[Computer Aided Chemical Engineering] Integrated Design and Simulation of Chemical Processes Volume 35 || Introduction in Process Simulation

CHAPTER INTRODUCTION IN PROCESS SIMULATION 2 2.1 COMPUTER SIMULATION IN PROCESS ENGINEERING 2.1.1 PROCESS FLOWSHEETING Simulation is a fundamental activity in Process Engineering. The following definition captures its es- sential features (Thomé, 1993): Simulation is a process of designing an operational model of a system and conducting experiments with this model for the purpose either of understanding the behaviour of the system or of evaluating alternative strategies for the development or operation of the system. It has to be able to reproduce selected aspects of the behaviour of the system modelled to an accepted degree of accuracy. Simulation implies modelling, as well as tuning of models on experimental data. A simulation model serves to conduct ‘virtual experiments’. Almost invisible in most cases, being incorporated in the soft- ware technology, modelling is the key feature in every simulation. It is important to keep in mind that the simulation is only an approximate representation of the reality, at a certain level of accuracy, and not the reality itself. That is why the user must always be able to evaluate the reliability of the results delivered by a simulator. Simulation in Process Engineering requires specific scientific knowledge amongwhich wemay cite methods for accurate description of physical properties of pure components and complex mixtures, models for a large variety of reactors and unit operations, as well as numerical techniques for solving large systems of algebraic and differential equations. The scientific and engineering activity that makes use of professional modelling and simulation for Chemical Process Industries (CPI) is designated by computed-aided process engineering (CAPE). Since 1991 the European Federation of Chemical Engineering (www.efce.info) organises each year a scientific congress of the worldwide CAPE community under the label ESCAPE. The main simulation activity in process engineering is flowsheeting. Following a previous defini- tion (Westerberg et al., 1979), flowsheeting is the use of computer aids to perform steady-state energy and mass balancing, sizing and costing calculation for a chemical process. This interpretation reflects the fact that flowsheeting has deep roots in process design. The impres- sive progress in the last 30 years, both in modelling and in simulation technology, has enriched this definition. Nowadays flowsheeting is involved not only in the design of new processes but also in the continuous improvement of existing technologies, by revamp and debottlenecking, in managing process operation and control, as well as in research and development (Dimian, 1994). Computer Aided Chemical Engineering. Volume 35. ISSN 1570-7946. http://dx.doi.org/10.1016/B978-0-444-62700-1.00002-4 © 2014 Elsevier B.V. All rights reserved. 35 http://www.efce.info http://dx.doi.org/10.1016/B978-0-444-62700-1.00002-4 In a complex plant, the units form a system that can be understood at best by simulation. Taking into account the evolution in the last decades, we may formulate a more extended definition as: Flowsheet- ing is a systemic description of material and energy streams in a process plant by means of computer simulation with the scope of designing a new plant or improving the performance of an existing plant. Flowsheeting can be used as an aid to implement a plantwide control strategy, as well as to manage the plant operation. According to the above definition, in flowsheeting the behaviour of the system has the highest pri- ority. The modelling of the individual units must be subordinated to the goal of modelling the entire system. Flowsheeting has different purposes in Design and Operation, but these converge when the knowledge acquired in the operation of an existing plant serves as the basis for improving its design or for developing a new process. In Design, the first objective of flowsheeting is a systematic investigation of different alternatives that can be developed for a given design problem. Modern design relies upon systematic methods whose main merit is to be able to set optimal targets well ahead the detailed design of units. From sev- eral sub-optimal alternatives a base-case is selected, and further submitted to integration, sizing and optimisation. In addition, combined steady-state and dynamic flowsheeting can help to understand the process dynamics, and on this basis to support the implementation of a plantwide control strategy. Mastering the constraints of modelling in flowsheeting can greatly help the designer to deliver a reli- able project despite the lack of data or tight schedule. In Operation, the flowsheeting is more demanding, because it has to mirror the behaviour of an existing plant submitted to various disturbances. The mathematical models embedded in simulator have to be reconciled against plant data. Again, the modelling and accuracy must be subordinated to goals, such as the monitoring of unit performance, maintenance, revamping or support in process control activities. It may be concluded that nowadays flowsheeting is heavily involved in both Design and Operation. Steady-state flowsheeting is a daily activity in engineering companies and technical services of process plants. Dynamic flowsheeting is involved increasingly in advanced engineering activities, as for example in the design of process control systems, or in Real-Time Optimisation and Computer- Integrated Manufacturing. Material and energy balances remain the most important results in flowsheeting. The stream report displays the way in which the raw materials are transformed in products for given performances of units. In addition, dynamic flowsheeting can mirror the time-variation of the component inventory in different locations, as well as the dynamics of energy streams. The economic analysis, namely the profitability, is the moment of truth of every design. Flowsheet- ing is the only way to solve accurately this problem, because it can account for the spread of all material and energy costs in a flowsheet. 2.1.2 APPLICATIONS OF COMPUTER SIMULATION The current revolution in information technology, as well as the impressive progress in modelling and simulation technology, has a significant impact on Process Engineering. A new paradigm is emerging, in which simulation is involved throughout all the stages of a process life cycle, from ‘idea’, through experiments in laboratory, during scale-up at different levels, to process design and plant operation (Edgar, 2000). 36 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION Figure 2.1 illustrates this new approach. Simulation is placed in the core of the three main engi- neering activities: Research and Development, Design and Operation. The unifying matter is the sci- entific knowledge embedded in universal models, as well as the generic character of computational methods. These activities, apparently disconnected, can share a large number of ‘first-principle’ models, such as thermodynamics, chemical kinetics, transport phenomena, etc. Note that in a larger extent process simulation should include other computer-based activities, such as Molecular Simulation and Computational Fluid Dynamics (CFD). However, in this book, we will limit the presentation to the capabilities offered only by the flowsheeting software, commonly called process simulators. 2.1.2.1 Research and Development Process simulation can guide and minimise the experimental research, but cannot eliminate it. Actu- ally, the calibration of models requires accurate experimental data. It is the experiment that proves the model, and not the opposite! Statistical planning of experiments is nowadays to a large extent obsolete. Instead, the experimental research should take profit from the power of rigorous models incorporated in simulation packages, particularly in the field of thermodynamics. For instance, simple vapour–liquid equilibrium (VLE) experiments in laboratory can be used to increase the reliability of a feasibility study in innovative processes. Conversely, industrial VLE measurements can be used to calibrate the ther- modynamic models incorporated in a simulator when experimental information is not available. The innovation of sustainable processes begins at the laboratory scale. The investigation of the fea- sible design space by simulation can reduce tremendously the experimental effort. Wemay speak about computer-aided experimental research. Modelling and simulation can serve also as the basis for computer-aided scale-up. Some models could pass unchanged from laboratory to the plant scale, while others should be modified to incorporate specific elements to each level. Research and Development Design Operation Simulation FIGURE 2.1 The new paradigm of Process Engineering: simulation as core activity in Research and Development, Design and Operation. 372.1 COMPUTER SIMULATION IN PROCESS ENGINEERING Simulation can explore innovative solutions that are difficult to be investigated experimentally. For example, the integration of process simulation with CFD can replace costly prototypes. The flow- sheeting of virtual plants involving rigorous hydraulic modelling of units is very likely in the next years. 2.1.2.2 Process design Globalisation and sustainable development set challenges for Process Design, such as high efficiency of raw materials and energy, flexibility and responsiveness to market dynamics, safe and clean manufacturing. Clearly, these characteristics must be intrinsic to the conceptual design itself, and not added later by costly modifications. In this respect, process simulation can bring significant contributions, as: • Development of novel sustainable technologies aiming to minimum energy and material requirements, as well as to zero waste and pollutants. • Ensuring absolute safe operation and resiliency by integrating controllability in the conceptual design at early levels. • Permanent improvement of existing technologies, by revamping, retrofitting and debottlenecking. 2.1.2.3 Process operation The advent of Real-Time Optimisation and of Computed-Integrated Manufacturing in the decade 1990 opened large opportunities for applying simulation directly in the manufacturing process. In addition to model-based process control, we may mention preventive maintenance by systematic computer monitoring of equipment performance. The integration of manufacturing with the supplying chain can be realised only by setting up a complex computer-based system, in which simulation plays a central role. Summing up, process simulation is a key factor in ensuring excellence in research, development, design and manufacturing. Table 2.1 illustrates the wide range of applications of modelling and sim- ulation technology in CPI. 2.1.3 SIMULATION OF COMPLEX PLANTS Nowadays, by means of commercial flowsheeting software, it is possible to produce a computerised tool for simulating complex process plants, called plant simulation model (PSM). Figure 2.2 illustrates the simplified structure of a complex plant, as often encountered in basic chemicals or petrochemical industry. Several interconnected recycle loops exist, each of them containing quite a large number of material, energy or process control loops. To facilitate the investigation, the plant is split into sub- flowsheets, named part ‘A’, ‘B’ and ‘C’, which could be analysed independently. Simulation models for sub-flowsheets are tuned and converged separately, and later, merged in a global model. Steady-state PSM serves to support both operation and design, including revamping and debottle- necking (Figure 2.3). A more complex dynamic PSM can be developed to support the design of the process control system and of the startup/shutdown procedure, but also for advanced applications such as operator training and real-time optimisation. The PSM should mirror the behaviour of the plant. The information of interest concerns the flow- sheet units (e.g. the temperature profile along a tubular reactor) and the complex network of material 38 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION 33 21 20 32 222324252627 313028 29 S2 Part “C” Part “B” P Hv Hv Sl Lt 42 Waste 4140 3 1 121110 1514 16 13 Waste 17 76 8 9 4 5 2 B A Raw material Part “A” FIGURE 2.2 Abstraction of a complex plant. Table 2.1 Process Simulation Applications in Chemical Process Industries Chemical Industries Process Applications Oil and gas Offshore exploration, surface treatment, pipeline transport, underground storage, gas processing Refining Gasoline and