Unified homogenization model for accurate prediction of effective thermal conductivity in multi-phase particle-reinforced composites

Particle-reinforced composites are commonly used in various industries, with effective thermal conductivity (ETC) being a critical property. This study develops representative volume element (RVE) models to calculate ETC, achieving a dispersion of less than 2.57 %, which indicates an appropriate RVE size. The feasibility of applying standard two-step mean-field homogenization (MFH) and differential two-step mean-field homogenization (TSH) methods to predict ETC in multi-phase composites is examined, highlighting their strengths and limitations. The impact of various TSH schemes on ETC prediction is further evaluated, identifying the optimal scheme (Opt-scheme) for multi-phase composites. By combining the strengths of MFH and Opt-scheme, the Unified Homogenization (UH) method is introduced, reducing the maximum error from 51.4 % to 6.73 %, significantly improving prediction accuracy. Additionally, the Lewis-Nielsen (L-N) model for two-phase composites is modified to enable more efficient and accurate ETC estimation in particle-reinforced composites, achieving an average error of 1.2 %, with an applicable volume fraction of up to 73 %. The L-N and UH models developed in this paper exhibit superior prediction accuracy, extrapolation, and generalization compared to existing models, as validated by experimental and simulation data.