Conceptual Design And Process Synthesis Semantic Scholar-Books Pdf

Conceptual Design and Process Synthesis Semantic Scholar
16 Feb 2020 | 27 views | 0 downloads | 50 Pages | 479.96 KB

Share Pdf : Conceptual Design And Process Synthesis Semantic Scholar

Download and Preview : Conceptual Design And Process Synthesis Semantic Scholar


Report CopyRight/DMCA Form For : Conceptual Design And Process Synthesis Semantic Scholar



Transcription

154 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION, inputs to and outputs from the process were fixed and the goal was to find flowsheets that. could transform the inputs into the outputs Preliminary analysis steps often based on short. cut calculations accompanied the synthesis steps to help make choices among alternative. flowsheet structures CACHE recognized the potential of this research and sponsored the first. Symposium in the emerging field of computer aided chemical process synthesis in 1972 Sead. Early Process Synthesis Explorations, Rudd 1968 suggested perhaps the first organized framework for design based on the no. tion of systems decomposition He suggested that for any original design problem a number of. simpler specific subproblems should be identified the solutions to which when put together. would plausibly solve the original problem Typical chemical process design subproblems. might include raw material receiving and storage feed conditioning reaction product isola. tion purification storage and shipping and so on Some sort of systems decomposition is a. common conceptual design practice, King and coworkers 1972 proposed an alternative evolutionary modification synthesis. approach patterned on another industrial conceptual design practice This approach starts with. an existing flowsheet for the desired product or a similar product and then identifies aspects of. the design that could be changed to greatest advantage to better meet the objectives of the spe. cific case at hand as well as alternatives to implement that change Various heuristic and ana. lytical methods were proposed to identify evolutionary opportunities but the success of the. scheme depended strongly on the initial flowsheet, As a third approach Ichikawa and coworkers 1972 viewed process synthesis as an opti. mization over process structure and approached conceptual design through the application of. operations research and mathematical programming techniques This analysis dominated ap. proach starts with a larger superflowsheet which contains embedded within it many redundant. alternatives and interconnections among them and then systematically strips the less desirable. parts of the superstructure away This superstructure optimization offers the promise of simul. taneous optimization of structural as well as other design parameters However it requires a. starting superstructure from somewhere which for some simple problems however may be. implicit in the formulation as well as very extensive computational capability since the super. structure optimization problem is in general nonlinear nonconvex and involves both continu. ous and discrete variables, Siirola 1971 suggested an alternative decomposition scheme based on an hierarchical or.
dering of physical properties In the resulting systematic generation scheme property differ. ences between raw materials and desired products were detected and resolved using the means. ends analysis paradigm The result was the application of chemical technologies in a sequence. such that the raw materials become transformed into the desired products The property hier. archy was molecular identity first followed by species amount concentration phase temper. ature pressure and possibly other form properties Species identity differences were attacked. first and resolved by the application of reaction technologies Species amount differences were. identified next and resolved by splitting or mixing or purchase Concentration differences were. identified next and generally resolved by separation technologies Finally phase temperature. Conceptual Design and Process Synthesis 155, and pressure differences were detected and resolved by a variety of enthalpy changing technol. ogies If a difference resolving technology were identified but could not be directly applied to. a given stream for example reaction that could not be applied because feeds were not pure. enough or a separation technology could not be applied because the feed stream was the wrong. phase recursive design subproblems were identified with the subgoal of producing stream. conditions suitable for the application of the suspended technology generally but not always. by altering properties lower in the hierarchy After all identified property differences were re. solved the property changing tasks were integrated into actual processing equipment The pro. cedure may be repeated if desired at increasing levels of detail making use of screening results. obtained at lower levels of detail, Alternative solutions are generated when more than one technology is identified that can. be applied to reduce or eliminate a property difference The selection of which technology to. choose might be made on the basis of rules or some evaluation at the time the technologies are. being examined Alternatively each may be chosen separately and the consequences followed. separately leading to alternative design solutions and then each final solution is evaluated. One additional possibility is that all feasible alternative technologies are selected and applied. in parallel leading to a redundant design or superstructure At the end of the design process the. superstructure is reevaluated in its entirety and the less economical redundant portions elimi. The hierarchical decomposition paradigm became the basis of a series of flowsheet syn. thesis programs including AIDES Siirola 1970 BALTAZAR Mahalec and Motard 1977. and PIP Kirkwood Locke and Douglas 1988 as well as the Pinch Technology Onion Lin. nhoff and Ahmad 1983 It has also formed the basis for the present classification of the major. process synthesis subproblems Reaction Path Synthesis Species Allocation or Input Output. and Recycle Structure Separation Scheme Synthesis and Heat and Power Integration. In 1990 the PIP Process Invention Procedure flowsheet synthesis software became. available through CACHE This software follows closely the hierarchical design procedure de. scribed by Douglas 1988 which involves alternating rule based synthesis and short cut anal. ysis and economic evaluation which enables rough estimates of the optimum design The. method generates for each reaction step in a process an input output and recycle structure a. vapor recovery system for the reactor effluent a liquid