Accurate formula for the effective conductivity of highly clustered two-phase materials

Two-phase heterogeneous materials arising in a variety of natural and synthetic situations exhibit a wide variety of microstructures and thus display a broad spectrum of effective physical properties. Given that such properties of disordered materials generally depend on an infinite set of microstructural correlation functions that are typically unobtainable, in practice, theoretical approaches must rely on rigorous bounds and approximations on the effective properties that depend on a limited set of nontrivial microstructural information. Torquato [Random Heterogeneous Materials, Microstructure and Macroscopic Properties (Springer-Verlag, New York, 2002)] derived such an approximation for the effective thermal/electrical conductivity of two-phase microstructures that perturb about those that realize the well-known self-consistent formula, a key feature of such microstructures being phase-inversion symmetry which can be observed in certain cellular, bicontinuous, and porous materials. Here, we show via extensive numerical simulations that this virtually unexamined approximation, which depends on up to three-point correlations functions, accurately predicts the effective conductivities of various two- and three-dimensional two-phase microstructures across phase volume fractions in which both phases can form large and well-connected clusters, including certain phase-inversion symmetric microstructures as well as phase-inversion asymmetric dispersions of certain fully penetrable particles. We also highlight the accurate sensitivity of this approximation to phase-connectedness properties by showing that it predicts percolation thresholds that are in good quantitative and/or qualitative agreement with known thresholds for the models considered here. Finally, we discuss how this formula can be used to design materials with desirable effective conductivity properties and thus aid in materials by design.