Bolt Preload, Load Sequence

[lardj]

I am running a model with 2 bolt preloads created using the beam elements and 1 external load.  Does the model apply the preload first in 1 load step and then the external load or are they applied during the same load step?  Is it possible to apply them in different load steps for more real world accuracy?

I am concerned about the accuracy of the bolted joint results because I'm seeing results which are counter intuitive.  At a preload of 38.6 kN my joint seemed to slip, while at 30.9 kN it did not slip.  It has me questioning in what sequence the preload is applied and how accurate the bolt beam element method is.

Thanks.

[S.LI]

Pre-load is there at the beginning (the first step). The external load could be applied after that (starting from the second step) by adjusting load curve or active period, if needed.

About the difference between two of your examples, I'm not sure the reason before seeing more details on models. 

Please consider to upload a simple model with similar issues.

Is your pre-load above yield stress?

[S.LI]

About pre-load, I'm talking about the one in MES, instead of Linear static.

[qzhu]

The pre-load, as its name suggests, is applied before the analysis starts (i.e. before time 0). Other loads are applied at user definition during analysis (i.e. when time >0). In my opinion, it is mainly designed for elastic beams. For elastoplastic model, one has to make assumptions about how the preload is applied, the simplest of which is to assume it loads monotonically to the final (preloading) state.

[lardj]

Just to clarify then, the preload is actually applied prior to the external loads, correct?  So I shouldn't have to mess MES then or load curves.

[qzhu]

Yes, the load curves will have no effect on preloads. Preloads are loads that a part (or element) is subject to under some unspecified constraints so that it will not affect connecting parts (elements). At time 0, the constraints are lifted instantaneously and thus the stress caused by preloads start to affect its connecting parts (elements). One thing to note is that strain at time 0 is assumed to be zero since reference for strain measurement is generablly arbitrary. It is known that sometimes the preloads will cause convergence difficulty for nonlinear cases when solver has difficulty to absorb the huge stress released instantaneously. In such cases, and in situations where you want to have more precise control on the releasing (or applying) of preloads, you may consider explicitly simulate the preloading.