Statistical Model to Predict Air Side Pressure Drop for Heat Exchangers 2018-01-0081

In a typical ground vehicle, airflow enters engine compartment through grille and carries heat from the engine, cabin and other auxiliaries through heat exchangers such as radiator, condenser, oil cooler and charge air cooler respectively. The amount of airflow entering the engine compartment is governed by their individual resistances, the grille and engine compartment resistances. Also, this flow adds to drag and deteriorates overall aerodynamic efficiency. It is known as cooling drag which contributes to 8 to 12 percent of overall drag. Aerodynamics and Front End Air Flow (FEAF) development happens through CFD and it demands accurate heat exchanger pressure drop data which is usually obtained from supplier at very early stages of a vehicle development. Historically, this data is found to have significant variations compared to in-house test data. To have an accurate pressure drop data for CFD simulation, a predictive (statistical) heat exchanger model was built around Kriging Gaussian method. In this study, the heat exchanger geometry was parameterized into six different parameters mainly as fin-density, fin height, fin thickness, fin louvers, tube height and core thickness. The porosity/air side pressure drop performance of any heat exchanger depends on one of the pattern of these parameters. This pattern repeats across the width and height of a heat exchanger and pressure drop across this individual pattern signifies total pressure drop of the entire heat exchanger. The first part of the paper discusses about establishing a correlation between in-house heat exchangers test and CFD for couple of heat exchangers with simplified geometries. Later sections discuss about setting up full factorial Design of Experiments (DOE), CFD models creation and analysis. In this work, DOE was used to generate data points for unique combinations of parameters and pressure drop were obtained across these data points. These parameters and data points were statistically analyzed by fitting Kriging response, thereby obtained a response surface, which was used as a tool to predict the pressure drop for any possible combinations of the heat exchanger parameters.