Negative values of relative permeability in lbpm_color_simulator #35
Comments
|
Can you attach the input file that you used and the output of the simulation? Usually if the relative permeability is negative it means the flow is blocked for some reason (i.e. pore space is blocked in the image). It would be good to check that the image is being read into the simulator correctly (by visualizing the input file in python). |
|
Here is the input file: ` Note: The simulation exited after the time limit assigned for the job has been reached. |
|
This is exiting because you are hitting the requested walltime (48 hrs = 172,800 seconds). The scheduler is killing the job once the walltime is exceeded,
For a CPU based workload, you can specify smaller sub-domains. You will be able to run this job on many more than 8 processor cores with smaller sub-domains assigned to each core. The time-to-solution will be significantly faster if you do this. For a CPU-based system it would be interesting to try the following domain decomposition This will require
And you will need to request 324 cores from LSF. I'm not sure if you have this many cores available on your system, but LBPM should scale well enough to run this way if you do. |
|
Do you think running like this will make the relative permeability values correct? Because I have run another case with only sub volume of Here is the input file: Here is the output file: and here is the timelog.csv: |
|
Note that you are running an unsteady simulation here, since you have specified The unsteady protocols are a bit tricker to interpret, since there can be capillary end effects as well as dynamic effects, so just be aware of this. For the negative relative permeability values, the sign was flipped based on the way the |
|
Can you tell which number of |
|
For steady-state you should set a periodic boundary condition with The other boundary conditions that are used the most are
|
|
So have the boundary conditions in morphological initial conditions to be the same as the one will be used in the subsequent steady state or unsteady state process? |
|
The output of the morphological step is an 8-bit image, so it can be read into the simulators in like any other image. You can specify different boundary conditions if you want, but you do not have to. If you are doing history labeling, it is useful to apply boundary layers prior to the morphological step so that the history labels extend into the mixing layers. This is really only important if you are using periodic BC / fractional flow simulations. |
|
Thanks very much. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I have run the
lbpm_color_simulatorpresented in the tutorial on 8 cores CPU but the results of relative permeability are negative as seen in thetimelog.csvfile:The text was updated successfully, but these errors were encountered: