Error Message - the Solver has exited Unexpectedly HELP!!!

[JMatteau]

Error Message—Analysis has stopped because the Solver has exited Unexpectedly

I desperately need help on this topic. My job revolve around using Simulation CFD and right now I can't do my job because of this. I’m using Simulation CFD 2013 SP1. I am simulating a large model and I keep getting this error message.  The problem with the message is that it happens randomly trough out the solving.

For the same Design study :

First it happens during the meshing.

Then I solve again and it happens at iteration 20/100

Then I solve again and it happens at iteration 80/100

Then I solve again and it happens at iteration 95/100

Then I solve again and it happens during the meshing phase

…

A quick fix that I’m using right now is to solve 10 iterations at the time. If I try to go with 20 iterations it won’t work. There is no logic to it since I can do 2X10 iterations without any problem but I can’t do 20 iterations. I don't want to spend all my time starting the solver if I want to do 300 iterations.

Does anyone have a clue on how to fix the problem? I’m out of solution and I have no hint on how to fix this random problem.

I can't attach my .cfz file because of the size 8.82Mb. But I would gladly send it by Email to someone from the support team.

Best regards,

Jonathan Matteau

Ingénieur jr. Mécanique / Jr engineer mechanical

[Jon.Wilde]

Dropbox link on its way!

[Jon.Wilde]

I have taken a look at this and suggest the exact same as the other case as it happens, you need to extend the inlets and outlets. Ideally we need to avoid recirculation over them.

Are you running this from a local drive also, it does need to be. I can confirm that it runs OK here, although with those extensions the results would likely be more reliable.

[JMatteau]

Hi Jon,

It run ok for a certain time but will eventually stop working (I got up to 233 itterations yesterday). I added two flag manager that seem to be helping the stability of the solving. If you look at the other scenario Nozzle Box No CSUNK, this is the less stable one. The problem with this  Design study is that it is the preliminary phase of a bigger project. This project is 20 time bigger that this one. So if I'm not able to get results for this Design study, there no chance for me to get the bigger project to converge. So it is very important that this design study is Bullet proof and this is not the case. From the last two weeks that I worked on it, I was never able to do a complete solve up to convergence without having the Error Message - the Solver has exited Unexpectedly.

***********************************************

About the local drive, yes I am working on my local drive.

I thought that setting the Fully developped Volume flow rate option on the inlet BC would do the same job as extending the inlet length ?

Another concern that I had was the thin solid volume. Should I make my geometry out of surfaces instead of thin volume?

*****************************************************************

I notice something interesting yesterday: We had a previous extremely complex contract (months of work) on CFD 2011 and I tried to run it on CFD 2013. Obviously if it was running fine on 2011 it should be able to solve without any problem on CFD 2013. THe model could not pass the meshing phase without getting the Solver has exited unexpectedly error mesage. In the log there was an error number:

[14:54:14]  output:C:\Program Files\Autodesk\Simulation CFD 2013>set E=-1073741819

[14:54:14] output:C:\Program Files\Autodesk\Simulation CFD 2013>IF -1073741819 NEQ 0 goto EOF

[14:54:14]  output:C:\Program Files\Autodesk\Simulation CFD 2013>exit -1073741819

[14:54:14] Finished with exit code = -1073741819

The only difference between 2011 and 2013 is the flag manager (and internal stuff that I'm not aware). With that last test I did, it seem obvious that something is happening in CFD 2013. THere seem to be sensitive parameters that need to be adress. I wouldn't blame the meshing because it was fine on CFD 2011. Right now I'm trying to identify which flag could help with the solving. Is there a complete description of the flag list because the help list is not complete ?

Best regards,

Jonathan Matteau

Ingénieur junior mécanique / Engineer junior mechanical

[Jon.Wilde]

Hi Jonathan,

We should be able to run this no problem. You are correct about the fulyl developed inlets, although it can still help to have a good inlet length, to ensure we have a good boundary layer as the flow exits around the corner.

The outlets do need extending though, this is almost certainly what the issue is, do you have flow entering and exiting over the boundary?

