A Gaussian Process-Based emulator for modeling pedestrian-level wind field

Wind tunnel tests and computational fluid dynamics (CFD) simulations remain the main modeling techniques in wind engineering despite being expensive, time-consuming, requiring special facilities expert knowledge. There is a clear need for fast, accurate, but, at same time, computationally economical substitute. This study proposes Gaussian Process-based (GP-based) emulator to predict pedestrian-level environment near lift-up building – an isolated, unconventionally configured building. The proposed GP-based transcends limitations of previous emulators as it can handle many inputs (8) output parameters (384) large dataset (150 CFD simulations). To increase efficiency, current data reduction method based on Principal Component Analysis (PCA) technique estimate hyper-parameters optimization. latter efficiently compute 250 requires no prior knowledge their probability distributions. faster, by factor 107 than predicting speeds, its accuracy substantiated using both qualitative quantitative analyses, which reveal that emulator's predictions all-prominent flow features have systematic bias, are highly great reproductivity.