CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   finite volume solver problem (http://www.cfd-online.com/Forums/main/103533-finite-volume-solver-problem.html)

ufocfd June 20, 2012 07:24

finite volume solver problem
 
I am developing a uniform compressible finite volume solver,

Runs ok when using a cell size of 2.0 but somethings
going wrong when I reduce the cell size to 0.1 units.

I think the CFL & timestep specification is ok but
I think its related to the update of the Q vector:

RHS = dE/dx + dF/dy + dG/dz

Q(n+1) = Q(n) + RHS*dt/vol, where dt = timestep and vol = cell volume

Is this correct ? Have I missed something obvious ? Thanks.

FMDenaro June 20, 2012 07:37

Quote:

Originally Posted by ufocfd (Post 367415)
I am developing a uniform compressible finite volume solver,

Runs ok when using a cell size of 2.0 but somethings
going wrong when I reduce the cell size to 0.1 units.

I think the CFL & timestep specification is ok but
I think its related to the update of the Q vector:

RHS = dE/dx + dF/dy + dG/dz

Q(n+1) = Q(n) + RHS*dt/vol, where dt = timestep and vol = cell volume

Is this correct ? Have I missed something obvious ? Thanks.

I see a discrepancy, dimensionally your equation is not correct ...

If V is your Volume, then the equation writes:

d/dt [Integral_V Q dV] + Integral_V Div Flux dV = 0

Therefore:

d/dt [Q_average] + (1/vol) Integral_V Div Flux dV = 0

Hence, your RHS must be vol*(dE/dx + dF/dy + dG/dz)

ufocfd June 20, 2012 11:15

thanks very much FMD !


All times are GMT -4. The time now is 20:45.