Removal Limits Below Column Footings

[kcall2]

Background: Using Schmertmann's method I estimated settlement below a proposed column footing and recommended removals below the footing to reduce maximum settlement to less than one inch. I recommended that removals extend beyond the footing a distance equal to the removal depth. This is based on the assumption that the stress from the footing load spreads out along lines with a 1:1 slope.

Problem: Utilities and easements will limit the horizontal extent of removals for the column footings (on one side) so that the removals will only extend beyond the footing a distance equal to half of the removal depth (or less).

Question: How will the estimated settlement be affected and can it be reasonably evaluated? Will this affect the behavior of the footing? Are my assumptions too conservative (although common practice here)?

Your thoughts are greatly appreciated.

 

[fattdad]

Just to clarify terms: I use undercut for the depth and I use overexcavation for the width. I'd undercut to limit settlement. I wouldn't overexcavate to limit settlement. I'd just remove the bad dirt and replace it with some highly frictional aggregate.

I understand that you used Schmertman's method, but have you looked at just solving this problem using Bousinesq's stress distributions? You may find it interesting to look at the actual stress distribution below the center, edge and corner of your footing. You'll see widely varying stress distributions. The actual settlement, however will be some "average" of the soils response to these stress increases. If you are dealing with compression (assumed as you said you're using Schmertman's method), then you can assign the modulus values to the various layers and then look a the stress increase in each layer (or sublayer) and integrate the stress increase with respect to depth (i.e., calculate the "area" of the stress bulb). Divide the "area" by the soil modulus (using consistent units) and voilla, you'll have the anticipated settlement.

There is no part of this determination that depends on overexcavation.

I haven't been sued yet (i.e., my foundation designs hold up). How did you get at soil modulus? That's the issue. . .
f-d

¡papá gordo ain’t no madre flaca!

 

[oldestguy]

I've run into this problem several times.

I try to clean off the nearby structure and then assume that the friction between the compacted material and that structure makes up for what you can't replace.

And I also have not been sued on this subject.

 

[BigH]

You have assumed a 1:1 distribution of the stress below the footing; normally for stress distribution - it is 2V:1H (fairly comparable to the Bousinesq that fattdad mentioned - of course you knew this. Assuming that you are replacing the natural material with a well graded sand and gravel - perhaps crushed stone (like a road subbase), then the stiffness of the placed material will be significantly higher than that of the original and will attract more stress than the Bousinesq or the 2:1. So, the edge effects will be smaller. Given that you can excavate to a line of 2V:1H, if you do, I would see very little danger to the existing services . . . don't undercut the existing service lines.

 

[kcall2]

I selected the Schmertmann method due to the low fines content of the soil. I do understand that the 1:1 stress distribution is conservative and that a 2v:1h is more realistic, but where is this supported in the literature?

Note that column footings are integrated into the continuous perimeter footings and will behave as strip footing with non-uniform load.

f-d, soil modulus was estimated based on soil type and SPT blow count data.

 

[BigH]

Just about every standard geotechnical textbook. See:

https://www.u-cursos.cl/ingenieria/2008/2/CI62D/1/material_alumnos/objeto/19147

 

[CarlB]

Actually wouldn't a stress distribution projection of 1:1 be less conservative than 2:1, when it comes to settlement prediction? 1:1 would show less stress increase at depth.

 

[fattdad]

I think we're contrasting 1:1 with 1/2:1.

f-d

¡papá gordo ain’t no madre flaca!

 

[CarlB]

Well I used 2:1 (v:h) to follow kcall's nomenclature.
1:1 is more conservative in evaluating lateral extent of influence, but less conservative if that spread is used to calculate stress increase.

 

[BigH]

This "ratio" is always troublesome - geotechs use H:V (as in a 3:1 slope) while structural types seem to use V:H - which is why I always (with a few forgets) use the H and V in the ratio.