Chemical Engineering Process Simulation

Chemical Engineering Process Simulation, Second Edition guides users through chemical processes and unit operations using the main simulation software used in the industrial sector. The book helps predict the characteristics of a process using mathematical models and computer-aided process simulation tools, as well as how to model and simulate process performance before detailed process design takes place. Content coverage includes steady-state and dynamic simulation, process design, control and optimization. In addition, readers will learn about the simulation of natural gas, biochemical, wastewater treatment and batch processes.

• Provides an updated and expanded new edition that contains 60-70% new content
• Guides readers through chemical processes and unit operations using the primary simulation software used in the industrial sector
• Covers the fundamentals of process simulation, theory and advanced applications
• Includes case studies of various difficulty levels for practice and for applying developed skills
• Features step-by-step guides to using UniSim Design, SuperPro Designer, Symmetry, Aspen HYSYS and Aspen Plus for process simulation novices

• Language: English
• Publisher: Elsevier Science
• Release date: Sep 29, 2022
• ISBN: 9780323984553

Book preview

Part I

Basics of process simulation

Outline

Chapter 1. Introduction to process simulation

Chapter 2. Registration of new components

Chapter 3. Physical property estimation and phase behavior for process simulation

Chapter 4. Simulation of recycle streams

Chapter 1: Introduction to process simulation∗

Dominic C.Y. Foo ¹ , and Rafil Elyas ²       ¹ University of Nottingham Malaysia, Semenyih, Selangor, Malaysia      ² East One-Zero-One Sdn Bhd, Shah Alam, Selangor, Malaysia

Abstract

Process simulation is the representation of a chemical process by a mathematical model, which is then solved to obtain information about the performance of the chemical process (Motard et al., 1975). It is alsoknown as process flowsheeting. Westerberg et al. (1979) also defined flowsheeting as the use of computer aids to perform steady-state heat and mass balancing, sizing, and costing calculations for a chemical process. In this chapter, some basic information about simulation will be presented. This includes the historical developments, basic architectures, degree of freedom, and solving algorithms. Besides, 10 good habits of process simulation are also provided at the end of the chapter, to guide readers in nurturing some good practices in using process simulation software.

Keywords

Degrees of freedom; Onion model; Process design; Process flowsheeting; Process synthesis; Simulation software; Simulation steps

1.1 Process design and simulation

1.2 Historical perspective for process simulation

1.3 Basic architectures for commercial software

1.4 Basic algorithms for process simulation

1.4.1 Sequential modular approach

1.4.2 Equation-oriented approach

1.5 Degrees of freedom analysis

1.6 Incorporation of process synthesis model and sequential modular approach

1.6.1 Ten good habits for process simulation

Exercises

References

Further reading

Process simulation is the representation of a chemical process by a mathematical model, which is then solved to obtain information about the performance of the chemical process (Motard et al., 1975). It is also known as process flowsheeting. Westerberg et al. (1979) also defined flowsheeting as the use of computer aids to perform steady-state heat and mass balancing, sizing, and costing calculations for a chemical process. In this chapter, some basic information about simulation will be presented. This includes the historical developments, basic architectures, and solving algorithms. Besides, 10 good habits of process simulation are also provided at the end of the chapter, to guide readers in nurturing some good practices in using process simulation software.

1.1. Process design and simulation

Many regard process simulation being equivalent to process design, which is indeed a misleading understanding. In fact, process simulation and process synthesis are two important and interrelated elements in chemical process design, which may be used to achieve optimum process design. The aim for process simulation is to predict how a defined process would actually behave under a given set of operating conditions. In other words, we aim to predict the outputs of the process when the process flowsheet and its inputs are given (Fig. 1.1). In the modern days, commercial process simulation software packages are often used for such exercises.

