Economic Model Predictive Control: Theory, Formulations and Chemical Process Applications PDF
Preview Economic Model Predictive Control: Theory, Formulations and Chemical Process Applications
Advances in Industrial Control Matthew Ellis Jinfeng Liu Panagiotis D. Christofides Economic Model Predictive Control Theory, Formulations and Chemical Process Applications Advances in Industrial Control Series editors Michael J. Grimble, Glasgow, UK Michael A. Johnson, Kidlington, UK More information about this series at http://www.springer.com/series/1412 Matthew Ellis Jinfeng Liu (cid:129) fi Panagiotis D. Christo des Economic Model Predictive Control Theory, Formulations and Chemical Process Applications 123 MatthewEllis Panagiotis D.Christofides Department ofChemical andBiomolecular Department ofChemical andBiomolecular Engineering Engineering University of California, LosAngeles University of California, LosAngeles LosAngeles, CA LosAngeles, CA USA USA Jinfeng Liu Department ofChemical andMaterials Engineering University of Alberta Edmonton, AB Canada ISSN 1430-9491 ISSN 2193-1577 (electronic) Advances in IndustrialControl ISBN978-3-319-41107-1 ISBN978-3-319-41108-8 (eBook) DOI 10.1007/978-3-319-41108-8 LibraryofCongressControlNumber:2016944902 ©SpringerInternationalPublishingSwitzerland2017 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpart 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 orinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar methodologynowknownorhereafterdeveloped. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfrom therelevantprotectivelawsandregulationsandthereforefreeforgeneraluse. 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 authorsortheeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinor foranyerrorsoromissionsthatmayhavebeenmade. Printedonacid-freepaper ThisSpringerimprintispublishedbySpringerNature TheregisteredcompanyisSpringerInternationalPublishingAGSwitzerland ’ Series Editors Foreword TheseriesAdvancesinIndustrialControlaimstoreportandencouragetechnology transferincontrolengineering.Therapiddevelopmentofcontroltechnologyhasan impactonallareasofthecontroldiscipline.Newtheory,newcontrollers,actuators, sensors,newindustrialprocesses,computermethods,newapplications,newdesign philosophies…, new challenges. Much of this development work resides in industrial reports, feasibility study papers, and reports of advanced collaborative projects. The series offers an opportunity for researchers to present an extended expositionofsuchnewworkinallaspectsofindustrialcontrolforwiderandrapid dissemination. The model predictive control (MPC) design philosophy may be used for a low-level loop controller, replacing regulating loop control systems based on the proportionalintegraldifferential(PID)controllers.Italsohastheflexibilitytowork atahigherlevelasatrackingcontrolsupervisortofollowareferencetrajectoryby providing the reference inputs to low-level loop controllers, typically of the PID variety.ThesetypesofindustrialformulationsforMPCusuallyinvolveaquadratic cost functional, a (linear) dynamic process model, and a set of physical system constraintsoninputsandoutputs.ThisclassofMPCproblemshastheadvantageof beingreadilynumericallysolvableandimplementedinrealtime.Clearlytheseloop controller and reference-tracking uses of MPC fit very nicely with the lower level/supervisory-level architectures of the traditional process control hierarchy based on process function and sampling-time-separation arguments. Sincethisparadigmhasmetwithconsiderablesuccessintheprocessindustries, the obvious question to ask is where next with industrial developments of MPC. The academic control community has certainly been busy extending the theory of MPC into the field of nonlinear predictive control. Indeed, our sister series v vi SeriesEditors’Foreword Advanced Textbooks in Control and Signal Processing was recently fortunate to publish Model Predictive Control by Basil Kouvaritakis and Mark Cannon (ISBN 978-1-319-24851-6, 2016) in this growing field of nonlinear MPC. However,analternativeandpragmaticwayforwardistofollowtheexampleof researchers Matthew Ellis, Jinfeng Liu, and Panagiotis D. Christofides and look again at the process trends and requirements of industry. These researchers are mainly involved with the chemical process industries and what they find is an increasing focus on “dynamic market-driven operations which include more effi- cient and nimble operations”. Their solution is a new re-formulation and re-interpretation of the MPC method termed economic model predictive control (EMPC) that is reported in this Advances in Industrial Control monograph, EconomicModelPredictiveControl:Theory,Formulations andChemicalProcess Applications. Their aim is to exploit the ever-increasing power of computing technology to enhancethetraditionalcontrolhierarchybyextendingandre-interpretingtheMPC methodinseveralrespects.Thisapproachmaybeappliedtoanupper-levelcontrol more concerned with management functions and scheduling, or to an intermediate level involving the multivariable loop controls. The cost functional is selected to capture some of the economic objectives of the process. The dynamic process modelisextendedtorepresentbotheconomicandphysicalvariablesintheprocess. Thephysicalprocessconstraintsetremainslargelyunchangedininterpretationand formulation, but a new set of “economic” process constraints is appended to