Modeling Lean-On Bracing for Highly Skewed Bridges

[BigRy]

I am designing a 228ft. simple span bridge, with both abutments at 36° from roadway tangent / girder centerline. Intermediate bracing is colinear per state standards for bridges with this large of a skew. Two-stage construction, with 5 and 7 girder structures for stages 1 and 2, respectively. The roadway and deck curve at the far end of the bridge. We're not positive yet, but we think the contractor can do each stage in a single pour. We are stuck with tight constraints on superstructure depth due to railroad underclearance and large costs that would be associated with raising the profile. Right now we're working with about 6.5' from top of deck to bottom of bottom flange at mid-span. Girders are all spaced at 6.25'.

We have been working with MDX grid models for final and staged configurations. Due to the geometry, we decided to use bent plate diaphragms instead of cross frames. As you might have guessed, our bracing forces are blowing up. We're getting up to 2400k-ft in our diaphragms, which results in connections so massive that the connection plates are nearly touching. And the larger we make the bracing, the larger the forces they attract. I'm aware of limitations/errors in the way MDX analyzes bracing and flange lateral bending. I've read through NCHRP Report 725, but it doesn't seem that their instructions for a modified 2D analysis can really be applied in MDX. Those instructions are very general, and I do not see a way to input a modified torsional constant. They also do not go into great detail about how exactly this modification is carried through to other MDX analyses and output. We've tried a number of different bracing designs and layouts with the regular 2D analysis. The best case we can get is to essentially eliminate as much bracing as possible, which obviously isn't realistic from a constructability standpoint.

We are now looking to investigate the use of lean-on bracing, so we can effectively eliminate many bracing forces while still providing lateral support to the flanges. The state is reluctant to adopt this sort of "new technology", so we need to make a compelling case with this structure. The problem is that I can't find a way to model the lean-on bracing. From what I understand, it could be done in Larsa or another FE program, but we have neither the time nor the budget to begin a modelling task like that. I was hoping the relationship between diaphragm size and diaphragm force would translate to individual cross frame member sizes and forces. That is not the case. The forces are independent of the member size, even for Type E which is statically indeterminate. I then tried going back to the custom bracing input, using J, I, and Wt, except I used values for a pair of angles, one near each flange. For custom braces, MDX analyzes it like a beam and outputs moments and shears. My hope was that the shear would follow the path of the stiffest element, which would be following along the girder until it reaches a diaphragm. It did not do that either. So lastly, I minimized all bracing inputs for each "lean-on" brace (i.e., I set J, I, and Wt to 0.0001). That forced MDX to transfer all the bracing loads to the nearest diaphragm. But that also includes transferring the lateral forces that should be taken by the struts. So we end up with a couple diaphragms that still have sizeable loads, not nearly 2400k-ft but still large.

Has anyone had to design lean-on bracing? If so, how did you do it? I think I'm out of ideas here.

 

[BridgeEI]

Haven't modeled the bracing before, but TxDOT has adopted it and they put on a presentation or two through the steel conference. If you look up the videos from the past steel bridge symposium you will find more helpful information.

As far as your bracing, can you stagger the braces instead of collinear?

 

[BridgeEI]

https://ftp.dot.state.tx.us/pub/txdot-info/brg/0212_webinar/romage_chambers.pdf

This is what I remember, I think the video is the same as this presentation. Hope it helps.

 

[BigRy]

Thanks. I've actually already reviewed that presentation, and unfortunately it doesn't get into the specifics of modeling. That presentation is actually why we're thinking to pitch this idea to the state. My project manager and the guy from the state went to the presentation together.

Staggered bracing is sort of on the table, but it is also incorrectly modeled in MDX. MDX doesn't properly calculate flange lateral bending stresses. The state also is not a fan of staggered bracing of course.

