Fluids

Fluids

Regarding Transient Simulation

    • KSD
      Subscriber

      I am solving a problem where first I had assumed it as steady case.

      For steady state simulation, I used to change a lot of parameters at certain number of intervals to get my solution converge.

      Usually first I start with low courant number, relaxation factor, first order scheme and then gradually changes it.

      But now I am little bit confused that there two separate settings for time and space.

      Now there are settings under solution control (which I am assuming it is used for space) and for time settings (like time step size, max iter/time step etc..) under run calculation.

      Means can I do the same procedure applied for steady case here as well, means by keeping very small value of under relaxation factor and using First order then gradually increase it.

      The thing which confuse me, is how these both (settings under solution control and run calculation for transient simulation) are interrelated.

      Normally how do we solve such problems, should I keep every parameters fix under solution control (like keep the default values) and change only the settings of time (like time step, max iter/time step) in order to make the solution converge.

      I haven't found that many videos regarding transient simulation where they change certain parameters, it would be helpful also, if I can get some references where they mention how these two are related to each other.

      And I think courant number of solution control is just mor making matrix more diagonal and courant number of time settings is the real CFL condition we should worry about.

      Actually, I am not clear, how to tackle such problems, are there any general guidelines like start with these settings, if residuals steady out then gradually increase the parameters.

      Please Help.

    • Rob
      Ansys Employee
      For steady flows we may alter time scale factor (or Courant Number) or UR factors to aid convergence. If you do, make sure you understand that dropping these factors may mean you need to run for more iterations to get a properly converged solution.
      For transient I'd tend to leave the above factors alone and control convergence on the time step. Ie keep dropping the time step until the solution is converging well.
    • KSD
      Subscriber
      Sir, Ok Sir, I will keep that in mind. Thank You.
Viewing 2 reply threads
  • You must be logged in to reply to this topic.