Process simulation

The process simulation is a tool for development and optimization of technical processes in process engineering or chemical plants .

Principles

Process simulation is essentially an image of chemical processes and basic operations in computer programs. A number of skills are required for modeling:

chemical and physical properties of pure substances and mixtures (through measurements, research in databases, correlations or estimates),
Material and heat transport,
Device properties ( reactors , columns , mixers , condensers , evaporators, etc.),
Reaction mechanisms and kinetics , see chemical reaction engineering ,
suitable mathematical, physical and chemical models
efficient solution processes.

The process simulation ensures that the material and heat balances are correct and brought into a stable equilibrium . Usually the processes are visualized at the same time.

history

The first attempts to simulate processes electronically were made on electronic analog computers as early as the 1950s . However, these approaches were quickly abandoned in favor of simulations on digital computers .

The first developments in the digital process simulation of chemical plants began in the 1970s, as this was the first time that suitable hardware and software (in particular the Fortran programming language ) were available. The modeling of physical properties began much earlier, for example cubic equations of state (see for example Van der Waals equation ) and correlations (see for example Antoine equation ), which were developed in the 19th century and some of them today still be used. Investigations into the kinetics of chemical conversions and reaction mechanisms were also well advanced. Apparatus properties had also already been largely modeled so that all tools were available to model and calculate complete chemical processes in silico (exclusively using computers).

At the same time, the development of process simulation has greatly accelerated the further development of the various models for the estimation of material properties , reaction mechanisms , their kinetics, device properties , etc., but in particular also the development of fact databases . Today, fact databases are used to further develop estimation methods and correlations.

Static and dynamic process simulation

Originally, process simulation was only applied to stationary systems. A complete mass balance and energy balance of a stationary state are obtained on the basis of models. This static simulation is now supplemented by dynamic simulation. In this context, dynamic means that time-dependent results are calculated. In principle, the flow diagram is viewed infinitesimally and calculated numerically as a system of differential equations . This method requires a significantly higher computing power , but also allows the transition to the control and management of chemical plants in real time . A simple example is filling or emptying a container. With dynamic simulation, control processes (PID controller), holdups and chemical reactions in particular can be represented very realistically.

Phase equilibria

The most common phase equilibrium is the vapor-liquid equilibrium (mostly abbreviated as VLE for Vapor-Liquid Equilibrium), which is particularly important for gas scrubbing and rectification . But it is also used when calculating boiling and thawing temperatures. For ideal substances, such as alkanes , the Raoult-Dalton law , which is based on the definition of partial pressure , is sufficient as a model . In the case of non-ideal mixtures, the activity coefficient is calculated in the liquid phase and the fugacity coefficient in the gas phase, thus correcting Raoult-Dalton's law. While the fugacity coefficient can be easily calculated from the equation of state (often Soave-Redlich-Kwong) of each individual substance in a mixture, the activity coefficient depends on the binary interactions. In a mixture with z. B. 10 ingredients exist 45 binary interactions. Therefore 45 VLE would have to be measured in this case. VLE measurements are in databases e.g. B. the DETHERM or the DDB and in the literature such. B. DECHEMA Data Collection. This also contains the associated parameters of suitable models such as B. Non-Random Two-Liquid Model (NRTL). For many binary mixtures that were not measured, the model parameters can be estimated using the UNIFAC method. The UNIFAC model may be a. described in the VDI heat atlas.

The more the activity coefficients deviate from one, the more clearly the xy diagram (x denotes the composition of the liquid, y the vapor composition) differs from that of an ideal VLE, until it finally intersects the diagonal or forms an S-curve, which is the case sign of azeotropy . and possibly a mixture gap is. This can easily be demonstrated using the porter model.

