Reappraisal of the ASTM/AASHTO standard rolling device method for plastic limit determination of fine-grained soils

Abstract

Given its apparent limitations, various attempts have been made to develop alternative testing approaches to the standardized rolling-thread plastic limit (PLRT) method (for fine-grained soils), targeting higher degrees of repeatability and reproducibility. Among these, device-rolling techniques, including the method described in ASTM D4318/AASHTO T90 standards, based on original work by Bobrowski and Griekspoor (BG) and which follows the same basic principles as the standard thread-rolling (by hand) test, have been highly underrated by some researchers. To better understand the true potentials and/or limitations of the BG method for soil plasticity determination (i.e., PLBG), this paper presents a critical reappraisal of the PLRT–PLBG relationship using a comprehensive statistical analysis performed on a large and diverse database of 60 PLRT–PLBG test pairs. It is demonstrated that for a given fine-grained soil, the BG and RT methods produce essentially similar PL values. The 95% lower and upper (water content) statistical agreement limits between PLBG and PLRT were, respectively, obtained as −5.03% and +4.51%, and both deemed “statistically insignificant” when compared to the inductively-defined reference limit of ±8% (i.e., the highest possible difference in PLRT based on its repeatability, as reported in the literature). Furthermore, the likelihoods of PLBG underestimating and overestimating PLRT were 50% and 40%, respectively; debunking the notion presented by some researchers that the BG method generally tends to greatly underestimate PLRT. It is also shown that the degree of underestimation/overestimation does not systematically change with changes in basic soil properties; suggesting that the differences between PLBG and PLRT are most likely random in nature. Compared to PLRT, the likelihood of achieving consistent soil classifications employing PLBG (along with the liquid limit) was shown to be 98%, with the identified discrepancies being cases that plot relatively close to the A-Line. As such, PLBG can be used with confidence for soil classification purposes.