 

[BridgeEI]

I think using the groupi,J etc are the way to model it. I assume you have a severely skewed bridge and that combined with the span length are really controlling the design. I would adjust spacing of braces to find something more manageable and then you'll have to design around it. A 228ft simple span is a monster span especially having a structural depth that small. I bet if you increased the girder depth your forces would decrease (knowing that you're already constrained).

 

[EQguys]

BigRy,

Some of the options that I can think of which will reduce your crossframe/diaphragms forces are as follows:

1. Consider staggering the crossframes. That really helps in a 54 degree skew situation. Start the first crossframe near the support at a distance of 1.5 to 2 times the girder depth from the supports. Once you have this preliminary layout and run the numbers, if the crossframe member forces are still high, then you will have to soften the system behavior by eliminating some crossframes.
2. Switch from Grid model to a Plate and Eccentric Beam (PEB) model. Grid models in high skew situation overestimates the crossframe forces by a factor of atleast 2 to 3.

 

[BigRy]

BridgeEI,
How would I model the cross frames using the I/J/Wt approach and get the bracing behavior I'm seeking? We have tried numerous collinear cross frame layouts, all the way down to a Z-patter with bracing only in the left half of the top bay, one continuous line of bracing full-width near mid-span of the structure, and bracing in the right half of the last bay. That produces the lowest bracing forces, but they're still very high. Increasing girder depth even a couple inches is not an option. That would be a major design change that would only be allowed as an absolute last resort.

EQguys,
1. Staggering is a potential option, but it's one that the state would really need some persuasion to adopt. We think we have a better chance with this lean-on bracing concept, which is why we're trying to model it now.
2. My coworker out of our Pittsburgh office had initially investigated the FE model and said it doesn't work with our geometry. Yesterday, out of desperation, I tried it myself. It appears to work fine. I just don't think he's familiar with all the nuances of developing a layout in MDX. MDX is so damn finicky about the geometry sometimes. You need to know how to play the game to get it to work. Anyway, initial runs seem to be showing lower bracing forces. I'm now changing the FE input to match exactly to the grid model input to do an apples to apples comparison. I will let you know how it goes.

 

[BigRy]

I guess I jumped the gun a little on the lower bracing loads with FEM. I only modeled Stage 1 for sake of time. Compared to the grid model, the max bracing moment reduced 8.12% and the max bracing shear reduced 13.98%. Overall, bracing moments reduced 22.23% on average and bracing shears reduced 16.34% on average. This is better, no doubt, but we're still working with a 909k-ft max bracing moment. This is also only Stage 1. The full configuration model results in larger loads.

 

[BridgeEI]

You may be able to create a bunch of groups and your condition items will have a list of TOP CHORD L3x3x0.5 FOR GROUP (whatever #). Same for bottom chord. You may be able to assign several TYPE A BRACING (or whatever) and assign dummy values to I/j/wt and check your results to make sure it's doing what you think it should be. Just a few ideas.

 

[1ManWolfpack]

BigRy,

Have you looked at the AASTHO-NSBA collaboration standards? Link

I haven't gone through it too much, but I think g-13.1 may have some information that would be useful to you. Link
Hope this helps some.

 

[BigRy]

BridgeEI,
I'm not sure I understand what you are saying. Are you saying to try using preset (Type A, B, etc.) bracing and custom I/J/Wt values simultaneously for a single brace? If that even works, I don't think it would solve the issue of splitting up moment and shear since neither the preset or custom inputs do that by themselves.

1ManWolfpack,
That publication is essentially a reproduction of the research I have read already. They provide [partial] guidance on modeling in 2D software, but the issue is that I can't translate that guidance into MDX-specific modeling. MDX appears to be too limited to handle those recommendations. On top of that, the guidelines would only help me get conservative bracing forces. There's no guidance on getting accurate deflections in a 2D grid analysis.

On another note, the state wants nothing to do with removing bracing near the abutments. They insist we maintain a maximum unbraced length of 25'-0", even though we've shown them we have performance ratios under 0.5 with 30'-0" max unbraced length. So now we're back to working with gigantic bracing forces again. They don't seem to care how big the bracing connections end up being. Perhaps we can get something to work, but it's going to be a solid mass of plates and bolts. I still think lean-on bracing is the way to go. We haven't presented the idea to them yet. But finding a way to model it is looking grim.

