Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
SYNTHESIS AND DESIGN OF CHEMICAL PROCESSES Lim Kuan Howe1, Dennis1, N.V.S.N. Murthy Konda1 and G.P Rangaiah1* ABSTRACT The goal of this work was to critically examine two hierarchical procedures used by Douglas [1] and Stanislav [2] for the conceptual design of chemical processes. Using the Hydrodealkylation of Toluene (HDA) process as a case study, it was found that the flowsheet developed using Stanislav’s procedure was more economical than the flowsheet developed using Douglas’ procedure. The main advantage of Stanislav’s procedure is that it is rich enough to allow all reasonable alternatives to surface, hence eliminating the possibility of letting the best flowsheet slip through the mind of the engineer during the design process. However, the drawback is that the process is more time consuming as more options have to be enumerated. This disadvantage was circumvented in this study by the formulation of a HYSYS-EXCEL interface using Visual Basic programming, which essentially combines the simulation power of HYSYS with the spreadsheet capabilities of EXCEL, thereby greatly shortening the time needed to evaluate a particular flowsheet. INTRODUCTION Process design involves not only technical knowledge but also a good dose of creativity. With the increasing complexity of chemical processes and the invention of novel operations (e.g. reactive distillation, membrane separations), it is not surprising that numerous alternatives can be generated. The goal of the chemical engineer then, is to choose the alternative that brings the largest economic benefits to the company. However, even with the advent of powerful process simulation tools, it still remains a challenge to develop an entire process flowsheet from the drawing block. Hence arises the need for a systematic approach to plant design that would help the chemical engineer to arrive at the best design in a logical fashion. Douglas [1] suggested a hierarchy of decisions (Figure 1), which shall be referred to in this paper as the conventional design procedure. Using this procedure to develop the process flowsheet will help the designer to arrive at the final design using a stage-bystage method. At each stage, he generates new alternatives and evaluates those using selected economic criteria. He then chooses the best alternative and proceeds to the next stage where he will have to make another decision among the new alternatives that are generated. Finally, he will arrive at the last stage where the final choice will lead to the ‘best’ alternative. Stanislav [2] suggested a new hierarchy, also shown in Figure 1, which shall be referred to in this paper as the modified design procedure. There are two key modifications from 1
Department of Chemical and Biomolecular Engineering, National University of Singapore, Singapore 119260.
*
Author for correspondence; Email:
[email protected]. Fax: (65) 6779 1936; Phone: (65) 6874 2187
39
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
the conventional procedure. Firstly, the reactor design is considered as a separate stage of the design procedure rather than as part of the recycle structure. Secondly, the decision to recycle is left to the last stage of the hierarchy, just before the heat integration.
Figure 1: Conventional and Modified Design Procedures The objective of this study is therefore to present a comparison of the conventional and modified design procedures, the aim being to see which design hierarchy would help the engineer to arrive at the ‘best’ process flowsheet in the shortest time. The HDA process is used as the example in this study, which involves the extensive use of a process simulator as well as sizing, costing and profitability analysis. FLOWSHEET EVALUATION In order to apply either of the two procedures listed in Figure 1, it is necessary to evaluate the profitability of the different flowsheets generated at each stage of the procedure so that we can select the most profitable alternative and proceed to the next stage. To determine the profitability, the total capital investment and the annual cost of sales have to be computed. These two components must in turn be calculated from various subcomponents including equipment cost, cost of land, site preparation, feedstock cost, utility cost, labor cost, maintenance, operating overhead, depreciation, selling expense etc. A summary of these computations is shown in Figure 2 and further details can be
40
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
obtained from Seider et al. [3] as well as Turton et al. [4]. The problem however, is that these computations are lengthy and even with the advent of commercial simulation software, it still takes a substantial amount of time to arrive at the profitability of a particular flowsheet e.g. the product selling price required to give a return on investment of 20%. This was the reason why Douglas [1], in demonstrating the application of his proposed design hierarchy, used shortcut calculations and estimates to determine the profitability of the different flowsheets generated at each stage of the hierarchy. This method, though time-saving, is undesirable because it might lead to the engineer eliminating a potentially more economical flowsheet.
