Following earlier publication of a model of crack flow and filtering based on a 2-D crack representation, the model has been improved and submitted to validation by comparison with data from a number of other theoretical and experimental studies. This process has led to improved understanding of gas flow and filtering in real crack geometry and provided evidence that the model predictions of barrier filtering capacity are likely to be conservative for realistic containment conditions. Our analysis also indicates that, because of drag effects related to the existence of reduced aperture sites, the viscosity limited flow cannot be approximated by the plane Poiseuille model. As a result, theoretical models employing this approximation over-predict measured flow-rates by around an order of magnitude.
