Procedure to run unsteady simulation?
 

Dear all I am extending my code from steady simulation to unsteady simulation. I want to know the procedure to run unsteady simulation.

First, What is the starting guess to run the simulation? Can I use stagnant fluid?

Second, When do I start to average the results?

Any suggestion is very please to me.

Thank you in advance.

STN

Re: Procedure to run unsteady simulation?
 

Often the start is a dirty, or incompletely converged steady solution.

Start the time-average when the start up transients have dissapeared. Monitoring a parameter such as drag can help you decide when this occurs. About 3 - 10 characteristic lengths, or the time for a particle to travel past the geometry about 3 - 10 times, is usually required. Reflective boundary conditions can make this time longer.

I make my time-averages very long for good accuracy, 50 - 100 characteristic lengths. This depends on how chaotic the flow field is.

Re: Procedure to run unsteady simulation?
 

Hi Linda

Thank for your response.

Anyway, you say that the starting guess may use an incompleted converged solution of steady flow. Is it mean that it is impossible to simulate the starting flow ?

And How long is the computation time that you used to get an good accurecy average results?

Thank you STN