Porous medium approach is widely used in modelling high resistance devices such as heat exchangers, automotive catalysts or filters, where details of flow distribution inside the channels are not important. This reduces the computational time considerably, as the whole length of the monolith does not need to be modelled, and the thin boundary layers in each channel do not need to be resolved. The drawback of the approach is compromised accuracy of the flow predictions downstream of the monolith, because the mixing of the individual jets coming out of the monolith channels is not accounted for. Very few studies exist where this issue has been addressed. The methods include artificial turbulence generation, inferring turbulence information from upstream, or using hybrid modelling approach to separate the flow into channels. In this study, a different technique is suggested, where the porous medium model is used simultaneously for imposing the axial flow resistance presented by the monolith, and forming the multiple jet flow downstream of the monolith. The results for a planar diffuser with a catalyst are compared with experimental data, and a marked improvement in mean flow and turbulence level predictions downstream of the monolith is demonstrated. Methodology for using this approach in different geometries and configurations is also suggested. The advantage of the proposed technique is a very simple implementation through customised resistance functions combined with the known benefits of the classical porous medium approach (smaller mesh size). This means that the monolith parameters can be changed quickly, thus speeding up the system design process, and custom channel geometries can be investigated easily, with higher accuracy of the flow predictions in the whole system containing the monolith (or multiple monoliths).