(130c) Single and Multi-Objective Optimizations Using Existing Rigorous Process Simulations: A New Hope

Today, traditional process simulation software does not take full advantage of advances in CPU architecture (multiple cores) for parallel processing. At the same time, industrial needs for achieving optimum process conditions for both single and multiple objectives are at their greatest. This led many industrial and academic groups to investigate equation-oriented (EO) solvers, custom written code, and/or reduced models to achieve the necessary calculation speed to perform large scale optimizations. The disadvantages have been extended setup time, significant domain expertise requirements, difficult model maintenance, and sub-optimal solutions.

This paper will discuss significant improvements in calculation time achieved through parallel processing techniques in a traditional sequential process simulation package combined with advanced optimization methods to ensure global optimums in single and multiple objective optimizations. The industrial benefit is a reduction in setup time, a reduction in required expertise, the ability to leverage existing process models, and the benefits of first principles’ simulation. Calculation time is competitive with rigorous EO methods. We will present examples of single and multiple objective optimizations of industrial processes and discuss setup and calculation times.