Although there are already many models for prediction of the thermal conductivity of nanoparticle suspension, most of them only consider the cases for high nanoparticle volume fractions (often [ 0.1%). Considering the cost of particle itself, pumping and transportation, dispersion stability, etc., low particle concentrations are obviously more preferred. The present study firstly experimentally investigated the thermal conductivity of some oxide colloid suspensions with low particle concentrations. It was found that the famous multi-sphere Brownian model (MSBM) showed a large deviation for thermal conductivity prediction when the particle concentrations are below 0.1%. By introducing an adjustable exponential con- stant, n, into the volume fraction item of MSBM, the predicted thermal conductivity can be in good agreement with experimental data for the particle volume fractions ranging from 0.001 to 10%. The dependency of n on the particle volume fraction and its possible physical significance were discussed in detail. It turns out that, when the volume fraction is larger than 1%, the modified model can be reduced to the original MSBM. The value of n approximately equals to 0.7 when the volume fractions of nanofluids are lower than 0.005%. Between these two volume fractions, n is found to follow a nearly linear relation with the logarithm of volume fraction. The validity of the modified MSBM model for practical supplication was further justified based on numerical method. The simulation results showed that our model has excellent agreement with Maxwell–Garnetts model in low volume fraction ranges due to the weak interaction between various nanoparticles in the system. Our study should be of value for numerical simulation and engineering design of nanofluids in the future.

Keywords