AnsweredAssumed Answered

a sharing violation occured while accessing an unnamed file- and other simulation issues

Question asked by John Rhodes on Jul 25, 2014
Latest reply on Aug 4, 2014 by Jared Conway

Like • Show 0 Likes0
Comment • 8

I'm a previous ANSYS user (from a while ago) and I'm going to be using solidworks simulation in my next job.

I'm really disappointed in the quality of the simulation software. I've got installed on three different computers (multiple times on each computer to make sure it's installed right), and I've had nothing but issues with simulation. This is a total opposite of the the modeling environment, which is works very nicely.

Everything was going well, when suddenly, my simulation wouldn't run. After meshing, it gave the error "Midside nodes on two edges of an element are merged, element: (reducing mesh tolerance or specifying a higher no. of points for Jacobian check for meshing may help)".

So, I messed around with the mesh settings. No luck. I removed most of model to see if I could still run any simulation, when I got the error message "A Sharing violation occured while accessing an unnamed file". Then, it stopped solving, but the solver window didn't close, so I couldn't do anything else. When I closed SW, the solver window stayed open, but I couldn't close it without going to task manager and ending it.

Right there, there were three bugs, the error message that is only meaningful to the SW programming team, the fact that solving can end without closing the window which leaves solidworks in a state that it needs to be restarted, the solving window does not close when SW closes, and the solving window can enter a state where it cannot be closed without going to task manager.

My questions are: why do the two errors occur, how do I fix them, and should I just use ANSYS 2007 instead? I'm doing very basic simulation, but I have no desire to waste hours of my life fighting crap software.

Jerry Steiger Jul 25, 2014 5:38 PM

Correct Answer

John,

I am an ANSYS user who has much less experience with SolidWorks Simulation. The few times that I or my colleagues have used SolidWorks Simulation it has worked well. These have usually been relatively straightforward; when we have more difficult analyses to run it usually falls on me to do it and I do it in ANSYS. That is partially because I am more familiar with ANSYS and partially because I am a bit leary of using SW Sim on the tough stuff. The other guys use SW Sim because it's easy to use, unlike ANSYS. Since you are used to lrunning ANSYS, this probably isn't much of an issue, although I always struggle after not using it for a few months.

From following this forum, it seems to me that SW Sim IS more buggy than ANSYS. On the other hand, a lot of people are getting acceptable results and you say that your simulations are pretty basic, so I wouldn't expect you to have many problems.

In a couple of cases we have run either the same or a very similar analyses in SW Sim and ANSYS and SW Sim was substantially faster. None of the analyses took more than a few minutes to run, so the faster solution time didn't really matter to us.

Meshing does seem to be one of SW Sim's weaknesses. We originally chose ANSYS over the competition, which included COSMOS, the predecessor of SW Sim, because the fellow who did the evaluation thought that ANSYS had the best mesher. (On the other hand, he was most familiar with ANSYS, having used it in school.) The other guys have run into a few cases where they had difficulty meshing parts, but they always managed to work their way around it.

I would worry about what operating system you will need to run such old code.

Jerry S.

See the reply in context

No one else had this question

Outcomes

Jerry Steiger Jul 25, 2014 5:38 PM

John,

I would worry about what operating system you will need to run such old code.

Jerry S.

Actions

John Rhodes @ Jerry Steiger on Jul 26, 2014 1:23 AM

Thanks. I still have the old ANSYS machine, so we'll just be using that.

I've had some better luck using very low quality meshes in SW, and I bet I could figure it out if I had enough time, but I know ANSYS very well, and as you said, the mesher is much superior.

Actions

Jared Conway @ Jerry Steiger on Jul 26, 2014 6:09 PM

I haven't used ansys so I can't speak from experience.

But regarding the mesher, I don't know if I'd say it is a weakness. It can mesh almost anything, it just doesn't have some options tha people want like adding noding a model manually or other elements. But is that important if it has a good GUI, meshes and solves quickly and gives accurate results?

Regarding buggyness, inherently when you make something more user friendly you also expose it to use by a wider variety of people with a wide skillset. I think a lot of buggyness you see is from people using it with little experience. This makes it look buggy when in fact it is generally user related. With ansys you generally have trained analysts, they have one way to do it, they do it the same way, very easy to test and document that which I think sometimes makes it look less buggy. But again, generally buggy ness is ui related, results are still accurate.

