Particulate filters are used to remove harmful particulate matter (PM) from the exhaust gas of automobiles. They are required in most modern vehicles to achieve compliance with legal limits set on vehicle emissions and are often combined with a catalyst to increase their functionality as an emission control technology. Knowledge of the exhaust gas hydrodynamics in filter systems is crucial for the optimisation of their structure and operation. The nature of most filters, i.e. opaque and brittle, means most anemometry methods are inapplicable. In this work, magnetic resonance imaging (MRI) is applied to measure gas hydrodynamics in such filters. This methodology is used as it provides a non-invasive means of measuring gas flows in complex geometries.

The first part of this thesis focuses on the fundamental hydrodynamics of gas at the ends and in the channels of filters. The first measurements of laminar and turbulent flows at the filter entrance and exit are made. These show the features expected from literature simulations but also additional turbulent effects not previously predicted. Possible challenges facing future CFD comparisons with this work are also explored. The gas flow fields in the channels are measured, allowing calculation of the through-wall or filtration velocity. This provides data for comparison with 1D and 3D CFD models at a range of flow rates. The 3D CFD model is validated for modest Reynolds numbers and allowed parameters such as flow profile shape and wall friction to be explored. The 1D model shows agreement with the measured data when these parameters were allowed to vary with the through-wall velocity. The MRI data is also combined with an analytical filtration model to predict the filtration behaviour and total filtration efficiency of PM.

The second part of this thesis explores the preparation and operation of filters. Catalysts are applied to filters using a washcoat slurry, a process which can cause non-uniform distributions of washcoat in the filter and perturb the gas flow. Washcoat distributions are difficult to measure non-destructively due to the filter opacity. MRI allows perturbations to the gas flow fields to be measured and changes to the wall permeability estimated using the aforementioned 1D model. This is performed for three model washcoated GPF samples. The permeability estimations agree with porosimetry measurements and reveal a variety of non-uniform washcoat distributions in the filters, which is predicted to impact the pressure drop and filtration behaviour. This method is applied to commercial filters which are loaded with PM from a real-world engine to three different levels. The PM-free filters show a non-uniform washcoat distribution and the PM deposits mostly in the regions of low washcoat loading and high through-wall velocity.