Drained Shear Strength of Clays

[GeoPaveTraffic]

Hello all, hope everyone's new year is off to a happy and profitable start.

I’m working on a project with one of our west coast offices and the principal in charge of the project wants to use fully softened shear strengths for some slope stability analysis. I’ll hold my opinions on this till the bottom of the post.
A few facts, at least as I understand them since I’m being added to the project team late in the process.

1. Project is a levee in the central valley of California, USA.
2. Levee is constructed primarily of cohesive soils with some sand.
3. Atterberg limits show everything: plastic silt, lean clay and fat clay.
4. Levee height is fairly small, 15 feet or so high with 2.5 to 3 horizontal to 1 vertical slopes.

It has been determined to use Stark, Choi and McCone’s method as presented in the May 2005 ASCE Journal of Geotechnical and Geoenvironmental Engineering titled Drained Shear Strength Parameters for Analysis of Landslides. While the paper mainly discusses ultimate (residual) strengths it also presents fully softened peak strengths. In rough terms it appears to lower the effective strength of the soils by about 20 percent.
I’m curious if others are typically using this method and what your experience has been.
Now my general thoughts:

1. I agree with the concept discussed in the paper and the need to be cautious when determining strengths for over consolidated clays.
2. However, it seems to me if you are going to refine the strengths that you also need to refine the target factors of safety.
3. If you refine the strengths downward because you have done a “better” job of determining the strengths; then the target factor of safety should also be refined.
4. In my experience very few slopes designed using peak strengths and a factor of safety of 1.5 fail. Those that do fail have typically revealed something in the repair that was not known during the original design or, more commonly, the slope was not constructed as designed.

Thanks to all for your thoughts.


Mike Lambert

 

[moe333]

Mike,
I have used the fully softened procedures previously, but for natural slopes with very stiff clay, and the strengths generally do seem low. For a levee, with compacted materials (presumably), I probably would have taken good undisturbed samples and run CU w/pore pressure to obtain drained strengths.

The required FS for the project will likely be set by the controlling agency, so I don't know that you will be able to use a lower FS with the use of fully softened strengths.

 

[BigH]

Mike - have I got a slope for you in residual soil . . cut back to 4:1 and it slides . . . residual friction angles (such as London Clay) can be as low as 11deg as you probably know. Fully softened is a bit higher. Most equilibrium analyses really are using average mobilized shear strength which for different parts of the slope are not the same. There is a classical paper on the subject - I'll see if I can find it for you (by classic - I mean pre-1975).

 

[GeoPaveTraffic]

moe333 - thanks for the comments. Pretty much agrees with how I would have approached the project. I also agree that getting a lower target FS is very unlikely.

BigH - Good point about most limit equilibrium analysis. Look forward to the paper, or at least the name. I can likely get it from our Library if you can locate the name.

Mike Lambert

 

[Mccoy]

Geopave, the 'fully softened' concept is maybe a vague one, but reading the article and sentences like this: "Skempton (1970) concludes that the fully softened shear strength is numerically equal to the drained peak strength of a normally consolidated specimen" I would call that condition as governed by phi_cv or phi constant volume. I'm more familiar with this definition related also to Roscoes's studies on critical state soil mechanics.
The following excerpt also is debatable, and apparently the first sentence (Terzaghi) is incomplete:

Terzaghi et al. 1996), for the analysis of first time slides in heavily
overconsolidated clays. However, Mesri and Shahien (2003) use
the back analysis of 107 first time slides in clay or shale to conclude
that the fully softened strength is mobilized in homogeneous
soft to stiff clay and where the shear surface cuts across
bedding planes and laminations. Therefore, it is recommended
that the effective stress cohesion be assigned a value of zero for
homogeneous soft to stiff clay. If the slope is comprised on nonhomogeneous
stiff clay or clay shale, at least a portion of the slip
surface is at residual and thus a cohesion of zero should be sued
for these materials (Mesri and Shahien 2003). In summary the
value of cohesion should be zero unless back-analysis of local
case histories suggest a value greater than zero (Mesri and Abdel-
Ghaffar 1993).

 

[Mccoy]

Continuing from the previous post-

So, the big question would be: "Is stability along a dam slope (your case), no previous sliding having occurred, governed by phi_peak and cohesion>0, or rather by phi_cv and cohesion=0?

My previous beliefs were that constant volume phi and cohesion = 0 are applied in cases where soil has been remolded, where progressive failure is involved, in cases where stiff clay is fissured, in cases where there is a stress concentration which causes the initiation of shear phenomenon (some offshore rigs foundations).

I don't know why an homogeneous OC clay, unfissured, with no progressive filure present, should necessarily exhibit a critical state shear strenght. The authors cite some evidence by Mesri though, such evidence should be anlyzed in detail.
A dam might be included in the remolded soil category though.

