High horizontal reactions in pot bearings on a skewed multi girder deck bridge

[schvetz]

Hi all,

I am currently assessing an existing composite steel girder-concrete deck bridge due to anticipated ground movements causing the abutments to tilt inwards.

The bridge has a span of approximately 39m and a skew of approximately 34 degrees - a fairly standard bridge. East end is the "fixed" end with pot bearings released in the transversal direction only and 1no. fixed pot bearing, and West end is the "free" end with free pot bearings and 1no. sliding pot bearing in the longitudinal direction (see attached articulation diagram).

However, when modelling the structure as a 3D plate and beam finite element model (as recommended in SCI P356) I get unrealistically high horizontal ULS reactions on the "fixed" end, before any ground movements are even applied, which pretty much mean the bearings are theoretically failing by up to 200%, based on UK assessment standards CS454. This is obviously not true and inspection reports do not flag up any bearing defects or damage.

The standard response to this is that this is probably a finite element modelling issue and engineering judgement says that the way the bridge was originally designed, with the articulation diagram attached, it was never intended to take such high horizontal reactions, other than the standard longitudinal braking loads and other accidental lateral loads.

On the other hand, no one has been able to provide an explanation as to why this is occurring, with 3no. different finite element models from 3no. different softwares highlighting the same issue.

The bearings have been modelled at their exact location i.e. bottom flange, so if this was modelled with beam elements they would've been modelled with offsets. I found this offset to be one of the reasons for the high reactions. The model is also staged to take into account construction sequence and locked in stresses after the wet concrete from the deck starts acting composite with the girders.

Has anyone ever come across this, or would someone have a plausible explanation for this? See attached bridge model and articulation diagram.

Thanks in advance!

 

[SlideRuleEra]

I plotted the software output. The line connecting the points has the look (more or less) of some out-of-place mathematical function... which indicates to me that the most likely cause is a software modeling issue. I know first hand that skewed bridge decks can behave in strange ways, but cannot come up with a reason for those results on an actual skewed bridge.

 

[Lomarandil]

I similarly can't come up with a reason why these reactions would exist, and suspect the software. Will be curious to hear what turns up.

 

[3DDave]

What does " 3no." translate to when not abbreviated?

 

[3DDave]

If I understand the diagram correctly, because the degrees of freedom on the fixed bearings are not co-axial, any distortion will produce a large amount of leverage. The bearings should match the skew of the beams to eliminate that.

 

[bugbus]

"3no." means "3 number [of]", to differentiate it from a unit of measurement or dimensions, basically to avoid confusion. Pretty commonly seen in civil/structural drawings etc.

 

[3DDave]

Ah, OK. QTY is out of favor? At least it doesn't affect the observation about fixity being a possible source of trouble.

 

[bugbus]

Could be a regional thing. In Australia, "no." and sometimes "off" is quite common, don't often see "qty".

 

[schvetz]

Thanks all, in case you're interested I decided to run another model allowing for a +/- 0.5mm gap from the guided bearings (from their record drawings), with "zero-gap" elements in Strand7, and although it improves the situation some of the bearings still theoretically fail just on self-weight + superimposed dead load - see attached graph.
Not something I want to convey to the asset owner!!
Would a grillage model be more adequate for this type of bridge? The 3D plate model results simply do not seem realistic. Bear in mind I will have to apply in-plane ground movements to this model as well.

 

[SlideRuleEra]

Would a grillage model be more adequate for this type of bridge?

IMHO, starting over with a simpler 2D grillage model is a real good idea:

1) Should be accurate enough to get needed info.

2) By starting fresh, perhaps whatever "errors" or factors that currently produce the weird results will be avoided.

Since the bridge is relatively high-skew (34[sup]o[/sup]), orthogonal mesh is recommended.

 

[triops1996]

schvetz,
I recently attended a webinar that discussed a similar situation. They were designing a new curved steel girder bridge and getting very high bearing lateral loads.

Their solution was simply to design the bridge to have free exterior bearings and keep the interior fixed. This dropped their loads considerably to something reasonable. I forget how they also verified the reactions, but I recall they did something that gave them confidence in the new lower reactions.

Doesn't help in your case since you're analyzing an existing bridge, so sorry there.

I'm curious if you free up those exterior bearings, do you get reasonable loads?

Best,