Probabilistic correlation between laboratory and field liquefaction potentials using relative state parameter index ( ξR)

Abstract

Cyclic triaxial tests on reconstituted sand samples are broadly employed in practice while they ignore the inherent characteristics of soil in field condition such as aging, fabric, and prior strain history. Relative state parameter index, ξ R , is utilized in a probabilistic framework to adjust the cyclic triaxial resistance ratio of sands at 15 uniform cycles ( CRR tx,15 ) to field condition. A wide-ranging database containing the results of cyclic triaxial tests conducted on reconstituted samples has been compiled to derive a correlation between relative state parameter index ( ξ R ) and triaxial cyclic resistance ratio. The adjustment coefficients proposed by researchers are employed to correct CRR tx,15 − ξ R relationship for actual field condition. The adjusted CRR tx,15 − ξ R relationships are applied to a database of field liquefaction case histories composed of both SPT and CPT based data and their performances in field condition are evaluated. It is demonstrated that constant triaxial-to-field adjusting coefficients cannot ever predict conservative results. Logistic regression method is employed to derive a field probabilistic criterion that obtains the likelihood of liquefaction initiation in terms of ξ R . The ξ R -based boundary curve standing for 20% likelihood of liquefaction initiation is found to be the most conservative limit state boundary to be used in field conditions. Finally, the triaxial and field CRR− ξ R relationships are composed and a probabilistic triaxial-to-field adjustment coefficient is proposed in terms of ξ R and a given liquefaction probability. It is anticipated that the proposed relationship could reasonably correct the results of cyclic triaxial testing on freshly reconstituted sand samples.