GT1R1A_Caracterisation of Rock Masses.pdf

AFTES GUIDELINES FOR

CARACTERISATION OF ROCK MASSESUSEFUL FOR THE DESIGN AND THE CONSTRUCTION

OF UNDERGROUND STRUCTURES

2 TUNNELS ET OUVRAGES SOUTERRAINS - N° 177 - MAI/JUIN 2003

Draft submitted by Jean-Louis GIAFFERI (EDF) - Chairman GT 1

With assistance in drafting this document from Alain AMELOT (SPIE Batignolles) - Daniel ANDRE (SNCF) - François BERBET (BOUYGUES TP)

Philippe BOUSQUET-JACQ (EURISK) - Stéphane CURTIL (TERRASOL) - Jean-Louis DURVILLE (CETE Rhône-Alpes)Denis FABRE (CNAM) - Jean-Alain FLEURISSON (Ecole des Mines de Paris - CGI) - Bernard GAUDIN (SCETAUROUTE)

Mehdi GOREYCHI (NERIS) - Françoise HOMAND (ENSG Nancy) - Gilles PARADIS (SNCF) Jean PIRAUD (ANTEA) - Alain ROBERT (CETU) - Philippe VASKOU (GEOSTOCK)

Christophe VIBERT (Coyne et Bellier) - Françis WOJTKOWIAK (INERIS)

Special thanks are due to Bernard GAUDIN for shouldering the heavy burden of writing the many drafts needed before arriving at this final report.

Thanks are also due to the proof-readersPierre DUFFAUT and Bernard GODINOT (GTM), Jean LAUNAY (VINCI Construction),Yann LEBLAIS (EEG SIMECSOL)

A.F.T.E.S. welcomes comments on this paper

Version 1 – Approved by technical Committe 29/04/2003

In 1978, AFTES issued its first Recommendations on the description of rock masses, using the following approach:

• Describe in detail all factors potentially influencing the stability of underground structures

• Classify field conditions with respect to each individual factor separately, without attempting to link them together.

This new version retains this basic principle, but with important additions.

Firstly, we make a clear distinction between the characterisation of (i) the rock matrix, (ii) discontinuities, and (iii) the rock masstaken in its entirety, dealt with in the three main chapters forming the backbone of the Recommendations.

We have placed the description of the influencing factors within the general context of the geotechnical survey. This description isdependent on a geological model. This model is made up of ‘homogeneous sub-units’ displaying the relevant characteristics. Wewere also concerned about the transition from instrumental and field data to data values used in the design analyses.

Lastly, we took the decision – and risk – of presenting general rock classification systems, not in order to advocate their systematicuse but to point out their limitations. Despite their apparent convenience, these classification systems (and more importantly, thecorrelations drawn from them) simplify to an outrageous degree a reality which is always complex. They can never be a substitutefor abundant observation, measurement and testing, and one must always bear in mind the value of the parameters on whichthey are based throughout the whole design and construction process.

The descriptive approach recommended here by AFTES applies not only to structural stability analysis but equally to the selec-tion of location, cross section and construction method. It is not confined to tunnels only, and these Recommendations may bethought useful for other types of rock engineering.

Jean PiraudChairman, AFTES Technical Committee

PREFPREFACEACE

3TUNNELS ET OUVRAGES SOUTERRAINS - N° 177 - MAI/JUIN 2003

Caracterisation of rock masses useful for the design and the construction of underground structures

PagesPages

1 - INTRODUCTION - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 41.1 - PURPOSE OF ROCK MASS CHARACTERISATION - - - 41.2 - GEOLOGICAL MODEL- - - - - - - - - - - - - - - - - - - - - - - - 4

1.2.1 - Geological survey - - - - - - - - - - - - - - - - - - - - - - - - 41.2.2 - Geological model - - - - - - - - - - - - - - - - - - - - - - - - 4

1.3 - GEOTECHNICAL CHARACTERISATION OF SUB-UNITS 4

2 - MATRIX CHARACTERISTICS- - - - - - - - - - - - - - - - - - - - - 52.1 - IDENTIFICATION PARAMETERS- - - - - - - - - - - - - - - - - 6

2.1.1 - Common names - - - - - - - - - - - - - - - - - - - - - - - - - 62.1.2 - Petrography and mineralogy - - - - - - - - - - - - - - - - 62.1.3 - Alteration of the minerals in the rock matrix - - - - - 62.1.4 - Densities (French standard P 94-410-1/2/3)- - - - - 62.1.5 - Volume weights - - - - - - - - - - - - - - - - - - - - - - - - - 72.1.6 - Moisture content (French standard P 94-410-1)- - 72.1.7. Porosity (French standard P 94-410-3) - - - - - - - - - 72.1.8 - Degree of saturation - - - - - - - - - - - - - - - - - - - - - - 72.1.9 - Permeability - - - - - - - - - - - - - - - - - - - - - - - - - - - - 72.1.10 - Ultrasound wave velocity (French standard p 94-411) - Continuity index - - - - - - - - 7

2.2 - MECHANICAL PARAMETERS- - - - - - - - - - - - - - - - - - - 72.2.1 - Deformability: instantaneous behaviour- - - - - - - - 82.2.2 - Deformability: time-dependent behaviour

related to creep - - - - - - - - - - - - - - - - - - - - - - - - - 82.2.3 - Time-dependent behaviour related to swelling - - 92.2.4 - Mechanical strength - - - - - - - - - - - - - - - - - - - - - - 92.2.5 - Triaxial test and failure criteria - - - - - - - - - - - - - - - 102.2.6 - Parameters for resistance to excavation - - - - - - - - 112.2.7 - Other tests - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 12

3 - CHARACTERISTICS OF DISCONTINUITIES - - - - - - - - - 123.1 - JOINT IDENTIFICATION PARAMETERS - - - - - - - - - - - 12

3.1.1 - Types and origins of the discontinuities - - - - - - - - 123.1.2 -Description of discontinuities - - - - - - - - - - - - - - - - 13

3.2 - CHARACTERISATION OF JOINT SYSTEMS- - - - - - - - - 133.2.1 - Directional joint set patterns - - - - - - - - - - - - - - - - 133.2.2 - Statistical analysis of geometrical parameters

for each joint set- - - - - - - - - - - - - - - - - - - - - - - - - 143.2.3 - Overall joint density indexes - - - - - - - - - - - - - - - - 15

3.3 - MECHANICAL PARAMETERS OF DISCONTINUITIES - 173.3.1 - Deformation parameters- - - - - - - - - - - - - - - - - - - 173.3.2 - Shear strength parameters - - - - - - - - - - - - - - - - - 173.3.3 - Hydraulic parameters - - - - - - - - - - - - - - - - - - - - - 18