fuels Petrochemicals Hydrocarbon-based chemicals, methanol, monomers Basic organic chemicals Intermediates, solvents, detergents, dyes Inorganic chemicals Ammonia, sulphuric acid, fertilisers Fine chemicals Pharmaceuticals, cosmetics Biotechnology Food and bio products Metallurgy Steel, aluminium, copper, etc. Polymers Polyethylene, PVC, polystyrene, fibres, etc. Paper and wood Paper pulp Energy Power plants, coal gasification Nuclear industry Waste treatment, safety Environment Water cleaning, biomass valorisation 392.1 COMPUTER SIMULATION IN PROCESS ENGINEERING and energy streams (e.g. the flow rate and purity of a product stream, energy, subject to variations in raw materials, energy utilities and product specifications). Among its objectives, we mention: • Deliver a comprehensive report of material and energy streams; • Determine the correlation between the reaction and separation systems; • Investigate the formation and separation of by-products and impurities; • Support preventive maintenance; • Study how to eliminate wastes and prevent environment pollution; • Evaluate the plant flexibility to changes in feedstock or products’ policy; • Validate the process instrumentation and enhance process safety and control; • Update the process documentation and prepare future investments; and • Optimise the economic performance of the plant. To achieve its goals, a PSM must be calibrated on plant data collected during special organised ‘test- runs’. Additional data reconciliation programs may be used to increase the reliability of the plant data, by minimising the errors in measurements and supplying estimations for non-measured variables. The basic document of a simulation is the stream report. This not only helps to understand the main problems in operation but also offers a quantitative basis for communication between different mem- bers of the plant team, particularly between technical services and plant management. A remarkable feature of a PSM consists of its diagnosis value. For example, inconsistency in the material balance of some components can be explained by a malfunction of an equipment item, so that this can be traced by fault detection (Figure 2.2 with shaded equipment); for example, a badly designed reboiler of the dis- tillation column (unit 24) produces an impurity, which by propagation affects the selectivity of the chemical reactor (unit 20). Plant simulation model (steady state) Operation - Operating window - Sensitivity - Diagnosis - Maintenance Revamping - Heat integration - Equipment upgrade - Waste minimisation Debottlenecking - System analysis - Interactions - Alternative flowsheets - Optimisation - Design Plant simulation model (dynamic) Operation - Start-up - Shut down - Safety Dynamics plantwide control Operator training Real-time optimisation FIGURE 2.3 Applications of steady-state and dynamic plant simulation models. 40 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION Hence, the development of a PSM is the proper approach to deal with industrial simulation prob- lems. The progress in software technology makes possible today the development of integrated steady- state and dynamic models. However, these require significant investment in qualified staff. Recently, generic simulation products have been proposed for applications in refining and petrochemical indus- tries, which can be customised for specific processes. 2.1.4 A HISTORICAL VIEW ON SIMULATION The story of process simulation began in 1966, when Simulation Science, a small company located in Los Angeles, USA, had the idea to commercialise a generic computer program for simulating distil- lation columns. This was the heart of a flowsheeting package, PROCESS, which might be considered the ancestor of process simulators. This software evolved into today’s Pro/II process simulation soft- ware (iom.invensys.com). Three years later, ChemShare (Houston, USA) released DESIGN, a capable flowsheeting program for gas and oil applications. In 1995,WinSim Inc. (www.winsim.com) purchased the rights to the program, which is today marketed as DESIGN II. At that time, the expansion of the refining and petrochemical industries motivated the advent of computer packages. During the 1970s, scientific computation had its golden age. The algorithms used today have deep roots in the methods developed at that time. FORTRAN programming language became the de facto standard among scientists and engineers. The simulations were executed on fast but expensive main- frame systems, to which the user was connected via a remote terminal. The input file was a bunch of cards, with instructions perforated on 80 columns format. Later the input of data became possible by editing a file on an electronic screen, the instructions for job execution being coded via a specific lan- guage based on ‘key-words’. The first world oil crisis in 1973 has greatly stimulated the interest in simulating processes with alter- native rawmaterials, such as coal and biomass. In 1976, the USDepartment of Energy andMassachusetts Institute of Technology jointly launched theAdvanced System for Process Engineering (ASPEN) Project, the root of today’s AspenTech AspenONE software. The advent of high-speed computation systems boosted the business of small companies specialised in modelling and simulation. Moreover, the major engineering bureaus aswell as some largemanufacturing companies in refining and petrochemical indus- tries developed in-house flowsheeting programs.Mostly adopted was the sequential-modular (SM) archi- tecture. However, some simulation packages were based on the equation-oriented (EO) approach, such as SPEEDUP at Imperial College in London (UK) and TISFLOatDSM inTheNetherlands.More generally, scientific computation evolved from individual programs to large packages designed as industrial prod- ucts. Beginning in 1980, several all-purpose steady-state flowsheeting software programs were available on mainframes and distributed in time-sharing on international networks. The worldwide time-sharing computer networks from the 1980–1990s were the precursors of today’s Internet. The Personal Computer arrived in 1982. Although the power of the first PCs was weak for flow- sheeting, the idea of a ‘personal’ tool was strong enough to incite enthusiasts. Thus, ChemStations de- veloped ChemCAD and HyproTech developed HYSYS as simulation packages. The challenge of leaving the elitist environment of mainframes was launched. The arrival of workstations (1985) and of a new multi-tasking operating system, UNIX, is at the origin of a revolution in the scientific computation that continues today. Few scientific software com- panies survived these dramatic changes. The reusability of the old proven FORTRAN routines in a new ‘object-oriented programming’ environment was critical. 412.1 COMPUTER SIMULATION IN PROCESS ENGINEERING http://iom.invensys.com http://www.winsim.com At the beginning of the 1990s, the domination of PC products was a fact. The relative stabilisation in operating systems, dominated nowadays byMicrosoft Windows, enabled the development of new gen- eration of simulation software. The graphical user interface (GUI) became a central part in the software development. The power of the former supercomputers was available on desktops. Today, process simulation is concentrated in a surprisingly low number of systems. On the other hand, the modelling needs of industrial users are much larger than those supplied by generic software capabilities. The solution to this contradiction involves a large cooperation between specialised soft- ware firms and the community of CAPE users. 2.2 STEPS IN A SIMULATION APPROACH The main steps of a simulation workflow and the most important flowsheeting concepts will be intro- duced by means of an industrial example. The flowsheet is rather simple, but it is challenging for a new user. The example will prove that simulation of recycle systems needs careful analysis. The user must be aware of flowsheeting concepts, such as sub-flowsheets, tearing, calculation sequence, convergence, etc. They will be introduced here and examined in more detail in Chapter 3. We adopt here the strategy of a sequential computation of units, which is most used in steady-state flowsheeting. EXAMPLE 2.1 APPROACH OF A SIMULATION PROBLEM Figure 2.4 displays the simplified flowsheet of a low-pressure methanol process (Ullmann, 2001). The user-friendly GUI of a modern simulation packages could tempt a user to enter the simulation problem directly on screen, without pre- liminary analysis. Examine this possibility, as well as the modelling issues that the simulation of this process could raise. Solution. The process operates at pressures between 5 and 10 MPa and temperatures of 200–300 �C. A typical syn- thesis gas obtained by methane steam reforming has the composition 15 vol.% CO, 8 vol.% CO2, 74 vol.% H2 and 3 vol.% CH4. The formation of methanol is described by the equilibrium reaction: CO+2H2>CH3OH; DH300K ¼�90:77kJ=mol (2.1) Because of CO2 presence, there is a second independent equilibrium reaction: CO2 +H2>CO+H2O; DH300K ¼ 41:21kJ=mol (2.2) Typically, the selectivity with modern catalysts is above 99%. The following impurities may be found: higher alcohols, hydrocarbons and waxes, esters, dimethyl ether, ketones. The hot synthesis gas at 3 bar and 700 �C, issued frommethane reforming, enters themethanol synthesis process by two heat exchangers, namely the reboilers of two distillation columns for methanol recovery and purification. After cooling, the fresh gas is compressed, mixed with recycled gas and brought to the required inlet reactor temperature. To ensure an op- timal temperature profile, the cold gas is injected at several points along the reactor. The reactor outlet is sent through a train of heat exchangers for heat recovery by feed preheating and steam generation. After cooling at appropriate temperature, the reaction mixture is submitted to liquid–gas phase split by a flash. The gas phase-containing unconverted reactants is recycled to the reactor. A small amount is purged to prevent inert build-up. The liquid phase containing crude methanol is treated in a sequence of two distillation columns. The first column removes light impurities, while pure methanol is obtained as overhead from the second column with water and heavy impurities in bottoms. 42 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION Now, let us see how to tackle the above process by simulation. Here wewill illustrate the working procedurewith Aspen Plus release 8.4, but the approach is similar for other simulators. We emphasise that introducing the units by employing the GUI, as they appear in the Process Flow Diagram (PFD; Figure 2.4) will not work. The reasons are twofold: (1) not all the physical units have a readily available simulation model. The user has to find workarounds, for example by combining library models; (2) the process involves recycles of material and energy, which require mastering of specific flowsheeting techniques in order to build a resolvable simulationmodel. Therefore, the problem needs careful analysis for converting the PFD into a Process Simulation Diagram (PSD). In a first approach the user has to identify the major sections of the flowsheet. As shown in Figure 2.4 these are (1) Feed conditioning by heat integration, (2) Reaction section, (3) Heat recovery around the reactor, (4) Reactants/product sepa- ration and (5) Product purification. Equally important to mark on the flowsheet are the input and output streams, in this case the synthesis gas, pure methanol, purge, lights and wastewater, respectively. Accordingly, a block diagram of the methanol process can be built, as presented in Figure 2.5. After feed conditioning, the reaction takes place, followed by the phase separation of products from unconverted reactants, which are recycled—a small amount being purged to avoid the accumulation of inert species. The product purification delivers themain product, as well as by-products and impurities. Based on the above analysis, Figure 2.6 illustrates the flowsheet for simulation, called here after PSD. It may be ob- served that PSD is different from the PFD. However, the PSD should capture the essential features of the flowsheet in view of getting reliable mass and heat balances around the key units and for the whole process. A first conclusion might be drawn: a preliminary analysis of the technological flowsheet must be done to translate it in a diagram compatible with the capabilities of the simulator. Simulation units (blocks) are employed for building a PSD. These may correspond directly to the physical units, or may be employed only asmodelling tools. In addition, setting up a PSD needs the interpretation of a PFD, in order to simplify the computational procedure without altering the goal. In this case, a major assumption is neglecting the heat recovery Continued Recycle gas W at er c c c c c d dd Wastwater Light ends Synthesis gas Purge gas Pure methanol d e ab g h f Reaction Product purification Reactants/products separation Temperature conditioning Heat integration Pressure conditioning Input Output Output Light ends Purge gas Pure methanol Wastwater Synthesis gas FIGURE 2.4 Flowsheet for the low-pressure process for methanol (Ullmann, 2001). 