Premium
Estimating critical state soil mechanics parameters from constant cell volume triaxial tests
Author(s) -
Kirby J. M.,
O'sullivan M. F.,
Wood J. T.
Publication year - 1998
Publication title -
european journal of soil science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.244
H-Index - 111
eISSN - 1365-2389
pISSN - 1351-0754
DOI - 10.1046/j.1365-2389.1998.00142.x
Subject(s) - isotropy , soil mechanics , geotechnical engineering , critical state soil mechanics , mathematics , deformation (meteorology) , constitutive equation , compression (physics) , constant (computer programming) , compaction , triaxial shear test , mechanics , geology , structural engineering , engineering , computer science , soil water , soil science , physics , finite element method , thermodynamics , petrology , oceanography , quantum mechanics , programming language , shear (geology)
Summary Compaction, tillage, stresses around growing roots and other soil deformation events may be predicted by the critical state model of soil mechanics, but estimating the parameters is time consuming and expensive. We develop a back analysis of the constant cell volume triaxial test, in which the critical state parameters are derived from the results of a single test. This both saves much labour and provides more information than traditional analyses, which require several triaxial compression tests and an isotropic compression test to yield the same information. The method finds, using a minimization algorithm and a quasi‐analytical solution to the stress–strain equations, the simulated soil deformation (and hence the properties used in that simulation) that best fits the test data. The minimization is a form of regression analysis. For normally consolidated samples the method provides stable estimates of the slope of the critical state line ( M ), the slope of the virgin compression line ( λ ) and elastic modulus ( E ). The standard errors of the estimates are small in relation to the means of these parameters. The estimates appear to be more reliable than those of more commonly used estimation procedures. The slope of the rebound line ( κ ) is estimated, but a measure of the accuracy of the estimate cannot be calculated.