separation system for the reactor efflu. ent and a heat recovery network Currently an expert system implementation of PIP is under. development,Heat Exchanger Network Synthesis, The pairing of complementary enthalpy increasing tasks with enthalpy decreasing tasks. so as to optimally recover and reuse energy within a process while decreasing dependence on. external utilities is known as heat integration The synthesis of heat exchanger networks is the. process synthesis subproblem that has received perhaps the most attention This synthesis prob. lem generally occurs late in the synthesis hierarchy after the various hot and cold streams have. been fairly well defined with respect to amount composition available temperature desired. temperature and other physical properties Also the tradeoffs between capital and operating. cost are well understood,156 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION. It turns out that the maximum amount of heat integration and the minimum heating and. cooling utility requirements can be determined for a process without actually designing a heat. integration network through a technique called Targeting Targeting procedures also exist for. estimating the optimal number of exchangers and the total heat transfer area These estimates. can also be made in the light of constraints requiring or forbidding energy matches between. specific pairs of streams CACHE has made available for educational use one such program for. calculating these estimates TARGET II developed by Linnhoff March Inc. Linnhoff and Hindmarsh 1983 popularized important representations including Com. posite Curves and Grand Composite Curves for visualizing the thermodynamics governing the. heat exchanger network synthesis problem They also introduced the concept of Pinch Tech. nology based on the point in the design where heat is transferred at the minimum approach. temperature which is especially useful in constraining the selection of energy matches which. preserve overall minimum utility usage Pinch technology also provides guidance on splitting. streams for parallel heat exchangers minimizing the total number of exchangers proper incor. poration of heat pumps and integrating heat and power tasks Within these constraints proce. dures for the actual synthesis of heat exchanger networks are somewhat less well developed. but all of the standard process synthesis approaches including systematic generation evolu. tionary modification and superstructure optimization using all the standard tools of heuristics. expert systems deterministic algorithms and mathematical programming have been suggest. ed THEN a heat exchanger matching algorithm developed by Carl Knopf is available through. CACHE The straightforward application of heat integration technology to a typical process. design generally reduces the net present cost on the order of 10 due largely to sensible heat. As mentioned heat exchanger networks have also been a testbed for superstructure opti. mization approaches to process synthesis In this case techniques from operations research. have been used to essentially convert the synthesis problem into one of analysis that can be. solved by mathematical programming CACHE s sixth Process Design Case Study Chemical. Engineering Optimization Models with GAMS contains three heat exchanger network synthe. sis examples including a linear programming transshipment model for predicting minimum. utility consumption with stream matching constraints a nonlinear programming model for glo. bal optimum search in the synthesis of heat exchanger networks and a mixed integer nonlinear. programming model for simultaneous synthesis and optimization of heat exchanger networks. Heat Integrated Distillation System Synthesis, Synthesis methods for separation tasks depend in part on the nature of the system to be.
separated In the case of distillative separations of relatively ideal systems simple volatility or. dering may be sufficient to represent the solution thermodynamics Alternative separation. trains can be synthesized using list processing and either heuristic rules Seader and Wester. berg 1977 or dynamic programming techniques Hendry and Hughes 1972 If the number. of components is not too large all possible separation orderings may be generated and evalu. ated as is implemented in the PIP software, Distillation schemes involve a fair number of auxiliary enthalpy changing tasks including. reboilers condensers feed quality adjustment and product coolers Such distillation schemes. Conceptual Design and Process Synthesis 157, can be heat integrated with any of the standard techniques However since the temperatures of. these enthalpy changing tasks depend upon the pressures purity specifications and other op. erating conditions of each column in a distillation sequence it is advantageous to simulta. neously perform heat integration within the column pressure optimization Siirola 1981. Utility requirement reductions in the neighborhood of 50 and net present cost savings of 35. are typical for such simultaneous distillation sequence synthesis operating parameter optimi. zation and heat integration CACHE s sixth Process Design Case Study also includes an ex. ample of a GAMS based MINLP approach for optimal sequencing and heat integration in the. synthesis of sharp separation sequences, For binary separations or processes that contain a single or a particularly dominant distil. lation significant energy recovery by the previous approach may not be possible However a. single distillative separation may be implemented as two columns in such a way that the latent. heat rejected from one might be recycled to the other approximately halving the net utility re. quirement In order for this to be accomplished the condenser of one column must be at a high. er temperature than the reboiler of the other This may be done by operating the columns at. different pressures possibly assisted by the degree of separation designed for each column. Blakely 1984 compared a number of such multiple effect designs against a single column as. a function of feed flowrate feed composition and relative volatility The specific results de. pend on the costs and temperatures of available utilities as well as on the capital cost correla. tions In fact at low flowrates the economics of the single column design are generally superior. to all of the multiple effect designs largely because at those scales the capital costs of the col. umns dominate the energy costs However at higher flowrates for all conditions of feed com. position and relative volatility there exists one or more multiple effect designs with lower net. present costs by as much as 35 similar in magnitude to the savings achievable in the