My bottom line:
We should determine first which are the parameters which govern the occurrence of the limit state under consideration, in this case slope failure. If peak strenght governs, then we should apply peak values in the analysis and no further reduction of FS, if critical state, or residual values govern, we should apply such values and again do not decrease our desired FS. It may not Always be easy to determine the correct values governing the occurrence of failure, so employing critical state values (fully softened) and a usual FS, we put ourselves in a safe condition, where we drastically cut the probability of failure due to our uncertainty on the actual conditions (peak or critical or residual) governing failure. Of course that also means Greater probability of being overcautious and increasing costs.
The above is less significant when the residual state governs, since that is the case when a large displacement has already occurred, literature says in the order of 2-3 feet at least and that's easier to dtermine.

 

[BigH]

GPT,

The paper had to do with a failure in London Clay at 11 degrees - when a 1 m high cut into a very flat hillside (11 deg) for a highway widening caused a failure - I think it is Skempton's paper at the London's 4th ISSMFE Conference "Stability of Natural Slopes in London Clay". you might also wish to check the Mexico City Conference's State of the Art paper by Hutchinson and also a few papers in the conference itself.

As I am scratching my head . . .

http://ascelibrary.org/doi/abs/10.1061/9780784412787.023

http://www-civ.eng.cam.ac.uk/geotech_new/people/bolton/mdb_pub/211.pdf

http://howland.co.uk/Articles/The n...nditions as illustrated by the weald clay.pdf

 

[GeoPaveTraffic]

Thanks Mccoy and BigH.

I'm going to continue to give this much additional thought.

Just to clarify, there is no evidence of failures or poor past performance at this site. The embankment (dam/levee) is constructed primarily of lean clay with some sand, however there are also fat clays, silts, and some very sandy or even clayey/silty sands present. The foundation consists of similar materials, however there are a few layers of poorly graded sand.

In the foundation the observed overconsolidation appears related to desiccation not previous overburden. In the embankment the overconsolidation is likely due to both compaction and desiccation.


Mike Lambert

 

[dgillette]

Where's fattdad? This is his kind of discussion.

I'm with McCoy on this. Residual friction angle is, IMO, way too conservative for a first-time slide. Phi_cv, used with "static" pore pressure, and zero or "token" cohesion, seems most "correct," once you figure out what phi_cv ought to be.

A few years ago, fattdad and I attended a workshop on softened strength, at Virginia Tech, organized by Mike Duncan. Unfortunately, CGPR charges nonmembers a fair amount of $ for the proceedings. Tim Stark made a case there for ring shear tests (ASTM D7608) to get what I think would be phi_cv at very large strain, but "The general consensus was that the fully softened shear strength is best measured with the direct shear apparatus." I'm not so sure about that, because the direct shear test would eventually reach residual friction angle once all the particles at the thin sheared surface are reoriented. Others thought it wouldn't be too far off to use the secant peak friction angle with no cohesion as an approximation of phi_cv.

On the other hand, if you are required to produce FS>=1.5, you might just use phi_peak with more than just token cohesion. In the warm California sun, there is bound to be a large portion of the embankment that isn't fully saturated, so you would have some capillarity to justify the cohesion (especially if this is a levee that only holds water occasionally, in which case assuming steady-state seepage may be unnecessarily conservative).

Cheers!
DRG

 

[GeoPaveTraffic]

Good comments dgillette.

As an update, preliminary analysis using fully softened strengths resulted in... wait for it...

Factors of safety below requirements. I'm sure you are all shocked by this. As of yesterday, laboratory testing consisting of CU triaxial tests are being setup to run. Will be interesting to see what those results look like.

I then expect a fight or at least a lot of discussion about what strength to use for the final analysis.

Mike Lambert

 

[fattdad]

. . . sometimes I get sidetracked. . .

fair warning, my agency is a member of CGPR, Tim Starke was a classmate of mine and Mike Duncan was on my committee. We (i.e., my agency) is also working with a consultant on an interchange project that'll expose the Potomac-group clays, which are prone to slope failure. The consultant has hired Tim.

The CGPR workshop on stiff clays had Steve Wright giving a presentation on levee failures. His work was pretty comprehensive and concluded that wetting and drying cycles during the service life of compacted levees obliterates overconsolidation and that the cohesion is reduced to zero.

Tim referenced his use of ring shear to obtain fully-softened strength. Problem is, (as you all likely know) where do you get ring shear testing performed? We had a break-out session to discuss the "best" way to measure fully-softened strength and we went with the DDS. Some of you may know about the Corps' method with multiple reversals and such. We are not talking about that approach to obtain residual strength. Rather, we are talking about getting your borrow material (i.e., CL/CH), pushing it through the No. 40 sieve (no air drying) and placing hydrated clay into the DDS machine. NO COMPACTION!

Rather than remolding the sample to some percent proctor, just place the normally consolidated "mud" into the DDS. From there apply your incremental normal load to get to your target confinment and shear. (Take your dial readings so the incremental loads are applied after primary consolidation has occurred.)

Using the earlier-posted definitation of fully-softened shear strength, you now have a normally-consolidated sample at some confinement in the DDS. Use the value for peak strength that results from the DDS. This will likely be greater than what you ultimately determine to be the residual strengh.

I think in today's practice, work by Starke and Wright and opinions as provide to the Corps would encourage the long-term performance of the levee be detmined by FSS.

f-d

¡papá gordo ain’t no madre flaca!

