What is the difference between the SOLU centered gradient and SOLU upwind gradient ?
To my knowledge (from how it's done under OpenFOAM) :
we add to the implicit upwind scheme an explicit correction to the form "Grad(phi) \cdot Pf".
The gradient is a cell centered value (Gauss or LS). The cell choice depends on the sign of flux : we choose the upwind cell and the distance Pf is calculated from this cell to the face center
What code_saturne means by "centered gradient" or "upwind gradient" ?
There is by default a slope test that is enabled by default :
What is this slope test and how the limiter is calculated ? Have you any reference ?
How does it work with SOLU scheme ?
Should we turn off the limiter with SOLU as we could add excessive dissipation ?
The "upwind gradient SOLU" should be closer to the SOLU in the litterature.
The legacy "SOLU" uses a centered/regular cell gradients to reconstruct values at a face's I' and J' points.
Regarding the slope tests, from discussions with colleagues, the "second-order" aspect of SOLU could lead to the maximum principle not being respected in some cases, so the slope test can help here.
For the default slope tests, I'll try to check an old report to see if it contains external references.