432.2 STEPS IN A SIMULATION APPROACH construction around the chemical reactor, which can be rebuilt after closing the energy balance of the recycle loop. The comments that follow will serve also for introducing some key concepts in flowsheeting techniques, which will be amply discussed in Chapter 3. Before entering the PSD, two preliminary phases has to be completed: entering the components andmodelling of phys- ical properties. In this case, all the components are chemically defined, but there are cases where some species non- registered in the software database, for example an impurity, should be introduced as user-defined. The second aspect implies in the first place the selection of a thermodynamic model. We may observe that the process takes place at two different pressure levels, higher pressure for methanol synthesis and lower pressure for methanol pu- rification. For the first part a standard selection is an equation-of-state (EOS) model, as for example Peng–Robinson or Soave–Redlich–Kwong. For the second part a liquid activity model is suited, such as Wilson, Uniquac or NRTL, but ca- pable of handling supercritical gaseous components, such as CO, CO2, CH4 and H2, by means of Henry coefficients. An- other option is to use for the whole flowsheet a modified EOS model capable of handling both polar and non-polar components. The thermodynamic modelling is of paramount importance for the reliability of results. For this reason, two chapters of this book, 5 and 6, have been allocated for presenting the key thermodynamic issues in process simulation. Entering the flowsheet is piloted by the GUI. The input in the process is the hot synthesis gas (stream S-GAS), which is firstly cooled in the heat exchangersHEX-1 andHEX-2by supplyingheat to the distillation columnsDIST-2 andDIST-1, and then in a final cooler HEX-3. This operation is part of the heat integration, which can be solved later. Note that the trans- mission of energy streams between the units permits a considerable simplification of the computational sequence. Thus, in a first step the heat exchangers are treated as simple coolers with specified outlet temperatures. It is important that the tem- perature after the last cooler be sufficiently low, here 77 �C, in order to comply with the compression operation that follows. The gas is compressed in the multi-stage compressor COMP-1 to 45 bar and 125 �C, mixed with the RECYCLE stream, and compressed further in the single-stage compressor COMP-2 at 50 bar. It is advisable to place before the reactor a heat ex- changer in order to ensure the proper temperature for starting the reaction. Later a more rigorous approach based on heat integration techniques, explained in Chapter 10, will solve the optimal heat recovery problem, as shown in Figure 2.4. At this point we reach the most important part of the flowsheet, the chemical reaction section. Several options are avail- able for the reactor modelling (see Chapter 8 for more details): (a) Stoichiometric reactor is the simplest one, requiring only information about reaction stoichiometry and reaction extents. (b) Equilibrium reactor is suitable when the chemical reactors are designed to operate close to chemical equilibrium, as in this case. Process simulators provide two different models for calculating the chemical equilibrium: – Equilibrium constant models by introducing explicitly the chemical reactions. – Minimising the Gibbs free energy of the reactionmixture. This method needs only specifying the chemical species at equilibrium, but it can give unrealistic results, when some species are in fact subject to kinetic controlled reactions. (c) When the reaction kinetics is reliable, the user can employ kinetic reactor models. Conditioning - Pressure - Temperature Reaction Reactants/ product separation Product purification Feed Recycle Purge Product By-products FIGURE 2.5 Block diagram of the methanol process. 44 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION COOL-2 Flash REACT-S React FSPLIT B22 HEX-1 COMP-1 COMP-2 DIST-1 DIST-2 V1 HEX-2 HEX-3 ROUT 3 Gas Liq ROUT-4 Rin 13 Purge Recycle S-GAS 1A 2 Lights Met+wat Methanol Water 1B 1 H2 H1 HEX 5 77 �C 45 bar 125 �C 50 bar CH3OH conversion = 0.02 Split fraction = 0.1 35 �C 25 bar Equilibrium temperature = 271.4 �C 15 stages B:F = 0.99 R = 2.47 kmol/h 20 stages D:F = 0.753 RR = 1.5 P = 3 bar T = 700 �C CO : 30 kmol/h H2 : 148 kmol/h CO2: 16 kmol/h CH4: 6 kmol/h 220 �C FIGURE 2.6 Process simulation diagram of the low-pressure methanol synthesis. In this exercise, the reactor is modelled by the equilibrium reactor REACT, followed by the stoichiometric reactor REACT-S. The first unit handles the main reactions (2.1) and (2.2) occurring close to equilibrium, while the second de- scribes reactions leading to by-products and impurities. In this case only the formation of dimethyl ether from methanol dehydration is accounted for. The reactor effluent is cooled and sent to the gas–liquid separation unit (FLASH). This simple unit has the important task of ensuring a sharp separation of the gas and liquid streams, firstly for recycling the unconverted reactants, secondly for separating the crude product. Therefore, the temperature and pressure of the flash should be carefully specified such that this goal is fulfilled. In this case an acceptable result is obtained by temperature of 35 �C, ensured by cooling water and pressure of 25 bar. The stream GAS is recycled, from which a fraction PURGE is withdrawn. Note that the presence of a stream purge is compulsory when some components tend to accumulate in a recycle because no exit point. After solving the recycle loop, the stream LIQ is sent to the first distillation column DIST-1, where light by-products are removed, and then to the second distillation column DIST-2 that delivers the methanol product in top and wastewater in bottoms. The specification phase for the most units from Figure 2.6 is straightforward. In a more general approach, this topic can be solved by means of degrees of freedom (DOF) concept. However, the chemical reactor needs a deeper analysis, pref- erably in a separate file, as shown in Figure 2.7. It should be remarked that methanol synthesis is an exothermal, reversible reaction for which the equilibrium con- version decreases with temperature, according to the Le Chatelier principle. If the temperature is too low, the reaction is slow. On the other hand, the conversion is low at higher temperature, because of the equilibrium limitations. Therefore, there is an optimum temperature profile along the reactor, which explains the solution adopted in Figure 2.4. The reactor consists of several catalyst beds adiabatically operated. Accordingly, the reactor inlet is split into several streams. The first inlet stream is heated, while the other colder streams are injected between the catalyst beds in order to decrease the tem- perature and bring the reaction mixture away from equilibrium. Figure 2.7 details the simulation model. The stream Q1 to the first reactor bed is heated to 220 �C, while streams Q2, Q3 and Q4 are fed at 107 �C, temperature set by the heat exchanger COOL-1. In order to find ratio of splitting, we make use of flowsheet controllers, which are simulation tools similar to process controllers. They may be used for feedback and feedforward control. These are implemented as ‘Design specification’ and ‘Calculator’ blocks in Aspen Plus, or ‘Adjust’ and ‘Set’ blocks in HYSYS. This topic is developed in Chapter 3. By means of three design specifications (T2, T3 and T4), the flow rates of the quench streams Q2, Q3 and Q4 are tuned such that bed-inlet temperatures of 230 �C are achieved. With respect to simulation strategy, there are several possibilities depending on the experience of user. In the case of modular-sequential simulators, mastering the concepts of tearing of streams and computational sequence is essential, and will be developed in Chapter 3. For beginners, we recommend developing the simulation step-by-step by adding few sim- ulation units at a time, running the simulation and carefully checking the results after each step. This approach is allowed by some simulators, where the user can run stepwise the simulation. We totally discourage the optimistic approach in which the user draws the entire flowsheet, specifies the inlet streams and the units, and then hits the ‘run’ button. Raising the number of iterations will not help either. Another observation concerns the relationship between how a unit is specified and how easy it is to be solved. The easiest specification is the ratingmode, in which the inlet stream and the unit sizing or performance is given, the simulator being asked to calculate the outlet stream (Figure 2.8). A more difficult problem arises in the design mode, when the inlet and outlet streams are known. Usually, solving for the inlet is a very difficult problem. This calculation mode is against the principle of sequential-modular simulation approach, but nevertheless allowed by some simulation software as long as the DOF are satisfied. In these cases the use of feedback controllers is recommended. On the contrary, this specification type is straightforward while working in equation-solving mode, although an initial guess for the unknowns is usually obtained by a SM approach. With respect to Figure 2.7, the first difficulty arises when specifying the SPLIT unit (streams Q2, Q3 and Q4 split fractions). Initially, equal fractions of 0.25 are guessed. At this point, it is wise to add three feedback controllers (design specifications), denoted here by T2, T3 and T4, which manipulate the split fraction of streams Q2, Q3 and Q4 in order to achieve the specified bed-inlet temperatures of 230 �C. The solution of these three control structures may be achieved simultaneously. From our experience, the fastest convergence is obtained with the Broyden method (see Chapter 3). The equilibrium model used for the main reactions is not capable of describing the whole chemistry, particularly the formation of light by-products. Therefore, the stoichiometric reactor REACT-S is introduced, assuming that 2% of the methanol formed is transformed to dimethyl ether. The development of the simulation proceeds without difficulty until the RECYCLE stream is obtained. Since the feed contains small amounts of methane, an inert, this should have a way 46 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION Design-spec T2 Design-spec T3 Design-spec T4 REACT-1 REACT-2 REACT-3 REAC T-4HEAT-1 SPLIT M2 M3 M4 COOL-2 Flash REAC T-S FSPLIT Split-2COMP-2 COOL-1 COMP-1B4 RIN-1 ROUT-1 A RIN-2 ROUT-2 B RIN-3 ROUT-3 RIN-4 ROUT-4 Q1 RIN-AUX Q2 Q3 Q4 ROUT 3 Gas Liq Purge Recycle 2 RIN 1S-GAS 45 bar 50 bar 230 �C230 �C 220 �C 125 �C 77 �C 107 �C 35 �C 230 �C 25 bar Split fraction = 0.1 CH3OH conversion = 0.02 FIGURE 2.7 Process simulation diagram of the multi-bed, adiabatic methanol synthesis reactor. to leave the plant. The purge fraction is a matter of optimisation: less reactant is lost by smaller purge, but the price to pay is higher recycling cost. Here, a purge fraction of 0.1 gives acceptable results. The convergence of the recycle loops may raise major difficulties in simulation. We advise firstly saving the file. The recycle loop is closed by connecting the RECYCLE stream to the inlet of the compressor COMP-2. Choosing as sequencing option ‘Design specification nesting – with tear streams’, the user may attempt to run the simulation. The question is which tear stream? The software may make itself a selection by a topological analysis (see Chapter 3), or the user may choose the streams she/he may estimate the best. In this case choosing RIN seems a good idea. For the problem at hand, the simulation eventually converges, but only after increasing slightly the number of allowed flowsheet evaluations from the default value. The reason is that one component, dimethyl ether, accumulates slowly to an equilibrium value. Thus, we advise examining the convergence history