[saeedi]

I am new to OpenFoam and had a quick question for you:

I want to do an incompressible simulation and I just need to save P and U but for many time steps (which will take more than 500 GB). So, I do not want to have any extra variable saved,

I assume that variables ( U_0, phi, phi_0) are for the internal use of OpenFoam and will be saved any ways. Right? Because having these three extra variables saved will more than double the occupied space.

Can I get rid of them? as I said, I just want to have instantaneous velocity and pressure to be saved.

Thanks in advance.

[ssss]

This files are creates with the backward and CrankNicolson schemes for the ddtSchemes. Whitout them you cannot restart the simulation

[saeedi]

Thanks for your reply.

So I have to tolerate them

In my own code, I also use CN method for time advancement and only save the last velocity. but unfortunately I cannot handle unstructured grid and have to use OpenFoam for that. Some times, they may not be the most optimum way. or may be there are other reasons.

[Coke Rivas Ordenes]

Dear SSSS,

Originally Posted by ssss

This files are creates with the backward and CrankNicolson schemes for the ddtSchemes. Whitout them you cannot restart the simulation

Hope you can help me out with a topic related to the *_0 files. I have run a 5000 seconds Large-Eddy-Simulation of the ABL with a kEquation turbulence modelo. I must continue with the simulation but from the 5000 second i have a diferent mesh
(
i remove and stl object from the domain so the new mesh has three blank Cubes, the ones in the image:

blocks.png

)

To continue with the simulation from the 5000 seconds fields i created a new simulation directory and import the fields (U, p, k and nut) from the last saved time-step of the previous 5000 seconds simulation with the mapFields utility. The problem arise when i notice that *_0 files and phi (and phi_0) fields where not imported to the new "0 folder", form where i intent to re-start the simulation. Never the less, the simulation restarted with out any problem and from a simple "eye inspection" the fields imported are in agreement with the last saved time fields:

Left: fields with mapFields. Right: last saved time-step fields.

mapFields.png

Do you know why the new simulation did not require the *_0 files or the phi field considering that the simulation uses a backward ddt scheme? (Using: pisoFoam solver, kEqn turbulence model, openFoam 4.1)

Thank you for your time! And to any one willing to "give me a hand".

[ssss]

Buenas Coke Rivas,

The _0 files are not required if the simulation starts from time "0". The solver will use the Euler Scheme in the first iteration and second iterations and then it will start using the backward scheme.

If you want to interpolate the _0 files you need to create files with their names and boundary conditions in the 0/ folder of your targe source. Doing something like this in your target testcase:

cp 0/U 0/U_0

Should allow to map the U_0 from your source testcase to your target

[Coke Rivas Ordenes]

Originally Posted by ssss

Buenas Coke Rivas,

The _0 files are not required if the simulation starts from time "0". The solver will use the Euler Scheme in the first iteration and second iterations and then it will start using the backward scheme.

If you want to interpolate the _0 files you need to create files with their names and boundary conditions in the 0/ folder of your targe source. Doing something like this in your target testcase:

cp 0/U 0/U_0

Should allow to map the U_0 from your source testcase to your target

Gracias SSSS,

That was very useful.

Regards,
Coke