heat. integration of other multiple distillation separations. Nonideal Distillation Scheme Synthesis, For systems which exhibit more nonideal behavior involving azeotropes and regions of. immiscibility a more detailed representation of the thermodynamics may be necessary even. for the synthesis of separation schemes at the targeting level of detail One such representation. is the Residue Curve Map developed by Doherty and coworkers 1985 Residue curve maps. are derived from an analysis of a single stage batch still but may be applied to the understand. ing of the thermodynamic behavior of solutions and the behavior of continuous distillative sep. aration operations on those systems Residue curve maps define regions of feed compositions. from which it is possible to reach the same lowest boiling composition and the same highest. boiling composition by alternative operation of a single simple distillative separation task. Some systems involving azeotropes exhibit more than one such region separated by boundaries. which normally can not be crossed by simple distillation However techniques have been de. vised for overcoming these distillation boundaries including pressure shifting exploiting. boundary curvature and exploiting separation technologies not involving volatility such as liq. uid liquid decantation solid liquid crystallization and kinetic phenomena such as membrane. diffusion The preparation of residue curve maps as well as the design of distillative separations. consistent with these maps is being implemented in the MAYFLOWER software under devel. 158 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION,opment at the University of Massachusetts.
An industrial separation synthesis procedure for systems exhibiting nonideal thermody. namics has been described in the recent CACHE sponsored conference Foundations of Com. puter Aided Process Design Siirola 1994 Key features of the method include the Residue. Curve Map representation of solution thermodynamics overlaid with pinched regions and liq. uid liquid or solid liquid equilibria data identification of thermodynamic Critical Features. to be avoided pinched regions overcome distillation boundaries or exploited for example. liquid liquid tie lines a Strategic Hierarchy to address issues raised by the critical features. first the identification of Interesting Compositions useful as mass separation agents which. must be regenerated and recycled within the process Opportunistic Resolution of remaining. concentration property differences by mixing decant or distillative separation as permitted by. distillation boundary constraints and pursuit of Multiple Alternatives heaviest underflow as. well as lightest overhead for every multicomponent distillative separation The method also. includes methods for pressure shifting distillation boundaries exploiting boundary curvature. producing compositions which are neither the highest boiling nor the lowest boiling in a re. gion extractive distillation handling recycle removing redundancy and heat integration A. similar rule based system called SPLIT Wahnschafft Jurain and Westerberg 1991 is cur. rently being incorporated into the ADVENT software by Aspen Technology Inc CACHE is. working to make methods and tools such as these and other aspects of Residue Curve Map tech. nology available for educational use,Outlook for the Future. Many of the methods and tools being developed in the process synthesis research commu. nity are still more concerned with analysis targeting representation operability etc than. with alternative generation More effort will be directed specifically to alternative generation. for conceptual design at all levels of detail, Many of the applications discussed here were based on algorithmic systematic generation. approaches or heuristic rule based or expert systems based approaches to process synthesis Be. cause of combinatorial difficulties neither the algorithmic nor the heuristic methods carry any. guarantees of structural optimality, Superstructure approaches on the other hand while not especially practical at the present. time do offer with the promise of structural optimality and have tremendous potential for the. future As mentioned several glimpses of how such approaches might be formulated are in. cluded in CACHE s sixth Process Design Case Study There are significant challenges remain. ing to improve computational efficiency to overcome combinatorial difficulties and to develop. global optimization strategies for the generally nonconvex mixed integer nonlinear problems. involved Steady progress is being made on these fronts and other process synthesis problems. such as reaction path and reactor scheme synthesis design for waste minimization and consid. eration of other important design criteria such as safety operability and controllability. CACHE will continue its efforts to make advances in all of these process synthesis methods. and tools available to the educational and industrial communities. Conceptual Design and Process Synthesis 159,Conclusions. Systematic approaches to conceptual design and process synthesis have begun to have. measurable industrial success Although there are many restrictions these techniques have led. to higher value lower energy lower environmental impact and sometimes even novel design. alternatives Process synthesis techniques have been successfully applied to the conceptual de. sign of total process flowsheets as well as to specific design subproblems including heat inte. gration heat integrated distillation trains multiple effect distillation and the separation of. azeotropic systems Typical energy savings of 50 and net present cost reductions of 35. have been achieved using systematic process synthesis methodologies. Certain features of systematic approaches to process synthesis appear to have special mer. it These include architectures which are hierarchical in scope and iterative in level of detail. targeting as a useful bounding exercise means ends analysis as a recursive problem solving. paradigm with hierarchical as well as opportunistic goals implemented by an iterative formu. lation synthesis analysis evaluation design strategy thinking specifically in terms of tasks to. be accomplished before equipment to be employed the importance of representations which. encapsulate analysis and the advantages of solving related synthesis problems simultaneously. Advances in problem formulation and in computational hardware and software capability. offer the promise of a new generation of practical process synthesis techniques based directly. on structural optimization Even greater benefits are expected to be realized as this next gener. ation of approaches are transferred to industry and incorporated into design education Soon. the goal of synthesizing provably superior conceptual process flowsheets may be at hand. References, Blakely D M 1984 Cost Savings in Binary Distillation through Two Column Designs M S dissertation Clem.
