Jump to content

Adriano A. Koga

Members
  • Content Count

    545
  • Joined

  • Last visited

  • Days Won

    29

Adriano A. Koga last won the day on September 17

Adriano A. Koga had the most liked content!

About Adriano A. Koga

  • Rank
    Super User

Profile Information

  • Are you University user?
    No

Recent Profile Visitors

1419 profile views
  1. Have you included PARAM,LGDISP,1 or in your loadstep Nlparm(LGDISP)? This also turns on large displacements. For convergence issues, try adding NLADAPT to your loadstep and change a few convergence parameters such as max number of cutbacks, maximum DT, etc.. Also request PARAM, NLMON, DISP so that OptiStruct will write a separate . H3D file showing the displacement for each attempt in load increments. You can load this in HV and try to figure out what is going on. Also in the .out file a more detailed description on convergence is shown.
  2. Hi, If this is material related only, you can only have elasto plastic behavior, and so the maximum that you will get is perfect plasticity, with near zero slope of the stress strain curve. The behavior in your curve is only related to material or also includes some nonlinear geometric behavior?
  3. Matf is usually applied for failure calculation for composites. What do you mean by failure in your case? Mats1 should model elasto plastic behavior, amd you should have plasticity and some decrease in stiffness, but in OptiStruct you will not see element deletion. If you want to see your elements failing and vanishing, you need to go to RADIOSS.
  4. It depends. They would be different. One of them will give you the bracket frequency alone, considering only ita stiffness as it is bolted. But normally this information alone is not so important, as you probably won't have it operating by itself. Usually we would model the assembly of bracket plus the motor as this is the operating condition. Take care with modal analysis as you will have influence both of the bracket itself but also the motor has stiffness added to the system. So if you model only mass, your frequency will probably be lower than the real life, as you add mass but not stiffness. As an analogy f~srqt(k/m) The bottom line is that depending on how you sinplify your model, you will get different results, and some modes might not be captured due to the stiffness representation.
  5. I would suggest you first to get the current optimization ready, and then try to get more complete setup. Crawl, walk and run approach.
  6. Just like your first function, dresp2 can be used to combine different basic responses. So just follow the same structure to implement these. There's another option that is dresp3, which can call excel, compose, fortran..but this needs some coding, although it gives you much more freedom. Your model is too big..i couldn't run it in my machine. do you need such a big model?
  7. Usually at the free size step you don't have a clear status on the mass and other responses such as strains, displacements,etc.. So, usually using compliance, as it is a global measure of flexibility, helps you to identify where the main layers and regions should be placed. After free-size, i'd say you can start looking at some other responses, such as mass. If you have a mass target, then it can be used as a constraint. Otherwise use it as objective, and start adding other design constraints, such as composite strains or failure indexes. There's no "rule", as it depends on your problem, but I hope this helps you.
  8. by the way, for some reason, i couldn't see your pictures on my side here.
  9. your model is big..it has a couple millions elements...but ok..just takes longer to run. But i understand that you're running a thermal transient analysis, and you get your temperature fields from somewhre else, maybe CFD.. How do you want to use these temperatures in your transient run? Do you want to apply them as initial conditions? or as imposed temperature curves? For what i've seen you can calculate temperature, but maybe it is not what you expected.
  10. if i'm not wrong you could just use objective: minimize compliance (response: compliace == inverse of stiffness), with no constraints. I believe topography accepts compliance (not quite sure right now...). Or if it is not allowed, then try to create a displacement response, for the region of interest (a single node), and add this a objective: minimize Displacement. For topography you might want to turn on some symmetry constraints (pattern grouping) to get more feasible designs.
  11. i've made a few modifications for testing. Still didn't finish, but you might want to take a look at it and check the idea. We better check how to make your model running more efficiently. I've changed from direct to modal FRF and adjusted the extracted modes. Also changed for EIGRA, as it is more efficient for large models. In theory you could. But it would complicate a little further your model. hibrid2_acceleration_contact_modal_opt_v3mod.zip
  12. in InspireStructure, you're able to break your bodies into smaller parts, and assign to each part a mesh size (manual instead of automatic), using the Browser in the left. Maybe this works as well for InspireCast. I will let some other specialists give you some other options, in this case.
  13. hi @Mirhan Ozdemir I believe you would need some script for doing this, as there's no automatic conversion for this case in HM. Have you any experience with scripts in HM? Try taking a look at the script section here in the Forum.
  14. gap elements or using freeze contact between 2 surfaces will provide conductive heat transfer accross the parts. For convection you would need to add a special convection group in the shell side, and define a ambient temperature and convective coefficient. There's a tutorial in OS Help for this kind of problems (convection). Contacts and gaps i don't recall having a tutorial for thermal.
  15. did your model run until the end or did you get any error? Can you share your .out file?
×
×
  • Create New...