Quasi-static vs non quasi-static in pretension models

Fatmac

I have a few questions about pretension analysis and the differences between quasi-static and non-quasi-static approaches. Rightly or wrongly, I feel that applying the full bolt pretension before the structure is loaded or deformed is important. It reflects what is most likely to occur in real life, and I have the impression that if full pretension is not achieved before surfaces begin to slide relative to each other, the results may become unrealistic (I imagine that in most cases this would lead to increased deformation). For this reason, I normally use a quasi-static nonlinear analysis and apply the pretension at the very beginning of the simulation. If possible, I avoid ramping up the pretension (whether it is applied directly or simulated by cooling the fastener) and only start ramping the external load or deformation afterwards. My first question is: if I use pretension in a single-step analysis, should I use displacement rather than load? Presumably, in a single-step analysis (which is what I normally use), if a pretension load is applied, the solver will try to keep it constant throughout the analysis. My workaround is to use displacement and, through trial and error, obtain the correct bolt tension before the structure is loaded. This is somewhat tedious, which is why I often shrink the bolt instead and keep it at the same temperature for the entire analysis. Am I misunderstanding how pretension loads work? If not, is there another workaround (other than introducing an additional step) to avoid the iterative process of achieving the correct bolt pretension? My next question is why a non-quasi-static nonlinear analysis seems to reach a solution much faster. In the example I just ran (in the metal forming discussion), it took only about 20% of the time. In some cases, no matter how I adjust the ramping, I cannot get a quasi-static analysis to converge, whereas a non-quasi-static one converges without difficulty. I had imagined that stepping through the analysis with smaller increments might help convergence, but often each small step in the quasi-static analysis takes longer than completing the entire simulation in one go with a non-quasi-static approach. Is there a difference in the analysis itself between simple nonlinear analysis and quasi-static nonlinear analysis?
My last question is if a pretension type analysis is carried out with no ramping in a simple non quasi static analysis and the pretension (particularly if using load) and load/deformation is applied together at the same time (although I guess there is no time concept in this type of analysis) is the result potentially flawed?

I would be interested in hearing others' thoughts on this topic.

Victor

Yes, the displacement pre-tension is probably the right one and it does have that problem of having to decide the displacement. One way is to first solve with the Force option, measure the resulting displacement, then set that as the Length adjustment. Hopefully that only takes one extra solve, instead of hunting for it.

Non-quasi-static nonlinear doesn't actually exist in CCX. It's just an abstraction in Mecway that hides the time steps. But it does several things that might contribute to faster convergence:

• No AMPLITUDE=STEP on the *STEP card, so loads are ramped automatically.
  
• If there's elastic contact, then no extra contact step, so contact stiffness might get initially reduced to help convergence.
  
• No DIRECT on the *STATIC card, so it always has automatic time stepping.
  

It sounds like you might not be using automatic time stepping so maybe that's the main cause of the slower solving?