4 - CHARACTERISTICS OF ROCK MASS - - - - - - - - - - - - - - 194.1 - IDENTIFICATION PARAMETERS- - - - - - - - - - - - - - - - - 19

4.1.1 - RQD- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 19

4.1.2 - Degree of alteration - - - - - - - - - - - - - - - - - - - - - - 194.1.3 - Rock mass continuity index ICM - - - - - - - - - - - - - 19

4.2 - MECHANICAL PARAMETERS- - - - - - - - - - - - - - - - - - - 204.2.1 - Rock mass deformability, rock mass deformation

modulus EMas - - - - - - - - - - - - - - - - - - - - - - - - - - 204.2.2 - Rock mass limit strength - - - - - - - - - - - - - - - - - - - 21

4.3 - HYDROGEOLOGICAL CONDITIONS - - - - - - - - - - - - - 224.3.1 - Identification of aquifers - - - - - - - - - - - - - - - - - - - 224.3.2 - Measurement of initial piezometric conditions - - - 224.3.3 - Measurement of rock mass permeability KM - - - - 224.3.4 - Gas - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 234.3.5 - Other parameters - - - - - - - - - - - - - - - - - - - - - - - - 23

4.4 - INITIAL STATE OF STRESS IN ROCK MASS - - - - - - - - - 234.4.1 - Initial state of stress and approximations - - - - - - - 234.4.2 - Characterisation of stress tensor - - - - - - - - - - - - - 244.4.3 - Commentary on field test methods - - - - - - - - - - - 244.4.4 - Classification of stress states - - - - - - - - - - - - - - - - 25

4.5 - TEMPERATURE - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 254.5.1 - Geothermal parameters - - - - - - - - - - - - - - - - - - - 254.5.2 - Methods for estimating temperatures for

underground structures - - - - - - - - - - - - - - - - - - - 25

5 - USE OF ROCK MASS CHARACTERISATION FOR UNDERGROUND STRUCTURE STABILITY ANALYSIS AND CONSTRUCTION - - - - - - - - - - - - - - - - - - - - - - - - - - - 255.1 - CHARATERISTIC VALUES AND PARAMETERS FOR DESIGN- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 25

5.1.1 - Individualisation of sub-units - - - - - - - - - - - - - - - - 255.1.2 - Geotechnical characterisation of sub-units - - - - - - 26

5.2 - GEOTECHNICAL CLASSIFICATIONS - - - - - - - - - - - - - 275.2.1 - General - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 275.2.2 - Bieniawski’s Rock Mass Rating- - - - - - - - - - - - - - - 275.2.3 - Barton’s Q index - - - - - - - - - - - - - - - - - - - - - - - - - 285.2.4 - Summary and precautions- - - - - - - - - - - - - - - - - - 28

5.3 - CORRELATIONS - - - - - - - - - - - - - - - - - - - - - - - - - - - - 295.3.1 - General - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 295.3.2 - Estimating rock mass deformability - - - - - - - - - - - 295.3.3 - Hoek’s GSI index- - - - - - - - - - - - - - - - - - - - - - - - - 295.3.4 - Estimating rock mass limit strength - - - - - - - - - - - 30

5.4 - PRESENTATION OF ROCK MASS CHARACTERISATION DATA - - - - - - - - - - - - - - - - - - - - - - - 31

5.4.1 - Basics and general remarks - - - - - - - - - - - - - - - - - 315.4.2 - Example of data presentation in tabular form- - - - 315.4.3 - Synoptic presentation of rock mass characterisation

data and cross-referencing to geological profile- - 31

CONTENTSCONTENTS



1 - INTRODUCTION

1.1. – PURPOSE OF ROCKMASS CHARACTERISATION

The most important goal in the characte-risation of rock masses is to provide theengineer with qualitative and quantitativedata to describe their structure andassess their mechanical and hydraulicproperties at a scale commensurate withthe volume of rock affected by the struc-tures. The overlying materials (sand,scree, moraine, etc.) are ignored in theseRecommendations.

It is essential to have precise knowledge ofthis data for project design, selection ofconstruction methods, support and liningthickness. The cost of the works is directlydependent on these points.

Whereas the mechanical properties of therock matrix can be determined from labo-ratory tests on small specimens, those of arock mass measuring several thousandcubic metres in size which may containwithin itself many discontinuities and hete-rogeneities cannot be determined directly.

In situ field tests, whose number is inevita-bly limited by their high cost, comesomewhere between laboratory tests andthe full size structure in terms of scale. Theyare instructive but still imperfect for fullyascertaining the mechanical properties ofthe rock mass at the relevant scale.

Not being amenable to direct measure-ment, the mechanical and hydraulic pro-perties of the rock mass must necessarilybe approached by indirect methods:

• either by trying to construct a model ofthe rock mass relevant to the size of thestructure under consideration, using testdata obtained at a smaller scale and thecharacteristics of the discontinuities,

• or by resorting to current rock classifica-tion systems and the mechanical characte-ristics which can be derived empiricallyfrom them, based on the back-analysis offull scale structures, as operated by variousauthors.

In both cases, it is vital to arrive at themost methodical and comprehensive cha-racterisation of the rock mass as possible.

1.2 – GEOLOGICAL MODEL

1.2.1 – Geological survey Before embarking on the stage of a rockmass characterisation properly so-called,as defined in these Recommendations, thedesign process normally begins with ageological survey to situate the projectarea within the general geological setting.

The survey rests essentially on field work byengineering geologists using the fullarmoury of tools and methods available tothe them:

• Desk study to collect published material,maps and data

• General mapping of the project area,detailed mapping of outcrops and indica-tors, collection of hydrogeological data

• Photography at various scales (satelliteimagery and aerial photographs)

• Geophysical methods: high resolutionseismic reflection and refraction, resistivityand electromagnetic tests, thermogra-phics, ground penetrating radar, etc.

• Exploratory boreholes, shafts and adits

• Information available from any nearbystructures, etc.

The geological survey locates the majorgeological units and their relations, the maindiscontinuities, the tectonic history, etc.

1.2.2 – Geological model

The first step in the characterisation of rockmasses requires constructing a tentativegeological model showing the geologicalstructure of the rock mass complete with itsconstituent units, boundaries, major fea-tures, heterogeneities and uncertainties.

It will ideally be a three-dimensionalconceptual model yielding cross sectionsto be used for understanding the structures

and identifying singularities and indetermi-nate points.

The geological model is the indispensablebasis for proceeding with the characterisa-tion of the rock behaviour parameters.With this model, the rock can be subdivi-ded into homogeneous sub-units whosemechanical and hydraulic properties cansubsequently be determined at projectscale.

What governs the size of a sub-unit withinthe rock mass is its uniformity of its geo-technical properties, producing uniformityof response to the structure to be built. Asub-unit may thus occupy part of a geolo-gical stage, the whole stage or severalstage. It may be homogeneous in terms ofits lithology, jointing, rock stresses, etc.

