[Alisa_W]

Hello, I am simulating the unsteady flow in a rotating duct. The model is simple, just a duct rotating around a axis (shown in Fig 1).

The software is fluent. Before unsteady simulation, I simulated the steady flow field (SRF model). However, when I transferred the "frame motion" to "mesh motion"(Fig 2) and began to calculate, the error "detected in AMG solver: x momentum ...." and "Floating point error" occurred. Meanwhile,the flow field also had unexpected high speed zone. It is strange because the steady process simulated very well and the unsteady process also went smooth for several time steps.

I guess that the problem may be caused by the set of simulation( mesh quality is up to 0.85).I am not familar with unsteady simulation of single rotating zone. Can anyone tell me if I made some mistakes when set the mesh motion?

1.I didn't tick the "dynamic mesh" and cannot Preveiw Mesh Motion in Run Calculation term.
2.If the time step size has effect on it? I set it as 0.00125s and the min size of my mesh size is e-4.
3.The turbulence model is SST-DDES.

I appreciate any help or suggestion!^_^

[Alisa_W]

other case set details

[RaiderDoctor]

I think you may have set your axis of rotation wrong...
What I understand about your setup, you have a rectangular duct: (first figure)


And then you are applying a rotation to it. Problem is, I think you want to be rotating it along the x-axis rather than the z-axis (what you have right now). The reason for this is the current rotation looks something like this: (second figure)

Which is a little difficult to visualize, and even more so if there's a device that does something like this. Back to your error message, it's saying that the momentum in the x-direction cannot reach convergence. Could this be due to the fact that, because of the centripetal force acting on the fluid, more fluid is exiting the domain than entering? If we just look at Reynold's Transport Theorem, does the input match the output?

[Alisa_W]

Thank you for your quick reply! Actually, I have set the right axis and the duct rotated around an axis outside the duct domain. Maybe you can image it as one of the channels of a impeller.
Another strange thing is that the steady simulation is very smooth. So maybe the reason for the wrong messege is not that water cannot enter the channel.

Originally Posted by RaiderDoctor

I think you may have set your axis of rotation wrong...
What I understand about your setup, you have a rectangular duct: (first figure)


And then you are applying a rotation to it. Problem is, I think you want to be rotating it along the x-axis rather than the z-axis (what you have right now). The reason for this is the current rotation looks something like this: (second figure)

Which is a little difficult to visualize, and even more so if there's a device that does something like this. Back to your error message, it's saying that the momentum in the x-direction cannot reach convergence. Could this be due to the fact that, because of the centripetal force acting on the fluid, more fluid is exiting the domain than entering? If we just look at Reynold's Transport Theorem, does the input match the output?

[RaiderDoctor]

Okay, cool. The impeller domain makes more sense to me. I'm afraid I don't know what you mean by "smooth" for the steady state simulation. The bit about water not being able to enter the domain is a little confusing to me as well: you set the inlet as a velocity inlet with a specified magnitude. So, therefore, fluid is entering the domain.



As a quick test, set your inlet and outlet to pressure inlet and outlet, respectively, and try to run a transient case. This will determine if all other setup parameters are okay.

[Alisa_W]

"Smooth" means it runs well and the velocity profile is right. Your suggestion (change inlet and outlet to pressure inlet and outlet) is a little confusing to me. My experience is that inlet & outlet should be velocity dependent one and pressure dependent another. Did I get wrong?

Originally Posted by RaiderDoctor

Okay, cool. The impeller domain makes more sense to me. I'm afraid I don't know what you mean by "smooth" for the steady state simulation. The bit about water not being able to enter the domain is a little confusing to me as well: you set the inlet as a velocity inlet with a specified magnitude. So, therefore, fluid is entering the domain.



As a quick test, set your inlet and outlet to pressure inlet and outlet, respectively, and try to run a transient case. This will determine if all other setup parameters are okay.

[RaiderDoctor]

No, I believe so long as there is a pressure gradient leading for the inlet to the outlet, the program should calculate correctly. In fact, I believe that even if there wasn't, it would still work due to the phenomenon of flow reversal on the inlets and outlets.

At this point, could you post a few more pictures about your steady-state results and overall setup?

[Alisa_W]

I am sorry that I has been a long time. I found that when I using coarse mesh the unsteady simulation is fine. So maybe I had set wrong time step for my case.(DDES model requires fine mesh and the corresponding time step is very small).
Thank you so much !!

Originally Posted by RaiderDoctor

No, I believe so long as there is a pressure gradient leading for the inlet to the outlet, the program should calculate correctly. In fact, I believe that even if there wasn't, it would still work due to the phenomenon of flow reversal on the inlets and outlets.

At this point, could you post a few more pictures about your steady-state results and overall setup?