Wolfgang Borutzky E ditor Bond Graphs for Modelling, Control and Fault Diagnosis of Engineering Systems Second Edition Bond Graphs for Modelling, Control and Fault Diagnosis of Engineering Systems Wolfgang Borutzky Editor Bond Graphs for Modelling, Control and Fault Diagnosis of Engineering Systems Second Edition 123 Editor WolfgangBorutzky Bonn-Rhein-SiegUniversityofApplied Sciences DepartmentofComputerScience SanktAugustin,Germany ISBN978-3-319-47433-5 ISBN978-3-319-47434-2 (eBook) DOI10.1007/978-3-319-47434-2 LibraryofCongressControlNumber:2016955827 ©SpringerInternationalPublishingSwitzerland2011,2017 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof thematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation, broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology nowknownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor theeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforany errorsoromissionsthatmayhavebeenmade. Printedonacid-freepaper ThisSpringerimprintispublishedbySpringerNature TheregisteredcompanyisSpringerInternationalPublishingAG Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland Foreword Prof. Borutzky has asked me to write a foreword to this second edition of the book Bond Graph Modelling of Engineering Systems. This put me in mind of the time when Ronald C. Rosenberg and I, as Assistant Professors at MIT, used to stop at a bar in Harvard Square to wait for the traffic to subside on our way homefromwork.(Ron’sdoctoralthesisresultedintheENPORTprogram,thefirst bondgraphprocessor,which,althoughrestrictedtolinearsystems,wouldsolvethe problems associated with differential causality in simulation.) We often discussed whattechnicalareasweshouldworkoninthefutureandwecametotheconclusion that bond graphs, recently invented by our advisor, Prof. Henry M. Paynter, were worthyofbeingbetterknowntothegeneralworldratherthanjusttoMITgraduate students. The result was our first book intended for self-study of bond graph methods, Analysis and Simulation of Multiport Systems: The Bond Graph Approach to Physical System Dynamics, The MIT Press, 1968. When the book was finished, Prof. Paynter asked us if he could provide a Foreword in the form of a Historical Note. We naturally agreed and the result is a fascinating two page history of Paynter’sthoughtstarting,hesays,inhighschool.Hispersonalreferencesgoback to the 1950s, but his references to chemical structure graphs go back another 100 years. HisfinalparagraphofhisHistoricalNoteisasfollows: ThusweredevisedonApril29,1959,thetwoideal3-portenergyjunctions(0,1)torender thesystemofbondgraphsacompleteandformaldiscipline.Thecompletesystemwasthen firstpublishedin1961. Asyoucansee,Prof.Paynterwasthinkingofhisreputationinthefuture. The publication mentioned was the MIT press book Analysis and Design of EngineeringSystems,1961,wasinfactapaperbackcompilationofclassnotestaken by a graduate student and, as Paynter admitted, the glued together book tended to fallapartasitspageswereturned.(Thiswasagreatdisadvantagetoslowlearners.) Thus, bond graphs remained for a while only known to those who had studied at MIT. v vi Foreword ProfessorPaynterwouldcertainlybegratifiedtoseetheinterestinbondgraphs thathasdevelopedinthepast60years.Thepresentvolumeisatestimonytolarge number of researchers who are presently contributing to extensions of bond graph techniquesandtonewapplicationareas.Healsomightbesurprisedtolearnthathis “complete and formal discipline” has been extended in directions he never would havethoughtpossible. Thereis,ofcourse,somedangerinextendingbondgraphtechniquestoinclude idiosyncratic notation and manipulations that apply to ever more special cases. In theearlyyears,thoseofusinterestedinpromotingbondgraphmethodshadtofight thosewholikedtheforce-currentanalogybetterthantheforce-voltageanalogysince they produced bond graphs with 0’s and 1’s and I’s and C’s interchanged, which provided a barrier to those newcomers interested in learning the benefits of bond graph methods. This volume is an example of the correct approach in which the authors take pains to make clear why special notation or special manipulations on a bond graph are necessary to provide insight into problems which don’t yield to standardbondgraphtechniques. WhenIcontemplatePaynter’scontributiontophysicalsystemdynamics,andof thoseofpresentday engineers whocontinue toadvance bond graph techniques as evidenced by the present volume, I think of the well-known quote from Goethe’s Faust: