Published Online: 10 December 2012
Page Count: 10
Student, School of Engineering and Physical Sciences, James Cook Univ., Townsville, Queensland
Associate Professor, School of Engineering and Physical Sciences, James Cook Univ., Townsville, Queensland
(Received 3 April 2012; accepted 8 October 2012)
The coefficient of consolidation (cv) is traditionally determined by fitting observed settlement–time data to the theoretical average degree of consolidation versus time factor relationship developed by Terzaghi. Although it is widely accepted that different curve-fitting methods can produce different values of cv, very few comparisons have been conducted to assess the validity of these methods. In this study, the settlement–time data gathered from conventional oedometer tests conducted on three different clays were analysed using three common curve-fitting techniques: the Casagrande log-time method, Taylor’s root-time method, and the Cour inflection point method. A new method proposed by the authors for calculating cv, which abandons the traditional curve-fitting approach in favour of a computational-based approach, was also used to compare these results. To assess the validity of each cv value, the experimental results were compared with the theoretical average degree of consolidation curve and quantified using the root-mean-square (rms) error. The efficacy of the designated curve-fitting method was found to significantly depend upon the “shape” of the settlement–time curve generated during testing. For example, in this study, clay containing a significant fraction of fine sand often resulted in settlement–time curves that exhibited no clear inflection point, which made analysis using the Cour method very difficult. In general, the Taylor method predicted larger values of cv than the Casagrande method, and correspondingly smaller rms errors. The variance method proposed by the authors resulted in values of cv that more closely matched those generated using the Casagrande method. However, smaller rms errors were achieved using the variance method, which suggests that this technique may produce a more realistic estimate of cv than the Casagrande method.
Paper ID: GTJ20120038