homogeneity 2

[AsbAli]

I try to find a definition for homogeneity of granular soils and here are my questions:

1- Is there any kind of definition?

2- How can I know if a constructed soil of a dam body (for example the soil is a widely graded soil) is homogeneous?

3- If there is a definition, how I can quantitavely estimate the homogeneity of soil at each point (or more exactly area) in the field.

4-I think, the common method is to take sampls and compare them, ( for instance, comparing the grain size distribution) but if I have two different grain size distribution, how can I estimate the degree of homogeneity?

I would really appreciate if anyone can give me some tips.

 

[dgillette]

Homogeneous soils don't exist in nature. Even very uniformly graded soils vary somewhat from point to point horizontally but especially vertically. The best we ever see is something that we can approximate as homogeneous in the analysis without affecting the results of the analysis too much, and that is site-specific. In your embankment dam example, you may be concerned with seepage, for which permeability is very sensitive to small changes in grain sizes (and the permeability is guaranteed to be anisotropic because of variations in the borrow pit and compaction). Or, you may be concerned with uniformity of the filter material, for its ability to act as a filter throughout the dam.

I don't know of any universally applicable procedure for quantifying homogeneity. You can do statistical analysis of differences in grain size, CPT, PI, etc., and look for a small standard deviation, but that won't tell the whole story because of the importance of spatial distribution, vertical and horizontal.

There, I have given you a vague answer that basically translates as "You will have to figure it out based on the specifics of your problem."

 

[Ron]

Agree with dgillette....homogeneity doesn't really exist. It is generally an assumption made, based on engineering judgment, for computational purposes. Homogeneity is related to consistency. If soils are of a consistent classification (i.e., SP, SP-SM, SP-SC, etc.) and there is a record of consistency of gradation, moisture-density relationships, unconfined compressive strength, direct shear or triaxial shear strength, then you are likely OK to assume homogeneity for most purposes.

Engineered fill is one instance where the assumption of homogeneity is usually appropriate.

 

[AsbAli]

Thanks a lot for the answers. I really appriciate it.

Imagine you have a grain size distribution consisted of just two fractions (D[sub]0[/sub]=0 ; D[sub]50[/sub]=1mm ; D[sub]100[/sub]=2mm) and if you consider a vertical segregation (for instance all of grains with 2mm diameter is on the top of the grains with 1 mm diameter) and homogenous material a very well mixture of these two grains, you can admit that the coefficient of uniformity or grain size distribuion can not give you any information about the homogeneity. Even the permeability is an average value that for both situations will give the same result and if we consider the third dimension it will be more abstract.

I like this sentence: "The best we ever see is something that we can approximate as homogeneous in the analysis without affecting the results of the analysis too much, and that is site-specific. [dgillette]" The problem is that for labratory experiments, it will be a good definition but in situ it seems to be impossible that we analyse the whole embankment to see if we have homogeneity.

it is really pitty that we can go to the mond but we are unable to define the homogeneity of soil.

Thanks

 

[BigH]

There is also the problem of placing materials (if fills) in a homogeneous fashion. It is always in the specs but hard to do. Take a grader leveling off a subbase layer with sizes up to 75 mm (and bigger in the other direction of the particle) - and they have to grade to plus/minus 10 mm. When you watch this, you will undoubtedly find that the "big stones" get graded off to the side . . . hence while your gradations of the "stock" may be similar from sample to sample, the placement factor is hard to control.