Fraunhofer Institute for Industrial Mathematics ITWM
Fraunhofer Institute for Industrial Mathematics ITWM
Customized Simulation - Software Package for Batch Distillation
© freepik
Customized Simulation - Software Package for Batch Distillation

Dynamic Simulation and Optimization for the Chemical Industry

Software Package for Simulation and Optimization of Batch Distillation Processes

© freepik

Customized Simulation – Software Package for Batch Distillation

Simulating and Optimizing Dynamic Processes in Chemical Engineering

In many industrial processes, there is still undiscovered potential for improvement - for example, in terms of the yield of a target product and the energy required to produce it. Identifying this potential is of crucial importance, especially in times of scarce resources. We develop reliable, easy-to-use and individually tailored software for the simulation and optimization of dynamic processes in process engineering. Our software products allow the users to interactively discover potential for improvement and support them in finding optimal solutions. 

 

Software for Batch Distillation

A team from the departments »Optimization - Technical Processes« and »Transport Operations« is developing a software package for the simulation and optimization of batch distillation processes at industrial companies. Batch distillations are used to produce high-purity chemicals from relatively small batches of an initial mixture composed of several substances. For this purpose, the mixture is filled into the bottom of a distillation column and heated there. Due to the different boiling temperatures of the individual substances, they evaporate at different rates and thus reach the condenser at the top of the column at different speeds. In this condenser, the incoming vaporous mixture of substances is liquefied again and then partly withdrawn into various fraction containers and partly refluxed to the column. 

Schematic drawing of a batch distillation column
© Fraunhofer ITWM
Schematic drawing of a batch distillation column with the heated sump at the bottom, the condenser at the top of the column and the stages between the sump and the top, in which the liquid and vapor phase(s) are in thermodynamic equilibrium. The mixture of substances liquefied in the condenser is partly withdrawn and collected in - in this case three - fraction containers and partly returned to the column.

Dynamic Simulation Model

Batch distillation processes are very dynamic in the sense that the mixture compositions, temperatures, pressures and flows in the distillation column can change very quickly and suddenly. In order to adequately describe such rapid changes, a dynamic simulation model is required. We always decide in close consultation with our industrial partners exactly which aspects are taken into account in the model and which are not. In general, the simulation model should be as coarse and simple as possible and only as fine as necessary. In the example mentioned, it was particularly important to our project partners that the simulation model - beyond the usual vapor-liquid equilibria - also depicts liquid-liquid equilibria. Reactions in the column sump or solid-liquid equilibria, on the other hand, were not to be modeled initially. 

Simulation Module as the Core of the Software Package

Our simulator allows to predict the mixture compositions, temperatures, pressures and streams in the distillation column at any given time, depending on the column geometry, the initial composition of the bottom mixture and on the various control variables such as the evaporator heating capacity or the reflux ratios from the condenser into the column. These control variables can be conveniently specified via a graphical user interface. Our simulator is implemented in the high-performance language C++. The implementation builds on a version of the simulator that has already been successfully used for the simulation of other dynamic network systems, namely for power and district heating networks.  

Apart from this basic simulation core, our software package also includes two optimization modules:

  • one for substance parameter estimation
  • one for the optimal control of the batch distillation process
Predicted Composition for a Substance Mixture
© Fraunhofer ITWM
Composition of the distillate withdrawn at the column head, as predicted by our simulator for a 7-component mixture. As can be seen, very sudden changes in composition occur in this system. Moreover, a second liquid phase is formed during distillation – whereas at the beginning and at the end there is only one liquid phase in the distillate.

Estimation of Substance Parameters

The first optimization module allows unknown or only imprecisely known thermodynamic substance parameters underlying the batch distillation process to be estimated. For this purpose, an external optimizer adjusts the substance parameters such that the predictions of our simulation model are as close as possible to the experimentally measured values. The choice of the substance parameters to be optimized and the experimental comparison variables is made via the user interface. 

Optimal Specification of Control Variables

The second optimization module makes it possible to set the control variable profiles of the batch distillation process under consideration in such a way that certain user-defined objectives are achieved in the best possible way, while at the same time strictly adhering to certain constraints. Control variables are, for example, the reflux ratios from the condenser to the column or the fraction boundaries. Objectives can be to maximize the target product yield, minimize the waste product amount, minimize the total energy, or minimize the total distillation time, and a typical constraint is that the purity of the target product in the main fraction must be at least 95 percent. 

Graphical User Interface for Analysis of Simulation and Optimization Results

The graphical user interface not only allows for a very simple and user-friendly definition of the simulation and optimization problems to be solved. The users can also visually analyze the calculated simulation and optimization results. In particular, it allows an easy comparison of different optimal solutions and the selection of a best solution for the current objective.

Related Links: