Hello everybody,

I am computing supersonic flows in transient mode with CFX. I took a time step giving small (around 5) CFL number and with maximal 5 iteration loops between each real iterations, as precognized by the CFX Manual. It gives a physical time step of around 5.10-5 s. The mean velocity of the flow is around 4000 m/s and the total length of the domain around 10 m. It means that a fluid particle takes around 2.5 ms to travel along the domain. However during the transient simulation, the code computes between each physical timesteps: 5 (loops) x 5.10-5 s = 0.25 ms . If I really get the thing, it means in this case that between each physical time step the convergence is not achieved? Must I increase the number of coeficient loops to simulate at least one characteristic time between every time step ? Or I may be did not get something other? Thank you for your help.

Hi,

I don' t know what you mean by 5 loops x 5e10-5=0.25ms. The 5 iterations inside a timestep are not time increments themselves, they are iterations of the equations to achieve convergence at the given timestep size (5e10-5s in this case).

If you are having problems with convergence then smaller timesteps again will probably help. Also adaptive timesteps can be very useful in this type of simulation to allow the timestep to grow as the flow settles out.

Glenn Horrocks