II.1 Modeling strategy

        Mesh design in problems with a moving mesh and changing cell connectivity is dominated by the need to keep the dynamic parts of the grid simple so that they can be easily changed during the transient run. In such cases, the formulation of mesh motion is divided into two conceptual steps. The first deals with connectivity changes, (cell removal, addition, reconnection, etc...) which are defined by PROSTAR `events'. The second step is to specify the grid vertex positions as a function of time by supplying a set of PROSTAR grid-manipulation commands to be executed at each time step.

        Moreover, the modelling  has to respect following considerations:

  The initial mesh must contain all the cells which will be used in the analysis. Thus, it will be seen that the initial grid construction contains overlapping cells in the vicinity of the valves. These cells are rearranged according to the valve position at time = 0, which is the start time.

   When cells are added (at the time of their activation), they are still deemed to the neighbours they had at the time of their removal (desactivation). This imposes sequencing constraints on the activation and desactivation processes.

   If any part of the solution domain becomes separated from the rest of the flow field during a transcient run (for example the intake port after the intake valve has closed), the cell material type in that domain must be changed. All material types used must have been defined in the initial setup.

            II.2 Mesh and boundary conditions

        The predefined geometry was read from external files : tut.vrt, containing the vertices, and tut.cel, containing the cells definitions. The vertices and cells files can be found here.The following mesh is obtained :

           On this image, one can see the global geometry of the cylinder: the intake port is green, the exhaust port is blue and the cylinder chamber is in red. There are two more elements which don't appear on this image, the two valves of the two port.
        The boundary locations are read the tut.bnd file. The boundary conditions used are wall, cyclic and symetry plane conditions. A constant pressure (zero-pressure) is assigned as boundary condition at the intake and exhaust ports end.
       The fluid originally in the intake port is assumed to be a different species called INTAKE. This fluid as the same properties than the air. The same way, the exhaust port is filled with another fluid species called EXHAUST.
       Since we have no further informations on the boundary conditions we think that the symetry plane conditions are used to simulate an infinite length for the third dimension which is not represented on the mesh. The cyclic conditions must be used for inlet/outlet. The tut.bnd file can be found here.
        Some boundary conditions must be defined between the various regions of the fluid domain. The boundary type used is the ATTACH type, it is used for boundary surfaces that are attached together dynamically during the course of the analysis.