5 Replies Latest reply on Sep 17, 2007 4:44 PM by Ian Hogg

    hysteresis spring Cosmos Motion

    Mark O'Connor

      I need to model a compressive spring with a hysteresis loopcharacteristic. This is where the force-displacement responseduring compression is different from that when it is released. Thearea constrained by the two curves is the energy lost. I need it todefine a springpack in a railway drawgear assembly. This is theapproximate desired response:

      I am using CosmosMotion education edition SP5 and have so far beenunable to generate any form of hysteresis response. I would love tohear from anyone with any suggestions!

        • hysteresis spring Cosmos Motion
          genexxer genexxer
          How about two springs in parallel, one that has a trigger release at x=3 and the other is not released but simply decompresses back to x=0.
            • hysteresis spring Cosmos Motion
              Mark O'Connor
              Thanks for your suggestion.

              I should add that it needs to be able to switch curves at any x, asmost of the loads absorbed by the springpack will causedisplacements much less than x=3.

              Basically I am modelling this springpack to incorporate it into akinematic model of a train which will then be tested under variousoperating scenarios to determine peak loadings on track anddrawgear. We aim to repeat this simulation using variousspringpacks and with varying states of wear in the drawgearassembly to determine the optimum springpack choice for oursponsor.

              I am rapidly coming to the conclusion that COSMOSmotion isunsuitable for such a project!

              Any suggestions on how I can make progress with COSMOSmotion, orrecommendations of other packages I should be looking into?
                • hysteresis spring Cosmos Motion
                  Ian Hogg
                  Hi Mark,
                  I'm not sure on exactly how you want the intermediate response tobe (ie if the spring were to start being compressed after comingpart of the way back on the hysteresis curve, but I would do the 2spring approach, but with a slight twist. Let's assume that thespring is getting taken all the way up to the end of the curve andthen unloaded all the way back down
                  One of the tricks is obviously getting the spline data in. Once youread it into COSMOSMotion, it is given a Spline ID. If you takenote of that spline ID, you can use it in other functions evenafter you delete the entity that read in the spline data.
                  Instead of using the built in spring entities, create an actionreaction force between the parts at the same location where thespring will act.
                  Load in the first spline data to this entity and accept it(function type as spline/curve data) . Take note of the spline ID.Edit the force and now read in the second spline (it should have adifferent ID).
                  Now change the function type to expression. What you're going to dois based on the relative velocity decide which spline curve willrepresent the stiffness response of the spring.
                  In the COSMOSMotion help, you can find details on the IF statementsyntax and options. Let's assume that if the relative velocity is<=0, we use the compression spline, and if the relative velocityis >0, we use the release spline
                  We use markers for the action reaction force to measure therelative velocity, VR(I,J). There is a button on the functionexpression dialog to let you get the marker IDs for the parts thespring attaches to.
                  IF(VR(I,J):CUBSPL (DM(I,J), 0, id compression spline),CUBSPL(DM(I,J), 0, id compression spline),CUBSPL (DM(I,J), 0, id releasespline)
                  Where DM measures the absolute distance between the 2 attachmentpoints and is the x-axis of the spline curves. I'm not sure if thisis a complete fit for what you want to model.

                  Actually thinking about it a bit more, maybe this is what you wantto happen. It just depends on the timeline.
                  1. Define a spring using the first curve (ie action reaction force,spline data, DM for the x-axis).
                  2. Then define a damper between the same 2 points, and adjust ituntil you get an over damped response that emulates this returnfunction. It may take a little iteration as the damper will be afunction of time (and velocity).
                  Anyway, see if this gives you any useful ideas for yourapplication.
              • hysteresis spring Cosmos Motion
                genexxer genexxer
                Hi Ian,
                Will this allow the switch to either spline at any X?