Jump to content

Simon Križnik

Members
  • Content Count

    985
  • Joined

  • Last visited

  • Days Won

    84

Everything posted by Simon Križnik

  1. Pneumatic structures can be modeled in Optistruct by applying outward facing pressure load.
  2. Thanks Prakash, your question actually resolved my issue. Membrane behaviour can be modeled by referencing an additional isotropic material of negligible stiffness with USER option on MID2 and MID3. optistruct_fabric2.hm
  3. Hi, I am trying to model fabric material in Optistruct with NLGEOM solver. The material model should be orthotropic, membrane behaviour (no bending and transverse shear) and have different modulus of elasticity for tension and compression. There are some issues: 1. Is there another way to define different modulus of elasticity in tension/compression for shell elements? 2. Membrane type behaviour is modeled with MID2 and MID3 set to blank as was suggested in this topic. Alternatively it could be set up with NIP=1 in PSHELLX extension. However, after analysis is completed the result display only the initial configuration. The file is attached for debugging purposes. optistruct_fabric.hm
  4. Hi George, thanks for response. Turns out it is quite simple to determine mass at specified pressure. In quasistatic simulation the mass/time function should slowly increase and the mass at the required pressure can be found in Hypergraph.
  5. Hi, The airbag injectors are defined by a function of mass/time so it takes some trial and error to converge on the pressure required. Is there a more direct way to determine the pressure at the end of inflation?
  6. Hi, I have a similar query regarding orientation of orthotropic property /PROP/TYPE9 (SH_ORTH). There is an option to orient the property by reference vector. That is fine for flat surfaces, but how it works for more complex curved surfaces like spheres?
  7. Vx, Vy and Vz are x,y and z components (global coordinate system) of a vector V which defines material orientation. Those are filled in automatically when you reference by nodes. Woven composite is modeled by inputting the same Young's modulus E11=E22 into material law definition.
  8. Hi, you should define type 11 line interface in addition to type 7 (or just type 19 which is a combination of type 7 and 11) in order to avoid nodes passing through master surfaces. Radioss user guide describes:
  9. Comparing energy error time history with animation it became apparent there was a modeling issue: shell elements were penetrating into solid edges.
  10. Hello, I have sent the requested files to dropbox. Some parts were modeled with under-integrated elements but hourglass energy is constant throughout. I also ran the fully integrated model and the same issue occurs.
  11. Increasing the min_gap worked, but during computation the energy error suddenly drops to -99.9% and unrealistic stresses propagate through the structure, when compared to the run with traditional mass scaling (at 1/10 the timestep of AMS).
  12. Thanks for being helpful as usual, George. The interface could also be made stiffer by increasing stiffness scale factor Stfac.
  13. Hi George, Thanks for the tip. The manual also suggests this contact deactivation can be avoided by using Spotflag=25 or 26 (TYPE2 penalty formulation). Both methods avoid WARNING ID : 852 ** WARNING SLAVE NODE OF AN INTERFACE TYPE2 & AMS; however some slave nodes of type 7 are still de-activated during simulation, eventhough they are not referenced in type 2. So this de-activation is probably caused by type 7 interface definition (ouput file):
  14. Actually, the deactivation of nodes was due to another interface (type 7) and it was solved by setting the Irem_gap to 1. I would still like to know the reason of WARNING ID : 852
  15. Hi, I am trying to run a model with tied interface (Type 2) and AMS. The interface is defined with contact surface as master and nodes on the slave side. During starter the following warning appears: WARNING ID : 852 ** WARNING SLAVE NODE OF AN INTERFACE TYPE2 & AMS and during computation: WARNING INTERFACE 2 NODE xxx DE-ACTIVATED FROM INTERFACE Can this deactivation of slave nodes be avoided when using AMS and type 2 interface? The same model was run successfully using DT/NODA/CST.
  16. Just wanted to update: the problem was I did not give reduction factor (R_E) which is 1 by default on material law 19 (fabric), so elastic modulus is the same in compression and tension. Bellow is the simulation with R_E at 0.001 giving good results.
  17. Hi, I am trying to convert Radioss file containing monitored volume (perfect gas) to Optistruct. In the convert browser the monitored volume is marked green i.e. "all active attributes for a given entity are mapped". After conversion there is control volume listed as type airbag, but with no parameters other than entity selection. When running analysis in Optistruct (static and explicit dynamic) it seems like monitored volume has no effect on the results. Am I missing something?
  18. Thanks again, Prakash. Actually, I was referring to minimum thickness (THICK), not member size (MEMBSIZ), but the syntax is the same. I am still having difficulty setting up radopt so looking forward to GUI.
  19. Hi matteoconese, you can check transparency in Surface Transparency Panel (shortcut ctrl+t).
  20. Hi Alexander, after checking the box, click return and then click update in the responses panel. What is the point of using weighted compliance with only one loadstep?
  21. Hi Prakash, thanks for the info. How is the minimum thickness defined in radopt syntax? I was unable to run with the following:
  22. Hi George, my issue with stress response is the run terminates at the end of first outer loop with ERROR # 382 (as described in my opening post) so I get no results. Apparently there is no such problem on your end since you managed to get some results at all. I have no idea what could be the cause, if you did not modify the files.
  23. Hi George, I tried as instructed but the issue remains- it looks like solver runs just as well with either *.rad or *.radopt file extensions. The issue seems to be with stress response in particular, because other responses worked fine.
  24. Hi all, I am trying to perform optimization in Radioss through -radopt command. The trouble is that computation terminates at the end of first outer loop when stress response is used: I have attached files, if anyone would like to check it out. test1_0000.rad test1_0001.rad test1.rad
×
×
  • Create New...