George's Impact

MIT Symposium 2040 Visions of Process Systems Engineering George’s Impact Manfred Morari Electrical & Systems Engineering University of Pennsylvania June 1, 2017 George’s Formative Years 1965 1974 1977 MIT Symposium 2040 Visions of Process Systems Engineering George’s Impact Manfred Morari Electrical & Systems Engineering University of Pennsylvania June 1, 2017 Outline • My Time with George at U of MN • Past Attempts to Predict the Future • Predicting the Future Outline • My Time with George at U of MN • Past Attempts to Predict the Future • Predicting the Future My advisors at the U of Minnesota 1975-77 George Stephanopoulos Rutherford “Gus” Aris 1929-2005 Accompanying Note ~ 1995 Technology: Digital Control Computer IBM 1800 (introduced 1964) [wikipedia] Computer Control ~1968 Standard Oil of California Courtesy: Chevron Advanced Process Control Textbooks with Origins at U of Minnesota 1969 1981 Critique of (Process) Control JOURNAL CRITIQUE Critique of Chemical Process Control Theory 1973 JOURNAL CRITIQUE A. S. FOSS Critique of Chemical Process Control Theory Department of Chemical Engineering University of California, Berkeley, California 94720 A. S. FOSS 84 When it is stated, as it has been inDepartment more than of Chemicalone instance,EngineeringIEEE TRANSACTIONSare complex enough ON to AUTOMATICtax one’s ability CONTROL, to un- VOL. AC-20, NO. 1, FEBRUARY 1975 1976 recent publication, that thereUniversity is a wide of California, gap between Berkeley, the Californiatangle 94720 the relationships even at steady conditions (Shah theory of process control and its application, one is left and Stillman, 1970). ‘The inert level in the synthesis loop with the unmistakable impression that those who conceive may be regulated by manipulation of the purge rate, but When it is stated, as it has been in morethe than theory one are in some sense leagues ahead of those who since the purge constitutes a large fraction ok the reformer would use it. Thatinstance, the contraryare complex is the enough case is to the tax thesis one’s ability to un- recent publication, that there is a wide gap between the tangle the relationships even at steady conditionsfuel flow,(Shah one finds that such an adjustment influences Perspectivesof this essay. Indeed, the theory of chemical process the temperature and methane content of the reformer theory of process control and its application, one is left and Stillman, 1970). ‘The inert level in the synthesis loop control has some rugged terrain to traverse before it process effluent stream. Steam production in the waste with the unmistakable impression that those who conceive may be regulated by manipulation of the purge rate, but meets the needs of those who would apply it. heat boilers is thus upset and so too are the compressors the theory are in some sense leagues ahead of those who since the purge constitutes a large fraction ok the reformer would use it. That the contrary is the case is the thesis fuel flow, one finds that such anDesign adjustmentdriven Conceptsinfluences by that steam. for The compressorsProcess are also Control upset by the change in recycle rate in the synthesis loop. The of1976 this essay. Indeed, the theory of chemicalPROBLEMS process OF CHEMICALthe temperature PROCESS and CONTROL methane content of the reformer control has some rugged terrain to traverse before it process effluent stream. Steam production in resultantthe waste flow changes influence temperature rises in the 84 The needsIEEE TRANSACTIONS are intimately ON AUTOMATIC related CONTROL, to the VOL. problems, AC-20, NO. and1, FEBRUARY synthesis 1975 reactor, leading perhaps to higher temperatures meets the needs of those who would apply it. heat boilers is thus upset and so too are the compressorsA. Kestenbaum,’ R. Shinnar,*l and F. E. Thau2 Design Conceptsthe problems,for asProcessdriven usual, bywear that a steam.sometimes TheControl effective compressors camou- are also at upsetwhich by the side reactions that produce ether and flage. Superficiallythe thechange pioblem in recycle of chemical rate inprocesy the synthesiscon- methaneloop. The contribute significantly to the total heat libera- PROBLEMS OF CHEMICAL PROCESS CONTROL trol appears to involve the regulation of complex, often tion andDepadments to reduction of Chemical in methanol Engineering production. and Electrical Further, Engineering, The City College of The City University of New York, Perspectives resultant flow changes influence temperature rises inNew the York. New York 10031 The needs are intimately related to the problems,poorly understoodand synthesis physicochemical reactor, leading processes perhaps in theto higherface temperaturesthe upsets in the reformer lead to concentration distur- the problems, as usual, wear a sometimes