Of course, even small features that are onlylocal singularities in the wider rock massmust be treated as individual sub-units andas such, be the subject of geotechnicalcharacterisation.

Subsequently, as the design studies pro-ceed, the latest results from the geologicalsurvey will be incorporated from time totime into the model.

1.3 – GEOTECHNICAL CHARACTERISATION OFSUB-UNITS

Characterisation of a homogeneous sub-unit in a rock2 mass always involves deter-mining the parameters of the rock matrixand of the discontinuities; with discontinui-ties, the geological survey must make itpossible to choose the most relevant scaleat which they must be analysed and cha-racterised with reference to the scale of theproject under design.

Some homogeneous sub-units as definedin para. 1.2.2. may consist of more or lessregularly alternating rock layers, each withhighly contrasting geotechnical properties(for example, marl limestone, flysch, etc.)which must be analysed separately beforeproceeding with the characterisation of thewhole sub-unit.

1 It must be noted that some homogeneous sub-units thus defined are liable to include random heterogeneities (karst cavities for example) unidentifiableby the exploratory works. It is the engineer's job to decide what steps are to be taken in respect of this risk, after assessing its probability.

2 Rock must be understood in a very general sense. It may be a mass of soft rock or rock crushed by tectonic action to the point where it becomes like asoil, as well as a highly competent rock mass.



Characteristic parameters for the rockmatrix and discontinuities appear at thetop of Table 1. Those for the rock mass,some of which derive from the former, arelisted in the bottom half of the Table.

2 – MATRIX CHARACTERISTICS

PRELIMINARY REMARK. Most laboratorytests used to characterise rock matrix para-meters are inexpensive compared to field

exploratory works (drilling) and even moreso, the cost of building the structure. It isalways advisable to perform enough tes-ting in order to obtain data that can bemanipulated by statistical methods and

2.1 IDENTIFICATION PARAMETERS2.1.1 Common names2.1.2 Petrography & mineralogy2.1.3 Alteration of minerals2.1.4 Densities2.1.5 Volume weights2.1.6 Moisture content2.1.7 Porosity2.1.8 Degree of saturation2.1.9 Permeability2.1.10 Ultrasound wave velocities, continuity index

2.2 MECHANICAL PARAMETERS2.2.1 Deformability: instantaneous behaviour

- Young's modulus - Poisson's ratio2.2.2 Deformability: time-dependent (creep) behaviour2.2.3 Time-dependent (swelling) behaviour2.2.4 Strength

- Uniaxial compressive strength σ- Tensile strength σ- Brittleness index FR- Point compressive strength (Franklin test)

2.2.5 Triaxial test and failure criteria: Mohr-Coulomb criterion,Hoek & Brown criterion

2.2.6 Parameters for resistance to excavation: hardness, drillability, abrasiveness, DRI

2.2.7 Other tests: fragmentability, degradability, LA & MDE tests

3.1 IDENTIFICATION PARAMETERS3.1.1 Types and origins of discontinuities3.1.2 Description of discontinuities: strike, spacing, persistence,

roughness, weathering, width, infill, water3.2 CHARACTERISATION OF JOINT SYSTEMS3.2.1 Directional joint set patterns3.2.2 Statistical analysis of geometrical parameters for each set:

orientation, spacing, persistence3.2.3 Lumped jointing density indexes: RQD, ID, FD

4.1.1 RQD 4.1.2 Alteration and weathering degree4.1.3 Rock mass continuity index ICM4.2.1 Rock mass deformability – rock mass deformation modulus

EMas

4.2.2 Limit strength of rock mass4.3.1 Identification of aquifers4.3.2 Measurement of initial piezometry4.3.3 Measurement of rock mass permeability KM4.3.4 Gas4.3.5 Other parameters4.4.1 Initial state of stress and approximations4.4.2 Characterisation of stress tensor4.4.3 Commentary on field test methods4.4.4 Classification of stress states

4.5.1 Geothermal parameters4.5.2 Temperature assessment methods

3.3 MECHANICAL PARAMETERS3.3.1 Deformation parameters: normal stiffness, tangential stiff-

ness3.3.2 Shear strength parameters: peak strength, residual strength,

dilatancy3.3.3 Hydraulic parameters

2 CHARACTERISTICS OF ROCK MATRIX 3 CHARACTERISTICS OF DISCONTINUITIES

4 CHARACTERISTICS OF ROCK MASS

4.1 IDENTIFICATION PARAMETERS

4.2 MECHANICAL PARAMETERS

4.3 HYDROGEOLOGICAL CONDITIONS

4.4 INITIAL STATE OF STRESS IN ROCK MASS

4.5 TEMPERATURE

Table 1 – Characteristic parameters of rock matrix, discontinuities and rock mass(Numbers refer to paragraphs in these Guidelines)



reveal the homogeneity or scatter in themeasurs. Meaningful laboratory test resultsare essential to draw full benefit from theinformation obtained by drilling.

2.1 – IDENTIFICATIONPARAMETERS

2.1.1 – Common namesRock names are based on chemical andmineralogical composition, texture and theway their were formed. There are threemain classes of rock: igneous, metamor-phic and sedimentary rock (Appendix 1).

Igneous rock is solidified magma.Solidification at depth produces plutonicrock which solidified slowly and permittedcrystals to grow large enough to be seenwith the naked eye; the most commonexample is granite. Extrusive rock is for-med from magmas emerging directly atthe Earth's surface; few crystals can beseen by eye because of the rapid coolingof the material. The most widespreadextrusive rock is basalt.

Sedimentary rock forms at the surface, onland or under water, by deposition of origi-nally near-horizontal beds. Sedimentaryrock subdivides into:

• Detrital rocks, resulting from the deposi-tion of debris from pre-existing rocks resul-ting from erosion and transport processes(running water, glaciers, wind); the mostwidespread representatives are sandstoneand the argillaceous rocks.

• Physical/chemical and/or biogenic rocksformed by precipitation of ions in solutionand/or living matter; the commonest arecarbonate rocks and saline rocks, still calledevaporites.

Metamorphic rocks are the result of pro-found transformation in the solid state ofpre-existing sedimentary or igneous rocksby elevated temperatures and/or pres-sures. They often exhibit schistosity orfoliation accompanied by lineation. Thecommonest are schist and gneiss in whichthe minerals are strongly oriented. Marbleand quartzite are massive, completelyrecrystallised rocks in which the orientationof the minerals (calcite or quartz) is hardlyvisible to the naked eye.

It is important to bear in mind possiblevariations in the facies of rocks belongingto the same geological stage and the factthat some common names deriving fromthe geological map do not always tally with

the lithology of the rock concerned.Reference should be made to the full des-cription in the description accompanyingthe map.

