4 Replies Latest reply on Feb 17, 2009 9:03 PM by Fred Lampela

    Buckling of Hydraulic Cylinder

    Fred Lampela
      I have a quiestion for you experts out there. I am currently working on a rather large hydralic cylinder, the engineer here has a spreadsheet for calculating buckling using Euler's formula, mostly concerned with rod dia., so in the spreadsheet we are using the extended length of the cylinder & the rod diameter for the entire length of the extended cyl. This 6" bore cyl operates at 4500 PSI has an extended length of 201.25". So the spreadsheet says we need a 5.16" rod dia to achieve a 2:1 factor of safety. I ran the cylinder with a 4.500" dia rod in simulation, & then ran a rod 5.160 dia. of the extended length to compare against what the spreadsheet was telling us.
      In the cylinder I restrained both end clevis's and applied 4500 PSI load to the inside cyl wall & the piston face. The pins were free to rotate in the clevis bores. & gravity applied, with cylinder in a horizontal position. I was careful to have the wear ring supports in the proper locations on the piston & gland to simulate the actual support.
      In the rod simulation I fixed one clevis, applied a load of 127,235 lbs to the other clevis, again gravity applied & pins able to rotate in the clevis bores. I have attached JPG's of the simulations, cyinder & 5.160" rod.
      Now the question: Is the hydraulic cylinder bucking in simulation reliable enough to say, yes a 4.500" cylinder rod adequate? Or shoud I continue to rely on Eulers formula for column buckling on hydralic cylinders. My though is that the way we are calculating with the spreadsheet isn't considering the added strength the cylinder tube adds for buckling considerations.
      I ran the rod simulation with a 4.500" dia on the rod also & stresses were near 40,000 psi., so a 4.500 column the ext length doesn't cut it, but a 5.160" does, and the cyl simulation with a 4.500" rod is very close to the stresses in the 5.160" rod.
      Any thoughts? Are the simulation methods I am using the right way to approach this?
        • Buckling of Hydraulic Cylinder
          sandeep pawar
          Fred,

          Out of curiosity, why are you adding gravity?

          Thanks,
          Sandeep
            • Buckling of Hydraulic Cylinder
              Fred Lampela
              These cylinders are for a large knucle boom type loader, they will cycle from vertical to horizontal, so added gravity & ran simulation with the cylinder horizontal to give side load as gravity would do. The deflection shown is at a scale of 20, you can see the center bowing down slightly. I believe the formula we use on our spreadsheet is a common method of calculating column buckling, I could be wrong but, I don't think it takes gravity into consideration, or any other external side loads, but maybe a there is some consideration given in the factor of saftey.
            • Buckling of Hydraulic Cylinder
              Bill McEachern
              Buckling is far more complicated and subtle than most people realize. The Euler computation implicitly assumes that the stucture is undeformed - the loads do not affect the stiffness and it is essentially a part or bonded assembly. In a FEA simulation you have 3 kinds of buckling: linear, non-linear and post buckling. If you want to cut it close you need post buckling - that is you do an NL ananlysis with RIKS (arc length in cosmos speak) or the Crissfield control scheme to have the numerics hang together so that the solution can proceed through the snap thru and see where it actually buckled. In cosmos contact is not supported with this control scheme. I have been to within less than 1% as compared to experimental. Then you have Non-linear buckling: you do a static analysis of the loads to prestress/deform the model and then you do an eigenvalue extraction on the resulting stiffness matrix to get the buckling estimate of the deformed and loaded structure. I believe that Cosmos Professional is capable of this type. You can tell if the problem looks like it is being solved twice. This is a closer estimate but can still result in significant uncertainties depending on how close you are to a geometric induced instability - ie load eccentricity induces the buckling as opposed straight compression. In linear buckling none of the loads are used other than to give you the buckling load factor. In ASME pressure vessels a factor of 4 is used on linear buckling just to give some perspective to the potential uncertainties. For the case at hand where load eccentricities seem difficult to come by due to the pinned ends you need to apply your engineering judgement to sort out how close you can go. I tend to devise expereiments that I can run either numerically of experimentally to bound my uncertainties. Hope that helps.