Jump to content

Robert Lawson

  • Content Count

  • Joined

  • Last visited

  • Days Won


Robert Lawson last won the day on November 22 2019

Robert Lawson had the most liked content!

1 Follower

About Robert Lawson

  • Rank

Profile Information

  • Are you University user?

Recent Profile Visitors

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

  1. I ran an analysis w/ just the modal subcase here and it worked fine as you have it. I tried to run the optimization, but I don't have the latest version on our solvers. I get an error "acceleration card not supported for optimization" (or something like that). I suspect that is updated/fixed for the current version. I may be at the end of my usefulness. Good luck!
  2. Those undercarriage beams are not connected to the RBE3's in a sensible way. The RBE3 doesn't have any stiffness of it's own, so those undercarriage beams are essentially just floating around. I suspect you ought to extend those beams out another element-ish and connect into the center (dependent) node on the RBE3. That's one problem. Next, you can't constrain the center node of an RBE3 because it's already dependent on the nodes it connects too. If you still really want to constrain it, you could put something like a stiff CBUSH between the undercarriage beams and the RBE3 to break up the dependency. Then you can constrain the CBUSH grid, just adjacent to the center RBE3 grid.
  3. Assuming you're analyzing in Optistruct... For 2D PCOMP analysis, you would want to use a MAT8 card. For 3D solid core, you want a MAT9OR card. HC manufacturers are only giving you the two pertinent shear modulus values. Assuming X and Y are in-plane, and Z is out-of-plane: For MAT8, the hear modulus values above are G1z and G2z. For MAT9OR, the shear modulus values above are G23 and G31. For MAT9OR, E3 is something significant. I would start with the modulus of aluminum (assuming aluminum core), and knock it down by the void fraction of the core. The other values are typically small, but probably can't be ignored. You can easily come up with combinations that don't satisfy the irritable constitutive relationships.
  4. Are you sure when you did your line mesh, you got actually got structural elements? Your picture above doesn't show any bars, beams or rods on those lines. Your pink rigids connecting to the green structure appear to just be connecting to single nodes on the green. That's likely going to result in some unrealistic stresses and deformation there. You'll likely need to grab a bigger footprint there.
  5. M.karthickraja - If you look at corner stress data instead of centroidal stress data, you will likely find the results are much closer. - Robert
  6. In a case like this, with one ill-tempered surface, I'd suggest just simply deleting the offending surface or surfaces. Then re-create a new one with ruled. With that, let's hope it'd behave well enough to create a solid. There's one more plane of symmetry here, so you could do a bit less work yet. Split this down the ZY plane. - Robert
  7. There's a lot of different ways to mesh this. You might start by simply partitioning the solid geometry into two or three separate blocks. Then mesh how you did the above attempt. Alternatively, create a 2D automesh on the interesting side, and then line drag that 2D mesh into 3D brick elements. - Robert
  8. Hi ElineH - In the help for DOPTPRM, note 1 says "optimization algorithms are automatically selected by OptiStruct based on the optimization type". https://altairhyperworks.com/hwhelp/Altair/2019/help/os/topics/solvers/os/doptprm_optmeth_bulk_r.htm I'm looking at a .out file from a run, and I see no information stating which method it actually used. Perhaps there's a setting to get more verbose output, but I'm not seeing anything obvious. - Robert
  9. Good day jang - You have a several parts here that are not attached to each other. You can use the tool "find" (shift F5) to find elements attached. For instance, "component1" is not attached to "auto1". So there is no load path from the loads to ground (SPC). So you are asking OS to solve x=F/K, but K=0 in this case. - Robert
  10. OS wants you to use the stress option on the DTPL card instead of a DRESP1. However, that means you won't be able to pass that response to your DRESP2/DEQATN. This might be a show-stopper for you.
  11. I'm not sure I follow this, but I'll try. Also, I'm not sure this is a reasonable request from the customer. Though "the customer is always right" as they say. So they are asking for each individual mode's contribution to the overall RMS stress? Typically, I would plot the Stress PSD for high stress elements of interest. From that you'd expect to see peaks at pertinent modes. Then you could go view those mode animations and confirm that the Stress PSD output makes sense and is believable. If that's not sufficient for this customer... hmmm. How many modes do you have in the frequency range of interest? If something small-ish like 10-20ish, you could use MODESELECT case control to select one mode at a time, and re-run the simulation. This is a bit clunky. If you've got >20 modes, this is busy work. If you have to do this for a lot of elements/responses --> really no fun at all.
  12. Francesca - Does this model run as an analysis run? I suspect the problem is in the analysis model itself, not necessarily the optimization setup. It could be a semi-infinite number of problems. I can't DL the model to look though -- I think only Rahul R can see what comes into the dropbox. - Robert
  13. It seems like this should work as you have it, but it doesn't. It must be being a bit lazy (or efficient?) about creating a mass matrix if it doesn't need to. Add an ANALYSIS=MODES and METHOD=? to your case control, and then an EIGRL to your bulk data. Then it will give you a MAAX.
  14. Failure criterion like Tsai-Wu are relevant to typical composite laminates (skins). Those failure criteria are not relevant to core materials. For core materials, typically you would recover the solid element stresses (in some HC material relevant coordinate system) and compare to some strength allowables. Those would typically come from a Hexcel (or other vendor) data sheet. Assuming X and Y are in-plane, and Z is out-of-plane, you'd look at XZ shear and YZ shear stresses. You should pay attention to the ribbon L direction and transverse W direction. Depending on your loads and design, you should probably look at Z normal stress, and compare to the compression (stabilized) allowable. They don't give you a tensile allowable because that would depend on how it's bonded to the skins. A Z normal tensile failure mode would likely be in the bond to the core or the 1st ply of the laminate, not the core itself. Depending on your manufacturing quality these might be some small numbers.
  15. Looks like you need to define the G's in your materials which are currently blank. In the case of the core, G23 and G31 are the key stiffness values relevant for honeycomb core. Typically, these would be grabbed from a Hexcel datasheet. For the carbon fiber material, you can estimate G1Z and G2Z to be the same as G12. - Robert
  • Create New...