Is this to be expected?
I have assemblies with contact in which I have defined localcontact conditions for all of the faces that interact or arebonded.
Almost by accident I found that the models mesh differently (andrun differently) depending on whether the Global Contact Conditionis set to Free vs. Bonded or No Penetration and that they also meshdifferently whether I mesh in SW2008 SP3.0 on X32 or X64.
I care because in some of my more complex models this is enoughdifference to make the model either run or not run at all (i.e. itsits at the first contact iteration forever trying to determinenumber of contact points).
To test this I loaded one of the tutorial models "Nl_Pipe_Holder"and meshed it both ways.
In SW2008 SP3.0 on X64, with Global set to Bonded and a defaultelement size/tolerance of .12"/.006" I got 3619 elements.
With Global set to Free I got 3594 elements.
To make it worse, the model meshes differently when run an an XP32machine where Bonded results in 3595 elements and Free gives 3597.(I've already reported this problem earlier and it was assigned SPR435168)
From the attached screen dumps you can see that the contact surfaceof the "pipe" has meshed differently at the spot the two partsinitially touch, even though a Local Contact Condition alreadyspecifies these faces as No Penetration.
According to the documentation the local conditions should overrideglobal conditions, so why does the global setting make a differencehere?
When you run the analysis it is obvious that the No Penetrationcondition has won out as the parts can slide over each other.
-MikeSolidworksSimulation
I have assemblies with contact in which I have defined localcontact conditions for all of the faces that interact or arebonded.
Almost by accident I found that the models mesh differently (andrun differently) depending on whether the Global Contact Conditionis set to Free vs. Bonded or No Penetration and that they also meshdifferently whether I mesh in SW2008 SP3.0 on X32 or X64.
I care because in some of my more complex models this is enoughdifference to make the model either run or not run at all (i.e. itsits at the first contact iteration forever trying to determinenumber of contact points).
To test this I loaded one of the tutorial models "Nl_Pipe_Holder"and meshed it both ways.
In SW2008 SP3.0 on X64, with Global set to Bonded and a defaultelement size/tolerance of .12"/.006" I got 3619 elements.
With Global set to Free I got 3594 elements.
To make it worse, the model meshes differently when run an an XP32machine where Bonded results in 3595 elements and Free gives 3597.(I've already reported this problem earlier and it was assigned SPR435168)
From the attached screen dumps you can see that the contact surfaceof the "pipe" has meshed differently at the spot the two partsinitially touch, even though a Local Contact Condition alreadyspecifies these faces as No Penetration.
According to the documentation the local conditions should overrideglobal conditions, so why does the global setting make a differencehere?
When you run the analysis it is obvious that the No Penetrationcondition has won out as the parts can slide over each other.
-MikeSolidworksSimulation
