Cooling performance of Newtonian and non-Newtonian nanofluids in a square channel: experimental investigation and ANN modeling

The cooling performance of Newtonian and non-Newtonian nanofluids in a square duct was identified experimentally. The flow regime was laminar. Two-step method was used to prepare stable dispersions of γ-Al2O3 and TiO2 nanoparticles. Water and ethylene glycol were used as the base fluids of Newtonian, and a 0.5 mass% carboxymethyl cellulose in water was used as the base fluid of the non-Newtonian nanofluids. Nanoparticle concentration was in the range of 0.1–1.5% by volume. Heat transfer coefficient of nanofluids enhances significantly relative to the base fluids. The enhancement of heat transfer coefficient of nanofluids is proportional to the Peclet number and nanofluid concentration. Nanoparticles type and the base fluid affected the Newtonian nanofluids performance, while for non-Newtonian nanofluids this is not the case. The improvement in nanofluids heat transfer coefficient is more than just related to their thermal conductivity enhancement. The obtained experimental data were modeled by the artificial neural network (ANN). Two empirical correlations were also fitted on the data. The experimental data are well predicted by the ANN models and empirical correlations. Statistical criteria show that the ANN models are more accurate.