[Lorenzo92]

Hello,

I' m trying to force openFOAM not to write during the calculation the fields U_0 and phi_0. I'm currently using backward schemes.
The reason for that is that I'am running a huge DNS on a HPC ( about 100milion cells) cluster and due to memory shortage and also due to speed issue related with the writing of such mentioned fields, I can't afford to store neither U_0 or phi_0.

I also tried to restart a simulation , a very small one just as a test, deleting folders U_0, phi_0 from the latest timeStep and the solver goes on without complaning so these fields are not necessary.

Does anyone know where istructions for writing U_0 and phi_0 are located?
I really appreciate any kind of help , thank you

Lorenzo92

[elones]

These field are indeed necessary if you want to preserve second-order accuracy from first time-step. To your equation, I guess it will be somewhere in src/OpenFOAM/ and look for write().

[Lorenzo92]

Thank you elones for reading,

so do you mean that if I do not write fields *_0 and then I restart the simulation I should lose second order accuracy?
My doubt arises since the simulation's restart is possible also deleting such fields from the latest time folder , I would expect that the program complains about the absence of these fields but this doesn't happen.

[elones]

Look at the code of backward time scheme. It use only first-order Euler scheme for first time step if "oldold" time level (*_0) is not provided. The program runs without complains because if you are start for the first time you have only initial condition (old) and not "old" + "oldold" needed for backward time scheme.