Differential Deflection in RCB

[apriley]

We're adding fill to an existing rcb at the end (existing fill tapered to 0 at the end, and we're putting 14' on it to match the rest of the existing). The geotechs are telling us that the settlement will be 3" on that section. The lead structural engineer on the project is worried about cracking in the box due to the differential. However, any amount of differential settlement in the box cranks in huge forces. So even if you can calculate the deflection accurately, the box doesn't "work" structurally in the longitudinal direction by a long shot.

I feel like this is a created problem, since I'm sure that there are plenty of existing box culverts that have had fill added in a future project. Furthermore, it would seem that some differential deflection would have occurred during construction, since the fill varied in the initial condition. Does anyone have experience with this? Is it a problem, or is it something that can be explained using engineering judgment?

Thank you.

 

[ishvaaag]

Making a 3D model in FEM with maybe brick elements for the soil and plate (or brick) elements for the culvert will show the effect of the additional load disperse out of the culvert and lesser differential settlement issues. You will be ensuring that you have a true rigid connection at the old-new interface ans see what the response of the old part is in this more accurate model. Upon the results of the model you may think on how more to proceed, because obviously such model isn't either perfect.

When you build something new besides something old whatever the loads, compatibility of deformations obliges the same settlement at foundation, point of contact. In a culvert you have the walls, deep beams along, and if longitudinal settlement becomes a problem you may add flexural strength by diverse ways by adding effectively collaborating steel for flexural stress; and if postensioned (even on brackets!) you may even control and correct the actual deflection to some wanted movement.

 

[apriley]

I'm not worried about a NEW structure next to an existing, I'm worried about adding load to an existing.

I'm trying to avoid modeling - the model I used showed that the box was overloaded even with minimal deflections (3/4"). I have very generic and useless foundation springs from the geotech, so modeling really doesn't get me a reasonable answer, unless the answer is to rip out the box and rebuild it, which I'm convinced is unnecessary.

So, have you (or anyone else) had experience adding load to an existing box? Were there deflection problems?

 

[cvg]

since the existing culvert has some ability to absorb and redistribute the loading, the soil beneath the culvert will see perhaps a different loading than what your geotech has used. Spreading the loading out to a wider area may reduce the differential settlement to an acceptable level. However, a lot depends on the soils and structural design of the original structure.

 

[JAE]

apriley,

The differential settlement can be used to calculate the forces that will result in the box, only they must be based upon some assumption of box stiffness.

Now, normally, you would initially use the overall gross moment of inertia of the box (assuming it is uncracked). But this means you have a very stiff box being deflected 3" and will produce the highest bending stresses in the section.

But most likely, with the uncracked, gross moment of inertia, you will exceed the cracking stress in the concrete and thus the stiffness of the box will immediately reduce. If the box is reinforced properly, the hope is that the cracks would be somewhat distributed along the length of the box but that might not be the case - you could get one large crack relieving all the stress at one spot.

So if it cracks, is there a problem? Might be that soil would slowly erode into the crack and cause sink-holes above. Some type of hydraulic grout plug might be attempted. Could be expensive.

If the cracks are well distributed and the rebar holds the concrete together, you may not have a problem.

 

[ishvaaag]

A sketch of the old and new situation may help, for I see me looking the thing more or less as JAE seems be doing. I was mainly saying that you can even control the general deflection by adding reinforcement. I once had to design a culvert extension for a lane extension in a 6 lanes hwy whilst on a consultancy, but in that case no problem of the kind you have at hands manifested, even if it was a big one, maybe 8 or 9 meters of span.

 

[apriley]

The box is standard. I've attached the standard plan from 1966 - as you can see, the longitudinal reinforcement is minimal. The box is a few hundred feet long (under a freeway), and most of it is covered by 14' of fill. The last 50' had minimal fill, and we're filling it to 14'. I don't have a good schematic for that.

I used cracked section properties in my model, but even with that reduction in stiffness, the loads were well beyond capacity.

The lack of longitudinal bars (#4 @ 18) makes me think that this isn't an issue for boxes. I doubt that any box would settle exactly the same in every location. It's notable that standard boxes are constucted with expansion joints every 30'...it's possible that maybe it's a small enough distance that all deflections are taken by the joints and each piece sees limited load.

 

[apriley]

I ran the analysis using releases @ 30', and the forces reduced dramatically. The box is OK. I think it's clear that the expansion joints are used for exactly this reason. The only thing left to do now is head to the field and check that the expansion joints actually exist.

Thanks for your help with this issue.