Gerova, K. and Savill, A., "On Optimum Choice of Forchheimer Equation Coefficients for Computational Fluid Dynamics Simulation of Heat Exchangers," SAE Int. J. Passeng. Cars - Mech. Syst. 8(2):779-785, 2015, doi:10.4271/2015-01-9110.
The simulation of heat exchanger air flow characteristics using Computational Fluid Dynamics requires knowledge of the experimental pressure drop across the element concerned. This is normally achieved through wind tunnel testing of either full scale heat exchangers or by using laminations of various porous materials and honeycombs to represent these. The current paper both continues and compliments prior work published by the lead author, which entailed a series of measurements of the pressure drop in both the near and far field, across screens with porosity (β) in the range 0.41 < β < 0.76. This experimental investigation established a relationship between the porosity and the pressure drop characteristics of a given material at various angles of inclination to the free-stream flow. In addition, the effect of screen depth was investigated using honeycombs. The present paper investigates the use of a weighted least squares regression model to correct the previously obtained Forchheimer Equation coefficients for residual errors associated with the data being a function of the underlying experimental measurements. Following a review of the available relations for pressure drop through woven screens, it was concluded that the best method to use in obtaining the Forchheimer Equation coefficient is when each screen is treated individually rather than using a general correlation that relates the permeability and inertial coefficient to the porosity of the screen. Porous jump conditions in the computational procedure were used to model thin membranes with known velocity vs. pressure-drop characteristics. The CFD results matched closely the experimental data, with only a slight variation when modelled using coefficients obtained through data fitting. In contrast, the simulation results varied from the experimental data when only the material thickness value was modified, and not the corresponding porous jump coefficients. A best practice recommendation was thus established.