COMPUTING AND SYSTEMS TECHNOLOGY DIVISION

VOLUME 7 No.2

NEWSLETTER

COMPUTING AND SYSTEMS

TECHNOLOGY DIVISION

SEVENTY-FIVE YEARS OF PROGRESS

AMERICAN INSTITUTE

OF CHEMICAL ENGINEERS

DECEMBER 1984

TABLE OF CONTENTS

Publication Board's Messages 1

CAST Awards ' ...............•......•................... 2

Nomination Form for CAST Awards 3

Programming Board Report 5

Houston Meeting Program Information 6

Chicago Meeting Program Information 7

Reflections on a Visit to JapanProf. Peter Rony, 8

History of Computersfrom a Chemical Engineering Point of View

Dr. Edward Gordon 10

The Great Plains ASPEN Model ProjectDr. I.H. Rinard '" 17

ASPEN User Group 19

Ordering Microfiche 20

Process Systems Engineering Conference 22

FOCAPD Proceedings Available 24

San Francisco Meeting CAST Related Papers 25

PUBLICATION BOARD'S MESSAGES

by Dr. Edward Gordon

Fluor Engineers Inc.(714) 975-3531

One disadvantage of the increased size ofthe CAST Newsletter is the additionaleffort it takes to prepare the material forpublication. Since the effort is performedalmost entirely by volunteers, there is agreater opportunity for unplanned delayswhen crises arise in the jobs we are beingpaid for. As a consequence, the Spring1984 issue was mailed just before theAnaheim National Meeting in May. This Fallissue has not done as well and it is beingsent out after the San Francisco Meeting.One small benefit is that the list ofMicrofiche for that meeting is included.

Starting January 1, 1985, Professor PeterRony will take over responsibi1i~y for theNewsletter and the CAST Publications Board.Peter has been the editor for the IEEEComputer Society publication called Microfor the past two years. The next item inthis Newsletter is a message for the CASTmembership from your new edit0r.

In this issue, there is a History of compu­ters from a Chemical Engineering point ofview. This summary is intended as back­ground for our younger members to help themunderstand the enormous changes which haveoccurred in the past forty years as Compu­ter-Aided Chemical Engineering has become areality. In the previous issue, Ted Lein­inger covered the history of the MachineComputation Committee which spawned CAST.

Your new editor has provided some reflec­tions on a recent visit to Japan. IrvRinard, the Area lOb Chairman, covers whatprobably is the largest effort to date inapplying the ASPEN program.

1

by Prof. Peter Rony, VirginiaPolytechnic Institute

The CAST Executive Committee elected me asChairman of the Publications Board for thetwo year period, January 1985 -- December1986. I am honored by this appointment,and will try to do my best to sustain thehard work, momentum, and imaginationprovided by Ed Gordon during his term inoffice.

To those of you who do not know me, myqualifications include: (a) I have ~een

Editor-in-Chief of IEEE MICRO, a publlca­tion (24,000 subscribers) of the IEEEComputer Society, during the period January1983 -- December 1985, (b) I have benefit­ted from my association with competent IEEEComputer Society personnel in the publica­tions area, among whom include True Seaborn,IEEE MICRO; (c) I believe that the editorialprocess is fundamental for a professionalorganization; and (d) I had the foolishnessto be vocal about my concerns for the AIChEflagship magazine, CEP, which I felt hasneither presented the totality of our pro­fession effectively nor been competitivewith other flagship magazines such as, forexample, IEEE Spectrum and IEEE CS Computer.It is of interest to note that Computermagazine started as a newsletter, much likethe CAST Newsletter, about fifteen yearsago.

Ed has been publishing and mailing theNewsletter from his office in Irvine,California. With the change of the AIChEstaff in New York, the CAST ExecutiveCommittee desires to have the AIChE head­quarters participate more strongly in thepublication of the magazine. CEP has a newpublisher, Diane Foster, and a new editor,Agnes K. Dubberly, who have exciting plansfor the magazine. I look forward to thecollaboration of the CAST Division withthem.

(CONTINUED ON PAGE 24).

1984 CAST AWARDS

The Computing in Chemical Engineering Awardis given in recognition of outstandingcontributions in the application of comput­ing and system technology to chemicalengineering. G.V. (Rex) Reklaitis receivedhis B.S. degree from the Illinois Instituteof Technology and his Ph.D. degree fromStanford University. Rex has spent a yearin Switzerland as a NSF post-doctoralfellow and a semester in Lithuania SSR asSenior Fulbright lecturer. He has taughtat Purdue University for 12 years. He isthe author/co-author of 2 books, co-editorof an ACS Symposium Series Volume, and ofan AIChE Symposium Series Volume, all inthe areas of computer-aided design, optimi­zation, and flowsheeting. He is the authorof a number of research papers in thesefields, focusing most recently on computeraids for batch/semi-continuous processdesign and analysis as well as on applica-tions of computer graphics. He iscurrently Vice-President of CACHECorporation and has been a Director of theCAST Division.

The Ted Peterson Award is given for a paperco-authored by a student. Chau-Chyun Chenis Principal Engineer at Aspen Technology.His areas of specialty are applied thermo­dynamics and computer-aided process simula­tion. He has published and presented manytechnical papers in the fields of enzymetechnology, electrolyte thermodynamics, andprocess simulation. He holds a B.S. degreein chemistry from National Taiwan Univer­sity and M.S. and Sc.D. degrees in ChemicalEngineering from the Massachusetts Insti­tute of Technology. He is a member ofAIChE, ACS and Chinese Institute of Engine­ers, U.S.A.

Both awards were presented at the CASTAwards banquet at the AIChE meeting in SanFrancisco. Professor Reklaitis gave a talkentitled, "Prospects for Computer-AidedBatch Process Engineering".

2

CAST AWARDS SOLICITATION OF NOMINATION

Please use the form on the next two pagesto submit your nonimination to Jeff Siirolaby March 31, 1985. Use a separate copy ofthe form for each nomination.

Computing in Chemical Engineering Award

This award is given to recognizeoutstanding contributions in theapplication of computing and systemstechnology to Chemical Engineering. It isnormally awarded annually and consists of aplaque and a check for $1500. Funding forthe three years (1982-84) has been providedby Simulation Sciences of Fullerton, Calif­ornia, and Intergraph Corporation, Hunts­ville, Alabama. The 1983 Awardee wasArthur W. Westerberg of Carnagie-Mellon.The 1982 Awardee was Lawrence B. Evans ofASPEN Tech. formerly Professor of Chemical~ngineering at MIT and leader of the ASPENProject at MIT. The 1981 Awardee wasRichard S.H. Mah, Professor of ChemicalEngineering at Northwestern University.The 1980 Awardee was Brice Carnahan, at theUniversity of Michigan. The 1979 Awardee,Richard R. Hughes at the University ofWisconsin, was the first recipient of theaward.

TED PETERSON STUDENT PAPER AWARD

This award is given to an individual forpublished work in the application ofcomputing and systems technology toChemical Engineering. The work must havebeen done by the individual while pursuinggraduate or undergraduate studies inChemical Engineering. The award consistsof $500 and a plaque and is normallyawarded annually. This is a new award andthe first award was made in 1983 at theDiamond Jubilee Meeting in Washington, D.C.It is currently being supported by IBM andChemShare, Inc. The 1983 winner was CarlosE. Garcia of Shell Development in Houston.He was a student at University of Wisconsin- Madison.

AMERICAN INSTITUTE OF CHEMICAL ENGINEERS

1985 AWARD NOMINATION FORM·

1. Name of the Award

2. Name of Nominee

___________________Today's Datei _

__________________ Date of Birth _

3. Present Position (exact title) _

Address ---,--cc:-c:-----:;,----,.--------=--c-=:-:----------:;;:---Institution or Company City and State Zip

4, Education:

Institution

5. Positions Held:

Company or Institution

Degree Received Year Received

Position or Title

Field

Dates

6. Academic and Professional Honors (include awards, memberships in honorary societies and fraternities,prizes) and date the honor was received,

7. Technical and Professional Society Memberships and Offices

8. Sponsor's Name and Address

Sponsor's Signature

• A person may be nominated for only one award in a given year.

.,3

B. CITATION

1. A brief statement, not to exceed 250 words, of why the candidate should receive this award. (Use separatesheet of paper.)

2. Proposed citation (not more than 25 carefully edited words that reflect specific accomplishments).

C. QUALIFICATIONS

Each award has a different set of qualifications. These are described in the awards brochure. After readingthem, please fill in the following information on the nominee where appropriate. Use a separate sheetforeachitem if necessary.

1. Selected bibliography (include books, patents, and major papers pUblished.)

2. Specific identification and evaluation of the accomplishments on which the nomination is based.

3. If the nominee has previously received any award from AIChE or one of its Divisions, an explicit statementof new accomplishments or work over and above those cited for the earlier award(s) must be included:

4. Other pertinent information.

D. SUPPORTING LETTERS AND DOCUMENTS

List of no more than five individuals whose letters are attached.

Name

1.

2.

3.

4.

5.

Please send the completed form and supplemental sheets byMarch 15, 1985 to the CAST Division Awards Chairman, Dr. JeffSiirola, Tennesse Eastman Company, Kingsport, TN 37664

4

Affiliation

range of technological areas. A reading ofReischauerls "Japan The Story of a Nation t1

indicated that such a feeling has not beentypical in post-war Japan. But it existstoday, and it means that we should be awareof and sobered by, the competitive pressurethat Japan will exert in many "high tech"areas ranging from megabit semiconductormemories to nuclear reactors. As a conse­quence of my visit, I developed a greaterappreciation of and respect for what theyhave accomplished during the last twentyyears, and what they plan to accomplish inthe next twenty. The United States hasevery reason to be worried about theJapanese industrial challenge to itsleadership.

At NEC, through the kindness of Mr. KiyoshiEmi, I was able to tour their model officebuilding (in Abiko) that exhibits state-of­the-art electronic office automation tech­nology. Thin conductors under rugs. High­speed fiber optic information buses in thewalls. Mechanical conveyors for documentsthroughout the building. A teleconferenc­ing room that permits exchange of visualimages (from electronic cameras, electronictablets, and electronic blackboards) fromAbiko to New York via satellite. Mostimpressive ..

