Sie sind auf Seite 1von 12

Computers and Chemical Engineering 26 (2002) 295 306 www.elsevier.

com/locate/compchemeng

Integrated process design instruction


D.R. Lewin a,*, W.D. Seider b, J.D. Seader c
Chemical Engineering Department, Technion, Israel Institute of Technology, Haifa 32000, Israel Chemical Engineering Department, Uni6ersity of Pennsyl6ania, Philadelphia, PA 19104, USA c Chemical and Fuels Engineering Department, Uni6ersity of Utah, Salt Lake City, UT 84112, USA
b a

Received 21 August 2000; received in revised form 2 January 2001; accepted 2 January 2001

Abstract As chemical engineering education moves into the new millennium, it is incumbent on educators to provide a modern curriculum for process design, yet mindful of the limited time for instruction that is available. This paper addresses three key components of a chemical engineering curriculum that prepare undergraduates to be effective process designers in industry: (a) a structured approach relying on fundamentals, integrated with instruction in the competent use of process simulators; (b) a balance between heuristic and algorithmic approaches; and (c) instruction in the integration of design and control. It is argued that these components should be included in an integrated fashion, with much of the material appearing gradually during the delivery of core courses, taking full advantage of computing capability and multimedia support for self-paced instruction. In this paper, each of the features is discussed in detail and demonstrated for the design of a typical process. 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Process design instruction; Heuristic and algorithmic approaches; Chemical process simulators; Interaction of design and control; Multimedia and web-based instruction

1. Introduction Instruction of chemical engineers should reect the challenges they face in industry. Young chemical engineers are required to assimilate rapidly new and emerging technologies to react in a exible manner to shorter production cycles and strict quality regulations. They are expected to improve product quality while at the same time reduce operating costs and environmental impact, improve operability, minimize waste production, and eliminate possible hazards. It is incumbent on chemical engineering educators to provide a modern curriculum for process design instruction that addresses these needs while being mindful of the limited time available. The rst issue involves the concept of a structured core curriculum that focuses on fundamentals as a basis for design. Typically, design is taught in the senior year and involves the integration and assimilation of core
* Corresponding author. Tel.: + 972-4-829-2006; fax: + 972-4-8230476; http://tx.technion.ac.il/ dlewin/pse.htm. E -mail address: dlewin@tx.technion.ac.il (D.R. Lewin).

course material as dictated by the needs of a design project. Section 2 describes how the core course sequence has impact on the needs of instruction in design. Furthermore, we discuss the need for students to support their developing knowledge of engineering fundamentals in general, and more specically their design activity, by mastering the use of a commercial simulator to a high level of competence. We suggest that adopting self-paced methods relying on multimedia tutorials, which assist the students in preparing simulations of process owsheets, can support this effort. In the second issue, which is discussed in Section 3, it is postulated that the teaching of design itself should strike a balance between heuristic and algorithmic approaches. While heuristics lay the foundations for acquiring the experience necessary to carry out practical process creation and equipment design, the importance of the latter is to ensure the generation of optimal designs. The last issue is the importance of dealing with interactions between the design and control of chemical processes when learning to prepare process designs. In Section 4, the current state of the art in the integration of process design and process control is reviewed with

0098-1354/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 9 8 - 1 3 5 4 ( 0 1 ) 0 0 7 4 7 - 5

296

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

particular emphasis on its impact on the education of undergraduates. Several textbooks are available to support a senior course in process design. The traditional textbooks focus on either hierarchical design relying on back-ofthe-envelope calculations (Douglas, 1988), or on detailed equipment design, costing, and economics (Ulrich, 1984; Peters & Timmerhaus, 1991). Of the more recent texts (Smith, 1995; Woods, 1995; Turton, Bailie, Whiting, & Shaeiwitz, 1997; Biegler, Grossmann, & Westerberg, 1997), only Seider, Seader, and Lewin (1999) additionally provide detailed support on the use of simulators, with an explicit treatment of the interaction of design and control. In this paper, it is an objective to discuss our view of several key aspects of how computer-aided process design can be taught to chemical engineering undergraduates. This topic has been treated previously by a number of chemical engineering educators, starting with Westerberg (1971), and with more recent treatments by Turton and Bailie (1992), Cameron, Douglas, and Lee (1994), Shaeiwitz, Whiting, and Velegol (1996), Bell (1996), Rockstraw, Eakman, Nabours, and Bellner (1997), Counce, Holmes, Edwards, Perilloux, and Reimer (1997). It is not intended in this article to provide a comprehensive coverage of instruction in process design with emphasis on the advantages and disadvantages of alternative approaches. Rather, it is our purpose to extend some old ideas and introduce some new ones that we have tested with our students.

advocate is in order. We therefore rst discuss the particular skills that need to be fostered, and the frame of reference used to dene goals for the student, couched in terms of educational objectives.

