Learning Forum

[RozinaNalbandian]

Hello all,

I am running various temperature cases of some Fluent simulations. All residuals are set to 4e-4, except for energy which is set to 1e-6. I'm using a realizable k-epsilon model with enhanced wall treatment with under-relaxation. The method I am using for a "faster" simulation is to obtain the steady state conditions for velocity and pressure by only having the Flow and Turbulence equations ON under "Controls" (Energy is Off), then to switch to transient with Flow and Turbulence equations OFF and only Energy equation on. This allows Ansys to save the steady state results obtained and only calculate the unsteady results of temperature. For some reason, I am not able to get my continuity residuals low enough for convergence in steady state. I have experimented with trying all different Schemes overnight as well as trying different initialization methods/solvers (momentum vs density-based) to no avail. I have over 2000+ iterations and the trend of the residuals for continuty seems to stay constant and have stopped dropping after a certain point. Strangely enough, I tested transient with a timestep size of 0.0001 seconds, with ALL three equations on and it converges okay, it's just very time consuming.

I am currently testing out a long transient run to see if I made a weird mistake with mesh interfaces but so far they look fine and I have heat transfer occuring through different bodies at reasonable rates. I increased my time step size to 0.01 seconds and each timestep converges within 100 iterations.

 

My current ideas are that

A) maybe my mesh isn't fine enough? But does being able to reach convergence in transient imply the mesh is fine enough, or is the solver "stricter" for steady state with energy equation off?

B) I am running cases where the material properties of the fluid (air) change, would slightly altering the properties of air cause steady state to not be obtainable?

C) My system may be inherritently unsteady which is why steady state convergence for continuity is so hard to obtain. 

 

I am running Ansys 2021 R2. Sorry if I used any incorrect terminology, I'm still learning. 

Any help would be appreciated and I can provide more information or screenshots if needed.

 

-Rozina Nalbandian

 

 

Screenshots:

Current Mesh 3.4 million elements:

 

 

[Rob]

Have a detailed look at the flow field, and post some contours. Residuals not hitting a limit can mean you have a problem, but can also mean there's a transient in the flow. However, looking at your model, how does flow pass from inlet to outlet? Doesn't the steel plate get in the way?

[RozinaNalbandian]

 

Hello I solved my issue. It turned out I had to add some inflation layers to some surfaces in my mesh and rename some named selections as well. But after running steady state, switching to transient (haven’t run anything yet) setting report definitions, and archiving a file as a backup, it looks like I’m out of space already on my OneDrive. Is there a way to simplify my file so it only has the final steady state data and settings on it? My archive file is huge at 2.37 GB (I guess due to my geometry and mesh being included).

 

 

Regarding the steel: The metal only goes part of the way through the model like baffles so the fluid snakes around the inner tube.

 

-Rozina Nalbandian

 

[Rob]

If you're in Workbench, then it's a little more complicated. In Autosave there's a "retain x previous cases" option. If you run in standalone you can overwrite/delete spare files as required. I do recommend saving all data locally as the solvers assume the disc/storage can write as quickly as data is dumped to them, if they can't you can lose packets and finish up with a corrupt file. That's less of an issue at present as USBs are much improved from 10 years ago but I expect it to more common again as the file write software improves faster than hardware.