In 2014, I could run a case and change a boundry conditon (i.e. inlet temperature) and solve from the end of the previous solution. This would result in an answer in a few interations. However in 2016, if I do the same thing it appears the prevous BC is used when restarting the solve from the end. To get updated results, I have to start the solver from interation 0. Is this a new feature in 2016 or do I have something set-up incorrectly?
Thanks,
Gary
Solved! Go to Solution.
Solved by cummingsga. Go to Solution.
Solved by Jon.Wilde. Go to Solution.
Hi Gary,
Is this working within the same scenario? What BC's are you changing and from what values?
Thanks,
Jon
Yes the same scenario. I completed a run and saw the temperature of the outlet water was low. I changed the inlet water temperature and re-ran from the end of the previous run. The results of this run yielded the same outlet temperature as the initial run and checking the inlet temperature it too was for the previous run. The Design Study Bar displayed the updated inlet temperature for the latter run. This is in CFD2016. When I would do this same thing in CFD2014, the results for the second run would be different from the initial run indicating the change to the temperature BC was used. In order to get correct results in CFD2016, I had to restart the analysis at step 0.
Thanks,
Gary
Are you running 2016 SP2? It works OK for me here.
Sounds OK to me, maybe this is down to something else in the set up. Are you able to share a CFZ?
Unfortunately I cannot share without a non-disclosure agreement and a non-public forum, but some background. This is a 3D model created in Inventor 2014. After setting it up I ran the solver once with the default auto-mesh. Then I ran again with adaptation set for 3 cycles. I asked it to save the intervals (BWT it did not save the iterations). The adaptation process ended prematurely when the solution diverged. I ran again without adaption and the model solved. This is the state of the model I am working with. There could be some form of corruption in the file. When I ran it after the solution diverged; a message was displayed in the output bar-message window something to the effect of "applying results to the adaptive mesh2" which seems strange since the adaptation is not selected for that run.
These messages are currently displayed when I run the model from the last iteration.
Control file processing complete
Check Out advanced License: Successful
Authorization verified
Processing input ...
Loading restart mesh input
Mesh input complete
Processing results input ...
Results input complete
Processing BC input ...
Loading restart BC data ...
BC restart input complete
Analysis Initialization Complete
Input processing complete
** FINITE ELEMENT SUMMARY FOLLOWS...
4793379 Total Nodes, 3111063 Fluid Nodes , 1682316 Solid Nodes
22278604 Total Elements, 12331482 Fluid Elements , 9947122 Solid Elements
7 Inlets 1 Outlets 0 Unknowns
*********** ANALYSIS STARTED ***********
Radiation surface factor calculation ...
Memory Allocated
Radiation Problem Set Up Complete
Radiation surface factor calculation complete
Turbulent Incompressible Flow is ON
Turbulence model: Standard k-epsilon
Intelligent Wall Formulation: ON
Thanks,
Gary
Update: I cloned the scenario and added a new BC (film coefficient) and started the solver. In the message window received "Interpolating results onto a new mesh 2". A few minutes later the process stopped The Status File is attached after it failed. I simply restarted the solver and it is currently running.
The memory issue I can understand, 22m elements is a lot for most machines to handle.
The re-running issue is a little strange though - I suspect it might be down to the version you are running. Is it possible for you to upgrade at all?
What software do you want me to upgrade? CFD2016 is the latest I have access to but I will check with the CAD Administrator. Memory is not a problem, the machine has 128GB of installed RAM.
Thanks,
Yes, try CFD 2016 SP2. Or I can test it here if you can give me some exact steps to follow 🙂
Here's a new idea: what happens if you turn off Intelligent Wall Formulation?
I turned off the Intelligent wall with no joy, see attached screen capture.
I simply restarted the solver after the previous message and it ran. However the mesh was reset (very course) so I restarted the mesh adaption. Now it is saving the data from the iterations of the mesh as requested. The solver is still running as I type this, but it looks like things are improving. I will test changing the BC's once the adaption process completes.
I installed the SP2 .msi file (again) and it completed but the version and build numbers did not change. See attached,
That says 16.2 so the upgrade when smoothly.
Please let us know how you get on, thanks!
Turning off the adaptive wall appeared to work, I can change the BC and restart from the end of the previous solution. I do not recall if I was using the adaptive wall in CFD2014 or not but that does not matter.
Some other questions:
Thanks for all the help
Gary
Hi,
This sounds loads better!
If your machine rebooted, it is likely that your results were lost. So much can be held in limbo.
Saying that, check out this article, as it might work here 🙂
To stop showing velocity: I guess it's still a global result (under the right click menu) or on a cut-plane or something else. You can also just show one legend per side:
The Summary Points should still work regardless 🙂
Did you right click to update them?
Thanks!
Jon
I will try the methods in the article to recover the data.
If you look at the picture I uploaded in my previous post, you will see the Legend count is set to 1 but two legends are displayed. I cannot get the Velocity Legend to go away. I even unselected the legend display and reselected it and both legends returned.
Yes I updated the points but I get a result of 0 unless the point is on a node. If I go to Results - Points and select a point that has a zero value in the summary table and plot it, a error message that the point is not on the geometry is displayed.
Thanks for the help,
Gary
Regarding the legends - this shouldn't happen. Can you at least get it down to one if you stop showing velocity on a cut plane, ISO surface or globally? I think it must be set somewhere. Strange that it still has two legends though, does it stay like that even if you change the 1 to a 2 and back again?
The Monitor Point issue is strange also! Are those points definitely within the model and not outside of it for some designs?
The only way I can get one legend is to select the Velocity Magnitude again, the temperature legend goes away. If I select temperature again, both are displayed. It acts as if the VM legend is a hard default and the number of legends is being ignored. If I select a different ledgend, the temperature one is replaced and the VM one remains.
When I placed the monitor points I used the "Align to Surface" feature to make sure it was on the surface. Given that, I would thing it would be on the geometry. In stead of putting it on the node, should I have placed the point between nodes but not on the mesh lines? Say close to the center of the facet? That way if the node moved due to meshing changes, worst case the point would be slightly under the "surface" of the geometry.
BTW, no luck with recovering the data via the first two methods. For this study it is not mission critical, I was asking for future reference.
Thanks,
Gary