Applications of data assimilation to forecasting indoor environment

Data assimilation (DA) is a technique to combine numerical predictions and experimental measurements, which has been commonly used by the community of numerical weather forecasting but is relatively new to the field of indoor environment. Among all data assimilation methods, Ensemble Kalman Filter (EnKF) is especially suitable to solve large-scale nonlinear problems due to its low computation requirement. In this paper, two new applications of EnKF to forecasting indoor environment are presented. The first study is to forecast the contaminant transport in a residential house based on a multi-room tracer gas decay experiment. The forecast model avoids calculating source term in the given problem by rapid updating model states with EnKF while still providing significant forecast lead time. The second study is to predict the fire smoke dispersion in a multi-room compartment fire without giving actual heat release rates (HRRs) as model inputs. Compared to conventional deterministic models, the EnKF models improve the predictions significantly.