A digital workflow for learning the reduced-order structure-property linkages for permeability of porous membranes

摘要: Abstract Quantitative relationships between the complex porous structure of a membrane (henceforth simply referred to as microstructure) and its effective permeability are critical for improving performance membranes used in filtration separation applications. This paper presents digital workflow learning structure-permeability linkages membranes. The presented establishes desired by bringing together recent advances (i) generators three-dimensional representative volume elements (3-D RVEs) reflecting large diverse set structures, (ii) numerical approaches reliable evaluation 3D-RVEs, (iii) low dimensional representation material internal using framework 2-point spatial correlations principal component analyses, (iv) Gaussian process (GP) regression with input-dependent noise (i.e., heteroscedasticity). It is seen that this study can systematically identify salient features 3-D microstructure train reduced-order heteroscedastic GP models on data generated physics-based simulations. will be shown structure-property able make high fidelity predictions assessment uncertainties new structures at minimal computational cost.