It is preferable to use the terms inAppendix 1 and avoid employing unusualcomplex names.

2.1.2 – Petrography and mineralogyA petrographic description covers the fol-lowing observations with the naked eye ormagnifying glass or (preferably) by inspec-tion of thin sections under the microscope:

• Identification of minerals present

• Size and arrangement of the minerals(texture)

• Proportions of the different constituents

• Voids and discontinuities (pores and fis-sures).

Mineralogical analysis of the constituentsestablishes the mineral composition of therock and yields information on its proper-ties such as weathering potential, swellingpotential, ability to stick, abrasion poten-tial, etc.

The mineralogical analysis is usually perfor-med by X ray diffraction on a small powde-red specimen. It allows the identification ofthe minerals present and, after interpreta-tion, yields the semi-quantified composi-tion. Special preparation is needed if it issuspected that swelling clay minerals mightbe present.

Additional quantitative determinations ofCaCO3, silica, sulphates, organic matter,etc. refine the identification.

The clay fraction, if present, must be cha-racterised from the Atterberg limits. Amethylene blue absorption test will esti-mate the activity of the clay fraction (Frenchstandard NF P 94-068).

2.1.3 – Alteration of the mineralsin the rock matrixAlteration of the matrix is the result of phy-sical/chemical changes in the constituentrock minerals. It is usually associated withmajor changes in the physical and mecha-nical properties of the rock. Some mineralsare subject to dissolution (e.g. calcite, gyp-sum), other to decompositione (e.g. bio-tite, plagioclase). As a general rule, therock loses cohesion. The process is usuallysubdivided into

• Hydrothermal alteration, usually confined

to the walls of major discontinuities inwhich fluids from deep-lying sources circu-late. It frequently causes major mineralogi-cal changes (appearance of special mine-rals such as chlorite, etc.), usuallyaccompanied by significant changes inmechanical properties.

• Weathering working down from the sur-face to sometimes considerable depth.Processes include solution of gypsum, cal-cite, etc., mechanical disruption (increasedmicrocracking) and mineralogical changesproducing clay minerals.

The intensity of alteration and weatheringof the matrix can be quantified by minera-logical analysis and indirectly by tests suchas the methylene blue test and measure-ment of ultrasound wave velocity.

A clear distinction must be made betweenthe degree of alteration of rock as takenfrom a borehole or at the tunnel face andits potential of alterability when exposed tothe atmosphere.

2.1.4 – Densities (French standardP 94-410-1/2/3)Different densities (M.L.-3 dimensions) areapplicable according to the condition ofthe material.• Natural density ρ = m/v

Ratio between dry mass md of the oven-dried sample and the volume V of thesample including any air it contains.• Dry density ρd = md /v

Ratio between dry mass md of the oven-dried sample and the volume V of thesample including any air it contains.

• Density of solid particles ρs = ms/vs

Ratio between the dry mass of solid par-ticles ms in a powdered specimen and thevolume vs occupied by the particles (mea-sured in a pycnometer). This characteristic

Photography 1– Hydrothermal alteration – Granite(Ghangzou China)



of the solid phase of the rock material isdirectly dependent on the mineral compo-sition of the rock. Appendix 1 lists valuesfor the more common minerals.

2.1.5 – Volume weightsCorresponding unit weights (M.L-2.T2

dimensions) γ, γd and γs to he above unitmasses are obtained by multiplying theunit masses by the acceleration due to gra-vity g = 9.81 m/s2:

γ = ρ x g

2.1.6 – Moisture content (Frenchstandard P 94-410-1)The moisture content by weight w is theratio, expressed as percentage of themass of water mw to the mass of the drymaterial md:

w(%) = (mw/md) x 100

2.1.7. Porosity (French standard P 94-410-3)The porosity n is the ratio, expressed aspercentage of the volume of voids vv to thetotal volume of the rock sample v:

n (%) = (vv/v) x 100

The porosity is governed mainly by thepresence of globular voids (pores) andmuch less by fissures (flat, thin voids). Somevoids may be inaccessible to saturationwater (occluded voids).

Values for porosity classes are listed inTable 2.

In the majority of sedimentary rocks, a dryunit mass ρd of the order of 2.7 t/m3 is agood indicator of the porosity.

2.1.8 – Degree of saturationThe degree of saturation with water Sr isthe ratio, expressed as percentage of thevolume of water in the sample vw to thevolume of voids vv; it is the percentage ofthe pore space filled with water.

Sr = (vw/vv) x 100

Rock is described as 'dry' when Sr = 0. It is'saturated' when Sr = 100%.

2.1.9 – PermeabilityThe permeability k of a rock sample is des-cribed by a coefficient relating the flow Qpassing across a surface S under a hydrau-lic head gradient i (Darcy's law).

Q/S = k x i

The dimension of k is a velocity (L.T1).

The permeability of the rock matrix is stron-gly influenced by microcracking (intercon-nected voids) and therefore varies with thestate of stress. The proper choice of repre-sentative samples and their pre-test condi-tion is particularly important. Laboratorytests are done with special longitudinal,radial, etc. permeameters or triaxial appa-ratus. If permeability is found to be aniso-tropic, tests should be made in severaldirections.

Knowledge of the matrix permeability isessential only for some underground pro-jects (mined storage, waste disposal bycontainment, etc.).

2.1.9 – Ultrasound wave velocity(French standard p 94-411) –Continuity indexUltrasound wave velocity yields informa-tion on alteration and weathering and/orfissuration and porosity.

Measuring wave velocity in several diffe-rent directions may reveal anisotropy dueto preferential orientation of microcracks orrock structure.

There are two types of wave:

• Compression ( or longitudinal or P) wavesnoted Vp

• Shear (or transverse or S) waves, not rou-tinely measured, noted Vs.

The continuity index IC of rock is definedas the ratio between Vp measured onsamples and the theoretical Vp* derivedfrom the mineral composition of thesample:

IC (%) = 100 x (Vp/Vp*)

Vp* is the harmonic mean of the Vpi wavevelocities in the constituent rock minerals(Appendix 2) multiplied by their proportionby volume ci:

1/Vp* = ∑ ci/Vpi

An approximation is usually possible. Itconsists of estimating the theoretical wave

velocity Vp* from the table in Appendix 3showing maximum theoretical wave veloci-ties for some rocks, assumed to be soundwithout pores and fissures.

The continuity index for the same petro-graphic composition falls as pore porosityincreases; the trend is even more markedas microcrack porosity increases.

Continuity classes based on IC value are lis-ted in Table 3.

2.2 – MECHANICAL PARAMETERS

The mechanical parameters relevant togeotechnical classification and to thechoice and optimisation of tunnelling tech-niques and plant are determined in thelaboratory from representative samples.

