triaxial CU tests vs SHANSEP

[AK92]

Has anyone compared the results of CU testing vs SHANSEP parameters using other methods?

With the consolidated undrained testing, you can get a set of effective stress parameters and a set of total stress parameters, which relates shear strength to consolidation pressure (similar to SHANSEP relationships).

However, I do realise that with the CU test, the samples are reconstituted to the original pressures (at just one particular depth), whereas with field vane the actual variation vs depth is found, so there might be some fundamental differences there.

 

[fattdad]

each point on a CU test is consolidated to a prescribed degree. They're not all consolidated to the initial overburden pressure.

Maybe I'm missing the OP's question.

f-d

¡papá gordo ain’t no madre flaca!

 

[AK92]

@fattdad

Sorry I wasn't too clear with my original question.

My point was that with the CU test, samples are reconstituted, whereas with other methods (field vane, CPT) the strengths are measured in-situ, so I understand that there's a fundamental difference between the CU tests and conventional SHANSEP methods using in-situ tests.

My question is whether anyone has tried to compare the SHANSEP relationship (Su = a+b*σ'v) with the total stress parameters from the CU test (Su = cu+ tanΦ * σ'v).

I'm asking this because I'm involved in a project with quite many CU tests performed, but no reliable tests for undrained shear strength(most of the "saturated" UU samples tested have significant friction angles, unreliable mechanical CPT results, etc..), so it's quite hard to get a SHANSEP relationship outside of the CU tests. I didn't assign the soil investigation unfortunately, and I don't think the client is going to spend any more money for SI at this point.

 

[fattdad]

CU testing may or may not be on reconstituted samples. You can certainly take Shelby tube samples to enable testing the strength at the in-situ void ratio, moisture content, etc.

Each point of the CU test will return an undrained shear strength that should be appropriate for the state of consolidation. So, for a three point test you should have three stress strain curves, excess pore pressure data and such. Each of these can be interpreted for total or effective strength. So, for a three point test you'd have three Su values (depending on your definition of failure) for three value of, "p'." Heck, you'll get three Su/p' relations from each three point CU-bar test!

Maybe I'm still missing something?

f-d

¡papá gordo ain’t no madre flaca!

 

[GeoPaveTraffic]

Why do you need/want SHANSEP parameters?

As fattdad pointed out, you have a number of undrained shear strength points to work with.

Mike Lambert

 

[AK92]

@fattdad

Sorry, I have used the wrong terminology. It should read "reconsolidated" not "reconstituted".
I know that I can get Su/p' relationships from the CU test, but how does it compare with the SHANSEP relationships for particular jobs? I have read Charles Ladd's Terzaghi Lecture on it, and he mentions that CIU tests are not very useful because the soils are isotropically consolidated, and not anisotropically consolidated with consideration of the K0 of the soil.

@GeoPaveTraffic

I am trying to model staged construction of a reinforced earth wall using vertical drains. The soil would not be able to support the final load with the current Su, so I need to be able to model the gain in strength with consolidation and dissipation of pore pressures. It doesn't help that none of the undrained strength testing in this project hasn't been reliable (UU tests on saturated samples have significant friction angles, probably samples have lost water during transport!) and they did not specify any other form of in-situ undrained strength testing such as field vane (we did 3 CPT's, specified CPTu testing but got mechanical CPT testing with questionable results). In general the workmanship is questionable. We do have tons of CIU testing for some reason unknown to me (what a waste of money!)

We're thinking of relying on SPT data to estimate Su using good old Su=5N since it has the least amount of operational error with automatic hammers nowadays. I realise that SPT is not a good method to estimate Su in clays. However, I estimated the Su using the undrained friction angle and undrained cohesion from the CIU test using the estimated current overburden pressure, and the results don't correlate well with the adjacent SPT's at all (SPT = 1 to 12Su), with some of the higher SPT's having really low Su's and vice versa. So I'm not too sure whether to trust the SPT's or the CIU tests, so that's a major issue too. It doesn't help that the soil is a residual volcanic soil. The SPT's tend to suggest soil strength which is much higher than what the CIU tests indicate. The UU tests (when I use the current overburden stress) shows even lower strengths than the CIU tests. However, I'm thinking of being on the conservative side and rely on the CIU test results instead, looking at the very high water contents, void ratios, Cc/Cr values and low OCR values.

The average undrained friction angle is about 15 degree and undrained cohesion is about 20kPa. Su/σ'v = tan Φu = tan 15 = 0.267 which is similar to estimated SHANSEP relationship of 0.267 estimated using empirical correlations using PI by Skempton.

So I'm thinking of using these parameters to model the staged construction process and fill stability evaluation. Would my approach be correct?