At Mitsubishi, I had the opportunity toobserve development work on low-costrobots. One of the engineers was out ofschool only a year, and the other, at adifferent site, had about ten or more yearsof experience. What impressed me was thatsuch engineers were responsible for theentire design of their respective robots:mechanical, microcomputer, interfaceelectronics, and robot operating systemsoftware. The Japanese had a word for suchan endeavor: mechatronics. There wasstrong product orientation, and a singleindividual was identified with a singleproduct. I am not an industrial or manu­facturing engineer, but my sense of Ameri­can education is that we do not instillproduct orientation in our students as do

9

the Japanese. Also, the mechanical skillsof undergraduate engineers are weaker thanthey were 30 years ago; many, if not most,U. S. engineering students no longer haveexperience in machine shop. The Japanesewhom I met were very product oriented:define it, develop it, and market it. Fora new engineer fresh from a university, asuccessful product would have annual salesof, for example, $1,000,000 to $2,000,000and would have a lifetime of at leastseveral years ..

A final topic is the matter of sales ofJapanese personal computers in the UnitedStates. I frequently asked representativesof the companies I visited, How are yourpersonal computer sales doing in the U.S.?The answer: To date, such sales have notmet expectations for reasons which they didnot completely understand. The weakAmerican reaction to their computers, whichthey considered to be competitive with theIBM PC and its clones, has come as asurprise--an exception, perhaps temporary,to the Japanese success stories in cameras,video tape recorders, watches, automobiles,and portable radios. When the questioningwas reversed and I was asked if wouldpurchase a Japanese computer, my answer wasno unless the PC was very inexpensive ornovel. When asked to elaborate, Iexplained that I had no confidence thatstandard third party hardware and softwarewould be available for a Japanese PC. Idid not want to become stuck with a machinethat was poorly supported by the U.S.computer industry.

I would like to thank my many Japanesehosts, who, in addition to the individualsmentioned above, included Mr. OsamuTakahashi, Mr. Tadashi Kurachi, Mr. ShojiHiroe, Mr. Tohru Haegawa, Mr. TeiichiTanaka, Mr. Fumihiro Nakamura, Mr. AkiraNaito, Mr. Toyoaki Taniguchi, and others.Our conversations enriched my appreciationof your country.

HISTORY OF COMPUTERS

from a Chemical Engineering Point of View

by Dr. Edward Gordon Fluor Engineers Inc.

In the previous issue of the CASTNewsletter, Ted Leininger presented anexcellent history of the A. 1. Ch. E MachineComputation Committee (MCC) which was thepredecessor to CAST (1) • That paperpresents a review of the dramatic changesin the role of computers in ChemicalEngineering over the past forty years.That review provides some background whichexplains the timing of many of the MCCevents.

WHERE WERE WE FORTY YEARS AGO?

Forty years ago, there Were very fewdigital computers in the world. The sliderule was the most commonly used aid toChemical Engineering calculations. Therewere a few calculations like tray-to-traydistillation calculations which requiredmore precision than a slide rule couldprovide (2). For those calculations, amechanical desk calculator which typicallycould handle 8 or 10 digit numbers was usedin 1944. Those calculators were limited tothe same four functions (addition, subtrac­tion, multiplication, and division) (3)that an inexpensive pocket calculatorperforms today with much less noise andeffort.

There were a number of mechanical analogcomputers available. They had been usedfor solving simultaneous equations back inthe twenties. Network flow was simulatedelectrically as far back as 1925. Theirusefulness in solving differential equa­tions was known in 1931. Bell Labs startedwork on electronic analog computers in themid-thirties (4).

10

The Sorter for punched cards was inventedby Hollerith for processing of the datafrom the 1890 U. S. Census. It providedsome very effective tabulating capabilitieswhich made it possible to complete process­ing of the data before the next census (5).Various printing and logical capabilitieswere added as those devices evolved intoAccounting Machines. In the late thirties,IBM introduced their Model 601 Calculatorwhich did mUltiplications much like a deskcalculator.

Before that, multiplications were performedby repeated additions without a convenientmechanism for shifting to the next decimaldigit. The IBM Model 602 which came alonga few years later was also capable ofperforming divisions.

Accounting Machines were computationaldevices which performed relatively simplemanipulations of the numbers or charactersthey read from punched cards. The resultscould be punched into the same card, intoanother card, or printed (6) . Thosemanipulations were much like today'ssubroutines and in many ways AccountingMachines behaved like computers. The usualAccounting Machine had very limited datastorage within the machine, typically muchless than 100 digits. A few unusualinstallations had rather substantialcapabilities for their time.

The best known of the sophisticated relaybased machines is the monster (AutomaticSequence Controlled Calculator) which IBMbuilt for use at Harvard University start­ing in 1938 (7). Zuse built relay basedcomputers for the Germans which receivedmuch less attention (8). Perhaps the mostimportant military application was theBritish work on Cryptography (9).

The first electronic digital computingdevice was initiated in 1938 by Atanasoffand Berry for the solution of up to 30simultaneous linear equations (10). Forall of these machines, programming was doneby plugging wires into a Control Panel(11). Thus, it was far more difficult toprogram then than with any computer in usetoday. I don't know of any publishedaccounts of applications of computers inChemical Engineering before 1944. However,the situation changed drastically very soonthereafter.

ACCOUNTING MACHINES 1944 - 1954

Although major advancements were made incomputer development (12), they had verylittle impact on Chemical Engineering inthat decade. The major advances in Chemi­cal Engineering computations came fromapplications of Accounting Machines. Thosemachines were relatively common particular­ly in the larger companies. Since mostaccounting operations are cyclical, therewere times in the various accounting cycleswhen the machines were idle.

A number of Chemical Engineers became awareof the availability of those machines on anon-interference basis and managed topenetrate the language barrier. In addi­tion, the capabilities of the IBM Calcula­tors were substantially improved. TheModel 602 was replaced by an advancedversion, the Model 602A and before long thearithmetic was done using vacuum tubes, bythe IBM 603 and 604. Furthermore, theseapplications started to attract attentionand IBM sponsored a number of symposia(13) •

A 1950 article by Rose and Williams at PennState describes how they did distillationcalculation using an IBM 604 (14). A yearlater Opler and Heitz described their doinga six component system distillation on anIBM 602A (15). However, distillation cal­culations were not the only ones requiringmore precision than a slide rule could pro­vide. Data analysis via regression alsorequired greater precision.

11

In 1949, I became aware that the facility Iwas working in had an IBM 601 and I wasspending most of my time doing data analy­sis. After much difficulty, I was finallyable to get our Accounting Machine expertto wire a Control Panel which I could useto generate the crossproducts needed forregression. The 601 took two three digitnumbers which it read from the punched cardand punched their product in to the samecard. These products were then summed by a402 Tabulator to generate the matrix ofcoefficients representing the normal equa­tions. I was fortunate that we had a MassSpectrometry group who spent most of theirtime solving sets of simultaneous equationsto generate useful analyses from themeasurements. They could solve my set ofequations in about 20 minutes. Sinceneither the 601 nor the 402 could divide,it was impractical to try to get either oneto solve the equations. The key point is

\ that inspite of the many frustrations in­volved and the many steps it took to gener­ate the desired results. in six months Iwas able to come up with more and bettercorrelations than I had in the previousfour years using graphical data analysismethods.

As scientific applications of AccountingMachines became more common, IBM thendecided that they should get into comput­ers. They also made available a datastorage box for use with the IBM 604 Calcu­lator and a 402 Tabulator. They calledthat configuration a Card Programmed Calcu­lator (CPC) and it was very popular in theearly fifties (16). Those boxes had astorage capacity of fifty 20 digit numbersfor a total of one thousand digits, a majorstep forward.

There was relatively free exchange of ideasand techniques and Prof. Rose and associ­ates started a series of review articles onComputers, Statistics, and Mathematics(17) •

I' '.

MOVING UP TO COMPUTERS 1954 - 1964

Although the first Univac Computer wasdelivered to the Census Bureau in 1951(18), digital computers not based onaccounting machines did not have muchimpact on Chemical Engineering until theywere more common. Relatively few were soldand they were used primarily for businesstype applications. The more importantearly computers in Chemical Engineering,were the ElectroData (Burroughs) 205, IBM704 and IBM 650 which Were first deliveredin 1954, 1955, and 1956 respectively.

Early computers were classified as large,medium and small based on price. Onehundred thousand dollars was the lowerlimit on medium and 750 thousand the upperlimit (19). The IBM 704 was in the largeclass so its Chemical Engineering applica­tion came largely from the Oil and ChemicalCompanies. The Engineering Companiesfavored the 205 over the 650 even thoughboth were in the medium price range. The205 had 4080 ten digit words of data andprogram storage on its drum. That corre­sponds to 20. 4K bytes which is small bytoday's standards even for a personalcomputer. The 650 was even smaller when itwas introduced though later models did have4000 words of storage.

When these computers were first delivered,they did not have any assemblers or com­pilers. Effectively, programming was donein absolute machine language. That was nottoo difficult for the 205 because it was adecimal machine and it used a singleaddress per command structure. In duetime, programming aids were developed (20).

When Fortran was first described at acomputing conference, it was described as alanguage developed for those mathematicianswho were unwilling to learn machine langu­age so they could write their own one-shotprograms for the IBM 704 without botheringthe programmers (21).

12

Fortran had very limited usage on the firstgeneration medium priced computers,although it made quick inroads on thelarger IBM machines. Univac had their owncompiler. Ultimately, versions of Fortranwere developed which could be used effici­ently on medium priced machines (22).However, by then the first generationmachines were largely obsolete. Tubemachines required much more maintenancethan the transistor machines of the secondgeneration.

Computer Graphics were largely limited toCalComp plots of graphic output orTektronix CRT displays. A few were usinglight pens for graphic input. The techno­logy goes back to the Whirlwind Project atMIT which was started in 1947 (23). Onearea which received much attention was AirTraffic Control and the development activi­ty was applied in the SAGE system. Thelatter used a monstrous array of firstger.eration computer hardware in a projectwhich required reasonably good reliability(24) . The modern approach to interactivegraphics was under development with theSKETCHPAD project at MIT being the bestknown effort (25).

