Unexpected deformed shape for a curved bridge

[fakharifar.mostafa]

I have modeled a curved bridge. Link to the model and problem schematic are as below.

https://www.dropbox.com/s/xfk2kq1t7pw4d ... -Trial.spf

https://www.dropbox.com/s/dzrnndgrf7gel ... ration.jpg

There is a link element ("BaseRigidArm-PierTop") between the pier top and rigid elastic arm ("ElasticRigidArm-PierTopDeckCente" connecting the deck C.G. to pier top.

This curved bridge has been giving me a lot of trouble. It seems when model is not coincident with the reference X-Y-Z axes, definition of link elements and support conditions get somehow messed up in SStruct.

My left support is restrained "x+z+ry+rz" and I still get displacement in X-direction. I have changed all of the parameters for the link element (F1,F2,F3,M1,M2,M3). unless I ignore the link element, there is always separation between the pier top and the deck. Is it a graphical flaw? Should I ignore this? Please advise.

In this coordinate system for this model, I also can't apply acceleration at the extreme abutment supports. Thanks.
Thanks.

[huffte]

I ran the model and I get no x-displacement at the abutments. I was also able to apply the ground acceleration in the x-direction at the abutments.

The links have a fairly soft F3 stiffness with rotational stiffness values very small, hence the separation it seems. I am not sure why that is surprising given the soft link spring and the 1g+ ground acceleration.

Perhaps I have missed something.

[fakharifar.mostafa]

Thanks much Tim.

I applied this arbitrary amplified record to make sure bridge is in the nonlinear range and see the link element works.
I needed to be more specific, the local axes that defines the link elements are in fashion that the F3 of the link element is in the Y direction (bridge longitudinal direction). This link is supposed to represent the seating bearing type and according to the values assigned for the link, it should move in Y direction during analysis. However, since the bridge is curved, thus the local axis 3 of the link element is deviated from the Global Y axis, and I guess that's why I got a different deformation as the one intended. Am I right?

[huffte]

Probably so. Note that you can specify the rotation angle of a link by explicitly entering the angle. You should be able to calculate the angle which would make either the 2 or 3 local link axis tangent to the curve at the substructure in question. I think this might help greatly in alleviating the problem.

It may also be more realistic to specify approximate actual rotational stiffnes values for the link instead of using very small values.

Interesting model. Best of luck fakharifar.mostafa.

[fakharifar.mostafa]

Thanks Tim.