https://www.mdu.se/

The use of biomass has increased in recent years due to the efforts to reduce the high emissions of greenhouse gases, primarily carbon dioxide from combustion of fossil fuels. At the same time industrial processes have become more complex because of increased production rates and profitability requirements. A higher degree of automation is needed when the processes are too complex to be handled manually. There is a need to find modeling strategies that can automatically handle the challenges that the conversion of biomass in an industrial process entails, such as operational changes, decreasing component and overall system efficiency, drifting sensors, etc. The objective of this thesis is to develop a methodology for on-line applications in industrial processes. Dynamic process models have been developed for continuous digesters and boilers. Process models have been evaluated for their use in continuous industrial process. Applications that have been studied are monitoring and diagnostics, advanced control and decision support. The process models are designed for on-line simulations. The results shows that the use of mathematical simulation models can improve the use of both process data and process understanding, to achieve improved diagnostics, advanced control and process optimization. In the two examples of industrial processes covered in this thesis, we can see that similar types of models can be used for completely different types of processes, such as pulp digesters and boilers. It also demonstrates the ability to combine soft sensors and hard sensors with physical models to take the information to a higher level of utilization.