| description abstract | In this paper, an optimization was performed to achieve uniform distribution of convective heat transfer coefficient over a target plate using two impinging slot (air) jets. The objective function is the root mean square error (Erms) of the local Nusselt distribution computed by computational fluid dynamic (CFD) simulations from desired Nusselt numbers. This pattern search minimized this objective function. Design variables are nozzle widths, jettojet distance, jettotarget plate distance, frequency of pulsations (for pulsating jets), and the flow rate. First, an inverse design is performed for two steady jets for simplicity and the obtained errors for three different desired Nusselt numbers, NuD = 7, 10, and 13, were 20.73%, 20.08%, and 22.92%, respectively. Uniform distribution of heat transfer coefficient for two steady jets was not achieved. Thus, two pulsating jets are considered. The range of design variables for pulsating state is as same as steadystate and heat transfer rates increased about 400% over steadystate due to the effects of pulsations in inlet velocity. Thus, in the pulsating state, optimization must be performed for the desired Nusselt numbers around fourtimes NuD in the steadystate, i.e., NuD = 28, 40, and 52. The Erms reduced less than 0.01% and distribution of heat transfer coefficient for all cases was uniform. An experimental study using an inverse heat conduction method (conjugate gradient method with adjoint equation) has been performed and the experimental results for the case of NuD = 52 are presented. The estimated distribution of Nusselt number on the target plate with the numerical distribution has around 3.2% relative error with optimal configuration. | |