son University Clemson, Doherty M F and G A Caldarola 1985 Ind Eng Chem Fundam 24 474. Douglas J M 1988 Conceptual Design of Chemical Processes McGraw Hill New York. Foucher E R M F Doherty and M F Malone 1991 Automatic Screening of Entrainers for Homogeneous. Azeotropic Distillation Ind Eng Chem Res 30 760, Hendry J E and R R Hughes 1972 Generating Separation Process Flowsheets Chem Eng Progr 68 6 69. King C J D W Gantz and F J Barnes 1972 Systematic Evolutionary Process Synthesis Ind Eng Chem Pro. cess Design Develop 11 271, Kirkwood R L M H Locke and J M Douglas 1988 A Prototype Expert System for Synthesizing Chemical. Process Flowsheets Comput Chem Engng 12 329, Linnhoff B and S Ahmad 1983 Towards Total Process Synthesis AIChE Meeting paper 26d Washington. Linnhoff B and E Hindmarsh 1983 The Pinch Design Method of Heat Exchanger Networks Chem Eng Sci. Mahalec V and R L Motard 1977 Procedures for the Initial Design of Chemical Processing Systems Comput. Chem Engng 1 57, Morari M and I E Grossmann Eds 1991 Chemical Engineering Optimization Models with GAMS CACHE.
Design Case Studies Volume 6 CACHE Austin, Rudd D F 1968 The Synthesis of Process Designs I Elementary Decomposition Principle AIChE J 14 342. 160 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION, Seader J D 1972 Computer Aided Chemical Process Synthesis CACHE Sponsored Symposium CACHE. Seader J D and A W Westerberg 1977 A Combined Heuristic and Evolutionary Strategy for the Synthesis of. Simple Separation Sequences AIChE J 23 951, Siirola J J 1970 The Computer Aided Synthesis of Chemical Process Designs Ph D dissertation University. of Wisconsin Madison, Siirola J J and D F Rudd 1971 Computer Aided Synthesis of Chemical Process Designs Ind Eng Chem. Fundam 10 353, Siirola J J 1981 Energy Integration in Separation Processes In R S H Mah and W D Seider Eds Founda.
tions of Computer Aided Process Design Vol II Engineering Foundation New York 573. Siirola J J 1994 An Industrial Perspective of Process Synthesis In L T Biegler and M F Doherty Eds. Foundations of Computer Aided Design AIChE Symposium Series Vol 91 No 304. Umeda T A A Hirai and A Ichikawa 1972 Synthesis of Optimal Processing System by an Integrated Ap. proach Chem Eng Sci 27 795, Wahnschafft O M T P Jurain and A W Westerberg 1991 SPLIT A Separation Process Designer Comput. Chem Engng 15 656,PROCESS SIMULATION,Lorenz T Biegler. Carnegie Mellon University,Pittsburgh PA 15213,J D Seader. University of Utah,Salt Lake City UT 84124,Warren D Seider. University of Pennsylvania,Philadelphia PA 19104, This brief discussion of process simulation is written from a standpoint of the impact.
of the contributions of CACHE on the educational community Emphasis is on steady. state process simulation and optimization with FLOWTRAN Some suggestions are. made for needed future developments,Introduction and Historical Development. One of the essential early steps in process design or analysis following the synthesis of a. process flowsheet and the establishment of a preliminary set of operating conditions suitable. for a base case study is the computation of material and energy balances so as to determine. 1 for every major stream in the process the flow rate composition temperature pressure and. phase condition and 2 for every major piece of equipment the overall size and the shaft work. or heat transfer requirements Before the availability of digital computers these calculations. were done manually with a slide rule or mechanical desk calculator If the process involved re. cycle streams multistaged multicomponent separation operations and or reactors with multi. ple reactions the manual computations could be extremely tedious and time consuming Often. the computations would be done by approximate methods and incomplete closure of material. balances was sometimes acceptable If optimization studies were made they were usually re. stricted to individual operations using one or two design variables because a more global ap. proach was not practical, Two significant events occurred in the mid 1950s that were to drastically alter the ap. proach to process simulation and design The first was the introduction of the IBM 704 digital. computer with its built in floating point arithmetic The second was the development of the. easy to use high level procedure based programming language of J W Backus and others. 162 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION, from IBM called FORTRAN The IBM 704 came with a FORTRAN language compiler and. subroutines written in that language were automatically handled Almost overnight chemical. engineers in several large companies and in universities began writing large scale FORTRAN. computer programs to replace the manual methods At first these programs were called com. puter aided steady state chemical process design programs Later they would be referred to. as simply simulation or flowsheeting programs The ability to replace tedious hand calculations. by computerized calculations also spurred tremendous development in the areas of thermody. namic properties of liquids and gases rigorous calculation of equilibrium stage operations and. design of nonisothermal multiple reaction reactors including effects of transport phenomena. At the same time it was quickly realized that the digital computer was not limited to linear sys. tems Consequently there was an explosion in the development of numerical methods for solv. ing all sorts of nonlinear as well as linear problems. By 1967 six commercial and two academic steady state simulation programs all written. in FORTRAN were in publicized use Undoubtedly other unpublicized commercial simula. tors were also in use One of the widely publicized academic programs PACER Process As. sembly Case Evaluator Routine developed by Professor Paul T Shannon at Purdue. University as reported in 1963 was made available to chemical engineering departments in. 1967 following an instructional PACER workshop at Dartmouth College April 23 28 1967. which was attended by 20 chemical engineering professors from the United States and Canada. In 1971 the author of PACER together with Professors Crowe Hamielec Hoffman Johnson. and Woods from McMaster University published the first book on simulation entitled Chem. ical Plant Simulation An Introduction to Computer Aided Steady State Process Analysis. which provided the necessary background to use PACER Crowe et al 1971. The academic version of PACER was mainly a modular executive program for