Dynamic Mesh and Transient convergence Issues

Malookiyat · September 10, 2018, 16:11

I am simulating an expeller device for design purposes and it is aerodynamically coupled to a gas turbine exhaust. I modeled it like a cross between the ball-check valve tutorial and the fan tutorials available on youtube using a subdomain, defining a rigid body (the rotor) and using dynamic meshing to assess the effects. I am encountering a few problems:

1. I have to keep the time interval really low (order of milliseconds) to avoid negative volumes during the run. This means that a 1 second run would take at least 250 iterations. Is there something else I can improve to save time?

2. As it is aerodynamically coupled, the RPM rises with time and that seems to be correlated with a divergence in results, I increase the max iterations during my run but the divergence isnt controllable. Will introducing an RPM or time based resistance to the rotation help? as in modeling a bearing of sorts.

Please let me know what more information I can provide.

ghorrocks · September 10, 2018, 19:38

1) 250 iterations does not sound like a lot to me. CFD is slow and computer intensive, so expect long run times.

2) Maybe, that would depend on exactly what the problem is and how you have modelled it.

The standard way of modelling devices like this where you do not know the device speed is to do a range of fixed speeds (that is, no rigid body motion) and plot the torque versus speed graph. The point where the net torque on the device equals zero is the steady state speed. Doing a range of known speeds like this is always far easier than letting the device find its own speed like you are doing.

Malookiyat · September 14, 2018, 16:56

Lets just say I am trying to model something similar to a cylinder spinning on its long single axis inside a high speed tunnel/cylinder in a concentric orientation. I do not want to create another cylinder around the ball and use the rotating fluid domain as I am interested in the immediate boundary layer and I cant seem to find a way to rotate the solid ball and have it actually interact with the surround fluid as I dont get to define interfaces as walls.

Malookiyat · September 15, 2018, 10:36

I managed to use a rotating wall to get this done since I had a symmetric body. Now I need to model condensation of wet air at high temperature and pressure in the fluid domain. Any help?

ghorrocks · September 16, 2018, 19:53

There are a few examples on condensation in the CFX tutorials, but they are on surfaces. This is a very specialised (and complex) topic and I think your only chance would be to contact ANSYS support to see if they have some examples available.

Malookiyat · September 18, 2018, 07:49

I have tried using the Physics modelled in the Steam Jet example but I am unable to produce any liquid water and the solver keeps failing to solve the enthalpy change between the two states (Im assuming thats because there is no liquid in the domain).

To further explain the problem:
At the inlet, I have 9/1 ratio of Air Ideal Gas and Water Vapor at 150 m/s and 650K. This flow is put through a crude compressor and the outer shell is cooled to 220K. I have outlet nozzles placed on the outer shell at the maximum Diameter point where the maximum separation due to centrifugal action occurs. I have to modify the rotor to achieve the saturation pressure for water vapor and see if additional cooling is required. The outlet on the other side is open to atmosphere as well.

I wish I could share design but I am not allowed.

Malookiyat · September 18, 2018, 09:17

I have also create an annular subdomain as a Mass sink where condensation is most likely as opposed to modelling the entirety of it as a sink.

September 10, 2018, 16:11	Dynamic Mesh and Transient convergence Issues	#1
Malookiyat New Member Saif Ul Malook Join Date: Sep 2018 Posts: 8 Rep Power: 8	I am simulating an expeller device for design purposes and it is aerodynamically coupled to a gas turbine exhaust. I modeled it like a cross between the ball-check valve tutorial and the fan tutorials available on youtube using a subdomain, defining a rigid body (the rotor) and using dynamic meshing to assess the effects. I am encountering a few problems: 1. I have to keep the time interval really low (order of milliseconds) to avoid negative volumes during the run. This means that a 1 second run would take at least 250 iterations. Is there something else I can improve to save time? 2. As it is aerodynamically coupled, the RPM rises with time and that seems to be correlated with a divergence in results, I increase the max iterations during my run but the divergence isnt controllable. Will introducing an RPM or time based resistance to the rotation help? as in modeling a bearing of sorts. Please let me know what more information I can provide.

September 10, 2018, 19:38		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	1) 250 iterations does not sound like a lot to me. CFD is slow and computer intensive, so expect long run times. 2) Maybe, that would depend on exactly what the problem is and how you have modelled it. The standard way of modelling devices like this where you do not know the device speed is to do a range of fixed speeds (that is, no rigid body motion) and plot the torque versus speed graph. The point where the net torque on the device equals zero is the steady state speed. Doing a range of known speeds like this is always far easier than letting the device find its own speed like you are doing. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

September 16, 2018, 19:53		#5
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	There are a few examples on condensation in the CFX tutorials, but they are on surfaces. This is a very specialised (and complex) topic and I think your only chance would be to contact ANSYS support to see if they have some examples available. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Convergence	Centurion2011	FLUENT	48	June 15, 2022 00:29
sliding mesh problem in CFX	Saima	CFX	46	September 11, 2021 08:38
Dynamic Mesh Problem.	Tom Clark	FLUENT	10	June 21, 2021 05:27
wrong transient solution in different time steps of Dynamic mesh	esicia	FLUENT	7	February 26, 2012 14:27
early stall, poor convergence, and mesh quality	everest	CFX	2	May 12, 2010 17:27

September 14, 2018, 16:56		#3
Malookiyat New Member Saif Ul Malook Join Date: Sep 2018 Posts: 8 Rep Power: 8	Lets just say I am trying to model something similar to a cylinder spinning on its long single axis inside a high speed tunnel/cylinder in a concentric orientation. I do not want to create another cylinder around the ball and use the rotating fluid domain as I am interested in the immediate boundary layer and I cant seem to find a way to rotate the solid ball and have it actually interact with the surround fluid as I dont get to define interfaces as walls.

September 15, 2018, 10:36		#4
Malookiyat New Member Saif Ul Malook Join Date: Sep 2018 Posts: 8 Rep Power: 8	I managed to use a rotating wall to get this done since I had a symmetric body. Now I need to model condensation of wet air at high temperature and pressure in the fluid domain. Any help?

September 18, 2018, 07:49		#6
Malookiyat New Member Saif Ul Malook Join Date: Sep 2018 Posts: 8 Rep Power: 8	I have tried using the Physics modelled in the Steam Jet example but I am unable to produce any liquid water and the solver keeps failing to solve the enthalpy change between the two states (Im assuming thats because there is no liquid in the domain). To further explain the problem: At the inlet, I have 9/1 ratio of Air Ideal Gas and Water Vapor at 150 m/s and 650K. This flow is put through a crude compressor and the outer shell is cooled to 220K. I have outlet nozzles placed on the outer shell at the maximum Diameter point where the maximum separation due to centrifugal action occurs. I have to modify the rotor to achieve the saturation pressure for water vapor and see if additional cooling is required. The outlet on the other side is open to atmosphere as well. I wish I could share design but I am not allowed.

September 18, 2018, 09:17		#7
Malookiyat New Member Saif Ul Malook Join Date: Sep 2018 Posts: 8 Rep Power: 8	I have also create an annular subdomain as a Mass sink where condensation is most likely as opposed to modelling the entirety of it as a sink.