2.1. Educational approach ad6ocated


An important goal of the undergraduate curriculum in chemical engineering is to develop the integration, design, and evaluation capabilities of the student. As shown in Fig. 1, Bloom (1956), characterized the six cognitive levels in the hierarchy: Knowledge Comprehension Application Analysis Synthesis Evaluation. The cognitive skills at the highest level are synthesis and evaluation, which rely on comprehension, application, and analysis capabilities in the knowledge domain, and are consequently the most difcult and challenging to teach. However, to prepare undergraduates to be effective designers in industry, it is important to ensure an adequate coverage of these higher-level skills, rather than limit their education to one based on just knowledge, comprehension, application, and analysis. To achieve the desired coverage in a cost-effective manner, it is important to dene instructional objectives in each undergraduate course in a manner such that the six skills are covered by the senior year. Note that Blooms taxonomy has been applied in chemical engineering by Fogler and LeBlanc (1995), Fogler (1999), Felder and Rousseau (2000). The focus of the learning activity is placed on the accomplishments expected from the student through the formulation of course goals in terms of instructional objectives. The key is to provide material that increases the abilities of the students, with the emphasis being on what the student is able to achieve rather than merely what he or she is aware of or understands. As an example of a possible approach, the instructional objectives for a typical course on process design might be: On completion of this course, the student should be able to: Carry out a detailed steady-state simulation of a chemical process using a process simulator (e.g. HYSYS) and interpret the results. Synthesize a network of heat exchangers for a chemical process such that the maximum energy is recovered or the minimum number of exchangers is used. Synthesize a train of separation units. Suggest reasonable process control congurations using qualitative methods. Formulate and sol6e a small-scale process optimization problem using a process simulator (e.g. HYSYS). E6aluate process alternatives at various levels: single units, complete plants, and the conglomerate level.

2. A structured approach relying on fundamentals Before discussing the building blocks that are an integral part of the toolbox of a process designer, a brief mention of the educational approach that we

Fig. 1. Blooms taxonomy of educational objectives (Bloom, 1956).

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

297

Exercise judgment in the selection of physical property correlations for design. It is noted that these objectives focus on the prociency in required skills expected from the student. Clearly, a precondition for exhibiting these skills is that the student understands the underlying material. Furthermore, it is our experience that students feel more comfortable with clearly dened objectives that quantify what is expected of them.

2.2. The design project and process simulator as means to integrate process knowledge
A designer must have a working knowledge of mathematics, chemical and physical technology, biotechnology, materials science, and economics, which are the building blocks used by the design engineer. This knowledge is developed in a structured fashion in the core chemical engineering courses. It is advantageous to develop the capabilities of the students with a process simulator, in conjunction with the core course materials, as will be discussed shortly. The integration skills of the students are developed through their solution of industrially-relevant design case studies. During the design project, teams of students are expected to call upon diverse aspects of their working knowledge to carry out an integrated process design, determining its feasibility with respect to environmental impact, safety, controllability, and economics. In so doing, the student designer integrates previously acquired knowledge in the engineering disciplines, as well as management skills. Due to the problem scale, this inevitably involves the use of a process simulator to formulate and solve the material and energy balances, with phase and chemical equilibrium, chemical kinetics, etc. and to size process equipment for cost estimation. Familiarity and competence in the use of a simulator permit the student to quickly develop a base-case design, which is veried against process and thermodynamic data. The availability of a reliable process model allows the design team to assess rapidly the economic potential for alternative designs, as well as to derive optimal operating conditions using optimization methods that incorporate economics. Moreover, competence in the use of the simulator allows process evaluation to go beyond economics alone; controllability and operability can be assessed using dynamic simulation, while some simulators automatically provide information to help determine the environmental impact of each of the product streams. Process simulators are an indivisible part of modern practice in chemical process design. This has been true for some time in the petrochemicals, bulk and ne chemicals industry, and is rapidly becoming true in

biotechnology and microelectronics manufacturing. The routine use of the process simulator in industry implies that chemical engineering graduates should be competent to utilize these tools in the analysis, synthesis, and evaluation of process designs. Once students have learned to use simulators intelligently and critically, they appreciate how easy it is to incorporate data and perform routine calculations, and master effective approaches to building up knowledge about a process. As discussed next, the level of simulation skills required of the students completing industrial-scale design problems imply sufcient exposure to the use of simulators during the core courses.

