5 Replies Latest reply on Mar 1, 2009 1:04 PM by Bill McEachern


    Sherif Saad
      I have an assembly of 9 parts and 16 bolt the computer takes more than 10 hours and it only achieve 20% in solving contact constraints. the contact is bonded and no penetration.
        • Contact
          Anna Wood
          What version and service pack of SolidWorks are you using?

          What are your computer specs?

          What is your question?

          • Contact
            Sherif Saad
            I have a cor 2 duo with 2 giga ram , windows vista SP1, SW2008 SP 5
            my problem is that it take more than 8 hours and it never continue only 20 % on solving contact constrains. I know there is something wrong with the contact but I don't know it. I tried each two parts only and it takes no time but when i try all of them together it never works. I don't know what is the problem the mesh is ok and restrains is ok but the contact is not. If any one have any advise I will be so greatfull.
              • Contact
                Bill McEachern
                In my opion something si moving. Run a very corse mesh with draft elements, use the FFE solver and show the convergence plot. IF it is not running nicely down hill it is typically a good indication that something is moving.

                I would attached an elastic support to every part or ones that you may have a suspicion may have unrestricted DOF's. Just use some low stiffness relative to the problem. This will get you a solution eventually and then you can see what is moving. Remember if things are not strictly conincident they will not get the default bonded connection or whatever is specificed as the default if not a bonded.

                Alternatively, you can set the default to free and manually apply all contacts pretty easily usig the select contacts option.

                In my experience, this does not look like a software issue but a restraint issue. FEA is a stiffness based method so things need tohave some stiffness or things don't converge very quickly or not at all.

                Also in the solver options check the apply soft springs and this, while a bit slow, will usually get you an answer so you can see what is moving.

                Also the contact iterations are not vectorized so you would see low CPU utilization if it is spending a lot of time on the contacts. The solvers are vectorized for the stiffness matrix inversion.

                Good luck.
              • Contact
                If you change no penetration to bonded, or use bonded as global contact, it will be faster,
                • Contact
                  Sherif Saad
                  I have done that but it is not the case.