Jump to content

Andy

Moderators
  • Content Count

    80
  • Joined

  • Last visited

  • Days Won

    8

Andy last won the day on December 10 2019

Andy had the most liked content!

About Andy

  • Rank
    Advanced User

Profile Information

  • Gender
    Male
  • Country
    United States
  • Are you University user?
    No

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Hi, Yes VisS is a damping coefficient on the interface stiffness which would effect the normal force. After reviewing your model, I am unable to find the source of the lost energy. Since the error only occurs with the contact model, there must be some part of the energy from the contact force that is not being tracked correctly. I'm not sure what you are simulating but if the results look reasonable to you then it is OK ignore the energy error. Thanks, Andy
  2. Hi, On your cluster you must have the HW desktop package installed and launch the Radioss simulation using the <install direct>/altair/scripts/radioss to automatically get the h3d file. If Radioss is launched directly using the executable then you must run hvtrans on the A001 files after the Radioss simulation is done. Running hvtrans could be after the job submission can be added to the submit script. This can be done using the command, altair/scripts/hvtrans -a *A001 *A001 -o model.h3d Thanks, Andy
  3. Hi, Thanks it doesn't seem to be the friction causing the issue. Can you run the model without damping and see how that changes the results? Also try setting in the model with FRIC=0 also set VISs = 1e-30 which will remove damping applied to the interface stiffness. Or if you can share the model, I can review it. Thanks, Andy
  4. Hi, The friction energy is also not included in the energy calculations. In the TYPE7 contact model try setting FRIC=0 to see if the energy error is low. If it is low, then the energy error is from the friction and it is OK to have that large of an energy error. Thanks, Andy
  5. Hi @gslee Thanks for the detailed information. The issue is that the damping energy is not included in any of the global energy output. The damping energy is also not included in the % energy error which is why your energy error is -34.4%. Since the total energy in your model is small (this is reasonable since your only load is the /CLOAD) thus the small amount of damping energy added creates a larger % energy error. Thanks, Andy
  6. Hi, There was some incorrect information in the post you linked to. External Work includes the work from Imposed movement (displacement, velocity or acceleration), force (/CLOAD or /GRAV), and pressure. It also include the frictional energy from a /RWALL (when FRIC!=0.0) What loads are applied in your model? Do you have Friction defined in the contacts or rigid wall? Also plot, contact energy, internal energy, kinetic energy, and hourglass energy. Also plot DTE which is TTE - external Work. Thank, Andy
  7. This is NOT correct. External Work includes the work from Imposed movement (displacement, velocity or acceleration), force (/CLOAD or /GRAV), and pressure. It also include the frictional energy from a /RWALL (when FRIC!=0.0) If there are only an initial velocity, gravity and a rigid wall in the model then I would assume the high negative external work is caused by the friction on the rigid wall and maybe gravity. It is difficult to say without seeing the model. Thank, Andy
  8. Hi, k= interface stiffness g = contact gap thickness p= penetration into the contact gap Ln is the natural logarithm. We will improve the description in the theory manual. Yes this is dealing with normal penetrations. The contact energy output in Radioss also include the tangential frictional force but I don't have a formula for that calculation. The best way to see if the contact energy is due to penetration or friction is to make FRIC=0 so there is no friction and rerun the model. Thanks, Andy
  9. Hi, Using LAW42 would be recommended since it is a rubber material. Using Ismstr = 10 is recommended with hyperelastic material laws. As HyperMan shows, /PROP/SOLID, Isolid=14 with Ismstr = 10 works. Or you can use /PROP/SOLID, IHKT = 2 with Isolid = 24 (HEPH) and Ismstr = 10 which runs a little faster and also works. We have this statement in the comments of /PROP/SOLID but I realize it is easy to miss. Thanks, Andy
  10. Hi Ken, You need to define /TH/NODE and /TH/RWALL in your input model. These are defined in HyperMesh as Output block. You also need to define the frequency of output which is the engine option /TFILE which is in the control cards in HyperMesh. See the tutorial Step 8 in RD-T: 3500 Tensile Test Setup in the 2019.1 help. Thanks, Andy
  11. The issue with LAW2 is that it is linear elastic. For Non-linear elasticity, you can use one of the hyper-elastic materials like LAW42, LAW69, LAW92, or LAW94. Example "RD-E: 5600 Hyperelastic Material with Curve Input" shows more details about these materials. To viscous (rate) effects to the material you can use the viscous input in LAW42 or add /VISC/PRONY to the other materials. However, data for the viscous input is difficult to determine. So then maybe Hyperman's suggestion of Rayleigh damping /DAMP would be easier. Thanks, Andy
  12. My guess is there is some issue with your engine file filename_0001.rad. Maybe you don't have an end time defined and thus there is no endtime and it just runs one cycle? RUN/Cantilever_beam/1/ 0.3 Thanks, Andy
  13. Hi, If you are using 2019 and modeling the bolts with solid elements then you can use the option /PRELOAD. See the /PRELOAD reference guide and the User's guide, Explicit Structural Finite Element Analysis, Loads, Preload Unfortunately the /PRELOAD option, setup is not supported by HyperMesh but it is supported in HyperCrash 2019. If you are not modeling your bolts with solid elements but instead with springs, then you need to use the method you mentioned in your post. The SPR_PRE spring only has stiffness in 1 degree of freedom (tension / compression between the two nodes.) The SPR_BEAM spring has 6 degrees of freedom (DOF). Since SPR_PRE is in parallel with SPR_BEAM, you set for SPR_BEAM, k1=0 and Stif0=10000 N/m. Then for DOF k2-k6 in the SPR_BEAM you must calculate the correct stiffness for the other directions for your bolt. Thanks, Andy
  14. Hi, I would suggest using an /RBODY instead of an /RBE2 to connect the items. Then you can disable the /RBODY by specifying a Skew_ID which would reference a /SENSOR/INTER or /SENSOR/RWALL depending on how you are modeling the floor. RD-E: 1200 Jumping Bicycle shows an example of using /SENSOR/RWALL. Thanks, Andy
  15. Hi, Unfortunately, we don't have a theory manual for the TYPE24 interface. Most things are work the same as TYPE7 except for the contact stiffness and gap. The contact stiffness constant in TYPE24. The initial stiffness is calculated and used. Versus in TYPE7 the stiffness increases as the node penetrates in to the Gap. All the other details about TYPE24 are in the comments in the Reference Guide. Thanks, Andy
×
×
  • Create New...