Figure 2: Profitability Analysis of a Flowsheet
Figure 3: Linking Object Libraries of HYSYS and EXCEL To circumvent this problem, a special program was developed in this study to automate the cumbersome process of determining the required selling price for a particular flowsheet. This program is essentially an interface, which combines the process simulation power of HYSYS with the spreadsheet capabilities of EXCEL by linking the
41
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
object libraries of these two applications through Visual Basic as shown in Figure 3. With this interface program, the time taken to evaluate a process flowsheet is thus greatly reduced and as a result, the use of the design procedures becomes simpler and more accurate. The HYSYS-EXCEL interface is based in the EXCEL environment, the main page for the HDA process as shown in Figure 4. On this screen, the interface captures the key process simulation results from HYSYS and displays it in a user-friendly interface, allowing the user to see at one glance whether all the key process constraints are satisfied. This interface also allows the user to change the simulation parameters in HYSYS directly from EXCEL. For example, the user can change the desired reaction conversion rate on the EXCEL interface, which will automatically transmit the new input to HYSYS, which will in turn run the simulation based on the new input and then send the new simulation results back to the EXCEL interface.
Figure 4: Main View of HYSYS-EXCEL Interface for HDA Process Example The main feature of the interface that allows it to speed up the evaluation of a process flowsheet is its ability to size and cost equipment using macros in the EXCEL interface. For example, sizing and costing a distillation column manually requires a lot of work because iterative computations have to be done to obtain the column diameter which can 42
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
give a vapor velocity that will not bring about flooding or entrainment. In addition, calculations will also have to factor in the tray efficiencies. However, all these calculations can be done automatically using macros in the HYSYS-EXCEL interface, which will capture the necessary data from the HYSYS simulation, perform the iterative calculations and compute the cost based on pre-entered cost equations. A similar algorithm is followed for the other process equipment. Similarly, macros can also be written for all the cost components in Figure 2 and hence, the profitability of a flowsheet can be obtained at the click of a button. A view of the results that was obtained for a HDA process simulation is shown in Figure 5 where the required selling price of benzene to give a return on investment of 20% is seen to be $29.23/lbmol. The detailed results on the equipment, utility and feedstock costs are also displayed in two other sheets in the EXCEL interface. More details on features and programmability of this HYSYS-EXCEL interface can be found in Lim [5].
Figure 5: Results View of HYSYS-EXCEL Interface for HDA Process Example CASE STUDY: HYDRODEALKYLATION OF TOLUENE PROCESS The HDA process was used as the case study in this work to compare the applications of the conventional and modified design hierarchies. This process involves the conversion of toluene to the more valuable benzene product, with methane and diphenyl as the byproducts. A production rate of 265 lbmol/hr of benzene at 99.99% purity was assumed. 43
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
Process constraints can be found in Douglas [1]. The two non-catalytic vapor-phase reactions are: Reaction 1: Reaction 2:
Toluene + Hydrogen → Benzene + Methane Benzene ↔ Diphenyl + Hydrogen
Stage 1 of both the conventional and modified design hierarchies involves choosing between a continuous and a batch process. In this case study, a continuous process rather than a batch process was chosen for two reasons. Firstly, we are concerned with a plant of a production capacity of 176x106 lb/yr. Such a large plant capacity would be more suited to a continuous production. In addition, the benzene product is not a seasonal product like fertilizers where demand is high only during certain periods of the year. Rather, we expect the sales of a petrochemical product like benzene to go on throughout the year. For this reason also, it is wiser to choose a continuous process. The main difference between the two procedures arises in stage 2. In stage 2 of the conventional procedure, we identify the input-output structure of the flowsheet and in doing so, decisions are also made on which output streams to recycle. In contrast, for the modified design procedure, no decisions on recycle are made at this stage of the hierarchy. For the HDA process, the basic input-output structure with no recycle streams can be developed as seen in Figure 6.
Figure 6: Input-Output Structure of HDA Process with no Recycle Streams If one uses the conventional hierarchy, before designing the reactor sub-system in stage 3, decisions would be made on which output streams from the process ‘black box’ in Figure 6 are to be recycled. For the case of the HDA process, benzene is obviously removed as a product. Toluene is a valuable reactant and hence it should be recycled to the process. Diphenyl is a by-product and a decision should also be made whether to remove or to recycle it to the process. If diphenyl is removed from the process, there will be selectivity losses of benzene to diphenyl and hence the toluene consumption rate and cost would increase. Furthermore, there is the additional capital and operating costs associated with the extra separation step of removing the diphenyl. However, it is also necessary to consider the equilibrium constant of the by-product reaction. Finally, a decision is needed as to whether there should be a recycle stream of the gas stream rich in hydrogen. Here, a trade-off exists between the savings in fresh hydrogen feedstock and the additional costs associated with the recycle compressor as well as the methane build-up in the loop. It is possible to make rough calculations in order to estimate the economic
44
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
potential of these alternatives and to make decisions on which streams to recycle. Douglas [1] in doing so developed the input-output structure shown in Figure 7.