The most spectacular computer developmenteffort in this time interval was theStretch Computer Project. Its goal when itstarted in 1956 was one hundred times theperformance of an IBM 704. It was soonfound that a ten fold improvement in basiccircuit speed was the best they couldachieve with the technology of that timeand a six-fold improvement in memoryperformance. Thus, they didn't quite maketheir goal except for double precisionfloating point operations and mesh typecalculations where their look ahead fea­tures were quite effective (26).

Although it was started in the firstcomputer generation, it was completed afterthe second generation was on the scene andit carried a second generation modelnumber, 7030. Because it didn't quite makeits performance goal, its selling price waswell below IBM's cost. That led to astatement attributed to Tom Watson that ifthey sold enough of that computer, thelosses could be sufficient to drive IBM outof business. The design advances wereadapted to raise the performance of the7094 model over its predecessor, the 7090,enough to offer a more favorable cost-per­formance ratio (27) and many customersswitched to the less expensive 7094.

A more practical supercomputer was the CDC6600 which was started in 1960. Theirfirst machine was delivered in 1964 (28).With faster circuitry than Strech used andgreater emphasis on number crunching, itwas a commercial success. CDC and theirChief Systems Designer, Seymour Crayquickly gained undisputed leadership insupercomputers (29).

DEVELOPING MAJOR SYSTEMS 1964- 1974

Certainly the most popular computer in thethird generation is the IBM 360 series,although for a short while there weredoubts. The 360 series was a completebreak from all members of the IBM 7000series second generation machines. The IBM7000 scientific machines used a 36 bit wordwhich was also interpreted at six 6-bitcharacters (30). That lead to the IBMseven track tapes.

In contrast, the IBM 360 series was basedon four 8-bit bytes each of which canrepresent a character (31). Thus, the 360machines used nine track tapes. Thisessentially complete incompatibility was atraumatic experience for IBM customers.However, as usual, IBM provided ampleprogram conversion support and successfullyweathered the storm.

13

Many companies which placed heavy emphasison their engineering calculations ended upwith either a 6000 series CDC computer or aUNIVAC 1108 type machine for thosecalculations. There were significantdifferences between the CDC and UNIVACversions of Fortran but even largerdifferences between those versions and theIBM Version (32). However, for documenta­tion of algorithms in technical journals,ALGOL was the standard language (33).Since very few of the programmers knewALGOL, this made the algorithm literaturemuch less useful for engineering applica­tions.

From a Chemical Engineering calculationpoint of view, the key point is that theselarger and more powerful computers made thedevelopment of major Flow Sheet SimulationPrograms practical. Computer Aided Chemi­cal Engineering became a reality. Pre­viously, ability to perform such calcula­tions was much more limited both in termsof what could be done as well as how manypeople were able to do them. Ed Rosenpresented an insider's view of the develop­ment of FLOWTRAN (34). Most major Oil andChemical Companies as well as Engineeringand Construction firms were developingtheir own major systems.

In the same time frame, independent soft­ware vendors were developing their ownsystems primarily used via Service Bureaus.However, the resources required normallyexceeded those an independent vendor couldput together and they had much difficultycompeting with in-house systems unless veryspecialized technology was involved.Interactive computing was quietly growingalso.

CONTINUED GROWTH 1974 - 1984

With continued growth in the usage andcapabilities of the computers and thesoftware packages, there was also continuedgrowth in the cost of supporting andmaintaining these systems. In addition,major cooperative efforts such as ASPEN andvarious Design Institutes made the computa­tional technology more readily available(35,36). As a consequence, the majorpackages from independent software vendorsstarted to become more competitive incapabilities with the large in-housesystems.

With the continued shrinking of computercircuitry based on large scale integration,it became feasible to put an entire CPU ona chip (37) and Microcomputers becameavailable. Minicomputers was the namegiven to what formerly were called Mediumsized computers. With the progress incircuitry the Minicomputers were competi­tive in performance with the large mainframes of just a few years earlier vintage.One of the more important applications ofthose Minicomputers is Computer-AidedDesign (CAD) which has been defined as theuse of interactive graphics programs todevelop diagrams, parts lists and workingdrawings (38).

As CAD developed and proliferated, thecapabilities were extended to the utiliza­tion of the CAD information in NumericallyControlled manufacturing devices to makeComputer Aided ManUfacturing (C~~) muchmore common. Tying the CAD system withProcess Simulation and Data Base Managersis leading to Computer Aided Engineering(CAE) • The progress has been especiallyspectacular in computer chip design (39).There are so many details to contend withand wiring lists have been in widespreaduse for many years.

14

However, Process Design differs in manyfundamental ways from Chip design andprogress in that kind of CAE has been muchslower (40) even though we are gettingclose to practical Ch.E. work stations.Relational Data Bases do offer much promise(41) and Local Area Networks (LAN) areproviding communication capabilitiescommensurate with work station needs (42).

One byproduct of the much greater use oflarge scale integration computer chips isthe gain in reliability. The firstcomputer I used required four hours per dayof hardware maintenance in addition toabout eight hours every weekend. When thecause of a program failure was not anobvious program or input error, there was afifty percent chance that the fault was ahardware malfunction. Today, the amount ofself testing which is automatically per­formed is sufficient to make undetectedhardware malfunctions an extremely unlikelycallse of software failures (43).

Consequently, Microcomputer reliability isnow sufficient that they are being widelyused in Process Control Systems. With themuch greater computational capability nowavailable to Control Systems, major changesin control philosophies are now practicaland receiving serious attention (44).

WHERE TO FROM HERE

The strong trends now underway are in someways incompatible so it is difficult topredict where the current trends will betaking us. There is much effort beingdevoted to CAE for Process Design with thetraditional Ch.E. software vendors playinga minor role and the CAD software vendorsbeing dominant. However, data base utili­zation is a critical part of the majorgains anticipated and the current CADsystems have their own specialized andrelatively inflexible data bases.

J.A., and Johnson,Statistics, and

Microcomputers are rapidly becoming muchmore powerful and widely used. Yet theprimary applications are Word Processingand Spread Sheet applications (45) .Software development costs inhibit thedevelopment of more specialized programswith a relatively limited market. To me,the most logical direction is distributedsystems where the Microcomputers do thosetasks which they are most cost-effectivefor and large Mainframes continue to dowhat they do best. However, there are anumber of major problems which need to besolved before distributed systems will bepractical.

Widespread use of computers in ChemicalEngineering education is anticipatedbecause of the substantial benefits pos­sible. The programming effort required maybe less than the computational effortavoided (46).

REFERENCES

1. Leininger, T., "Upstream of CAST - ABrief History". CAST Newsletter, Vol.7, No.1, April 1984, pp. 5-11.

2. Opler, A... , and Heitz, R.G., lIMachineCalculation of the Plate-by-PlateComposition of a MUlticomponentDistillation Column". Proceedings,Industrial Computation Seminar, IBM,1951, pp. 18-23.

Ralston, A., and Meek, C.L.,Encyclopedia of Computer Science,Petrocelli Charter, 1976, pp. 192-193.

Ibid. pp. 59-62.

Awad, E.M., Automatic Data Processing,Prentice-Hall, 1966, pp. 22-24.

Ibid. pp. 61-128.

Ralston, loc.cit. pp. 34-35 and 488.

15

8. Ibid. pp. 1463-1464 and 488.

9. Kahn, D., The Codebreakers, MacMillan,1967.

10. Ralston, loc.cit., pp. 136-137 and 489.

11. Awad, loc.cit, pp. 115-117.

12. Ralston, loc.cit., pp. 474-478 and488-495.

13. Eckert, W. J., "The Role of the PunchedCard in Scientific Computation". IBMProceedings, pp. 13-17.

14. Rose, A., and Williams, T.J. t "PunchedCard Devices for DistillationCalculations" .. Industrial andEngineering Chemistry, Vol. 42, 1950,pp. 2494-2497.

15. Opler, A.. , and Heitz, R.G., "PunchedCard Calculation of SiX-ComponentDistillation Columns with HeatBalancing".. Industrial and EngineeringChemistry, Vol. 43, 1951, pp.2465-2471.

16. Rose, A., Williams, T.J., and Dye, W.S.III, ttcontinuous Distillation DesignCalculations with the IBMCard-Programmed Electronic Calculator".IBM Proceedings, pp. 24-31.

17. Rope, A., Schilk,R.C., "Computers,Mathematics".UIndustrial and Engineering Chemistry,Vol. 45, 1953, pp. 933 - 940.

18. Sharpe, W.P., The Economics ofComputers, Columbia University Press,1969, pp. 185 - 187.

19. Awad, loc.cit., p. 28.

20. Rosen, S., Programming Systems andLanguages. McGraw-Hill, 1967, pp. 3-7

21. Backus, J. W., et al. liThe FortranAutomatic Coding System". Presented atthe Western Joint Computer Conference,Los Angeles, 1957.

39. Swerling, S., and Langeler, G.R., "CAETool Supports both Design and RoutineTasks". Electronic Design, May 27,1982, pp. 117-125.

22. Rosen, loc.cit., pp. 8 - 18.

23. Bell, C.G., and Newell, A., ComputerStructures: Readings and Examples.McGraw-Hill, 1971, pp. 137-144.

40. Cherry, D.H., Grogan, J.C.,G.L., and Perris, F.A., llUseBases in Engineering Design".Engineering Progress, Vol. 781982, pp. 59-67.

Knapp,of Data

Chemical(5) May

24. Awad, loc.cit., p. 6.

25. Ralston, loc.cit., p. 300.

26. Bell, loc.cit., pp. 421-429.

41. Westerberg, A.W., and Chien, H.H.,Proceedings of the Second InternationalConference on Foundations of Computer­Aided Process Design. CACHE, 1984, pp.75-165.

27. Ralston, loc.cit., p. 1359.

28. Bell, loc.cit., pp. 489-503.

29. Ralston, loc.cit., pp. 481-484.

42. Gordon, H. R., "Engineering WorkstationNetworking" . Paper 56f presented atA.LCh.E. Conference, Anaheim, 23 May1984.

43. Bell, loc.cit., pp. 166-168 and328-329.

Edgar, T. F . ,Chemical

30. Bell, loc.cit., pp. 515-525.

31. Ibid., pp. 561-603.

32. Rosen, loc.cit., pp. 23-28.EngineeringP-33, 1981.

and Seborg, D.E., Ed.,Process Control II,Foundation PUblications,

33. Ibid., pp. 8-12 and 48-117.

35. Ibid., pp. 48-49.

45. Ferris, D. , "The Micro-MainframeConnection". Datamation, Vol. 29 (11),November 1983, pp. 126-138.

46. Finlayson, B.A., "The Impact ofComputers on Undergraduate Education".Chemical Engineering Progress, Vol.80(2), February 1984, pp. 70-74.

