Numerical Analysis and Scientific Computing Seminar | Lucian Ivan, Progress in the Polydisperse Gaussian-Moment Model for Disperse Multiphase Flows

Tuesday, May 16, 2023 1:00 pm - 1:00 pm EDT (GMT -04:00)

MC 5417 and online (For Zoom Link please contact ddelreyfernandez@uwaterloo.ca)
 

Speaker

Dr. Lucian Ivan, Computational Research Scientist Canadian Nuclear Laboratories

Title

Progress in the Polydisperse Gaussian-Moment Model for Disperse Multiphase Flows

Abstract

Despite their prevalence in countless engineering applications, multiphase flows characterized by particles of variable properties, such as those encountered in aerosolized sprays, atomized fuel injection, and atmospheric aerosol and airborne bioaerosol dispersion, remain a computational challenge. Although Lagrangian-based models have been relatively successful and widely adopted for multiphase flow simulations, they remain prohibitively expensive to generate statistically relevant predictions – often requiring an excessive number of sample particles and numerical experiments. Alternatively, Eulerian-based methods have the potential to mitigate the computational cost associated with predicting multiphase flows, but they traditionally exhibit modelling artefacts. This talk provides a progress update on the development of the Eulerian-based polydisperse Gaussian-moment model (PGM) for the description of particle-laden multiphase flows. The PGM includes a direct treatment of the local higher-order statistical moments related to particle velocity, namely variances and covariances. Furthermore, the PGM also tracks the local variance and covariance of each distinguishable property of interest (e.g., particle size, temperature, bioaerosol concentration of viroload) as well as the covariances associated with each property variable and particle velocity. Such statistical information is not used in traditional Eulerian models, and holds the promise of improved modelling accuracy at reduced computational cost. Derived within an entropy-maximization moment-closure formulation, the PGM yields a set of first-order robustly-hyperbolic balance laws which are efficiently solved with a massively parallel discontinuous-Galerkin-Hancock framework. A range of multiphase flow simulations, including dispersion of atmospheric plumes, bioaerosols, and fuel sprays are presented to illustrate current capabilities of the computational framework.