It is important to obtain reliable estimates of potential problems as early as possible when tunnelling through heterogeneous rock masses, such as flysch. The tunnel designer can then select optimum routes and provide resources to investigate those areas in which problems are anticipated.
One of the problems in deep (more than 30m below surface) tunnels is squeezing. Engineers need a methodology by which to estimate the severity of potential squeezing and the range of solutions to be considered.
The first job is to estimate the strength and deformation properties of weak heterogeneous rock masses.
Hoek and Brown drew up criteria for estimating the strength and deformation characteristics of rock masses. However, any realistic criterion for estimating rock strength and deformation can be used, provided that the same process is used in deriving the final curves relating tunnel deformation to rock strength.
The geological model
Fookes’ geological model – whether used as conceptual, hand-drawn or a computer generated three-dimensional solid model – is the basic building block for the design of all major construction projects. A good model will enable geologists and engineers to understand the interactions of the components of the earth’s crust and so make rational engineering decisions. If there is no adequate geological model, ad hoc decisions have to be made and there is a high risk of construction problems due to unforeseen geological conditions.
In most developed countries, reliable regional geology maps exist and the geological libraries may contain more detailed maps where investigations have been carried out for resource development or other purposes. Consequently, the starting point of any tunnel route assessment should be a thorough literature survey.
This should be followed by a walkover survey in which topographic forms, rock outcrops and any other significant geological features are noted and used in the construction of the first geological model. Such a model, although still crude, may be adequate for comparison of alternative routes and for avoiding obvious problem areas such as landslides.
Once the route has been selected, the next step is the construction of an engineering geology model. This will almost certainly involve a diamond drilling programme in which the rock mass is explored at the depths of the proposed tunnel. On the basis of a carefully planned drilling programme and the crude geological model, it should be possible to build an engineering geology model that is sufficiently detailed for final tunnel design.
Estimation of rock mass properties
A critical step in the methodology is the selection of reliable rock mass properties that can be used, with the depth of the tunnel, to estimate the response of the rock mass to the stresses induced by tunnel excavation. Hoek and Brown’s criteria are the most widely used and can be adapted to weak heterogeneous rock masses.
This and other criteria assume that the rock mass behaves isotropically. In other words, while the behaviour of the rock mass is controlled by movement and rotation of rock elements separated by intersecting structural features, such as bedding planes and joints, there are no preferred failure directions.
These failure criteria should not be used when the rock mass consists of a strong blocky rock, such as sandstone, separated by clay-coated and slickensided bedding surfaces. The behaviour of such rock masses will be strongly anisotropic and will be controlled by the fact that the bedding planes are an order of magnitude weaker than any other features.
In such rock masses the predominant failure mode will be gravitational falls of wedges or blocks of rock, defined by the intersection of the weak bedding planes with other features, which act as release surfaces. However, if the rock mass is heavily fractured, the continuity of the bedding surfaces will have been disrupted and the rock may behave as an isotropic mass.
In applying the Hoek and Brown criteria to isotropic rock masses, three parameters are required for estimating the strength and deformation properties. These are:
Uniaxial compressive strength σci of intact rock
With heterogeneous rock masses it is extremely difficult to obtain a sample of intact core for testing in the laboratory. Practically every sample obtained from such rock masses will contain discontinuities in the form of bedding and schistosity planes or joints. The samples will probably also contain several of the component rock types that make up this heterogeneous rock mass. Consequently, any laboratory tests carried out on core samples will be more representative of the rock mass than of the intact rock components. Using the results of such tests in the Hoek-Brown criterion will impose a double penalty on the strength (in addition to that imposed by GSI) and will give unrealistically low values for the rock mass strength.
In some special cases, where the rock mass is closely jointed and where it has been possible to obtain undisturbed core samples, uniaxial compressive strength tests have been carried out directly on the ‘rock mass’ (Jaeger, 1971).
These tests require a high level of skill on the part of the driller and the laboratory technician. The large scale triaxial test facilities required for such testing are only available in a few laboratories in the world and it is not worth considering such tests for routine engineering projects.
