Learning Forum

[DeHoodedOne]

Hi, My name is Louis Oguike, a Masters student in Mechanical Engineering at University of Cape Town, I am working on investigating the performance of 3 stage cascade thermal energy storage that employs micro encapsulated spheres containing PCM KOH, KNO3 and NaNO2 sequentially from the inlet during charging and discharging phases. I have some issues with simulating the melting of a PCMs. It's in a domain where there is a hot incoming HTF at 400C. The PCMs are encapsulated in copper spheres. I hybrid initialized the simulation and patched the PCM temperatures at 27C. The simulation proceeds as expected up until the liquidus temperature of the PCMs is reached the the temperature values oscillate to as high as 3000C. All the walls are coupled and shell conductions were used on spherical encapsulations. The liquid fraction of PCM does not form in all the spheres for reason I know not. I have attached images of the settings I used the proceed with the simulation. Can you render some assistance, please? Also how can constrain the maximum possible temperature to 400 degree and prevent overshoots

[Rob]

If you have a look for Solver Limits you'll find a way to cap the temperature: use with caution and read up on what it actually does.
Re the solver stability, I assume the PCM material is constant density?

[DeHoodedOne]

The PCM density varies and I accounted for that

[Rob]

If the PCM is encapsulated and changes density where does the extra volume go (or come from)?

[DeHoodedOne]

I succeeded in getting the simulation to work but consequently when I try to run the simulation, ANSYS when I hybrid initialize my model, there will be a message saying: 'Node 999999: Process 15040: Received signal SIGSEGV', and fluent will close by itself automatically.
the error log contains
Node 999999 Fatal signal raised sig = Segmentation fault
29900320 CX_Primitive_Error
59540cc0 seh_filter_exe
299a41b0 logical_right_shift
3937eb20 _C_specific_handler
5b612060 _chkstk
5b5c1020 RtlRaiseException
5b610c80 KiUserExceptionDispatcher
283ec5b0 Fill_Capillary_Pressure

And the same settings were used in the successful simulations yet I can't figure out why they now bring errors. When I try to access Ansys after the error I get this prompt:
exception of type 'ansys.fluent.cortex.cortexnotavailableexception' was thrown

Kindly assist

[Rob]

If you use hybrid do you patch anything into the domain? You may need to be more careful with the initial conditions.

[DeHoodedOne]

Yes I do. I patch the initial temperature conditions of everything in the domain except the heat transfer fluid

[Rob]

How good is the mesh?

[DeHoodedOne]

Good Day Rob. I have been able to sort all my prior issues concerning the domain and numerical errors. I have been carrying out CFD post analysis on some models which have gone smoothly without a hitch but I've begun to encounter problems I can explain with other similar models. While performing CFD post analysis, I have been trying to develop contour plots for some of my models and as soon as I click "apply", I get nothing, the program halts as though it is generating the contour but never completes. Why I am baffled is because I have done this with 6 similar model and there were no issues, the fluent simulation generated no error either, yet I can't get past this contour generation phase which is crucial to my work.

[DeHoodedOne]

Good Day Rob. Please is there any update on this ?

[Rob]

I didn't see the post, I suspect it's from when I was on holiday.
Can you plot the contour in Fluent? You've got a fixed range set so check that's suitable too.

[DeHoodedOne]

I haven't tried plotting contours on Fluent. I've used the range for similar models and I was able to generate contours on CFD post without incident. I will try plotting on Fluent directly to see if it works

[DeHoodedOne]

Good Day Rob. Apologies for getting back to you so late, i have been trying out some tweaks to my settings to check if that might be the issue but to no avail. From the Image below, The contours can be generated via FLUENT but even after allowing for the contour generation process to run for 14hours on CFD post, it still didn't generate. Is there any possible reason why this is happening ? I would I employed using this FLUENT contours but it doesn't allow for a lot of operations that are avail on CFD post. Kindly advise

[Rob]

CFD Post may be trying to read in all the data sets: how many are there? It's also less effective with the newer HDF format so that may also be a factor. Other than the movies I've not needed CFD Post for several releases, and even then I tend to create the animations via Fluent during the simulation.

[DeHoodedOne]

I'm not quite clear on which data sets you're referring. My primary challenge is I have successfully used CFD Post to handle similar models like this one without a hitch. I can't select time steps in fluent as well so that poses a major challenge to me.

[Rob]

Did you load a single case & data into CFD Post or the project file that's generated during a Fluent transient run? It's flpj or similar.

[DeHoodedOne]

The project file that's generated during a fluent transient run

[DeHoodedOne]

.wbpj

[DeHoodedOne]

Hello Rob. Any updates?

[Rob]

Yes, my garden looks much better for a long weekend.
wbpj is the Workbench project, not anything to do with Fluent. Not sure why Post is struggling. If you export the case & data can you open the files in CFD Post outside of Workbench?