[nikhil108]

Hallo Foamers,

I am doing a leak simulation using rhoReatingBuoyantFoam, Which is a transient solver. I have to run like 30 cases with different velocities, and different leak flow rates. I have to run all the cases until 240 seconds so they are comparable. But, for some simulation cases that I am dealing with, they show steady behavior after some time like 20 seconds. In cases like this is it possible to switch simulation from transient to a steady-state automatically if there is no change in the respective field I am interested in. Transient for 240 seconds is taking days. As for some case variants, it's a complete waste of time to run after they reach a steady state. Changing ddt scheme from euler to steadyState gives an error. Can transient solvers be used in steady-state? I don't know, how people deal with this kind of stuff, how to proceed further, any ideas?

Please enlighten me if my question itself is wrong ?

[ErenC]

Well, you can define convergence criteria in fvSolutions and when it converges simulation will stop.

[nikhil108]

Yeah! but the thing is, I am using Courant number of 10, residuals are in the range of 10e-3, and they are not going down throughout the simulation time. As the behavior of the results are physically making sense, I didn't cared much about residuals. In the beginning, I tried using the Piso loop to make sure explicit terms converge at every time step, but there is not much change in residuals irrespective of how many n correctors I use. As the simulation is running without errors, I ignored the residuals and carried on running the simulation cases :\. well, I know it's not completely the right way, but I didn't had a choice. Is there is any approach to deal with my situation? Thank you.

[ErenC]

Oh so you have steady-periodic solution. Well, I have no suggestion at this point. Lets hope someone will answer and I'll learn another thing.

[CFD-HSNR]

Originally Posted by nikhil108

Yeah! but the thing is, I am using Courant number of 10, residuals are in the range of 10e-3, and they are not going down throughout the simulation time. As the behavior of the results are physically making sense, I didn't cared much about residuals. In the beginning, I tried using the Piso loop to make sure explicit terms converge at every time step, but there is not much change in residuals irrespective of how many n correctors I use. As the simulation is running without errors, I ignored the residuals and carried on running the simulation cases :\. well, I know it's not completely the right way, but I didn't had a choice. Is there is any approach to deal with my situation? Thank you.

Hello Nikhil,

Since you are calculating transiently, the recorded values are the initial residuals, i.e. the residuals of the flow variables to be solved for each iteration step made by the solver before the actual solving of the flow field. For a transient simulation it is generally sufficient if the variable to be calculated converges within the iteration step, i.e. fulfils the convergence criterion specified in the fvSolution after the calculation has been carried out.

To comply with the CFL criterion, the assumption of a Courant number < 1 is also recommended, which will also have a positive effect on the convergence of your solution.