[roo]

hi

my problem relates to ground heat exchanger, where the water in the pipes exchange heat with the ground and vice versa. Make a steady calculate the input values ​​into the pipe defined mass flow rate 1.2 kg.s-1 and a temperature of 328.15 K. After 200 iterations when the solution is converged, switch model for unsteady state and set the input to the pipe by means of mass flow rate 0.1 kg . s-1 and a temperature of 293.15 K. time step is set after 86400 s immediately after starting the calculation begins diverge solutions and writes "divergence detected AMG solver: temperature" and several faces shows reversed flow.
I think the problem is the rapid change in mass flow and large time steps.
Any ideas how it could be solve this problem, while maintaining a low mass flow and large time steps?

thx for your attention

[oj.bulmer]

Do you mean you set timestep as 86400 s? That is insane!

You can not guess the timestep. You need to either to a timestep sensitivity study, or use adaptive timestepping algorithm to let FLUENT find the right timestep. There are no shortcuts in transient simulations, unless of course you don't want correct results

OJ

[roo]

If i set lower time step (60, 100, 1000 or 3600 s) or adaptive timestepping the results is same. Immediately, solution diverges.
I used to explain the time step 86400s since the problem is solved in day mode after a period of 151 days.

[delaneyluke]

To give you some perspective, for a simple unsteady flow analysis ( no energy), the timesteps used are never more than 0.1-1 s. In your case it would have to be even smaller.

Regards
Luke