before increasing the number of iterations. If the convergence is monotone, then some more it- erations help solve the problem. If the convergence is hieratic, or extremely slow, other reasons should be searched. We want to emphasise that this simple procedure for closing a recycle loop does not always work. Often, one has to provide good initialisation of the tear stream(s) or to employ complex design specifications, manip- ulating the plant inlet streams in order to achieve control of the flowsheet mass balance, as shown in Example 2.2. This issue is deeply connected to the topic of Plantwide Control, and will be discussed in detail in Chapter 15. Moreover, because the convergence process implies repeated evaluation through the recycle loop, the user should attempt this only after getting robust convergence of units for a wide range of input streams. Key results of this exercise are the temperature atwhich equilibrium is calculated (271.7 �C) and the condition (flow rate, temperature, pressure, composition) of the liquid stream LIQ which enters the separation section. These are presented in Table 2.2 (the quench streams Q2, Q3 andQ4 can be easily calculated bymass balance around the mixersM2,M3 andM4). We can now proceed with the liquid separation system. At this point, the user may extend the simulation shown in Figure 2.7 by adding the distillation columns, or he can return to PSD from Figure 2.6, which now can be easily specified and converged. We will choose the latter option. The insertion of the valve V-1 for pressure reduction is necessary. Then, the problem is simulating the two distillation columns. The models for the separation units are of two types: design or rating. In design mode, the unit is defined by its performance, sizing characteristics being computed by shortcut methods. In rating mode, the performance of the unit is computed for a given design. In a first attempt the design mode is more suitable: specify the desired separation and ask for the number of theoretical stages and the reflux ratio. In this way, the desired outputs are always achieved. This DS Unit Unit Unit Rating mode Easy Design mode Difficult Unknown inlet Very difficult Use of design specification to solve a difficult problem. The rating-mode provides an initial guess. FIGURE 2.8 Different ways of specifying a simulation problem. 48 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION Table 2.2 Stream Table of the Multi-Bed Adiabatic Reactor for Methanol Synthesis S-GAS RIN-1 ROUT-1 RIN-2 ROUT-2 RIN-3 ROUT-3 RIN-4 ROUT-4 ROUT RECYCLE PURGE LIQ Temp (�C) 700 220 291.8 230 281.4 230 275.5 230 271.7 272 31.5 31.5 31.5 Press (bar) 3 50 50 50 50 50 50 50 50 50 25 25 25 Mole flow (kmol/h) 200 294.7 278.0 425.7 409.7 592.4 572.5 799.7 774.9 774.9 652.4 72.5 50.0 CO 30 20.20 15.87 25.99 20.02 32.55 25.17 40.75 31.60 31.60 28.44 3.16 0.003 H2 148 233.4 212.6 329.5 311.6 456.3 433.8 613.7 585.6 585.6 527.0 58.56 0.025 CH3OH – 1.78 10.16 11.05 19.03 20.13 30.10 31.47 43.89 43.02 5.143 0.571 37.32 CO2 16 17.53 13.48 22.27 20.26 31.12 28.533 42.04 38.77 38.77 34.68 3.854 0.232 H2O – 0.15 4.192 4.267 6.278 6.371 8.957 9.073 12.35 12.79 0.434 0.048 12.31 DME – 1.032 1.032 1.549 1.549 2.189 2.189 2.984 2.984 3.423 2.984 0.332 0.107 CH4 6 20.63 20.63 30.97 30.97 43.76 43.76 59.67 59.67 59.67 53.67 5.963 0.037 is important especially when the product streams are recycled. However, a rating model would be necessary in order to model the heat integration between the synthesis gas cooling and the reboilers of the distillation columns (Figure 2.4). As the separation of the dimethyl ether/methanol/water mixture is easy, we proceed by considering the rigorous RADFRAC rating model. The NRTL model with Henry components is used for the distillation columns DIST-1 and DIST-2. Spec- ifying the distillation units is easy: number of trays, feed tray, pressure profile, distillate (or distillate to feed ratio) and reflux ratio. Again, we recommend adding one unit at a time, selecting appropriate specifications and carefully checking the results. For example, robust convergence of the column DIST-1 is obtained by using reflux flow rate instead of reflux ratio, since the top vapour distillate is very small, depending on the separation in the unit FLASH. Using ratio specifica- tions, such as distillate to feed, ensure easier convergence. We also recommend checking the temperature and composition profiles, in order to follow the separation progress along the trays and detect inappropriate sizing. The specifications from Table 2.3 lead to 35 �C in the top of the first column, while the molar purities of the methanol and water streams are 99.8% and 99.2%, respectively. Now, we can also include in our simulationmodel the heat integration between the feed stream S_GAS and the reboilers of the two distillation columns. Again, a step-by-step approach is recommended: First, cooling duties of heat exchangers HEX-1 and HEX-2 are specified as being equal to reboiler duties of columns DIST-2 (0.788 Gcal/h) and DIST-1 (0.112 Gcal/h), respectively. Then, the simulation is run to initialise the heat streams H1 and H2, which are afterwards connected to the heat exchangers HEX-1 and HEX-2. Now, we may leave the steering wheel to the simulator. The question that rises again is: where the simulation should start and what would be the order of computation of units? The topological analysis will determine the calculation sequence by identifying the tear streams that must be initialised to allow a sequential computation of units. In this case, the simulator chooses streams RECYCLE, H1 and H2 as tear streams and performs the calculations in the following order: $OLVER01 (converge tear streams RECYCLE H1 H2) HEX-1!HEX-2!HEX-3!COMP-1!COMP-2! HEX!REACT!REACT-S!COOL-2!FLASH! B22!V1!DIST-1!DIST-2 RETURN $OLVER01 After a successful run and before saving and closing the simulation, it is wise to ‘reconcile’ the tear streams, which saves the stream results to be used as initial guesses for the next run. As an extension of this exercise the user may try to incorporate the heat saving solution proposed by the flowsheet 2.4 – but not to skip chapter 3. Approach of a simulation problem Example 2.1 points out the methodological levels for setting-up a simulation problem (Figure 2.9). These are briefly described hereafter. Table 2.3 Specification of the Distillation Columns DIST-1 DIST-2 Number of stages 15 20 Feed stage 7 10 Distillate to feed ratio – 0.753 Bottoms to feed ratio 0.99 – Reflux (kmol/h) 2.47 – Reflux ratio – 1.5 Pressure (bar) 1.1 1.1 50 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION 1 Problem analysis A real PFD must be translated in a scheme compatible with the software capabilities and with the sim- ulation goals. The flowsheet scheme built up for simulation purposes will be called in this book PSD. In general, PSD is different from PFD. Simple units may be lumped together, for example a pump fol- lowed by a heat exchanger using steam as utility may be modelled as a heater with specified duty and outlet pressure. Contrary, complex units, such as distillation columns or chemical reactors, may need to be simulated as small flowsheets. Hence, a preliminary problem analysis is necessary. The steps in defining a simulation problem are as follows: 1. Problem analysis 2.Input 3.Execution PFD Simulation model Components Properties Specifications PSD Solution options Initialization Convergence report Stream report Unit report Tables, plots 4. Results analysis - Convergence, reliability - Sensitivity, case studies - Optimisation S im ul at io n up da te C on ve rg en ce tr ou bl es ho ot in g P ro ce ss d es ig n FIGURE 2.9 Methodological levels in steady-state simulation. 512.2 STEPS IN A SIMULATION APPROACH – Analyse the chemistry in order to identify the components (chemical species, petroleum fractions) to be included in the simulation. – Convert PFD in PSD. Choose simulation models for each unit/group of units. Split the flowsheet in several sub-flowsheets, if necessary. – Analyse the process conditions to decide the appropriate thermodynamic model(s). Look at the global flowsheet, sub-flowsheets and key units. – Analyse the specification mode (degrees or freedom) of the flowsheet. This is of outmost importance and will be discussed in more detail in Example 2.2. 2 Input The input of a flowsheeting problem depends on the software technology. This activity is normally supported by a GUI. A part of the input data is available from Problem analysis: – Select the components, from standard database or user defined – Select the thermodynamic models. Check model parameters – Draw the flowsheet – Specify the input streams – Specify the units (DOF analysis) and initialise the difficult units A second category of inputs are related to convergence options: – Determine the computational sequence and choose the solution algorithm – Initialise the tear streams A simulation model is obtained after the input step is completed. 3 Execution The simulation is successful when the convergence criteria are fulfilled both at the flowsheet and at the unit level. A simulation delivers a large amount of results. The most important are: – Stream report (material and heat balance), including flowsheet convergence report – Unit report, including material and heat balance, as well as unit convergence report – Rating performances of units – Tables and graphs of physical properties The graphical presentation of results may take various forms. Generally, advanced software provides its own analysis tools, but the exchange of data with all-purpose spreadsheets is usually available. De- tailed results, such as internal flows or tables of properties, may be exported to specialised design packages. 4 Results analysis Analysis should start by checking the convergence and the reliability of the results: – If the simulation converges, verify the mass balance of the key units and of the whole flowsheet. Look at the flow rate of recycle streams, they should have reasonable values. Check the flow rates and purities of product streams. 52 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION – If the simulation does not converge or the results appear unreliable, carefully examine the convergence history in order to pinpoint the reasons for lack of convergence. Look at all simulation messages, including warnings. Troubleshooting actions include: – Check/revise the components specification and the properties calculation methods. – Consider the possibility of using simpler but more robust models in the PSD. – Check the specifications, taking into account the whole flowsheet. In general, avoid specifying the process-outlet streams. Instead, specify recycle flow rates or product to feed ratios. – Build the flowsheet gradually and check the results after each step. – Provide better initialisations of the tear streams and difficult units. – Check the convergence algorithms and parameters, and change them if necessary. – Check the convergence errors and the bounds of variables. After the user is confident that reliable results have been obtained, flowsheeting analysis tools can be employed to get more value from the simulation results. The most used is the sensitivity analysis. This consists usually of recording the variation of some ‘sampled variables’ as function of ‘manipulated vari- ables’. The interpretation of results can be exploited directly, as trends, correlation or pre-optimisation. Case studies can be performed to investigate combinations (scenarios) of several flowsheet variables. Finally, the simulationworkmay be refined bymulti-variable optimisation. As a result, the designer can suggest improvements/developments of the original PFD, and a new simulation cycle may start. EXAMPLE 2.2 DOF AND FLOWSHEET SPECIFICATION The apparently simple problem described in this example will emphasise the importance of carefully analysing the process before implementing a simulation model in flowsheeting software. The example illustrates how a preliminary mass balance can be obtained before building a process simulation by using general-purpose software. Moreover, it demonstrates several aspects of process simulation that were not covered by Example 2.1: initial use of stoichiometric reactor models, followed later by kinetic models, setting separation targets, an- alyzing the flowsheet DOF, employing feedback controllers (design specifications) for converging reactor–separation– recycle problems. Problem statement. Acetals are oxygenated compounds, obtained from the reaction between an aldehyde and an al- cohol, which have been proposed to be used as combustion enhancers for