9 Replies Latest reply on Jun 13, 2018 5:40 AM by Nikhil Phatak

    FEA Best Practice for Stress Analysis

    Samantha Shillinglaw

      So,

      I told my boss I wasn't happy the simulation method we were using was accurate and he directed me to CadTek to get their advice. Their advice supported my concerns but apparently they have previously directed us to conduct FEA in the way I think is inaccurate. I have been "volunteered" by my boss to research the different FEA practices to find a suitable method.

       

      I have an SW FEA book I can refer to but I am looking for some pointers of where I can find examples to support my research as my boss will expect me to explain every minute thing to him to prove what works.

       

      The basics of our FEA was a skid, simulating a drop test, load test (lifted by lugs on base frame), and 4 x simulations at 45 degrees either side of the front and right planes. The method specified by my boss was to multiply gravity by the factor of safety, (so a factor of safety of 2 would set the gravity to 9.81 x2).

      I challenge this method because we have load tested one skid to 4 Tonnes in real life and it passed, yet for this FEA method it fails with a load of less than 0.6 Tonnes! Given the multiplication factor of 3 that would be less than 2 Tonnes.

       

      The method I was advised to use is have all the loads and gravity as accurate and use the Factor of Safety plot to set a multiplication factor.

      This method produced results that correlated with our real life load test. For the other method I have had to bulk up the skid to a ridiculous amount and it still has not passed. I am convinced the Factor of Safety Plot is the correct way to do this, but I need irrefutable evidence for my boss.

       

      Thanks

        • Re: FEA Best Practice for Stress Analysis
          Bill McEachern

          Assuming that a linear static analysis was done both methods should produce the same result. What analysis procedure was used? What failure criteria was used? Onset of yield = FOS =1? If this is true then passing a test does not mean a linear static analysis would match as much residual strength can remain post onset of localized yielding. Some clarification could be helpful in sorting it out.

          • Re: FEA Best Practice for Stress Analysis
            Matt Peneguy

            I'm not an expert on this, and your description seems pretty thorough (good job on that), but if you could post your file/s, I'm sure you will receive much better responses.

            • Re: FEA Best Practice for Stress Analysis
              Kevin Chandler

              Hello,

               

              For what they're worth, there are two simulation PDF files included with SW in:

              C:\Program Files\SOLIDWORKS Corp\SOLIDWORKS\lang\English

               

              "introduction_to_simulation.pdf" & "simulation_theory_manual.pdf" (attached).

               

              Like I typed, FWIW.

               

              Cheers,

               

              Kevin

              • Re: FEA Best Practice for Stress Analysis
                Dan Golthing

                Samantha, it sounds like you have an uphill battle. 

                 

                SolidWorks doesn't make it too easy to take a peek behind the curtain, as it were.

                 

                What are all these settings?  How do they affect the results?  Etc.

                 

                Apparently, few people ask, because a few years back when I was trying to find out, it took an act of congress to get some in-depth info out of them.

                 

                Are you all able to do some hand calculations to verify the simulation results?  I think this is critical to help expose the weaknesses of the software. 

                 

                For instance, I found out awhile ago that their buckling results only works for Euler columns.  Shorter columns produce wildly invalid buckling results.

                 

                Are you running through a mesh-refinement process to help test the quality of the output?

                 

                Can you replicate your problem with simpler structures?

                 

                We can see how intellectually challenging it is to properly use this software, and then SolidWorks goes and markets it to the unwashed masses.

                • Re: FEA Best Practice for Stress Analysis
                  Roland Schwarz

                  Any good analysis and testing regiment combines and correlates predictive analysis with actual testing. Use both to refine one another--testing to refine analysis methods, and analysis to refine testing. testing can teach you where your analysis falls short, and analysis can show you much about what your testing might fail to reveal.

                   

                  You already have one good solid case in your favor. Can't argue with evidence. Make more.

                   

                  What is stopping you from running both methods and seeing which is more accurate?

                  • Re: FEA Best Practice for Stress Analysis
                    Samantha Shillinglaw

                    Hi all,

                     

                    Thanks for your helpful suggestions.

                     

                    It is not worth risking my job to post the files, however I can do a sketch with the sim details to better illustrate.

                     

                    Nothing is stopping us running both methods to see which is more accurate, this has already been done to a degree, and why I raised the issue in the first place. The testing was done by our manufacturer and is only for a load test so there is not a lot to go on. We don't have the facilities or the time to do the kind of tests we need to compare to the other 3 simulations.

                     

                    Looking at hand calculations to verify the results is something I will research. I think really if they were that simple to do, that would be a preferable method instead of wasting weeks doing FEA. Yes we are using mesh refinement and I don't know how I might simplify the design for FEA. It's pretty simple already to me. The loading is what's more complicated. It's a design on a ship so we are trying to replicate any forces the model might be subjected to in a storm for example.

                     

                    Thanks for the PDF's I will print those off to read and see if I can get a better understanding.

                     

                    I am assuming it was a linear analysis, I have no idea, just did what I was told and didn't think it was correct. My boss forwarded info from CADTEK that also suggests that both analyses should produce the result, provided large displacement was turned off (it was not). I suggested to my boss my next area to look into is whether we should be performing a linear analysis in the first place and whether we need to use FEA at all.

                      • Re: FEA Best Practice for Stress Analysis
                        Nikhil Phatak

                        Hi Friends,

                        Personally I feel that Solidwork's FEA is best used as a tool or as a part time seldom used program and suitable to make quick decisions about the design when it is being created or during its initial stage or when ever we are creating something totally new out of the box.

                        That time whatever excellent tools which will save considerable amount of time during inception stage of that design project for  decision making e.g. we have like auto meshing, quickly adding boundary conditions and visualizing results and displacements and specifically presenting those results backed up by simple hand calculations will be enough for its worth.

                         

                        I have used Solidworks & simulation in different organisations for different roles from past 5 years and I found that it needs huge in depth knowledge of Math's. FEA is tremendously huge and I frankly can not grasp all of it.

                        I am not a mathematician who can afford to spend all of my time solving equations and matrices and to worry about what models to use for what kind of materials and which displacements are valid and why they are not behaving as they should .

                        I am a Mechanical Engineer

                         

                        If somehow I could determine exact method of solving a system using FEA or equations which will closely represent a real time problems I ill be called  Dr. Nikhil :-)

                         

                        So I think FEA is a guessing tool and programs like Solidworks or any FEA will help us guess right faster than others.

                        Best method is to test your systems and trouble shooting them before you produce it

                        And frankly math's is boring, I would prefer building a machine and test it on the floor instead wasting my time behind FEA

                        at least i will learn interesting new things about my machine which are real.

                         

                        P.S. - This is my personal opinion about using FEA, after all i just love solid-works and simulation and its technology

                        as it saves lots and lots of my time

                         

                        Regards,

                        Nikhil

                      • Re: FEA Best Practice for Stress Analysis
                        Samantha Shillinglaw

                        Well 2 minutes time with the INTRO TO SIM PDF has me convinced that we are well of track. We should be doing a dynamic analysis not a static, as we are looking at impact forces (sort of) and waves are oscillatory in nature, two of the contexts the document says must be performed as a dynamic study.

                         

                        I guess I better find out if we can calculate the results before doing too much research into dynamic analysis (assuming it doesn't cost extra).