directing. material and energy calculations for flowsheets which could include recycle streams PACER. had a limited library of unit operation models and lacked a physical property estimation pack. age However it was open ended so that users could add operation and property models In. 1968 Professors R L Motard and Ernest J Henley at a workshop at the University of Houston. that was attended by 42 professors introduced a more advanced simulation program called. CHESS Chemical Engineering Simulation System Like PACER CHESS was based on a. sequential modular architecture in which models of the operating units were solved one at a. time For each unit the inlet stream conditions were known or assumed and the outlet stream. conditions and heat and work transfer rates were computed by the operation model If the pro. cess did not involve recycle loops the unit calculations began at the feed end of the process and. progressed to the product end so that the calculated outlet stream s from one unit became the. inlet stream s to the next unit s If the process included one or more recycle loops the user. could provide a guess for the conditions of one stream in each loop and the program would au. tomatically iterate each loop to convergence to some specified tolerance The CHESS program. included an expanded library of unit operation models and a physical property package mak. ing it easy to use However it lacked a general rigorous multistage multicomponent separa. tion unit for modeling distillation type operations. By the early 1970s a number of new and mostly commercial steady state simulation pro. grams that contained rigorous absorption stripping distillation and liquid liquid extraction. models began to emerge The CACHE Computer Aids for Chemical Engineering Education. Process Simulation 163, Committee which had been formed in 1969 directed its Large Scale Systems Task Force. chaired by Professor J D Seader and the Program Distribution Task Force chaired by Profes. sor Warren D Seider to conduct a survey of academic and commercial simulation programs. that might be made available at a reasonable cost for use by chemical engineering faculty and. students That survey identified a number of such programs and these two Task Forces then. evaluated the programs with a comprehensive test problem that determined their ease of use. and robustness The FLOWTRAN program of Monsanto Company emerged as the preferred. FLOWTRAN was conceived in 1961 by the Applied Mathematics Department of Mon. santo Company when they like many large chemical processing companies began contemplat. ing the development of a generalized modeling computer program for process design and. simulation The project gained considerable impetus late in 1964 when Robert Bob H Cavett. a brother of the entertainer Dick Cavett left Pure Oil Company and joined the Applied Math. ematics Department of Monsanto Company While at Pure Oil Bob Cavett had developed a. physical property package for vapor liquid systems and had investigated numerical analysis. procedures for converging nonlinear systems of equations and recycle loops Monsanto pur. chased from Pure Oil all rights to Cavett s work By 1965 the name FLOWTRAN had been. selected for the name of Monsanto s program which was to be based on the sequential modular. architecture In 1966 the initial version of the program was available for use by chemical en. gineers at Monsanto From 1969 to 1973 more than 70 outside companies used FLOWTRAN. through Monsanto Enviro Chem Systems Inc via commercial computer networks However. this type of outside use was terminated by Monsanto in 1974 in favor of licensing the program. On June 13 1973 in a letter to Mr John W Hanley President of Monsanto Company. CACHE requested that Monsanto consider allowing chemical engineering faculty and students. to use FLOWTRAN in course work and research With the assistance of Professor John J. McKetta of the Advisory Committee of CACHE and F E Reese M C Throdahl J R Fair. and S I Proctor of Monsanto approval of the use of FLOWTRAN via a national computer. network was granted by Monsanto Company on December 10 1973 This approval included. assistance in implementation of the system by providing a grant and loaning specialists from. the Monsanto Corporate Engineering Department Starting on May 10 1974 Dr Allen C. Pauls of Monsanto directed a three day training course on FLOWTRAN which was attended. by Richard R Hughes H Peter Hutchison J D Seader Warren D Seider and Arthur W. Westerberg On June 5 1974 at a meeting of the CACHE trustees in Seven Springs Pennsyl. vania Dr Proctor of Monsanto formally presented to CACHE the Monsanto grant and condi. tions for making FLOWTRAN available to universities. On July 1 1974 with significant contributions from Pauls and Hughes Seader and Seider. completed the draft of a student textbook entitled FLOWTRAN Simulation An Introduction. That book included material on the principles of simulation detailed instructions on the appli. cation of FLOWTRAN to a wide variety of problems including a comprehensive design study. by Hughes numerous examples and homework problems and descriptions of all of the equip. ment unit and physical property models By August 1974 this book was published by. CACHE in a soft bound edition and stocked for sale by Ulrich s Bookstore in Ann Arbor. 164 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION, The academic version of FLOWTRAN was almost identical to the commercial version.
that was used by Monsanto and licensed to other companies The program consisted of 9 exec. utive routines tailored to the particular computer on which the program was to run 7 main. FORTRAN programs 2 data lists 86 FORTRAN subroutines and input data and output list. ings for 27 test problems that illustrated most of the features and capabilities of FLOWTRAN. The program was the result of 60 man years of effort at a cost to Monsanto of more than two. million dollars At the time of the approval for university use Monsanto engineers were run. ning about 700 FLOWTRAN jobs per month The application of FLOWTRAN to design and. simulation was restricted to vapor liquid systems The library of operations models called. blocks included 5 flash blocks 5 distillation blocks 2 absorber stripper extraction blocks 8. heat exchanger blocks 3 reactor blocks 2 pump and compressor blocks 5 stream mixing di. viding manipulation blocks 4 control blocks 1 recycle convergence block 16 equipment cost. blocks 4 financial analysis blocks and 6 report blocks Included in two of the flash blocks was. the first algorithm for calculating a vapor liquid liquid three phase flash which was de. scribed in the 1969 textbook Material and Energy Balance Computations by Professor E J. Henley