11 Replies Latest reply on Mar 4, 2014 11:54 AM by Jared Conway

    Part failure detection

    David Dearing

      Hello guys, I've been using simulation for a while, but in a very limited way.  I design a part, assign it a material, apply the load, check the results against the material properties to make sure it hasn't failed, and move on.  My client wants to know how much force his part can take before it fails.  Is there any way to do that OTHER than just increasing the force over and over until the resultant stress is greater than the material's property allows?  In other words, is there a way that Solidworks will automatically determine the force required to make the part fail? 

       

      Thanks.

        • Re: Part failure detection
          Shaun Densberger

          You could make a Factor of Saftey plot (the Factor of Safety is the Induced Stress divided by the Allowable Stress), and then multiply your applied load by the lowest Factor of Safety in your model. For example, if you have a part made out of A-36 (yield of 250 MPa) with an applied load of 100 N and your peak stress is 125 MPa, then you lowest Factor of Safety is 2 and your maximum load before failure (assuming failure is defined by plastic deformation) is 200 N (100 N * 2).

           

          Now, given that you've asked this question I need to give you a warning. Before you go and do this, you (and your client) need to understand that you need to be extremely careful with how you use this "failure load" value. Do not take the results from a simulation as the definitive failure load. Errors in how you set up the simulation (loads, constraints, etc.), numerical errors in the finite element method, and errors associated with assumptions made in the model (linear, homogenous material, etc.) means that the failure load could be higher or lower by some unknown amount (assuming you have converged results; if you don't, then all bets are off).

          • Re: Part failure detection
            J. Mather

            David Dearing wrote:

            ...resultant stress is greater than the material's property allows?....

             

            Thanks.

            This is not a definition of material failure.

            Safety factor is not a definition of material failure.

             

            All you are finding is where the part goes beyond elastic deformation into plastic deformation.  The actual area of elastic deformation might be insignificant.  Far more information is needed.  Can you attach the file here?

             

            What is your definition of "failure"?  Fracture is a different topic.

              • Re: Part failure detection
                David Dearing

                Thank you for the replies.

                 

                My client is building a carbon fiber reinforced wing for a road racing team in California.  It will see speeds up to 220mph, and produce a significant force onto the car, as much as 800 pounds or more in some situations.  We want to make sure that the wing's structure can handle the load, will not break (CF is strong but brittle, and will pretty much shatter when it fails), and that the struts will remain relatively ridged with high drag and downforce.  I'm mostly dealing with three forces: lift (downforce), drag, and surface pressure.  All parts are being made of 12K carbon fiber. 

                 

                I would post the part file, but our company just purchased SW2014, and most people can't open the file yet whenever I've posted it on this forum.  They took SW2011 off my work station, unfortunately.

                 

                What approach should I take to ensure that the wing won't fail?  Thanks again.

                  • Re: Part failure detection
                    Bill McEachern

                    model the wing strut & wing structure with surfaces. Use laminated shell elements, specify your lay ups, loads and the rest.

                    To mitigate the britle fracture check out this: http://www.innegratech.com/. They have been evaluated by some F1 team to mitigate brittle fracture of F1 wings so stuff doesn't end up all over ht place when and accident occurs. The white water sports guys love it as well.

                    • Re: Part failure detection
                      Shaun Densberger

                      I'm not racing expert, but if the wing were to catastrophically fail at high speeds, I would think that this could cause the driver to lose control of the vehicle (especially if it's during a turn). If this is the case, and injury/death could result from a catastrophic failure of the wing, then I strongly advise against using your simulation as the determining factor.

                       

                      Computer simulations should really be used to study the behavior of a design and identify trends. Using it to predict failure points is possible (in a limited sense), but it requires a detailed analysis with a robust code and a competent engineer that understands the physics at play and the inner workings of numerical simulations. If there are serious consequences if your simulation results don't match the physical system, and the simulation software is a "magic black-box" to the user, then we have a recipe for disaster. As the old adage goes, "FEA makes a good engineer great, and a bad engineer dangerous."

