(235f) A Surrogate Optimization Algorithm for Red Wine Reactor Simulation-Based Design

Design and analysis of complex reactor systems often involves the use of computational fluid dynamics (CFD) simulations. However, because many design variables are used in simulation, and their effect on process responses are usually nonlinear, design optimization is time-consuming and difficult to perform. This difficulty can be managed by combining a surrogate model with an optimization method to select response-optimal design parameters. Here, we use a radial basis function to accumulate process knowledge from a red wine fermentation reactor simulation, and iteratively suggest optimal design parameters using a hybrid global/local stochastic optimization method. We find that this framework guides the simulation to optimal designs in fewer experimental queries than traditional response surface methods. By utilizing this framework, fermentation practitioners can take advantage of advances in CFD modeling to quickly design processes even when simulations are expensive and time-consuming.