biodiesel. The requirement of this exercise is to perform the conceptual design of a plant producing 10 kmol/h of 1,1-diethoxy butane, by the reaction of butanal and eth- anol, and to build a simulation model. The chemical reaction is: 2C2H5OH+C3H7�CH¼O>C3H7�CH OC2H5ð Þ2 +H2O Eð Þ Bð Þ Að Þ Wð Þ The reaction takes place in liquid phase at 40 �C, in the presence of homogeneous or heterogeneous acid catalysts. When Amberlyst 47 is used, the reaction rate is given by the following relationship (Agirre et al., 2010): r¼ k1c2EcB�k2cAcW The reaction constants follow Arrhenius law: k1 ¼ 1:08exp � 35,505 8:31�T � � m3ð Þ3 kmol2 skgcat and k2 ¼ 1:06�105exp � 59,752 8:31�T � � m3ð Þ2 kmolskgcat Continued 532.2 STEPS IN A SIMULATION APPROACH The information given up to this point is enough to perform a conceptual design of the plant using Aspen Plus as flow- sheeting software. The reader is encouraged to test his/her skills before reading further, and to use the rest of this section to confirm the approach and the results. Solution. The equilibrium constant at the reaction temperature 313 K is: K¼ k1 k2 ¼ 0:1139 m 3 kmol Because the reaction is equilibrium limited, complete reactants conversion cannot be achieved. Hence, a separation section and recycle of reactants are necessary. As first separation alternative, we consider distillation. Relevant boiling points at atmospheric pressure are presented in Table 2.4. Because the reactants are the lightest components in the mixture, they will be separated and recycled together. Moreover, as water forms several light-boiling azeotropes with the reactants, it will also be present in the recycle stream. A conceptual flowsheet is presented in Figure 2.10. Note that the product purification, which involves breaking the water–1,1-diethoxy butane azeotrope, will not be considered in this exercise. In the following, we will develop a simplified linear model of the Mixer–Reactor–Column 1–Recycle part of the plant. This will give a good approximation of the plant mass balance. More importantly, analysis of the linear model will lead to valuable insight concerning the fulfilment of the DOF, which will prove to be very useful for building the rigorous Aspen Plus simulation model. In the first step, we write the mass balance equations for each chemical species and each unit of the plant. We will denote by FK, j the molar flow rate of species K in the stream j. nK are the stoichiometric coefficients and x is the reaction extent (kmol/h). Mixer: FK,0�FK,1 +FK,3 ¼ 0, K¼E,B,A,W (2.3) Reactor: FK,1�FK,2 + nKx¼ 0, K¼E,B,A,W (2.4) Column 1: FK,2�FK,3�FK,4 ¼ 0, K¼E,B,A,W (2.5) Note that total mass balances (for all species over one unit, and for each species over the entire plant) are not necessary, as they are not independent equations. In the next step, we set targets for the flowsheet units. Table 2.4 Boiling Points of Components Temperature (�C) Composition (Molar) Destination 69.55 Butanal (0.748)–Water (0.252) 73.15 Ethanol (0.303)–Butanal (0.697) 74.88 Butanal Recycle 78.15 Ethanol (0.895)–Water (0.105) 78.31 Ethanol 94.06 Water (0.881)–1,1-diethoxy butane (0.119) 100.0 Water Product purification 153.69 1,1-Diethoxy butane 54 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION Reactor: The conversion of the limiting reactant, which cannot exceed the equilibrium conversion, is a legitimate re- actor target. We choose to use an excess of ethanol, for exampleM¼FE,1/FB,1¼4 (twice the stoichiometric amount). The following equation can be used to calculate the equilibrium conversion: Kc�cW,eqcA,eq cB, eqc2E,eq ¼ 0 (2.6) where cK,eq ¼ nK,eq Veq (2.7) nK, eq ¼ nK, in + nKxeq, K¼E,B,A,W (2.8) Veq ¼ X K¼E,B,A,W Vm,KnK,eq (2.9) XB, eq ¼ xeq nB, in (2.10) The molar volumes Vm,K (in 10 �3 m3/kmol) are: 89.568 (butanal), 58.173 (ethanol), 150.342 (1,1-diethoxy butane) and 18.05 (water). As conversion is an intensive variable, we can do the calculations for any initial number of moles, for ex- ample nE,in¼4 kmol, nB,in¼1 kmol, nW,in¼0, nA,in¼0. The equations can be solved with any suitable software. Figure 2.11 presents a possible Matlab implementation. Running the program gives the equilibrium conversion, XB,eq¼0.7589. We choose XB¼0.5. Therefore, the equation describing the target performance of the reactor becomes: x�XBFB,1 ¼ 0 (2.11) Continued Mixer Reactor Column 1 Product purification Ethanol (E) Butanal (B) Ethanol, butanal, water Water (W) 1,1 Diethoxy butane (A) Water 1,1 Diethoxy butane 1 2 3 4 5 6B + 2E A + W FIGURE 2.10 Conceptual design of the 1,1-diethoxy butane plant. 552.2 STEPS IN A SIMULATION APPROACH Column 1: The separation targets can be set in terms of component recoveries. Thus, butanal and ethanol will be found in the distillate stream 3, while 1,1-diethoxy butane will leave the column with the bottoms stream 4. However, due to azeotropes formation, water will be distributed between the two streams. A rough estimation of the amount of water in distillate can be obtained by considering the azeotropes as pseudo-components: FW,3 ¼ 0:252 0:748 FB,3 + 0:105 0:895 FE,3 (2.12) Therefore, the column separation targets are given by: FE,2�FE,3 ¼ 0 (2.13) FB,2�FB,3 ¼ 0 (2.14) FA,2�FA,4 ¼ 0 (2.15) 0:34FB,3 + 0:12FE,3�FW,3 ¼ 0 (2.16) At this point, we count 21 variables (flow rates FK, j, K¼E, B,W, A; j¼0 . . . 4 and reaction extent x) and 17 equations (3�4mass balance, 1 reactor target, 4 separation targets). Therefore, four more specifications are needed, of which two are obvious: no water and no 1,1-diethoxy butane are fed to the process: FW,0 ¼ 0 (2.17) FA,0 ¼ 0 (2.18) The remaining two specifications are more subtle. One could employ the problem requirement which was not used yet, namely the throughput, in two different ways: specify the flow rate of 1,1-diethoxy butane in the product stream 4, or equivalently the flow rate of butanal at plant inlet (because each mole of butanal leads to 1 mol of 1,1-diethoxy butane). We chose the latter one, being closer to the way the flowsheet is built and specified (from fresh feeds to plant products): % main program clear all close all global n10 n20 n30 n40 M = 4; % reactants ratio n20= 1; n10= M* n20; n30= 0; n40= 0; x0= 0.5; %initial guess x = fzero(@echil,x0); % solve the equations conv = x / n20; % calculate and print conv % the conversion function y = echil(x) global n10 n20 n30 n40 Kc = 0.1139; %m3/kmol, 313 K V1m = 58.173e-3; %molar volume,etanol V2m = 89.568e-3; %butanal V3m = 150.342e-3; %acetal V4m = 18.05e-3; %water n1 = n10- 2*x; V1= V1m*n1; n2 = n20- x; V2=V2m*n2; n3 = n30+ x; V3= V3m*n3; n4 = n40+ x; V4= V4m*n4; V = V1+ V2+ V3+V4; C1 = n1 / V; C2 = n2 / V ;C3 = n3 / V; C4 = n4 / V; y = Kc - C3*C4/C1/C1/C2; FIGURE 2.11 Matlab code for calculating the equilibrium conversion. 56 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION FB,0 ¼ 10kmol=h (2.19) Finally, the inexperienced designer might reason that the reaction requires 2 mol of ethanol for each mole of butanal, therefore: FE,0 ¼ 20kmol=h (2.20) In conclusion, the model of the plant contains the mass balance of each species over each unit, Equations (2.3)–(2.5), the performance targets for the reactor (2.11) and distillation column (2.13)–(2.16), and feed conditions (2.17)–(2.20). The model unknowns are the species flow rates in each stream and the reaction extent. The equations are linear and can be written as: Ax¼ b (2.21) where the vector b contains the right-hand side of the equations, the only non-zero entries corresponding to Equations (2.19) and (2.20). A solution of the linear system (2.21) can be easily obtained in Matlab by defining the matrix A and the vector b, fol- lowed by the simple statement x¼A\b. Unfortunately, Matlab complains about matrix A being singular and returns a so- lution x where most entries are Inf or NaN (Infinity or Not-A-Number). Equation (2.20) is the reason why the model cannot be solved. In fact, the information that ‘one mole of butanal reacts with two moles of ethanol’ is already included in the stoichiometric matrix n. Therefore, Equation (2.20) is redundant. The correct way to complete the problem description is specifying, directly or indirectly, one flow rate in the recycle loop, for example the ratio between reactants at reactor inlet: FE,1�MFB,1 ¼ 0 (2.22) with M>2 (excess of ethanol), for example M¼4. The results obtained for the correct problem setting are presented in Table 2.5. A final remark concerns development of the simulation with the favourite package. We recommend starting with the chemical reactor, using results from Table 2.5 for specification of the reactor-inlet stream. After introducing the kinetic data, one can adjust the reactor volume such that a conversion around XB¼0.5 is obtained. The distillation column should adjust the distillate rate such that the amount of ethanol in the bottoms stream is limited (do not specify the bottoms stream!). The distillate is recycled and mixed with the fresh feeds. Add a design specification which adjusts the flow rate of fresh ethanol such that the ratio ethanol/butanal in the mixed stream has the desired value, for example M¼4. Then close the recycle. Table 2.6 presents results obtained from a rigorous Aspen Plus simulation. The reactor uses 25 kg of catalyst. The dis- tillation column has 25 stages with feed on stage 13 and is operated at reflux ratio R¼2, the distillate rateD¼83.80 kmol/h being adjusted such that the ethanol mole fraction in the ethanol+water bottoms mixture is 0.002. The agreement with the simplified model (Table 2.5) is excellent. Continued Table 2.5 Mass Balance of the 1,1-Diethoxy Butane Plant (Simplified Model) Species Flow Rates (kmol/h) Feed (0) Reactor Inlet (1) Reactor Outlet (2) Recycle (3) Product (4) Ethanol (E) 20.0 80.0 60.0 60.0 0.0 Butanal (B) 10.0 20.0 10.0 10.0 0.0 1,1-Diethoxy butane (A) 0.0 0.0 10.0 0.0 10.0 Water (W) 0.0 10.6 20.6 10.6 10.0 572.2 STEPS IN A SIMULATION APPROACH As an extension of this exercise, the user may consider the case when the fresh ethanol is available as ethanol–water mixture with a composition that is close to the azeotropic one. 2.3 ARCHITECTURE OF FLOWSHEETING SOFTWARE 2.3.1 COMPUTATION STRATEGY The architecture of a flowsheeting software is determined by the strategy of computation. Three basic approaches have been developed over the years: – Sequential-Modular (SM) – Equation-Oriented (EO) – Simultaneous-Modular In SM architecture, the computation takes place unit-by-unit following a calculation sequence. The sequence contains unit operations and convergence blocks. Each unit operation or convergence block can be solved individually, provided the input streams are given and the units are correctly specified. A convergence block also contains unit operations and other convergence blocks, but these can be solved only simultaneously. Convergence blocks arise due to recycles and design specifications. When recycles are involved, the solution strategy identifies a set of tear streams. If initial guesses of the tear streams are provided, the units from the convergence block can be solved sequentially. Then, the guesses are checked and updated by an appropriate algorithm, until convergence is obtained. Similarly, the manipulated variable from a design specification is changed until the specification of the controlled variable is fulfilled. The SM architecture was the first used in flowsheeting, but still dominates the technology of steady-state simulation. Among the advantages of the SM architecture, we may cite: – Modular development of capabilities – Easy programming and maintenance – Easy control of convergence, both at the units and flowsheet level There are also disadvantages, as for example: – Need for topological analysis and systematic initialisation of tear streams – Difficulty to treat more complex computation sequences, such as nested loops or simultaneous flowsheet and design specification loops Table 2.6 Mass Balance of the 1,1-Diethoxy Butane Plant (Aspen Plus Simulation) Species Flow Rates (kmol/h) Feed (0) Reactor Inlet (1) Reactor Outlet (2) Recycle (3) Product (4) Ethanol (E) 20.02 80.48 60.48 60.46 0.0 Butanal (B) 10.0 20.12 10.12 10.12 0.0 1,1-Diethoxy butane (A) 0.0 0.0 10.0 0.0 10.0 Water (W) 0.0 13.21 23.21 13.21 10.0 58 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION – Difficulty to treat specifications regarding internal unit (block) variables – Rigid direction of computation, normally ‘outputs from inputs’ – Not well suited for dynamic simulation of systems with recycles Some modifications have been proposed to improve the flow of information and avoid redundant com- putations. Among these, we may mention the bi-directional transmission of information implemented in HYSYS™. In the EO approach, all the modelling equations are assembled in a large system producing Non- linear Algebraic Equations in steady-state simulation, and Differential and Algebraic Equations in dy- namic simulation. Thus, the solution is obtained by solving simultaneously all the modelling equations. Among the advantages of the equation-solving architecture, we may mention: – Flexible environment for specifications, which may be inputs, outputs or internal unit (block) variables – Better treatment of recycles and no need for tear streams – Suitable for an object-oriented modelling approach However, there are also substantial drawbacks, as: – More programming effort – Need of providing good initial guesses for the unknowns. This initialisation is often done by an SM run of a closely related problem. – Need of substantial computing resources, but this is less and less a problem – Difficulties in handling large DAE (differential-algebraic equations) systems – Difficult convergence follow-up and debugging In Simultaneous-Modular approach, the solution strategy is a combination of SM and EO approaches. Rigorous models are used at units’ level, which are solved sequentially, while linear models are used at flowsheet level, solved globally. The linear models are updated based on results obtained with rigorous models. This architecture has been experimented in some academic products. It may be concluded that the SM approach maintains a dominant position in steady-state simulation. The EO approach has proved its potential in dynamic simulation, and real-time optimisation. The solution for the future generations of flowsheeting software seems to be a fusion of these strategies. 