Time steps and step size errors wrt curve editor input

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • sumiray
    Junior Member
    • Aug 2012
    • 23

    Time steps and step size errors wrt curve editor input

    I have to set and run a experiment as of the following. A pressure has to be applied and sustained on the inner surface of the hemisphere; 4 Pa. The pressure has to be sustained throughout the experiment duration; hence I applied the pressure in the initial step with curve editor value as (0,4) to (1,4).

    Subsequently, a pressure of -85.5 Pa is to be applied onto the base surface of the extruded part on the hemisphere. I added the pressure as an additional step with the curve editor value as (0,0) to (1, -85.5). The number of time steps and step size was set as the default of 10 and 0.1 respectively for the 2nd pressure loading. However, negative jacobian pts were encountered during the running in febio (filename: exp1.feb). So I modified the time step and step size to 10 and 0.001 respectively. This allowed the file to run in febio (filename: exp2.feb). However, technically with these modified magnitudes, the time duration for the 2nd pressure loading should be 10 x 0.001 = 0.01. With that assumption, I further changed the curve editor value now to (0,0) to (0.01, -85.5) instead and tried to run it in febio again. However, the file ran into error termination again (filename: exp3.feb) with negative jacobians.

    This has left me quite confused about what the relation is of the curve editor values wrt the time step/step sizes.

    Exp1.feb (error)
    Intial step: 1st pressure load of 4 Pa applied (0,4) – (1,4)
    Step 1: 2nd pressure load (0,0) – (1,-85.5)
    Steps/step size = 10/0.1

    Exp2.feb (runs in febio)
    Step 1: 2nd pressure load (0,0) – (1, -85.5)
    Steps/step size = 10/0.001


    Exp3.feb (error)
    Step 1: 2nd pressure load (0,0) – (0.01,-85.5)
    Steps/stepsize = 10/0.001
    Attached Files
  • sumiray
    Junior Member
    • Aug 2012
    • 23

    #2
    urgent help needed!

    Hi,

    I have posted this thread for a week and yet to garner any replies. I would sincerely appreciate it if i could get some guidance for this problem. Thank you!

    Comment

    • dsrawlins
      Developer
      • Dec 2008
      • 366

      #3
      Hi Sumitra,

      Just an update that I'm working on this problem. See the attached exp6.zip. Instead of using two time steps I suggest using loadcurves to control the pressure loads. The pressure on the hemisphere needs to be ramped up over a finite time period. I'm increasing the pressure from t = 0 to t = 1 and then holding it constant. The pressure on the extruded part is 0 to t =1 and ramps up to -85.5 at time t = 20. At this point it's still error termination so I need to play with the run parameters a bit.

      Cheers,

      Dave

      exp6.zip
      Department of Bioengineering, University of Utah
      Scientific Computing and Imaging institute, University of Utah

      Comment

      • maas
        Lead Code Developer
        • Nov 2007
        • 3400

        #4
        Hi Sumitra,

        Exp1 and Exp3 are essentially the same problem. Since this is a static analysis, time is somewhat arbitrary and letting your model run for 10 steps of 0.1 or 10 steps of .001 makes no difference, as long as you define the loadcurve consistently. In the first case, the end time will 1.0 and in the second time .01. Thus, if you define your loadcurve to end at 1 in the first case and .01 in the second case (as you did for Exp1 and Exp3), you effectively have two identical problems. I hope this kind of makes sense. This also explains why Exp2 run, since you only ran the model to a load of -0.855.

        Looking at your model, I'm also a little concerned that you are running into a geometrical instability. You're applying a buckling load so there will be a max load for which FEBio will be able to solve. You may consider adding additional constraints to stabilize the model.

        Hope this helps.

        Cheers,

        Steve
        Department of Bioengineering, University of Utah
        Scientific Computing and Imaging institute, University of Utah

        Comment

        • sumiray
          Junior Member
          • Aug 2012
          • 23

          #5
          HIHI,

          Thank you for the much needed help! and yes, i did play around with the force application and realized that maybe the 2nd pressure load is causing instability that causes the running in febio to run into error termination. So i changed the application of load type to traction load of x=0, y=0, z=1 so that the parts will not be clashing/intersecting when under deformation. I adopted the steps kindly provided by Dave and it managed to run without any buckling issues.

          With the timesteps as 200 and size as 0.1 as inputs, however after running the feb file in febio and opening the xplt file in postview, i only see the completion of 104 steps. Does this mean that my program has not run completely despite the normal termination or is this due to utilising the must points?

          And as evident in the boundary conditions input, the rim of the hollow hemisphere was fixed stationary for x/y/z directions to prevent the model from flying away into space. But is there any way i could actually prescribe the displacement of the rim about the center of the hemisphere such that the hemisphere model rotates about its axis when the 2nd load is applied downwards?
          Attached Files

          Comment

          • dsrawlins
            Developer
            • Dec 2008
            • 366

            #6
            Hi Sumitra,

            By using the automatic time stepper, FEBio is able to take larger time steps which explains why you get less than 200 steps. I tried attaching the hemisphere to a rigid body which was allowed to rotate around the y-axis, but this proved to be an under-constrained problem. Note that if you constrain the y-rotation for the rigid body that you get the same solution as you did constraining the nodes in x,y,z. Hope that helps.

            Cheers,

            Dave
            Attached Files
            Department of Bioengineering, University of Utah
            Scientific Computing and Imaging institute, University of Utah

            Comment

            • sumiray
              Junior Member
              • Aug 2012
              • 23

              #7
              ahmmm, i tried to tweak the boundary conditions bearing your advice in mind. And i changed the fixed displacement on the rim to fixing all the displacement except for Ry. I then ran the feb file and it managed to reach normal termination. So from the understanding that i have, i am assuming that this would mean that face of the rim would only be allowed to rotate about the y-axis? I am looking at enabling the model to rotate/displace with the pulling down force of the 2nd load.

              Thank you, Dave!
              Attached Files

              Comment

              • sumiray
                Junior Member
                • Aug 2012
                • 23

                #8
                Is there any way i could impose a boundary condition or a spring element such that the model rotates about its center of origin when the downward traction force (second load) is applied on the extruded part of the hemisphere? I tried different conditions but i still cannot manage to make the model rotate.
                Thank you!

                ps. the .feb file is uploaded above this post.

                Comment

                • dsrawlins
                  Developer
                  • Dec 2008
                  • 366

                  #9
                  Hi Sumitra,

                  If the edge of the hemisphere is not constrained, any force would just cause it to spin, leading to an underconstrained model. I tried adding a spring, but that didn't help. It depends on what you're actually trying to model, but you might try using tied contact between the hemisphere and a disk with a deformable material which is allowed to deform in the xz plane.

                  Cheers,

                  Dave
                  Department of Bioengineering, University of Utah
                  Scientific Computing and Imaging institute, University of Utah

                  Comment

                  Working...
                  X
                  😀
                  😂
                  🥰
                  😘
                  🤢
                  😎
                  😞
                  😡
                  👍
                  👎