Kaufman, D.J. ,& Monsanto It •

2, 1981 Pl'.

andDesignIndustry

41-57.

Rosen, E. M. ,"Computers,Computers in

34.

36. Mah, R.S.H., and Seider, W.D.,Foundations of Computer Aided ChemicalProcess Design., United EngineeringTrustees, 1981, pp. 425-510.

Micro intoVol. 24,

37. Cassell, D.A., "Putting thePerspective". Datamation,(8), August 1978, pp. 97-102.

38. George, B., "Leoking at CAD/CAM/CAEThen and Now" .. Software News,September 1984, pp. 50-52.

L16

The project involved five major activities.These were:

In late 1983, the U.S. Department of Energy(DOE) contracted with the Halcon SD Group(HSD) of New York to develop a steady-statesimulation model for the Great Plains CoalGasification Plant. The model was to bedeveloped using the ASPEN (Advanced Systemfor Process Engineering) program developedfor the DOE by the Massachusetts Instituteof Technology. The plant, now in start up,is designed to produce 125 MM SCFD ofsubstitute natural gas from the gasifica­tion of North Dakota Lignite. It was theDOE's intention to demonstrate that theASPEN program could be used to provideuseful simulation models of the plant.

by

Inc.

Project

Group,

PROPERTIES

A second phase of the physical propertiestask was to establish parameters that couldbe used to predict the highly non-idealvapor-liquid and liquid-liquid behavior inthe process sections of the GPGP. Thepreferred approach would be to use a singleequation of state to predict the phaseequilibria for all sections of the plant.This was not possible. Some of the sec­tions use activity coefficient models butthe majority use the Peng Robinson equationof state modified for water.

An evaluation of the pure component proper­ties in the ASPEN data bank for the thirtyfive compounds required to simulate theGreat Plains Gasification Plant (GPGP) wasmade. Both unary and temperature dependentproperties were checked. The DIPPR (DesignInstitute for Physical Properties Data)data bank was used as the basis of compari­son. It was found that the ASPEN data bankis for the most part reliable, its mainproblem being lack of documentation.

ASPEN Model

Halcon SD

PHYSICAL

Plains

I.H. Renard

Grea t

by Dr.

Physical properties evaluation andmodeling,

Incorporation of a rigorous moving-bedgasifier model into ASPEN,

Cost estimation system evaluation andenhancement,

3.

2.

1.

4. Development of flowsheet models foreach of twelve plant sections from rawgas quenching through gas drying andcompression, and

5. Case studies using the flowsheet modelsto determine sensitivity to keyoperating and design parameters.

GASIFIER MODELThis task called for the incorporation intoASPEN of a rigorous model for a moving-bedgasifier. This was taken to mean a modelthat can predict results of changes inoperating conditions on raw gas flow rate,composition, temperature; carbon conver­sion; and maximum solids temperature withinthe bed. An extensive survey of theliterature yielded a number of gasifiermodels which appeared to be reasonablecandidates for use in ASPEN. Of these, twomodels best satisfied the various evalua­tion criteria; one was developed at WestVirginia University (WVU) by Wen and Desaiand the other at the University of Delaware(UD) by Denn, Wei, and Yoon. Both have

17

ASPEN COST MODELS

been included in the gasifier unit opera­tions block RGAS that has been incorporatedinto the DOE version of ASPEN. RGAS alsoincludes a detailed devolatilizationmodel. The UD model has been found to beeasier to tune for a specific coal andsomewhat more robust that the UWV model.

A new method for the estimation of commod­ity material and labor costs, which areassociated with ME (Major Equipment)installation, has been developed andincorporated into the ASPEN CES. Theoriginal ASPEN method used cost-independentfactors supplied by the user to determinethese OE (Other Equipment) costs. The newmethod which has been implemented in ASPENcalculates OE factors from correlationsbased on purchased equipment cost.

The ASPEN program contains within it a CostEstimation System (CES) which estimates thepurchase cost and utility consumption ratesfor major pieces of equipment in a processflowsheet. Each cost model contained inthe CES was screened by making severalstand-alone runs for equipment that hadbeen purchased by HSD in the past. Wherelarge discrepancies resulted between theprice calculated by the CES and the actualpurchase cost of the equipment, adjustmentsor changes to the CES models were made.These included either adjustment of thefactors built into the models, incorpora­tion of new cost correlations, or thedevelopment of new models altogether.

Almost all the technology in this list isproprietary. No detailed technical infor­mation such as heat and material balancesor equipment design specifications wereavailable from the licensors. All thefLowsheet models are based entirely onpublicly available information includingthat presented during the course of variouspublic hearings as well as what is avail­able in the literature.

The length of this article does not permitdetailed descriptions of these models,their reliability, and the case studyresults obtained thereform. Needless tosay, the simpler the flowsheet, the morerobust the model. For more details, referto the more detailed summary paper given atthe 1984 Pittsburgh Coal Conference ("ASteady-State Simulation Model for the GreatPlains Coal Gasification Plant," by I.H.Rinard, S.S. Stern, M.C. Millman, B.W.Benjamin, K.J. Schwint, and D.R. Carnegie,all of HSD; L.E. Graham of DOE; and J.S.Dweck and M. Mendelson of JSD Consultants,Inc.). There are also a series of 20topical reports for this project whichshould be available by the end of 1984.

1. Oxygen Plant

2. Gasifier Quench, Shift Conversion, GasCooling, and Gas Liquor Separation

3. Rectisol (Acid gas removal)

4. Methanation

5. Gas Compression and Drying

6. Phenosolvan (Phenolics removal)

7. Phosam (Ammonia removal)

8. Cooling Tower

9. Methanol Plant (Rectisol solventmakeup)

10. Stretford (Sulfur recovery)

11. Steam Distribution

beenThese

flowsheet models havepart of this proj ect.

Eleven ASPENdeveloped asinclude:

PROCESS SIMULATION

18

ASPEN USER GROUP

this tool is to let every ASPEN Users GroupThe public ASPEN Users Group met onNovember 26, 1984 in conjunction with theNational AIChE meeting in San Francisco,California. The next general meeting willbe in Houston, Texas on March 24th.

member haveactivitiesCommittees.tions aboutboard.

access to information about theof the various Users GroupContact Jim Henry for instruc­how to implement the bulletin

The other committee chairman appointed are:

The tentative agenda for the March meetingis:

Documentation and Training - Alfred Dyson(205-729-3643)

The ASPEN Users Group now has about 250members. There is only one official membername and address list, the one maintainedby the secretary. If anyone is interestedin joining the group or wants informationabout any of the User Group's activitiesplease contact any of the officers.

The Physical Properties committee iscomparing the unary parameters of the ASPENdata bank for the first 250 components withthe values in DIPPR. This should becompleted by the March meeting so thecommittee can recommend changes in the databank to the Custody and Corrections commit­tee. Anyone interested in helping withcomparing DIPPR vs ASPEN PCD values shouldcontact Martin Millman.

Due to the Great Plains Coal Gasificationproject, there has been a significantimprovement in the Costing and EconomicsSystem. The changes have been incorporatedinto the public version and are availablethrough NESC (National Energy SoftwareCenter).

Stearns Catalytic Corporation has written aCondensed Users Manual for use in theircompany. It has been donated to the UsersGroup and for a reproduction/tabs/bindercharge of $52.00 they will deliver a manualto you. Contact either Alfred Dyson orTrina J. Ige1srud to place an order.

DOE/METC has put the ASPEN documentation onthe Wang system. The manuals need to beproofread for both spelling and conceptualerrors. Anyone interested in helping outwith this task please contact Rita Bajuraor Alfred Dyson.

BusinessTechnical

Martin Millman

has a 'bulletinHenry from theThe purpose of

March 24 5:30 8:00Committee Reports,

Every November new officers are elected bythe ASPEN Users Group. The new officersare:

Monday,Meeting,talks.

Custody, Corrections and EnhancementsRobert MacCallum, (918-660-3153)

Trina J. Igelsrud, SecretaryStearns Catalytic CorporationBox 5888Denver, Colorado 80217303-692-4286

The ASPEN Users Groupboard' supplied by JimUniversity of Tennessee.

Physical Properties(212-689-3000)

Rita A. Bajura, ChairmanDepartment of EnergyBox 880Morgantown, W.Virginia 26505304-291-4714Jim Henry, Vice ChairmanUniversity of Tennessee at ChattanoogaChattanooga, Tennessee 37402615-755-4398

19

Anaheim, CA May 20-23, 1984

MICROFICHE

Session 29 Process Modeling With Compu­ters-II a,b: Fiche 7; c/d,e:Fiche 6

j

Human Factors in Control SystemDesign a,c,d: Fiche 17; b:Fiche 36

Session 41 Microcomputers in Education a:Fiche 12, b,c,d,f: Fiche 13

Session 42 Computer-Aided Process PlantDesign b,c,d,g: Fiche 12

Session 43 Process Data Reconciliation andRectification-I a,c,d: Fiche 19

Session 44 Advanced Computer Control inToday's Refinery a: Fiche 37,b,c,d: Fiche 13

Session 55 Process Data Reconciliation andRectification-II a,c,d: Fiche16

session 5

session 30 Microcomputers in ProfessionalPractice: II b,c,d: Fiche 29

Session 31 Software for Advanced Controla,b,d,e: Fiche 34; c/f: Fiche22

Session 15 Microcomputers in the Labora­tory a,f: Fiche 16; b: Fiche36, h: Fiche 15

Session 16 Process Modeling With Compu­ters-I a: Fiche 1, c,d: Fiche 2

Session 17 Microcomputers in ProfessionalPractice: I b,d: Fiche 22

Secosion 18 Software for Control SystemDesign a: Fiche 3; b,d: Fiche 4

Session 28 Microcomputers-Are They Usefulin Project Management? d,e:Fiche 31

Session 60 Computer Aided Engineering inthe Chemical Process IndustryII: Project Executionb,c:Fiche 51

