The Use of GPU to Decrease Simulation Time of PBM Modeling of Particle Aggregation Using Ansys Fluent

A common process in industry is the production of latex using emulsion aggregation, a system containing a discrete phase of latex in a continuous aqueous phase. Throughout this process smaller particles flocculate into larger more uniform particles, establishing a particle size distribution (PSD) which is required for quality standards. The colloidal interactions are governed by Derjaguin, Landau, Verwey, Overbeek (DLVO) theory and these fluid and flow properties can be modeled using the Computational Fluid Dynamics (CFD) software ANSYS Fluent. Population Balance Modeling (PBM) is used to track the PSD of the latex particles. The main limitation of using CFD to model these processes is the computational time required to run the simulations. Previously forty seconds of simulation data required 6 hours and 22 minutes to run. Central processing units (CPUs) and graphics processing units (GPUs) are used in unison to improve calculation speed, achieving a speedup factor of 13. The aim of this project is to understand the complexities associated with emulsion aggregation as well as minimize the computation time required to run the simulations. Through the examination of velocity profiles and change of PSD over time, two tanks of different sizes are modeled using ANSYS Fluent and compared to assist in proper scale up.