Figure 7: Input-Output Structure of HDA Process (Conventional Hierarchy) Subsequent to the input-output structure, Douglas [1] continued with the stages in the design procedure in Figure 1 and arrived at the flowsheet shown in Figure 8. This flowsheet was re-simulated and evaluated in this work and the required selling price of benzene to give a return on investment of 20% was found to be $29.25/lbmol benzene at an optimum reactant conversion of 75%, similar to what was obtained by Douglas [1].
Figure 8: Flowsheet from the Conventional Design Procedure The question arises with respect to the conventional procedure then as to how sure one can be about the recycle decisions made in stage 2? At such an early stage of the hierarchy, it might not be so easy to make decisions on which streams to recycle based on simple heuristics or estimated calculations. It is more or less certain that toluene, being a valuable feedstock should be recycled back to the process. However, the decision might 45
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
not be so clear for diphenyl, hydrogen and methane because such recycle streams might have a great impact on the design of the reactor and separation subsystems. For example, if one had decided on recycling the hydrogen/methane stream at stage 2 of the hierarchy, the possibility of replacing the stabilizer column with a simple flash unit due to the elimination of methane build-up in the system might not have surfaced. This shows that potentially more efficient designs of the reactor and separation subsystems could have been missed if recycle streams were fixed a priori. In other words, the conventional hierarchy of decisions is not rich enough to allow all potential designs to surface.
Figure 9: Modified Design Procedure with Additional Iterative Loop In contrast, the modified design procedure avoids this flaw of the conventional hierarchy by putting off the recycle decision to a later stage. In this manner, the reactor and separation subsystems can be designed without any limitations and recycle decisions on each output stream can be taken on its own merit i.e. whether recycling a particular output stream will yield a cheaper flowsheet than the base flowsheet without any recycle streams. However, we must not neglect the impact of a recycle stream on the reactor and separation subsystem. For example, the decision to recycle diphenyl would mean that the diphenyl concentration in the process would build up to an equilibrium level and therefore the concern of increased selectivity losses at high reactant conversions is eliminated. As such, the reactor system can be re-designed to operate at a higher optimal reactant conversion. Therefore, it would be more appropriate to add an iterative loop to
46
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
the modified hierarchy used by Stanislav [2] to reflect the re-consideration of the reactor and separation subsystems whenever a new recycle stream is considered. This modification is shown in Figure 9. Applying the modified design procedure in Figure 9 to the design of the HDA process, one would logically start by evaluating a flowsheet where diphenyl, excess toluene, hydrogen and methane are all removed from the process. However, as previously mentioned, the recycle of a valuable feedstock of toluene is almost certainly economical and one might choose a flowsheet with only toluene recycle as the base case instead. This was done in this study and the required benzene price of this flowsheet was evaluated to be $42.05/lbmol benzene (at optimum reactant conversion of 95%). This price was then used as the base price with which the next recycle decision can be compared.