20

ORDERING

Session 44 Major New Directions in ProcessControl: I a,c: Fiche 43

Session 50 Application of Computers in theControl and Management of Bio­logical Processes-II a, e,f:Fiche 32; b: Fiche 33, d: Fiche31

Session 43 Computer Aided Engineering inthe Chemical Process Industry-IProcess/Project Evaluation b:Fiche 36; c,d: Fiche 35

Washington, D.C., Oct. 30-Nov. 4, 1983

Session 19 Chemical Engineering Educationin the Next 25 Years: II-Gradu­ate c: Fiche 30

Session 21 History of Fractional Distilla­tion b,f, Fiche 44; c: Fiche 45

Session 26 Computer-Aided Process Analysisand Synthesis e,f: Fiche 5

Session 33 Application of Computers in theControl and Management of Bio­logical Processes-I a/b/c,f:Fiche 29

Session 2 Chemical Engineering Educationin the Next 25 Years: I-Under­graduate a,c: Fiche 51

Session 9 Innovations In Applied Mathe­matics In Chemical Engineeringb: Fiche 54

Microfiche are available from AIChE Head­quarters for at least one year after themeeting. Prices: $2.50 per microfiche forAIChE Members, $4.50 per microfiche fornon-members. All sales are final noreturns are accepted. For additionalinformation, call AIChE Technical Publica­tions department (212) 705-7335

Session 56 Progress Toward Process Engine­ering Workstations a,d: Fiche20; c: Fiche 31

Session 61 Pilot Plant Instrumentation andControl a,e: Fiche 9; b,e:Fiche 10

Philadelphia, Aug., 19-22, 1984

Session 43 Emulsion Polymers and EmulsionPolymerization - I ComputerModeling of Emulsion Polymeri­ization a,b,d: Fiche 27

Session 53 Investment Decision AnalysisTwenty-Five Years Later - Arethe Numbers More or Less Help­ful Now? - I a,b·,c,d,e: Fiche2

Session 54 Investment Decision AnalysisTwenty-Five Years Later - Arethe Numbers More or Less Help­ful Now? - II a,b: Fiche 20

San Francisco, CA November 25-30, 1984

Session 8 Advances in Applied Mathematicsb: Fiche 78; e,f: Fiche 108

Session 27 Advances in Numerical Analysisand Applications a,e,g: Fiche47; c,d,f: Fiche 48

Session 42 Use of Computer Executive Pro­grams in Plant Design b: Fiche30

Session 45 Computer-Aided Design of Batchand Semicontinuous Processesa,d: Fiche 118, b,e: Fiche 72:c,f: Fiche 71, g: Fiche 119

Session 57 Pilot Plant Seminar I c,f:Fiche 54

Session 63 Computer-Aided Process Designand Analysis b,e: Fiche 41, c:Fiche 39, d: Fiche 29, f: Fiche112

,'" 21

Session 76 Pilot Plant Seminar II a,c:Fiche 106

Session 82 Topics in Process Control Ib,f: Fiche 68, d,e,g: Fiche 69

Session 88 Advances in Process Synthesisa: Fiche 54; b: Fiche 52, c:Fiche 107, d: Fiche 55, e:Fiche 53, f: Fiche 51

Session 91 Modeling, Simulation andControl in the Forest ProductsIndustry-I a,b,c: Fiche 15,d,e,f: Fiche 113

Session 101 Topics in Process Control IIa,c,f: Fiche 90; b: Fiche 10

Session 104 Combinatorial Optimizationb:Fiche 110, c,d,f: Fiche 103;e: Fiche 109

Session 110 Modeling, Simulation and Con­trol in the Forest Products In­dustry-II a,b: Fiche 73; c:Fiche 51, d,f,g: Fiche 34

Session 120 Modeling and Identificationa: Fiche 45; c,d,e: Fiche 44;f: Fiche 43: g: Fiche 76

Session 123 Production Scheduling a,d,f:Fiche 94, c: Fiche 121

Session 131 Process Fault Detection andDiagnosis a,b,c,d: Fiche 26; e:Fiche 32

PROCESS

-4th

PSE 85 will pick up the key themes of thefirst of the triennial series entitled PSEheld in Kyoto in 1982, concentrating onOPERABILITY CONSIDERATIONS AT DESIGN STAGE:THE DESIGN OF FLOWSHEETS: AIDS FOR pLANTOPERATION and THE INFLUENCE OF NEW CADTECHNOLOGY. While traditional subjectssuch as flowsheeting will be considered,the emphasis of this multi-national pro­gram will be very much on new topics andnew directions.

PSE 85 is being organized by the Institu­tion of Chemical Engineers on behalf of theEuropean Federation of Chemical Engineer­ing, in association with the Inter AmericanFederation of Chemical Engineering, and theAsian Pacific Confederation of ChemicalEngineering. As EFCE Event No. 313 thisconference will continue a successfulseries. of symposia held at the invitationof the EFCE Working Party on ComputerApplications, taking place in Montreux in1979, Heviz 1980, Vienna 1981, Antwerp 1982and Paris 1983.

Registration will take place on the after­noon and evening of Sunday 31st March 1985.Technical sessions will be held from themorning of Monday 1st April to lunch timeThursday 4th April. The afternoon of the4th will be devoted to technical visits anddiscussions.

MEETING SITE

Robinson College, Cambridge will be thevenue for this multi-national event. Anarchitecturally award winning complexRobinson College provides excellent confer­ence facilities including accommodation ofan exceptionally high standard. Particularattractive in the Spring, Cambridge offersa wealth of tourist interest to delegatesand their partners.

22

TECHNICAL PROGRAM

The response to the Call for Papers hasbeen excellent and many papers of a highcaliber are still being considered by theTechnical papers Committee. A finalprogram is not available at this stage, butan outline of the sessions and theirtechnical content is given below. Completedetails about papers and authors will bepUblished in the Final Announcement whichwill appear during the early part of 1985.All papers selected for presentation willbe published as No. 92 in the IChernE' sSymposium Series.

OPERABILITY CONSIDERATIONS AT DESIGN STAGE

Ths primary theme of this session is thekntroduction of flexibility, dynamicresilience and controllability at thedesign stage and the influence of uncer­tainty upon them. New approaches aredescribed Which promise substantiallyimproved plant operation resulting from theintegration of these measures into theearliest stages of design, rather thantrying to compensate for them if problemsarise later.

THE DESIGN OF FLOWSHEETS

This session reports on the latest advancesin the development of technology forflowsheet simulation and the optimum designof process networks. Various techniquesare examined, including the potentialapplication of Artificial Intelligence tosuch design. The handling of uncertaintyin design information is also covered, asis the treatment of multi-phase systems.

AIDS FOR PLANT OPERATION

Papers under this heading cover trainingsimulators, methods of detecting theprimary causes of failure alarms and waysof monitoring plant performance, includingthe analysis of measured data. The sessionis particularly valuable in that it in­cludes a number of contributions whichreport on industrial experience.

THE INFLUENCE OF NEW CAD TECHNOLOGY

The papers in this session describe the wayin which new CAD technology is now becomingavailable for application. Ideas such asgraphical input/output, data management andspecial methods for the use of Super­Computers are discussed here, as well asthe possible exploitation of entirely newtechnology like Artificial Intelligence andFuzzy Set Theory. The practical employmentof this new technology is illustrated bydescriptions of its introduction to and usein major industrial organizations.

POSTER SESSION

Papers best suited to a poster stylepresentation are included under thisheading. Moreover, the session allowsinteresting and valuable papers, whichcannot be accommodated within the confinesof the above formal sessions, to be pre­sented. The range of subjects is thereforediverse, from cost diagrams to programdesign methodologies.

23

EXHIBITION

The conference is expected to attractdelegates throughout the field of PSE fromallover the world. Floor space is avail­able for approximately 12 companies to setup their own displays exhibiting productsrelating to process engineering CAD.

TECHNICAL PAPERS COMMITTEE

F.A. Perris (Chairman)Air Products

P.S. BanksBP

D. DepeyreEFCE CAD Working Party Secretary

D. W. GillingsConsultant

J.D. Perkins (Vice-Chairman)Imperial College

J.W. PontonUniversity of Edinburgh

M.L. PrestonICI

G.V. ReklaitisPurdue University, USA

L.M. RoseETH Zurich, Switzerland

R.W.H. SargentImperial College

T. TakamatsuKyoto University, Japan

P. WinterProsys Technology

FURTHER INFORMATION

Miss Julie Wearne, The Conference SectionThe Institution of Chemical Engineers165-171 Railway TerraceRugby, CV21 3HQ, UK

Tel: 0788 78214 - Telex 311780

"FOCAPD PROCEEDINGS"

Proceedings of the Second InternationalConference on Foundations of Computer-AidedProcess Design held at Snowmass, Coloradoin the Summer of 1983 are now available.

The week long meeting, sponsored jointly bythe CAST Division of the AIChE and CACHE,with support from the National ScienceFoundation and Chiyoda Engineers and Con­struction Co., EXXON Research and Develop­ment Co., the Halcon SD Group, MonsantoCompany, Olin Chemicals Corp., ProcessSimulation International, Shell CompaniesFoundation, Tennessee Eastmen Co., andWeyerhauser Co.

Professor Arthur Westerberg (CarnegieMellon University) and Dr. Henry H. Chien(Monsanto Co.) were the meeting co-chairmanand served as editors of the Proceedings.The Proceedings (a hardbound book of morethan 1000 pages) contain all 22 papers pre­sented during the meeting, plus summariesof the discussions held during each session.The major sessions were:

1. Keynote (Expert Systems)

2. Overview and Outlook for ProcessSystems Engineering

3. Progress in Data Base Development

4. Computational Algorithms

5. Physical Properties for Design

6. Nonsequential Modular Flowsheeting

7. Design and Scheduling of BatchChemical Plants

8. Complex Single Unit Design (columns,reactors)

9. Operability in Design

10. Contributed Papers (various topics) ,

The text is available at $37.50 from:

Professor Brice Carnahan, ChemicalEngineering Dept., The University ofMichigan, Ann Arbor, MI 48109

24

MESSAGES'Continued from Page 1

Ten days spent in isolation in Blacksburgis insufficient time to develop a plan forthe Newsletter during the 1985-86 period.A few thoughts have occurred that are worthsharing. First, I served a two-year ap­prentice period as Associate Editor-in­Chief of IEEE MICRO during 1980-1982. Sucha period prepared me for my selection asEditor-in-Chief during the fall of 1982.Continuity in a publication is important,so I would recommend the appointment ofseveral colleagues, one of whom would bethe candidate for Chairman of the Publica­tions Board in 1987-88.