the MPC problem description. The resulting formulation involves nonlinear mathe- matical representations and constrained nonlinear system optimization. The researchers put their ideas and solutions to a test with a set of applications from the chemical process industries and there are three challenges in this work: i. how to capture mathematically the “economic dimension” of a process in the construction of the cost functional, the process model, and the constraints; ii. how to provide an EMPC control theory to guarantee essential control prop- erties such as closed-loop stability; and iii. how to develop the numerical computational algorithms that will allow the application of the desired control actions in real-time operation. The EMPC method is a challenge for control theory analysis, and the resulting nonlinear optimization problems are difficult to solve and implement. The mono- graph presents new results in these areas that are original to the authors and they demonstrate their results with detailed chemical process simulation examples. The success of MPC in the process industries has been largely due to the economic benefits provided using linear system and quadratic cost problems. However, SeriesEditors’Foreword vii further improvements will require more accurate nonlinear plant models and cost measures whichthis textexplores.The Advances inIndustrialControlmonograph series was originally created for the promotion of new methods for industrial applications. The level of originality and the new research results presented in this monographmeetthisseries’aimandmakeanexcellentcontributiontoAdvancesin Industrial Control. Michael J. Grimble Michael A. Johnson Industrial Control Centre University of Strathclyde Glasgow, Scotland, UK Preface Traditionally, economic optimization and control of chemical processes have been addressed with a hierarchical approach. In the upper layer, a static economic optimization problem is solved to compute an optimal process steady state. The optimal steady state is sent down to the lower feedback control layer to force the process to operate at the optimal steady state. In the context of the lower feedback control layer, model predictive control (MPC) has become a ubiquitous advanced control methodology used in the chemical process industry owing to its ability to control multiple input, multiple output process/systems while accounting for con- straints and performance criteria. Recent pressure to make chemical processes operatemoreefficiently,costeffectively,andreliablyhasmotivatedprocesscontrol researchers to analyze a more general MPC framework that merges economic optimization with process control. In particular, economic MPC (EMPC), which incorporates an economically motivated stage cost function in its formulation, has attracted significant attention and research over the last 10 years. The rigorous designofEMPCsystemsthatoperateprocessesinaneconomicallyoptimalfashion while maintaining stability of the closed-loop system is challenging as traditional notions of stability may not apply to the closed-loop system under EMPC. ThisbookcoversseveralrigorousmethodsforthedesignofEMPCsystemsfor chemical processes, which are typically described by nonlinear dynamic models. The book opens with a brief introduction and motivation of EMPC and a back- groundonnonlinearsystems,controlandoptimization.Anoverviewofthevarious EMPC methods proposed in the literature is provided. Subsequently, an EMPC scheme designed via Lyapunov-based techniques, which is the main focus of this book, is described in detail with rigorous analysis provided on its feasibility, closed-loop stability and performance properties. Next, the design of state-estimation-based EMPC schemes is considered for nonlinear systems. Then, several two-layer EMPC frameworks are presented that address computational efficiencyandindustriallyrelevantcontroldesigns.Thebookcloseswithadditional EMPCdesignsthataddresscomputationalefficiencyandreal-timeimplementation. ix x Preface Throughoutthebook,theEMPCmethodsareappliedtochemicalprocessexamples to demonstrate their effectiveness and performance. The book requires some knowledge of nonlinear systems and nonlinear control theory. Because EMPC requires the repeated solution of a nonlinear optimization problem,abasicknowledgeofnonlinearoptimization/programmingmaybehelpful in understanding the concepts. This book is intended for researchers, graduate students, and process control engineers. WewouldliketoacknowledgeDr.MohsenHeidarinejad,Dr.XianzhongChen, Dr. Liangfeng Lao, Helen Durand, Tim Anderson, Dawson Tu, and Anas Alanqar all at UCLA who have contributed substantially to the research efforts and results included in this book. We would like to thank them for their hard work and contributions. We would also like to thank our many other collaborators and col- leagues who contributed in some way to this project. Inparticular,wewouldliketothankourcolleaguesatUCLAandtheUniversity ofAlberta,andtheUnitedStatesNationalScienceFoundationandtheDepartment of Energy for financial support. Finally, we would like to express our deepest gratitudetoourfamiliesfortheirdedication,encouragement,patience,andsupport over the course of this project. We dedicate this book to them. Los Angeles, CA, USA Matthew Ellis Edmonton, AB, Canada Jinfeng Liu Los Angeles, CA, USA Panagiotis D. Christofides
The list of books you might like