2.3.2 SEQUENTIAL MODULAR APPROACH The SM approach is mostly used in steady-state flowsheeting software, among which we may cite as major products Aspen Plus, ChemCAD, HYSYS, UniSim, Pro/II, ProSim, SimSci and WinSim. However, there are some dynamic simulators built on this architecture, the most popular being HYSYS. The basic element in a modular simulator is the unit operation model. A simulation model contains non-linear algebraic and differential equations describing the conservation of mass, energy and mo- mentum. These can be represented in the following condensed form: f x, u, d, pð Þ¼ 0 (2.23) y¼ g xð Þ (2.24) where the following notations was used (Figure 2.12): 592.3 ARCHITECTURE OF FLOWSHEETING SOFTWARE – x, internal (state) variables, such as temperatures, pressures, concentrations; – u, input variables, connecting the unit to other upstream units. Typical examples are the inlet streams specified as temperature, pressure, flow rate and composition, duty, work; – d, variables defining the geometry, such as volume, heat exchange area, etc. (unit parameters); – p, variables defining physical properties, such as specific enthalpies, K-factors, etc.; – y, output variables, connecting the unit to the downstream unit (e.g. temperature, pressure, flow rate). In order to improve the reliability of the solution algorithm, the above system could be completed with constraints on the variables involved. Note that the system of equations (2.23) has a strong non-linear character, particularly due to the interdependence between physical properties and state variables. It is important to keep in mind that physical properties may consume up to 90% of the computation time. The difference between the total number of non-redundant variables in the system (2.23) and the number of independent algebraic equations gives the DOF. These are usually specifications that a user must supply to run a simulation. By combining Equations (2.23) and (2.24), the basic rule representing a simulation unit in the SM approach becomes: output variables¼ function input variables,unit parameters, physical propertiesð Þ (2.25) The functional relation is specific for each unit, such as flash, pump, reactor, distillation column, etc. Because of a large variety of physical situations, it is rational to use solution algorithms which are tailored to the specific unit. Such an example is the inside-out algorithm used for solving the multi- component, multi-stage vapour–liquid separation models. Note that the specification mode of a unit operation is important for convergence. This aspect, related to degree of freedom analysis, will be discussed in Chapter 3. Specifications Parameters Inlet streams Duty Work Internal routines Sizing Duty Work Outlet streams Geometry Unit operation model FIGURE 2.12 General layout of unit operation model. 60 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION The flowsheeting software is a very sophisticated computer-based system. In addition to the col- lection of algorithms for solving different unit operations, flowsheeting software includes a GUI and a database with physical properties. Moreover, it should offer facilities to include user-defined models, physical properties and solution algorithms. The architecture of a process simulation software is designed with computer science development and management tools. It is interesting to note that in the total cost the software maintenance (typically more than 70%) is by far more important than the cost of programming (typically under 10%). 2.3.3 EQUATION ORIENTED APPROACH In EO approach, the software architecture is close to a solver of equations. Solution of dynamic sim- ulation problems by an EO approach is more efficient, because dynamic systems are naturally modelled by a system of DAE of the form: dx1 dt ¼ f u, x1, x2, d, pð Þ (2.26) 0¼ g u, x1, x2, d, pð Þ (2.27) The overall DAE system (2.26)–(2.27) is sparse and stiff, and its size varying between 103 and 105 equations. The steady-state solution is obtained by setting the derivatives to zero. Dynamic simulation is more demanding as its steady-state counterpart. First, it needs much more sizing elements. Then, the pressure variation cannot be neglected or lumped in the specification of sim- ulation unit. Moreover, process control elements must be introduced, at least for the purpose of stabi- lising the unit inventory. However, in steady state the specification of variables is more flexible. Any flowsheet variable could be set as, irrespective if this regards input or output streams, internal unit vari- ables (see later in Chapter 3). EO flowsheeting software (such as gPROMS and Aspen CustomModeller) provide comprehensive facilities for developing, validating and executing simulation models, by performing activities such as steady-state and dynamic simulation, optimisation and parameter estimation. Features for problem debugging and post-processing of the results are also available. The modelling language has an object-oriented character and is supported by a user-friendly GUI. Thus, the user defines classes of Models which are instantiated by Units, connected by Streams (flow rate, composition, pressure and temperature) according to the flowsheet topology. TheUnits can be aggregated to formcomplexModels. For example, a counter-current, tube-in-tube heat exchanger can be modelled as the following time- dependent boundary-value problem: @T1 t, zð Þ @t ¼�u1@T1 t, zð Þ @z + 4 dtKTr1cP1 T2 t, zð Þ�T1 t, zð Þð Þ (2.28) @T2 t, zð Þ @t ¼ + u2@T2 t, zð Þ @z � 4 d2m�d2t � � KTr2cP2 T2 t, zð Þ�T1 t, zð Þð Þ (2.29) F1 ¼ pu1d 2 t 4 r1 (2.30) 612.3 ARCHITECTURE OF FLOWSHEETING SOFTWARE F2 ¼ pu2 d2m�d2t � � 4r2 (2.31) with the initial and boundary conditions: T1 0, zð Þ¼T1,0 zð Þ; T2 0, zð Þ¼T2,0 zð Þ (2.32) T1 t, 0ð Þ¼ T1, in; T2 t, Lð Þ¼ T2, in (2.33) where T1(t, z), T2(t, z) are temperature of the tube and shell fluids; dt, dm are tube and shell diameters;KT is heat-transfer coefficient; r1, r2 are tube and shell densities; cP1, cP2 are specific heats; F1, F2 are mass flow rates; u1, u2 are fluid velocities; T1,in, T2,in are inlet temperatures; L is length; z is axial co- ordinate; and t is time. Figure 2.13 illustrates a possible implementation of this model using gPROMS. 2.4 INTEGRATION OF SIMULATION TOOLS Computer simulation covers today practically all the activities devoted to Process Engineering. Flow- sheeting is the key activity, but not the only one. Other computerised tools are necessary to take a true advantage from it. Some of more conceptual nature may be situated upstream, such as thermodynamic analysis and process synthesis; other may be downstream, such as detailed design of units and eco- nomic analysis; and some should be executed at the same level of design, such as heat integration. Moreover, the simulation is involved more and more in sophisticated computer-based systems for Real-Time Optimization (RTO) and operator training. Software for management and business planning can be also integrated with flowsheeting packages, taking into account that these tools need essential information about heat and material balance, best available via rigorous simulation. These components form the core of today’s Computer-Integrated Manufacturing systems. Hence, the applications of pro- cess simulation are shared in two categories, design and operation. These are largely interdependent, but distinctive activities may be identified inside each one. In Computer-Aided Design, the key activity is Conceptual Design that includes Process Synthesis (development of the process flowsheet diagram) and Process Integration (optimal valorisation of ma- terial and energy resources). Other computer-supported activities deal with the detailed design of equipment and with the production of documents, such as drawings and plans, including piping and instrumentation diagram, up to 2D and 3D plant layout. For batch processes, the key activity is planning the timing of different manufacturing tasks, called Scheduling. In Computer-Aided Operation, we can mention the real-time monitoring of material and energy balance, managed nowadays by means of data reconciliation programs. The plant operation can be adapted and optimised in real time by means of computerised tools based on dynamic flowsheeting. Other advanced applications are simulators for safety studies and operator training. This variety of applications mentioned above is reflected by a large number of tools, of smaller or larger extent. However, the only way to ensure an efficient use is their integration in a coherent system. Three approaches may be imagined: 62 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION 1. Integration of complementary products around a central flowsheeting system 2. Interfacing of products supplied by different vendors by means of a dedicated file system (application program interface) 3. Clustering of several packages around a graphical environment and database environment, driven by general accepted standards The first approach dominates today, but only few integrated systems have survived in the rude competition of the 1990s. The second approach was considered occasionally as a partnership between generalist simulator suppliers and smaller developers specialised in niche applications. In practice, inter- facing different systems, sometimes in direct competition, has proved to be unworkable. A reason might be the high cost of developing specialised interfaces that became rapidly obsolete because of fast changes in information technology. The third approach seems themost rational. Figure 2.14 describes the concept. Heat exchanger MODEL PARAMETER RHO1, RHO2 AS REAL CP1, CP2 AS REAL dt AS REAL KT AS REAL L AS REAL T0 AS REAL DISTRIBUTION_DOMAIN zUp AS ( 0 : L ) zDn AS ( 0 : L ) VARIABLE Dm1 AS MassFlow Dm2 AS MassFlow u1,u2 AS Velocity Tin2,Tin1 AS Temperature T1 AS DISTRIBUTION (zUp) OF Temperature T2 AS DISTRIBUTION (zDn) OF Temperature BOUNDARY T1(0)=Tin1; T2(L)=Tin2; EQUATION FOR k:= 0|+ TO L DO $T1(k)=-u1*PARTIAL(T1(k),zUp)+4/dt*KT/RHO1/CP1*(T2(k)-T1(k)); END FOR k:= 0TO L|- DO $T2(k)=+u2*PARTIAL(T2(k),zDn)-4*dt/(dm*dm-dt*dt)*KT/RHO2/CP2*(T2(k)-T1(k)); END Dm1=u1*3.14*dt*dt/4*RHO1; Dm2=u2*3.14*(dm*dm-dt*dt)/4*RHO2; UNIT M1 AS Heatexchanger SET WITHIN M1 DO RHO1:=1000; RHO2:=700; L:=2; dt:=0.05; dm:=0.1; T0:=20; KT:=1; CP1:=1; CP2:=1; zUp:=[BFDM, 1, 10]; zDn:=[FFDM, 1, 10]; END ASSIGN WITHIN M1 DO Dm1:=0.2; Tin1:=20; Dm2:=0.2; Tin2:=80; END INITIAL WITHIN M1 DO FOR k:=0|+TO L DO T1(k)=T0; END FOR k:= 0 TO L|-DO T2(k)=Tin2; END END FIGURE 2.13 gPROMS model of a counter-current tube-in-tube heat exchanger. 