2.3. Use of the simulator in core courses: opportunities and challenges


The high level of competence in the use of simulation expected of the students in the design project relies on their having obtained exposure to simulation in parallel with the core courses. One way to accomplish this is to require students to solve at least one exercise involving the use of simulators as part of each core course. Indeed, recent articles by Russell and Orbey (1993), Bailie, Shaeiwitz, and Whiting (1994) discuss the addition of design projects in the sophomore and junior years. Table 1 provides a typical simulator-based exercise for core courses in the chemical engineering curriculum. Adoption of such a sequence goes far in preparing students to use a simulator in solving largescale problems in the senior design course. With the wide availability of commercial process simulators to educators, the working knowledge of mathematics, chemical and physical technology, and economics can be put to effective use in solving meaningful problems, starting in the sophomore course on material and energy balances, by solving various parts of a complete process with a process simulator. The third author of this paper recalls vividly his experience as a junior when taking the rst course in chemical engineering, based on material in Chemical Process Principles Part 1 Ma terial and Energy Balances (Hougen & Watson, 1943). The instructor rst covered the fundamentals in Chapters 1 9, with application to and homework exercises for small closed-end problems. The last 2 weeks of the course were spent on Chapter 10, which involved material and energy balance calculations by hand for a complete process. Although the calculations were tedious and very time consuming, students developed an appreciation of what chemical engineering was all about and a desire to proceed to the next level of instruction. Today, the tediousness and time-consuming aspect of process calculations can be eliminated and some time can be spent on teaching synthesis and evaluation skills, even in the sophomore year. The material and energy

298

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

Table 1 Core course sequence and typical exercises using simulators Course Mass and energy balances Heat transfer Exercise Analysis of methanol synthesis loop Objectives Convergence of material and energy balances for processes with recycle and purge streams Analysis of sensitivity to degrees-of-freedom Selection of economically optimal operating conditions Designing a heat exchanger for vaporizing uid (computing temperature approaches) Optimal selection of heat-transfer area, weighing reduced energy demands in furnace against increased cost of exchanger Avoidance of temperature crossovers Impact of estimation method on the accuracy of thermodynamic properties, including K -values and enthalpies. Impact of design variables (e.g. number of ideal trays, feed tray location) on performance of the column Impact of selection of degrees of freedom on attaining column specications Difculties in converging multicomponent, multistage separation models Learning to set up a dynamic simulation Denition of controlled and manipulated variables and the installation and tuning of control loops Testing the dynamic resiliency of the column Learning to use the simulator to set up and solve an optimization problem Observing the importance of selecting the appropriate manipulated variables for optimization Observing the impact of process constraints

Heat-integrated toluene dehydroalkylation (see Fig. 1(d))

Thermodynamics Constructing Txy diagrams for alcoholwater systems Separation Simulation of a depropanizer column processes

Dynamics and control

Dynamics and control of a binary distillation column

Process design

Optimization of a multi-draw column

balance course is taught in the sophomore year, using textbooks such as Himmelblau (1996), Felder and Rousseau (2000). Both of these books cover essentially the same fundamentals as presented in the Hougen and Watson textbook. In addition, Himmelblau (in Chapter 6) and Felder and Rousseau (in Chapter 10) cover the solution of material and energy balances for continuous, steady-state processes with a process simulator. Both texts leave to the instructor the choice of a process simulator and instruction on how to use it, so unless he or she is knowledgeable in the use of computer-aided process simulation programs, it is probable that this material will not be covered. In Chapters 12 and 13 of Felder and Rousseau, two fairly complex processes are described and problems given for making material and energy balances, as well as other chemical engineering calculations. Calculations for the methanol synthesis process in Chapter 13 are particularly suitable for the use of a process simulator and serve as an excellent introduction in the sophomore year to process design. The use of a process simulator in the sophomore year introduces the student to the importance of being familiar with a large number of chemical species; the use of physical properties such as density, vapor pressure, specic heat, enthalpy, and K -values; the ease of chang-

ing units; the ease of drawing process ow diagrams with systematic ways of numbering streams and equipment units; and methods of handling recycle. If students are introduced to the use of a process simulator in the sophomore year, their skill in using simulators can be further enhanced in the junior year in courses in uid mechanics, heat transfer, separations, thermodynamics, and reaction engineering. The course in uid mechanics can include simulator calculations of pipeline pressure drop, sieve-tray pressure drop, and power requirements of pumps, compressors, and turbines. The study of heat exchangers in the heat transfer course can include the detailed design of a heat exchanger, including considerations of the complex variation of the temperature driving force, temperature crossover violations, and prediction of bubble and dew points for multicomponent mixtures. Process simulators are quite useful in the solution thermodynamics course because the tedious calculations of activity coefcients, K -values, bubble and dew points, vapor liquid equilibria, liquid liquid equilibria, and data correlation are readily carried out, and property graphs and tables are easily prepared. When a process simulator is used in a thermodynamics course, less time need be spent on the myriad of equations that

