[punyaplaban]

Hi guys,
Hope you all are doing well. I am trying to simulate a wing-fuselage combination using SU2 at low Mach numbers (~ 0.3) and I am using AUSMPLUSUP for the simulation. I have also set USE_ACCURATE_FLUX_JACOBIANS= YES. I get the lift and drag co-efficients being converged but the residual keeps increasing (e.g. from -1.52 to -1.51 in the next iteration) and the simulation runs forever. Any suggestion would help.

[aero_amit]

You can take clues from available test cases in -
SU2-7.1.0/TestCases/rans folder

Roe scheme should also do a satisfactory job for M=0.3 condition simulations. you can try VENKATAKRISHNAN_WANG limiter as well. Be careful with CFL values (higher initial values may sometimes lead to problems with convergence).
Also if your force/moment coefficients are converged to the right value (even though residuals are stalled), it's fine. There is another convergence criterion in SU2 (CONV_CRITERIA=CAUCHY), which works on the above instead of residual drop, you can give a try to that as well.

[punyaplaban]

I tried the above setting but the lift and drag co-efficients just alternate between two values and the residuals still increase. Any suggestions?

[aero_amit]

For many of the geometries (practical) I have tried, it was rare to get 6 order residual drop or CACHY criterion meeting at 6th or 5th decimal place. You can plot the convergence of force/moment coefficients w.r.t. iterations and see if the mean appears to be stabilized between two small values (graph will look flattish in later iterations). In such case, you can take mean of coefficients for last 100, 500 or 1000 iterations (whatever seems appropriate). Many times I end up with 2 to 4 orders of residual fall (than it stalls) and small oscillations in coefficients. If one is getting considerably large oscillations, it may indicate the need to run an unsteady simulation.
Regarding residual rising, if first it is falling few orders, than rising a bit and becoming flat, it should not be a problem.