Evolution of stress-induced thermal anisotropy in granular materials: A directed network perspective

External stress can induce anisotropic effective thermal conductivity lambda eff in granular assemblies. However, few studies have used a proper microstructural parameter to study the evolution of thermal anisotropy. In this paper, deformed particle assemblies under different stress paths were generated using discrete element method. By considering each particle as a node, and each heat transfer path via interparticle contact or particle-air-particle as a directed edge, a novel directed thermal network was constructed and a sample-scale parameter named "directed network thermal resistance R" was introduced. Results show contact attributes (i.e., connectivity, quality and orientation) not only dominate the value but also the evolution of anisotropic lambda eff that calculated using thermal network models. R has the merit of considering the three contact attributes and shows a general inverse linear relationship to anisotropic lambda eff regardless of the dominant heat transfer direction, and its ratio can capture the evolution of thermal anisotropy.