I am working on the friction between steel plates and wooden blocks. The model is made of two large rectangular steel plates separated by numerous small cubical wooden blocks. The blocks are fixed to one plate, the wood/steel friction is studied on the opposite plate. One connector per wooden block has been created, and of course the regions that come with the connectors were defined too. The contact property is the same for every connector.
Displacements are expected to be very small compared to the size of the plates, but not that small compared to the wooden blocks (and its mesh size). The wood and the steel are defined as linear materials. Finally, a few blocks are expected not to be touching the steel plate they face, which means a few connectors are opened at the end of the analysis.
My question is about the solver choice. Is this model suited for a Linear static analysis SOL101 ? Or does the unknown state of some contactors make it necessary to use the Non-Linear Static analysis SOL601 ?
The results I acquired rather support the SOL601, but I would like to understand why.
Sol101 is suitable for initially touching or inter-penetrating contact simulation. If you are concern about the opening gap between the possibly touching surface, then this is out of the realm of Sol101, and I would recommend using Sol601.
Thanks for your answer ! However, according to the NX User's Guide, pages 617-618, the linear solution 101 allows contact elements to be active or inactive (no contact), the contact status of a region is determined during the outer loop of the algorithm. What makes you say SOL101 can't handle contact openings ?
The User's guide clearly describes the contact algorithm for SOL 101, but the official documentation says nothing about the SOL 601 contact algorithm, even in the NL guides. All I could find on the internet is an excerpt from a book about non-linear simulation, which details the non-linear contact algorithm (pages 12-13). And from what I understood it works the same way as the linear one...
I think the paragraph in the manual is illustrating about how to solve penetration problem, which requires iteration loop. I do agree Sol101 can solve penetration problem in my previous reply.
What I meant was your expectation of the contact simulation. If you expect to see the contact definition produce a gap (from close to open gap), then not in Sol101. Just my experience...
If you would like to read more about Sol601, you should check on NX Advanced Nonlinear Theory and Modeling Guide, not NL Guide. The NL guide is covering Sol106 by the way.
I'm glad you answered because Experience is what I need !
My results from SOL 101 show some open contact regions, with 0 contact forces output.
The SOL 601 give roughly the same contact status for those regions, that is why I have trouble admitting SOL 101 can't handle open contact as you explained. Moreover, the following sentence from the User's Guide proves SOL101 can do with open contact regions: "Once the inner loop is converged, the status of all contact elements is determined in the outer loop. A contact element is set inactive and removed from the global equation if a tensile traction is required to close the contact element"
Can you describe the model(s) which lead you to this conclusion ?
Sorry, by NL guides I meant Advanced NL Guide and NL Guide. Neither of these refer to outer or inner loops of the contact algorithm used in Non-linear solutions...
Using the contact force output to check for contact status is what I neglected, and yes this can be done for both solution 101 and 601. What i was thinking is if simulation animation is used to check for the contact change. Sol101 cannot do animation to check contact status, but Sol601 can.
So I am thinking back on your original question, how to choose between these 2 solutions. The main differentiator between these 2 solutions is nonlinearities. There are 3 types of nonlinearities, namely material, geometrical, and boundary. In your case, the nonlinearity of concern is boundary. This is to say the boundary of your model is changing with time. For example, at time 0 second, the boundary is stationary (model at rest); then at time 5 second, boundary is subjected to external force due to contact. Sol601 allows you to see the contact change according to time. Sol101 neglects the time factor.
Your explanation is very clear. The loads applied on my model are not time dependent, yet time effects in Sol601 can be observed using Auto Time Stepping - Total Load, in which the solver automatically applies the load through a ramp time function, and increases or reduces the time step size depending on how well the solution converges. Time becomes a dummy variable which is used to indicate different load levels in this incremental load analysis.
So if time is removed from the Sol601 analysis (no Auto Time Stepping), and the loads are applied within one time step, what difference remains between the Linear and the NonLinear analysis ?
SOL 101 contact is linear in the sense that the contact pairing generated once and never updated and the orientation of the contact elements is not updated.
The lack of pairing update means that results may be invalid if large relative sliding occurrs. Models where bodies are initially far apart (distance greater than the contact search distance) cannot be solved.
The lack of orientation update means that large displacement/rotations of the model can yield invalid results.
Within these constraints, the actual contact solution itself is a "true" nonlinear process. The status of the existing contact elements is updated in the outer loop. Any contact element can switch between active (closed) and inactive (open) in any outer loop iteration. The forces and penetrations are adjusted in the inner loop.
Contact elements on the periphery of the contact area may switch between open and closed status several times before the process converges. The statement that SOL 101 contact cannot produce an open final status is simply incorrect.
Thanks for your answer. I've tried to run the SOL 601 analysis with the parameter Iterations for Pairing = 1 (NSUPP=1 in the NXSTRAT entry). So the contact pairs are created at the first contact iteration, just like the SOL 101.
The results don't change much for SOL 601, they are very, very close to the previous 601 anlysis, and they remain quite different from the SOL 101 ones.
However, this pairing parameter proves one thing : no large relative sliding occurs, no large displ/rotation is computed, so there is no need for for pairing and orientation update.
So which analysis gives the better results ? And why is the other one so different from the "good" results ?