 

[dgillette]

If I understand the problem right, regardless of the OCR in the ground now, you will be concerned with NC soils when the fill is being constructed. That makes analysis easier (because you don't need so much testing to find the exponent applied to the OCR), but of course the soil is weaker if NC. Do you have oedometer tests? Those are much more valuable in figuring out the existing strength than the SPTs are! Probably more valuable than the UUs also. If you have oedometers, you can use the main SHANSEP equation tau/sigma'vc = S(OCR)^m, and look at sensitivity to reasonable variation in S (remembering that S varies with the orientation of the shear plane). Most of the time, you can just assume that m=0.8. Refer to CCL's Figure 12, with the portions of the shear surface marked C, D, and E. That's IMPORTANT when trying to assess reasonable values for S. Check whether your CPT data seem to fit with any reasonable value of S. That will give you some idea about whether to discard them completely. Note that CCL discouraged the use of undrained friction angle, phi-cu, although once in a while you can get away with it if the geometry is right.

CIU tests tend to over-predict the strength of anisotropically consolidated material with sigma1'c equal to the consolidation pressure in the main SHANSEP equation: tau/sigma'vc = S(OCR)^m. The "tau" there is tau-ff, shear stress on the failure plane at failure.

The strength you have to work with is basically whatever it is now, outside the footprint of the embankment. Within the footprint, it will increase to S*sigma'vc as consolidation progresses. If the material is now OC, there will not be much gain in strength until sigma'vc exceeds the maximum past pressure. This makes sense, because there isn't much decrease in void ratio on the unload-reload curve in a consolidation test; you don't get much more strength until the void ratio decreases significantly.

BTW, I recently read that Prof. Ladd had died last month. He was a consultant on a project that I worked on long ago, and I learned a lot.

Gotta go home for dinner now.
DRG

 

[AK92]

@dgillette

Thanks for your help!

I do have oedometer tests and tons of them! OCR tends to be higher near the ground surface, then it drops with depth until it reaches about 0.8-1.0 at depths of around 8m or so (looks like the soil is still undergoing one-way drainage consolidation). I don't have sensitivity data unfortunately, but I suspect that it is quite reasonably sensitive (high PI, LL and water contents). According to Skempton's equation, I'm predicting a S of about 0.267 (similar to tan phi-cu). I think I would just go with that considering the lack of good information. Do you assess the stability of the slope using limit equilibrium methods, by dividing the soil into the improved part within the footprint and the unimproved part outside the footprint, using the estimated pore pressures to estimate the increase in effective stress and thus the increase in undrained strength (using the SHANSEP correlation) in the improved area?

From advice from a very senior engineer, he doesn't think that the soil is anisotropic as it is a residual soil made of volcanic ash (weathered in place from the parent volcanic ash) which is unlike the formation process of sedimentary soils. However, there have been lots of prior human activity on the site (construction, filling, vehicle loads from heavy buses and trucks) with a problem with regional groundwater dewatering. From site pictures, I do see quite some evidence of differential settlement occurring. I honestly do not know what to make of it though.

Sorry to hear about Prof Ladd's passing. He had contributed a lot to the field.

 

[dgillette]

Yes, limit equilibrium would be the logical thing to do, unless you are also going to have a FLAC or PLAXIS model set up for deformation analysis, in which case you could analyze the stability by "relaxation." You're probably not planning to do that however. You will have to divide the foundation into zones with different undrained shear strength, vertically and horizontally, with the number depending upon how SHANSEP-friendly your software is, and how conveniently the OCRs vary. Draw a big section through the foundation and mark σ'v and σ'vc for each location. From that and SHANSEP and whatever other data you can bring to bear, find the undrained strength that exists now. Put on the first stage and check undrained stability. If it's stable, let consolidation happen and find the new values of sigma'v and strength. Put the next stage on and check undrained stability. Let consolidation happen, and so on. Bon chance!

Unfortunately, there isn't be a single value of S that works for the whole thing because it varies with the stress rotation, not just because of anisotropy of the soil fabric. That's seen in reconstituted lab samples as well as samples from the ground. I'd be hesitant to assume 0.267 for the average? value of S (≠ tan φ-cu) without backup from actual lab shear data. (If that's from Skempton's lab results, be careful, because he probably only used CIUC tests, not DSS, CAUC, and CAUE.) You are working with something that is geologically different from the typical sedimented soils and tailings that SHANSEP was made for. The soil may not be layered, but its stress and strain history has been anisotropic, so I would expect that its strength is also.

That's about as far as I can go with this "remotely."

 

[BigH]

As dgillette states, this is tricky in that you are dealing with a residual volcanic soil. There will be remnant joint planes within the volcanic - we have found them on our site - and these may still effect stress propagation.

 

[AK92]

@dgillette

This is the approach that I'm currently using. What representative value of S would you recommend, perhaps about 0.23? Do we have literature to adapt SHANSEP to residual volcanic soils? I always wondered why the undrained shear strength of a clay wasn't plotted against the void ratio (or water content), this way we would have a very convenient method to estimate the shear strength at any stage of loading.

@BigH

How do you identify the joint planes if it has already been weathered to become a residual soil? How would this impact the design and how would I take those into account?

 

[dgillette]

I'm not going to recommend a specific S, because I have pretty close to zero experience with residual soils.