Dynamic Physical Simulation Models Lead to Improved Process Operations

The project IMPROVISE was finished in October. TU/e in collaboration with DSM, FC, Corbion have worked on model based operation technology based on detailed physical models and used it to improve daily operations of important chemical processes in the Dutch Chemical Industry,

Being able to anticipate what a chemical process is going to do next allows you to make the right decisions. By doing so, you can produce better products, save energy, reduce waste and make processes safer. TU/e and industrial partners have demonstrated the benefits of the dynamic physical models for process control and process monitoring over a wide range of processes including whey separation process, resin production in a reactive batch distillation column and crystallization.

This work has been undertaken as part of IMPROVISE, a research project, within the cluster of Process Intensification, Process System Engineering and Advanced Process Control (PSE, PI, APC) of ISPT. It has demonstrated that dynamic simulation models play a significant role in the reduction of operational costs by use of the model based operation technology and decision making. Led by Dr Leyla Özkan of TU/e, the project consortium involved Royal FrieslandCampina, Corbion and DSM.

For each of the case studies, the project has delivered dynamic physical models, developed and tested several advanced control and monitoring strategies in simulation.  In certain cases, the process operation strategies have already been taken up by the industrial partner and implemented in the pilot scale.

The IMPROVISE project has thus far been so successful that the consortium partners will continue their collaboration together in a new project called: INSPEC. This project will also welcome one new academic partner Radboud University and one new industrial partner, chemical producer Huntsman. This follow-up project aims to combine sensor based process monitoring with advanced process control technologies.

This video shows the performance of the Ultrafiltration Membrane Unit. On the left we see two sections, the top two plots show the performance of the UF unit under the manual control of operators. This is the operation previously implemented . Operators turn  4-5 membranes  on and when they observe a drop in the output flow of retentate they increase the baseline pressure or  they add a new membrane stack . Adding a new membrane stack with full of water results in disturbance for the overall system and operating at high pressures actually contributes the evolution of the fouling of the membranes. With a rigorous dynamic model we have shown that actually it is better to turn all the membranes at the same time if we want to extend the operation of UF membrane unit. This is shown on the last two plots on the left side. Here we operate all the membranes on and at low baseline pressures.

The effect of both types of operation is shown on plot at the right hand side.  We designate the fouling with the membrane resistance. The use of series of resistances concept is common in modeling of membrane fouling. In this plot, the evolution of membrane resistance is shown for two different types of operation. When all the membranes are on and low pressures are used, the speed of increase in the membrane resistance is slower. This indicates that we can operate membranes longer time.


This project is co-funded with subsidy from the Topsector Energie by the Ministry of Economic Affairs and Climate Policy.