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

299

appear in the textbooks and more time can be spent in solving practical problems that demonstrate the importance of thermodynamics to students. Regrettably, the use of a process simulator in a solution thermodynamics course does not appear to be considered in the leading textbooks on the subject. Instead these textbooks either provide their own computer programs for computing physical properties or suggest the use of popular numerical-method programs. Thus, the opportunity to integrate the important lessons learned in the solution thermodynamics course for the later benet of the capstone design course is often missed. The separations course can prot greatly from the use of process simulators to solve both binary and multicomponent, multistage separation operations such as distillation, absorption, stripping, and liquid liquid extraction. It is suggested that less time be spent on graphical methods that are limited to binary and ternary mixtures, with more time spent on multicomponent separations that are readily handled by process simulators. The reactor-engineering course also affords an excellent opportunity to tackle practical problems in reactor design after completing instruction on the ideal plugow and CSTR reactors. Using an enthalpy datum of the elements (rather than the compounds), simulators readily handle reactor energy balances without the need to supply heat of reaction information. Simulators also readily compute chemical or simultaneous chemical and physical equilibrium using either the equilibrium-constant method for specied stoichiometry or the minimization of free energy method for specied product chemicals. Activity coefcients can be taken into account and complex kinetic expressions can be specied. Here too, the use of process simulators to design chemical reactors appears to be ignored in the leading textbooks on chemical reaction engineering. As discussed by de Nevers and Seader (1992), the use of process simulators prior to the senior design course provides students with an opportunity to develop a critical attitude towards chemical process calculations. They cite a problem involving the condensation and subsequent single-stage ash separation at 100 psia of a vapor mixture of ammonia and water, initially at 290 F and 250 psia. The student rst solves this problem graphically using an enthalpy concentration diagram. The result, which is considered to be reasonably accurate, is a vapor of \ 99 wt.% ammonia and a liquid of about 68 wt.% ammonia at a temperature of about 80 F. The student then solves the problem numerically with a process simulation program. He or she is required to select at least four different pairs of K -value and enthalpy correlations for comparison with the graphical solution. Many students are shocked by the

widely varying results. For example, with one set of four pairs of correlations, the ash temperature ranges from 91.2 to 83.4 F with an average of 0.5 F. From then on, students pay careful attention to the selection of correlations for physical properties. The educational importance of discussing errors is also presented by Whiting (1987, 1991). Students who have used process simulators throughout the chemical engineering curriculum are in a position in the senior design course to concentrate their efforts on synthesis and evaluation aspects of process design. Instructors can devote more time to instruction in the synthesis of heat-exchanger systems using pinch analysis, the synthesis of nearly- and non-ideal separation trains, second-law analysis, economic evaluation, optimization, waste minimization, safety, environmental impact, and controllability. During the senior design project, teams of students are better prepared to call upon diverse aspects of their working knowledge to carry out an integrated process design and determine its feasibility from all aspects, not just economics.

2.4. Effecti6e instruction in process simulation: the role of self -paced approaches
The quality of training may be enhanced, and instruction resources used more efciently, through the use of multimedia and web-based approaches. Such self-paced methods of training undergraduates allow them to obtain the details they need to use the simulators effectively, saving instructors class time, as well as time answering detailed questions as the students use simulators to make calculations. In a typical situation, when creating a base-case design, students can use the examples in the multimedia tutorials to learn how to obtain physical property estimates, heats of reaction, ame temperatures, and phase distributions. Then, students can learn to create a reactor section, using the simulators to perform routine material and energy balances, and in some cases kinetic calculations, to size the reactor. Next, they can create a separation section, which often involves multicomponent, multistage distillation-type calculations (Seader & Henley, 1998), which almost always leads to the addition of recycle streams. Using the coverage of process simulators in the multimedia tutorials accompanying the textbook by Seider et al. (1999), the instructor needs only to review the highlights of simulator usage in class. This invariably leaves time for the discussion of more advanced issues. Furthermore, through installation of the multimedia materials on the web, students gain access to the material from remote locations. Our experience is that the response of students to self-paced multimedia instruction has been very positive.

300

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

3. A balance between heuristic and algorithmic approaches The teaching of design should strike a balance between heuristic and algorithmic approaches. Since design invariably involves signicant designer intervention, it is important to teach both heuristics as well as computer-aided algorithmic methods. The former lay the foundations for acquiring the experience necessary to carry out practical process design, while the latter is critical to ensure the generation of optimal designs. Process synthesis is generally introduced rst by example and by instructing students to rely on heuristics (Douglas, 1988). These heuristic rules are important in that they provide a framework for workable designs, based on easy to understand rules of thumb (Walas, 1988). For example, consider the synthesis of a process to hydrodealkylate toluene using a number of heuristic rules, which lead to the sequence of ow diagrams shown in Fig. 2 (Seider et al., 1999). It is noted in Fig. 2(a) that an undesirable side-reaction to biphenyl accompanies the principal reaction, and the conversion of toluene is incomplete. The selection of the reactions conditions is motivated by a desire to minimize the production of the unwanted side-product, while maximizing the yield. The reaction conditions lead to the distribution of chemicals shown in Fig. 2(b), in which

