[cute]

Dear members,

I am writing a 3D solver that has one ODE and one PDE.

dC/dt = -a*X, with C(0)=1
dX/dt = laplacian(DX, X) + k*X, with no slip boundary condition everywhere and X(0) = 0.05 at the bottom

I am confused how to setup this model specially ODE and need help. So far I can think as follows:

while (simple.loop())
{                 
    Info<< "Time = " << runTime.timeName() << nl << endl;         
    solve(fvm::ddt(C) == -a*X);           
    solve(fvm::ddt(X) == fvm::laplacian(DX, X) + k*X); 

    runTime.write();      
}

FoamFile
{
    version     2.0;
    format      ascii;
    class       volScalarField;
    location    "0";
    object      C;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

dimensions      [0 0 0 0 0 0 0];

internalField   uniform 1;

boundaryField
{
    AllWalls
    {
        ??????? //I don't have boundary conditions
    }
    
}

My questions are:
(1) is my setup correct?
(2) I don't have boundary conditions for ODE, what should I do in above code?
(3) I don't want to use simple.loop() in while condition, what are other options to iterate through time-steps?

(4)Plz suggest, how can I improve this setup?

Thanks for reading.

[cute]

Anyone has any suggestion.

[cute]

Hello everyone,

Any comment or suggestion in my case?

Thanks.

[marupio]

I don't think finiteVolume is set up for ODEs. OpenFOAM has an ODE solver library, src/ode. Have a look at applications/test/ODE to see how one is set up.

If you take an example from the chemistry solvers, ODEs are solved by first looping through all the cells in the mesh, then solving the ODE. That is:

// given that rho is a volScalarField
// given that C is a PtrList<volScalarField> of your ODE variables
forAll(rho, cellI)
{
    scalarField y(nEqns); // number of variables in your ODE
    forAll(y, varI)
    {
        y[varI] = C[varI][cellI];
       // solve ODE here, see Test-ODE for example
    }
}

But you should also consider the fact that you are solving a complex numerical system in two uncoupled steps. You might need to do something to couple them. I describe this in section 2.1.8 here http://mspace.lib.umanitoba.ca/handle/1993/22096

[cute]

Thanks marupio. I am not sure how to implement this. I am using OpenFoam for the last three months and still learning.

If I setup ODE using FVM, will it be wrong? I think that it should produce the same result, but less efficient.

I need to know, how we can make it more efficient?

Thanks.

[Tobi]

But you should also consider the fact that you are solving a complex numerical system in two uncoupled steps.

This is in fact something you should take into account but to be honest. We are always solving complex stuff uncoupled. Even the momentum equation is segregated and the pressure - momentum coupling is also not coupled (only if you have the explicit coupled solver that we are not using normally).

I solved something similar -> the Gray-Scott-Model <- Both equations are coupled but I also solved it segregated and without any problems. Even your equations above should be an easy task to implement.

[cute]

Thanks Tobi, I will look into it.