632.4 INTEGRATION OF SIMULATION TOOLS The core of an integrated system is the database system and the GUI. The assembly can be inter- faced with simulation packages, primarily for physical property and thermodynamic computations, as well as for steady-state flowsheeting. Packages devoted to conceptual activities, such as process syn- thesis or heat integration, or to engineering tasks, such as design of heat exchangers or piping, can be incorporated at a second level of complexity. Even more complex is the connection of a dynamic sim- ulator. The integrated system should also be ready to accept user models. Several projects for developing commercial process engineering databases have been launched in the years 1980–1990, but none has survived. Rapid changes in technology made the development of such products very risky. Instead, each major supplier developed a proprietary database. However, the absence of standards affects both software producers and users. A collaborative approach based on open software architecture becomes necessary. An example is the CAPE-OPEN initiative (www. colan.org) that will be described later in this chapter. 2.4.1 INTEGRATED SYSTEMS Several major integrated simulation systems will be shortly presented. Updated information may be found by consulting the respective web sites. 2.4.1.1 AspenONE engineering suite (AspenTech) The integrated system includes both the all-purpose flowsheeting system and specialised packages. Different packages communicate via specific files, but share the same physical property methods and data. Here we mention only the major components. A more complete and updated list can be found by consulting the website (www.aspentech.com). – Aspen Plus: steady-state simulation environment with comprehensive database and thermodynamic modelling; feasibility studies of new designs, analysis of complex plants with recycles, optimisation. GUI & database Steady-state flowsheeting Dynamic flowsheeting Process synthesis Sizing Safety hazard Piping Thermodynamics Heat integration User unit Physical properties FIGURE 2.14 Integrated software system around a database environment. 64 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION http://www.colan.org http://www.colan.org http://www.aspentech.com – Aspen Plus Dynamics: dynamic flowsheeting interfaced with Aspen Plus. – Aspen CustomModeller: modelling environment for user add-on units and programming in dynamic simulation. – Aspen Polymers: first-principles polymer modelling technology, fully integrated with Aspen Plus, Aspen Plus Dynamics and Aspen CustomModeller. Industry-leading physical properties, employing state-of-the-art polymer activity coefficient models and equations of state; includes extensive databases of polymer segments, initiators and phase equilibrium parameters. Polymerisation kinetics includes the most comprehensive set of polymerisation kinetic schemes available in the market. User-defined reactions can be included to account for new or unusual chemistries. – Aspen Batch Modeller: a comprehensive simulation tool for both identifying appropriate reaction kinetics models, and then for the conceptual design, analysis and optimisation of batch distillation and batch reaction processes in chemicals, specialty chemicals, pharmaceuticals, food and beverage and other industries. – Aspen Properties: state-of-the-art physical property methods, models, algorithms and data, including regression capabilities and estimation methods. – Aspen Distillation Synthesis: enables rapid conceptual design of distillation schemes for the separation of chemical mixtures with non-ideal VLE. – Aspen Adsorption: enables process simulation and optimisation for a wide range of industrial gas and liquid adsorption processes, including reactive adsorption, ion exchange and cyclic processes: for example, pressure-swing, temperature-swing and vacuum-swing adsorption. – Aspen Chromatography: a comprehensive flowsheet simulator for the simulation and optimisation of batch and continuous chromatographic processes. It allows engineers and scientists to model and understand the separation and purification processes normally found in the pharmaceutical, biotechnology, fine chemical and food product businesses. 2.4.1.2 HYSYS product family (AspenTech) Aspen HYSYS is a comprehensive process modelling system widely used by oil and gas producers, refineries and engineering companies (www.aspentech.com). HYSYS was created by HyproTech, but later on it has been acquired and modified by AspenTech and by Honeywell – where it is known as UniSim Design. The main components of HYSYS are: – Aspen HYSYS: steady-state simulation – Aspen HYSYS Dynamic: integrated dynamic simulation especially aimed at oil and gas, gas processing, petroleum refining – Aspen HYSYS CatCracker, HydroCracker, DelaydCoker, Reformer, VisBreaker: simulation of units specific to a refinery – Aspen Flare System Analyser: design, rate and debottleneck flare and vent systems – Aspen HYSYS Acid Gas Cleaning: rigorous rate-based calculations and property packages suitable for simulation of amine-based absorption processes – Aspen HYSYS Upstream and Upstream Dynamics: simulation of transient multiphase flow of oil, water and gas pipeline networks; integration of the pipeline network models with processing facilities models – HYSYS.Process: steady-state flowsheeting for optimal new designs and modelling of existing plants, evaluate retrofits and improve the process 652.4 INTEGRATION OF SIMULATION TOOLS http://www.aspentech.com – HYSYS.Plant: steady-state and dynamic simulation to evaluate designs of existing plants, and analyse safety and control problems – HYSYS.Operator Training: startup, shutdown or emergency conditions, consisting of an instructor station with DCS interface, and combined with HYSYS.Plant as calculation engine – HYSYS.RTO+: real-time multivariable optimisation; online models may be used off-line to aid maintenance, scheduling and operations decision-making – HYSYS.Refinery: rigorously modelling of complete refining processes, integrating crude oil database and a set of rigorous refinery reactor models – HYSYS.Ammonia: full plant modelling and optimisation of ammonia plants 2.4.1.3 ChemCAD (ChemStations) Its scalability allows users to tailor add-ons and features as needed for their specific industry and pro- cesses. This results in ultimate flexibility and affordability (see www.chemstations.com). • CC-STEADY STATE: Chemical process simulation software that includes libraries of chemical components, thermodynamic methods and unit operations to allow steady-state simulation of continuous chemical processes from lab scale to full scale. • CC-DYNAMICS: Process simulation software that allows dynamic analysis of flowsheets. Applications to operability check-out, PID loop tuning, operator training, online process control and soft sensor functionality. • CC-SAFETY NET: Piping and safety relief network simulation software that allows rigorous analysis of any piping network. Combines two-phase relief device calculation, rigorous pressure drop calculation, rigorous physical property calculation and rigorous phase equilibrium calculation. • CC-FLASH: Physical properties and phase equilibria calculation software that is a subset of the ChemCAD suite (all of the ChemCAD suite products include CC-FLASH capabilities). This program allows rigorous calculation of physical properties and phase equilibria (VLE, LLE, VLLE) for pure components and mixtures. • CC-BATCH: Batch distillation simulation software that, when used as an add-on or standalone program, makes batch distillation simulation and design easy with intuitive, operation step-based input. CC-BATCH is extremely flexible, with many operating modes and the capability to model any number of operating steps and conditions. CC-BATCH optimises batch operation, minimises intermediate ‘slop’ cuts and increases productivity. 2.4.1.4 Pro/II (Invensys SimSci) Pro/II process simulation software is a steady-state simulator enabling improved process design and operational analysis. It is designed to perform rigorous heat and material balance calculations for a wide range of chemical processes. Spanning oil and gas separation to reactive distillation, Pro/II offers the chemical, petroleum, natural gas, solids processing and polymer industries the most comprehensive process simulation solution available today. Pro/II is now integrated with Spiral Crude Suite, providing accurate crude feedstock information to process design simulations to greatly increase the accuracy of the models. Integrating these products results in time and labour savings, increased design accuracy and truly integrated tools for the refining industry. 66 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION http://www.chemstations.com 2.4.1.5 ProSimPlus (ProSim) ProSim is a solution widely used by world’s leading oil, gas, chemicals and engineering companies. ProSimPlus is process engineering software that allows steady-state simulation and optimisation of processes, and performs rigorous mass and energy balance calculations for a wide range of industrial steady-state processes. It is used in design as well as in operation of existing plants for process opti- misation, units troubleshooting or debottlenecking, plants revamping or performing front-end engi- neering analysis. The key features include: a comprehensive set of unit operations including complex models, powerful thermodynamic package able to model highly non-ideal systems and a wide range of applications, unique GUI allowing instant usability, convenient drawing of the flowsheet and quick access to results, open system to expand capabilities (user-defined unit operations, VB scripting, CAPE-OPEN thermo and unit operation interfaces). Remarkably, ProSimPlus provides over 70 unit operations (see www.prosim.net for details). 2.4.1.6 Design II (WinSim Inc.) Design II for Windows (www.winsim.com) is a rigorous process simulation for chemical and hydro- carbon processes including refining, refrigeration, petrochemical, gas processing, gas treating, pipe- lines, fuel cells, ammonia, methanol, sulphur and hydrogen facilities. Its main features include: 60+ thermodynamic methods and 1200+ component database, crude and multi-component distillation towers, flash vessel sizing, two-phase heat exchanger rating, three-phase thermodynamic calculations, crude feed specifications, ChemTran data regression and property constants, automatic output to MS Excel, Visual Basic/Visual C++ interface, inline Fortran/process optimisation. 2.4.1.7 gPROMS (Process Systems Enterprise Ltd.) PSE’s modelling software products enable complex process design and operational decisions to be based on a detailed understanding of the process (www.psenterprise.com). Key features include the following: • General process modelling tools provide an environment for custom modelling, steady-state and dynamic simulation and optimisation, and estimation of model parameters from data. • Advanced model libraries provide state-of-the-art, advanced models that have been validated in industrial application such as: non-equilibrium gas–liquid contactors; fixed bed, bubble columns and trickle bed reactors; Fischer-Tropsch reactors; fuel cells. • Sector-focused tools are aimed at domain specialists: crystallisation, solids, carbon capture. • Model deployment tools allow execution of gPROMS model within other engineering software: CAPE-OPEN compliant systems, Matlab, Excel, Fluent, web interface. 2.4.1.8 Mobatec Modeller (Mobatec) MobatecModeller (www.mobatec.nl) is a software tool that lets model builders construct dynamic (and steady-state) process models of any size in a short time – from single units to entire plants. Users can quickly set-up complex models that are transparent, without much documentation. The modelling methodology implemented in Mobatec Modeller is based on hierarchical decomposition of processes – in which material and energy exchange play a predominant role during normal operation – into networks of elementary systems and physical connections. 672.4 INTEGRATION OF SIMULATION TOOLS http://www.prosim.net http://www.winsim.com http://www.psenterprise.com http://www.mobatec.nl 2.4.1.9 SuperPro Designer (Intelligen Inc.) SuperPro Designer (www.intelligen.com) facilitates modelling, evaluation and optimisation of inte- grated processes. The combination of manufacturing and environmental operation models in the same package enables the user to concurrently design and evaluate manufacturing and end-of-pipe treatment processes and practice waste minimisation via pollution prevention as well as pollution control. Some of the key features of SuperPro Designer include: models for over 140 unit procedures/operations, rigorous reactor modules, material and energy balances, extensive chemical component and mixture database, extensive equipment and resource databases, equipment sizing and costing, thorough process economics, scheduling of batch operations, throughput analysis and debottlenecking, resource tracking as a function of time, waste stream characterisation, environmental impact assessment, OLE-2 support, PFD customisation, enhanced compatibility. 