Characterization of geotechnical variability and Evaluation of geotechnical property variability: Discussion

The two papers by K.-K. Phoon and F.H. Kulhawy are very interesting. However, the Discussers would like to point out some considerations. In a preceding article (Cherubini 1997) one of the Discussers collected data regarding friction angle, plotting a graph of the mean coefficient of variation (COV). In this and a further article (Cherubini and Orr 1999) three fields of COV variability are evident, not including some “outliers”: 10–50% for clay, 5–25% for silt, and 5– 15% for sand. These values substantially confirm the results reported in Fig. 5 of the “Characterization” paper. Regarding the variability of cu, the Discussers think it is better to refer to the normalized value (cu /σ vo ′ ) instead of cu to avoid problems related to data detrending procedures. For the evaluation of trends, the Discussers think that it is necessary to examine the possibility of utilizing techniques like the EDA resistant line (see, for example, Velleman and Hoaglin 1981) which are not significantly influenced by outliers (Cherubini and Giasi 2000). Lastly, it is necessary to test transformation laws to find those capable of better “fitting” the measured values. It is then necessary to collect numerous data to be managed according to proper procedures for evaluating “precision” and “accuracy” of the law (Cherubini and Orr 2000). References