SUMMARY

An important class of materials design problems involves optimization of the transport properties of a material to satisfy some target specification. In many cases, these transport properties depend sensitively on the detailed microstructure of the material across a wide range of length scales. In addition, relevant transport phenomena may span a broad range of time scales, and multiple complexly-interacting physical processes may contribute to the phenomena of interest. Computational simulation at the level of detail necessary for accurate prediction of behavior under practical usage conditions is therefore often intractable, so instead decades and millions of dollars’ worth of experimental trial and error are frequently required to develop a single new material.

Novel data science strategies have the potential to dramatically shift the balance between computer simulation and real-world testing in the materials development process, significantly reducing the cost and time-to-market for new materials. One particularly promising approach in this area is the materials knowledge system (MKS) framework for property localization, a relatively new technique which has already been shown to offer dramatic efficiency gains in both mechanical and transport-related test cases.

The proposed research will extend previous work in the area to address challenges of particular interest in prediction of transport in microporous materials, namely: the high contrast in transport properties between pores and matrix, transient behavior with possibly widely-separated timescales, and the multiphase/multiphysics nature of many important transport problems. The planned end result is a realistic multiscale test case demonstrating multiphase transport of water and gas in a microporous medium, including modeling of heat transport and phase change, under conditions analogous to those present in the diffusion media of hydrogen-fueled proton exchange membrane
fuel cells.