SIMULATION MODELS INCREASE PROCESS EFFICIENCY
More efficient daily process operation with rigorous simulation models
In chemical engineering, the use of rigorous simulation models is common practice in the case of new process design, optimization studies, or for trying out different operation scenarios. However, once a process in a chemical plant is up and running, limited improvements are generally made. In 2013, the Institute for Sustainable Process Technology (ISPT), in collaboration with Eindhoven University of Technology, FrieslandCampina, Corbion and DSM, started the project IMPROVISE (Improved Process Operation via Rigorous Simulation Models) to develop model-based technology for improving daily process operation. The IMPROVISE-project was successfully completed in 2017, and proves that the use of dynamic models not only leads to better control of the plant, but also saves valuable resources. A follow-up project, called INSPEC, is already running.
MORE SUSTAINABLE AND EFFICIENT PROCESSES
“In a world in which resources are scarce and there is a growing public demand for greener production methods, industry needs to move to more sustainable and efficient processes", says Leyla Özkan, Project Leader. “This is where academia and business need to work together. The scientific field of dynamic systems and control provides essential, effective tools for the high-performance operation of a wide variety of processes. IMPROVISE offered an ideal platform and meeting place where industry could benefit from the existing expertise of researchers."
FROM OFFLINE TO ONLINE
Rigorous simulation models are used regularly in an offline manner for the optimal design of new processes and the investigation of operational scenarios. “They are built using fundamental conservation laws, and offer a high-fidelity digital representation of their physical twin", explains Leyla. “Over the past decade, disciplines such as automotive and high-tech systems have increasingly been using such dynamic models in an online fashion for control implementations and operations. This means the models can be updated on the spot if necessary, and as a result, the whole model-based operation technology used in these systems can be adjusted. Model-based design and control are in fact indispensable components of operations in these disciplines."
“In the IMPROVISE project, we aimed to improve daily operations at three different companies by using the knowledge available in dynamic rigorous models", says Leyla. “The online use of dynamic rigorous models has several benefits. With these models, we can monitor plant behaviour and get information about key process variables. In essence, they make the immeasurable visible in real time. Dynamic models can also be used for predictive maintenance by providing information on how the plant is going to perform in the future."
ULTRAFILTRATION OF WHEY PROTEINS
One of the companies involved in IMPROVISE was FrieslandCampina. “As a dairy cooperative, FrieslandCampina valorises more than 10 billion litres of milk every year", says Marc Jacobs, responsible for Global Process Technology R&D at FrieslandCampina's Innovation Centre. “A substantial part of this involves the production and sales of various dairy ingredients derived from cheese whey. One of the key technologies for whey processing is ultrafiltration, which is used to concentrate the whey proteins. These ultrafiltration systems are cyclically operated between production and cleaning. Our case in the IMPROVISE-project centred around the optimal control of the system to reduce or slow down the build-up of fouling, thereby increasing productivity and product quality by reducing process variability."
MORE EFFICIENT AND BETTER QUALITY
Together with Eindhoven University of Technology, FrieslandCampina built a validated rigorous dynamic simulation model of the ultrafiltration system. “With this model, control strategies were designed centred around more fine-grained control of the loops", says Marc.
“Evaluation of these strategies showed us that we could reduce the fouling build-up during processing, and therefore use the installed membrane more efficiently. Moreover, by preventing the flow and concentration disturbances due to the introduction of clean membrane stages during production, we achieved much better control of product quality."
PUTTING LEARNINGS INTO PRACTICE
Although no model-based control systems have been implemented yet at FrieslandCampina, the learnings from the control studies have been worked out further, says Marc. “They've been implemented in the existing control system of a microfiltration system for milk protein separation, where it greatly improves control. For the ultrafiltration system, a project is now running to implement the necessary changes to the hardware and controls. The learnings from the project are also used in the design of the control systems in several new membrane plants." FrieslandCampina now also participates in the follow-up INSPEC project, which focuses on linking chemometric PAT methods to rigorous dynamic models for process control. “We hope that with the introduction of advanced statistical modelling methods, we can make steps in improving the control of product properties during processing," says Marc.
POLYESTERIFICATION IN REACTIVE BATCH DISTILLATION COLUMNS
One of the other participants in the IMPROVISE-project was nutrition and materials multinational DSM. “DSM produces polyester resins in a batch process, where water is continuously removed from the process to drive it to maximum conversion", says Ruud Barendse, Modeling Engineer at DSM Material Science Center. “As a batch process, it is inherently dynamic and therefore ideally suited for model-based process control. As part of the IMPROVISE-project, a generic model of the process was built, based on a combination of publicly available information and DSM insights from production and prior internal modelling efforts. The first simulation results showed a good fit with real-life observations from our production practice."
TAKING MODELLING TO THE NEXT LEVEL
DSM is also taking part in the follow-up INSPEC-project. “In fact, DSM has been using rigorous modelling for process development and support for decades", says Ruud. “However, over the past few years, there has been a substantial shift from continuous bulk processes to more batch-oriented production. This also requires a change in modelling approach, in which the IMPROVISE- and INSPEC-projects can play a useful role. In general, we're continuously striving for process optimisation, and changes will be implemented in practice if sufficient incentive is generated."
WHO IS DR. LEYLA ÖZKAN?
Dr. Leyla Özkan is an Assistant Professor in the Control Systems Group of TU/e. Before joining TU/e, she worked as a consultant at IPCOS B.V. Her research interests include modelling, optimization, model-based control of chemical systems, interaction of process design and control. She is an Associated Editor for Journal of Process Control and chairs the PSE-NL Network.