10 Replies Latest reply on Jul 8, 2014 2:49 PM by Shaun Densberger

    BUCKLING OF PISTON RODS

    Raghav S

      Hello,

      I have to simulate buckling in case of a hydraulic piston rod. One end of the rod is fixed whereas the other end of the rod i.e the piston head is guided inside the cylinder tube. The force is applied axially on the piston.I tried using the cylindrical faces fixture under advanced fixtures but i am unable to understand what the values of x,y,z in that mean physically. My object can move only in the axial direction and changing the value of 'z' in the cylindrical face fixture gives very different results. What does the value of 'z' exactly signify here? I have tried various other types of fixtures on the piston head but none of the results are convincing. Also with a decrease in load my displacements increase which makes no sense at all. Can anyone please suggest what changes i can make to obtain a better result? Please find attached a screenshot of my simulation.

        • Re: BUCKLING OF PISTON RODS
          Shaun Densberger

          First things first. What do you want to do with these results? Are you trying to determine the load at which buckling will occur? If so, then keep in mind that the values calculated from a "Buckling Analysis" are non-conservative; i.e. the buckling value calculated by the software will be higher than the real buckling value (by an unknown amount). More realistic values can be obtained with a nonlinear analysis.

           

          Now, regarding your questions on constraints, it'd help if you could post the model itself. The mode(s) of buckling within a structure are directly related to how the structure is constrainted, so it's critical that you accurately represent the constraints in simulation. Does a "fixed" constraint on one end of the rod accurately represent the real system? Trying out different constraints can be a good way to understand what each does, but it's better if you do it on a very basic model that you'll understand all of the outcomes.

           

          Typically, the x,y, and z values corresponded to enforced displacements with respect to the world coordinate system. For SolidWorks, if the box is left blank, then that degree of freedom is free (i.e. it's allowed to move) and any numeric value is the enforced displacement.

            • Re: BUCKLING OF PISTON RODS
              Raghav S

              Thanks a lot Shaun. My aim is to change the angle of cylinder,material and calculate the Load Factor/Critical load for buckling. I am facing problems mainly with applying fixtures to the piston head. It would be of great help if you could suggest some alternate fixture for the same. Also please find attached the model and a few other results that i got when i changed the fixtures on the piston head.Sliding contact.png Piston Rod Force-fixture.png

                • Re: BUCKLING OF PISTON RODS
                  Shaun Densberger

                  My aim is to change the angle of cylinder...

                   

                  Are you saying that your material isn't isotropic?

                   

                  I am facing problems mainly with applying fixtures to the piston head. It would be of great help if you could suggest some alternate fixture for the same.

                   

                  You're issue is that you're not applying the constraint properly. The constrain on the piston head in the first picture is preventing axial motion while allowing radial and theta. Use the "Advanced Fixtures" and set the radial displacement to 0 while leaving the theta and axial free; this will give an idealized representation of a piston head in a cylinder.

                   

                  Radial Constraint.png

              • Re: BUCKLING OF PISTON RODS
                Jared Conway

                None of the results are convincing > what is your test case that you are comparing against?

                 

                Note that displacements will basically be meaningless in buckling. The output is the buckling load factor.

                 

                And Shaun is 100% on the restraints. I might suggest going through the static tutorials and running a static analysis on this first before moving onto buckling. We just did a big buckling project for a customer and it turned out great, so I know the software can do it. Just needs the right setup.

                  • Re: BUCKLING OF PISTON RODS
                    Raghav S

                    Thanks a lot Jared.

                    The piston rods have been designed (stroke lengths) theoretically with a FOS of 3.5 against buckling (based on the Euler's formula).

                    I also have a few questions regarding buckling

                    1. Are buckling load factor and FOS the same thing?

                    2.Is it true that buckling happens under non equilibrium conditions whereas the displacements in the software are calculated assuming equilibrium condition?

                    3. Do all load values above the critical load lead to buckling or does it follow a phenomenon similar to resonance wherein buckling occurs at various multiple of the critical load?

                      • Re: BUCKLING OF PISTON RODS
                        Jerry Steiger

                        Raghav,

                         

                        The displacement shown in a linear buckling analysis is just showing you the initial shape that the buckling would take, how the buckling would start. As Jared said the displacements themselves are meaningless.

                         

                        Any load above the buckling load will cause buckling. It is not at all like resonance. As Shaun said, buckling can occur at loads below the calculated value.

                         

                        Jerry S.

                        • Re: BUCKLING OF PISTON RODS
                          Shaun Densberger

                          As both Jared and Jerry have said, the displacement values in a (linear) buckling analysis are meaningless; it's just like the displacement values from a modal (frequency) analysis. This is because both a linear modal and linear buckling analysis involve solving an eigenvalue problem.

                           

                          The eigenvalues are the modal frequencies of the structure in a modal analysis and the buckling load factors in a linear buckling analysis. The eigenvectors are the mode shapes of the structure in a modal analysis and the buckling mode shapes in a linear buckling analysis. If you've ever solved an eigenvalue problem by hand, then you'll know that you typically normalize your eigenvector (i.e. you define the largest value to be 1), hence the "meaningless" displacement values.

                           

                          Just as a modal analysis results in multiple modes of vibration, a buckling analysis will result in multiple modes of buckling. It's recommended that you ask the solver to calculate more that just the first mode of buckling (it very well might do this internally even if you've only requested the first mode) and that you look at the other modes of buckling. As I said earlier, a linear buckling analysis is non-conservative by an unknown amount, so using this analysis type to determine the load that will cause buckling is very risky. A linear buckling analysis should be used to understand what modes of buckling could exist in your design, and how design changes modify the structure's response.

                            • Re: BUCKLING OF PISTON RODS
                              Raghav S

                              Thanks Shaun, your inputs helped a lot. One last thing if i apply a load that is greater than the critical load for the 1st mode and lesser than the critical load for the 2nd mode, can i still expect buckling?

                                • Re: BUCKLING OF PISTON RODS
                                  Shaun Densberger

                                  Yes. Real world buckling is an instability problem due to effects such as load eccentricity, non-homogenous material, and various other factors. This is the breakdown between highly idealized theory taught in beginining engineering courses and the real world.

                                   

                                  Besides, how could you apply a load beyond the critical buckling load without first applying the critical buckling load?