Your second issue is different, what flags are you using and why? I cannot see any setup in this project.

There are some differences between the 2 versions, the meshing is also improved in 2013 and will give a slightly different mesh.

[JMatteau]

Hi Jon,

We should be able to run this no problem. You are correct about the fulyl developed inlets, although it can still help to have a good inlet length, to ensure we have a good boundary layer as the flow exits around the corner.

That is my opinion also, but right now I don't see the light out of the tunnel...

I do need to make it work, that is what I'm getting paid for.

The outlets do need extending though, this is almost certainly what the issue is, do you have flow entering and exiting over the boundary?

No the outlets are the 2 BC set to 0 KPa gage pressure, and no flow are around it.

Also what do you think about switching the solid volume to surface?

Your second issue is different, what flags are you using and why? I cannot see any setup in this project.

There are some differences between the 2 versions, the meshing is also improved in 2013 and will give a slightly different mesh.

DRSmoothing -2

mesh_boundarylayer_blend 0

[Jon.Wilde]

We should be able to run this no problem. You are correct about the fulyl developed inlets, although it can still help to have a good inlet length, to ensure we have a good boundary layer as the flow exits around the corner.

That is my opinion also, but right now I don't see the light out of the tunnel...

I do need to make it work, that is what I'm getting paid for.

See below, make soem geometry changes and then we will have a model that we can work with.

The outlets do need extending though, this is almost certainly what the issue is, do you have flow entering and exiting over the boundary?

No the outlets are the 2 BC set to 0 KPa gage pressure, and no flow are around it.

What I mean is, when you run it, does flow both enter and exit the surfaces - it may not in reality but in simulation this might be a cause of the issue - try the extensions please.

Also what do you think about switching the solid volume to surface?

Sure, you can do. It might save some mesh when you move to a larger model, although the fluid mesh around the solids may not be quite as accurate.

Your second issue is different, what flags are you using and why? I cannot see any setup in this project.

There are some differences between the 2 versions, the meshing is also improved in 2013 and will give a slightly different mesh.

DRSmoothing -2 - This is used when we have recirculation near resistance parts, I would delete this.

mesh_boundarylayer_blend 0 - This is used to go back to the older meshing style, personally I would run without it for now and only use it if you cannot mesh the model.

[JMatteau]

Trying your solution right now.

I'll be back with update later today

Thank you sir for the support

[JMatteau]

Hi Jon,

About the outlet,

The only BC set on the outlet is Pgage = 0 psi. Is there a way to prevent recirculation ? Do I need to insert another BC to prevent this behavior ? I don't think there is recirculation in my outlets but I was wondering if there was a way to force it ?

BEst regards

[Jon.Wilde]

Hi,

The best way is simply to extend the geometry, just to ensure that we have nice smooth/laminar flow exiting

[JMatteau]

For the moment the simulation is running smoothly, I'm keeping my fingers crossed

[JMatteau]

HI Jon,

It just crashed again. It reached iteration 67 and then I got the error message. I've putted my cfz file in the drop box and I also added the log file in it. Shoul I increase the lenght of th I/O?

Best regards,

*************************************************************

[13:07:18]  output:job aborted:
[13:07:18]  output:[ranks] message
[13:07:18]  output:[0] terminated
[13:07:18]  output:[1] process exited without calling finalize
[13:07:18]  output:[2-4] terminated
[13:07:18]  output:---- error analysis -----
[13:07:18]  output:[1] on wks-a063
[13:07:18]  output:C:\Program Files\Autodesk\Simulation CFD 2013\SimCFDSolver.exe ended prematurely and may have crashed. exit code 0xc0000005
[13:07:18]  output:---- error analysis -----
[13:07:18]  output:C:\Program Files\Autodesk\Simulation CFD 2013>set E=-3
[13:07:18]  output:C:\Program Files\Autodesk\Simulation CFD 2013>exit -3
[13:07:18] Finished with exit code = -3
[13:07:21] Processing solver exited

*************************************************************

[Jon.Wilde]

Please could you share the summary (sum) and status (st) files?

If you have not made the recommended changes of extending the inlet and outlets then yes, please do so.

