An integrated RBFN-based macro-micro multi-scale method for computation of visco-elastic fluid flows

Abstract

This paper presents a numerical approach for macro-micro multi-scale modelling of visco-elastic fluid flows based on the Integrated Radial Basis Function Networks (IRBFNs) and the Stochastic Simulation Technique (SST). The extra stress is calculated using the Brownian configuration fields (BCFs) technique while the velocity field is locally approximated at a set of collocation points using 1D-IRBFNs. In this approach, the stress is decoupled from the velocity field and computed from the molecular configuration directly without the need for a closed form rheological constitutive equation. The equations governing the macro flow field are discretised using a meshless collocation method where the IRBFN approximants improve the accuracy of the numerical solutions by avoiding the deterioration of accuracy caused by differentiation. As an illustration of the method, the time evolution of the planar Couette flow and the steady state Poiseuille flow are studied for two molecular kinetic models: the Hookean dumbbell and FENE dumbbell models.