Design of process industry plants and their automation is a challenging task, especially when cost-effectiveness pressures are ever increasing. This leads to challenges for engineers responsible for this work. This thesis investigates how those challenges could be alleviated by extended use of dynamic process simulation. The current trend of digital twins both calls for and enables this work. As this work relies on simulation models and is computational in nature, digital twins are an enabler. On the other hand, the digital twins call for approaches to extract added value from data, in the case of this thesis, simulation-generated data. The present work consists of four case studies from the process industry. Dynamic process simulation is combined with a novel model comparison method, with global sensitivity analysis and with multiple objective optimization. By conducting a massive number of well-planned simulations and analysing the resulting data, it is shown that the challenges could be alleviated. The cases pertain to certain phases of a process industry plant's life cycle, namely early design and operation. Two cases, Paper production and Tower control, target the early design phases, while the Filtration and Bottleneck cases concentrate on the operation phase. The Paper production case shows the utility of the proposed model comparison method. This led to the conclusion that it helps in gaining confidence in optimization results from simplified models, focusing the designer's attention as well as providing insight into the operation of the plant. The Tower control case, combining dynamic process simulation and global sensitivity analysis, highlights process areas where the control designer's attention should be focused. Similarly, the Bottleneck case shows where retrofit actions on an operational plant should be focused. Finally, the Filtration case shows the feasibility of combining dynamic simulation with interactive multiple-objective optimization in providing insight into the process operation. A synthesis of these contributing results then supports the main hypothesis of this thesis: Extended added value or utility can be extracted from simulation models when they are combined with other mathematical methods.
|Translated title of the contribution||Laajennetun dynaamisen prosessisimuloinnin lisäarvo prosessi- ja automaatiosuunnittelussa|
|Publication status||Published - 2019|
|MoE publication type||G5 Doctoral dissertation (article)|
- dynamic simulation
- global sensitivity analysis
- model comparison
- multiple objective opimization