(134b) Efficient Modeling of Polydisperse Solid-Liquid Suspensions in Stirred Tanks

Predicting the movement and interaction of dispersed solids with the continuous phase in stirred tanks is of key interest in a variety of industrial applications in the chemical and biopharmaceutical industry such as heterogeneous catalysis, dissolution, or microcarriers in cell culture bioreactors. Key parameters of interest are the minimum required agitation speed for off-bottom suspension (Just-suspended speed) and inhomogeneities in the particle concentration across the reactor volume in order to determine optimal feeding and sampling locations. Complexity in the prediction of this movement arises due to the simultaneous interaction of a large number of particles, making predictions based on analytical solutions of the settling behavior of single spherical particles impossible. Additional difficulty arises for non-spherical particles where the forces on the particles depend on the orientation of the particles within the flow field. While classical RANS-based CFD modeling approaches proved to be sufficiently accurate for the prediction of just-suspended speeds, transient effects, and local inhomogeneities may not be captured sufficiently accurately [1]. In this work, we use the Lattice Boltzmann Method to predict the transient three-dimensional flow field inside stirred tanks and employ an Euler-Lagrange approach via the discrete particle method (DPM) for modeling the dispersed solid phase. This approach was shown to be suitable for small particle diameters below the numerical grid size. In order to decouple the size dependency of the dispersed particles from the numerical grid, several models are reported in the literature [2]. Here we present a coupling approach to generalize the Lagrangian dispersed phase and resolve this grid dependency based on the two-grid method. The simulation runtimes are shown to be suitable for application in the aforementioned industrial use cases by integrating and running the presented model on the commercial GPU-based modeling software SIMVANTAGE (simvantage.com).

[1] Delafosse et al. “Solid-liquid suspension of microcarriers in stirred tank bioreactor – Experimental und numerical analysis”, Chem. Eng. Sci 180, 2018, p52-63

[2] Zhang et al. “Grid-indipendent Eulerian-Lagrangian approaches for simulations of solid fuel particles combustion”, Chemical Engineering Journal 387, 2020