Slope Stability

[SuperSandman]

Hi Guys

Once again im refreshing my memory with some basic fundamentals.

I have numerous software programs that can calculate the factor of safety for slope stability and retaining walls. Most software is pretty user friendly. However my question is based on the input of cohesion and angle of friction.

I personally carry out shear box or triaxial tests (Drained) to obtain the drained cohesion and angle of friction. A typical slope stability program will ask me to input these parameters. I would then input my results for C and Phi and continue the stability analysis. A few experienced engineers told me that i should not trust the "cohesion" value and that i should enter this as zero (0) value into the program and run the analysis based purely on the drained angle of friction. So basically they told me that i should always use 0 for the cohesion.

In some cases i have tested cohesive soils with very low angle of friction (Between 17 - 23) but the same sample also has gives me high cohesion (between 30 - 40). If i input the cohesion as 0, i get a very low factor of safety for the slope (due to the low Phi). And many cases i obtained a factor of safety of less than 1 (>1) BUT the existing slope has been standing at its current angle for decades without any slip failures.

I personally understand that even though i did not use cohesion in the analysis, it does play an important role in the stability of the slope, especially when i have high cohesion values from my lab tests.

I never questioned the "experienced" engineers about this but i thought maybe you guys can enlighten me a bit more. Should i trust their recommendation of ignoring the cohesion in the stability software, or do you feel that i should in fact use a conservative value for cohesion???

I would appreciate all responses!!!

 

[fattdad]

What happens is the upper 5 ft or so undergoes repeated wetting and drying cycles that in many instances reduces the shear strength to some "fully-softened" state. This is most likely to occur in fat clay or elastic silt. Fully-softened shear strength shows no cohesion intercept (the initial portion of the failure envelope is curved).

I guess when you get into the lower depths (deep seated failure planes and such) you could apply some drained cohesion, however.

f-d

¡papá gordo ain’t no madre flaca!

 

[GeoPaveTraffic]

Your example of existing slopes demonstrates the problem with always assigning a cohesion value of 0 for drained conditions. I have read and understand the various theories that have been put forth saying there is no such thing as cohesion in a fully drained material. However, I too have seen many slopes that should not be standing if cohesion = 0 was always true.

That said, the best option is to use laboratory tests with "a grain of salt" and a whole lot of experience when doing slope stability analysis. You indicated that your results indicated between 30 and 40 for cohesion but you didn't include units. If the units are psf, then I would typically use it in stability analysis. If the units are in psi, then I think you have a problem with your laboratory tests.

In general it takes experience with the local soils to assign strength parameters for slope stability analysis.

Mike Lambert

 

[SuperSandman]

Thanks for the responses guys!

My unit for cohesion kPa

 

[GeoPaveTraffic]

Those are fairly high values for cohesion for drained testing. I recommend you review your testing proceedures and ensure that the tests were turely drained.

Mike Lambert

 

[dgillette]

There's neglecting cohesion, and then there's neglecting cohesion.

If you measure a high cohesion and low friction angle, you are probably seeing the effect of overconsolidation, capillarity, etc. at low normal stresses. At higher normal stresses, those effects become less important compared to the value of sigma' times tan(phi'). If you simply translate the apparent strength envelope downward until it intersects (0,0), you may be throwing out a big chunk of the frictional component of strength at higher normal stresses.

It is not exactly correct to do so, but sometimes people (including me) run a MC envelope through, or very close to (0,0) and tangent to the Mohr circle for the specimen with the highest normal stress, if it's high enough that the strength is primarily frictional. In effect, this throws out the apparent benefit of OCR and capillarity, without throwing away the frictional component. If the normal is high enough, OCR effects and capillarity become minor to nil, and this will give a number that's at least halfway almost reasonable.

 

[GeoPaveTraffic]

ddgillette makes a good point. In the end, what is important is to use a strength envelope that models the behavior of your material in the stress range that you are working. The specifics of how you enter that envelope into the stability model is much less important than modeling it correctly.

Mike Lambert

 

[dirtsqueezer]

You might just to go your local city engineering office and look up public records of previous studies to see if they've handled the same conundrum just to get a check of your data. They may even get specific. If you're dealing with a problem region slopewise, and you're just getting a refresher, a half hour out of your day might be worth the trip.