On the other hand, when an unknown process flowsheet is to be created for given process input and output streams, this entails the exercise on process synthesis (Fig. 1.2). Process synthesis has been an active area of research in the past 5decades, with some significant achievements in specific applications, e.g., heat recovery system, material recovery system, and reaction network (El-Halwagi and Foo, 2014). Process synthesis and process simulation supplement each other well. In most cases, once a process flowsheet is synthesized, its detailed characteristics (e.g., temperature, pressure, and flowrates) may be predicted using various process simulation tools, so that an optimum flowsheet may be developed. In other words, process simulation tools are useful in guiding process synthesis exercises (Foo et al., 2005; Lott, 1988). One may also explore the use of process simulation tools for other related activities, such as waste minimization (Sowa, 1994; Hilaly and Sikdar, 1996), debottlenecking (Koulouris et al., 2000), etc.

Figure 1.2  A process synthesis problem (El-Halwagi, 2006; Foo, 2012).

Figure 1.1  A process analysis problem (El-Halwagi, 2006; Foo, 2012).

Within process synthesis, one of the important models to guide flowsheet synthesis is the onion model first reported in Linnhoff et al. (1982). As shown in Fig. 1.3, process design exercise begins from the core of the process and moves outward. In the center of the onion, the reactor system is first designed. The reactor design influences the separation and recycle structures at the second layer of the onion. Next, the reactor and separator structures dictate the overall heating and cooling requirement of the process. Hence, the heat recovery system is designed next, in the third layer. A utility system at the fourth layer is next designed, to provide additional heating and cooling requirements, which cannot be satisfied through heat recovery system. At the final layer, the waste treatment system is designed to handle various emissions/effluents from the process, prior to final environmental discharge.

In a later section of this chapter, the onion model will be used to guide the simulation of the integrated flowsheet of chemical processes.

Figure 1.3  The onion model.

1.2. Historical perspective for process simulation

With the introduction of computers in the 1950s, we see the start of commercial process simulation software a decade later. The first generic process simulation software known as PROCESS was launched by Simulation Science based at Los Angeles (US) in 1966, for the simulation of distillation columns (Dimian et al., 2014). This software evolved into PRO/II and is marketed by AVEVA in recent years (AVEVA, 2022). Another commercial software for gas and oil applications, known as DESIGN, was launched in 1969 by ChemShare Corporation based at Houston (US) (Dimian et al., 2014). This software is marketed as DESIGN II for Windows by WinSim Inc. since 1995 (WinSim, 2022).

Stepping into the 1970s, which was generally known as the golden age of scientific computing, several important historical milestones mark the active developments of process simulation tools. First, FORTRAN programming language (introduced by IBM between 1954 and 1957) became the de facto standard among scientists and engineers (Evans et al., 1977). Two important books on process simulation (Crowe et al., 1971; Westerberg et al., 1979) described some important developments in the 1970s. Of particular importance is the formal introduction of sequential modular (SM) approach by Westerberg et al. (1979), which is commonly utilized in most software. Next, the first oil crisis in 1973 simulated the development of simulation tools that can be used for solid handling, i.e., power generation with coal. Following this was the important ASPEN (Advanced System for Process ENgineering) project at Massachusetts Institute of Technology (MIT) between years 1976 and 1979, sponsored by the US Department of Energy (Evans et al., 1979; Gallier et al., 1980). Aspen Technology Inc. (AspenTech) was then formed in 1981 to commercialize the technology, with Aspen Plus software being released in 1982 (AspenTech, 2022). Another important achievement in 1970s is the development of software based on equation-oriented (EO) approach. Important EO-based software includes SPEEDUP developed at Imperial College, London (UK) (Hernandez and Sargent, 1979; Perkins et al., 1982), which was later succeeded by gPROMS (Siemens PSE, 2022). Note that in the 1970s, simulation was mainly executed on fast but expensive mainframe systems, where user was connected via a remote terminal.

