A Bridge Load Rating and Theory vs. Reality 1

[anominal]

The topic structure is a RR bridge that has been in service for 100+ years. The bridge is best described as a continuous ballasted concrete deck slab (please see attached). Until recently the controlling normal rating for the bridge had been acceptable to the owner. In 2010 a new firm performed a finer analysis of the bridge, cutting sections and calculating stresses at locations not considered before.

The controlling normal rating suffered. The owner was not pleased. Their "O.K. for normal use" bridge turned into a "Maybe you shouldn't run those ballast cars..." bridge.

This most recent load rating and analysis uses a number of simplifying assumptions that are in agreement with AREMA. It models LL effects using a suitable computer program. It calculates stresses at appropriate locations. In all respects, the new load rating and analysis seems theoretically correct.

However, it does not match reality.

Engineers who have worked with a contractor have probably heard this before: "It ain't goin' nowhere". This bridge ain't goin'. Aside from deficiencies addressed in a rehabilitation during the 90s, the bridge has purportedly had few maintenance concerns.

My task is to reconcile the new unsatisfactory load rating (from a respectable firm that performed a reasonable analysis) with the observed satisfactory condition of a bridge that has been in service for decades. Where to begin...

 

[JedClampett]

I'm no bridge nor railway engineer, but it seems that using a simplified analysis for derating a bridge is not sufficient. If a more complete analysis shows overstresses than that's when you need to break the bad news to the owner.
On the other hand, if a simplified conservative analysis showed the bridge was OK, that would be fine. Although that didn't work out too well in Minneapolis.
What I don't understand is why the analysis is using some more refined portions (previously unanalyzed sections) on one hand, yet simplifying the analysis per AREMA on the other hand.
If it's a budget thing, the owner needs to cut loose more funds to complete the analysis.

 

[Qshake]

Well, I'd say that you've got quite a task on hand. This probably doesn't go to your overall question but you should align similar portions of the simplified analysis with those of the more sophisticated analysis and compare/contrast the assumptions, results, behavior as modeled and as in practice. You should be able to show, in the end, what's different and whether it's good, bad, conservative, or not.

Now as to the overall question of why structures don't often perform as analyzed - here's a simplified example but one often used in our realm.

In many cases, especially here in the USA, bridges are designed and rated using a "line girder" assumption. That is a single girder with it's tributary area and portion of live load is used to design or rate the structure. Before LRFD, this sort of analysis was far more conservative. LRFD has taken in some of the considerations of load distribution to superstructure elements and overall continuity in the cross-section of the bridge. But a structure doesn't behave as a single line girder element. Nope, far from it. It behaves more as a plate that is stiffened with some type of longitudinal elements, steel girders or prestressed concrete girders or the old concrete t-beams. Of course, the wider teh spacing between these longitudinal girders and the wider the overall deck the more it will act like singular beam elements. But that's hardly the case, ever, especially with RR bridges.

I've noted that this is mainly in the USA. It is apparently more common to use a grillage or plate representation in the UK. At least that's been my experience.

Another example of why things don't behave as analyzed is load distribution. We assume simple means of distribution that are convenient for analysis but may not reflect reality. I once analyzed an old (late 1800) slab on girder system where the giders were framed into floor beams. According to my calculations at one particular floor beam the combination of live load, dead load and rust meant imminient failure of the floor beam. So I had the owner close a portion of the structure while this floor beam could be replaced. It turns out that the slab, which had large haunches to the stringers, weren't actually always bearing on the stringers and that the slab, which was thick, was redistributing it's load elsewhere.

Sometimes the difference is in the way we analyze structures and sometimes the difference is in the way the structure actually performs. We do try to build the structures to perform the way we design them, but in the case of the late 1800 structure, the maintenance and upkeep just wasn't there to ensure the actual performance kept up.

I hope that helps. Good luck.

Regards,
Qshake

Eng-Tips Forums:Real Solutions for Real Problems Really Quick.

 

[bridgebuster]

SSVK - there's a blast from the past. I thought the Peck Bridge was replaced?

 

[paddingtongreen]

Check the deflection under various high loads, then backfit them into actual stresses and strains. Then blame the simplifying assumptions for the difference.

You will have to send the loaded car in slowly or you won't see it in the deflection.

