The existing CABARET code was around 4000 lines of Fortran 90, so this was used as an outline for the parallel version. The code included routines for; reading in the computational grid, reading from /writing to restart files, data output to Tecplot format, as well as the main computational part which can be separated into the following five steps:
- - This is the conservative predictor step. This calculation updates the neighbouring conservative type variables. Interpolation of the cell based variables to the grid points is also performed.
- - Computation of the cell centred viscous terms. This step updates the cell based variables.
- - Extrapolation step where the local cell-based characteristic splitting is performed. This is where the most intensive computation lies.
- - Applying conditions for the boundary cells. This determines the new values of the flux-type cell face values and only involves the physical boundary.
- - The conservative corrector step. This also updates the cell based variables with their values from the previous time step.