[nskelly]

Hi,

I am looking to use a laminar steady state solution as an initial condition for a laminar transient case. I know it is possible to do so on CFX, but unsure how to do it on OpenFOAM.

I have done the steady state case with simpleFoam, and I guess I need to change the solver to pisoFoam for transient, and the ddt scheme to a transient scheme. I am a bit unsure about the startFrom and startTime, because the steady state case is in iterations, but a transient case is in time steps.

I have read through some posts but I can't find one where someone has managed to do this (please let me know if there is a thread about it)

I think I can use mapFields but unsure how to use if I have the same mesh. Most probably more straighforward than I am thinking. Also is it possible to do this with a decomposed case?

[arvindpj]

just change the startFrom flag in the controlDict also with the solver.

 startFrom       latestTime;

Cheers,
Jay

[shereez234]

Originally Posted by arvindpj

just change the startFrom flag in the controlDict also with the solver.

 startFrom       latestTime;

Cheers,
Jay

Jay ; I don't think that this is the only thing to change.

OP;

You need to first change the delta T to a smaller value representing courant number <1 if pisoFoam is your choice of solver.

YOu need to change the fvSchems time marching scheme from steady state to Euler, backward or Crank nickolson (or etc).

If it is the same mesh you don't need to map the results as well just take the saved data for the latest time step in simpleFoam. Inside the data files there will be a file called uniform > time. Edit the index of time to one, time delta value to the desired delta T value.

From there I think your simulation can start.

[nskelly]

Originally Posted by shereez234

Jay ; I don't think that this is the only thing to change.

OP;

You need to first change the delta T to a smaller value representing courant number <1 if pisoFoam is your choice of solver.

YOu need to change the fvSchems time marching scheme from steady state to Euler, backward or Crank nickolson (or etc).

If it is the same mesh you don't need to map the results as well just take the saved data for the latest time step in simpleFoam. Inside the data files there will be a file called uniform > time. Edit the index of time to one, time delta value to the desired delta T value.

From there I think your simulation can start.

Thanks for your answer.

I've change fvSchemes to make it transient, and fvSolution for PISO loop. I have set my required timestep 1e-5 and changed the solver to pisoFoam.

So in the last time directory e.g. 100. I change index - 1, deltaT - 1e-5 and deltaT0 - 1e-5?

And in the controlDict I just change startTime to latestTime (which should be 100 directory)?

[LuckyTran]

Originally Posted by nskelly

So in the last time directory e.g. 100. I change index - 1, deltaT - 1e-5 and deltaT0 - 1e-5?

And in the controlDict I just change startTime to latestTime (which should be 100 directory)?

The uniform=>time folder contains info for statistics and you can even leave this alone if you reset your statistics or delete this folder entirely. index is how many time-steps have been averaged and deltaT and deltaT0 is the current and previous time-step values. These are needed for resuming calculating averages when you restart a simulation. You should never change these values. The size of your time-step is placed in the controlDict.

Only thing you might want to do is change 100 to 1 or 0. But this is not absolutely necessary because OF will just think that the current time is 100s.

When you do startFrom latestTime;
OF runs a utility that scans your case directory and looks for the biggest time. You can also do:

startFrom startTime;
startTime 100;

which will have it start from time 100.
This is same for transient as it is for steady state except that the meaning of time is iteration for steady state and time for transient.