Michael.
Timing has a lot to do with the outcome of a rain dance.

 

[hokie66]

Agree with Michael. Not a bridge engineer myself, but verifying structures by load testing is a long accepted approach. Trying to analyze structures built 100 years ago by using modern methods is often not logical, as the materials were so different.

 

[anominal]

Qshake: you nailed it by describing the limitations of using a line girder assumption. This slab is being analyzed in 1' longitudinal strips. The lateral live load distribution is another simplifying assumption that clouds the water. In accordance with AREMA, the lateral live load distribution is limited to the spacing of track centers (12').

My first exercise is reinterpreting the code. Buying into uniform LL distribution between the outermost track centers and considering multiple track loading will reduce LL effects. It may alter the normal rating enough to make this go away - without condemning another firms work.

While I don't think that anyone will be too keen on calculating stresses from strains gages, I'll keep it in my back pocket. Would that type of study be expensive?

 

[anominal]

bridgebuster:
I think you're right about the Peck Bridge. This is the Bridgeport Viaduct, which leads up to the Peck.

 

[bridgebuster]

thanks

 

[ACtrafficengr]

This might be an approach you could take:

NYSDOT had a case where a concrete T-beam bridge had a low load posting due to missing record plans. Without knowing what reinforcing steel was in the bridge, they load-rated it conservatively.

When the 20T posting became a problem for truck movement in the area, they instrumented the bridge and measured the deflections while driving a truck of known weight over it, and satisfied themselves that they could raise the load posting.



"...students of traffic are beginning to realize the false economy of mechanically controlled traffic, and hand work by trained officers will again prevail." - Wm. Phelps Eno, ca. 1928

 

[paddingtongreen]

@anominal, You don't need strain gauges, you are measuring deflections. They have systems that use lasers to measure deflections. I don't know the details but the bridge inspection department of the last place I worked had the capability. From the deflections you can calculate curvature due to that load, and from there, the strain in the extreme fibers and from there, the stresses.

I imagine you could put a target on the side of a bridge, aim a laser on the center and then load it up and read the difference.

I forgot, you have those abandoned piers there to use as fixed points. You should be able to accurately measure the distance between the old pier and the girder, both loaded and unloaded (the bridge loaded, not you).

Michael.
Timing has a lot to do with the outcome of a rain dance.

 

[VoyageofDiscovery]

First thing that strikes me of the original design is that there is a lot of ballast, second the number of spans compared to the abandoned piers.

The deep ballast reduces the LL pressure. Have you taken this into account?

Here in Canada for slab railway bridges we distribute loading at 45 degrees through the slab to main bottom reinforcing.

VoD

 

[VoyageofDiscovery]

Beyond all this, though arching is not considered in design in AREMA, this may well be occurring with the over 6ft of ballast depth. Finding a geotech willing to give you arching values for this will be a problem though.

VoD

 

[wiktor]

As Qshake presented the main topic, I will just add some to it.
The key issue with this type of bridge is determination of the effective width of the slab, but with 4 tracks this should be non issue.
Calculations of the forces in different sections also should not affect the rating. I do believe that the rating done previously was performed using simplified method, so more rigorous analyses, such as calculation of the moments using computer program shall increase the rating, not opposite.
The bridge was designed for Cooper 60 loading, and until major changes have occurred, such as section loss of the reinforcement or deterioration of the concrete, the rating shall be unaffected.

In my opinion the following should be verified:
1. Controlling forces used in the rating. Perform simple manual sanity check of the moments used in the rating, for both positive and negative moments. If discrepancy exceeds 10% use manually computed moments. Couple of popular structural programs have major problem with moving loads for multiple spans, and results shall be viewed very carefully….
2. It is not reputable company doing the load rating. It is typically inexperienced ASCE II or III, acting without any guidance, or sanity check. So carefully review the rating calculations, and do not fill belittled by the greatness of the company. They also do make mistakes, sometimes of major consequences.

If you can post “sanitized” version of the rating calculations, including the load distribution, effective width of slab for positive and negative moments, cap beam rating calculations, and most important clearly marked location of the new rating sections.
Remove all headers, company names, and bridge and owner details.

I have a strange impression, that after practicing design and rating of bridges for more than 30 years, I will learn something new from this exercise.