One of the few courses of action that can be taken to resolve this dilemma is to use the point load test on samples in which the load can be applied normal to bedding or schistosity in samples. The specimens used for such testing can be irregular pieces or pieces broken from the core. The direction of loading should be as perpendicular to any weakness planes as possible and the fracture created by the test should not show signs of having followed an existing discontinuity.
Photographs of the specimens, before and after testing, should accompany the laboratory report, since these enable the user to judge the validity of the test results. The uniaxial compressive strength of the intact rock samples can be estimated, with a reasonable level of accuracy, by multiplying the point load index Is by 24, where Is = P/D². P is the load on the points and D is the distance between the points.
In the case of weak and/or fissile rocks such as clay-like shales or sheared siltstones, the indentation of the loading points may cause plastic deformation rather than fracture of the specimen. In such cases the point load test does not give reliable results.
Where it is not possible to obtain samples for point load testing, the only alternative is to turn to a qualitative description of the rock material to estimate the uniaxial compressive strength of the intact rock.
Constant mi
The Hoek-Brown constant mi can only be determined by triaxial testing on core samples or estimated from a qualitative description of the rock material as described by Hoek and Brown (1997). This parameter depends upon the frictional characteristics of the component minerals in the intact rock sample and it has a significant influence on the strength characteristics of rock.
When it is not possible to carry out triaxial tests, an estimate of mi can be obtained. Most of the values quoted have been derived from triaxial tests on intact core samples and the range of values shown is dependent upon the accuracy of the geological description of each rock type. For example, the term “granite” described a clearly defined rock type and all granites exhibit similar mechanical characteristics. Hence the value of mi is defined as 32 ± 3. On the other hand, the term “breccia” is not precise in terms of mineral composition and hence the value of mi is shown as 19 ± 5, denoting a higher level of uncertainty.
Influence of groundwater
The influence of groundwater on the behaviour of the rock mass surrounding a tunnel is important and must be taken into account in estimating potential tunnelling problems.
The most basic impact is upon the mechanical properties of the intact rock components of the rock mass. This is particularly important when dealing with shales, siltstones and similar rocks that are susceptible to changes in moisture content. Many of these materials will disintegrate quickly if they are allowed to dry out after removal from the core barrel. Hence, testing of the “intact” rock to determine the uniaxial compressive strength σci and the constant mi must be carried out under conditions that are as close to the in-situ moisture conditions as possible. Ideally, a field laboratory should be set up close to the drill rig and the core prepared and tested immediately after recovery.
In one example in which a siltstone was being investigated for the construction of a power tunnel for a hydroelectric project, cores were carefully sealed in aluminium foil and wax and then transported to a laboratory in which high quality testing could be carried out.
In spite of these precautions, the deterioration of the specimens was such that the test results were meaningless. Consequently, a second investigation programme was carried out in which the specimens were transported to a small laboratory about 5km from the exploration site and the samples were tested within about an hour of having been removed from the core barrel. The results of this second series of tests gave consistent results and values of uniaxial compressive strength sci and constant mi that were considered reliable.
When laboratory testing is not possible, point load tests should be carried out as soon after core recovery as possible to ensure that the moisture content of the sample is close to the in-situ conditions.
The second impact of groundwater is that of water pressure. This manifests itself in a reduction in the strength of the rock mass due to the reduction in stress acting across discontinuities. This “effective stress” effect is taken into account in the analysis of stress induced progressive failure surrounding the tunnel. Many numerical programs incorporate the capability for effective stress analysis and one of these programs should be used for the final tunnel design.
In many cases, the effective stress effects are not significant during construction since the tunnel acts as a drain, and the water pressures in the surrounding rock are reduced to negligible levels. However, if the groundwater conditions are re-established after completion of the final lining, the long-term effects of water pressure on rock mass strength should be investigated.
A final effect of groundwater occurs when high water pressures or flows are encountered during construction. This gives rise to practical construction problems and facilities for dealing with these problems should be provided in the contract. The practical issues of water handling in wet tunnels are not dealt with in this paper.
