Optimizing the Conceptual Process Design of an Entire Plant Using Multi-Scale Models

Many of the important decisions during the design of a new plant are made in the early stages. Yet, models used during FEED are typically standard flowsheet models. Furthermore, optimization of the design is often done by manual trial and error or, if a formal optimization is performed, only a limited number of design decisions and constraints is typically considered. Although in many cases the models are adequate for the purpose at hand, even properly optimized designs using simplified/standard models may lead to poor results. In a similar fashion setting up a simplified optimization problem typically results in a suboptimal solution. In this presentation we will demonstrate the importance of using detailed, multi-scale models in an optimization with 49 decision variables on an industrial case study. We will compare the results with using simplified models and trial-and-error optimization. We will demonstrate how uncertainty in some of the parameters can have a significant effect on ROI/IRR and how this uncertainty can be assessed. Finally, some new developments for simultaneous product and process design as well as structured product process design are discussed.