• RecurringMeetings
• 2014-11-26

Recurring Meeting of Cleveland Clinic - University of Utah

Date: November 26, 2014

Time: 2:00 PM EST

Means: Conference call

Attendees:

1. Ahmet Erdemir (Cleveland Clinic)
2. Jeff Weiss (University of Utah)
3. Ben Ellis (University of Utah)
4. Steve Maas (University of Utah)

Agenda:

1. Update on feature implementations.

   1. In situ strain - http://wiki.simtk.org/openknee/Febio/InSituStrain

   2. Set definitions - http://wiki.simtk.org/openknee/Febio/SetDefinitions

   3. Connector elements - http://wiki.simtk.org/openknee/Febio/RigidBodyMotions

2. Recap on FEBio segmentation fault issues during simulations.
3. Decide action items for next meeting.
4. Other.

Immediate Action Items:

• Cleveland Clinic (Ahmet & Craig)

  • Start testing in situ strain feature in relevance to its use in Open Knee(s).
  • Continue testing the sets feature.
  • Provide performance metrics on Open Knee(s) - Generation 1 model.
• University of Utah (Steve)
  • Prepare a draft of a manuscript about in situ strain feature.
  • Try to implement text file output for desired sets by referring to set label.

Notes:

1. Update on feature implementations.
   • Based on the last conference call and CMBBE 2014 presentation, in situ strain feature is in good shape. The feature is currently provided as a plug-in. There were discussions to incorporate this this capability in the core of FEBio code (probably not). Steve will start working on the manuscript, the goal is to get a draft ready by the next meeting. The Cleveland Clinic team will start testing this feature to assure it fits their needs. This testing will also help evaluate the plug-in framework.

   • Steve changed how node sets are defined. This was necessitated by the need to handle XML file more effectively and it conforms to the way that element and surface sets are defined. Ahmet also noted that the new definition syntax may help Python scripting as well. Currently both the newer and older formats are supported. In PostView, one can select surfaces, node sets, or parts; PreView utilizes the new format now. For the next meeting, Steve may be able to implement text file output for desired sets by referring to set label. As noted by Ahmet, this will be great convenience for future simulations using Open Knee(s). The Cleveland Clinic team will continue testing the sets feature.

   • Steve and Gerard Ateshian talked about connector element functionality. It seems like Gerard had good progress in the implementation of this feature. Currently, the issue is getting FEBio converge with these constraints implemented.
2. Recap on FEBio segmentation fault issues during simulations.
   • Ahmet noted that Steve's implementation to capture NaN and cutback step size had resolved segmentation fault problem. Ahmet ran 20 simulations with the problematic Open Knee(s) model (new release candidate updated for FEBio version 2) with coarse and fine convergence tolerances. All these simulations terminated normally. He is currently exploring convergence characteristics of this model under various loading scenarios. Steve is interested in focusing on performance and convergence issues.
3. Decide action items for next meeting.
   • See Immediate Action Items above.
4. Other.
   • Due to holiday schedule, the group decided to have the next meeting on January 7, 2015 (@ 2 PM US/Central). University of Utah team will need to identify possible dates and times for ongoing meetings in view of their 2015 academic schedule.
   • Ahmet ran Open Knee(s) - Generation 1 model (new release candidate updated for FEBio version 2) on the in-house high performance computing cluster at the Cleveland Clinic, with different number of threads (1, 2, 4, 8, and 16). He will provide performance metrics to the University of Utah team for reference.
   • Ahmet reminded the University of Utah team about the possibility to visit Cleveland Clinic site; all seems to be interested in this.