8 Replies Latest reply on May 9, 2013 9:20 AM by Chris Michalski

    Gravity gone wild?

    Chris Michalski

      Okay I'm in 2009 SP4.1 so this might be a glitch that no longer exists.  I was having problems getting 1 component of an assembly to mesh so I decided to investigate it separately to find usable mesh settings.  Found those, but when I did a simple simulation to verify it was sufficient I get some screwy results.


      It's a welded shaft with differing cross-sections.  I started with 2 bearing supports then switched to 2 circular fixtures, both give weird results.  It is fine using a centrifugal load only, but applying gravity or a pressure/force on the OD face towards a reference plane is being misinterpretted somewhere. 


      Hints?  Am I missing something here or is this simply a glitch in the Matrix?  I tried gravity and centrifugal on a plain tube and things were fine.

      20130507 grav1a.jpg

      setup of tube with gravity and 2 cylindrical fixtures


      20130507 grav3a.jpg

      with centrifugal forces the results are reasonable, very small radial expansion


      20130507 grav2a.jpg

      But when I apply gravity it acts like it is a HUGE centrifugal load.  It should be bending downward, in the middle, not expanding radially.

        • Re: Gravity gone wild?
          Jared Conway

          any errors when running like large displacement?


          what solver are you using?


          is soft springs enabled?


          centrifugal load is a body load, it just goes outwards. if you want to get results that aren't bogus, you need to restrain the circumferential direction. so for example if you had a collar that rotates, you'd want to fix the center still and let it "spin" the material outwards from the center. remove any of those restraints and likely you'll throw an error.

            • Re: Gravity gone wild?
              Chris Michalski

              With gravity active it does throw the large displacement warning (as expected by the 1600m radial expansion calculated)

              I tried both the Direct Sparse and FFEPlus - same result.

              I also tried both curvature based and standard mesh.

              No soft springs.


              The centrifugal portion seems fine.  It is the gravity that is wrong.  Why does gravity produce results like a large centrifugal load?  Gravity is shown to produce a purely radial expansion when the centrifugal is supressed.  That bottom picture (with gravity) should sag in between the two supports to some minor extent as gravity acts in a single direction, not radially outward. 


              Those supports are cylindrical fixtures with zero radial motion allowed (and one of them zero axial).  Defining them as Bearing Supports gave the same results (centrifugal reasonable, gravity completely wrong).

                • Re: Gravity gone wild?
                  Jared Conway

                  large displacement warning = results are no good unless you run with large displacement enabled

                  and my guess is enabling large displacement will result in other problems


                  if you want to send me the model, i can take a look in the latest version but that's my guess at this point.


                  does it happen on a simple cylinder?

                    • Re: Gravity gone wild?
                      Chris Michalski

                      No, a simple cylinder with gravity drooped in the middle as expected.  That was my first test to verify I wasn't crazy. 

                      I even started a new study to make sure it wasn't inadvertently confusing the gravity and the centrifugal. 

                      I also tried a fixed hinge support to let it rotate freely.

                      Applying a force perpendicular to a plane (similar to gravity) induced radial expansion also (but without the large displacement warning).


                      Here is the file with the results files included.

                        • Re: Gravity gone wild?
                          Jared Conway

                          Hi Chris, in 2013 SP3.0, each of those studies suffers from a lack of restraints which is causing the behavior you're seeing. Lack of restraints > large displacement warning > if you click "no", the results are no good. They are for your information only. (Something I've been asking the developers to eliminate for a few years now) A quick test would be to make one of your restraints fully fixed instead of a hinge or bearing, you'll see what I'm talking about.


                          I haven't completely thought this through, but based on your loading conditions, you might be able to get away with half symmetry. Otherwise you need to restrain a point somewhere on the model to stop it from rotating which is causing your strange results.