Modeling the Unsteady Flows in I.C. Engine Pipe Systems by Means of a Quasi-3D Approach

Increasing demands on the capabilities of engine simulation and the ability to accurately predict both performance and acoustics has lead to the development of several numerical tools to help engine manufacturers during the prototyping stage. One dimensional simulation tools are widely used during this phase and they allow the simulation of several engine configurations within a short time. Certain components, however, such as the intake and exhaust manifolds, exhibit a high degree of geometric complexity, which cannot be accurately modelled by ID codes, unless equivalent ID models are adopted. The need of achieving good accuracy, along with acceptable computational runtime, has given the spur to the development of a geometry based quasi-3D approach. This is designed to model the acoustics and the fluid dynamics of both intake and exhaust system components used in internal combustion engines. Models of components are built using a network, or grid, of quasi-3D cells based primarily on the geometry of the system. The solution procedure is an explicitly time marching pseudo staggered grid approach, where the equations of mass and energy are solved at cell centers while the momentum equation at cell connections or boundaries. The quasi-3D approach has been fully integrated into a ID research code in order to study the behavior of intake and exhaust devices under real engine pulsating flow conditions. This approach was mainly developed to model the acoustic behavior of complex shape silencers, however, in this work it has been extended and applied to the prediction of the fluid dynamic behavior of intake and exhaust systems. The validation was carried on a high performance V4 motorbike engine. In particular, the silencer and the air box have been modeled resorting to a quasi-3D reconstruction. Calculated results of instantaneous pressure traces and volumetric efficiency have been compared to measured data, highlighting a good capability in capturing dynamic effects with a computational runtime much lower than the one required by the integration of fully 3D models with the ID.