Modelling flow and pressure losses in a gasoline particulate filter

  • Marco Prantoni

    Student thesis: Doctoral ThesisDoctor of Philosophy


    The reduction of transport particulate matter emissions is crucial for improving air quality. Although particulate filters are efficient in removal of particulate matter, their inclusion in modern exhaust systems results in high back-pressures, which then causes higher fuel consumption as well as other problems to the engine performance. Consequently, there is an ever-increasing demand within the automotive industry for more accurate and reliable filter design tools.

    This thesis investigates both numerically and experimentally different sources of the pressure losses in particulate filters in order to develop a fundamental understanding of the complex physics of the flow in wall-flow filters and to develop tools for modelling filter flows, with particular focus on Gasoline Particulate Filters (GPFs) and their operating conditions.

    Although friction losses and contraction and expansion losses in laminar duct flows have been studied extensively, there are still some discrepancies in existing correlations. In some flow regimes in particulate filters, the contribution of these losses may be significant. Therefore, a numerical investigation of developing flow losses and losses due to the flow path contraction has been carried out. Improved correlations for both types of losses have been suggested, and supporting experiments for friction losses in filter channels with porous walls have been carried out. These provided useful information about filter friction factor and contraction loss coefficient, which supported the physical base of the new pressure drop models developed in this work. The results of these studies can also be used for the prediction/optimisation of the pressure drop in other applications such as catalyst filters, multi-channel systems, wind tunnels and flow meters.

    In order to be able to validate the developed filter models, experimental data from a joint project with Jaguar Land Rover has been used. This data, collected as part of a team of researchers, allowed to get a better insight into the filter losses at high mass flow rates and temperatures up to 680[C]. This unique data set can be used for the assessment of particulate filter models in a wide range of filter parameters and flow conditions.

    From an industrial/practical point of view and with regard to the modelling of par­ticulate filters, the main outcomes of this work include the development and assessment of two new physics-based particulate filter models.

    Although a number of 1-D models exist, none of them consider turbulent flow regime which may be present in the filter under certain conditions. To address this challenge, a new 1-D particulate filter model has been developed which covers both laminar and turbulent flow regimes.

    Because of the large pressure variations along the filter channel at high flow rates and temperatures, density change effects have been included in the model, which allows the model to be used at both low and high temperatures and filter back-pressures. The model predictions agree well with the experimental data and at high mass flow rates and temperatures the new model can improve the pressure drop predictions up to 30−40% with respect to the original model developed by Kostandopoulos and Johnson (1989), which is often used for high Reynolds number flows despite having been developed for laminar flow regime only.

    While 0-D and1-D models of a representative filter channel are useful when it can be reasonably assumed that all filter channels have similar flow rates, this is not the case where there is  considerable variation in channel properties or the upstream flow is highly non-uniform. In order to be able to account for the difference between channels, a new multi-channel particulate filter model and modelling approach, including coupling the model with CFD simulations, have been developed. This modelling approach allows to: (i) capture complex flow interactions between channels, (ii) account for density variations within individual channels, (iii) prescribe individual channel properties (i.e. wall thickness, hydraulic diameter and wall permeability) and (iv) investigate the overall effect of a given filter configuration on the exhaust system in 3-D (i.e. upstream and downstream effects).

    The potential of both models in terms of practical applications has been demon­strated by carrying out parametric studies. The results of the new 1-D particulate filter model have shown that for most of the considered filter geometries there is a range of filter sizes providing minimum losses in the given mass flow rate range, while the results of the new multi-channel modelling approach have shown that upstream flow non-uniformity may persist through the filter, which needs to be taken into account in filter design. The effect of different wall permeability between different channels has also been demonstrated. Such insights would allow filter designers to select the best filter configuration within other constraints used in the development process.

    Date of AwardMay 2021
    Original languageEnglish
    Awarding Institution
    • Coventry University
    SupervisorSvetlana Aleksandrova (Supervisor), Stephen Benjamin (Supervisor) & Humberto Medina (Supervisor)

    Cite this