Using real-time sensing data for predicting future state of building fires

Abstract

The development of sensor technologies in recent years makes it possible to acquire real-time states of building environment and its systems with a major trend towards big data and wireless data transmission. How to use these vast real-time data sets to achieve a safer, more comfortable and energy efficient building becomes a major challenge for building engineering. This paper investigates one of the possibilities of using the real-time data for the prediction of future fire development states. Beyond simply reflecting the real-time states of a system, the sensing data will be able to forecast a highly dynamic problem of building fire growth and smoke dispersion inside building environment. This paper presents the forecasting method based an ensemble Kalman filter (EnKF) to predict building fire smoke temperature and smoke layer height at the real time. Detailed formulations of the zonal fire smoke models and the EnKF model are presented. The proposed real-time forecasting method is demonstrated and validated by a 1:5 scaled compartment fire experiment. The results indicate that real-time forecasting of building fires is achievable while the accuracy is noticeable which can be applied to assist emergency evacuation and firefighting.