Numerical simulation and predictive modeling of an inextensible filament in two-dimensional viscous shear flow using the Immersed Boundary/Coarse-Graining Method and Artificial Neural Networks

Abstract

This article presents a hybrid IBM/CGM scheme for the study of two-dimensional orbital regimes of flexible and inextensible filaments in low-Reynolds number shear flows. The study includes a parametric analysis of the mechanical response for a range of the filament length, bending rigidity and shear-strain rate. The results are compared to theoretical estimates and previous numerical and experimental analyses. These results are subsequently used to develop a prediction model using Artificial Neural Networks to effectively forecast the orbital regime of a filament immersed in shear flow.