CBR vs. R-Value 1

[howardoark]

I have a fairly fundamental question about pavement design which I'd like to get some input on. The CBR is conducted on a soaked sample of subgrade material at a specific density - the sample is compacted to whatever you're going to require in the field (say 100% standard proctor) and then soaked because no one can prevent a pavement subgrade from becoming saturated at some point.

In R-Value testing, you make up 4 different samples at 4 different densities and moisture contents and then compress them until they're saturated (saturation evidenced by water being squeezed out of them). The R-Value reported is the R-Value that corresponds to water being squeezed out at 300 psi pressure.

I'm looking at a result now that has an R-Value of 24 at an initial dry density of 107.4 pcf (17.6% moisture content 800 psi exudation pressure) and an R-Value of 2 at an initial dry density of 95.9 pcf (27.3 % moisture content, 165 psi exudation pressure). The reported R-Value is 4.

I can see that using an R-Value of 4 is conservative and likely to represent the worst case. But if I'm going to use the worst case anyway, what's the point in requiring any compaction of the subgrade (other than my earnest desire to spend as little time in the company of lawyers who aren't my clients as possible)? It seems to me that the CBR represents a testing of actual conditions and the R-Value represents some theoretical worst case. Am I missing something important? Not that it matters because I'm in California and have to base designs on R-Value.

And how do they come up with correlations between CBR and R-Value when CBR can be based on anything from 80% of standard proctor to 105% of modified proctor?

Thanks

Jeff

 

[emmgjld]

Your post contains several misunderstandings of both the general pavement design process and concepts behind both the CBR and the Hveem-Carmany (R-Value) tests.

The design of a pavement section (AASHTO) assumes/requires the subgrade be sufficiently compacted or otherwise improved as to be a relatively consistent 'layer' within the section.

The CBR & Hveem-Carmany (R-Value)tests are two very different animals. The CBR utilized a 'punching' shear failure and the Hveem-Carmany uses material deformation.

Due to the methods of sample preparation and the final testing, the argument can be made that final subgrade performance is better approximated by the Hveem-Carmany method.

My own experience is the CBR is not appropriate for some soils conditions and the Hveem-Carmany method, when all aspects of the test results are considered, is appropriate with the more difficult soils.

After 35+ years, I can make either test method work, in the soils I am familiar with. Proper use of the Pavement Design \method (AASHTO, in my case) is helpful. Engineering is NOT 'Plug and Crank'.

 

[howardoark]

Thanks.

Would you please expand on this:

My own experience is the CBR is not appropriate for some soils conditions and the Hveem-Carmany method, when all aspects of the test results are considered, is appropriate with the more difficult soils.

 

[howardoark]

I mean, please expand on it if you can.

 

[emmgjld]

I apologize, been real busy for a week. Always dangerous to take a vacation. The Catch up is a bear.

The Hveem, when properly interpreted, seems to do a good job picking out unstable soils, such as Silts and well sorted Sands (including many mill tailings). These soils, when confined, will exhibit a lot of strength, giving moderate to high CBR values and Hveem 'R Values'. When unconfined (such as roads with narrow or soft shoulders and associated borrow pits), these materials roll and easily displace under repeated traffic loading. The Hveem testing and computation includes a Displacement Value which, if over 4.00 and in the case of Silts or Sands in my part of the world, indicates a distinct potential for lateral movement. Some means of additional stability needs to be considered, oftentimes in spite of a moderate to moderately high 'R Value'.

I have also found the Displacement Value to be useful when evaluating Aggregate Base materials. Some of these tend to 'roll' and are not as stable as assumed in the normal design process.

Another condition includes Silty Clay with a high Sulfate and/or sodium content. these soils will also exhibit an exceptionally high Displacement Value (often over 5.0 and as high as 6.5).

Remember, the Hveem-Carmany test is basically a triaxial test, with the emphasis on soil deformation, rather than actual soil shear. The final 'R-Value' is only part of the information this test is providing.

 

[howardoark]

I discussed this issue with the head of one of Caltrans's geotechnical laboratories. He said that CBR is more of a shear strength test and R-Value shows how susceptible a soil is to rutting.

He didn't have a good answer for why a soil compacted to 105% of modified Proctor 3% dry of optimum had the same R-Value as the same soil compacted to 85% of standard proctor 5% wet of optimum. I don't think there is a good answer to that question as those soils, in my experience, will behave entirely differently. I think we use R-Value because it is very conservative and roads don't fail when we design based on it.

I have a feeling, though I can't support this, that CBR designs are more rigorous, but they lead to more failures because soils are i) heterogeneous and ii) it's difficult to get uniform compaction over linear projects 25 miles long. It isn't obvious to me that using overly-conservative design methodologies save society money in the long run compared to using rigorous methods with strict quality control.

 

[emmgjld]

Two quotes from your last post goes to the heart of my discussion.
"...discussed this issue with the head of one of Caltrans's geotechnical laboratories. He said that CBR is more of a shear strength test and R-Value shows how susceptible a soil is to rutting."
and
"I think we use R-Value because it is very conservative and roads don't fail when we design based on it."

I generally agree the R-Value deals with rutting. After 35 years, I am convinced that is PRECISELY what most road sections should be designed for. I do not believe the Hveem-Carmany method presents a very conservative method. Rather, I think it actually presents the right method, based upon present knowledge.I believe the CBR method is not correct in many circumstances, and is therefore UNconservative.

An advantage to the CBR method is that it is rather inexpensive to buy the equipment and to run. Not always the best criteria for design purposes.

 

[howardoark]

I can't say that you're wrong, but I disagree with you.

As far as I can tell, R-Value is an index test. There has to be a better (i.e., more economical) method for designing roads based on the as-constructed properties of the subgrade.

 

[BigH]

http://www.mincad.com.au/Arrb98Paper/Arrb98Paper.htm

http://www.mincad.com.au/APSDSPaperChicago/ChicagoPaper.htm

http://www.mincad.com.au/circly.htm

http://www.mainroads.qld.gov.au/web...ing Amendment 2/$file/PDM_ManualIncAmend2.pdf

 

[emmgjld]

BigH, Thanks for the links. The last one

http://www.mainroads.

...does not work for me.
In one respect, the papers and info from MINCAD make my point. The CBR is 'calibrated' to match reality and rutting is the major problem to avoid.

Of Course, calibration is required for any test method. I just feel the HVEEM method is better suited for pavement & subgrade evaluation. I may be like concrete, All mixed up & set in my ways.

 

[jheidt2543]

To get the last link to work you will have to highlight the enitre line all the way to .pdf and then paste it into your broweser. It worked for me that way.

 

[howardoark]

If there's nothing better, you have to use it. Still, you should hope someone comes up with something better.