FEA on a really Thick Box culvert 1

[Althalus]

For reasons I can't discuss, we have an extremely large box culvert with a 2' thick wall (all around). This box culvert is large enough to drive a car through. When applying load to any of the four planes, I'm wondering how one would take into account the corners.

Say the box is 10' wide on the inside, 14' on the outside.
We have soil applied to 14'. But the FEM has the nodes at 12' (CL of wall). So, if you apply the soil load to the top plates, it only loads 12' instead of 14'. We'd be missing 2' of load.

We can add the additional load in a variety of ways to sort of mimic the additional load. And it would be accurate for the total vertical load. But it would not accurately depict the moment (for example).

I've got an idea of how I would do this. But it isn't perfect.

How would you do this modeling to make it resemble reality?

 

[SWComposites]

Modelling with what code?
A 10' wide 2' deep plate is probably outside the limits of simple shell theory, so some sort of solid model is probably more appropriate.

 

[GC_Hopi]

What FEM software are you using? A few options are offset nodes or extend the slab past the nodes or hand calc the load and apply the load at the nodes as point load (line load if you are fancy and put in dummy beams).

 

[JoshPlumSE]

A 10' wide 2' deep plate is probably outside the limits of simple shell theory, so some sort of solid model is probably more appropriate.

I'm not sure I agree. Yes, I can see why some might want to go with a solid element model. But, I mean that using shell elements for analysis (bending, reactions and such) should give you good behavior and probably be conservative. To me, it's about whether the extra modeling complexity (and the potential error that goes with it) of using solids, is worth any improved behavior you might get.

How would you do this modeling to make it resemble reality?

Honestly, I think you've got a pretty decent model. You've even done a good job of identifying of what I'd guess is the main weakness with it (i.e. potentially missed loading).

My tendency would be to do one of the following:
a) Apply the extra load and moment directly to the joints at the top (and bottom?) of the wall.
b) Create a "dummy" frame / beam member along the top (or bottom) of the wall where you want to apply some additional load and moment.
c) Create another group of plate elements that cantilever up off the wall (or down past the wall) and load them directly.
d) You might even decide to ignore that load model, but acknowledge it in the design by checking the slab for this amount of compression.

 

[IDS]

Standard practice for FEA of long buried structures like culverts is to do a 2D plane strain analysis modelling the soil with plate elements and the concrete with either plates or beams on the CL with offset elements to the soil/concrete interface. Considering the large depth to span ratio, using plates would probably be the best option in this case.

Possible reasons for doing a 3D analysis include modelling distribution of point loads, or interaction with headwall structures etc, but those don't seem to apply in this case.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/