Second, I invite all members of CAST toparticipate in providing contributions toour Newsletter. We shall continue what wealready do well, shall offer our bestmaterial for publication directly in CEP,ahd shall experiment, with your assistanceand guidance, with new ideas that have astheir objective an expanded role for theCAST Publications Committee. We expect to"win a few, and lose a few" in our attemptsto move the CAST publication activity for­ward, but hope to identify several innova­tions that will have lasting value.

Third, and finally, I will be on sabbaticalleave at the University of Delaware untilJune 1985. My office phone number will be(302) 368-5580. I invite you to sendinformation for the spring 1985 CAST News­letter immediately. Meeting announcements,feature articles, extended abstract of yourpresentations at AIChE meetings, informa­tion on user's groups, messages, opinions,commentaries, polls, advertisements, and soforth are solicited, as are your ideas onnovel contributions to the Newsletter andthe names of individuals who can do thewriting.

CAST Related Papers from San Francisco Meeting

M. GolubitskyUniversity of Houston

Paper No. 8a

The strategy of focusing on de­generate bifurcations and thenperturbing often helps one findinteresting quasi-global be­havior. Singularity theory pro­vides methods for implementingthis strategy. These methodswill be illustrated by discussingtwo examples: The CSTR andclamped Hodgkin-Huxley equations.These examples illustrate howsingularity theory may be appliedto bifurcations of both steadystates and periodic solutions.

rQMRllTER=AID~D_ARRLIED

MATHEMATlrS~_ANALYSlSQ~_rQMRLEX_YlSrQllS_EREE

SllR~Ar.lL~LQIY

K. N. Christodoulou andL. E. Scr i ven

University of Minnesota

Paper No. 8b

Linear stability theory and bi­furcation analysis are brought tobear by similar computer-aidedmethods, for which the CRAY-l andCDC CYBER 205 supercomputersprove cost effective. Thesemethods, lead to a sequence ofgeneralized asymmetric eigenprob­lems. The paper also exemplifiescomputer-aided functional anal­ysis of viscous free surfaceflows.

25

8N_QRER8TQR=THEQRETlr8gRRQ8~H_IQ_11NE8R

IR8NSgQET_RRQ~ESSES

IN_Qr~QTE_QUSgENSIQNS

A. K. Kulshreshthaand J. M. Caruthers

Purdue University

Paper No. 8c

Dilute suspension theory draws onthe solution of boundary valueproblems formulated on a singleparticle in the infinite contin­uous medium. Linear operator-­theoretic methods are shown to bea powerful tool in that spectralsolutions of associated self-­adjoint operators can be obtainedfor a variety of physicallyrealistic boundary conditions.

~1~llR~8TIQN_8NQ_ST~111IY

IN_~L8ME_RRQR8Q8TIQN

B. J. MatkowskyNorthwestern University

Paper No. 8d

We derive two simplified modelsfrom the general equations des­cribing flame propagation. Wethen describe sequences of bifur­cations, and correspondingchanges in stability, as parame­ters of the problem are varied.Such sequences of bifurcationsare thought to represent stagesin the transition from laminar toturbulent flame propagation.

8ggLl~8TlQN_QE_THE_IHEQRY

QE_GENER8LlkEQ_SlMR1E_~8YES

IQ_SllRf8~T8NT_INJE~TIQN

gEQ~~SSES_IN_RQRQllS_MEQ18

Y. C. Yortsos and S. Saneie,University of So. California

Paper No. 8e

We develop solutions to three-­component, two-phase surfactantinjection processes in porous me­dia for enhanced recovery of oil.The method of generalized simplewaves is to show that concentra­tion and saturation profiles canbe obtained with minimal computa­tional requirements for systemsfollowing suitable phase equilib­ria.

~8Elb!lQ~8L_A~ALXSlS_QE

~Q~Oll~!IQ~_l~_!~Q=£HbSE

EANDQIU:lEIUA

W. Strieder and D. S. TsaiUniversity of Notre Dame

Paper No. 8f

Variational methods are consid­ered which bound effective trans­port properties of random media.Dispersed spheres and randomlyoriented fibers were used tomodel a number of materials ofengineering interest. Estimatedthermal conductivities of plasticmetal composites, fiber glass in­sulation, and diffusivities ofgranular and fibrous materialsare compared with measuredvalues.

SJ:;SSlQ~_21

Q~_~HJ:;_llS.E_Qr_ADA£Il~.E

ME!HQDS_l~_~HE_SQLll!lQN

Qr_~QMallS!lQN_~EQaL.EMS

M. SmookeSandia National Labs.

Paper No. 27a

Efficient numerical solution com­bustion problems requires thatgrid points be placed adaptivelyin the regions in question. Weselect the points by equidistri­buting a positive weight function

26

over consecutive mesh intervals.Both lD and 2D steady state andtransient combustion problems areconsidered.

ALGQEI!HMS_EQB_IN!EGE8!lQNQE_~QMgLEZ_EE8~!IQN_SXS!.EMS

c. W. White IIIand W. D. Seider

University of Pennsylvania

Paper No. 27b

Strategies to improve efficiencyand reliability in the numericalintegration of mass and energybalances for a PFTR are evalu­ated. For combustion reactionsystems, modern stiff integratorsare very effective. The steady-­state, pseudo-steady-state, andchemical equilibrium approxima­tir~s, and exchanges of indepen­dent variables, scaling and arc-­length transformations offerlittle improvement in efficiencyand reliability.

SENSl~1~lXX_8N8LXSIS~Qf

SXS~EMS~QE_DIEEEB.ENXI8L~8N~

8LGEaBAl~_EQll8!lQNS

M. A. Kramer and J. R. LeisMassachusetts Inst. of Tech.

Paper No. 27c

Parametric sensitivity equationsoften arise in dynamic modelingof equilibrium stage processesand solution of partial differen­tial equations. Formulae are de­veloped which can be used to ef­ficiently produce the modelsensitivity equations.

~EGBEE_~HEQEX_AN~_HQMQ~Q£XL

~QQLS_EQB_~QMEQ~EB=AI~E~

EEQ~ESS_~ESIGN

T. L. WayburnClarkson University

Paper No. 27d

This paper presents some of thetheory behind homotopy- continua­tion methods, which have beenused recently by some investiga­tors to solve separation prob­lems. Conditions under whichthese methods are likely to suc­ceed are developed and a few ofthe difficulties are illustratedby simple examples.

IN~EMTIEI~ATIQN_QE_SEA~lALLX=

~ARXIMG_EAEAME~EES_IN

EAR~IAL_DIEEEEEN~IAL

EQQA~lQN§-BX_DIS~EE~E

EEGllLARIZA~lQN

C. KravarisUniversity of Michigan

Ann Arbor, MIand J. H. Seinfeld

Cal Inst. of Technology

Paper No. 27e

The present work refers to the i­dentification of spatially-­varying parameters in partialdifferential equations. Thediscrete regularization approachinvolves the use of the principleof regularization together withan a priori discretization of theunknown parameters via finite­dimensional convergent approxima­tions. Applications to petroleumreservoirs are considered.

27

A_NQNLINEAE_ERQGEAMMINGAEEEQA~H_~Q_~XNAMI~

QE!lMIZA~lQN_EBQaLEMS

L. T. BieglerCarnegie-Mellon University

Paper No. 27f

Chemical process problems des­cribed by differential- algebraicmodels are currently optimized byalgorithms requiring repeatedsolution. A simultaneous solu­tion and optimizating strategythat uses orthogonal collocationand an SQP method will be used.Applications to reactor designwill be presented.

~XNAMl~S_QE_EEBlQ~l~ALLX

EQR~E~_~HEMl~AL_BEA~!QBS

I. G. KevrekidisL. D. Schmidt and R. Aris,

University of Minnesota

Paper No. 27g

A method for studying the dynamicbehavior of periodically forcedchemical reactors is presented.The method yields stability in­formation on harmonic bifurca­tion. bifurcation to tori andfrequency locking, and routes todeterministic chaos.

A_NE~_8EEEQA~H_~Q

~A~~H_£RQ~ESQ_~ESIGN

==I~I~g_a8T~HMAS~EB

G. W. Frank and M. L. PrestonImperial Chemical Industries PLC

Paper No. 45a

The paper reviews experience andunderstanding of the requirementsof batch process design gained

the specifications ofTER, ICI's new batch

process flowsheeting system.This system is built on thefoundation of ICI's schedulingsoftware, Process EngineeringDatabase developments, and over15 years' experience with dynamicdiscrete event simulation lan­guage.

!tlQ_~AQE_S!QDIEQ_IN

fl8!~H_EL8N!_DESIGN

S. Ali and M. MalanchiniInstituto Guido Donegani

S. A. Novara, Italyand T. Ernst

M. Hofmeister and D.W.T. RippinTechnisch-Chemisches Labor

Paper No. 45b

The application of an academical­ly developed program MULTIBATCHintended for optimization ofbatch plant design is discussedbased on two case studies carriedout cooperatively with industry.Difficulties inhibiting the useof such programs for the solutionof industrial problems in thebatch process area are analyzedand successful remedies outlined.

D~!~EMINIS!I~_Y8EIAalLI!X

8NALXSIS_fQE_INIEEM~DI8T~_SXQ£=

AGE_IN_NQN~QNXINQQllS_EEQCEaSES

1. A. KarimiPurdue University

Paper No. 45c

Part 1: Allowability Conditions

Intermediate storage is commonlyused to mitigate the effects ofprocess parameter variations innoncontinuous processes. A tax­onomy and analysis is presentedof the various types of varia­tions. Sufficient conditions aredeveloped which ensure continuity

28

of periodic operation in thepresence of these variations.

Part 2: storage Sizing forSerial Systems

The allowability conditions ofPart 1 are applied to develop in­termediate storage sizing expres­sions for serial systems subject­ed to process parameter varia­tions. Multiple variations ineither starting moments, transferflow rates or transfer fractionsare considered first. Theseresults are then combined using aworst case analysis to developsize estimates under generalvariations.