WasduererbtvondeinenVäternhast,erwirbes,umeszubesitzen. (Thatwhichyouinheritfromyourfathers,youmustearninordertopossess.) Theauthorsofthepresentvolumehaveclearlytakenthisadvicetoheart. Davis,CA,USA DeanKarnopp July2016 Preface By the beginning of 2011, Springer published a compilation text of a number of researchers from all over the world on Bond Graph Modelling of Engineering Systems. The book covers theoretical and methodological topics as well as some applicationsandtheuseofsoftwareforbondgraphmodellingandsimulation.The aimwastopresentthestateoftheartinbondgraphmodellingbyaddressinglatest developmentsandintegratingvariousworksinaunifiedmanner.Insummer2014, theeditorwasinvitedtoprepareasecondeditionofthe2011compilationtext.Due totheactiveongoingresearchofmycolleagues,theircommitment,theirexpertise, and their contributions the outcome of this new book project, in my view, has becomemorethanarevisionofthefirstedition.Thisnewbookagaincoverssome theoretical issues, applications and software support as the first edition. However, thefocusofthesecondeditionisonlatesttopicsaswellasonsubjectsthathaven’t been covered in the first edition. Moreover, new young excellent researchers have joinedtheteamtogetherwithco-authorswhohavebeenusingbondgraphintheir researchandinteachingforalongtime.Therefore,theresultofthisprojectmaybe consideredanovelbookthatreflectstheactiveresearchinthebondgraphmodelling communityandlatestachievementsoverarathershortperiodofthepastfiveyears. AnumberofthecontributedchaptershavearisenfromrecentlyfinishedPhDtheses. Bond graphs were devised by Professor H. Paynter some 60 years back at the Massachusetts Institute of Technology (MIT) in Cambridge, Massachusetts, USA, and were developed into a methodology by his former PhD students Professor D. Karnopp and Professor D. Margolis (University of California at Davis) and ProfessorR.Rosenberg(MichiganStateUniversity,EastLansing,Michigan).Since thenthebondgraphapproachtophysicalsystemmodellingandsimulationhasbeen adoptedbymanyresearchers.Thecontributionstothisbookwelldemonstratethat today, 60+ years later, the bond graph methodology has spread from MIT all over theworldandissuccessfullyusedinvariousengineeringfields. The book is organised into five parts. Each of them starts with an introduction thatmotivatesthesubjectsandestablishesrelationsbetweenthechaptersofapart. The contributed chapters show that bond graphs are more than just a method- ology for modelling and simulation of multidisciplinary systems described by vii viii Preface continuous time models. Beyond that application area, bond graphs can also be used to represent hybrid models with continuous time and discrete state variables. Furthermore,bondgraphmodellinghasprovenusefulincontrolandasonepossible starting point for model-based fault diagnosis and fault prognosis in engineering systems. These areas become more and more important with an increase of the complexityoftoday’smechatronicsystemsandatrendtoatleastpartlyautonomous systems. Representing hybrid models in a bond graph framework and bond graph model-basedfaultdiagnosisandfaultprognosisaretopicsinthefocusofthisnew book.Accordingly,thetitleofthefirsteditionhasbeenadapted. Another topic that is not new but of still ongoing interest is the question of howopenthermodynamicsubsystemswithcompressiblefluidflowandmechanical subsystems described by a distributed parameter model can be integrated in a lumpedparametermodelofacomplexoverallsystem.PartIIIpresentsanextension of common bond graphs for modelling thermodynamic subsystems introduced by ProfessorBrownthatcanbecombinedwithconventionalbondgraphsforotherparts of an engineering system and is supported by a mature modelling and simulation software environment newly developed over the past years. As to distributed parametersubsystems,anapproachbasedonfinitesegmentsandanactivitymetric is presented that aims at a lumped parameter model of reduced order sufficiently accurate. Part IV illustrates the remarkable wide range of applications in which bond graphs have been used for modelling mechatronic systems of current interest by contributed chapters ranging from a wheelchair, to robots for in vivo surgery, or walking machines that can be used for operations in hazardous fields, to wind turbinesintheareaofrenewableenergysources. ProfessorHenryPaynter,theinventorofbondgraphs,envisagedapplicationsof bondgraphmodellingtochemistry,electrochemistryandbiochemistryalreadyback in1992.Sincethennottoomanyresearchershaveusedbondgraphsformodelling chemical reactions. One chapter in Part IV extends bond graph modelling even to