News: Contact us to upgrade your software!

Author Topic: Inconsistency in Load Case Results  (Read 27242 times)

Mike T.

  • Client
  • **
  • Posts: 5
    •  
Inconsistency in Load Case Results
« on: July 30, 2014, 03:02:01 PM »
I'm working with an FEA project, about 200 load cases that I imported via Excel, and about 50 components that are a mix of honeycomb sandwich and 2-stack unstiffened.  When I run detailed sizing on the project using all load cases, 2 components are returned with negative margins.  Looking at the Excel stress report, I see about 5 or 6 load cases listed with negative margins.  If I select only these load cases (un-checking all others on the project setup form) and re-run the sizing, I now see positive margins for these same load cases.  Also, if I de-select this same set of load cases that originally produced negative margins and re-run including all others, load cases that were originally listed with positive margin are now negative.  I wouldn't expect that the results for a given load case would be dependant on which other load cases are also selected, but that seems to be what I'm seeing.  Any suggestions or help would be appreciated.

Thanks,
-Mike

Mike T.

  • Client
  • **
  • Posts: 5
    •  
Re: Inconsistency in Load Case Results
« Reply #1 on: August 01, 2014, 10:01:22 AM »
To add more detail here, I was using the "Element Peak" load extraction method in the scenario I described above.  When I switched it to "Element Based", this problem went away, and so far I am receiveing the same results for each load case regardless of how many other load cases are activated.

I would still be interested in hearing a way to fix this issue in Element Peak mode if possible, because it would save a huge amount of processing time for me (with all ~200 load cases active, a few minutes per component turns into 8+ hours per component).

Thanks,
-Mike

Ryan

  • Administrator
  • *****
  • Posts: 145
    •  
Re: Inconsistency in Load Case Results
« Reply #2 on: August 07, 2014, 08:47:47 AM »
There was a bug with how the sizing code was processing the peak loads. There was a memory allocation error which was causing incorrect loads to be used for situations of many load cases. This bug was fixed in an internal version several weeks ago.

As a workaround until the update is available, you can try the new Load Case Filtering feature. In V7, HS is sampling which load cases are most damaging per component based on how low the margins are relative to each other. This data shows up in the "Controlling Load Cases" tab of the Excel stress report. This data can be used to filter out these load cases on the next sizing. Go to the backdoor data form (right-click the project in the tree) and enable “Apply filtered load cases if available when sizing”.

Thanks,
Ryan

Mike T.

  • Client
  • **
  • Posts: 5
    •  
Re: Inconsistency in Load Case Results
« Reply #3 on: August 07, 2014, 11:56:35 AM »
Ryan,

Thank you, I will use this workaround until the next version of Hypersizer is available.  To clarify, the Load Case Filtering feature is a way to make the optimization run faster when using Element Based loads (where the bug does not occur), and does not prevent incorrect results when using Element Peak loads.  Is that correct?

-Mike

Ryan

  • Administrator
  • *****
  • Posts: 145
    •  
Re: Inconsistency in Load Case Results
« Reply #4 on: August 08, 2014, 08:15:23 AM »
Load case filtering works with both the Element Based and N-Sigma loading methods.

1) The first step is to size or analyze the component as you normally would using N-Sigma or Element Based loads. During this initial run, load case sensitivity results are tracked based on the margins. You can view this data in the Excel stress report.

2) Next enable "Apply filtered load cases..." option in the Backdoor form. The next time you size or analyze the component only the dominant load cases determined from step 1 are considered.

-Ryan