 

[bridgebuster]

It's probably too late but CSI Bridge may be a better way to go. In my group we used it to analyze a two span bridge with a 45-degree skew. Our largest cross frame force was around 800ft-k. The spans are only 80 feet but we also have a railroad bridge supported on ours.

I'm not a big fan of MDX. Two years ago we were doing a two stage deck replacement. The program couldn't give us staged deflections but someone found a way into "tricking" the program to do it. A colleague told me that the current version is supposed to be able to handle staged deflections.[pre][/pre]

 

[BigRy]

bridgebuster,
Yeah, I think it's too late for other software. We have other options here, including Larsa. But we don't have the time or budget to start a new model in more advanced software such as that. We're not familiar with it either, so there would be a major learning curve involved. It's hard to make a comparison of your bridge to mine. Was yours commuter or freight? How many tracks? Ours is 5 lanes of HL-93, so it adds up fast. And the girder moments are a function of the square of the span, so that's an eight-fold increase in moment from 80' to 228'. As far as staged construction, do you mean staging in one single model? I've never heard of that in MDX. My bridge is two-stage, and we simply have 3 separate MDX models: Stage 1, Stage 2, and Final. On that note, the state also wants us to widen the closure pour, which results in one of the girders having no slab on it at Stage 2. It gets no concrete until the closure pour. So I'm trying to fool MDX into analyzing that now ha ha.

 

[bridgebuster]

BigRy,

My bridge is a superstructure replacement, carry 4 lanes of traffic (HL-93), above this are three commuter tracks. Work is done in four stages and there re narrow closure pours between the stages.

On the other bridge I mentioned, the deck was replaced half at a time. It's a narrow bridge 30' wide. MDX couldn't handle the calculation of the staged deflections without some trickery. I believe the young lady who modeled it put the existing deck in as a line load. I'll have to ask her on Monday. Supposedly the latest and greatest MDX has a better capability for staged construction.

A few years ago, a group in the office designed a two span continuous curved girder bridge that was built in two stages, but they published cambers based on a single, full width deck placement, no closure strip (no room for it). OOPS! The deck didn't close; there was a 3" differential.

 

[BridgeEI]

I don't want to derail the topic, but how did the differential get corrected?

BigRy - I was thinking if you specified Type A bracing then you could use I,j,wt in conjunction and specify for separate members. I don't know if it would work like I said, just have to verify. Essentially you would assign dummy properties to some of the members.

Another thought is to not tighten the bracing until after the pouring. That may help reduce forces in the cross frames

 

[bridgebuster]

BigRy, BridgeEI - I'm not looking to derail the OP either; the problem was fixed by cutting out a longitudinal section of deck and re-pouring it. A $500,000 fix. OUCH!

Deflections for staged construction can get tricky especially with high skews.

 

[BridgeEI]

Ouch...you won't make that mistake twice!

 

[bridgebuster]

For the record - I didn't make the mistake and had nothing to do with that project.

 

[BigRy]

BridgeEI, with the way MDX is configured, you can't specify a type of bracing with general geometric properties. It's one or the other.

The state has requested we field-drill and bolt the bracing in the bay between the two stages, so that issue is out of the way. Now I'm working to model Stage 2 that has the one girder with no deck until the closure pour. MDX does not like this one bit. Some of the output is calculated correctly, some is not. The deflections in particular are wrong, so I'm creating STAAD models right now for the deflections and cambers.

Bridgebuster, the lady you worked with was able to model a bridge with no deck in MDX? She just added line loads for the concrete? What about composite deadloads and live load? We have no live load on our Stage 2, so I guess that's a non-issue for us.

 

[bridgebuster]

BigRy - She's going to look at her model, probably in a few days, things are very hectic in the office this week.