Load Arching and Uplift Failure around Rock Anchors: Experimental investigation on rock anchor failure modes, load transfer mechanisms and load capacity

Abstract

Rock anchors play an important role in geotechnical engineering, providing stability and reinforcement to large-scale infrastructures. A comprehensive literature review revealed that the knowledge and design practices of rock anchors have remained relatively unchanged since the 1970s, indicating a lack of progress in developing the design criteria. Four potential failure modes of rock anchors have been identified: failure in the anchor steel, failure in the anchor–grout bond, failure in the grout–rock bond, and failure along a conical surface in the rock mass. Among these, the design against the last two failure modes is considered to be the least satisfactory. The primary objective of this research is to enhance our understanding of the load transfer and failure mechanisms in rock anchors, with a specific focus on failure in the rock mass and the grout–rock bond. The goal of the research is to provide knowledge which can help improve the design, reliability, and overall performance of rock anchors.

The thesis presents a comprehensive experimental investigation on the load transfer and failure mechanisms of rock anchors. In the research, two specifically designed laboratory testing rigs were used, 18 full-scale failure tests were done in the field, and a series of numerical models were developed and analysed. The tests were designed to investigate the bond failure of the interfaces and the uplift failure of the rock mass, which gave insights into the load transfer mechanisms of rock anchors and their failure modes.

The main findings of the research can be summarised as follows:

The bond strength of the anchor–grout and grout–rock interfaces were investigated. Along the anchor–grout interface, the shear stress was highest at the top of the grout column and gradually decreased towards the distal end of the anchor during loading. The bond shear strength at the anchor–grout interface was approximately 20% of the UCS of the grout with the used test configuration. Bar anchors with an endplate were used to test the grout–rock interface. The shear stress distribution along the grout–rock interface was highest at the endplate and decreased towards the top of the grout column before slip occurred. The bond shear strength at the grout–rock interface was estimated to be around 5% of the unconfined compressive strength (UCS) of the grout or the rock, whichever was weaker, in the tested rock mass.

Two-dimensional uplift tests of small-scale block models demonstrated the formation of load arches within each individual layer of blocks. Load arches was the most compressed part of the joints. The load arches transferred the anchor load to the side frames of the testing apparatus. The load capacity of the block models was found to increase proportionally with increasing confining stress and greater depth, represented by an increased number of block layers. Notably, the load arching was also observed in discontinuous numerical modelling of the tests. The numerical models demonstrated that load arching was achieved through block rotations and displacement. Furthermore, the capacity of the arches increased with higher confining stress.

The influence of joint patterns on the strength and capacity of the rock mass was investigated through large-scale laboratory tests. The testing demonstrated that the joint orientation and block size impact the rock mass capacity and apex angle. The capacity of the block models increased with increasing in-situ stresses. An increase in horizontal stress was measured for all patterns with horizontal and vertical joints, which was interpreted as the formation of load arches in these models. The failure followed the joints in all the tested patterns. The influence of the joints on the rock mass uplift capacity and failure mode demonstrated the need for considering them in the design criterion of rock anchors. Full-scale uplift field tests also demonstrated that the observed failure crater followed the joint sets. Additionally, the field tests showed that the current design methods significantly underestimate the uplift capacity of medium hard rock masses.

The numerical modelling of the large-scale laboratory models demonstrated the formation of load arches in rock masses under anchor loading. The load arches were observed through the trajectory of the major principal stress. The arching was symmetrical in the models with vertical and horizontal joints, but asymmetrical when the joints had an angle relative to the loading direction. The numerical modelling was expanded to large-scale anchors using equivalent material properties found from calibrating the model against the results from the laboratory. The large-scale models demonstrated that the uplift capacity of a single anchor within a row was reduced when the load arches of adjacent anchors intersected.

Based on these findings, several recommendations for further work were proposed. It was suggested to incorporate joint orientation and spacing in the design criteria to establish a more accurate criterion against rock mass uplift failure. Further testing in various rock masses was recommended to determine the relationship between the rock mass capacity and joint characteristics. It was recommended to establish a database for all documented uplift failure tests and observed uplift failures, which could be used to develop a statistically accurate design method against rock mass uplift failure when a large enough amount of data has been included.