Finite Element Modeling of Continuous Span Steel I-girder bridge

[Jhabindra]

Dear All,
I am new member in this discussion group. I have one question. I want to develop the FE model of the continuous span Steel I-girder bride which has 5 Spans with four continuous support in between the two abutments. The total length of bridge is 184 meter.There is one modular type expansion joint near the abutment. My final objective will be to predict the noise generation(Acoustics) by the vibration of concrete deck and steel girders by using this FE model and its modal properties.I think It will be not so reasonable to model whole bridge since I am interested only in one edge span with the expansion joint.And there is node no limitation in ANSYS. I need to consider finer mesh to analyse in higher frequency range.So my question is: Is it possible to model only one span of continuous type bridge? How can I give the boundary condition at the continuous edge of the bridge if I take only one span. Or is it possible to get real behavior of the bridge by taking only one span.

Your suggession will be highly appreciated.

 

[corus]

In general it's good practice in FEA to assume symmetry where ever possible to reduce the model size. With modal analysis, however, symmetry doesn't exist (if you want to include all possible mode shapes). I'd get a FE package that allows a greater number of nodes.

corus

 

[GregLocock]

I'd add that noise prediction is one of the most difficult things to do with an FEA package.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[rb1957]

maybe you could simplify the model by doing a coarse model of the whole structure and a fine mesh model of one bay, loaded by the reactions from the coarse grid model.

this works for statics, may not work for acoustics.

maybe you could use superelements (NASTRAN term, ANSYS will call it something different) to model the remote part of the bridge, and a detail model for the span in question.

again, not my area of expertise, but do you need a fine mesh for acoustic analysis ... you may need to model small details of the structure

good luck

 

[johnhors]

As corus said, you can't reduce your model without losing mode shapes (at the lower frequency end), but modal analysis only needs to be accurate for mass and stiffness. Thus you can have a much coarser mesh than that required for stress analysis, but still have a good representation for your purpose. Higher frequency modes will tend to be very localised in most structures with single geometric features resonating whilst the rest of the structure undergoes little or no movement. It is very unlikely that a high frequency mode will involve the whole structure (other than for repetitions of the same localised spot within each span). For this reason it is not unreasonable to model a single span for the higher modes, so long as you are aware that lower modes are being omitted. But as Greg pointed out, turning modal results into a noise prediction is rather challenging.

 

[Dinosaur]

This must be an academic problem since I believe the vehicle noise of the engine, exhaust system and tire noise have to be much greater than the "sound" emanating from the bridge. Earthquake analysis used to check bridge designs show that several modes are below the threashold of hearing, and I assume quickly pass through the audible spectrum, before becoming inaudible on the other side. These higher frequencies don't have much energy either. If this is a serious analysis, I would put the effort in analyzing the noise generated by the vehicles. If there are cross frame or lateral bracing members that rattle, that should be addressed through other means.