Snap-fit analysis: CNTSTB SCALE/S1

Francesco Porcari

Hello everybody,

today I'm studying a snap-fit analysis. The model is linked at the end of the post. 
I'm trying to change some setting parameters and observe the behavior of the system. In particular, I focus on SCALE, S1 in CNTSTB load collector. 

I don't understand how to set SCALE and S1 for obtain correct results, because in some online examples (like OS-E: 0155) SCALE=0.01 and S1 = 0, but in others (like this: snap-fit.pdf (hubspot.net)) SCALE=1 and S1=0.0001. 

The problem is that, varying SCALE, I get different results in the force insertion (SPCF along Z at the center of rbe2 at the snap's base). Furthermore, varying SCALE and S1, the load step can converge or not depending on their numerical values.

I imagine that SCALE influences the contact stiffnes, so the force insertion. S1, on the other side, is only a parameter for stabilize contact at the end of simulation.

Can anyone explain me how to choose correct numerical values for obtain consistent results?

Thanks

 

Francesco

 

 

asm_snap3.hm

Adriano_Koga

Hi,

 

first of all, please, take a look at the descripton at the help for the CNTSTB (contact stabilization) card.

SCALE is the global scale factor for the contact stiffness stabilization factor. S0 and S1 are the initial and final scalar factors inside the f(t*) function, meaning that they're also multiply the final stiffness value.

The basic idea of CNTSTB is to stabilize the contacts numerically, by adding some stiffness/damping to the parts, avoiding them to "fly away" at the beginning of your analysis, as usually contacts are not closed yet.

BUT stabilization techniques should be avoided when possible, or used with caution, to avoid non-physical behavior, or some small changes might occur. You can monitor the stabilization energy in the end of your analysis to understand how much of stabilization energy was added into your problem. Usually we would seek for the minimum energy as possible, that still can help or problem to converge.

So first, I'd run without  stabilization. IF the problem struggles to converge, and I'm confident of the BCs, i'd add some stabilization, trying to use it the minimum possible.