Nowadays quasi-3D approaches are included in many commercial and research 1D numerical codes, in order to increase their simulation accuracy in presence of complex shape 3D volumes, e.g. plenums and silencers. In particular, these are regarded as valuable approaches for application during the design phase of an engine, for their capability of predicting non-planar waves motion and, on the other hand, for their low requirements in terms of computational runtime. However, the generation of a high-quality quasi-3D computational grid is not always straightforward, especially in case of complex elements, and can be a time-consuming operation, making the quasi-3D tool a less attractive option. In this work, a quasi-3D module has been implemented on the basis of the open-source CFD code OpenFOAM and coupled with the 1D code GASDYN. The flexible OpenFOAM platform has been exploited for the development of both the quasi-3D cell-network generator and the implementation of the non-linear explicit quasi-3D solver. The grid generator is based on a fully automatic strategy, which takes as input the definition of the geometry in STL format. It works in two steps, firstly creating a refined background mesh, including all the small details of the geometry, and then agglomerating the cells, reducing the number of elements, in order to reach the desired cell density. The implemented quasi-3D solver, relies on a high accuracy numerical method specifically developed for working on such a coarse grid and it is adopted for the solution of the 3D domain. The coupled 1D/quasi-3D simulation tool has been firstly validated considering different silencer configurations, comparing the computed Transmission Loss with experimental values. Then, as further step of the validation, an entire engine configuration, including complex intake and exhaust systems, has been simulated. Furthermore, the performances of the quasi-3D numerical method are compared with those of a standard CFD solver, with particular focus on the numerical accuracy and computational burden of the two approaches.
