Implementation of Computational Fluid Dynamics and Response Surface Methodology to Study Nanofluid Heat Transfer

Methods of computational fluid dynamics and design of experiments are implemented in modeling the heat transfer of an Al2O3/water‐ethylene glycol (EG) nanofluid as a coolant in a car radiator system. The main purpose of the study is to propose a model to predict the Nusselt number using the design of computer experiment. The response surface methodology is applied to investigate the effects of the flow rate, the nanoparticle concentration, the water/EG ratio, and the inlet temperature on Nu. Analysis of variance is performed to obtain a suitable correlation that could predict Nu, leading to the elimination of insignificant terms in the initial proposed quadric model and the attainment of a reduced model. Moreover, the reduced model is converted to a simpler linear model using Box‐Cox transformation, which predicts Nu with an accuracy of ±20 %. The design of computer experiment (meta‐model) technique, which avoids the experimental difficulties with computer‐assisted models, has become increasingly popular in studying the effects of factors on a particular response in various engineering applications. Combining computational fluid dynamics simulation results with a powerful statistical model is an effective technique to investigate the behavior of a nanofluid in a heat transfer system.