[didomenico]

Hi everybody,
I have tried to manage a viscous version of rhopSonicFoam to solve high-Mach number flows.
Since the equations are written in conservative form, the viscous terms have to be formulated in an explicit manner (as far as I know):

fvVectorMatrix rhoUEqn
(
fvm::ddt(rhoU)
+fv::gaussConvectionScheme<vector>(mesh, phiv, rhoUScheme).fvmDiv(phiv, rhoU)
);

solve(rhoUEqn == - fvc::grad(p)
+ fvc::div(mu*dev2(fvc::grad(U)().T()))
+ fvc::laplacian(mu, U)
);

fvScalarMatrix rhoEEqn
(
fvm::ddt(rhoE)
+ mvConvection.fvmDiv(phiv, rhoE)
);

solve(rhoEEqn == - mvConvection.fvcDiv(phiv, p)
+ fvc::laplacian(mu * Cp / Pr, T)
);

Question #1: is this the only way to include them in a conservative form of N-S equations?

The computation of a bluff body at Mach 15 is performed easily, and a grid-independent solution is reached. On the other hand, a flat-plate parallel to the flow (Mach 9.9) shows instabilities at the first cells as soon as the grid is refined (gradients are sharper).
I have also tried to increase the nOuterCorrectors and nCorrectors but the solution diverges within the timestep!

Question #2: any idea (or reference) about how to solve the problem?

Question #3: have you references where I can get a description of the multivariateGaussConvectionScheme? I would like to know how it works and why it is used in this solver...
Thanks
Max