Attention must be given to any anisotropyin measured properties. Parameters descri-bing mechanical behaviour very often dif-fer according to the orientation of thesample with respect to bedding (in sedi-mentary rocks) or foliation (in metamorphicrocks, see example in figure 1). With somerocks, the anisotropy ratio, defined as theratio between maximum and minimumvalues of a parameter, measured at diffe-rent orientations, may be in excess of 5.

P 1 0 % < n < 1 % Very low porosity

P 2 1 % < n < 5 % low porosity

P 3 5 % < n < 15 % Moderate porosityP 4 15 % < n < 30 % High moderate

porosity

P 5 n > 30 % Very high porosity

CLASS POROSITY N VALUES DESCRIPTION

Table 2 – Porosity classes for rock matrix

IC 1 IC > 90 % Very high continuity

IC 2 75 %< IC < 90 % High continuity

IC 3 50 %< IC < 75 % Moderate continuity

IC 4 25 %< IC < 50 % Low continuity

IC 5 IC < 25 % Very low continuity

CLASS CONTINUITYINDEX IC

DESCRIPTION

Table 3 – Rock matrix continuity classes

Figure 1 – Example of anisotropic uniaxial compressivestrength (Lias shales, Grand'Maison, France, 1980)



Mechanical tests should be made in seve-ral directions; this must always be donewith metamorphic rocks.

2.2.1 – Deformability: instantaneous behaviour

2.2.1.1 – Young's modulus (French standard P 94-425)

In a uniaxial compression test, the Young'smodulus of elasticity E is defined as theslope of a loading/unloading cycle on theaxial stress/strain curve σ1 = f(ε1) at half thefailure stress of the sample σc.

Classes based on stiffness values (thereciprocal of deformability) are listed inTable 4.

2.2.1.2 – Poisson's ratio (French standard P 94-425)

In the uniaxial compression test, thePoisson's ratio n is defined as the ratio bet-ween the slopes of the straight-line limbsof the σ1 = f(ε3) and σ1 = f(ε1) curves.

ν = dε3/ dε1

ε1 = axial strain

ε3 = transverse strain.

Values of the Poisson's ratio for differentrocks generally fall between 0.15 and 0.40.

2.2.2 – Deformability: time-dependent behaviour relatedto creep

2.2.2.1 – Definition

Time-dependent effects take the form ofchanges in strain and/or stress over time.They have three causes:

1. They may be related to an intrinsic rheo-logical property of the rock whose defor-mation under constant load increase overtime. It is mainly encountered in evaporites(rock salt, potash, gypsum, etc.), argilla-

ceous rocks (marl, claystone) and some car-bonate rocks such as chalk.

2. They may be the consequence ofdamage. In any rock, time-dependenteffects more or less appear when themicrocracking threshold is exceeded.

3. They may be due to changes over timein the pressure of fluids in pores and fis-sures, due to changes in boundary condi-tions of flow patterns caused by construc-tion of the works (effect of consolidationand fluid flow pressures).

Only time-dependent effects associatedwith creep mechanisms 1 and 2 are consi-dered here. They produce viscoelastic(reversible) or viscoplastic (irreversible) res-ponses in the rock matrix. They havenothing in common with responses due tofluid flow (mechanism 3).

Time-dependent behaviour is more mar-ked as higher temperatures reduce visco-sity. This is particularly noticeable in evapo-rites. High stresses have the same effect,especially deviator stresses.

The long-term performance of under-ground structures may be affected bychanging stresses and strains, and progres-sive loss of strength. The magnitude ofthese processes is frequently different insitu from in the laboratory, although labora-tory tests do yield a significant approxima-tion of the importance, pattern and orderof magnitude of time-dependent beha-viour to be expected in situ.

2.2.2.2 – Laboratory characterisation oftime-dependent behaviour: creep test

In studying time-dependent rock beha-viour in the laboratory, the commonest testis the creep test which is a stress-controlledtest to measure strain under constantstress. The relaxation test, a strain-control-led test which measures stress changesunder constant strain, is not often used.

The creep test is usually preferred. It offers areasonable simulation of conditions obtai-ning around underground openings; afterthe transient tunnelling period, the state ofstress around the opening thereafterremains substantially unchanged over time.

If laboratory tests characterising time-dependent or long-term rock behaviourare to have maximum significance, testconditions (temperature, stress range,degree of saturation, drained or undrainedboundary condition, etc.) must reflectactual in situ conditions of the structureand the host materials.

The creep test may be performed underuniaxial or triaxial pressure and drained orundrained conditions, preferably at severaldifferent deviator stresses, to determinethe following parameters (figure 2):

• Creep threshold, the state of stress belowwhich creep is negligible (the thresholdmay be near-zero for some evaporites).

• Creep rate, which generally increaseswith deviator stress.

• Creep acceleration threshold beyondwhich "tertiary creep" occurs and culmi-nates in failure; it characterises the longterm strength of the rock. In rocks exhibi-ting no (or little) creep under normal civilengineering conditions (granites, etc.),long term strength is of the order of 90% ofthe strength measured in quick tests (suchas the standard uniaxial compressivestrength test); in chalks, it is of the order of50%; it falls to 30% or less for evaporites.

Creep rates cover a very extended range,depending on the stress ranges concerned:

• The fastest rates (>10-5s-1) cover the rangeof instantaneous strain as measured in thelaboratory.

• The slowest rates (<10-11s-1) are scarcelymeasurable but represent deformationover geological time.

Creep rates relevant to actual tunnels,chambers, storage cavities, etc. will liesomewhere between the two extremes.Feedback from real projects is still scarce

DE 1 E > 50 GPa Extremely stiff matrix

DE 2 20 GPa < E < 50 GPa Very stiff matrix

DE 3 5 GPa < E < 20 GPa Stiff matrix

DE 4 1 GPa < E < 5 GPa Fairly stiff matrix

DE 5 0,1 GPa < E < 1 GPa Low stiff matrix

DE 6 E < 0,1 GPa Very low stiff matrix

CLASSES YOUNG'S MODULUS

DESCRIPTION

Table 4 – Rock matrix stiffness classes (reciprocal of deformability)

Figure 2 – IIncremental creep test

I No creep: applied stress lower than creep thresholdII Transient (primary) creep decreasing over time:

applied stress slightly greater than creep thresholdIII Creep slowing down over time: applied stress lower

than long-term strengthIV Three-phase primary, secondary (stationary) and

tertiary (culminating in failure) creep: applied stressgreater that long-term strength



and possible rates cover such a wide range(being governed by the characteristics ofthe rock mass and the underground struc-ture) as to make any forecast of time-dependent creep around a real tunnelhighly problematical (see para. 4.2.1.5below).

