Cyclic Properties of Natural Soil Containing Pumice Particles

Abstract

Sands containing pumice particles are commonly found in the Waikato Basin in the North Island of New Zealand. The pumice particles originated from a series of volcanic eruptions centred in the Taupo and Rotorua regions. As a result of flooding and erosion along the Waikato River, the pumice particles have become mixed with other materials and have been distributed over the Waikato Basin; these mixtures are referred to herein as natural pumiceous (NP) soils. Due to their vesicular nature and the presence of internal voids, pumice particles are highly crushable, compressible and lightweight, rendering them problematic from an engineering point of view. For the purpose of laboratory testing, undisturbed and reconstituted NP soil samples were obtained from four sites along the Waikato River. Small-block sampling and gel-push sampling were used to obtain undisturbed NP soil samples. In this study, a method is proposed to crush the pumice particles and then based on the degree of particle crushing the pumice contents of natural soils were estimated. To investigate the effect of relative density (Dr), pumice content (PC), soil fabric and structure, particle shape and crushing on the undrained cyclic behaviour of NP sands, a comprehensive set of undrained cyclic triaxial tests were performed on undisturbed and reconstituted NP sands and reconstituted Toyoura sand with different relative densities. It was found that the reconstituted NP specimens showed significantly different cyclic behaviour compared with hard-grained sand. For instance, during cyclic triaxial testing the NP sands underwent deformation from the start of cyclic loading with the axial strain gradually increasing to a double-amplitude axial strain of 5%. In contrast, Toyoura sand underwent negligible deformation for a significant number of cycles, followed by a sudden increase in deformation to reach 5% double-amplitude axial strain within a few additional cycles. In addition, the NP sands showed a very contractive behaviour under initial cyclic loading, changing to a very strong dilative behaviour after a few cycles. Due to the formation of a stable soil skeleton inside the NP specimens, the instability condition was not observed and the liquefaction resistance of the NP sands was found to be considerably higher than Toyoura sand with the same relative density. In addition, based on a limited number of tests, it appeared that slight differences in the behaviour of the NP soils was mostly dependent on their composition, such as their pumice content. Finally, under similar initial conditions, the undrained cyclic responses of undisturbed and reconstituted NP sands was significantly different, and the liquefaction resistance of the undisturbed NP specimens was higher than that of reconstituted ones, indicating soil structure (e.g. stress history, fabric cementation) to be an important factor in the cyclic behaviour of NP soils. Regarding post-cyclic monotonic response, three parameters — initial shear modulus (G1), shear modulus at recovery (G2) and recovery strain (ɣr) — were used to model the stress–strain behaviour of NP and Toyoura sands that had undergone liquefaction. Due to the formation of a stable soil skeleton during cyclic loading and the high angularity of the pumice particles, the liquefied NP soils recovered their strength at a considerably faster rate than the Toyoura sand; the ɣr value of NP sands was smaller than that of Toyoura sand. In addition, the G1 of liquefied NP sand was higher than that of Toyoura sand; however, the G2 of liquefied NP sand is lower than that of Toyoura sand. The post-liquefaction monotonic responses of undisturbed and reconstituted NP sands were found to be different, indicating that the inherent structure of the undisturbed samples was unaffected by cyclic loading.