M.G. Smith

Doctor of Philosophy, Keble College, University of Oxford, Michaelmas Term 1993

A Laboratory Study of the Marchetti Dilatometer


The purpose of this study was twofold: to design, construct and commission a testing chamber for the calibration of in-situ devices in clay and to use the chamber to carry out a programme of research into the factors affecting the results of the Marchetti dilatometer test.

A calibration chamber system was developed which was capable of producing one metre high by one metre diameter cylindrical beds of clay. The preparation technique involved an initial phase of one-dimensional consolidation in a rigid tube, followed by a second phase of consolidation in a chamber with independent stress control. The stress control was achieved through water-filled flexible membranes in the side and at the base of the chamber. Reasonably uniform distributions of the water content and the undrained strength were obtained from investigations carried out in each clay bed after they had been tested.

The Marchetti dilatometer is an in-situ testing device the results of which have been interpreted chiefly through empirical correlations based on the results of field tests. The interpretation has mainly involved the use of two readings, and , though recently a third reading , has been introduced. Dilatometer tests were carried out in nine clay beds. The stress history and stress state of each clay bed were systematically varied in the test programme to allow their individual effects on the dilatometer readings to be assessed.

The study revealed that in clay the dilatometer reading is controlled by the undrained strength and the horizontal stress and was independent of the degree of overconsolidation. Through use of this result it was shown that the applicability of many of the empirical correlations presently used to evaluate the dilatometer readings is restricted to deposits with one-dimensional stress histories. It was also found that was dependent on the undrained strength and that the reading was close to the total horizontal stress of the sample though the latter finding does not appear to be supported by the few field results that are available. The results have highlighted the redundancy of the reading for property evaluation. It was found that a timed sequence of , and readings could be used to detect the variation of the consolidation properties between clay beds.

In addition to the tests in clay, the results of 31 dilatometer tests carried out in a calibration chamber for sand have also been analysed. An inter-relationship between , and was found suggesting that they were all measuring the same soil response; that of the pressure required to open a cavity in the sand. The reading was shown to be dependent on the horizontal stress and the state parameter, though this dependence could not be exploited to back calculate the horizontal stress. Overconsolidation of the sand specimens had no significant influence on the readings.

Thesis (13.61MB, pdf)

This thesis can also be downloaded from the ORA website