SXN!HESlS_Qf_fLE~I~LE_H~A!

E~CHANGEE_NEX~QEKS_fQE

MllLXI=EEEIQQ_Q£EEAXIQB

C. A. FloudasCarnegie-Mellon University

Paper No. 45d

Heat exchanger networks insemi-continuous processes musthave the flexibility to handlediscrete changes in flowrates andtemperatures of the processstreams. Using as a basis a mul­tiperiod transshipment model, aprocedure is proposed to derivenetwork configurations thatrequire minimum utility consump­tion with the fewest number ofheat exchanger units.

AEELICAXIQN_Qf_SIMULAXIQNXQ_XHE_CAEACIXX_ELANNING_Qf

A-SEECIALXX_CHEMICALSEBQQUCXIQH~fACILIXX

Richard M. FelderGeorge B. McLeod, IV

North Carolina State university

Paper No. 45e

A method has been developed to e-

I

valuate the effects of plant andprocess modifications on theoverall productivity of a multi­product specialty chemicalsplant. Simulation studies arefirst performed on individualprocesses in each of the produc­tion areas in which the processescan be carried out. A linearprogramming algorithm is thenused to determine the maximumimprovement in overall plant pro­ductivity that can be expected toresult from the implementation ofany process modification. Theresults allow management todetermine cost-to-benefit ratiosof proposed equipment oroperating personnel additionsbefore committing any capital tothe changes.

alMllLA!lQH_Qf_A_LAEG~

S~&L~_MllL!lgEQDllC!

fQLXM~£_gEQ~~Sg_llgIHG_~§QaS~

G. Joglekar, CAE Inc.G. V. Reklaitis

Purdue University

Paper No. 45f

The BOSS flowsheeting system fornoncontinuous processes is usedto model and study the operationof a large scale mUltistage, mul­tiunit batch/semicontinuous pro­cess involving concurrent produc­tion of mUltiple products. Theeffects of schedule on utilityconsumption rates, operator util­ization and assignment, and in­termediate storage distributionare investigated.

29

AH_INDllg!£X_~~8LllA!lQH

Qf_S£~~Dllf

P. K. Gupta & R. C. LavoieExxon Corporation

Paper No. 63a

The paper presents the results ofExxon's evaluation of SPEEDUP, anequation-based process flowsheet­ing system developed at ImperialCollege for the steady-state anddynamic modeling of chemical pro­cesses. The objectives of theevaluation were to determinewhether equation-based systemscan be used effectively forsteady-state simulation in acommercial environment and tocompare SPEEDUP with a commer­cially proven sequential modularprocess flowsheeting system.

A_gXE8!~GX_fQE_gEQC~Sg

8NALXSla_8ND_D~SIGN~

IMfL~MEMX8!IQN

Jorge R. PaloschiRafiqul Gani

Jose A. RomagnoliPlanta piloto de

Ingenieria Quimica

Paper No. 63b

The implementation of a strategyfor process analysis and designis discussed. This strategy ana­lizes the operation throughstudies that relate sensitivityand operability of the process toconditioning of the system ofequations representing it. Fur­thermore, an algorithm isproposed to obtain the initialpoint for the simulation problem.Applications to practical exam­ples is shown by means of twodifferent types of simulationpackages, CHESS and SPEEDUP. One

Paper No. 63c

H. N. Pham and M. R. DohertyUniversity of Massachusetts

8NALXSlS_QE_H~TgBQGENEQll~

A~EQTRQfl~_Ql~TlLLATlQNQ

A robust technique for liquid-­liquid-vapor equilibrium calcula­tions is able to distinguishbetween the stable, metastableand unstable regions of a phasediagram. The method has beenused to compute residue curvemaps for a selection of ternaryheterogonous azeotropic mixtures.

MllLTl~LE-SQLllTlQNQ~TQ

SXSTEMQ_Qf_lNTEBLlNKEQQlSTILL8TIQN_~QLllMNS~X

QlffgE~NXlAL_HQMQTQ~X

~QNTINllAXIQN

T. L. Way burnClarkson University

J. D. Seader and R. ChavezUniversity of Utah

Paper No. 63f

nificance of simulation results.Algorithms for sensitivity analy­sis of complex flowsheets arepresented which are 10 to 100times more efficient than contin­uation methods. Both equation­oriented and modular approachesare discussed.

examples consists of aof an existing ethylene

of thesectionplant.

THEEMQQXN8Ml~8LLX

~QNSlSTgNT_Qll8S1=NE~TQN

lQEMllL8~

A. Lucia, D. C. Millerand A. Kumar

Clarkson University

Paper No. 63d

New quasi-Newton formulae arepresented that obey secant condi­tions and sparsity constraints,while satisfying thermodynamicconstraints. Numerical experi­ments show that these new formu­lae can result in improved relia­bility and efficiency.

M. A. Kramer

This paper describes a computerimpJ2mentation of homotopy con­tinuation which was used to modeltwo differentn arrangements ofinterlinked distillation columns.Using product-purity specifica­tions, so that interlink flowrates were computed rather thanspecified, mUltiple solutionswere found for both arrangementsfor a number of reflux ratios.

QECQM£QSITIQN_ASgE~TS_IN

IHE_QESl~M_Qf_MllLIl~AB18§LE

~EQkESS~tQNIBQLLEBS

G. StephanopoulosMassachusetts Inst. of Tech.

M. NikolaouNat. Tech. University of Athens

Paper No. 63e

Sensitivity analysis aids processdesign by elucidating the impor­tant parameters in a processflowsheet, and by profiding sta­tistical information on the sig-

Paper No. 82a

The concept of block diagonaldominance has attracted a greatdeal of attention for the designof multivariable control systems.In this paper, an attempt is made

30

to elucidate the framework ofanalysis and interpret theseideas, especially as they pertainto the design of steady state ordynamic controllers.

E~ALllAXIHG_SIGNIEI~AMI

~~QMQMl~_XEAOE=QEES_EQE

gEQ~gSQ_DgSIGN_ANO_STEAOX=SIAIE

~QNXEQL_QRXIMIZAI1QM_fEQ5LEMS

W. R. Fisher, M. F. Dohertyand J. M. Douglas

University of Massachusetts

Paper No. 82b

An order-of-magnitude optimiza­tion procedure allows rapidscreening of flowsheet alterna­tives. The method exploits theunique characteristics of thechemical process design problem.Quantitative parameters are de­fined which specify the most im­portant design variables, whilepreventing the rigor of the opti­mization from exceeding theaccuracy of the economic modelsused.

Q~~llEEEN~ES_QE_ZEEQS_IM

QgTIMIZEO_~aEMI~AL_gEQ~ESSgS

W. J. Schmidt, J. M. Fox andJ. M. Douglas

University of Massachussetts

Paper No. 82c

Steady state optimization ofchemical processes results in dy­namical system zeros near theorigin. One consequence is alack of control system robust­ness. This is shown by an exam­ple. The phenomenom of inputmultiplicity is demonstrated.

H1.G!LggEJ!QEMAN~E

MllLTl~AEIAaL~_~QNTEQL

SIEAIEGlgS_EQE_QXSTgMSHAYIHG_TIME_DELAXS

N. F. Jerome and W. H. RayUniversity of Wisconsin

Paper No. 82d

A high performance time delaycompensator has been developedwhich contains the Ogunnaide-Ray(1979) compensator as a specialcase. reduces to the Smith pre­dictor for a single delay. andapproaches the realizable part ofthe process inverse as a limit.A simple design and implementa­tion procedure is described.

QgEEAIQE_~QNIEQL_XHEQEX

AND_A_NE~_~LASS_QE

NQNLIMEAR_~QNXBQLLEES

C. G. Economou and M. MorariCal. Institute of Tech.

Paper No. 82e

Some strong connections betweenoperator and control theories areestablished. They motivate a re­examination of the controller de­sign method in the light of iter­ative operator equation solutionprocedures. Contraction Princi­ple and Newtonian controllers aredeveloped and their performanceand stability characteristics aredirectly investigated.

QllAOBAXI~LQXNAMI~MAIRIX_~QNTEQL_QE

NQNLIMEAB_~QNIEQLLEEQ

C. E. GarciaShell Development Company

Paper No. 82f

31

An application of(Quadratic/Dynamictroll algorithm has

Shell's QDMCMatrix Con­

been used to

control a nonlinear process. Themethod employs differential equa­tions of the system in order to:1) calculate projected deviationsof variables from a specifiedtrajectory; 2) calculate localstep response coefficients. Im­plementation on a semi-batchreactor process is presented.

QN=LIN~_gbRAM~X~E

~SXIMA!IQN_~I_S~QD~N!lbL

~AR!lAL_R~GR~SSIQN

O. A. AsbjornsenUniversity of Houston

Paper No. 82g

"Normal regression equations forparameter estimation are solvedby standard Gauss eliminations,where the input/residual covari­ances serve the pivot selection.Parameter updates stop when theimprovement in the explanation ofthe residual becomes insignifi­cant. Practical experienceproves the method to be extremelyrobust. Outlayers indicatedramatic parameter changes."

SXNXH~SlS_Qf_HQMQGEHEQUS

bZ£QXRQgl~_DlSIlLLA!IQNS

S. G. Levy and M. F. DohertyUniversity of Massachusetts

Paper No. 88a

An exact algebraic method forcalculating minimum reflux ratiosin double-feed azeotropic columnshas been developed. The result­ing procedure allows for asystematic design and synthesisscheme for multi-column azeotro­pic distillations. Competingentrainers and competing columnsequences can be efficiently

,v.32

screened using this approach.

.llS~_Qf._.cQMgL£L~QLllMN12

XQ_Jmlilit:l..c~_H£liI

nlI~GEAnQ~LQf

DlSXILLATlQN_l2ISI£MS_lNXQTQTliL_fLQIil2li£~TS.

E. Hindmarsh. ICI Energy andProcess Synthesis Consultancy

paper No. 88b

A systems approach and newrepresentation is presented forheat integration of distillationcolumns into total flowsheets.Process changes are identifiedfor which discrete and reducingutility targets are calculated.The method highlights thosecomplex columns which furtherenhance heat integration possi­bilities, as demonstrated by ex­ample.

SINxn£SlS_Qf_Q~IIMAL

SAXllBbX~a_S~fABbTIQN

SJ:;QllEN.c£.6.