With the arrival of personal computer (PC) in the 1980s, several other important software packages such as ChemCAD (developed by ChemStations) and HYSYS (originally by Hyprotech) were launched. These software packages no longer operate on the mainframe systems but are PC based. Late 1980s also saw the needs of developing simulation software for biochemical processes (Petrides et al., 1989). This leads to the introduction of BioPro Designer (Petrides, 1994), which later evolved into SuperPro Designer marketed by Intelligen Inc. in the 1990s (Intelligen, 2022). It is worth mentioning that up to the 1980s, most basic developments of process simulation architectural are quite established, with review papers outlining their state-of-the-art techniques (Evans, 1981; Rosen, 1980). It is also worth mentioning the introduction of spreadsheet software during the 1980s, which allows quick solving of equation sets that follow a sequence (Seader, 1985). Spreadsheet software such as MS Excel ¹ are still highly welcomed by industrial practitioners in solving day-to-day computational tasks to date.

In the 1990s, with the domination of Microsoft Windows, most software packages were migrated from their previous mainframe/keyword input versions into the more attractive graphical user interface (GUI). Another important milestone happened in the early 21st century. In 2002, Hyprotech was acquired by AspenTech, which resulted with the ownership of HYSYS software. However, the US Federal Trade Commission judged that acquisition of Hyprotech was anticompetitive and ruled AspenTech to divest its software to the approved buyer—Honeywell (Federal Trade Commission, 2003). This later leads to the introduction of UniSim Design (Honeywell, 2017), which shares the same GUI as HYSYS (of Hyprotech), while AspenTech continues to market its Aspen HYSYS software (with different GUI in year 2016).

Some of the commonly used process simulation software packages are listed in Table 1.1.

1.3. Basic architectures for commercial software

Fig. 1.4 shows the basic structure of a process simulation software and sequential steps in performing the simulation task (Turton et al., 2013). As shown at the upper side of the figure, a typical commercial simulation software includes the following components, i.e., component database, thermodynamic model database, flowsheet builder, unit operation model database, and flowsheet solver. Note that some other elements, e.g., subflowsheet, financial analysis tools, and engineering units option, are software dependent and hence are excluded in this figure.

The bottom side of Fig. 1.4 presents a list of sequential steps in solving a simulation problem. In step 1, the basic information for a simulation problem is first provided. This includes chemical components and thermodynamic model selection, which can be done easily through their associated databases of the software. Note that it is advisable to select all components needed for the flowsheet, even though some components will only be used at the later part of the flowsheet. The selection of thermodynamic model is a crucial step, as different thermodynamic models will lead to very different mass and energy balances for some processes. Next in step 2, the process flowsheet is constructed using the flowsheet builder. This involves the selection of appropriate unit operation models (from the unit model database) and the connections among them with the process streams (some software may need to have energy streams connected too). In step 3, specifications are to be provided for the unit models as well as important inlet streams (e.g., flowrate, temperature, pressure) to execute the simulation. Note that in all modern simulation software, users may choose to display the simulation results in various forms. Finally, it is important to cross-check the simulation results, either through some empirical model, mass, and energy balances or through reported plant/experimental data. Doing so will increase the confident level of the simulation model.

Figure 1.4  Basic structure of a commercial simulation software and sequence in solving a simulation model. Adapted from Turton et al. (2013).

Table 1.1

1.4. Basic algorithms for process simulation

Two main classical techniques used in solving process simulation models are SM and EO approaches. Most commercial simulation software packages in the market, e.g., Aspen Plus, ChemCAD, and PRO/II, are using SM approach and hence will be discussed more in depth in the following sections.

1.4.1. Sequential modular approach

The term sequential modular was formally introduced in the late 1970s (although commercial software packages were found in the market prior to that) by Westerberg et al. (1979). The concept of SM may be explained using Fig. 1.5. Each of the unit modules contains some algorithms that are utilized to solve a set of process models, provided that the inlet stream information and unit specifications are given. Once a module is solved to convergence, it will generate the results for the outlet stream(s). The latter is then connected as a feed stream for the following unit module, which is then solved for convergence (Turton et al., 2013). The same process is repeated until all process units in the flowsheet are solved and converged. Note that certain unit modules may require iterative solution algorithm to achieve convergence; the overall process is, however, sequential in nature, i.e., no iteration is required (Turton et al., 2013).