Ultimately, it is also possible to use the non-random two-liquid model (NRTL) to calculate a liquid-liquid equilibrium (LLE), provided the parameters are known. An LLE can be calculated approximately with VLE-NRTL data. The larger the miscibility gap , e.g. B. Benzene - water , the smaller the error. In the case of the n-butanol- water system with a smaller miscibility gap, the approximation is not acceptable. With suitable data, even complex LLEs such as 3-methylpyridine water with elliptical equilibrium lines can be calculated.

On the basis of suitable data for the heat of fusion , the NRTL model can even be used to calculate solids solubilities (solid-liquid equilibrium, SLE for solid-liquid equilibrium). For many substances such as B. the very close-boiling substances 1-methyl-naphthalene and 2-methyl-naphthalene results directly in a eutectic , the location of which is a good approximation of reality.

Database

The substances used in the process simulation are selected from a database . The database contains gases, liquids, solids. Polymers and Electrolytes . It can be expanded with your own substances and data using regression. The database offers temperature-independent data such as B. critical pressure and temperature functions for, for example. The vapor pressure , specific heat capacity , etc. Known databases are DETHERM and the Dortmund database , which essentially contain experimental data for pure substances and mixtures, and the DIPPR database, which mainly parameters for equations for pure Contains substances. With the help of mixing rules, the substance data of mixtures can be approximately calculated from known basic substance data. For substances that are not included in the databases mentioned, the substance data are often calculated using incremental methods such as B. Joback generated.

rectification

Rectification, often also called distillation, is one of the central basic operations in process simulation but also in chemical process engineering . The older FUG model (Fenske-Underwood-Gilliland), which represents a quick and good approximation for ideal mixtures, hardly plays a role anymore. Rather, the Simultaneous Correction System (see Perry's Chemical Engineering Handbook) has established itself, which can model almost all types of rectification well, such as B. azeotrope, extraction, reactive distillation, dividing plate column, gas scrubbing, absorption, desorption, electrolytes, side column. The inside-out model is still in use for petrochemical distillations, as it converges quickly and the mixture mainly consists of alkanes.

Batch distillation can also be simulated. The algorithms of continuous distillation are mostly used. With batch distillation, a multicomponent mixture can be divided into individual fractions in chronological order. The mathematical description of the batch distillation takes place with the help of the Rayleigh distribution .

Reactors

The best-known types of models are the stoichiometric, equilibrium and kinetic reactors . The equilibrium reactor can be modeled on the one hand according to Gibbs' theory and according to van't Hoff. The Arrhenius model is usually used for the kinetic reactor. In combination with VBA, kinetic approaches can be represented as desired, e.g. B. for bioreactions. In the batch reactor, also known as a discontinuous stirred reactor, reactions are simulated as a function of time by solving differential equations.

Interfaces

For optimum operation of the process simulation serve interfaces such. B. Excel for data transfer to a project database or a plant. With the help of MS-COM technology, the process simulation can even be controlled from Excel, i. H. be started. This even enables online simulation in which data from running systems is continuously fed into the process simulation. The results are used for optimal process management.

Process simulation software

There are large numbers of commercial process simulation programs. Larger companies often have their own developments in use that are only used in-house.

While many systems can usually be used to simulate pure fluid processes, some simulation programs can also be used specifically to simulate solid processes.

literature

Gmehling, Kolbe, Kleiber, Rarey: Chemical Thermodynamics for Process Simulations, Wiley-VCH, 2012
Wang, Schmidt: Calculations in Chemistry and Process Engineering, Wiley-VCH, 2015

Individual evidence

^ ^A ^b ^c Gmehling, Jürgen, 1946-, Kolbe, Bärbel ,, Kleiber, Michael ,, Rarey, Jürgen ,: Chemical thermodynamics: for process simulation . Weinheim, Germany 2012, ISBN 978-3-527-31277-1 .

[:0-1] A ^b ^c Gmehling, Jürgen, 1946-, Kolbe, Bärbel ,, Kleiber, Michael ,, Rarey, Jürgen ,: Chemical thermodynamics: for process simulation . Weinheim, Germany 2012, ISBN 978-3-527-31277-1 .