unreacted toluene and hydrogen are recovered by installation of two material recycle streams. The two reaction products (benzene and biphenyl) are removed from the unreacted toluene and hydrogen by installation of a separation section. One possible arrangement consists of the ash vessel and three distillation columns shown in Fig. 2(c). It is noted that heuristics dictate that column operating pressures should be selected to allow the usage of cooling water whenever possible. Finally, Fig. 2(d) shows a possible instantiation of task integration, in which a preheater is installed to supply much of the heat duty required to bring the reactor feed to the high temperature that favors the primary reaction, by exchange with the hot reactor products, which need to be cooled. This arrangement signicantly reduces the heat duty required in the furnace. As the heuristic ideas are mastered, the students should be directed to computer-aided algorithmic approaches that assist them in the generation of better designs. Several algorithmic approaches, which have great practical value, should be presented. These include heuristic and evolutionary synthesis of nearly ideal vapor liquid separation sequences (Seader & Westerberg, 1977), synthesis of separation systems for non-ideal liquid mixtures (Malone & Doherty, 1995), the application of second -law analysis (Seider et al., 1999) to identify opportunities for improved energy

Fig. 2. The evolution of the owsheet for a process to hydrodealkylate toluene.

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

301

Fig. 2. (Continued )

302

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

utilization, and the application of methods to compute heat recovery targets (Linnhoff & Hindmarsh, 1983), and to assist in the design of optimal or near-optimal heat-exchanger networks (Smith, 1995). For example, the following algorithmic approaches can rene the design in Fig. 2(c): 1. Compare the separation sequence in the base-case design to alternative sequences by branch-andbound search. 2. Check the utility requirements against the thermodynamic MER (maximum energy recovery) target using the temperature-interval or graphical methods. Then, a mixed-integer non-linear program (MINLP) can be implemented to derive an optimal design for implementation. There may be additional opportunities for energy savings. For example, a number of alternative heat-integration congurations can be considered for the column sequence proposed in Fig. 2(c). In these congurations, the heat of condensation in a column operating at high pressure is used to supply the heat of vaporization in a column operating at a lower pressure, requiring careful selection of column operating pressures to ensure sufcient temperature driving forces. In selecting between these alternatives, the economic benets need to be weighed against their impact on the operability of the process, as discussed next.

4. Integration of design and control Traditionally, plant controllability and operability has been considered late in the design process, often leading to poorly performing chemical plants. The indisputable fact that design decisions invariably impact the process controllability and resiliency to disturbances and uncertainties is driving modern design methods to handle owsheet controllability in an integrated fashion. Several recent articles, including Rhinehart, Natarajan, and Anderson (1995), Edgar (1997), stress the need to integrate process control with process design. The model of an industrial chemical process for studying process control technology presented by Downs and Vogel (1993) has proved to be very valuable in helping to bridge the gap. Morari and Perkins (1995) stress the importance of steady-state and dynamic analysis in the determination of controllability. Perkins (2000) cites the need for educators to develop a systematic process systems approach that considers design, operation and control. Lewin (1999) describes the state-of-the-art and suggests that two alternative approaches, controllability and resiliency (C &R ) screening methods and integrated design and control, can ensure that chemical plants meet design specications. While C&R analysis is used for screening early in the design process, the integrated design and control approaches can be applied to fully

optimize and integrate the design of the process and its operation. Lewin focuses on three critical aspects that are predicted to characterize future activity in integrated design and control: 1. The quantitative assessment of chemical process controllability and resiliency has generated considerable interest, both academically and in industry. The vendors of commercial simulation software equate controllability assessment with dynamic simulation, and ultimately, plant-wide operability and controllability needs to be veried using this tool. However, it is more important to initiate C&R diagnosis without this expensive and engineering-intensive activity. It has been shown that controllability analysis reduces the alternatives early in the design process (Perkins & Walsh, 1994; Weitz & Lewin, 1996; Solovyev & Lewin, 2000). The challenge to the vendors is to build these tools directly into their simulation software. 2. Approaches for integrated design and control are important for improving a nal design (Bansal, Mohideen, Perkins, & Pistikopoulos, 1998). To effectively use a MINLP, it is necessary to develop methods to prune the network of congurations evaluated by the MINLP solver. The commonly used heuristic approach for MINLP network pruning can be replaced by adopting C&R analysis. 3. The training of chemical engineers, who should be taught to view design and control as an integrated activity, is a precondition to the future advancement of this eld (Seider et al., 1999; Luyben, Tyreus, & Luyben, 1999). To this end, both the fundamentals of process dynamics and control, and the impact of design on control, should be covered adequately in the undergraduate curriculum. The concern here is the need to bridge the gap between traditional process control courses, which emphasize theory, and applications to actual processes. As an illustration, consider potential control problems in the owsheet in Fig. 2(d), and their resolution by adopting C&R diagnosis during the design process: 1. Impact of recycle: The positive feedback loops associated with the material recycles in the owsheet can amplify feed disturbances. Careful controllability assessment indicates that the control conguration needs to account for the dynamic interaction between the process units. More specically, to eliminate the disturbance amplication caused by the material recycles, it is recommended that the ow rate of the recycle streams be controlled, either directly or indirectly by manipulating the purge stream. 2. Impact of heat -integration: The loss of degrees-offreedom associated with heat integration may cause the quality of control to deteriorate, depending on the conguration selected.