2.2.3 – Time-dependent behaviourrelated to swelling

2.2.3.1 – Swelling potential

Rock swelling is due to the rock increasingin volume over time, simultaneously withincreasing moisture content (wetting undersaturated or unsaturated conditions).Changes in the state of stress is a promo-ting factor. When rock expansion isconstrained, high stresses may result.

Rock swelling has two causes:

1. Water may be taken up by hydrophilicminerals, mainly swelling clay minerals suchas smectite, some hydroxides and sul-phates.

2. Change from anhydrite (CaSO4) to gyp-sum (CaSO4,2H2O).

Smectite-type minerals are relatively com-monplace and might be present wheneverthe rock contains clay minerals: clay, marl,molasse, fault gouge, karst infill, weatheredigneous and metamorphic rock.

Anhydrite and gypsum occur either aslarge masses in sedimentary beds or intru-ded into large tectonic discontinuities, oragain, finely scattered within marl or othermaterial.

Three conditions must be fulfilled for swel-ling to occur:

1. Potentially swelling minerals in the rock.

2. Available water.

3. A state of stress that permits volumeincrease.

2.2.3.2 – Identification of swellingpotential

IIt is strongly recommended to identifythe risk of swelling at a very early stage, asfollows:

• Qualitative observation of behaviour byimmersing specimens in water and seeinghow fast they break up.

• Methylene blue tests to determine thespecific surface and clay character of thematerial.

• Total or semi-quantitative (R.X.) mineralo-

gical analyses for the proportion of swel-ling clay mineral and/or anhydrite and gyp-sum present.

• If anhydrite is present, microcrackingshould be investigated, since water take-up is directly linked to the exchange sur-face area available.

Additional information can be obtainedwith the electronic microscope to gain abetter insight into the distribution of theclay flakes and any other special featuresliable to favour or oppose swelling (such assmectite being encapsulated in a calciteenvelope for example).

2.2.3.3 – Quantification of swellingpotential

To confirm the risk of swelling, mineralogi-cal analysis must be supplemented withlaboratory tests.

The International Society for RockMechanics ISRM recommends three labo-ratory tests to characterise rock swellingpotential (Int. J. of Rock Mech. and Min.Sci., 1999, vol. 36, 291-306):

1. Measurement of axial swelling pressureat constant volume (determination of swel-ling pressure σg).

2. Determination of unrestrained axial andradial swelling strain.

3. Measurement of axial pressure versusaxial strain (to determine swelling index Cg).

The third test derives directly from theHuder-Amberg test developed in the earlyseventies to characterise swelling in soils(cf. Appendix 4).

The swelling test for soils described inFrench standard P 94-091 uses the oedo-meter to measure swelling on four speci-mens assumed to be identical. The speci-mens are wetted and subjected to fourdifferent levels of axial stress. The testconsists of measuring the thicknessincrease ∆h/h of each of the four speci-mens and correlating thickness changewith the corresponding axial stress levelsby fitting a straight line on a semi-log plot(∆h/h – log σ).

The Huder-Amberg test uses the sametype of interpretation but has the advan-tage of requiring only one test specimen,and this limits the ever-present problem ofscatter inherent in irregularities in materialtype and condition, when using multiplespecimens.

In practical terms, the first step should beto measure the axial swelling pressure atconstant volume. Next, axial pressure ismeasured versus axial strain (Huder-Amberg type), the specimen being wettedfor the first time with axial pressure sub-stantially equivalent to the previouslydetermined swelling pressure.

Axial swelling pressures may vary greatly indifferent materials, from near-negligible(less than 0.1 MPa) to several MPa in somemarls. There is frequently severe scatter inthe swelling pressures measured on anygiven material. This is why it is advisable toperform several tests, regardless of thetype of test used to quantify swellingpotential.

2.2.3.4 – Comments

Materials containing clay minerals (clay,marl, molasses) frequently exhibit conside-rable anisotropy, due to the way they wereformed. This anisotropy may lead to verydifferent swelling potentials when tested indifferent directions, especially the direc-tions parallel and perpendicular to theplane in which the clay particles weredeposited. Tests must be made in thesetwo directions for a proper characterisationof swelling.

The composition of the water used in thetest may exert a strong influence on thedevelopment of swelling. Certain chemi-cals may promote or inhibit swelling. Thetest water should therefore be clearly iden-tified. Swelling may develop very diffe-rently in situ, depending on whether thewater comes from the surrounding rock orfrom some external source (from inside thetunnel).

2.2.4 – Mechanical strength

2.2.4.1 – – Unconfined compressivestrength σc or uniaxial compression test(French standard P 94-420

The failure stress in uniaxial compression isdefined as

σc = Fmax/A

Fmax = maximum axial force reached in thetest

A = area of the pre-test circular cross sec-tion of the test specimen.

Rock strength classes used in the presentRecommendations3 and ISRM appear inTable 5.

3 They differ from the classes used in the AFTES 1978 Recommendations



Soft rock and stiff soils often classify asClass RC6 and RC7.

Rock tensile strength σtb is determined bythe indirect Brazilian test (figure 3) usingthe procedure described in French stan-dard P 94-422.

The test breaks a cylinder of diameter Dand height H by applying a compressiveload F at diametrically opposed sides ofthe cylinder. The stress at failure of theBrazilian test specimen σtb is given by theequation

σtb = 2 Fmax/πx D x H

Fmax = load at failure

H = cylinder height

D = cylinder diameter.

2.2.4.2 – Brittleness index FR

The brittleness index FR is defined as theratio σc/σtb.

It is useful as characterising the drillabilityand failure type for hard rocks (σc > 25 MPa).

FR usually ranges from 5 to 30. Brittlenessclasses are listed in Table 6.

2.2.4.3 – Point load compression test,so called– Franklin test (French standard P 94-429

The point load compression tests involvesbreaking rock fragments of indeterminateshape or core pieces between two coneswith spherical tips (dimensions standardi-sed). Specimen thickness between points

may range from 25cm to 100cm. Test spe-cimens are usually pieces of rock takenfrom 50mm dia. cores, although charts areavailable to correct results from other coresizes (Broch & Franklin, Int. J. of RockMech. and Min. Sc., 1972, vol 9, 669-697).

Test results are expressed as a strengthindex Is (in MPa):

Is = F/D2

F = load at failure

D = specimen diameter or distance bet-ween points.

The index obtained with a 50mm diameteris written Is50.

The point load compression test can beperformed with very lightweight apparatuson site. Correlations yield an estimate ofrock unconfined compressive strength:

20 Is50 < σc < 27 Is50

However, the point load compression testcan never be considered as a substitute forthe uniaxial compression test.

2.2.5 – Triaxial test and failure criteria

2.2.5.1 – Triaxial test (French standard P 94-423)

The triaxial test is an axial compression testin which a constant confining stress σ3 isapplied to the specimen. The axial com-pressive stress is continued to failure.

