Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


Alessandro last won the day on June 2 2017

Alessandro had the most liked content!

About Alessandro

  • 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. Hi Prakash, Unfortunately, I cannot share the entire out file. Is there anything specific you would like to check? So I can extract and share it... In the meantime, I have updated the solvers from version 2017.1 to version 2017.2 and the error message is slightly changed. You can see below an example of the error I had after 147 iterations of an optimisation: [...] Lower limit of variable 44458 is larger than its upper limit: 0.1006E+01 > 0.1000E+01 Lower limit of variable 44460 is larger than its upper limit: 0.1004E+01 > 0.1000E+01 Lower limit of variable 44461 is larger than its upper limit: 0.1006E+01 > 0.1000E+01 Lower limit of variable 44463 is larger than its upper limit: 0.1004E+01 > 0.1000E+01 *** ERROR # 6896 *** Programming error, variable move limit calculated by movelmt0 is invalid. This error occurs in module "appdrv". Moreover, in the .fem file I do not have any of the variables called in the error (44458, 44460, 44461, etc..) Thanks again, Alessandro
  2. Hi Prakash Pagadala, Thank you for your reply. Anyway, I double checked and I don't have any similar bounds in the free-size optimisation. In my model, I have bounds only for the LAMTHK and the PLYPCT, and lower and upper bounds are different. Moreover, I do not have the error when the run starts but after several iterations. Do you have ideas about any other source for the error? Thanks you again, Alessandro
  3. Dear Community, I am running a free-size optimization with a quite complex model and the run is aborted after a number of iteration for the ERROR 703. I receive the following message: Error in input data - Lower bound larger than upper bound44879 0.100408E+01 0.100000E+01 Error in input data - Lower bound larger than upper bound44882 0.100408E+01 0.100000E+01 Error in input data - Lower bound larger than upper bound44897 0.100408E+01 0.100000E+01 Error in input data - Lower bound larger than upper bound44900 0.100408E+01 0.100000E+01 Error in input data - Lower bound larger than upper bound44918 0.100408E+01 0.100000E+01 *** ERROR # 703 *** number of design variables = 47541 number of constraints = 37 error code from optimizer = 900 The optimizer failed to find a solution of the optimization problem at the current iteration. This error occurs in module "appdrv". Other information: If I change design parameters, the iteration where I have the error also change. If I remove ply drop-off constraint, the error disappears. If I reduce the design domain, the error disappears. (unfortunately, I cannot share the model) Any idea on how to solve this issue? Thank you in advance, Alessandro
  4. Dear Community, I would like to present you a possible bug I found in Hypermesh 2017 when it creates the input file for optistruct including a buckling analysis (due to limited time, I have not checked if the same problem appears with other analyses). Please find attached an example and refer to the example for the description below. When you try to create the Optistruct file to run (for analysis only, -optskip), Hypermesh creates the following subcase card for buckling: SUBCASE12345678 LABEL Linear Buckling ANALYSIS BUCK SPC = 1 METHOD(STRUCTURE) = 3 STATSUB(BUCKLING) = 1 The first row is missing a space and the solver gives you the following error if the space is not added manually: *** ERROR # 1461 *** in the input data: Card "LOAD" is not allowed for this subcase type. The problem seems to be present only if the loadstep ID is made by 8 digits. I hope this will be helpful to somebody. Regards, Alessandro buckling_issue.hm
  5. I would like to add something else that helped me to bypass the problem keeping zero as lower bound in the DESVARs. I set a constraint on the composite failure and this stops the solver to go for zero-thickness laminates to avoid failure. I hope this is helpful for other users. Regards, Alessandro
  6. Hi Prakash, I have found that the problem is not the sum of all the TMANUF but the fact that, having zero as lower bound in the DESVARs, the solver is allowed to go to zero thickness for the laminates during the optimization. When this happens for some elements, the solver stops and gives the error 789. Of course, I cannot solve setting a non-zero lower bound for all the DESVARs. In that case the resultant structure will not be optimised at all, forcing the solver to have at least one TMANUF thickness for every ply. Instead, I solved setting TMANUF as lower bound only for the DESVARs related with ply shapes covering the entire laminates, assuring that the laminates cannot go to zero in any elements. Finally, the only thing I do not understand is why the behaviour described above happens even if a minimum laminate thickness is defined for the laminates in the DCOMPs. This is probably due (I don't know if you can confirm) to the fact that the solver does not try to satisfy all the constraints together but priorities based on violation percentage. It could be nice (for future developments) having the possibility of manually defining a set of constraints that needs to be satisfied, with maximum priority, at each iteration. Also the possibility of defining low priorities for others could be useful. I would consider this issue solved. Thanks for the support. Regards, Alessandro
