the Turbulence spectrum. Therefore, the SMS project achieved a “Smart wing design through Turbulence manipulation”. The bio-inspiration was only partial, because the high performances obtained, correspond to much higher speeds than faced from these birds, which never fly in transonic regimes during their cruise speed. Thanks to the strong multi-disciplinarity of the partnership, efficient actuators have been applied to realise this multiple-scale morphing, which goes far beyond the state of the art. The successful prototypes of the SMS project permitted the demonstration of quite significant benefits in the aerodynamic performances and this, by including near scale 1 design, the LS prototype. The synthesis of the obtained benefits is presented in the following paragraph.

6.1 Obtained Benefits and Conclusion Summary Disruptive design and construction of morphing wing prototypes attained performances beyond current limits and more conventional designs. • Increased performances in all flight phases: take-off, landing, cruise. • Aerodynamic performance increase (lift-to-drag ratio): order of 5–7% and higher in specific cases, realised thanks to feed-back control.

6 General Conclusions

273

• Lower angles of incidence associated with the morphing flap produce higher performance than conventional high-lift configurations in higher angles: a considerable benefit saving engine power in Take-off and Landing and this, in ~ scale 1. • Realisation of cambered shapes achieving simultaneous lift-to drag increase (6%) and drag reduction (5%) in landing thanks to optimal cambering shapes and to the hybrid morphing: a “first” in the state of the art and in ~ scale 1 where usually it is known that the cambering increases drag in other projects and studies. • Simultaneous noise sources decrease by an order of 10% and more. • Considerable drag decrease by 4% in cruise, leading to a considerable fuel’s consumption reduction and contributing to the greening of aircraft transport, thus contributing to the greening of aircraft’s transport. • Confirmation of the increased performances concerning the whole A320 aircraft by Hi-Fi numerical simulations in take-off and demonstration of clear benefits from the hybrid morphing, whose superior value has been demonstrated by the SMS project: 7% in lift increase, 1.3% in lift-to-drag increase. • Achievement of an economic electrical power supply systems for the morphing: order of a decade of W for the RS prototype and of 50 KW for the LS one, as well as ability of force bearing of order 1.5 t for the LS prototype. Moreover, the added weight on the wing’s structure is quite light comparing to other heavier systems based on MEMS or conventional hydromechanical actuators. Therefore, the disruptive morphing design studied in the SMS project enabled a considerable reduction of emissions, meeting the targets fixed by Flightpath 2050: European Commission, Directorate-General for Mobility and Transport, DirectorateGeneral for Research and Innovation, Flightpath 2050: Europe’s vision for aviation: maintaining global leadership and serving society’s needs, Publications Office, 2012, https://data.europa.eu/doi/10.2777/15458. The designs studied in SMS lead to “new horizons” in the morphing concepts towards multiple degrees of freedom multi-scale electroactive actuations, creating a “new material”, forming a powerful dynamic system composed of the solid and the fluid, optimally inter-operating in all flight phases. These elements open an unlimited way of “thinking” and realising the future wing design.

6.2 Impact: Communication and Dissemination of the Results The SMS project led to a wide communication and dissemination activity. It has been invited to animate speciific morphing stands and exhibitions at European and international level, including invitations by INEA (Innovation and Network Executgecy) of the European Commission. The non-exhaustive list is as follows:

274

M. Braza

• Organisation of the International Symposium IUTAM: www.smartwing.org/ iutam on 17–22 June 2018, Santorini with 3 SMS Sessions and participation of internationally-renown experts of the field: 10 SMS presentations. 95 participants. • SMS invited Session in the AIAA Aviation forum—Dallas, 17–21 June 2019 with 6 oral presentations: https://arc.aiaa.org/doi/book/10.2514/MAVIAT19. • Organisation of the 5th International Symposium FSSIC2019, «Flow Structure Sound interactions and Control», www.smartwing.org/FSSIC2019, on 27–30 Aug. 19, in Chania, Crete. 7 SMS presentations. 130 participants. • SMS invited session in ECCOMAS July 2020 with 6 oral presentations—postponed due to COVID19. • SMS invited session in the 10th EASN—Salerno, 2–5 September 2020, on “Innovation in Aviation & Space to the Satisfaction of the European Citizens”, https://easnconference.eu/SMS EASN sessions also in the 8th and 9th EASN. • Participation in the organisation of the 13th ETMM Congress, 23–25 Sept. 2020 http://etmm.ercoftac.org/etmm13/. • Invited Scientific and Technology Minip-symposium “Novel Wing Digitalized design” in ECCOMAS CM3–TRANSPORT 2021 held in Barcelona (https://con gress.cimne.com/cm3-2021/frontal/default.asp): “Numerical Study and Physical Analysis of Trailing Edge Electroactive Morphing on an A320 Type Morphing Wing in the Transonic Regime Including Wobulation Effects”. • Organisation and chairing of session: "Emerging Fields. Bio-inspiration & Morphing”, 56th 3AF International Conference on Applied Aerodynamics held in Toulouse in march 2022 (https://www.3af-aerodynamics.com): “Physical analysis on the transonic interaction of electroactive morphing concepts on an A320 type wing by numerical simulation at high Reynolds number”. • Invited Scientific sessions in ECCOMAS 2022 Congress held in OSLO (https:// www.eccomas2022.org/frontal/default.asp): – STS-05 “Shock Wave Boundary Layer Interactions in Aeronautical Applications”: “Numerical simulation and turbulence modelling of a 3D transonic regime around a supercritical wing involving strong separation”. – STS-06 “Disruptive aircraft’s wing configurations towards climate neutrality”: “Physical analysis on the transonic interaction of electroactive morphing concepts on an A320 type wing by numerical simulation at high Reynolds number”. Principal communication activities Participation in scientific and technical events organized by INEA: – – – – –

TRA—a new era for transport, Vienna, April 2018 with SMS stand ILA 25–30 May 2018 in Berlin: SMS stand and Conference by the coordinator AERODAYS–Bucharest, 27–30 May 2019 “Science is wonderful”, Brussels, 24–26 September 2019 European night of researchers 2018 and 2019 with SMS exhibition

6 General Conclusions

275

– SMS stand in the Rapacious Exhibition of the Museum of Natural History Toulouse, 11 October 2017–29 April 2018. All these activities are displayed in the public part of the SMS website, www.sma rtwing.org/SMS/EU.