The configuration that is going to be studied in this report is a free surface flow (e.g. see Tutorial 15, StarCD 3.15 User Guide). The model consists of a water column collapsing under its own weight and hitting an obstacle situated a small distance downstream, as shown on figure below.

**Click on picture to see the
grid ...**

The grid has been built as
described in Tutorial 15, then refined twice in the surroundings
of the obstacle. The final version of the grid has **100989 cells** : using the sequential code, thus one single processor, it is obvious that the
calculation time will reach several hours, and even several days.

In order to study the calculation time per iteration, the simulation
has been run simply on 10 iterations. In this case, the time step per iteration is rather
dissuasive : indeed, the CPU time for 1 iteration is equal to **125.40 seconds**, whereas the elapsed
time is **133.33 seconds**. Since this first stage of the computation has already needed approximatevely
20 minutes and does not go further in the simulation, it is not even conceivable to make the
calculation on a greater number of iterations.

Therefore, it is now relevant to run the parallel computation to enhance the performances of the machine.