22222222,
    I wouldn't use a value for cohesion when dealing with the soil you described-it's not technically correct.  My answer assumes you had pore pressure measurements and were able to calculate the effective stresses for the tests.  I think the cohesion you observed in the CIU test is most likely some slight curvature of the effective stress vs. shear stress plot.  This curvature occurs at higher confining stress levels.  The extent of the curvature depends on a number of things especially confining stress.  Some might incorrectly interpret the cohesion intercept to be above the origin based on your curves.  Perform the same test at lower confining stress levels and the shear stress intercept will probably be closer to zero.  If you are actually dealing with a problem involving high stress levels, then knowing the curvature will help solve strength related problems for your project.

Thanks a lot MRM for your answer.
CIU test was with pore pressure measurement to get effecive parameters and results represented by P-q diagram so line was stright. Confining pressures were 0.5/1.0/2.0/ and 4 kg/cm2.What do you think would be lower confining stress levels? Soil was ASTM, SW-SM with 90 % sand 9% silt and 1% of clay.
10 Kpa is close to zero but of course depends on the way you look at it.Effective frictional angle was 32 degrees.
I have repeated the test with the same confining stresses for SM soil with 70% sand 22% silt and 8% of clay.Result were c efect.= 12 Kpa and effective friction angle 30.6 degrees.

Be careful. I have seen embankments stand of fine to med. sand that shouldn't stand. Often times it is due to the adhesion of the moisture in the fine soil. After a while the soil dries out and the bank comes down. This same process could be effecting your results, esp. w/ 10% fines & of that only 1% clay. I don't know what your application is, but it is often difficult to predict how the cohesion and the friction will act together. Often friction increases as cohesion decreases over time. For slopes and retaining structures, I design only for friction or cohesion

Good point DRC1-It's true that when granular soils are partially saturated they can temporarily stand at angles greater than their internal friction angle-in fact it's the only reason we're able to build sandcastles.  However, in this case though, the test was run undrained with pore pressure measurements so the sample should have been saturated during the test.  There probably wouldn't have been any "apparent cohesion" as it is commonly called.

Also 22222222, I was looking at the stress levels you used in the tests.  They are moderately high stresses, but certainly not the highest that have ever been used in a triaxial test.  Depending on the sand type (grain type) and the gradation and possibly a few other factors, I still think it's possible to have some curvature to the shear stress-effective normal stress plot (the points on this plot are usually the peak stress ratios from your p-q diagrams.)  The curve of the Mohr circle plot shouldn't be ignored when evaluating the strength and everything else, but just know that it is probably not caused by genuine cohesion, which is rare in most cases-even with most clayey soils.

There's another interesting thing to consider for your project too...if you need to perform a settlement analysis for your embankment, look into whether or not particle crushing will be a concern at these high stress levels and if it should be considered in addition to your normal settlement analysis which usually assumes the soil is a linear elastic material, without particle breakage.  There is an good article by Lade, Yamamuro, and Bopp called "Significance of Particle Crushing in Granular Materials."  It is in the Journal of Geotechnical Engineering, April 1996 edition.  It may or may not be a concern, but wouldn't hurt to look at.

DRC1.
Thanks for your answer. Cohesion is always a bit of problem.Everything will depends on quality control during construction where no good quality will bring about failure.As for the slope design I use both cohesion and friction but only for slopes that are not active or old landslides and off course not made of sand and gravel only. If they are active or old landslides I always use zero cohesion and residual friction angle.So far it is working well. For retaining structure small cohesion is good to check your failure extent under the structure.In my case, recently we replaced all loose or soft soil with cement stabilized soil and therefore we have good foundation and can sleep well.

Hi MRM.
Thanks for your answer again.Recently I dug through several manuals dealing with compacted soils and some japanese standards along with NAVFAC DM 7.2 (1982) that describe cohesion for similiar soil as we have: cohesion as compacted up to 46 KPa and cohesion as saturated 19 Kpa (420 Psf). Since the lab shows in my case only 10-12 Kpa for design we will use 5 Kpa plus try to reinforce the edges of our embankments with geogrids, so far our client does not like because of the 100 years design life. Our embankments will sit on the soil cement replaced soil about 1 Mpa UCS. Thanks for refference about partice crushing. I will try to get that paper because the topic seems very interesting to me. I will try to re test again under the lower confinig pressure as you suggest.As I said to DRC 1, the quality control will be crucial during construction.

Hi 22222222,
Whatever parameters you decide on, it sounds like you're approaching the problem from a good conservative standpoint, so there shouldn't be a problem.  You may not want to spend the money to run your triaxial test again at lower stress levels unless you are interested enough in what the results would be and can do it yourself with your own equipment (for free).  Although I would be interested in hearing the results if you did run it again.  The stress levels you chose for your original test are more probably more in-line with the stress levels you'd expect from a high embankment as you described, which is why you chose them in the first place I assume.

Thanks for the interesting discussion!

I'm curious: why a 100 year design life for a well graded silty sand?

If the SW-SM material is primarily composed of silica, grain crushing is not an issue.  Did you have any trouble achieving a high B-value when saturating your samples?  Is the silt fraction micaceous?  If so, the embankment may be more deformable than usual.  This is seldom a problem, but can occur (although I have not seen any case studies.)

Side Note:
Compressibility is a problem with flue dust from coal-fired power plants.  I tried to run CIU tests on ash samples 15 years ago; could never get the B value above 0.7 even after a week of backpressure at 160 psi and flushing three pore volumes through the sample!  I finally realized that the carbon-based particles violate a basic assumption of soil mechanics: that the individual grains are incompressible.

Hi Focht3,
I don't know anything about the 100 year design life, but I thought your other comments pertaining to the sand composition were interesting and valid.  Those things should certainly be considered when examining the CIU test results for any soil.

The problem with not getting a good B value on the flue dust was also interesting.  Did you finally conclude that B=0.7 corresponds to "near saturation" for the dust and take it from there?  Was that testing for flue dust (waste) embankment design? Did you observe anything else unusual during the testing?  Just curious.
Mark

Hi Foch3.
Thanks for your comments and also interesting side notes. In my case the saturation was not a problem.I always got more than 95 percent saturation. Silty sand is without mica by only eyes observation. I did not performed any special test to check if sandy particles are silica or other minerals. 100 hundred years design life is for embankment as a structure, so we have to assure client that it will stay there for 100 years. Problem we have that they do not like geogrid reinforcement saying how to prove 100 years life.Probably are affraid of decaying of this material and even metioned the problem with fire on embankment and damaging geogrid. I think that this was more or less a joke.

Hi MRM.
Thanks for your comments.So far I did not performed the new CIU at lower confining stresses as the money is problem, but I will push for it and will inform you on results. I have chosen the stress level according to the embankment hight as you correctly assumed.
Also thank you for good ideas and discussion. I am going to open a new thread on Pile load test so you may have a look and comments are welcomed. Roughly explain, I am looking for information how to distinguish between plastic and elastic settlement after unloading the pile. Pile settlement was 52 mm on 1900 ton lading.Pile diameter 2 meters and installed in mudstone, sandy mudstone and sandstone stratas to depth 30m. After unloading there was residual settlement 44 mm. People say it was plastic deformation, but I am not sure. What is the percentage of residual settlement to say is plastic or elastic? Mudstone UCS is between 500 Kpa to 5 Mpa.