and Dr E M Rosen of Monsanto Company The control blocks allowed the user to. manipulate certain equipment unit parameters to achieve desired values of stream variables as. sociated with that unit or at another upstream unit. With respect to thermodynamic properties the academic version of FLOWTRAN includ. ed a property constant data bank for 180 chemical species Mixture property models were pro. vided to estimate density enthalpy and K values using the gamma phi formulation The. Redlich Kwong equation of state was used for non ideality in the vapor phase and the equa. tions of Scatchard Hildebrand regular solutions van Laar Wilson and Renon Prausnitz. NRTL were available for non ideality in the liquid phase The latter three equations required. binary interaction parameters that had to be supplied by the user A special program called. VLE included in the FLOWTRAN system could be used to compute these parameters from. experimental phase equilibria data For species not in the data bank the user could add their. property constants to a private file If the constants were not available they could be estimated. with a special program called PROPTY which was also included in the FLOWTRAN system. When the FLOWTRAN program was being conceived considerable effort was directed. to the choice of architecture With respect to the handling of recycle streams both the sequen. tial modular and simultaneous modular two tier methods were debated Although the latter. was preferable for processes with a large number of recycle loops the former was selected be. cause it could be more readily understood and appeared to be more efficient on a number of. test problems A very flexible recycle convergence block was written which could converge. up to three recycle loops simultaneously and provided a number of options to the user includ. ing the methods of successive substitution bounded Wegstein delayed Wegstein and damped. Wegstein The user was required to specify the location of the convergence block in the recycle. loop s A novel feature of the FLOWTRAN architecture was its variable internal structure. User input consisted of both data and commands in the form of ordered keywords pertaining. to selection of chemical species and thermodynamic property models and equipment unit. blocks Also the user could insert FORTRAN statements among the block commands The in. put was parsed into data and commands The data were stored and the commands were trans. lated into a main FORTRAN program consisting mainly of block calls which was then. Process Simulation 165, compiled and linked to just those subroutines that were needed for that particular simulation. This type of architecture made FLOWTRAN efficient in mass storage requirements and open. ended so that user supplied equipment unit subroutines could be readily added. On August 13 1974 Hughes Seader and Seider with Dr Pauls and Professor Richard. S H Mah held at Northwestern University the first of three three day FLOWTRAN work. shops sponsored by the AIChE The workshop was attended by 38 faculty members represent. ing 35 different universities in the United States and Canada One attendee remarked that. FLOWTRAN was destined to alter the way process design was being taught to senior students. in chemical engineering Today 20 years later that prediction has come true most if not all. chemical engineering departments now teach computer aided process design with any of a. number of available steady state process simulators. The 1974 edition of FLOWTRAN Simulation An Introduction was followed by a second. edition in February 1977 which added a few subroutines and a new chapter on writing cost. blocks A major revision was published in May 1987 which included a new SQP optimization. capability a new Broyden based convergence block and a new chapter on optimization all by. Professor L T Biegler as discussed in the section on optimization below In January 1975. CACHE published the booklet CACHE Use of FLOWTRAN on UCS by Hughes This booklet. which was revised in January 1982 with the assistance of Biegler provided all the instructions. necessary to access the FLOWTRAN program over the United Computing Systems UCS net. work with computers located in Kansas City Missouri In December 1977 CACHE pub. lished Exercises in Process Simulation using FLOWTRAN edited by Dr J Peter Clark This. book which was revised in December 1980 with the assistance of Dr T P Koehler and Pro. fessor Jude T Sommerfeld contained problem statements and solutions for 8 short exercises. 14 workshop exercises and 5 more comprehensive problems. By mid 1975 one year after the publication of the FLOWTRAN textbook and the com. pletion of the faculty workshop at Northwestern University 25 universities were using. FLOWTRAN During the period from 1974 to 1983 an average of 21 universities year were. using FLOWTRAN on the UCS network By 1985 59 universities in the United States and. Canada had used it This represented about one third of all the departments of chemical engi. Major deterrents to a wider use of FLOWTRAN were 1 the inconvenience of having to. submit card decks of input data in a batch mode to a remote batch machine 2 the turn around. time for a run which was typically one to two days from the time the student submitted a card. deck to the time the student received a print out and 3 the cost of accessing the program on. the UCS network Typically a small to moderate size run would cost from 2 00 to 10 00 If. access was from a remote job entry device the cost would be more Rarely did a student pre. pare an error free input on the first or even the second try Therefore it might take a week to. get a successful run On the average each university using FLOWTRAN was spending 1 500. In response to a request from CACHE Monsanto Company announced on August 19. 1982 that load modules of FLOWTRAN on magnetic media would be made available to uni. versities worldwide through CACHE for installation on in house computers The use of. FLOWTRAN on the UCS network would be discontinued on January 15 1986 By 1993. 