Discussion of “Accuracy of determining pre-consolidation pressure from laboratory tests”

Abstract

The authors (Umar and Sadrekarimi 2017) present results from an interesting experimental study to validate the accuracy of different methods available for interpreting pre-consolidation pressure ( p ) from oedometer testing. For this purpose, laboratory oedometer tests were performed on block samples taken from three Canadian clays and specimens were subjected to cycles of one-dimensional compression loading and unloading.The authors are to be commended for providing a high-quality database of laboratory consolidation tests. The discusser would, however, like to raise the following points in relation to the paper.1. The authors provided brief summaries (in Table 1) for the 11 graphical methods of interpreting pre-consolidation pressure, one of which was referred to as "Boone (2010)". Boone's method, however, has been oversimplified by the authors so that the method described in the paper does not match the instructions given by Boone (2010). More specifically, according to Boone (2010), p is the intersection point obtained from the following two lines: (i) a line coincident with C cmax (maximum value of the slope of the virgin compression segment of the void ratio (e)-log( v ) curve) and (ii) a line parallel to C r (average slope of the unload-reload cycle of the recompression segment of the e-log( v ) curve) passing through the point ( v0 , e v0 ) where v0 and e v0 are the in situ vertical effective stress and void ratio at in situ vertical effective stress, respectively. The method described and used by the authors (as the intersection of tangent lines to the recompression and virgin compression segments of the e-log( v ) curve) seems to be appearing for the first time in the literature in Zeevaert (1973) and thus it will be referred to as " Zeevaert's (1973) (2005), intercomparisons of average model-performance error should be based on the mean absolute error (MAE). The overall accuracy of models is usually assessed using model performance measures that quantify the agreement along the identity line (perfect prediction line). The widely used model performance measures in geotechnical engineering are the coefficient of determination (R 2 ), coefficient of efficiency (E), and the ranking index (RI). Coefficient of efficiency, proposed by Nash and Sutcliffe in 1970 (Nash and Sutcliffe 1970), has several advantages over R 2 (see, e.g., Oommen and Baise 2010) and should be used instead of R 2 in assessing model performances (Kootahi and Moradi 2017). Ranking index, proposed by Briaud andTucker in 1988 (Briaud andTucker 1988), is an overall index of model performance and takes into account both the accuracy and precision of a model to assess its ability in providing acceptable predictions. These measures are defined as follows:where n is the total number of measurements, P i is the predicted value, M i is the measured value, M is the mean measured value, is the mean of the series of analyzed data, and is the standard deviation of the series of analyzed data. The range of E values lies between −∞ and 1 with a valu...

Reply to the discussion by Kootahi on “Accuracy of determining pre-consolidation pressure from laboratory tests”

“…As these calculations (as well as those presented in the paper) include consolidation test data from several other studies (Becker et al 1987;Grozic et al 2003Grozic et al , 2005Jose et al 1989;Sridharan et al 1991), the values of MAE, E, and RI as well as the corresponding rankings differ from those reported by the discusser. This is also indicated by the smaller number of analyses in Table D1 of the discussion (Kootahi 2017) compared to those in Table R1. Nevertheless, the overall performance rankings based on the RD index are very similar to those from RMSE values.…”