Fluids

Fluids

Question on Checking Convergence Criteria for Transient Simulation

    • helen.durand
      Subscriber

      Good morning,

      We have been running a transient simulation, but the "k" residual does not drop to a value below 0.002. Would you have any thoughts on why this might be happening?

      We know how to efficiently check convergence with steady-state simulations (by checking both convergence criteria with respect to residuals and quantities of interest). How to check for convergence in transient simulations besides monitoring the residuals?

      In advance, thank you for your support,

      RGDiscovery

    • DrAmine
      Ansys Employee
      Same applies fir transient but now assessing convergence every time step. First check if you a good mesh and that you are using the right models ( example using standard k epsilon model on fine near wall mesh is a bad choice depending on the wall function you are using or using k rpsilon model to predict separation will also fail).
      Also consider like in steady state the behavior of monitored variables but now versus iteration: ideally the variables should not change any more towards the end of the time step.
    • helen.durand
      Subscriber
      Thank you very much for the great feedback.
      For a transient simulation, we have been trying to set up the condition you suggested for convergence by monitoring variables with respect to the number of iterations. However, ANSYS closes immediately once the stop criterion is met which happens after a few iterations within the first time step (the number of time steps is greater than one). Below is the attempt to set up the suggested condition using TUI. The idea is that we check the stop criterion for convergence at every iteration after a reasonable number of iterations have passed (e.g., 50 iterations). Would this capture the intention to check for convergence towards the end of the time step? If so, why did ANSYS finish the simulation immediately after the stop criterion is met?
      ;------------------------------------------------------
      ;TRANSIENT (MONITOR SURFACE FOR CONVERGENCE CRITERIA AS WELL) check-for convergence every iteration (option 2)
      /solve/convergence-conditions/check-for 2 q
      /solve/convergence-conditions/conv-reports/add r1 report-defs f1out-hfile stop-criterion 0.000001 initial-values-to-ignore 50 previous-values-to-consider 10 active? yes q q q
      /solve/convergence-conditions/conv-reports/add r2 report-defs f2out-hfile stop-criterion 0.000001 initial-values-to-ignore 50 previous-values-to-consider 10 active? yes q q q
      ;------------------------------------------------------
Viewing 2 reply threads
  • You must be logged in to reply to this topic.