Thermal Conductivity of Composites: How COMSOL Revealed an Omission in a Classical Paper

P. Berne [1],
[1] University Grenoble Alpes, Grenoble, France
Published in 2015

The initial motivation for this work was to explore the relationship between the shape of particles and the thermal conductivity of nanofluids or nanocomposites containing them. Since the possibility for manufacturing exotically-shaped particles is ever growing, it was thought useful to devise a way to select which materials and shapes have a potential for better thermal properties.

Literature [1] suggests that a wide range of experimental data on nanofluids can be well represented by a standard conduction model taking into account an interfacial resistance between the particles and the matrix in which they are embedded. In practice, the heat conduction problem is solved in a cubic cell containing one particle and submitted to a unit temperature gradient on two opposite sides and a zero-flux condition on the others. The heat flux through the cell is computed and translated into an equivalent conductivity K. The calculation is repeated for various particle sizes, corresponding to various particle volume fractions v. The K-v relationship is then analyzed as a series expansion K/Km = 1 + [K].v + O(v2) where Km is the conductivity of the matrix and [K], termed the “intrinsic conductivity”, is the final result. The computations were done using COMSOL Multiphysics with two Heat Transfer in Solids interfaces coupled by a flux condition. An “extra-fine” physics-controlled meshed was used in most cases.