Identification Of Particulate Contaminant Sources In Enclosed Spaces With Inverse CFD Modeling

Suspended particles are regarded as one of the major airborne pollutants, and can have adverse effects on human health. Accurate identification of particulate source locations is always helpful to minimize the pollutant exposure by removing or isolating the sources, if sensors have detected the release of particulate contaminant. In addition, knowing the pollutant source information can be valuable to deploy detection sensors and help improve the indoor ventilation strategies. Towards these aims, this study is in attempt to identify the particulate source locations with inverse modeling, through recovering particulate contaminant transport history based on known concentration field provided by some hypothetical sensors.This thesis proposes two inverse computational fluid dynamics (CFD) modeling methodologies:the quasi-reversibility (QR) method and Lagrangian-reversibility (LR) method. The QR method was developed previously to identify indoor gaseous contaminant sources, through reversing the time proceeding direction and replacing the second-order diffusion term with a fourth-order stabilization term in an Eulerian reference frame. In order to better capture the particulate sources, the method is modified by considering the gravitational settling effect; hence the deposition flux towards the floor is reversed. The QR inverse model assumes that the indoor airflow is steady and all particle concentration information can be provided by the sensors.In addition, the LR method is also developed to identify the particulate sources by recovering the particulate transport trajectories in a Lagrangian reference frame. The reversed particulate trajectories are obtained by reversing the airflow and the forces on these particles and then solved backward until the initial release. Some particles may be trapped or escaped at the boundaries in the forward transport process, so the boundaries should be specially treated by compensating back the trapped and escaped particles. To reduce the required sensor numbers, the track particles are proposed to put at the zone boundary lines which enclose the largest level of known particle concentration contour within the domain at a time. The scope of these particle positions at t=0 s is thought to be capable to encircle the actual contaminant source.The proposed methods are applied to identify particulate contaminant source released instantaneously from a point on the bottom surface or in a left bottom corner of a two-dimensional square cavity. Then the methods are further tested in a three-dimensional aircraft cabin where a pollutant was released from the nose of the middle passenger. The results show that both the QR method and the contaminant source zone prescription method using the LR model can accurately locate the particulate sources released instantaneously at a spot. The QR method performs slightly better than the LR model but is much more computationally demanding. Therefore, the LR model together with the contaminant source zone prescription strategy is more appropriate to be applied in practice.