A unique thermal conductivity model (ANN) for nanofluid based on experimental study

The alumina, copper oxide and zinc oxide nanoparticles (40 nm) were used to prepare the distilled water based nanofluids. The particle weight concentration varies in the range of 0.02% to 2%. The thermal conductivities were measured in the range of 20 °C to 90 °C. The input data for the present artificial neural network (ANN) model were nanoparticle weight fraction and nanofluid temperature and the output data was thermal conductivity of the nanofluid. The ANN used one hidden layer and it was optimised by varying number of neurons. The statistical approach has been employed to find out the coefficients in the proposed correlation using ANN validated experimental results. The estimated data obtained by the ANN model is in good agreement with the experiments. The proposed theoretical correlation is able to find out thermal conductivity ratio of nanofluids in a wide range of particle concentrations and temperatures. [Display omitted] •Thermal conductivities of metal-oxide/water based nanofluids are studied.•Artificial neural network architectures are developed for numerical prediction.•Nanofluid concentrations and temperatures are taken as input variables.•A generalised correlation was proposed to estimate thermal conductivity.•The mean square errors and correlation coefficients are used to evaluate accuracy.