 

[dgillette]

Ah! There you are. I knew you couldn't stay away.

Now, how much difference is there between the peak you measure that way, and Phi_cv, which I suppose would correspond to critical state void ratio with drainage (no excess PWP)?

 

[fattdad]

DG,

I can send Tim an email and see if he's looked into that. I really don't know, but I'd think Phi-cv would be greater, then again. . .

f-d

¡papá gordo ain’t no madre flaca!

 

[GeoPaveTraffic]

Thank you fattdad.

I'm curious if you would propose to use or require traditional factors of safety using FSS?

Mike Lambert

 

[Mccoy]

In the past 4 years I've been discussing the Phi_cv issue with at least 3 lab guys, one of them very experienced. The result was frustrating. They've not been able to yield a plausible value of phi_cv compared to the type of soils and peak values. The tests have been carried out with shear boxes mainly but one of'em involved a shear ring apparatus. Apparently there is no accepted lab procedure to determine phi_cv. I wonder how the literature authors have been able to determine this parameter so accurately and if it is a useful parameter at all since it cannot be reproduced in ordinary lab tests. I thought it was the case here in Italy, but now I learn that's common in the States as well.
From now on I think I'm going to save the money. I'm going to determine phi_cv from correlations with PI or choosing an intermediate parameter between peak and residual values.
Or maybe I'll try with Fattdad's procedure, but only where stability issues of embankments and earthworks are involved.

 

[BigH]

With respect to fattdad's third paragraph, might want to look at Take and Bolton's paper "Seasonal ratcheting and softening in clay slopes, lesding to first-time failure" - Geotechnique 61, #9 (or:

http://www-civ.eng.cam.ac.uk/geotech_new/people/bolton/mdb_pub/211.pdf)

- also look at Picarelli et. al., "softening and instability of natural slopes in highly fissured plastic clay shales" (

http://hal.archives-ouvertes.fr/docs/00/29/93/23/PDF/nhess-6-529-2006.pdf).

We are having problems too in fractured volcanic residual (red) highly weathered clay slopes. We have installed drains - but these go perpendicular into the slope and it seems that some work and some not as if there is preferred paths of water flow (along remnant fissures perhaps) so am suggesting that any future drains be drilled at say 30 deg to 45 deg from the perpendicular line so that the full face of the slope would be covered given proper drain spacing.

Good discussion so far . . . and thanks for the tip fattdad on the DDS - I may be trying that method . . .

 

[fattdad]

regarding safety factors. . .

What a policy mess! From my Agency's perspective we "require" a safety factor of 1.5 for "critical" slopes and 1.3 for "non-critical slopes." We base "criticality" on slope height or the consequence of failure (i.e., whether a structure or the travelling public will be exposed to risk). We would certainly consider a robust testing program, comprehensive testing, and reliability assessment in tempering our project-specific safety-factor expectations.

Some agencies may struggle with case-by-case exceptions.

Seeing as how this forum is "off-line," I'll share my recent correspondance with Tim:

From me:

I’ve gotten engaged in a discussion about shear strength for levee construction. From your work, and the work of Dr. Wright (well and the CGPR workshop), I can see a prudent engineer using FSS strength. Now we can discuss whether the DDS or the TSS is the “better” measure. . .

Some dude asked me, “So, is the reconstituted, normally-consolidated “FSS” similar to phi-cv?” I don’t know. I just appreciate the question.

from Tim:

A quick response is [that] Phi-cv is mainly used for granular soils because critical state is frequently used to correlate with insitu density.

Unfortunately Skempton (1970) somewhat equates the reconstituted, normally consolidated FSS to the critical state but I think you should consider FSS to represent an overconsolidated clay that has lost all of its overconsolidation due to weathering, softening, expansion, etc. Remember the FSS is approximated by the PEAK strength (small shear displacement) of a reconstituted, normally consolidated clay and not the strength at significant shear displacement (constant volume).

<end>

I snipped out all the "how's it going" stuff. . .

f-d


¡papá gordo ain’t no madre flaca!

 

[Mccoy]

Tim Stak's and alii article was practically useful to me in a report I'm working on presently.
It's not about a slide, but I need the Phi_cv parametrs for the foundations sliding limit state.

I have phi_res = 18° measured by ring shear test (it was originally meant to measure phi_cv but it actually yielded a phi_res). I also have the atterberg limits, specifically the W_L= 36%

This is a sample with sand 42, silt 38, clay 17%. Phipeak = 28°, c' = 0.3 atmospheres. USCS Inorganic clay medium plastic. Sounds a little weird so much cohesion with so little clay. PI = 12%

Entering Wl in figure 6, I get Phisoftened-pohires = about 5°, so my phi_cv would be about equal to phi fully softened = 18+5 = 23°, which I find a plausible value.
I realize it sounds overkill for a foundation analysis but in a stability analysis that would be a good legal backup to a choice of a Tricky parameter.

In the Italian building code (stability analysis) it is specified that we need to determine phi_peak, phi _post-peak (identifible with phi fully softened or phi cv) and phi _res-.