[bou]

Dear All,


I am using a modified version of scalarTransportFoam for simulating advection and diffusion of scalar field C in 2D over a rectangular domain. The velocity is constant over the whole domain (oriented vertically, along Y) and I use Dirichlet BC (C=0) on top and bottom and zero-gradient BC on left and right.


The problem is consequently very simple.



I keep using the solver "PBiCGStab" and preconditioner "DILU" as suggested in the tutorial scalarTransportFoam for my case.



Running my case works perfectly well as far as the mass conservation is concerned as long as the mesh is perfectly orthonormal (cells are squares). This means that summing at any time "t", the content inside the domain and what has left the domain through the boundary, gives me always the initial content (at t=0).


However, when I am using a slightly different mesh, still orthogonal but with varying mesh elements sizes (cells are rectangles) using a simpleGrading value in one direction of 10 in direction X for instance, the mass conservation is not respected anymore. What I don't understand is that from the beginning of the run, the mass content even increases...


Here is what I observe:


Time / Content inside / Flux through the BC

1.00000000e+04 2.41000001e+15 -2.94558017e-135
2.19999962e+04 2.41000002e+15 -1.67119996e-132
3.63999862e+04 2.41000004e+15 -5.17939549e-130
5.36799662e+04 2.41000006e+15 -1.16562613e-127
7.44159310e+04 2.41000008e+15 -2.13671699e-125
9.92990755e+04 2.41000011e+15 -3.39332677e-123
1.29158840e+05 2.41000014e+15 -4.84963157e-121
1.64990529e+05 2.41000018e+15 -6.39794638e-119
2.07988528e+05 2.41000023e+15 -7.93275153e-117
2.59586112e+05 2.41000029e+15 -9.36652874e-115



May you please explain me what is going on?


Best regards,