Settlement - Conventional/Log (Cr, Cc, eo) vs. Linear (mv) Pressure Scales 1

[geomane]

I just recently read "Geotechnical Engineering in Residual Soils" and based on the statements made in the text, settlement analyses on residual soils are best performed using the coefficient of compressibility (mv). The author is not a fan of working with consolidation test data when plotted on the log scale. It is actually an interesting read.

I am currently working on a project with residual soils and decided I would plot the consolidation data on both a log and linear pressure scale. I ran the settlement analyses using the conventional inputs (Cr, Cc, eo, max past pressure, etc.) and estimated a max total settlement between 0.75 and 1 inch. I then performed another settlement analyses using the coefficient of compressibility (mv) and estimated the settlement to be about twice as much, 1.75 to 2 inches. I know the coefficient of compressibility changes based on stress level; however, my plots indicate the slope to be relatively straight within the working stress range.

For those of you that routinely use mv to estimate settlements, do you typically estimate larger settlements than when using the conventional method?

 

[EireChch]

You should get the magnitude of settlement with Mv and Cc methods. Plot you data graphs and lets have a look.

 

[fattdad]

I think I'd use modulus for residual soils. There are a few exceptions, like the epi-Karst elastic silts. Then I'd use Cc, Cr, Cv, Pp and Po.

f-d

ípapß gordo ainÆt no madre flaca!

 

[geomane]

See attached spreadsheet. Let me know if you have any questions.

 

[GeoEnvGuy]

I took a quick look at your data and have the following comments.

The recompression loop occurs prior to the normally consolidated portion of your test. I don't believe this is the best way to get your recompression curve.

The second and fourth test appear to be highly disturbed due to the quality of the chart in e log sigma space. I would see how the plot looks in Log (1+e) vs Log sigma space and follow the Oikawa method for preconsolidation pressure determination.

For your Cc determination, Schertman 1953 says to run the test to 0.4eo and adjust the preconsolidation pressure void ratio up to the NC eo or estimate the OC adjusted eo by following Cr from insitu sigma to preconsolidation sigma. Likely your lab equipment wont be able to get to 0.4eo but you can still run the trend out to that point and construct the adjusted Cc for disturbance. See below photo from Das. Was this considered?

 

[geomane]

@GeoEnvGuy,

I used the strain energy method to determine the preconsolidation pressures. And yes, I plotted the field consolidation curves to get Cc, but my stresses from foundation loads plus insitu will fall below the pre-consolidation pressure.

 

[geotechguy1]

I think Janbu has also expressed similar views on the topic. Putting it into historical context, the log scales help make things look linear and easier to calculate, at a time when hand calculations was all that was available. A modulus based method - either non-linear, or linear (but making sure you are selecting an appropriate stress range) is perfectly acceptable, especially in a world of computerized calculations, and spreadsheets, and, as Wesley notes, for some soils the log scale can be extremely misleading.

 

[BigH]

Actually when I have done so, plotting log(e) vs Log(p) gives interesting results - seems to pick up the "preconsw Lawrence Wesley?

An old adage I was told by an old-timer. In consolidation testing, if you get the result within 30% you've had a good day - and in clays it will take a good many years to learn if you are right.

 

[LRJ]

I don't know why anyone would use the m[sub]v[/sub] approach - you need numerous different values depending on the stress/strain level. With the C[sub]c[/sub]/C[sub]r[/sub] approach you just need those two values. Both approaches should give the same answer, so the decision should be about using the simplest of the methods - which is clearly the C[sub]c[/sub]/C[sub]r[/sub] approach.

 

[geotechguy1]

The dependence of stiffness on strain level doesn't go away just because someone stuck it on a log plot and drew a straight line. It's still there, you just chose the stress / strain levels on a log plot.

 

[LRJ]

The point is that the C[sub]c[/sub]/C[sub]r[/sub] approach is a simpler approximation. Too often I have seen engineers use a single value of m[sub]v[/sub].