[JMatteau]

Hi Jon,

Yes I did extended the inlet and outlet. I was wondering if I should extend them even more.

Length is approximately 5"

There is no recirculation in the outlet.

As for the sum and st files, I restarted the solving at iteration 60 and they were erase by the neww solving. I soved correctly from iteration 60 to iteration 100. I will retart the solving from iteration 100 to iteration 400. More than likely there will an error and I will add to the drop box the sum and st files. The CFZ(with the extend I/O) file is still in the dropbox.

Best regards,

Jonathan Matteau

[JMatteau]

Hi Jon,

As predicted the sover has exited unexpectedly.

It achieved loose convergence at iteration 128. Then I restarded the solving with a tighter convergence. I converged at iteration 168 but exited with the error message : the Solver has exited Unexpectedly. The good news is that I got all my results anyway.

[08:35:17]  output:Flat lines detected in Convergence Monitor - Auto Stop.
[08:36:08]  output:job aborted:
[08:36:08]  output:[ranks] message
[08:36:08]  output:[0-3] terminated
[08:36:08]  output:[4] process exited without calling finalize
[08:36:08]  output:---- error analysis -----
[08:36:08]  output:[4] on wks-a063
[08:36:08]  output:C:\Program Files\Autodesk\Simulation CFD 2013\SimCFDSolver.exe ended prematurely and may have crashed. exit code 255
[08:36:08]  output:---- error analysis -----
[08:36:08]  output:C:\Program Files\Autodesk\Simulation CFD 2013>set E=-3
[08:36:08]  output:C:\Program Files\Autodesk\Simulation CFD 2013>exit -3
[08:36:08] Finished with exit code = -3
[08:36:11] Processing solver exited

Summary since the Inlet / Outlet extension

Error message at iteration 67

Restarted solving

Loose convergence at iteration 128

Restarted solving

Normal Convergence at iteration 168

Exited with the error message

I can confirm that extending the I/O made the model a lot more stable and converged way more easily. The error message is still there but seems to appear less offen. I will now make the same changes on my other geometry that was less stable. I will make sure that when it crash I will put the sum and st file in the dropbox

BEst regards,

[JMatteau]

Update concerning Error Message - the Solver has exited Unexpectedly during the Meshing phase.

If you keep getting this message during the meshing phase I found a quick way to fix it. For large model, it is often hard to define all the meshing manually. If you reach the point where you don't know what to do to solve this issue, just go in the flag manager and set the flag

-> mesh_boundarylayer_blend : 1

It worked for me and I hope it can help other users.

Best regards,

Jonathan Matteau

[JMatteau]

Hi Jon,

I am still getting some error messages: the Solver has exited Unexpectedly. It usually happens before solving. I don't know what is causing the crash because if I restart the solving it will usually solve succesfully. I've putted my files in the dropbox:

CFD_Hole_B4_support.cfz

mesh

mesher

Scenario 1.st

Scenario 1.sum

SimCFDJobManager

Simulation CFD

solver

Any ideas on what causing this ?

Best regards

Jonathan Matteau

Ingénieur junior mécanique / Engineer junior mechanical

[Jon.Wilde]

Are you running from a local drive yes? If it fails before the solve this is probably the issue.

Either that or you are running out of memory.

Have you considered running a 1/2 or even 1/4 model to speed things up here?

I have it running fine but I did tweak the mesh. Autosized and then turned on Gap Refinement (3 Gap Elements set in the Advanced window).

Press the Refine button and then Solve!

Seems to be working great, please try that.

[JMatteau]

Hi Jon,

Yes I am on the local drive. Like I was saying it will work and in fact I got it to converge this morning without any changes. When I got the error message all I had to do is to restart the solving and it worked. My question is why would it crash the first time and not the second time ? What is causing that random crash ? Why when I press solve there is a chance of a possible crash when in fact the model work fine?

[Jon.Wilde]

OK, I suggest you make those changes to help with speed and accuracy anyway.

The crashes could be write access/admin issues. Have you tried right clicking and launching CFD as admin, I wonder if you still get the error then?