Implication of end restraint in triaxial tests on the derivation of stress–dilatancy rule for soils having high compressibility

Constitutive models for soils are developed and validated against laboratory tests assuming these give representative information on the true material behaviour. However, data from standard laboratory tests reflect the sample response rather than the true material behaviour, due to nonuniformities in stresses and strains generated over the experimental test. The work examines the implications of end restraint on the definition of the stress–dilatancy rule of highly compressible soils with a finite element numerical approach. The numerical model replicates a reconstituted peat, typically characterized by a combination of high compressibility and high friction angle, which increases the severity of end restraint effects. Simulated results show that the global measurements from standard triaxial tests with rough end platens would not give the proper stress–dilatancy rule, if they were interpreted as the response of a single soil element at the constitutive level. Both overestimation and underestimation of the true dilatancy compared to the material response can be observed, depending on the deformation mode. To support the validity of the numerical results, experimental findings from drained triaxial tests on reconstituted peat are presented. Practical indications are given on how the standard interpretation of drained triaxial tests data on peats can be improved.