166 COMPUTERS IN CHEMICAL ENGINEERING EDUCATION, FLOWTRAN load modules had been prepared for 14 different computer and operating system. combinations 190 141 USA 11 Canada and 38 for 21 other foreign countries load modules. had been distributed worldwide and approximately 15 000 copies of the three editions of. FLOWTRAN Simulation An Introduction had been sold,Current Status of Steady State Simulation. By the late 1980s as discussed by Biegler 1989 computer aided chemical process sim. ulation had become virtually indispensable in process design and analysis A large number of. commercial simulators had become readily available and some of them notably ASPEN PLUS. of Aspen Technology Inc in Cambridge Massachusetts ChemCAD of Chem Stations Inc. in Houston Texas DESIGN II of ChemShare Corp in Houston Texas HYSIM of Hyprotech. Ltd in Calgary Alberta Canada and PRO II of Simulation Sciences Inc in Brea California. were being licensed at low cost to universities Compared to FLOWTRAN some most or all. of the latest versions of these simulators 1 can be installed on a wider variety of computer. operating systems including PCs with sufficient main and mass memory running under MS. or PC DOS or one or more versions of Windows 2 have much larger data banks with more. than 1000 chemical species 3 have a much larger library of thermo dynamic property mod. els including the ability to estimate K values from equations of state 4 have physical prop. erty models for entropy transport properties and other properties used to characterize. materials 5 have built in tables for binary interaction parameters of activity coefficient cor. relations 6 allow the user to specify any system of input and output units 7 can generate. graphs 8 can handle solids 9 have improved algorithms for converging multicomponent. multistage separations 10 have models for interlinked three phase and reactive distillation. 11 have models for kinetically controlled reactors and 12 can compute simultaneous chem. ical and phase equilibrium by minimization of free energy Thus on April 4 1994 Monsanto. Company announced that it was their desire to bring to an orderly conclusion the CACHE. FLOWTRAN project Although FLOWTRAN can continue to be used for educational purpos. es on existing computers no more load modules will be prepared or distributed In 1993 at the. AIChE Annual Meeting in St Louis the CACHE Corporation presented an award to Monsanto. Company in recognition of its significant contribution to chemical engineering education. FLOWTRAN ushered in a major change in the way process design is taught to chemical engi. neers More than any other individual Robert H Cavett of Monsanto was responsible for the. successful development of sequential modular process simulation which he implemented in. FLOWTRAN For this achievement in 1987 he was the first recipient of the Computing Prac. tice Award of the CAST Division of AIChE,Optimization in FLOWTRAN. With the availability of FLOWTRAN from Monsanto in 1974 there was early interest by. the late Professor Richard R Hughes to also demonstrate and implement an optimization ca. pability with process simulation and design Professor Hughes had directed the development. of the CHEOPS optimization package for Shell Development s process simulator and a similar. capability was envisioned in the mid 1970s for FLOWTRAN Fundamental work by Wilson. and Beale in the mid 1960s led to Quadratic Approximation Programming QAP which was. prototyped by Hughes and Isaacson in 1975 With QAP reduced quadratic models for the ob. Process Simulation 167, jective function and constraints were constructed from the rigorous flowsheet models and.
these reduced models were optimized to suggest a new base point As a sequel to this work A. L Parker and Dick Hughes proposed an implementation of this optimization strategy to Mon. santo and a working version of QAP was demonstrated with an ammonia synthesis project us. ing Monsanto s version of FLOWTRAN Using about 65 simulation time equivalents on the. ammonia process this version took advantage of FLOWTRAN s data structures as well as the. existing keyword interface Still this was the only problem solved by QAP on FLOWTRAN. Parker and Hughes 1981, Because the CACHE version of FLOWTRAN was available only through the UCS net. work it took some time to implement Parker s approach for that version As part of this exer. cise a similarity between QAP and the evolving Successive Quadratic Programming SQP. algorithm was noted and this led to a simplified optimization algorithm which was termed. Quadratic Linear Approximation Programming Q LAP Biegler and Hughes 1981 This ap. proach was actually an SQP algorithm in disguise and improved the performance fourfold over. the QAP strategy However developing an optimization approach proved expensive and clum. sy on the UCS network and further development was done with a simpler prototype simulator. called SPAD at the University of Wisconsin, The QAP Q LAP and virtually all previous flowsheet optimization approaches shared. the same disadvantages First they required full flowsheet evaluations for objective and con. straint function evaluations This was expensive and also potentially unreliable difficult inter. mediate points can lead to convergence failures Second these optimization approaches were. implemented so that they needed to control the information flow in the overall process calcu. lation This led to a number of difficulties that deterred implementation of optimization strate. gies on existing process simulators, However the application of the SQP algorithm overcame both of these draw backs This. algorithm can be interpreted both as the optimization of a quadratic model at each basepoint. iteration as well as a quasi Newton method applied to the optimality conditions of the process. optimization problem This second interpretation leads to the insight that both equality and in. equality constraints could be converged simultaneously with a systematic improvement of the. objective function In particular convergence of tear constraints which often represented the. least efficient part of a flowsheet simulation could be incorporated into the formulation of the. optimization problem and handled directly by the SQP algorithm This insight further leads to. a simplified implementation of an optimization algorithm to existing process simulators Vir. tually all process simulators and most prominently FLOWTRAN have convergence blocks. that appear as artificial units in the flowsheet calculation sequence Upon convergence a message. is passed to the executive program of the simulator that signals the next calculation phase This. leads to the inclusion of sophisticated convergence algorithms largely independent of the archi. tecture of the simulator and also allows the simple addition of an optimization capability it is. simply an extended convergence block Moreover this flexibility appears to have been antici. pated by early SQP codes as well e g VF02AD in the Harwell library These were structured. through reverse communication and thus allowed the information flow to be controlled by the. user or a comprehensive modeling package It appears that both recycle convergence blocks. and the SQP code were made for each other Not surprisingly the insights of linking SQP with. process simulators was also discovered independently at about the same time at the University.