2.4.1.10 ProTreat (Optimized Gas Treating, Inc.) ProTreat software (www.ogtrt.com) is a flexible flowsheeting package for use in the simulation of gas sweetening, dehydration and sour water stripping operations, including CO2 and H2S removal, from mixtures with other gases in such applications such as natural gas, refinery gases, EOR gas streams, syngas streams like ammonia synthesis gas and hydrogen, Claus unit tail gas, CO2 recovery from flue gas, Acid Gas Enrichment (AGR), Acid Gas Reinjection. 2.4.2 OPEN SOFTWARE ARCHITECTURE Despite the existence of powerful integrated software systems, there is a need for the integration of process modelling activity on a larger basis. This should take profit from Internet, as a worldwide net- work of knowledge and business. Following experts in information technology, two types of tools may be of interest for users: process modelling components (PMCs) and process modelling environments (PMEs). The first category is very large, including not only the unit operation models of major software suppliers but also specialised models of engineering and operating companies, as well as developments of consulting firms and academic research centres. The offer in PMC’s in a sharing environment could be seen potentially huge compared with the options available today in commercial products. On the other hand, the offer in simulation environments is very limited. One of the recent initiatives is the CAPE-OPEN consortium (www.colan.org). This involves the collaboration of some major operating companies, software suppliers and universities, with the support of the European Union (Braunschweig et al., 2000). The ultimate vision is to transform process model- ling in a cooperative activity consisting of sharing a large number of components from a variety of sources. Moreover, the simulation might be conceived as an interactive process executed on different computers via Internet or Intranet facilities. Figure 2.15 presents the concept of an open integrated system. The simulation environment could come from the vendor A, to which in-house unit operations can be coupled. One of the user-model unit calls the library of the vendor B, which is linked to a thermo-server supplied by the vendor C, who in turn could consider an EOS model from the supplier D. The vendor E might supply a special solver, while the database with information about physical properties could come from the vendor F, etc. It is clear that the PMC’s should be provided with compatible plug-and-play interfaces. This can be done nowadays at best with an object-oriented technology. 68 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION http://www.intelligen.com http://www.ogtrt.com http://www.colan.org Currently, the follow-up activity has the label of Global CAPE-OPEN (GCO) project. Another stan- dardisation initiative is PDXI (Process Data Exchange Institute), which has been set-up already in the decade 1990 by AIChE. The mentioned initiatives claim looking for a close scientific and technical cooperation. Opening the access to third parties asks for standards. More recently, the standard XML emerged, more convenient for transmitting structured information and queries. XML is a sort of meta-language that enables to build the structure of a personal document, but remaining compatible with the standard Web browsers. The adoption of XML can reconsider the idea of defining a standard process simulation database interface that failed in the past, but which is highly required for a truly open environment. The progress in communication can be spectacular, as for instance using the results of one simulator as starting data for another one. The interface of third-party design package or user program- ming with an all-purpose simulator could be tremendously simplified. The probable adoption of CAPE-OPEN standards will federate the interests of both software pro- ducers and users. At the time where this book is written, several projects are in development. COCO Simulator (CO-LAN and AmsterCHEM) is a free-of-charge, non-commercial, graphical, modular and CAPE-OPEN compliant, steady-state, sequential simulation PME. As an open flowsheet modelling environment, it allows anyone to add new unit operations or thermodynamics packages. COCO Simulator uses a graphical representation, the PFD, for defining the process to be simulated. COCO thermodynamic library ‘TEA’ and its chemical compound data bank are based on ChemSep Lite, a free equilibrium column simulator for distillation columns and liquid–liquid extractors. COCO’s thermodynamic library exports more than 100 property calculation methods with their ana- lytical or numerical derivatives. COCO includes a Lite version of COSMOtherm, an activity coeffi- cient model based on ab initio quantum chemistry methods. The simulator entails a set of unit operations such as stream splitters/mixers, heat exchangers, compressors, pumps and reactors. COCO features a reaction numerics package to power its simple conversion, equilibrium, CSTR, Gibbs mini- misation and plug flow reactor models (www.cocosimulator.org). DWSIM is an open-source CAPE-OPEN compliant chemical process simulator for Windows, Linux and Mac. DWSIM is built on top of the Microsoft .NET and Mono Platforms and features a GUI, ad- vanced thermodynamics calculations, reactions support and petroleum characterisation/hypothetical Unit library vendor B In-house unit Solver vendor E In-house unit Thermo-server vendor C Physical properties database F EOS vendor D Simulation executive GUI vendor A FIGURE 2.15 Vision of a typical CAPE-OPEN modelling tool. After Braunschweig et al. (2000). 692.4 INTEGRATION OF SIMULATION TOOLS http://www.cocosimulator.org component generation tools. DWSIM is able to simulate steady-state, vapour–liquid, vapour–liquid– liquid, solid–liquid and aqueous electrolyte equilibrium processes with several thermodynamic models and unit operations (dwsim.inforside.com.br). 2.4.3 INTERNET SIMULATION Nowadays, most of the simulators are available via local networks. Typically, each organisation has the licence for one simulator, but sometimes it may use a second or a third package. In a globalisation environment, sharing knowledge across remote offices demands a significant investment in commu- nication facilities. In addition, a large category of potential users, small companies or consultants, are excluded because of licence procedures. The advent of the Internet as a worldwide network of transmitting and exchanging information will change profoundly the way of using process simulation. Remote access to a simulation system (e.g. via a web interface) is possible from any point at any time. The availability on a global powerful server can relieve the users from the obsession of changing yearly the hardware and spending a fortune for main- tenance. These are material advantages. But there are also even more important advantages in com- munication and productivity. The attraction of a much larger and motivated category of users will have significant positive effects on the chemical engineering profession. At the time of writing Internet simulation is just beginning. Some applications in operating and managing remotely process plants have been described (Zeng et al., 2000). There are also sites offering Internet access to remote use of simulation and design software. 2.5 SUMMARY AND CONCLUDING REMARKS Process simulation is a key activity in Process Engineering covering the whole life cycle of a process, from Research and Development to Conceptual Design and Plant Operation. In this context, flowsheet- ing is a systemic description of material and energy streams in a process plant by means of computer simulation with the scope of designing the plant or understanding its operation. Steady-state flowsheeting is an everyday tool of the chemical engineer. The generalisation of the dynamic simulation in the design practice is the next challenge. By means of a capable commercial flowsheeting system, it is possible to produce a comprehensive computer image of a running process, a PSM, which can combine both steady-state and dynamic simulation. This tool is particularly valuable in understanding the operation of a complex plant, and on this basis can serve for continuously improv- ing the process design, or for developing new processes. Process simulation is based on models. A model should mirror the reality at the degree of accuracy required by the application. Having a good knowledge of the modelling background is essential for getting reliable results and using the software effectively. The difference between a successful and failed computer-aided project should be attributed more to an insufficient capacity of the user to benefit from the modelling environment than to inadequate performance of the simulator. That is why a prob- lem simulation must be carefully prepared. Flowsheeting is still dominated by the SM architecture, but incorporates increasingly features of the EO solution mode. A limited number of systems can offer both steady-state and dynamic flowsheeting simulators. 70 CHAPTER 2 INTRODUCTION IN PROCESS SIMULATION http://dwsim.inforside.com.br The integration of simulation tools is necessary to cope with the variety of needs in process engi- neering. It is desirable to open the access to simulation technology to a larger number of model sup- pliers. This can be realised by a cooperative approach between the community of users and of software producers. The availability of simulation systems on Internet can boost the use of simulation technol- ogy in a global environment. REFERENCES Agirre, I., Barrio, V.L., Güemez, B., Cambra, J.F., Arias, P.L., 2010. The development of a reactive distillation process for the production of 1,1 diethoxy from bioalcohol: kinetic study and simulation model. Int. J. Chem. React. Eng. 8 (1), A86. Braunschweig, B., Pantelides, C., Britt, I.B., Sama, S., 2000. Process modelling: the promise of open software architecture. Chem. Eng. Progr. 96 (9), 65–76. Dimian, A.C., 1994. Use process simulation to improve your operation. Chem. Eng. Progr. 95 (9), 54–63. Edgar, F., 2000. Process information. Achieving a unified view. Chem. Eng. Progr. 96 (1), 51–59. Thomé, B. (Ed.), 1993. Principles and Practice of Computer-based Systems Engineering, Wiley Series in Software Based Systems. John Wiley & Sons, Chichester. Ullmann’s Encyclopedia of Industrial Chemistry, sixth ed. 2001. Wiley-VCH, Weinheim. Westerberg, A.W., Hutchinson, H.P., Motard, R.L.,Winter, P., 1979. Process Flowsheeting. Cambridge University Press, Cambridge. Zeng, Y., Jang, S.M., Weng, C.C., 2000. Consider an Internet-based process simulation system. Chem. Eng. Progr. 96 (7), 53–60. SOFTWARE Aspen Plus & Dynamics, version 7.3, User Manual, 2010, www.aspentech.com. HYSYS User Guide, version 7.3, AspenTech, 2010, www.aspentech.com. CHEMCAD Use Guide, release 6.5, ChemStations, 2012, www.chemstations.com. Pro/II Use Guide, release 9.0, Invensys SimSci, 2008, www.invensys.com. gPROMS User Guide, version 3.7, Process Systems Enterprise, 2013, www.psenterprise.com. Mobatec Modeller User Guide, version 4.1, Mobatec, 2013, www.mobatec.nl. ProTreat User Guide, version 5.2, Optimized Gas Treating, 2013, www.ogtrt.com. 71SOFTWARE http://www.aspentech.com http://www.aspentech.com http://www.chemstations.com http://www.invensys.com http://www.psenterprise.com http://www.mobatec.nl http://www.ogtrt.com Introduction in Process Simulation Computer simulation in process engineering Process flowsheeting Applications of computer simulation Research and Development Process design Process operation Simulation of complex plants A historical view on simulation Steps in a simulation approach Outline placeholder Approach of a simulation problem Problem analysis Input Execution Results analysis Architecture of flowsheeting software Computation strategy Sequential Modular approach Equation Oriented approach Integration of simulation tools Integrated systems AspenONE engineering suite (AspenTech) HYSYS product family (AspenTech) ChemCAD (ChemStations) Pro/II (Invensys SimSci) ProSimPlus (ProSim) Design II (WinSim Inc.) gPROMS (Process Systems Enterprise Ltd.) Mobatec Modeller (Mobatec) SuperPro Designer (Intelligen Inc.) ProTreat (Optimized Gas Treating, Inc.) Open software architecture Internet simulation Summary and concluding remarks References Software