Determination of the failure criterion for arock involves performing several triaxial testswith increasing confining stresses. Theremust be at least four test runs, including auniaxial compression test in which σ3 = 0.

The full σ1 - σ3 = f(ε1) curve is plotted foreach test at a different confining stress tofind the peak value of the deviator failurestress σ1 - σ3, plus, possibly, the residual(constant) stress at which the specimenshears along the failure plane.

Measurement of axial and transverse strainin the course of the test leads to values forthe Young's modulus and Poisson's ratiounder confining stress conditions.

This is the standard soils mechanics crite-rion written as

τ = c + σn x tan ϕ

• σn is the normal stress and and τ is theshear stress on the failure surface

• c is cohesion

• ϕ is the internal friction angle.

σc > 200 MPa RC1 Extremely strong matrix

100 MPa < σc < 200 MPa RC2 Very strong matrix

50 MPa < σc < 100 MPa RC3 Strong matrix

25 MPa < σc < 50 MPa RC4 Moderately strong

5 MPa < σc < 25 MPa RC5 Low strong matrix

1 MPa < σc < 5 MPa RC6 Very low strong matrix

σc< 1 MPA RC7 Extremely low strongmatrix

CLASSUNIAXIAL

COMPRESSIVESTRENGTH σc

DESCRIPTION

Table 5 – Uniaxial compressive strength classes

Photo 3 – Specimen after failure in triaxial test(French standard p 94-422)

Figure 3 – Stress state at failure at specimen centre in compression test

FR 1 FR >25 Very brittle matrix

FR 2 15 < FR < 25 Brittle matrix

FR 3 10 < FR < 15 Moderately brittle matrix

FR 4 FR < 10 Low brittle matrix

CLASSBRITTLENESS

INDEX FR VALUES DESCRIPTION

Table 6 – Rock matrix brittleness classes

Photo 2 - Uniaxial compression test



In terms of the principal stresses, with σ1 > σ3,the Mohr-Coulomb criterion is written

σ1 = [σ3(1 + sin φ) + 2c(cos φ)] / (1 – sin φ)

This limit strength criterion can apply to theelastic limits, peaks or plateaus on the stressstrain curves from triaxial tests (total or effec-tive stresses, drained or undrained tests).

The Mohr-Coulomb criterion may suit themechanical behaviour of certain rocks inthe moderate confining stress range. Moregenerally, it may be acceptable for repre-senting the behaviour of a given rockwithin a certain limited range of confiningstresses (linearisation of a parabolic crite-rion).

2.2.5.2 – Hoek & Brown criterion

This parabolic criterion is well suited to themechanical behaviour of rock and is usuallyused for these materials.

The criterion for intact rock is written

σ1 = σ3 + σci x [(mi σci) + 1]1/2

in which

• σci is the uniaxial compressive strength ofthe intact rock material

• mi is a constant dependent on rock type.

From the standard expression of a parabo-lic criterion expressed in terms of uniaxialcompressive σci and σti tensile failurestresses (Appendix 5), it can be seen that

parameter mi is very close to the brittlenessindex FR:

mi ≈ σci/σti = FR

2.2.6 – Parameters for resistanceto excavationThe parameters examined in this sectionconcern the response of rock to variousmeans of breaking it up (its decohesion).They are useful for assessing rock in res-pect of excavating and crushing tech-niques, in order to refine forecasts as to

• mechanised excavation, through hard-ness and drillability tests,

• excavating tool wear and consumption,through abrasion tests,

• crushing performance, through fragmen-tation tests.

Test results are interpreted with the aid ofmulti-criteria correlation models which alsorefer to other parameters (mineralogy,mechanical properties, discontinuities),investigated in another location.

2.2.6.1 – Hardness and “ drillability”

These tests can be classified under threeheadings based on the techniques used.

1. Tests assessing the penetration of drillbits and cutter tools, such as the CERCHAR,Siever and other tests.

2. Static indentation testswhich assess the mark left by apoint applied to the rock, suchas the punch test, Vickers,Knoops and Schreiner tests,micro-indentation test, etc.

3. Rebound tests, in which therebound of a known mass hit-ting the rock is measured(Schmidt hammer).

The commonest test is thepenetration test derived fromthe test first described byFrench research institute CER-CHAR: "Drill bit penetrationresistance index" – Frenchstandard P 94-412.

It characterises rock hardnessvia its resistance to penetrationby a drill bit under standardtest conditions. The hardnesstest is suited to fine grainedrocks of moderate to lowstrength.

Rock hardness classes based on the CER-CHAR-INERIS rules are listed in Table 7.

2.2.6.2 – Abrasiveness

The abrasion potential of a rock is determi-ned by its mineralogical composition,especially the percentage of quartz itcontains, and its intergranular cohesionand grain size.

Abrasiveness can be characterised by twoconventional, standardised indexes; it mustbe recognised that they bear no relation-ship to each other.

2.2.6.2.1 – Point scratch test – French stan-dard P 94-430-1

The first index, AIN4 comes from the CER-CHAR-INERIS point scratch test. Testresults are expressed as an abrasivenessindex characterising the ability of a rock tocause wear of the cutting tool.

Rock abrasiveness classes derived from thisindex are listed in Table 8.

2.2.6.2.2 – Rotating drilling bit test (Frenchstandard P 94-430-2)

The other index is derived from the LCPCabrasiveness test. This test is suitable forrocks whose tensile strength is greater than

Photo 4 – Triaxial test apparatus

DU 1 > 120 Extremely hard matrix

DU 2 80 - 120 Very hard matrix

DU 3 40 - 80 Hard matrix

DU 4 20 - 40 Moderately hard matrix

DU 5 5 - 20 Soft matrix

DU 6 < 5 Very soft matrix

CLASS HARDNESSVALUES

DESCRIPTION

Table 7 – Rock matrix hardness classes based onCERCHAR-INERIS test results

AIN 1 >4.0 * Extremely abrasive matrixAIN 2 2.0 – 4.0 Very abrasive matrixAIN 3 1.0 – 2.0 Abrasive matrixAIN 4 0.5 – 1.0 Low abrasiveness matrixAIN 5 < 0.5 Very low abrasive matrix

CLASSABRASIVENESS

INDEX AINVALUES

DESCRIPTION

*Quartz and gemstones register higher than 6

Table 8 – Rock matrix abrasiveness classes based onresults of CERCHAR-INERIS test

4 CAI (CERCHAR Abrasiveness Index) in English-speaking countries



1 MPa. The result is expressed by the abra-siveness index ABR calculated from wear ofa drill bit rotated in a body of 4/6.3 aggre-gate obtained from the rock under test.

Rock abrasiveness classes based on theLCPC test are listed in Table 9.