Figure 1.5  Concept of sequential modular approach (Turton et al., 2013).

For process flowsheet that contains recycle stream(s), tear stream strategy is commonly used with SM approach to converge the recycle stream. As shown in Fig. 1.6A, the flowsheet consists of six operations, i.e., units A–F and a recycle stream that connects units C and F. If SM approach in Fig. 1.5 is adopted for the simulation exercise, one will start to simulate and converge unit A. This is followed by unit B and then unit C. However, because of the existence of the recycle stream, unit C can only be simulated once the recycle stream contains the necessary properties (e.g., pressure, temperature, and flowrates) after unit F is converged. However, unit F cannot be simulated without first converging unit C. In other words, the convergence of units C and F involves iterative steps.

To cater the iterative procedure, a tear-stream strategy is used. As shown in Fig. 1.6A, the recycle stream is virtually torn into two parts—r 1 (inlet for unit C) and r 2 (outlet from unit F). Some estimated data (e.g., temperature, pressure, and flowrate) are provided for r 1 to simulate unit C. Once unit C is converged, simulation then proceeds to unit D, E, and finally unit F. Once unit F is converged, the simulated results from outlet stream r 2 are then compared to those estimated data given to r 1 earlier on. If their values agree to the specified convergence tolerance (typically given in terms of percentage difference), the simulation is converged, or else the simulated results from r 2 will substitute the estimated data in r 1 and the iterative procedure is carried on. ²

Figure 1.6  (A) A process consists of a recycle stream; (B) concept of tear stream (Turton et al., 2013).

The main advantage of SM is that it is intuitive and easy to understand. It allows the interactions of users as the model develops. However, large problems (with many recycle streams) may be difficult to converge.

1.4.2. Equation-oriented approach

In EO approach, a set of equations are solved simultaneously for a simulation problem. For instance, for a problem with n design variables, p equality constraints, and q inequality constraints, the problem is formulated as follows (Smith, 2016):

(1.1)

(1.2)

The main advantage of EO approach is its ability to be formulated as an optimization problem. However, complex EO problems are difficult to solve and diagnose. It is also not as robust as SM approach (Smith, 2016). Hence, it has not been favored among commercial simulation software in the past few decades, until recent years where it is embedded in solving complex models in SM-based software (e.g., gPROMS).

One of the convenient ways of solving problem of linear equation set is through matrix inversion, which has the general structure as in Eq. (1.3):

(1.3)

where A is the coefficient matrix, X is variable matrix or an unknown vector (i × 1), and B is a constant matrix or a known solution vector (i × 1). The variables in variable matrix X can be determined by the product of inverse matrix A (i.e., A −¹ ) with matrix B, given as in Eq. (1.4):

(1.4)

The use of matrix inversion for equations solving is demonstrated with an example at later section (see Example 1.2).

1.5. Degrees of freedom analysis

DOF analysis is a useful tool to determine if a system has the sufficient information before it can be solved. Eq. (1.5) shows the basic equation of DOF ( ) of a system, which is given as the difference between its numbers of variables ( ) and the numbers of independent equations ( ).

(1.5)

For a system to be solvable, its n df must be equal to zero. In other words, when numbers of independent equations (e.g., =3) are equal to its numbers of variables (e.g., =3), the system can be solved readily.

On the other hand, when the numbers of variables (e.g., =3) are more than the number of independent equation (e.g., =2), this system has a positive DOF (e.g., =1). For such a case, design variables are to be specified before other state variables can be calculated. There are also over-defined cases where the number of variables is less than the numbers of independent equations (e.g.,  ≤ 1). For such a case, the redundancies are to be identified and removed, before the system can be solved.

A more complete form of DOF analysis is given in Eq. (1.6), to account for chemical reaction, molecular balances, and additional equation relations (Felder and Rousseau, 2005).

(1.6)

where is the number of chemical reactions, is the number of independent molecular species balances, and is the number of other equations relating the unknown