[Paul95]

Hey guys,


I'm currently simulating airflow in a simple test room. The room has a velocity inlet near the ceiling, a pressure outlet on the same wall near the floor, and a human figure in the centre (represented as a cylinder with a CO2 velocity inlet). Opposite the inlet and outlet, there's a wall section with a constant temperature, and another wall section that acts as a heated surface with a heat flux.
Now onto the key physics:

• Steady
• 3D
• Incompressible ideal gas
• Multi-component gas (air and CO2)
• Non-reacting
• Radiation (DOM)
• Turbulent - RANS (K-Epsilon)
• Realizable K-Epsilon Two-layer
• Two-layer All y+ wall treatment
• Segregated flow
• Separated Temperature
• Segregated species

Because of the multi-component gas, I've chosen the incompressible ideal gas model, which doesn't require an equation at the pressure outlet (as per the Star CCM+ guide).


Here's my problem:
The simulation gives good results above a certain velocity at the velocity inlet near the ceiling. The residuals converge nicely and I am monitoring the velocity and temperature at various points in the room over iterations. I see convergence to constant values. However, when I reduce the velocity at the ceiling inlet by 50%, my solution doesn't converge and the velocity and temperature values at all observed points fluctuate significantly. With Gravity on, the velocity at the observed point fluctuates irregularly and with Gravity off, the values fluctuate regularly and the residuals do not converge either.


Does anyone have any experience with this problem? Could it be related to the reduced flow rates at the inlet resulting in less forced airflow? I've tried solving this problem with both the Coupled Solver and the Boussinesq model (without CO2) at lower inlet velocities, but the same problems persist.


Thanks for your help in advance

[cwl]

Originally Posted by Paul95

Because of the multi-component gas, I've chosen the incompressible ideal gas model, which doesn't require an equation at the pressure outlet (as per the Star CCM+ guide).

What exactly do you mean by that? - I can't recall any requirements like that

Because except for that reason - I'd say that your model is over-complicated.

As for the question itself:

Originally Posted by Paul95

However, when I reduce the velocity at the ceiling inlet by 50%, my solution doesn't converge and the velocity and temperature values at all observed points fluctuate significantly.

Maybe within that range of reduced velocities it just has unsteady unstable (in terms of flow pattern) nature?

[Paul95]

Thank you for your reply and sorry for my late response.

Originally Posted by cwl

What exactly do you mean by that? - I can't recall any requirements like that

If gravity and e.g. the ideal gas model are used, a pressure function should be added for the pressure outlet. I have already read a lot about this here in the forum.

Originally Posted by cwl

Because except for that reason - I'd say that your model is over-complicated.

I am a friend of simple models, but in this case I don't know what I could simplify. Do you have a spontaneous idea?

Originally Posted by cwl

Maybe within that range of reduced velocities it just has unsteady unstable (in terms of flow pattern) nature?

I have also thought about this, but I have little to no experience with unsteady CFD simulation. I tried simulating the model unsteadily, but the results were not useful.

Could I solve the problem by simulating it unsteadily, or will the solution not converge because the flow pattern is unsteady and unstable?

[cwl]

Originally Posted by Paul95

If gravity and e.g. the ideal gas model are used, a pressure function should be added for the pressure outlet. I have already read a lot about this here in the forum.

Sorry, I've missed the "incompressible" in the original post.

Originally Posted by Paul95

I am a friend of simple models, but in this case I don't know what I could simplify. Do you have a spontaneous idea?

If it's airflow in the facility - I'd expect the concentration of non-air gases to be low and to barely affect the solution; temperature variation (like min vs max) should not be large.
Thus it looks like it can be simplified to single-gas with small gas components as passive scalars and Boussinesq approximation instead of ideal gas should be sufficient.

Originally Posted by Paul95

I tried simulating the model unsteadily, but the results were not useful.

What exactly do you mean by "not useful"?

Originally Posted by Paul95

Could I solve the problem by simulating it unsteadily, or will the solution not converge because the flow pattern is unsteady and unstable?

That's the thing - if the flow is strongly unsteady you'll see it in the unsteady results, if it reaches some steady flow pattern .. you'll also see it %)