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

303

As an example of the impact of C&R diagnosis on the synthesis of heat-integrated designs, consider the selection of an appropriate conguration for heat-integrated columns for the separation of an equimolar mixture of methanol and water, a typical product of a methanol synthesis loop. To provide commercial methanol, nearly free of water, dehydration is achieved commonly by distillation, a process in which energy is invested in return for separation. To reduce the sizable energy costs, heat-integrated congurations are considered commonly as alternatives to a single distillation column, three of which are shown in Fig. 3: FS (Feed Split): The feed is split nearly equally (FH : FL) between two columns to achieve optimal operation. The overhead vapor product of the highpressure column supplies the heat required in the low-pressure column. LSF (Light-split/Forward heat-integration): The entire feed is fed to the high-pressure column. About half of the methanol product is removed in the distillate from the high-pressure column, and the bottoms product is fed into the low-pressure column. Heat integration is in the same direction as the mass ow. LSR (Light-split/Reverse heat-integration): The entire feed is fed to the low-pressure column, with the bottoms product from the low-pressure column fed into the high-pressure column. Heat integration is in the opposite direction to that of the mass ow. These congurations reduce the energy costs by using the heat of condensation of the overhead stream from the high-pressure column (H) to supply the heat of vaporization of the boilup in the low-pressure column (L). Although more economical, assuming steady-state operation, they are potentially more difcult to control because the congurations: (a) are more interactive; and (b) have one less manipulated variable for process control, since the reboiler duty in the low-pressure column can no longer be manipulated independently. To show the energy savings, the owsheets for a single column and for the three heat-integrated alternatives in Fig. 3 were simulated on the basis of an

equimolar feed of 45 kmol/min, producing 96 mol% methanol in the distillate and 4 mol% methanol in the bottoms product, assuming 75% tray efciency and no heat loss to the surroundings, and using UNIFAC to estimate the liquid-phase activity coefcients. The total energy requirements for the four alternatives were computed as follows:

SC LSF

0.353106 kcal/ min 0.222106 kcal/ min

LSR FS

0.205106 kcal/ min 0.205106 kcal/ min

Clearly, the LSR and FS congurations save the most energy, and on the basis of steady-state economics alone, one of these two congurations would be selected. It makes sense to consider controllability and resiliency diagnosis to select the most appropriate conguration, as did Chiang and Luyben (1988), using the relative gain array (RGA) and minimum singular values based upon linear approximations to detailed non-linear process models. Although their ndings were inconclusive, they showed the FS conguration to be far less desirable using closed-loop simulations with their non-linear model. It is preferable, however, to perform C&R diagnosis using the results of the steadystate material and energy balances in a procedure suggested by Weitz and Lewin (1996), involving the following steps: 1. After the alternative owsheets are synthesized, control structures are considered, rst by selecting the process outputs to be controlled, y 6 {t }, the manipulated variables, u 6 {t }, and the disturbance variables, d 6 {t }. These are related by the model: y {s } = P {s }u {s } + Pd {s }d {s }. 2. Steady-state simulations of the owsheets are carried out using a process simulator.

Fig. 3. Three heat-integrated alternatives to a single distillation column.

304

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

Fig. 4. DC maps for the SC, FS and LSR congurations to dehydrate methanol. The bounds on the disturbances are 9 20% from their nominal values. The DC maps for each manipulated variable are computed separately.

3. The owsheets are decomposed into component parts. These are MIMO subsections of the owsheets that are approximated by matrices of low-order transfer functions (usually rst order with deadtime). This decomposition permits process units to be modeled in sufcient detail, allowing inverse response and overshoot phenomena to be represented. 4. Steady-state gains for the component parts are computed by perturbation of each input, one at a time. Time constants and delay times are estimated assuming perfect mixing or plug ow, as appropriate, with the ow rates at steady state. At this point, transfer-function matrices are dened for each component part. 5. The transfer-function matrices, P {s } and Pd {s }are generated for each complete owsheet. This involves computing the frequency response of each component part, and recombining the component parts, as dictated by the plant topology. 6. The frequency-dependent C&R measures are computed using the approximate linear model, P { j }

and Pd { j }. Following this approach, C&R diagnosis is carried out on the single column, as well as the best heat-integrated congurations, in terms of steady-state economics, namely the FS and LSR congurations. Fig. 4 shows the Disturbance Cost contour maps (Lewin, 1996) computed for each of the manipulated variables associated with the congurations: SC, FS and LSR, where the abscissa is the frequency, and the ordinate is the direction of the disturbance [F, xF ]T. For clarication, a disturbance direction of 0 denotes a positive disturbance in the feed ow rate alone, of 90 denotes a positive disturbance in the feed mole fraction alone, and of 45 denotes that both disturbances in the feed ow rate and feed mole fraction are at their maximum positive values. Since DC = 0.5 corresponds to a saturated manipulated variable, the disturbances are rejected adequately by all of the designs at the steady state (when 0). However, for a wide range of disturbance directions, the FS conguration has disturbance costs that exceed the 0.5 constraint beyond 0.1 rad/min (LH, QRH and FH/FL), and thus, disturbance

