(320d) Towards the Computational Optimization of Processing Equipment: Application to Static Mixers

The engineering performance of static mixer systems across a range of Reynolds numbers is predicted using time-accurate three-dimensional DNS and LES computational fluid dynamics (CFD) simulations. These simulations are executed using a lattice-Boltzmann algorithm over a voxelized solid-domain. Since this approach requires no user-meshing and minimal parameter specification, it is ideally suited for computational optimization. We begin by discussing the underlying physics governing the algorithm, and the difference between lattice-Boltzmann and conventional CFD. Next, for the Kenics UltraTab mixers operating in a turbulent regime we validate the predicted pressure drop and examine the transient eddy properties in the immediate vicinity of the mixer. We then validate the predicted mixing cup coefficient of variation (COV) against the experimentally measured values. Next, for laminar systems, DNS simulations are performed on Kenics KM mixers to predict the pressure drop and downstream mixing properties. The predicted pressure drop is compared to experimental measurements and literature values. The predicted COV is also validated against experimental data. We conclude by discussing how this modeling approach can be combined with computational search heuristics to identify optimized static mixer topologies.