[Pyrokrates]

Hey,


I wrote a program (OF version: v2006) to do some simulations with porous media. To reduce the jump and oscillations nearby the "inlet" and "outlet" of the porous media, I reduce the relaxationFactor of U (from 0.1 to around 0.0001) which also helps to correct the calculated pressure drop. Now I have some problems simulating loading processes. When Darcy coefficient inside my porous media changes locally, the velocity magnitude should change too (some regions get increased velocity other reduced...). Problem here is that with very small relaxationFactor, U can not compensate this change which is quite logical.


Do you have any suggestions how to be able to reduce/remove the velocity jump for larger relaxationFactors to make changes of U possible?


1) I thought about some manual correction (structured hex mesh):
Loop over the cell layer right before my porous media. Find all cells in thickness direction right behind the specific cell. Set U of the cell in front of the porous media to the mean value of the cells behind.


I´m not sure, if this is possible and converge since it changes the result of the regular calculation of field U. Another problem is following: How can I get the cells after the specific cell in front of the porous media... For flat porous media I could loop over my filter cells and find the cells with matching x and y values (flow direction is z) but for e.g. more complex filters like pleated ones this is kind of hard to achieve.


2) A second idea would be a split of solving my equations over different time steps. I have UEqn, pEqn and the new one for the loading process XEqn. My idea would be following: Calculate Ueqn and pEqn inside a inner loop where I can use e.g. three relaxation factors: 0.1 for the velocity change, 0.01 for lower osciallation and 0.0001 for a stable result of velocity U paired with the correct pressure p. After this inner loop I can do one single calculation for updating XEqn.


So what I imagine is something like this (just simple example):

while(simple.loop())
{
    while(i<1000)
    {
        //relaxU = 0.1
        #include "UEqn.H"
         #include "pEqn.H"
        i++

     }


     while(j<500)
    {
         //relaxU = 0.001
        #include "UEqn.H"
         #include "pEqn.H"
        j++
     }

     while(k<100)
    {
         //relaxU = 0.001
        #include "UEqn.H"
         #include "pEqn.H"
        k++
     }

    #include "XEqn.H"

    runTime.write();

    runTime.printExecutionTime(Info);
}

Do you have any suggestions about my ideas or maybe another plan?



Thanks in advance


Pyro