Search for: All records

This paper presents the development, implementation and experimental evaluation of a new crack detection mechanism for centrifuge modelling. The proposed mechanism is a brittle conductor bonded to cement providing a binary indication of if, and when, a sensor is cracked. The results of a pair of large centrifuge tests were used to evaluate the effectiveness of the proposed crack detection mechanism. Each test model included a soil profile consisting of a 23 m thick layer of lightly over-consolidated clay, underlain and overlain by thin layers of dense sand. The centrifuge models had two separate zones, a zone without reinforcement and a zone with an ‘embedded’ soil–cement grid, which had a unit cell area replacement ratio A r  = 24%. Models were subjected to 13 different shaking events with peak base accelerations ranging from 0·01 to 0·55g. The performance of the proposed crack detection mechanism was examined using (i) post-test crack mapping in the soil–cement grids, (ii) results of the crack detection system and (iii) time series of accelerations, displacements and footing rotation. The results from the centrifuge test showed that the proposed crack detection method accurately captured if, and when, cracking occurred in the soil–cement grid at the locations of the sensors. 

The effect of soil interlayering on the measured cone penetration resistance was examined in a layered soil model tested on a 9-m radius centrifuge. The soil profile consisted of a layer of sand between overlying and underlying layers of low plasticity clayey silt. The sand layer thickness varied from 0 to 240 mm (model scale) along the length of the model. The sand was loose with a relative density of 44% on one side of the model, and dense with a relative density of 88% on the other side. The clayey silt had a plasticity index (PI) of 6 and over-consolidation ratio (OCR) of about 1.5. Multiple cone penetration soundings were performed along the width and length of the model using cone penetrometers with diameters of 4, 6 and 10 mm. The model construction procedure, data processing, and cone penetration testing results are described.