15 Replies Latest reply on Feb 22, 2008 5:41 PM by Gary Garrison

    Insufficient Memory

    Gary Garrison
      I am attempting to run an analysis on an assemlby of ours. I really do wish I had more Cosmos experience, but my computer is old enough to not like Cosmos very much. I have run the /3GB Switch, I have adjusted the page file to the max allowed, and I have turned off all the WinXP GUI extras.

      When I try to mesh the assembly I always run out of memory. Is there anything else I can do?

      Here are my specs:
      P4 2.8
      2.5GB DDR-400 (not DDR2)
      /3GB switch on
      pagefile set to 4096MB (max allowed under WinXP)
      all add-in's other than Cosmos removed

      Mesh is attempted after a fresh restart, with no other programs open when I leave for othe day (it takes 2+ hours at least to error out). In task manager I am sitting at about 1.6GB the PF usage graph in use at the time of the crash. The only time I actually watched it to find out where it crashed I kept an eye on the PF USage under "Task Manager>Performance Tab" it went as high as 2.5GB but then drupped back to 1.6-1.7 GB for several minutes (15 or so) before it errored "Insufficient Memory", then when I click retry it says insufficient memory to complete task, and SW exits.

      We make wire grid products, like shelving, fencing, and warehouse pallet shelves. Since I am unable to make a flat grid out of "welded" wires and then bend it like I want to, I made a sheetmetal part, cut my grid openings, and then added bends.
      I am attaching an example picture of the grid I am attempting to run. The "sheetmetal formed grid .jpg" file is the shape of the formed grid without the cuts to form the grid. This is rigidly attached to a frame at the bottom of the V shaped bends and at both ends of the strait sections.

      The "sheetmetal formed grid 2.jpg" is the grid with the cuts to show what I am trying to mesh.

      The assembly consists of a 1-piece grid (single part), and the frame (3x0.75) rectangular tube forming a flat ladder type frame, 1 sheetmetal approximately W shaped piece (run the long axis of the frame at centerline, and a rectangular block that sits flat on the formed grid representing a piece of machinery. The frame is being only supported on the outside edges.

      The formed mesh grid is 60x80 inches.
      The frame OD is the same.
      The rectangular block (machinery) weighs 300lb that will be sitting on the grid and has a 12x30x72 inches size envelope.

      The mesh grid is basically making an airspace under the machinery. What I want is to run a gravity test to check displacement before we actually make a sample. I wish I could be more specific about what it is but at this time I am restricted on what I can discuss regarding specifics.


      Does any of this make any sense?
        • Insufficient Memory
          I assume that you are defragging the drives and the page file before running the file? Make sure to defrag the drive the files are on, not just the drive that has solidworks.
          • Insufficient Memory
            Hi Gary,

            Have you tried reducing the mesh size? or using draft mesh?

            Another approach could be model as shells and use the spot weld contact at the welded points. Since you are using shells this would reduce the mesh size.

            Regards,

            Stuart

            • Insufficient Memory
              Pete Yodis
              Gary,

              Is the design and the loads placed on this symmetric in any way? If they were. then you could half or quarter the geometry and perform the analysis on just that...
                • Insufficient Memory
                  Gary Garrison
                  I adjusted the shape of my "load" to make it symmetric. It is just a generic profile as it is. I did cut it into quarters and ren the analysis.

                  I left for the weekend and when I came in this morning I had yet another error where it caused a crash in the analysis package. The mesh applies fine now though. The TaskManager showed 1.90GB in the PF usage graph, less than the 4096MB I have allocated.

                  I will try running it again... again.

                  I have tried every trick I can think of to squeeze every bit of RAM out of this thing I can (without adding more).
                    • Insufficient Memory
                      Gary Garrison
                      I tried it again, with shell meshing and I still didn't work. This is after sitting here for well over an hour just waiting. I am thinking that without more resources (or a faster computer) I may just have to abandon this effort for now.

                      If I can get time, I may go visit my VAR to see what I am missing... I don't get a lot of COSMOS time because up until recently COSMOS couldn't run our assemblies (formed and welded wire meshes).

