Perspectives of Soft Computing in Multiscale Modeling for Fluid Flow Systems

Abstract

The modeling of a system containing many models with different scales is known as multiscale modeling. If in that system models are governed by many physical laws constituting the physics of the total problem, then we define the solution of the system as multiphysics solution. Multiscale modeling with multiphysics solutions is challenging task in the domain of mathematical and statistical modeling. However, in practice, governing physical laws on which structure and operation of systems stands are uncertain due to imprecise and heterogeneous information on governing parameters. This kind of uncertainty that presents in multiscale modeling can be managed with soft computing where uncertainty of the parameters is addressed using fuzzy set theory. On the other hand, computation with fuzzy set rather soft computing also facilitates machine learning technique. Therefore, addressing uncertainty of the models of interest using fuzzy set provides a new computing platform which can be called as multiscale modeling with soft computing. Advancement of this kind of computing technology facilitates a new domain, where artificial intelligence with risk informed decision-making methodologies can be formulated to solve real-life complex problems. Soft computing in general can handle imprecision, uncertainty, ambiguity and nonspecificity. The scale of many parameters of a fluid mechanical problem is different and also uncertain due to very small number of associated experiments. Accordingly, fluid mechanical models are labeled as multiscale multiphysics problems which possess uncertainty too and this uncertainty is quantified using soft computing method. In short, this invites to have a perspective of soft computing in multiscale modeling of fluid flow problem. This chapter will explore the perspectives of soft computing in multiscale multiphysics modeling of fluid flow. Multiscale modeling will be explored using statistical mechanics-based lattice Boltzmann method and soft computing will be addressed using fuzzy set theory.