Geological Strength Index GSI
The geological strength index (GSI) was introduced by Hoek, Kaiser and Bawden (1995), Hoek and Brown (1997) and extended by Hoek, Marinos and Benissi (1998). It is based upon an assessment of the lithology, structure and condition of discontinuity surfaces in the rock mass and it is estimated from visual examination of the rock mass exposed in tunnel faces or surface excavations such as roadcuts and in borehole core.
The estimated GSI value of the rock mass is incorporated into calculations to determine the reduction in the strength of the rock mass compared with the strength of the intact rock components.
The term flysch, attributed to the geologist B Studer, comes from the German word “fliessen”, meaning flow, probably denoting the frequent landslides in areas consisting of these formations.
Flysch consists of alternations of clastic sediments that are associated with orogenesis. It closes the cycle of sedimentation of a basin before the “arrival” of the poroxysme folding process. The clastic material is derived from erosion of the previously formed neighbouring mountain ridge.
Flysch is characterised by rhythmic alternations of sandstone and fine grained (pelitic) layers. The sandstone may also include conglomerate beds. The fine-grained layers contain siltstones, silty shales and clay-like shales. Rarely and close to its margins, limestone beds or ophiolitic masses may be found. The thickness of the sandstone beds range from centimetres to metres. The siltstones and schists form layers of the same order but bedding discontinuities may be more frequent, depending upon the fissility of the sediments.
The overall thickness of the flysch is often large (hundreds to a few thousand metres). Different types of alternations occur in this thickness, such as persistence of sandstone or typical alternations or siltstone persistence. The overall thickness has often been reduced considerably by erosion or by thrusting. The formation is often affected by reverse faults and thrusts. This, together with consequent normal faulting, results in a degradation of the geotechnical quality of the flysch rock mass.
Selection of σci and mi for flysch
In addition to the GSI values, it is necessary to consider the selection of the other “intact” rock properties σci and mi for heterogeneous rock masses such as flysch. Because the sandstone layers are usually separated by weaker layers of siltstone or shales, rock-to-rock contact between blocks of sandstone may be limited. Consequently, it is not appropriate to use the properties of the sandstone to determine the overall strength of the rock mass.
On the other hand, using the “intact” properties of the siltstone or shale only is too conservative, since the sandstone skeleton contributes to the rock mass strength.
Therefore, it is proposed that a “weighted average” of the intact strength properties of the strong and weak layers should be used. Suggested values for the components of this weighted average.
Estimating rock mass properties
Having defined the parameters σci mi and GSI as described above, the next step is to estimate the mechanical properties of the rock mass. The procedure making these estimates has been described in detail by Hoek and Brown (1997).
It will not be repeated here. This spreadsheet can be used for shallow tunnels and slopes (less than 30 m depth) but, in the context of this paper, the values for “deep” tunnels will be used and a value of greater than 30 should be inserted for the depth below surface in the input data section.
The uniaxial compressive strength of the rock mass scm is a particularly useful parameter for evaluating potential tunnel squeezing problems.
It can be calculated directly from the spreadsheet (as was done in plotting the curves in the top graph) or it can be estimated by means of the following equation which gives an estimate of scm for selected values of the intact rock strength σci, constant mi and the geological strength index GSI :
σcm=(0.0034mi0.8)σci{1.029+0.025e(-0.1mi)}GSI (1.1)
In order to estimate the deformation of a tunnel subjected to squeezing, an estimate of the deformation modulus of the rock mass is required. This can be obtained from equation 1.2, originally published by Serafim and Pereira (1983) and modified by Hoek and Brown (1997).
E(GPa)√(σcI ÷100) .10 ((GSI-10)/40) (1.2)
The values obtained from equation 1.2 are plotted in the lower graph.
Related Files
Geological Strength Index
Table 4: GSI estimates for heterogeneous rock massxes such as Flysch
Table 3: Geological strength index for blocky jointed rock masses
Line Graph 1
GSI Estimates
Line Graph 2
