Aerodynamic Flow Simulation in an Internal Combustion Engine Using the Smoothed Particle Hydrodynamics Method 2011-24-0029
The numerical simulation of internal aerodynamic of automotive combustion chamber is characterised by complex displacements of moving elements (piston, intake/exhaust valves…) and by a strong variation of volume that cause some problems with classical numerical based mesh methods. With those methods (FEM, FVM) which use geometric polyhedral elements (hexaedron, tetrahedron, prismes…), it is necessary to change periodically the mesh to adapt the grid to the new geometry. This step of remeshing is very fastidious and costly in term of engineer time and may reduce the precision of calculation by numerical dissipation during the interpolation process of the variables from one mesh to another.
Recently, the researcher community has renewed his interest for the development of a generation of numerical to circumvent the drawbacks of the classical methods. Among the large variety of innovating meshless methods, the so-called Smoothed Particle Hydrodynamics (SPH), appears to be suitable to describe the fluid dynamic equations which are generally the most studied.
This paper presents the SPH formalism used in order to compute compressible flow with solid boundary conditions. The physical model used for compressible simulation is first described. In particular the treatment of compressible fluid is detailed with the use of a Riemann solver and its implementation in SPH formalism. The strategy chosen is to use only boundary particles to model walls and this specific solid boundary numerical treatment for compressible fluid is described. This original hybrid boundary treatment is based on a partial Riemann problem at walls.
In order to test the accuracy of the solid boundary condition and of the SPH formalism in presence of moving wall and to handle a fluid with increase of density and pressure i.e a compression, relevant test cases are presented. Finally, a basic experimental engine is computed with the method presented in this paper and results are compared to experimental data (PIV).