[pablitobass]

Dear All,

I would like to put to your attention an issue that I have encountered during some tests.

As the title of the thread suggest, I have noticed pretty much different results evaluating the diffusive term of the momentum equation using two different, but (mathematically) equivalent, ways.

If mu is the viscosity of the fluid, U its velocity and D = 0.5*(grad(U) + grad(U).T()) is the deformation tensor, we can write

div(2*mu*D) = div(mu*grad(U)) + div(mu*grad(U).T())

where grad(U).T() is the transpose of the velocity gradient.

For an incompressible flow, and considering mu = constant, the second term of the previous equation is zero, hence we finally have:

div(2*mu*D) = div(mu*grad(U))

On the other hand

div(mu*grad(U)) = laplacian(mu,U),

if I am not wrong, so I have implemented a piece of code to evaluate these two variables:

volVectoField laplacian = fvc::laplacian(mu,U);

volVectorField div = fvc::div(2*mu*D);

During my tests I have seen difference (~20% and larger) in the x-components and a smaller difference in the other two components (the error here might be due to a different numerical treatment).

I have also tried all the numerical schemes available but the results are always quite different.

Unless I made some mistakes during the calculations, the two terms I wrote should be "equivalent", otherwise there might be a bug somewhere.

Does anyone else came across a similar issue?

Many thanks!

Regards,
Paolo

[pablitobass]

The problem has been solved. There is a mistake in the code I am using. I did the same tests in icoFoam and the operator laplacian and div work fine.

Thanks and sorry for the wrong report.

Paolo