IFAC PapersOnLine 51-13 (2018) 210–215 ScienceDirect Available online at www.sciencedirect.com 2405-8963 © 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2018.07.280 © 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. 1. INTRODUCTION Tropospheric PM10 is one of the most important air pollu- tants, in particular due to its impact on human health and ecosystem equilibrium (Zhang et al., 2007). PM10 forma- tion and accumulation are heavily nonlinear processes, de- pending on a high number of chemical reactions, physical transformations and meteorological conditions. They are driven by the emission of a number of pollutants(primary PM, nitrogen oxides, sulfur dioxide, ammonia, and volatile organic compounds). For these reasons, the selection of suitable PM10 control strategies is a really challenging is- sue that regional Authorities have to address (Pisoni et al., 2010), (Vlachokostas et al., 2009). Mathematically, this problem can be formalized as the control of a distributed parameter nonlinear system, subject to a forcing due to a not manipulable input set (essentially meteorological vari- ables) on one side and to a controllable input set (emission levels of a number of pollutants due to different activities) on the other. Due to the high economical and social impact of any decision (Carnevale et al., 2016), the definition of priority actions to be taken is a key issue (Carnevale et al., 2008). Source-apportionment (Belis et al., 2013), allowing the computation of the PM10 concentration caused by emissions coming from a certain activity, is one of the most important methodologies to address this point. In litera- ture, a large number of source-apportionment techniques, starting from the chemical analysis of collected PM10 samples, are considered and reviewed (Belis et al., 2013). The main drawback of these techniques is related to the fact that they give information only on the limited number of points where the samples are collected. In recent years, a series of new source-apportionment techniques have been developed, mainly based on the application of mathemati- cal models, relying on emission inventory and meteorologi- cal fields. Source-apportionment (SA) for primary PM can be performed using relatively simple approaches, because source-receptor relationships can be considered linear, and any Gaussian steady-state or Lagrangian puff model can be used. In this case, source apportionment is based on the hypothesis that PM coming from different sources do not interact. This assumption breaks down for secondary PM pollutants (e.g., sulfate, nitrate, ammonium, secondary organic aerosol) where nonlinearity, in particular due to chemical transformations, cannot be neglected. Eulerian photochemical grid models are the only possible tools to meet these issues. Nevertheless, grid models do not nat- urally provide source-apportionment because the impact of all sources has been combined in the total pollutant concentration in each cell. Different source-apportionment techniques have been developed to retain the advantage of using a grid model to describe the chemistry of secondary PM formation and also provide source-apportionment. In literature, SA evaluation usually follows two approaches: zero-out modelling (Yarwood et al., 2007) and reactive tracers method (Dunker et al., 2002). The first method is based on the fact that a specific emission is set to zero and the consequent change in output is evaluated. Zero-out Keywords: nonlinear models, air pollution, mathematical models, validation Abstract: In this work a source-apportionment approach is presented and applied in order to define priority on emission control strategies to limit PM10 concentrations in atmosphere. Emission control priority is defined starting from the results of the application of the source- apportionment module of the CAMx chemical transport model. The module, based on reactive tracers approach, allows to take into account the effect of the heavy non linearity affecting the formation and accumulation of PM10 in the atmosphere. The methodology has been applied on Northern Italy, a region often affected by high PM10 levels both in cold and warm season. The application results are presented in two different steps. In the first one, the validation of the overall CAMx model is presented, in terms of its capability to describe the phenomena involved in PM10 accumulation and formation. The second step concerns the application of the model to define priority on PM10 control actions. The validation results show high performances both in terms of mean values and other statistical indexes, with correlation around 1 and normalized errors close to 0 for almost all the considered validation stations. The source-apportionment methodology application highlights that control of emissions due to domestic heating, transport, agriculture and industrial processes is the main priority for a decision maker, accounting for the 70% of the PM10 levels. Department of Mechanical and Industrial Engineering, University of Brescia, Via Branze 38, 25123 Brescia, IT (e-mail: claudio.carnevale@unibs.it). C. Carnevale E. De Angelis G. Finzi A. Pederzoli E. Turrini M. Volta A non linear model approach to define priority for air quality control