  7. Hi Prakash, Thank you for your reply. I was probably not clear in explaining. Unfortunately, I have the same problem even if I delete the minimum and maximum laminate thickness constrains. Can you see other possible sources of the error? Thanks again, Alessandro
  8. Hi, I would like to add that sometimes the solver seems to be able to go beyond the first iterations if I set the TMANUF value only for a limited number of plies (not all of them) and, then, it stops later (e.g., iteration 8), writing the same error 789. Thanks again, Alessandro
  9. Dear community, I have a problem with the composite size optimization if I set a value for TMANUF for the plies. After the free-sizing, I use the sized model to perform a sizing optimization setting a specific manufacturable thickness for the plies involved (TMANUF). The solver is able to run the iteration zero, but it stops giving me the following error when iteration 1 starts. *** ERROR # 789 *** Invalid design properties w.r.t. the design variables were found. number of invalid property items = 6 This happens only if I set a value for TMANUF and also if I delete all the settings for maximum and minimum laminate thickness. Unfortunately, I cannot share the model I am working on. Any suggestion on the matter? Thanks in advance, Alessandro
  10. Dear Community, I find a problem in imposing minimum ply thickness constraint (PLYTHK) during free-size optimization. Whatever value I put in the DSIZE card as minimum ply thickness, this seems to be ignored by the solver and I get, in the resultant size model, plies with thickness lower than the value I specify. I have attached a simple model as example. In the example: Although I set 0.05mm as minimum ply thickness for all the plies, I get plies with lower thickness (e.g, ply 205200 is 0.02mm thick, ply 310300 is 0.0096mm thick, and so on...) in the sized resultant model. Has anybody found the same behaviour? and, has anybody found a solution? Thanks for the support, Alessandro plythk_issue_sizing.13.fem plythk_issue.hm
  11. I withdraw my question. I have found the following behaviours: 1)if I run the model with -optskip I cannot obtain any response in the output file 2) if I run the model as an optimization (without using -optskip) but deleting all the design variables, I will have a normal static analysis as result but I will also have the responses in the output file. The solution 2) is what I was looking for. I hope this can be helpful to somebody. Thanks, Alessandro
  12. Dear community, I am wondering if it is possible to print the responses defined for optistruct (as per optimization) to output file even if optimization is not performed (-optskip). This could be very useful to verify a model, in terms of constraints fulfilment, without performing an optimization. I know that it is possible to print all unretained responses to output file with "RESPRINT = option", but this is not working if optimization is not running. Any suggestion? Thanks in advance, Alessandro
  13. Dear Community, I have a problem when creating an Optistruct model for freesize optimization including a ply drop-off (PLYDRP) constraint. In particular, I created such a model in Hypermesh 2017 with different design variables (DSIZE). If I export the model in Optistruct format (.fem) and I import it again in Hypermesh 2017, I am not anymore able to see all the DSIZEs but I can see only one of it. I have created a simple model to give you an example. In the file attached, you will find (among the others) four DSIZEs: top, bottom, left, and right. The PLYDRP constraint is activate for all of these design variables. If I export the model to an Optistruct input file (.fem) and I import it again in Hypermesh, I cannot see anymore the four DSIZEs, but only one of them (top) is still there. This happens only if PLYDRP is activated. I would really appreciate if anybody could give me an explanation about this strange behaviour and what to do to avoid it. Thank you in advance, Alessandro plydrop_issue.hm
  14. Hi Prakash, Thank you for your reply but this doesn't work. I have found, instead, that you need manually to write the line "OUTPUT,FSTOSZ,ALL" in the .fem input file (the optione "ALL" is not available in Hypermesh interface) in order to get the _sizing.#.fem file at each iteration. Now this part is solved. Please let me know if a script able to calculate plies area from .fem file is available (if able to calculate elements area, it could be then modified to sum areas of elements related to the same plies). Regards, Alessandro
  • Create New...