FE modelling of bolted connections 1

[izax1]

I have always struggled with realistic modelling of bolted connections in dynamics (modal) analysis. Is there a smart method for getting both the correct stiffness of the connection and the response forces in the bolts from a modal response analysis??

 

[tld23]

In modal analyses there are two main issues (1) MASS and (2) STIFFNESS.

If your bolt, rivet etc is heavy compared to the structure you may need to add a mass to account for this. However mostly the mass of fasteners is small.

More significant is stiffness and with this there can be a problem. Fastened joints use contact forces to maintain the joint. In general contact is a non-linear problem, ie the stiffness may change with load. Now modal analyses are by definition linear and cannot accommodate nonlinearity. Hence in theory it is not possible to do a modal analysis of a fastened component. However in reality the non-linearity may be small and it reduces to a simplifying assumption.

There are many types of fastened joints and that I know of there are no general rules on representing them in FEA models. The approach I would take is to experimentally measure the stiffness of a real joint and build this stiffness in to the model. You may need a model of the experiment to match the model joint stiffnesses. Alternatively build a sub-model of the joint - with fastener and all.

Remember the following:
1) Most fastened joints are overlap joints and
a)the extra stiffness of the overlap
b)and mass
c)and offset should be included
2) Consider potential contact nonlinearity

TERRY

 

[izax1]

Thanks for your input.

Mostly I can model the bolted connection as linear with an equivalent stiffness. (mostly join bolted parts together). That will give me the correct global stiffness. But my problem is: How do I get my correct bolt forces?? They will depend on my local connection stiffness including bolt preload which is, as you say, non-linear. How do you capture that in a modal analysis. How can I estimate the fatigue life of my bolt from a random vibration loading? That is the trick I am looking for!!

 

[pjhype]

bernt,

A modal analysis will not give you forces. It provides you only with the solution of the eigenvalue problem concerning the stiffness and mass of the structure. Unless you are doing a direct frequency response analysis, the usual FE solution scheme uses the modal analysis as a pre-cursor to dynamic analysis (i.e. modal frequency response), and then post-processes the frequency response results for the random analysis.

Many systems behave like narrow band oscillators when subjected to broad band random excitation. If you believe this to be the case with your system, then you can approximate the RMS acceleration of the system c.g. by the following:

a-RMS := sqrt((pi*f*Q*W)/2)

where f is the resonant frequency, Q is the amplification factor, and W is the input power spectral density (at the resonant frequency). You can verify the narrow band behavior of your model and calculate Q by performing a frequency response analysis of your model.

Once you calculate the RMS acceleration, convert it to a peak acceleration and apply this acceleration as a body load to your model in a static analysis. Calculate the strains in the bolts, then use the strains in your strain-life equation to calculate the fatigue life.

pj