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306

305

rejection is expected to be very sluggish for this conguration. The other two congurations have low disturbance costs, and can be expected to reject these disturbances about as well as a single column. These results are corroborated by dynamic simulations using HYSYS. Plant (Seider et al., 1999), and are in agreement with those of Chiang and Luyben (1988). In this case, C&R analysis is effective for screening, enabling the FS and SC congurations to be rejected without the need for dynamic simulations. This approach has been used successfully for screening owsheets featuring exothermic reactors (Naot & Lewin, 1995), and polymerization reactors (Lewin & Bogle, 1996), azeotropic distillation columns (Solovyev & Lewin, 2000) and material recycles (Lewin, Gong, & Gani, 1996). In all cases, the conclusions obtained by others using rigorous dynamic models have been conrmed. It is promising as a short-cut diagnostic tool, and is well suited for integration into owsheet simulation software. When such analysis tools become available within the framework of commercial simulators, owsheet operability can be checked routinely.

References
Bailie, R. C., Shaeiwitz, J. A., & Whiting, W. B. (1994). An integrated design sequence sophomore and junior years. Chemical Engineering Education, 28 (1), 52 57. Bansal, V., Mohideen, M. J., Perkins, J. D., & Pistikopoulos, E. N. (1998). Interactions of design and control: double-effect distillation systems. Proc. of DYCOPS5, Corfu. Bell, J. T. (1996). Implementation of multiple interrelated projects within a senior design course. Chemical Engineering Education, 30 (3), 204 209. Biegler, L. T., Grossmann, I. E., & Westerberg, A. W. (1997). Systematic methods of chemical process design. NJ: Prentice Hall. Bloom, B. S. (1956). Taxonomy of educational objecti6es. Handbook I : Cogniti6e domain. David McKay Co. Cameron, I. T., Douglas, P. L., & Lee, P. L. (1994). Process systems engineering the cornerstone of a modern chemical engineering curriculum. Chemical Engineering Education, 28 (3), 210 213. Chaing, T., & Luyben, W. L. (1988). Comparison of the dynamic performances of three heat-integrate distillation congurations. Industrial Engineering Chemistry Research, 27, 99 104. Counce, R. M., Holmes, J. M., Edwards, S. V., Perilloux, C. J., & Reimer, R. A. (1997). A quality-driven process design internship. Chemical Engineering Education, 31 (2), 100 105. de Nevers, N., & Seader, J. D. (1992). Helping students develop a critical attitude towards chemical process calculations. Chemical Engineering Education, 26 (2), 88 93. Douglas, J. M. (1988). Conceptual design of chemical processes. New York: McGraw-Hill. Downs, J. J., & Vogel, E. F. (1993). A plant-wide industrial process control problem. Computers and Chemical Engineering, 17 (3), 245 255. Edgar, T. F. (1997). Process control from the classical to the postmodern era. Chemical Engineering Education, 31 (1), 12 17. Felder, R. M., & Rousseau, R. W. (2000). Elementary principles of chemical processes (3rd ed.). John Wiley & Sons. Fogler, H. S. (1999). Elements of chemical reaction engineering. Prentice Hall PTR. Fogler, H. S., & LeBlanc, S. E. (1995). Strategies for creati6e problem sol6ing. Prentice Hall PTR. Himmelblau, D. M. (1996). Basic principles and calculations in chemi cal engineering (6th ed.). Prentice Hall PTR. Hougen, O. A., & Watson, K. M. (1943). Chemical process principles. Part 1: material and energy balances. John Wiley & Sons. Lewin, D. R. (1996). A simple tool for disturbance resiliency diagnosis and feedforward control design. Computers and Chemical Engineering, 20 (1), 13 25. Lewin, D. R., Gong, J. -P., & Gani, R. (1996). Optimal design and controllability assessment for plant-wide benchmarks. Proc. of the 13th IFAC World Congress, 6ol. M (pp. 79 84). San Francisco. Lewin, D. R., & Bogle, D. (1996). Controllability analysis of an industrial polymerization reactor. Computers and Chemical Engi neering, 20(S), S871 S876. Lewin, D. R. (1999). Interaction of design and control. Proc. of the 7th IEEE Mediterranean Conference on Control and Automation (MED99). Haifa. Linnhoff, B., & Hindmarsh, E. (1983). The pinch analysis method for heat exchanger networks. Chemical Engineering and Science, 38, 745. Luyben, W. L., Tyreus, B. D., & Luyben, M. L. (1999). Plantwide process control. New York: McGraw-Hill. Malone, M. F., & Doherty M. F. (1995). Separation system synthesis for nonideal liquid mixtures. Fourth International Conference on Foundations of Computer -Aided Process Design, AIChE Symp. Series, 91. No. 304(91), 9 18.