biomolecular systems of living organisms. As the latter ones are open systems that are not at thermodynamic equilibrium, bond graph representations of their biomolecular systems are not evident. The chapter introduces appropriate bond graph variables. Molecular species are represented by a nonlinear C element and reactionsbyanonlineartwo-portRelement.Asaresult,bondgraphmodellingasa graphicalapproachgivesmoreintuitiveinsightthanpurelynumericalapproaches. Astotheresultsofsimulationruns,ithasbecomecommonthatsomesoftware providesapostprocessingmodulethatenablesavisualisationofmechanicalmotion in 3D under different angles of views. The last chapter of this compilation text presentsanovelapproachthatlinksabondgraphmodellingandsimulationprogram to a program for the geometrical design of complex multibody systems such as robotswithseveraldegreesoffreedomsothatthe3Dmotionofarobotisvisualised simultaneouslywithsomelittledelaytoitsnumericalcomputation. Although this contributed book aims to reflect the current state of the art in bond graph modelling by presenting and discussing advanced recent topics, all chapters have been written in such a way that newcomers to the methodology Preface ix with some knowledge of the basics may easily get into the vast fascinating and open field of advanced bond graph modelling. Readers who may want to have a closer look at bond graph fundamentals may find references to latest monographs and textbooks. Furthermore, each chapter provides many references to conference papers, journal articles and PhD theses on advanced topics. This multiple authors bookwellcomplementslatestmonographsandtextbooksonbondgraphmodelling andmayserveasaguideforfurtherself-studyandasareference. Bond Graphs for Modelling, Control and Fault Diagnosis of Engineering SystemscontinuesthepresentationofitspredecessortitledBondGraphModelling ofEngineeringSystemsandlikethelatteroneaddressesreadersinacademiaaswell as practising engineers in industry and invites experts in related fields to consider the potential and the state of the art of bond graph modelling. It is hoped that the bondgraphmethodologymaycontributetomoreinsightintophysicalprocessesand tothedevelopmentofusefulmodelsintheirengineeringfield. Acknowledgements This volume would not have been possible without the commitment, the expertise and the efforts of my colleagues. I would like to take this opportunity to thank all my co-authors for their contributed chapters. Likewise, I wish to thank all colleagueswhokindlytookthetimetoreadthechapters.Theirvaluablecomments andsuggestionsaregratefullyacknowledged. My special thanks go to Professor Dean Karnopp, University of California at Davis, one of the leading pioneers of the bond graph methodology, who kindly honoured all authors of this compilation text by writing the foreword with some personalnotes. Also, I appreciate the support this multiple author book project received from Professors Yuri Merkuryev, Riga Technical University, Latvia; David Murray- Smith, University of Glasgow, Scotland, UK; Ronald Rosenberg, Michigan State University,EastLansing,MI,USAandGregorioRomero,UniversidadPolitécnica deMadrid,Spain. Last but not least, I wish to thank my Editor with Springer at New York, NY, USA, Mary James, for her invitation to this book project, for her kind constant support during this book project and Brinda Megasyamalan, Project Coordinator, SasiReka,ProjectManager,forthecooperationthroughoutthebookproduction. SanktAugustin,Germany WolfgangBorutzky July2016 Contents PartI BondGraphTheoryandMethodology 1 DecompositionofMultiports.............................................. 5 PeterC.Breedveld 2 AMethodforMinimizingtheSetofEquationsinBond GraphSystemswithCausalLoops ....................................... 27 JesusFelez 3 Controlled Switched Structures for Bond-Graph ModellingandSimulationofHybridSystems........................... 47 MatíasA.NacusseandSergioJ.Junco 4 DynamicCausalityinHybridBondGraphs ............................ 87 RebeccaMargettsandRogerF.Ngwompo PartII BondGraphModellingforFaultDiagnosis,Fault TolerantControl,andPrognosis 5 IntegratingBondGraph-BasedFaultDiagnosisandFault AccommodationThroughInverseSimulation........................... 139 W.Borutzky 6 Model-Based Diagnosis and Prognosis of Hybrid DynamicalSystemswithDynamicallyUpdatedParameters .......... 195 OmPrakashandA.K.Samantaray 7 Particle Filter Based Integrated Health Monitoring inBondGraphFramework ............................................... 233 MayankS.Jha,G.Dauphin-Tanguy,andB.Ould-Bouamama xi