effectiveSuperiorityof many camou- unknown at whichand uncharacterized the side reactions disturbances. ofthat produce TheTransfer bancesether andin the synthesis loop makeupFunction gas, which also Over flage. Superficially the pioblemA. Kestenbaum,’of chemical procesyword regulationcon- R. Shinnar,*lmethane has several contribute interpretations and significantly F. E. when Thau2 to appliedthe total heatdisturbs libera- the methanol production rate. trol appears to involve the regulation of complex,to industrial often processes,tion and butto reductionit usually inimplies methanol the desireproduction. NowFurther, add to these static interactions the influence of a 1975 to hold constant certain states or perhaps a time-average 20-minute thermal lag in the reformer, the composition poorly understood physicochemical processes in the face the upsets in the reformer lead to concentration Sixdistur- criteria are offered for the evaluation of the performance of process controllers. Classical techniques for of the states. Disturbances may consist of slow drifts such and thermal lags in the carbon dioxide absorber-stripper of many unknownSuperiority andState-Variable uncharacterizedDepadments disturbances. of of ChemicalTransfer The Engineeringbances in Functionthe and synthesis ElectricalMethods loop Engineering, makeupOver gas, which Thethe Cityalso design inCollege of PID ofLinear controllersThe City University and modern ofTime-Invariant control New York,theoretic techniques for controller design are reviewed. It is word regulation has several interpretations whenas thoseapplied originating disturbs in thediurnal methanol temperature production changes, rate. of system, the treacherous feed-effluent heat exchange in a New York. Newmonadnocks York 10031 illustrated by the explosive vaporization of synthesisshown reactor that modernthat can optimalexhibit wrong-waycontrol algorithms (nonmini- often do not lead to useful controller designs. They are useful to industrial processes, but it usually implies the desire Now add to these static interactions the influenceonly of asa tools in heuristic design methods. Their main problem is sensitivity to the nature of the process transfer to State-Variablehold constant certain states or perhaps Methods a time-averagea slug of water in 20-minutein theLinear feed thermalto a crude Time-Invariantlag still, in theor ofreformer, persistent the compositionmum phase) temperature effects, and the long-lived com- of the states. Disturbances may consist of slow randomdrifts such upsets suchand thermalas the step-like lags in fluctuationthe carbon in dioxide pressure absorber-stripper positionfunction transient and inthe the lack synthesis of conservative loop. Even design were methods.the The problems faced in developing better algorithms are Feedback Systemoutlined. Design as those originating in diurnalFeedback temperature changes,in utility System ofheaders. system, Sometimes the Design treacherous disturbances feed-effluent result from heat a exchangeprocess in well a understood, the dynamic cause-effect rela- monadnocks illustrated by the explosive vaporizationpurposeful of changesynthesis in the reactorlevel of thatoperation, can exhibit but with wrong-way the tionships(nonmini- would be difficult to untangle. a slug of water in the feed to a crude still, or ofexception persistent of batchmum operations,phase) temperature these changes effects, are and infre- the long-livedBut com-such processes are not completely understood. The ISAAC M. HOROWITZ, FELLOW, IEEE, AND URI SHAKED random upsets such as the step-like fluctuation quent.in pressure In additionposition toISAAC transientthe continuous in M.the regulationsynthesisHOROWITZ, loop.tasks, Even coke were FELLOW, level the on the IEEE, reformer AND catalyst URI is likely SHAKED unknown, in utility headers. SometimesSix disturbances criteria result arethere from offeredare athe tasksprocess for of start-upthewell understood,evaluation and shut-down. the dynamicof In1. theIntroductionall ofcause-effect performancedifferent rela- from furnace of process to furnace, andcontrollers. drifting; the firing Classical It is trivial techniques in the sense thatfor classical frequency response these activities, there is the dominating necessity for safe distribution of the burners in the reformer changes in an methods lead to satisfactory designs. Interestingly enough, purposefulAbstm-The change objectives in the and level achievements of operation, of state-variable but with methods the in tionships would be difficult to untangle.In recent years there has been a growing literature on the the design ofoperation PID controllersin thehis eventwork ofon andmalfunction controllability modern or failiueand control observability. of any theoreticundefinedState

Load more