                        • Re: Part failure detection
                          Bill McEachern

                          Safety is all ways priority #1. However, this should not discourage you from giving it a whirl, just realize what you don't know - which could be quite a bit. Just make sure you have a test program that ensures the article has reasonable strength compared to what is required and this is verified by an adequate test program. Then you can get a feel for how your analysis did. At a lift coefficient of (CL=) 1 your wing is about 1'x 6'. You will have drag loads as well. If the test costs a lot to do it will be worth your while to pay some one to check your work or have your customer pay to get another analysis done. If the cost of a failed test is not a big deal then see how you do. Composites are not the simplist things to figure out if you don't have some knowledge of them. You might need to take a course.

                          FEA should only be used in the presence of a quality system that ensure hazardously mis-leading information or hazardous hardware is not released.

                          They break airliner wings in order to certify them. The test costs what $150 million? Yeah you don't want to bet the farm on ones guys say so - no matter who that guy is. You need a system that ensures a good estimate has been made and things are as well understood as they can practically be. If it beaks between 150 and 155 things are good, below is obviosuly not good and above 155 your a bum as the bird is to heavy.

                            • Re: Part failure detection
                              Shaun Densberger

                              "However, this should not discourage you from giving it a whirl, just realize what you don't know - which could be quite a bit."

                               

                              I should have be more clear on this point. I'm by no means trying to discourage David from doing a simulation (or learning how to do a simulation), but instead trying to make sure he's on the same page as everyone else when I come to using the results of a simulation in the real world. I have this knee-jerk reaction when it comes to simulation results and using them in the real world, because all too offten I see people try to take the results from a simulation as definiative proof of something and try to use them in a rather unsettling ways. I can't tell you the number of times I've heard people say, "you did an FEA study and it showed that the design is strong enough, so we'll forgo a load test."

                                • Re: Part failure detection
                                  David Dearing

                                  Thanks guys.  I very much appreciate it.

                                   

                                  Perhaps I wasn't clear, as I am known to not be.  I know as a fact that the design is sound.  It's the exact same structure and skin design we use for top fuel dragsters that routinely exceed 300 MPH at maximum AOA.  It could probably handle 500 mph, as at 300mph the wing barely flexes at all and has no flutter- this is real life, not a computer simulation.  But my client wants some cool graphics and a chart showing a full computerized FEA with failure predctions, etc. and  I just am having trouble creating such things.  He is in China, and he wants computer charts.  My experience with simulation is very limited.  I'm not worried in the slightest that the wing will fail- but he is.  So I have to give him something.  And no, I can't just break one and say "see, here you go- 500mph or 2000 pounds" old school.  He wants predictive data from the computer even though it is a tried and true design.  I completely understand that it isn't terribly usable.  I don't care.  I know the wing works.  It's been around for a very long time.  I just need to do a simulation showing what the computer predicts the failure point to be.  (yes, I'm frustrated)

                                   

                                  Thanks again for all your input.  Some of it is over my head, but I'm trying to learn. 

                                    • Re: Part failure detection
                                      Jared Conway

                                      for your original question, you could use a design scenario to run through all the possible iterations until you reach whatever you end up defining as "failure".

                                       

                                      for your actual application, your approach to the problem and your client's request seem to be out of sync. he's asking you to show the max design load, you're saying that there is no max design load.

                                       

                                      my suggestion would be to find a known data point, get your simulation setup and get them to match. this is going to require having some data like displacements strains..etc because right now all you know is "it doesn't break" but you don't know to what level it was stressed which is what you need to compare in simulation.

                                       

                                      then you can iterate towards the "max design load".

                                        • Re: Part failure detection
                                          David Dearing

                                          EDIT:

                                           

                                          Okay, here's what I did.  I decided instead of finding the "failure" forces, which as you all stated are hard to quantify on a real world object, I followed the follwoing steps.

                                           

                                          1. The race car we are installing this on will never see over 200 mph, so 200 X 1.65 = 330mph

                                          2. Put the wing at most severe angle of attack it will ever be in and started a flow simulation with the airspeed of 330 mph.

                                          3. Saved the results and exported them to Simulation.  Used the pressure and shear forces as the forces for that simulation.

                                           

                                           

                                          Now, I have the simulation successful.  Can I compare the stresses to the component properties values to make sure that the stress numbers don't exceed them? 

                                            • Re: Part failure detection
                                              Jared Conway

                                              if the materials have failure criterias applied to them, you can look at FOS as shaun mentioned

                                               

                                              or you could just evaluate the stress reported by the simulation to the allowable stress

                                               

                                              if you haven't reached it yet, since you have a linear analysis, you can linearly interpolate load to failure