(191b) Minimizing Batch Cycle Time Using Evolutionary Design of Dynamic Experiments

Due to the nonlinear characteristics of batch processes, optimizing their performances, such as minimizing the cycle time, usually relies on a knowledge-driven model consisting of a set of algebraic differential equations. However, as the configuration and behavior of industrial process varies one from another, the knowledge-driven model is not always available. In such a case, a data-driven model which can be developed after a limited number of experiments is an attractive alternatives. Here we apply the Design of Dynamic Experiments (DoDE)¹ methodology to model the process behavior and minimize the batch cycle time using the obtained model. DoDE generalizes the classic design of experiment approach and allows time-varying variables, such as feeding profile of reactant, to be considered in the experimental design.

The DoDE approach has been applied to optimize an industrial simulation at Dow². The obtained optimal operating condition was very promising but some safety and product quality constraints were violated of some designed experiments. To satisfy these constraints while running the experiments, we apply the evolutionary operation of DoDE³ in this presentation. This methodology also reduces the initial number of experiments. The initial design is selected conservatively in the small vicinity of the previous operating conditions. After the initial data-driven model has been estimated using the collected data, an optimal operating condition with satisfying uncertainty statistics⁴ is selected and implemented. The new optimal condition also serves as a new data point that will be applied to update the data-driven model. An updated optimal operating condition will then be selected to further improve the process performance. The above steps are iterated until the best process performance is achieved. We examine the evolutionary DoDE approach in an industrial simulation of polymerization process at Dow.

1. Georgakis, C., Design of Dynamic Experiments: A Data-Driven Methodology for the Optimization of Time-Varying Processes. Industrial & Engineering Chemistry Research 2013, 52 (35), 12369-12382.
2. Georgakis, C.; Chin, S.; Hayot, P.; Wassick, J.; Chiang, L. H. In Optimizing an Industrial Batch Process Using the Design of Dynamic Experiments Methodology, AICHE Spring Meeting, Houston,TX, April 10-14; Houston,TX, 2016.
3. Wang, Z.; Georgakis, C. In Data-Driven Optimization Using an Evolutionary Design of Dynamic Experiments for Biopharmaceutical Processes, AICHE Annual Meeting, San Francisco, CA, Nov 13-18; San Francisco, CA, 2016.
4. Wang, Z.; Georgakis, C., An in silico evaluation of data‐driven optimization of biopharmaceutical processes. AIChE Journal 2017, 63 (7), 2796-2805.