5. Conclusions We recommend that a curriculum that prepares chemical engineering graduates for the challenges they will face in industry should include the following features: 1. A structured approach relying on fundamentals. In this approach, students use process simulators starting in the sophomore material and energy balance course, applying their knowledge to practical problems. Students will then be better prepared for the challenges of the capstone design project and can spend more time on synthesis, controllability, safety, environmental concerns, waste minimization, optimization, and economic evaluation. Tutorials prepared in multimedia format can support this goal efciently and spare instruction time in the classroom. 2. A balance between heuristic and computer -aided al gorithmic approaches. Since design invariably involves signicant designer intervention, it is important to teach both heuristics as well as algorithmic methods. The former lay the foundations for acquiring the experience necessary to prepare practical process designs, while the latter is critical to ensure the generation of optimal designs. 3. Integrated design and control. Instruction should reect the current state-of-the-art in the integration of process design and process control. The concern here is the need to bridge the gap between traditional process control courses, which emphasize theory, and applications to actual processes.

306

D.R. Lewin et al. / Computers and Chemical Engineering 26 (2002) 295 306 principles : synthesis, analysis and e6aluation. New York: John Wiley and Sons Inc. Shaeiwitz, J. A., Whiting, W. B., & Velegol, D. (1996). A large-group senior design experience teaching responsibility and life-long learning. Chemical Engineering Education, 30 (1), 70 75. Smith, R. (1995). Chemical process design. New York: McGraw Hill. Solovyev, B., Lewin, D. R. (2000). Controllability and resiliency analysis for homogeneous azeotropic distillation columns. Proc. of ADCHEM 2000, Pisa. Turton, R., & Bailie, R. C. (1992). Chemical engineering design problem-solving strategy. Chemical Engineering Education, 26 (1), 44 49. Turton, R., Bailie, R. C., Whiting, W. B., & Shaeiwitz, J. A. (1997). Analysis, synthesis and design of chemical processes. New Jersey: Prentice Hall. Ulrich, G. D. (1984). A guide to chemical engineering process design and economics. New York: Wiley. Walas, S. M. (1988). Chemical process equipment selection and design. Boston: Butterworths. Weitz, O., & Lewin, D. R. (1996). Dynamic controllability and resiliency diagnosis using steady state process owsheet data. Computers and Chemical Engineering, 20 (4), 325 336. Westerberg, A. (1971). A course on computer aided process design. Chemical Engineering Education, 5 (4), 180 186. Whiting, W. B. (1987). Textbook errors: a rich source of problems and examples. 1987 ASEE Annual Conference Proceedings (p. 1631) Reno, NV. Whiting, W. B. (1991). Errors a rich source of problems and examples. Chemical Engineering Education, 25 (3), 140 144. Woods, D. R. (1995). Process design and engineering practice. New Jersey: Prentice Hall.

Morari, M., & Perkins, J. D. (1995). Design for operations, Fourth International conference on foundations of Computer-Aided Process Design, 304(91), 105 114. Naot, I., & Lewin, D. R. (1995). Analysis of process dynamics in recycle systems using steady state owsheeting tools. Proc. of the 4th IFAC Symp. on Dynamics and Control of Chemical Reactors, Distillation Columns and Batch Processes (DYCORD + 95). (pp. 203 208). Helsingor. Perkins, J. D., & Walsh, S. P. K. (1994). Optimization as a Tool for Design/Control Integration. Proc. of the 2nd IFAC Workshop on Integration of Design and Control (pp. 1 10) Baltimore, Maryland. Perkins, J. D. (2000). Education in process systems engineering past, present and future. Computers and Chemical Engineering, 24, 1367. Peters, M. S., & Timmerhaus, K. D. (1991). Plant design and econom ics for chemical engineers (4th ed.). New York: McGraw-Hill. Rhinehart, R. R., Natarajan, S., & Anderson, J. J. (1995). A course in process dynamics and control. Chemical Engineering Education, 29 (4), 218 221. Rockstraw, D. A., Eakman, J., Nabours, N., & Bellner, S. (1997). An integrated course and design project in chemical process design. Chemical Engineering Education, 31 (2), 94 99. Russell, T. W. F., & Orbey, N. (1993). The technically feasible design. Chemical Engineering Education, 27 (3), 166 169. Seader, J. D., & Westerberg, A. W. (1977). A combined heuristic and evolutionary strategy for synthesis of simple separation sequences. American Institute of Chemical Engineering Journal, 23, 951. Seader, J. D., & Henley, E. J. (1998). Separation process principles. New York: John Wiley and Sons Inc. Seider, W. D., Seader, J. D., & Lewin, D. R. (1999). Process design

Das könnte Ihnen auch gefallen