Study on estimation of instantaneous airflow distribution in cubic building group model using POD-LSE

Abstract

The authors have been developing a method to estimate the airflow distribution in urban area with limited observation information by combining POD (Proper Orthogonal Decomposition) with LSE (Linear Stochastic Estimation). In this paper, the effectiveness of this method was verified by reconstructing and predicting instantaneous airflow distribution obtained from LES (Large-Eddy Simulation) for cubic building group model. We validated the reconstructed and predicted values with POD-LSE by comparing with the LES data as the true value and investigated the sensitivity of the accuracy to the size of analysis area and the number of observation points.