Estimation of an aerosol source in forced ventilation through prior identification of a convolutive model

Florent Chata (1, 2) , Emmanuel Belut (1) , Denis Maillet (2) , Francois-Xavier Keller (1) , Anne Taniere (2)

1 INRS ( Vandoeuvre lès Nancy) - Institut national de recherche et de sécurité (Vandoeuvre lès Nancy)
2 LEMTA - Laboratoire d'Energétique et Mécanique Théorique et Appliquée

This article presents a method for estimating the time dependent generation rate of an aerosol source starting from a transient concentration signal measured at a distant point. The method is made up of two distinct steps: a calibration phase, followed by an estimation phase. The calibration phase consists in identifying a transfer function (termed "impedance") between a known source (the " calibration source ") and its measured concentration response. In the second step the unknown source generation rate, in the configuration of interest, is estimated by inversion of the corresponding measured concentration signal at the same point, using the previously identified impedance. The time integral of this generation rate, the emitted aerosol dose, can be calculated directly, starting from the integral of the (transient) impedance. Here both simulation of inversions and application to a real experiment have been implemented. The results confirm that it is possible to estimate the temporal pattern of injection and the global emitted mass of pollutant.

Mots clés: truncated singular value decomposition; aerosol source; unsteady conditions; inverse problem; computational fluid dynamics; convolutive model

Domaines: Mécanique des fluides [physics.class-ph]; Génie des procédés

Florent Chata, Emmanuel Belut, Denis Maillet, Francois-Xavier Keller, Anne Taniere. Estimation of an aerosol source in forced ventilation through prior identification of a convolutive model. International Journal of Heat and Mass Transfer, 2017, 108, pp.1623 - 1633. ⟨10.1016/j.ijheatmasstransfer.2017.01.015⟩. ⟨hal-01501952⟩