                      This project is a proposed product and we would like to produce some sort of virtual analysis before we move ahead.
                        • Insufficient Memory
                          Bill Reuss
                          Gary,

                          Aside from using symmetry, have you tried to simplify the model by reducing the amount of geometry you're trying to mesh? (looking at the bottom picture in your original post) I would start by removing some of the spars (L-to-R 'W' wires) and get the part to mesh. As a start, I'd remove three of every four spars and then try to mesh the part. Hopefully you will get to a point where you are not crashing your system and the part will mesh or, if it fails, you can at least look at the mesh failure diagnostics for insight into what is causing the problems.

                          Good luck.

                            • Insufficient Memory
                              Vince Adams
                              Hi Gary, have you tried using beam elements for this? I scanned the responses and didn't see this suggestion. I've done these problems in the past and had great, FAST success.

                              You'll want to get up to SP2 of 2008 for this...

                              You may want to search for threads on beam and mix meshing before attempting this. There have been several suggestions that can improve your success with this model type.

                              Vince
                                • Insufficient Memory
                                  Gary Garrison
                                  I would love to jump to SW08, but I don't have a second station to try it out on. I have to use the Sw07 unit for production, so I am not able to "learn" SW08 without a major interruption in workflow.

                                  There has been talk of getting a new computer but I don't know when that might happen.
                        • Insufficient Memory
                          Culley Parris
                          You can have both pieces of software installed on the same machine... if you have access to '08 it's probably worth the try...
                            • Insufficient Memory
                              Gary Garrison
                              Yeah, I know it's just that for me SW installs typically take longer than they should, and I just didn't have time to mess with it. I am caught up now with my projects (sh-h-h-hh...don't tell anyone) so I will look into that install tomorrow morning.
                              [/*goes to look up dual SW install procedures]
                                • Insufficient Memory
                                  When you get the error, what does it actually say? Is it something like "Looking for this size block and ony can find this size block"
                                  ?
                                  • Insufficient Memory
                                    Gary Garrison

                                    Gary Garrison wrote:

                                     

                                    Yeah, I know it's just that for me SW installs typically take longer than they should, and I just didn't have time to mess with it. I am caught up now with my projects (sh-h-h-hh...don't tell anyone) so I will look into that install tomorrow morning.

                                    [/*goes to look up dual SW install procedures]

                                    Well Tuedsay I installed SW08 in a seperate directory from SW07. It seems o.k. so far, but I didn't get too involved with it yet (I had Jury Duty Wednesday). I started the install at 8:30am from the Source Disks (sp0.0). The install and updated SP dowlload and install took until Noon. So nearly 4 hours. Is that normal? Or is it just a slow machine?


                                    The memory error I got was always "Insufficient Memory To Continue" Select Retry or Cancel to continue. Click Retry, and BANG! insufficient resources to continue Solidworks will close now. Then it's back to desktop.

                                    I will look into the Beam element analysis for my parts today.
                                • Insufficient Memory
                                  Larry Kutcher
                                  Gary,
                                  I've got bits and pieces of down time here and there right now, if you'd like, send the file over.
                                  Tell me "exactly" how you want the product to be run.
                                  If it works, great. If it doesn't well...I'm willing to give it a shot.



                                  Kutch
                                    • Insufficient Memory
                                      Gary Garrison
                                      Using the beam suggestion I am getting it to mesh what seems to be easily, albeit time consuming with my older computer.

                                      It begins the analysis run, will pop up four messages about my restraints failing to transfer. It then tells me the that it has encountered an error and cannot continue. After a bit it is then back to SW.

                                      What does that message mean? Does it have something to do with my beam elements? It also said my assembly was unstable. It is supported with a "fixed (no translation)" restraint? Is that the source of my error message?




                                      Previously I had been getting this message:

                                      "Midside nodes on 2 edges of an element are merged, element 33752. (Reducing mesh tolerance or specifying a higher number of points for jacobian check for meshing may help.)"

                                      I changed the mesh tolerance from .01875" to .01" and it seems to mesh ok now.