Related Books

ENGLISH AS A SECOND LANGUAGE (SPEAKING ENDORSEMENT)

ENGLISH AS A SECOND LANGUAGE SPEAKING ENDORSEMENT

Cambridge International General Certificate of Secondary Education 0510 English as a Second Language (Speaking Endorsement) June 2015 Principal Examiner Report for Teachers

Water Quality Report

Water Quality Report

Where Does My Water Come From? C ity of Clearwater residents use approximately 11.15 million gallons of potable water every day. Approximately 40 percent is ...

Chapter 1 - Russell Sage Foundation

Chapter 1 Russell Sage Foundation

In chapter 9, we present a summary of the major findings of the study and their implica-tions for the public debate over educational opportunity and persistence of inequality. Why Education Matters One of the dominant images of American history is the little red school-house. The one-room building with students of all ages was a symbol of

LNCS 6972 - Designing and Evaluating Casual Health Games ...

LNCS 6972 Designing and Evaluating Casual Health Games

Designing and Evaluating Casual Health Games for Children and Teenagers 199 This issue may be addressed by introducing casual game design principles to health

John Deere 4.5L, 8.1L & 9.0L Tier III Denso Common Rail ...

John Deere 4 5L 8 1L amp 9 0L Tier III Denso Common Rail

1) From the tractor cab looking forward, locate the fuel system on the left side of the engine. You will see the common rail with the injector lines coming out of it. 2) Locate Fuel Pressure sensor as shown in Pictures below 4.5L, 6.8L, & 8.1L John Deere Fuel Pressure Sensor

discover - Mermaids is a PADI 5 Star Scuba Diving Center ...

discover Mermaids is a PADI 5 Star Scuba Diving Center

Navigator Underwater Photographer Underwater Search and Recovery Diver Specialties AWARE - Coral Reef Conservation Project AWARE Specialist Emergency Oxygen Provider Digital Underwater Photographer (for S norkelers) Specialties Cavern Diver Ice Diver Rev 7 8 Scuba Review. PADI Medical Questionnaire Scuba diving is an exciting and demanding activity. To scuba dive safely, you must not be ...

Fundamentals of Puzzle and Casual Game Design

Fundamentals of Puzzle and Casual Game Design

About the Author Ernest Adams is a game design consultant and part-time professor at the University of Uppsala Campus Gotland in Sweden. He lives in England and holds ...

Analysis of Pile under Lateral Loads Using ALP Module

Analysis of Pile under Lateral Loads Using ALP Module

given in IS 2911 (Part 1/Sec 2) (2010). The pile head is assumed to be fixed connection with the pile cap. IV-WORKING PRINCIPLE OF ALP SOFTWARE Analysis of Laterally Loaded Piles (ALP) module under the OASYS software is used for the prediction of the effects caused due to the lateral loads on the piles in the form of pressures, ground movement in horizontal direction, induced shear force and ...

Monitoring and - INTERTANKO

Monitoring and INTERTANKO

information is recorded in the oil record book, you need not record these data in other recordkeeping documentation. The 2013 VGP requires bilge water analysis of oil and grease sampled after the bilge water treatment system. The 2013 VGP requires that a reading of the OCM be recorded at the time of sample collection so that the

IilTE]tTI]{}(O

IilTE tTI O

r*. 1. FOREWORD FOREWORD The INTERTANKO Safety, Technical and Environmental Committee (ISTEC) has developed this Guide to assist ship operators with the proper and consistent filling in of the Oil Record Book Part I (ORB) and to avoid possible enors in logging information in it, u'hich has been one of the main reasons for the increasing problems of ships being detained and fined in the US and ...