D. W. TedderGeorgia Inst. of Technology

paper No. 88c

A refinery light endsfractionation problem was studiedusing a computer optimizationprogram (RUNOPT), dynamic pro­gramming, and rank order listingto generate 17 near optimal sepa­ration sequences. A total of23,232 alternative sequences werescreened through the evaluationof 437 subproblems. Stateoptimization. tower interactions,and heat exchanger matching wereall found to be important.Although the unintegrated ranklist of sequences was similar tothe ranked list when integrationwas permitted, sequence shiftswere observed. The optimalsequence included heat exchanger

~

I

reviewed with particular atten­tion to computer-monitored chemi­cal systems.

£EQ~~SS_DXNAMI~S_AND_XHE

~~QrHAQ~I~_aEHA~lQB_QE

~A~~H_LDMaEB_EILNS

Ferhan KayihanWeyerhaeuser Technology Center

Paper No. l20b

A Monte-Carlo model is developedto predict the moisture contenttransients of boards during thedrying of dimension lumber inbatch kilns. Product value de­pends strongly on the final mois­ture content distribution whichis a function of lumberproperties and the drying sche­dule used. Owing to the inherentproperty variations, measurementlimitations and external processupsets, the same schedule doesnot yield the best result forevery batch. The control policyfor this process is defined interms of an end-point optimalcontrol problem. The paper willdiscuss the development of themodel, the process characteris­tics and the requirements for theoptimal control policy.

DXNAMI~_~S~lMA~IQN

Qf_T~M£~BA~llBE_AND

~QN~~N~BATIQN_gBQflL~S

EQB_~QNTBQL_Qf_A

£A~!~D_a~D_B~A~TQB

L. C. Windes and W. H. RayUniversity of Wisconsin

A. Cinar,Illinois Inst. of Tech.

Paper No. l20c

Effective control and yield maxi­mization of an exothermic packedbed reactor is made possible byestimation of conversion and sel­ectivity from on-line temperature

37

measurements. A two-dimensionaldistributed parameter optimalstate estimation algorithm is e­valuated through simulation andexperiments.

QaS~B~r;.!LD~SlGN

all,S@_QILMQD~L

B~Dll~XIQN_X~kHNIQD~S

J. E. Gupta, J. C. VogelJ. A. Porras and J. RomagnoliUniversidad Nacional Del Sur

Paper No. l20d

The use of reduced order dynamicmodels, based on structuraldominance analysis together withfiltering techniques. allows thedevelopment of an "observer" forefficient control purposes. Thisapproach is applied to an indus­trial column, and the numericalresults thus obtained illustrateits salient features.

~QM£bElSQN_Qf_LlN~AB

DIS!BlallT~D_£ABAM~T~B

fILT~ES_TQ_LDM£~D

A££EDX1MANXS

D. J. Cooper, W. F. Ramirezand D. E. Clough

University of Colorado

Paper No. l20e

Optimal distributed parameterfilters are commonly implementedvia and approximating system oflumped Kalman filters. Theeffect of such an approximationis investigated. The theoreticaldevelopment shows a loss inspatial correlation for thelumped approximants resulting inslower convergence. This pheno­menon is illustrated by two nu­merical results thus obtainedillustrate its salient features.

MQD~LING_~Y-bEQR~TQR_SXST~MQ

fQR_R8DIQ8~TIY~_~8ST§

Habib Amin, Bechtel Power Corp,K. C. Chiu, Bechtel Petro., Inc.David James, Bechtel Power Corp.

Paper No. 120f

A generic computer model isdeveloped for the dynamic simula­tion of the radioactive wasteevaporator systems in nuclearpower stations, The developedsystem component models areintegrated with Bechtel's DynamicAnalysis Program to simulate thesystem's dymarnics, The presentedexample illustrates the strongpotential of the model forsolving the system's controlproblems.

gEEECI_Ql_RE~XCLE

~IRU~TUBE_QN_DlSIlLL!TlQll

IQ~ER_IIME_~QN~T8NIS

T. J. Mc Avoy and N. Kapoor,University of Maryland

and T. E. MarlinExxon Research and Engineering

Paper No. 120g

Recycle can change the dynamicsof a process. Distillationtowers are inherently recycleprocesses. Published time con­stant estimates of towers can besubstantially in error whencompared with actual responses.The reason for this is therecycle nature of distillationtowers. A greatly improved meth-'od of estimating time cOustantsis presented.

81':L8LGQRnH/:LEQR~~HgQQLING_RgQDll~TIQN_IN

8_!1!J.L!IRRQI1U£T_£1E.t;£l8LTX~B.]:;MI.ctU;','LEMNX

R. M. Felder and P. M, KesterNorth Carolina State University

Paper No. 123a

A procedure has been developed toschedule production for amUltiproduct chemical manufactur­ing facility. The algorithmdevelops a schedule thatminimizes inventory costs whilemeeting a specified set ofproduct demands. The code hasbeen structured to fit on amicrocomputer, to solve fairly}arge problems, and to beaccessible and easy for plantpersonnel to use. Applicationsinclude determining whether pro­duct demands can be met withexisting plant capacity, calcu­lating the capacity increasesneeded to meet demands for whichpresent capacity is inadequate,and estimating the impact ofshutting down production areasfor specified periods of time.

1. SuhamiExxon Corporation

Paper No. 123b

CRUST is an interactive schedul­ing system, developed for theSupply and Transportation Depart­ment of Esso A. G., German af­filiate of Exxon. Its purpose isto computerize a large portion ofthe schedUling and coordinationof crude supply for the threeGerman refineries. CRUST allowsthe scheduler to evaluate

different refinery, pipeline, andship schedules.

D~I~RM!N8IIQN_Qf_~QMfL~IIQN

I!M~~_lQR_~ERI8L_MllLI!fBQQQCI

fRQC~~~~S_QNQ~R_M!ZEQ_INI~E=

MEQIAI~_STQE8GE_fQLICIES

W. Wiede Jr., and G. V. ReklaitisPurdue University

Paper No. l23c

A recursive scheduling procedureis reported which determines theearliest completion time andintermediate storage assignmentsfor a given product processingsequence on a serial processinvolving a variety of interstagestorage provisions. Any combina­tion of unlimited finite no-in­termediate storage or zero-waitprocessing policies can beaccommodated.

8ffLIC8TIQN_Ql_INX~B8CTIYE

Yl~U8L_SIMllL8IlQN_TQ

EElINEEX_QfEE8IIQNSCHEQULING

J. StewartImperial Oil Ltd.

Paper No. 123d

A new comput~rized system hasbeen developed by Imperial OilLimited for the scheduling ofrefinery operations. The programinteracts with the scheduler toprepare for crude arrivals,select unit operating conditions,and schedule product blending tomeet product demand for up tothree months in advance. Volumesand qualities of all products aretracked over time via simplifiedprocess and blender models.

The system is micro-computerbased, and makes extensive use of e

interactive computing, simula-

39

tion. animation and colour gra­phics.

AN_QfERAIIQN_RLANNING_8NQ~QNIRQLS_SXSTEM_lQE_~8I~H

M8NQlA~Ill1UNG

N. StreitHewlett-Packard

Paper No. 123e

H-P's Manufacturing ProductivityNetwork for managing informationin a manufacturing environmentwill be described. The systemconsists of user-customizable.interactive modules for managingmaterials and production planningand control. A major objectiveis to facilitate access toinventory and production datafrom order entry through productshipping.

EQllAIIQN_QEIENIED_SX~IEM_lQE

fRQQllCTIQN_SCHEDQLING_MQQ~LS

P. J. McLellan, G. S. Mueller &G. R. Sullivan

University of Waterloo

Paper No. l23f

The SPEED-UP equation orientedsimulatiom systems are used forsolution of nonlinear refineryprocessing unit models within aproduction scheduling framework.For a wide variety of processunits. it was found that amixture of purely equation-basedand purely procedural-basedapproaches yielded convergencyproperties most suitable forproduction scheduling systems.

S~SSIQN_nl

lE.MU~LLlTX_T~ST_QIL8_lAQLX

DIAGNQSIS_8LGQEITHMQN_8_RILQT_RLbNT

E. Oshima, Tokyo Inst. of Tech.Nagatsuta, Yokohama, H. Matsuyama

Hakozaki, Fukuoka, J. ShiozakiKyusyu University

Hakozaki, Fukuoka,and M. MatsushitaDiacel Chern Ind.

Paper No. 131a

A feasibility test of the faultdiagnosis algorithm based on thesigned digraph was performed on apilot plant. The plant was com­posed of two reactors and theirperipheral equipment. The effec­tivness of adding supplementalmeasured points in reducing thenumber of possible candidates toobtain the source of failure wasinvestiga ted.

P. AndowLoughborough University of Tech.

Paper No. l31b

All modern process plants areequipped with alarm systems. Thealarm system often provides theoperator with the opportunity totake remedial action prior totrip operation. Alarm systemscan be improved by the use ofdiagnostic aids such as "AlarmAnalysis. " Alarm analysis canimprove operator performance andhence avoid plant shutdowns.

A£gLI~AIIQ~_Ql_EX£EEX

SXST~M_I~~H~IQll~S_IQ

fAllLT_l)IAG~QSIS

Ernest J. HenleyUniversity of Houston

Paper No. l31c

In fault-diagnosis applications,IF-THEN production rules organizethe expert knowledge regardingfault-isolation. The classifica-

40

tion of system states can beformulated by:

If [system state i]& (observable fact)

Then system state

The example provided, has 22rUles for cause-isolation. Theinteractive conversation informsthe operator of:

1. The IF-THEN rule applied.2. The prerequisite states.3. A fault-diagnosis.

MQl)ELI~G_Qf_gEQ~~SS_gLANTS

fQE_EIS~_ANl)_E~LIABILIIX

A ShafaghiBattelle Columbus Laboratories

Paper No. l31d

A method is presented forsynthesis of fault trees and aninductive tree for processplants. The method is based ondecomposition of the plant intocontrol loop structures. Failuremodels of the loops are developedand synthesized through theirinteractions. The plant failuremodel is used for risk andreliability analysis.

D. R. WoodsMcMaster University

Paper No. 131e

How to solve trouble shootingproblems is examined particularlyfrom the viewpoint of trainingstudents.