Quadratic porous media expression causes divergence.
User tried to apply general quadratic polynomial resistance law like S = -C0 - K*V -C*Rho
*Vmag*V. Residuals decrease quite nicely but all of a sudden, it diverges. Double precision delayed the divergence but it still blows up in both single and double precision.
Here is an explanation:
User specifies power-law porous resistance with C1 = 0.
It means that resistance source term is constant for all directions and does not depend on velocity at all. In this case FLUENT still conducts the linearisation of this term with coefficient of linearization C0*Vmag^(C1-1).
If Vmag is not zero, math libs function call does not fail.
If Vmag is 0 (due to round off) then the power low function gives INF. This is exactly what happens in this case.
To resolve this, the porous media setup needs the following change:
"whenever there is an expression of the form (S = -C0 - K*V -C*Rho *Vmag*V) for porous media then user must put the constant term (C0) in the momentum source and not in the porous zone BC panel".
Under "porous" media, simply enter the K and C coefficients for viscous and inertial resistance and enter C0 as constant "source" term for momentum equation.