And regarding running harder problems in ansys. If solidworks simulation had the same physics represented, I would think they are comparable. If you need other physics or a specific solver, sure , take it to another tool. For example super nonlinear with contact, I might take to abaqus

Actions

Jared Conway Jul 26, 2014 6:02 PM

Use simulation every day basically and support many customers with simulation and have seen those errors probably 10 times in 10 years. Both well documented in the swx kb. Both generally related to feeding bad geometry into the mesher. Sometimes a bad results location like Dropbox or something. Any chance you are attacking this without getting some guidance from your car on best practices? I'd guess the same ones from ansys would apply so I'd be curious how you'd imposed and checked your geometry before starting the analysis. I'd guess it isn't that the mesher in ansys is better, I'd guess that from your experience you inherently plan around its limitations based on your experience.

Actions

Alessandro Tornincasa Jul 27, 2014 12:32 PM

Midside nodes on two edges of an element are merged, element: (reducing mesh tolerance or specifying a higher no. of points for Jacobian check for meshing may help)".

I've seen this error occuring only with a couple of customers:

- in one case one customer wasn't aware he was meshing his assembly with incompatibile mesh. Mesh was failing all the time, therefore he activated an option in the mesh called "incompatibile mesh for failed parts". Mesh was bad quality and the solver also was having a hard time using constraint equations to bond parts

- in another case customer had an assembly with thin parts but they didn't have uniform thickness, therefore he couldn't use shell elements. His mesh had very bad quality with very high aspect ratio elements. Reducing mesh size in order to have elements with aspect ratio no more than 10 solved the issue.

Generally speaking ansys mesher is much more powerful than solidworks simulation. The other Dassault Systèmes tool that can compare to that capability is SIMULIA ABAQUS.

SolidWorks Simulation on the other hand is easy to use, and has an unmatched level of integration, and a much lower cost than high end tools.

Unless you're trying to mesh a very complicated assembly (like a car engine, I'm sure you'll be able to carry out your analysis.

By the way, I've seen solidworks simulation meshing a car engine with the curvature based mesher. All you need is enough Ram to account for the small elements generated by the small curvature regions.

Alex

Actions

Jared Conway @ Alessandro Tornincasa on Jul 27, 2014 11:23 PM

Do you have some examples that show the Ansys meshers strength over simulation?

Actions

Alessandro Tornincasa @ Jared Conway on Aug 3, 2014 5:56 PM

One example is sweep mesh. Very useful on slender parts that can't be treated as beams or sheels, it generates fewer elements than tetrahedrons.

http://www.computationalfluiddynamics.com.au/tips-and-tricks-sweep-meshing-in-ansys/

By the way the two tools are not comparable: there are certain taks where the job can be done easily with solidworks simulation, and other tasks with very demaning analyses where a high end tool is needed like ABAQUS or ANSYS, Usually those cases are like 1 out of 100.

Actions

Jared Conway @ Alessandro Tornincasa on Aug 4, 2014 11:44 PM

"generates fewer elements" > I could agree on that, but probably similar accuracy. solve time may be somewhat affected but significantly enough to make the leap from a tool that is easy and reasonable to use?

1/100 cases seems to support what I said about simulation relative to abaqus/ansys.

Actions

8 Replies

Jerry Steiger Jul 25, 2014 5:38 PM
Unmark Correct
Correct Answer
John,

I am an ANSYS user who has much less experience with SolidWorks Simulation. The few times that I or my colleagues have used SolidWorks Simulation it has worked well. These have usually been relatively straightforward; when we have more difficult analyses to run it usually falls on me to do it and I do it in ANSYS. That is partially because I am more familiar with ANSYS and partially because I am a bit leary of using SW Sim on the tough stuff. The other guys use SW Sim because it's easy to use, unlike ANSYS. Since you are used to lrunning ANSYS, this probably isn't much of an issue, although I always struggle after not using it for a few months.

From following this forum, it seems to me that SW Sim IS more buggy than ANSYS. On the other hand, a lot of people are getting acceptable results and you say that your simulations are pretty basic, so I wouldn't expect you to have many problems.

In a couple of cases we have run either the same or a very similar analyses in SW Sim and ANSYS and SW Sim was substantially faster. None of the analyses took more than a few minutes to run, so the faster solution time didn't really matter to us.

Meshing does seem to be one of SW Sim's weaknesses. We originally chose ANSYS over the competition, which included COSMOS, the predecessor of SW Sim, because the fellow who did the evaluation thought that ANSYS had the best mesher. (On the other hand, he was most familiar with ANSYS, having used it in school.) The other guys have run into a few cases where they had difficulty meshing parts, but they always managed to work their way around it.

I would worry about what operating system you will need to run such old code.

Jerry S.
Like • Show 1 Like1
Actions
- John Rhodes @ Jerry Steiger on Jul 26, 2014 1:23 AM
  Mark Correct
  Correct Answer
  Thanks. I still have the old ANSYS machine, so we'll just be using that.
  
