The Torlesse composite terrane is an important geological unit in Canterbury, New Zealand, making up the backbone of the Southern Alps. It consists of a large group of rock that exhibits a range of engineering geological conditions. This study has been undertaken to characterise the range in engineering geological conditions throughout the Torlesse of Canterbury in order to develop a rock mass classification scheme specific to this abundant and complex rock type. The classification is aimed to aid in TBM tunnelling assessment in the Torlesse, which enables sub-division of an area or tunnel alignment into rock mass domains. Furthermore the classification enables the prediction of rock masses through geological controls in areas of poor outcrop coverage.

Four sites throughout Canterbury were selected for mapping to represent Torlesse terrane types, metamorphic facies and a range of regional fault settings: the Elliott Fault, Hurunui River, Ashley River Gorge and Opuha Dam. A preliminary desktop study was carried out with a landscape lineation analysis to develop 1) a conceptual geological model at each study site and 2) field mapping sheets to provide a check list to ensure consistency of information collected between outcrops and sites. Lineations and conceptual models identified a series of structural blocks within sites, which were further validated by field mapping. Outcrop field mapping was carried out across selected extents of study sites using the field sheets from the desktop study. Using NZGS (2005) and ISRM (1978) derived parameters, rock mass characteristics, including lithology and defect information, were recorded on the field sheets. A laboratory testing programme on selected outcrop intact rock was undertaken to support field work and later classification development.

Data from field work was plotted to derive rock mass trends. Trends were used to develop a classification framework. It was found the rock mass could be defined by bedding thickness, degree of fracture and the combination of discontinuities such as persistent jointing and shearing, which defined dominant rock mass control. The rock mass could therefore be classified based on: blockiness, defined by bedding thickness and density of non-systematic jointing (fractures); and defect structure, defined by the combination of systematic discontinuities such as persistent jointing and shearing.

The two principle rock mass governing controls were related together on an XY plot to form the conceptual Torlesse rock mass classification (TRC). Six classes encompassing the range of conditions observed in the Torlesse were devised for blockiness and defect structure. Blockiness classes range from: thickly bedded to massive sandstone with slight to moderate fracture, to very thin to thin bedded sandstone that is fragmented. Defect structure classes range from rock masses defined by: dominant systematic, persistent jointing with rare faulting, to rock masses typical of major shear zones, where material geotechnically behaves as a soil with no principle defect sets. Individual outcrop plotting then allowed rock masses typical of each site to be grouped on the TRC.

Clusters of each study sites’ outcrops were overlaid to characterise all rock mass types observed throughout this research. This allowed representative identification of eight distinctive rock mass types (Types 1-8) that are indicative of the Torlesse composite terrane of Canterbury. Each type has a series of geological controls that influence the nature of the rock mass. Geological controls can aid in the prediction of rock mass conditions for tunnel alignment selection.

Lithostructure and proximity to major structures were defined as major rock mass type controls. Lithostructure defines the effect of lithology on bedding thickness and fracturing by non-systematic jointing. Medium to massive bedding as part of rock mass Types 1 and 2 result in the best rock mass. In the sandstone-rich rock mass, systematic jointing dominates with less shearing and faulting and a lower occurrence of short, discrete, non-systematic jointing. Conversely, the thinly bedded Torlesse represented by rock mass Type 5 lacks persistent jointing. This type, being mudstone dominant, fractures more easily, is characterised by short, discrete jointing, and tends to localise faulting, shearing and some folding. Modern tectonic stress fields are also a major control. The size of the tectonic structure can impact different volumes of rock. Rock outside the direct fault zone can also be impacted giving rise to rock mass Type 6. For example, increased levels of shearing are observed in adjacent rock at both the Elliott and Opuha Dam Faults. Rock mass Types 7 and 8 represent the rock masses directly affected by large tectonic structures.

Sub-dividing proposed tunnel alignments by rock mass type allows assessment of tunnelling parameters. Dependant on project specific rock mass types expected, different TBM design will be suited. This has significant implications on support measures. Open gripper TBM’s are likely to be suited to rock mass Types 1 and 2. This rock mass is expected to represent the best rock mass stability but will be the hardest to excavate. As a result, rock bolt, mesh and shotcrete will likely prevent significant block failure through gravity release. Rock mass Types 3 and 4 are expected to represent a favourably interlocked rock mass, resulting in increased penetration rate but whose advance rate is likely to be hindered by the need for more extensive support. As rock mass Types 5-8 increase in abundance, shielded TBM’s will likely be best suited due to questionable thrust generation and support requirements toward the poorer rock masses. Penetration rates will be high but advance rates are expected to be low. Significant potential for failure exists in the poorer rock mass types without adequate support, including running ground. The selection of a shielded or gripper TBM will depend on the proportion and lengths of each TRC rock mass type anticipated along a tunnel alignment.

The opportunity exists for future work to refine and validate the TRC classification through increased data input, more extensive laboratory testing and its application to tunnelling projects. Furthermore it is hoped the TRC can be used for other types of geotechnical applications, at a variety of scales where Torlesse is concerned. To do this the TRC interpretations with respect to rock mass behaviour must be adapted to different scales.