Identification of constant contaminant sources in a test chamber with real sensors

Abstract

Identification of contaminant sources is important for controlling indoor air pollution. Previous studies of source identification are mainly based on computational fluid dynamic simulation results, and studies based on experimental measurements results are scarce. In this paper, six source identification experiments, five with one source and one with two sources, were conducted to investigate the effectiveness of source identification in real situations. A multi-source identification model was developed based on real measurements. The effects of number, layout, sampling duration and sampling interval of the sensors and the number of potential sources on the identification accuracy were analysed. Our findings showed that most of the source scenarios can be identified effectively, including both one- and two-source scenarios. The identification method is most accurate when the potential sources have different effects on sensor network. The identification accuracy can be improved by increasing the number of sensors and sampling duration, and proper arrangement of the sensors, but the sampling interval has a minimal effect. The identification accuracy may decrease with the increase in the number of potential sources. Our research demonstrates the feasibility of applying the source identification method under realistic indoor conditions for various scenarios of buildings.