  I've had some better luck using very low quality meshes in SW, and I bet I could figure it out if I had enough time, but I know ANSYS very well, and as you said, the mesher is much superior.
  Like • Show 0 Likes0
  Actions
- Jared Conway @ Jerry Steiger on Jul 26, 2014 6:09 PM
  Mark Correct
  Correct Answer
  I haven't used ansys so I can't speak from experience.
  
  But regarding the mesher, I don't know if I'd say it is a weakness. It can mesh almost anything, it just doesn't have some options tha people want like adding noding a model manually or other elements. But is that important if it has a good GUI, meshes and solves quickly and gives accurate results?
  
  Regarding buggyness, inherently when you make something more user friendly you also expose it to use by a wider variety of people with a wide skillset. I think a lot of buggyness you see is from people using it with little experience. This makes it look buggy when in fact it is generally user related. With ansys you generally have trained analysts, they have one way to do it, they do it the same way, very easy to test and document that which I think sometimes makes it look less buggy. But again, generally buggy ness is ui related, results are still accurate.
  
  And regarding running harder problems in ansys. If solidworks simulation had the same physics represented, I would think they are comparable. If you need other physics or a specific solver, sure , take it to another tool. For example super nonlinear with contact, I might take to abaqus
  Like • Show 0 Likes0
  Actions
Jared Conway Jul 26, 2014 6:02 PM
Mark Correct
Correct Answer
Use simulation every day basically and support many customers with simulation and have seen those errors probably 10 times in 10 years. Both well documented in the swx kb. Both generally related to feeding bad geometry into the mesher. Sometimes a bad results location like Dropbox or something. Any chance you are attacking this without getting some guidance from your car on best practices? I'd guess the same ones from ansys would apply so I'd be curious how you'd imposed and checked your geometry before starting the analysis. I'd guess it isn't that the mesher in ansys is better, I'd guess that from your experience you inherently plan around its limitations based on your experience.
Like • Show 0 Likes0
Actions
Alessandro Tornincasa Jul 27, 2014 12:32 PM
Mark Correct
Correct Answer
Midside nodes on two edges of an element are merged, element: (reducing mesh tolerance or specifying a higher no. of points for Jacobian check for meshing may help)".

I've seen this error occuring only with a couple of customers:
- in one case one customer wasn't aware he was meshing his assembly with incompatibile mesh. Mesh was failing all the time, therefore he activated an option in the mesh called "incompatibile mesh for failed parts". Mesh was bad quality and the solver also was having a hard time using constraint equations to bond parts
- in another case customer had an assembly with thin parts but they didn't have uniform thickness, therefore he couldn't use shell elements. His mesh had very bad quality with very high aspect ratio elements. Reducing mesh size in order to have elements with aspect ratio no more than 10 solved the issue.

Generally speaking ansys mesher is much more powerful than solidworks simulation. The other Dassault Systèmes tool that can compare to that capability is SIMULIA ABAQUS.
SolidWorks Simulation on the other hand is easy to use, and has an unmatched level of integration, and a much lower cost than high end tools.

Unless you're trying to mesh a very complicated assembly (like a car engine, I'm sure you'll be able to carry out your analysis.

By the way, I've seen solidworks simulation meshing a car engine with the curvature based mesher. All you need is enough Ram to account for the small elements generated by the small curvature regions.

Alex
Like • Show 1 Like1
Actions
- Jared Conway @ Alessandro Tornincasa on Jul 27, 2014 11:23 PM
  Mark Correct
  Correct Answer
  Do you have some examples that show the Ansys meshers strength over simulation?
  Like • Show 0 Likes0
  Actions
  - Alessandro Tornincasa @ Jared Conway on Aug 3, 2014 5:56 PM
    Mark Correct
    Correct Answer
    One example is sweep mesh. Very useful on slender parts that can't be treated as beams or sheels, it generates fewer elements than tetrahedrons.
    http://www.computationalfluiddynamics.com.au/tips-and-tricks-sweep-meshing-in-ansys/
    
    By the way the two tools are not comparable: there are certain taks where the job can be done easily with solidworks simulation, and other tasks with very demaning analyses where a high end tool is needed like ABAQUS or ANSYS, Usually those cases are like 1 out of 100.
    Like • Show 0 Likes0
    Actions
    - Jared Conway @ Alessandro Tornincasa on Aug 4, 2014 11:44 PM
      Mark Correct
      Correct Answer
      "generates fewer elements" > I could agree on that, but probably similar accuracy. solve time may be somewhat affected but significantly enough to make the leap from a tool that is easy and reasonable to use?
      
      1/100 cases seems to support what I said about simulation relative to abaqus/ansys.
      Like • Show 0 Likes0
      Actions

a sharing violation occured while accessing an unnamed file- and other simulation issues

Outcomes

Related Content

Recommended Content