[CeesH]

Hi all,

I have to model a tank with a recirculation (cooling) loop. Since there are no in/outlets, I attempt to force the flow through the loop by setting a small fixed-velocity flow profile in a section of the loop where we know there is unidirectional, turbulent and developed flow. I do so via a UDF which supplies the typical (1-r/R)^(1/7) profile. However, the solution explodes as soon as I start solving.

At first I tried with a fixed velocity of 1 m/s, I though in that case the exploding solution may be due to forcing a non-zero velocity at the wall. However, with the UDF this should not be an issue - and the solution still explodes even if I force the outer layer of cells to have 0 velocity. Even when I set the velocity in all grid cells of the fixed-velocity section to 0, the mean velocity in the domain soon reaches c^2

EDIT: setting no-shear b.c. to the relevant cell zone walls doesn't help either, neither does using the transient solver instead of steady-state

Any ideas on why this may be?

[CeesH]

Solution appearantly was to not just fix x-velocity to desired flowrate, but also y- and z- velocity to 0