# How to check whether an unsteady simulation is convergent or divergent

 Register Blogs Members List Search Today's Posts Mark Forums Read

 April 19, 2022, 10:42 How to check whether an unsteady simulation is convergent or divergent #1 New Member   Join Date: Jan 2022 Posts: 20 Rep Power: 3 Hey there Foamers! When dealing with steady simulations I check the residuals to conclude whether the simulation converges or not: if most of the parameters are associated to a low residual value (without remarkable oscillations) after 1500/2000 timesteps then we can conclude that the simulation is convergent. For example, I run a rhoSimpleFoam simulation and obtained the following residuals https://ibb.co/qnTyCY7. The above is my understanding, please let me know if it's not right. OK, but what if we deal with an unsteady simulation? How to check in a trustful manner (i.e. in a quantitative way if possible) whether the simulation is convergent or not? The residuals are not helpful in this case, because all variables oscillate within certain residual value range (for instance, see the pressure variable https://ibb.co/gzX4YnR; obtained using interPhaseChangeFoam). I guess that it depends on what unsteady solver are we using. I am running interPhaseChangeFoam. Thank you!

 April 20, 2022, 03:15 #2 Senior Member   Join Date: Apr 2020 Location: UK Posts: 488 Rep Power: 11 For an unsteady simulation, "convergence" means that the solution has achieved an accurate enough solution at that time step before moving on to the next time step. The solver (you are using interPhaseChange, so it is running PIMPLE) performs a number of inner interations (PIMPLE loops) to solve at each time step - you can look at the residuals of those inner iterations to see whether they have dropped significantly from their initial values; if they have not, then that is an indication that your solution is not very well converged at that time step ... and of course any errors at that time step will propagate to later time steps. In that case, you might benefit from increasing the max number of inner iterations to make the solver work harder, or you may need to reduce the time step and/or improve the grid. Hope that helps. JD_PM likes this.