Computer-aided placement of air quality sensors using adjoint framework and sensor features to localize indoor source emission

With the improvement in sensor technologies, air quality is increasingly being monitored. Two major factors in obtaining relevant information are the optimal placement and the number of air quality sensors. Moreover, in cases of poor air quality, the information of the pollution level given by the deployed sensors is not sufficient. An advanced understanding of the data is required to precisely identify the source pollution and thus propose effective solutions. In this article, a virtual testing strategy based on computational fluid dynamics (CFD) is presented for the optimal placement of indoor air quality sensors. We determine the placement of sensors in view of localizing the maximum of sources emitting on the indoor environment surfaces. Therefore, an adjoint framework is used to obtain the observable region associated with a given sensor position. The proposed method takes into account technical sensor features, such as the limit of detection (LOD). Two applications are studied: a simple 2D case and a real 3D room. In these examples, we first show that reducing the LOD of the sensors by one order of magnitude can increase the observable area by more than 50%. Then, we note that one-fourth of the potential sensor placements observe almost nothing and that 80% of the potential sensor placements have an observable area two times smaller than the optimal sensor position determined by the proposed CFD-based strategy.