Should CPT shear values be supported by testing? 1

[sgsibob]

I would like some opinions on whether the standard of practice is met if one relies exclusively on cone penetrometer (CPT) correlations to derive soil shear strength properties for design of critical facilities such as tall retaining walls, bridges, and large culverts, or whether such correlations should necessarily be supported by some site-specific measurements of shear strength, such as direct shear or triaxial tests. I also was wondering if the perceived reliability, and therefore the acceptaqbility in design, of stand-along CPT data depends on whether the materials are natural or man-placed fill, cohesive or non-cohesive.

 

[singingk]

Both the interpretation and the deriving of soil poperties from CPT-data are highly empiriclly based business. I would never give any geotechnical recomendation for buildings as mentioned in your post without getting the soil in my hands and doing some basic testing.
If I know
- the soils which have been penetrated,
- their specific cpt-profiles (cone resistance,sleeve friction as well as their characteristic curves) and
- their properties by my own experience (or reliable experience of others geotechnical engineers)
and (!)
if I got some basic test results (grading curves, water content, LOI),
I estimate strength and deformation parameters taking the cpt-data into account without further cost intensive testing. Without this it's hazardous!

 

[sgsibob]

You mention basic (index) tests like Atterberg Limits, gradation and moisture. I agree that these should be part of a program for any major structure. More specifically, though, for critical structures, do you feel direct shear or triaxial tests are necessary to check the CPT correlations, or do you think the CPT plus index tests is adequate for design of such facilities?

 

[fattdad]

You'd likely be O.K. in sands, but less so in clays. Most of the engineering correlations for clay behavior are more empirical. When you reference triaxial testing of shear strength, I tend to think of more clay soils. For cohesive soils, I'd consider undisturbed sampling and testing consistent with the intended stress path (i.e., TXC or DD).

Hope this helps.

f-d

¡papá gordo ain’t no madre flaca!

 

[emmgjld]

Actually a very good question and discussion for most indirect testing, to include SPT.

I am amazed at what decisions are made with average to poor quality SPT values, by engineers working in 'out of town' areas. I am also amazed at the poor quality of field testing, sampling and laboratory testing which is often used, whether 'in town' or 'out of town' for individual engineers. I understand the canard of 'lots of local experience'. I also am currently redoing several jobs, as the 2nd Geotech, since the original work was insufficient.

 

[msucog]

i think a point should be made that the definition of "insufficient" lies mostly with the owner. i've seen many situations where the owner had a set budget...so the geotech gave them an exploration that met that budget. we do it occassionally and note that other "stuff" should be done. quite often, you do not know that the spt values are poor quality...hence, a safety factor comes in to play. a lot of times, we're the second geotech and we see that the recommendations from the first geotech was due to the fact that the owner only wanted four holes and bare bones recommendations for a fee that was almost nothing. (the owner will get what they pay for quite often). i am absolutely against doing work with the input of local experience (as i've seen a lack this being a source of litigation).

all in all, i think engineers should be careful in different geologies (whether 5 miles or 500 miles apart). likewise, the project specific "issues" will probably be the biggest factor. here in the Piedmont, you could go 500 feet and find a completely different geology...so with the way litigation goes these days, i suggest that all professionals be careful before they try to go out on a limb to help a client.

"no good deeds go unpunished"

 

[dgillette]

"I am absolutely against doing work with the input of local experience..."

That's a typo, right? Should say "without the input..."?

 

[msucog]

lol..yes indeed. that is a slip of the worst degree there. my fingers get away from me sometimes when i manage to get home early (on time for a change) with kids running around and work still on my mind. thanks for catching that and pointing it out.

 

[BigH]

Without going into a long diatribe, no investigative program should solely be based on CPT (or piezocone). You need always to have a well defined and sampled "traditional" borehole so that you can correlate the CPT, in actuality, to the soils encountered. Always good to have correlations of CPT results with CU or vane tests.

 

[singingk]

"More specifically, though, for critical structures, do you feel direct shear or triaxial tests are necessary to check the CPT correlations, or do you think the CPT plus index tests is adequate for design of such facilities?"

If you get a budget for intensive testing, take it and improve your experience. Most time the promoters don't want to spend money for real qualified soil testing, hence desingnig such facilities (e. g. deep excavations in soft to very soft soils) based on cpt's and index tests is our daily business. Don't try to base any geotechnical judgement about shear strength on single shear tests (it doesn't matter wether triax or direct sehr tests) cause you can easily be fooled by it's results due to the unavoidable variations of soil mechanics test results. Single shear tests are only suitable if you have got enough experience correlate their (otherwise maybe missguiding) results with.

 

[coneboy]

Great question Sgsibob!

Many times my friends and I have discussed and debated the question you are asking. On smaller projects where we have lots of local experience we have just used the CPT data (combined with auger samples to get a soil PI) to model the Su values using a estimated Nkt value. Usually this invloves modeling the CPT data until a believable soil OCR value is reached. We then back calculate the Nkt factor to give us our Su values. This method has served us well when we are in area that we fairly certain the soil OCR values are near to 1.

On larger sites with more heavily loaded structures we always conduct some Nilcon vane or field vane tests. The Su values from the vane testing can then be used to fine tune your Nkt factor so that you determine the Su values more accurately. Using the site specific Nkt factor you can then get a good estimate of the soil OCR values. In the event that the client/developer wants a building with some form of unique or heavy loads we will get some piston tube samples.

All the soil modeling and back calculation of Nkt factors can be done very quickly with a CPT data analysis package or spread sheet program.

In the past what approach have you used?
I would be interested to hear your comments.

Coneboy

 

[sgsibob]

Where I practice, the soil conditions are highly variable. We routinely have to deal with inhomogeneous sand-gravel-clay (SGC) mixtures having variable amounts of carbonate cementation. In those cases I believe the presence of coarser clasts (gravels) can throw off CPT values and if the soil is not clast-supported, some direct shear or in situ tests are warranted. (Some practitioners discount the value of direct shear tests as not representative, subject to loading complexities, susceptible to sample disturbance, not realistic, and so on. Such statements have some truth to them, but have always struck me as no more compelling than similar statements that can be made about other, widely-practiced soil laboratory tests.)

It's not just about non-ideal soils (SGC mixtures), though. It's about using published correlations for any type of soil without thoroughly understanding the applicability of those correlations.

What I am really concerned about is the implication that mechanical properties are derived in a deterministic way from CPT data, when the CPT correlations used for them are actually empirical. For example, what is the justification for basing the in-situ shear strength of an SGC mixture on published CPT correlations developed on sands, if the data supporting those correlations do not include representative SGC mixtures?

On the other hand, if one references large-scale in situ testing of representative soils, or conducts one's own shear testing program to correlate the properties in a site-specific way, then I believe the CPT data can be useful to widen the scope of the testing and the results could be called deterministic.