Variations in Membrane Contact Patterns of Reconstituted Sand Specimens

Abstract

Abstract The penetration of the latex membrane into the peripheral voids of a sand specimen during application of confining stress can cause significant errors in subsequent triaxial test results. Analytical solutions are commonly used to account for this source of error by predicting the shape of the deflected membrane based on a sand particle-membrane contact pattern assumed to be represented by four spheres of diameter D50 whose centers form a square of length D50 on a plane parallel to the undeflected membrane. This paper presents the findings of a study that used image analysis to study actual contact patterns and compare these with the generally assumed configuration. The results show that the generally assumed constant configuration does not reflect the variability of actual patterns, which are a function of the specimen relative density and preparation method and may therefore grossly underestimate the amount of membrane penetration. The study shows that an increase in relative density results in a decrease in inter-contact distance, regardless of the specimen preparation method, and that the inter-contact distance at the sand-membrane interface is larger and more variable for moist tamped than air-pluviated specimens at any given density. These observations suggest that the sand-membrane contact patterns assumed by future analytical solutions, as a minimum, will need to take specimen relative density and preparation method into account if the amount of membrane penetration is to be more accurately predicted.