Figure 10: Flowsheet from Modified Design Procedure Next, the economic advantage of recycling diphenyl was examined and the simulated flowsheet was found to give a benzene price of $40.20/lbmol (at optimum reactant conversion of 98%). This price is lower than the base price and therefore, it is economical to recycle diphenyl. The option of recycling hydrogen was then examined by incorporating a recycle gas stream in the flowsheet with a membrane separator to remove methane. This new flowsheet was found to give a benzene price of $29.31/lbmol (at optimum reactant conversion of 98%). Hence, it is certainly economical to recycle the excess hydrogen. It follows naturally then that one should next consider recycling methane along with the excess hydrogen thus saving on the need for a membrane separation unit. The results of this flowsheet showed a benzene price of $28.45/lbmol (at optimum reactant conversion of 90%) and since this is lower than $29.31/lbmol, we conclude that it is not worthwhile to separate and remove the methane from the recycle
47
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
gas and the best flowsheet is that where toluene, diphenyl and flash gas (without further separation) are all recycled. This flowsheet is shown in Figure 10. The main differences between this flowsheet and that of Figure 8 are the elimination of the toluene column and the operation of the reactor at a higher reactant conversion of 98%. Hence, the results show that the flowsheet developed using the modified procedure is more profitable than the flowsheet designed using the conventional procedure. Specifically, for a desired return on investment of 20%, the required selling price of benzene for the former flowsheet is $0.80/lbmole lower than the latter. The advantage of using the modified hierarchy of decisions in Figure 9 is that one would systematically consider all the possible recycle structures of the flowsheet, without the possibility of missing out a better alternative if one had fixed a recycle decision a priori in stage 2 of the hierarchy. This modified design procedure is richer than the conventional one in the sense that it allows all possible alternatives to be included. This is an important characteristic of an effective design hierarchy. On the other hand, as pointed out by Kraslawski et al. [6], another important problem in process design and synthesis is the strategy problem – whether a strategy can be developed to quickly locate the best alternative without totally enumerating all the options. In this respect, the modified procedure loses out to the conventional procedure because by not considering any recycle streams in the first ‘run’ and the iterative loop made around the hierarchy with the consideration of each new recycle stream, more time is taken in simulation and evaluation of the different flowsheets. In contrast, in the conventional procedure, heuristics, experience and shortcut estimations are used to fix recycle decisions early in the hierarchy. As such, fewer alternatives have to be screened in detail and the ‘best’ alternative is quickly identified. However, the shortcoming of the modified hierarchy is overcome by the advent of powerful simulation and costing programs such as the custommade HYSYS-EXCEL interface used in this study. Such software greatly speeds up the process of simulating and costing a process flowsheet and hence time becomes less of a concern. It becomes more important then for an effective hierarchy to be rich enough to allow all alternatives to surface in the design process so that the most economical flowsheet can be developed. As such, the modified design procedure shown in Figure 9 would be better than the conventional one. CONCLUSIONS Using the HDA process as an example, it was found that the modified hierarchy of decisions provides a better way to develop a chemical process flowsheet. In the conventional design procedure, recycle streams are fixed a priori and puts a constraint on the design possibilities for the reactor and separation subsystems. In contrast, by leaving recycle decisions to a later stage of the process, one can develop a base-case flowsheet with no recycle streams first and then subsequently consider the individual economic effects of the addition of different recycle streams. This is a relatively more time consuming approach but it provides a more logical manner in developing a process flowsheet and prevents the design engineer from missing out on any advantageous modifications in the reactor or separation systems that could have been made had not a decision to recycle been fixed in an earlier stage of the hierarchy. However, the
48
Journal of The Institution of Engineers, Singapore Vol. 45 Issue 6 2005
application of a custom-made HYSYS-EXCEL interface will help to greatly reduce the time needed for evaluating each alternative flowsheet generated during the procedure. It is recommended to further test this modified design procedure by applying it to another chemical process. One can also consider incorporating safety, environmental and process control elements into the hierarchy of decisions. For example, diphenyl is known to cause damage to the liver and the nervous system. In addition, it has high acute toxicity to aquatic life. Hence, in the development of the input-output structure, it is necessary to consider whether diphenyl should be removed or allowed to recycle to extinction in the process. While designing complex flowsheets, it is also necessary to take into account the process control considerations. The most profitable flowsheet would be of no use if it is difficult to control. A reactor runaway or frequent off-specification production will make the plant not only unprofitable but also an unsafe place to work in. REFERENCES [1] Douglas, J.M., Conceptual Design of Chemical Processes, New York: McGraw-Hill (1988). [2] Stanislav V.E., An Examination of a Modified Hierarchical Design Structure for Chemical Processes, http://www.che.ttu.edu/classes/che5000/EmetsThesis.pdf (2003). [3] Seider W.D., Seader J.D., Lewin D.R., Process Design Principles, New York: John Wiley & Sons (1999). [4] Turton R., Bailie R.C., Whiting W.B., Shaeiwitz J.A., Analysis, Synthesis, and Design of Chemical Processes, New Jersey: Prentice Hall, PTR, Upper Saddle River (2003). [5] Lim D., Murthy Konda N.V.S.N., Rangaiah G.P., Synthesis and Design of Chemical Processes, Department of Chemical and Biomolecular Engineering, National University of Singapore (2004) [6] Kraslawski A., Li X., Conceptual Process Synthesis: Past and Current Trends, Chemical Engineering and Processing 43: 589-600 (2004)
49