An inter-comparison of three urban wind models with Oklahoma City Joint Urban 2003 Experiment wind measurements

Three wind models are compared to time-averaged wind measurements obtained in downtown Oklahoma City during the Joint Urban 2003 Field Experiment. The models cover several levels of fidelity and include an empirical-diagnostic code, a Reynolds-averaged Navier-Stokes (RANS) computational fluid dynamics model, and a Large Eddy Simulation (LES) computational fluid dynamics code. The models were run with the same grid resolution and used identical inlet conditions. All three models compare favourably to the near-surface wind measurements, although there are several instances of winds being calculated poorly in specific locations. For the three experimental cases evaluated in this study, the RANS model appears to perform the best. Although in principal the LES code should be the top performer, it is likely that inadequacies in how we specified the inlet profile and the necessity of finer resolution around building edges explain the LES code not performing as well as the RANS code. The empirical-diagnostic code performed slightly worse than the two computational fluid dynamics codes, but given that this code runs 2-3 orders of magnitude faster it is a viable option for applications where fast turn around time is required or where many cases must be run.