2.2.6.3 – Norwegian tests: DRI index

The Norwegian Drilling Rate Index definedby the University of Trondheim (Movinkel &Johannessen 1986) combines the results ofan S20 fragmentation test and an SJ drillbit penetration test. The DRI chart and adescription of the S20 and SJ tests appearin Appendix 6.

2.2.7 – Other tests2.2.7.1 – – Fragmentability (French stan-dard P 94-066) – Degradability (P 94-067)

These standard tests characterise the sus-ceptibility of the rock to fragment anddegrade under the action of drilling andexcavating machines. They are mainly usedwhen rock cuttings are to be used asabove-ground building material, but theyare undoubtedly useful in tunnelling workto characterise spoil muck.

2.2.7.2 – Los Angeles test (French stan-dard P 18-573) – Micro-Deval test inpresence of water (P 18-572

It will be remembered that the Los Angelestest characterises the resistance of rock tofragmentation while the micro-Deval(MDE) test characterises the rock's resis-tance to wear in the presence of water.These are essential parameters when dea-ling with hard rocks, for characterising thepossibility, when applicable, of using muckas something more elaborate than plain fill(ballast, road-making and concrete aggre-gate, road foundation courses, etc.).

3 – CHARACTERISTICS OFDISCONTINUITIESThe system of discontinuities in the rockmass must be investigated in detail at pro-totype scale of the project, mainly throughstatistical analysis in order to consider thenatural variability in their geometrical andmechanical parameters. Different dataacquisition methods are appropriate to dif-ferent scales of observation:

• aerial photography and surface geophy-sics at regional scale

• outcrop and borehole surveys at local scale.

When core drilling, it is always recommen-ded to accompany the geologist's logs(geological drill logs which include recordsfrom the various borehole loggings andjoint mapping) with colour photographs ofall the core boxes; photographs shouldinclude a scale, colour chart and legibleidentification markings. Photographicrecords may indeed provide the engineerwith abundant valuable information, espe-cially about discontinuities. However, theycould not replace the direct inspection ofthe cores which is essential , even if it isoften difficult.

3.1 – JOINT IDENTIFICATIONPARAMETERS

3.1.1 – Types and origins of thediscontinuitiesThe term discontinuity is used in rockmechanics in a very general sense, to desi-gnate any interuption of the continuity inthe rock material with its attendant mecha-nical, hydraulic and thermal properties.The surface can be treated, over a certaindistance, as a plane displaying zero or lowtensile strength.

Discontinuities represent a wide variety oftypes of surfaces whose geological identifi-cation conveys important information onsome of their geometrical and mechanicalfeatures (parameters). Instead of theconventional terminology of joints, frac-tures or cracks, preference is given to thefollowing terms:

• Bedding planes are surfaces separatingsedimentary rock strata. They are very per-sistent and may contain clay material resul-ting in very low shear strength.

• Joints properly so called, are discontinui-ties between two rock compartmentswithout any apparent relative movementbetween them. The joint may be tight oropen with generally planar, relativelysmooth walls, and may extend over a dis-tance measured in decimetres to deca-metres. They often occur as joint sets run-ning in two or three directions.

• Faults are the result of relative movementbetween the two rock compartments bet-ween which they lie, caused by the stressfield during the relevant tectonic episode(normal, reverse and strike-slip faults).Persistence is very variable (metres to seve-ral kilometres long) and they frequentlycontain infilling material (gouge) with poormechanical properties.

• Schistosity planes cause the rock to breakinto thin parallel layers under the action oftectonic stresses. Persistence is variableand small unitary schistosity planes mayform surfaces persisting for very long dis-tances through localised failure of rockbridges between closely-spaced parallel orstepped discontinuities.

• Fracture zones are assemblages of dis-continuities of variable persistence andorientation, organised in a more or less pla-nar pattern.

• Lithological contact surfaces betweenhost rock and veins often form virtual dis-continuities, although they may be real sur-

ABR 1 ABR > 2000 Very highly abrasive

ABR 2 1500 < ABR < 2000 Highly abrasive

ABR 3 1000 < ABR < 1500 Moderately abrasive

ABR 4 500 < ABR < 1000 Low abrasiveness

ABR 5 0 < ABR < 500 Very low abrasiveness

CLASSABRASIVENESS

INDEX ABR

VALUES

DESCRIPTION

Table 9 – Rock matrix abrasiveness test based on results of LCPC test

Photo 5 – Rock mass with complex tectonic history.Tibi region, Spain



faces where differential weathering hasoccurred.

Note : If in French, discontinuity is the genericterm, in English, “joint” is the term currently usedfor discontinuity in this recommendation. Forexample, joint set means set of discontinuities.

3.1.2 –Description of discontinuitiesDiscontinuities within a homogeneous zoneare systematically mapped in order to charac-terise, statistically, a joint system usuallyconsisting of several joint sets with differentstrikes and possibly a few singular units whichmay be important from the rock mechanicsviewpoint (faults, altered veins, etc.).

These observations are made along ascanline marked on natural outcrops, tun-nel walls or the sides of excavations, orfrom boreholes, whether or not cores havebeen recovered (down-the-hole viewingtechniques).

In order to ensure these measures arerepresentative, surveys must be plotted inseveral spatial directions and must cover asufficiently large volume, commensuratewith the average joint density. Lastly, it isimportant that the method used to collectthe data be clearly described.

The full procedure for the systematic analy-sis of joint systems is based on characteri-sing each discontinuity with the followingeight parameters:

1. Strike: describes the position of the dis-continuity plane in space with respect toNorth. Two conventions are used (figure 4):

• The first, based on the direction of thehorizontal line of the plane α (0-180°), thedip angle β (0-90°) and the direction of thedip vector, is normally used in field work.

• The second, which uses the azimuth ofthe dip vector αp (0-360°) and dip angleβ (0-90°), is the method recommended byAFTES for plotting results.

Strike is the basic parameter for establi-shing the initial identification of joint sets. It

also determines the shapes of the indivi-dual blocks comprising the rock mass, andthereby, the anisotropy which will governhydraulic and mechanical behaviour.

2. Spacing: is the perpendicular distance(modal or average) between adjacent dis-continuities in a set. In fact, the distancebetween two successive intersections ofthe joint trace with the survey line can bemeasured in the field. This measurement isfrequently biased because it depends onthe joint persistence (for any given numberof traces on a surface, the longer ones havemore chance of being cut by the line of thesurvey and therefore appear to be moreclosely spaced) and the direction of thesurvey line.

3. Persistence: Joint persistence or sizeconcerns the total area of the discontinuityin all directions. It is an important parameterbecause, along with joint spacing, it governsthe connectivity of the discontinuity net-work and therefore the permeability of therock mass and the volume of the intact rockblocks. It cannot be measured directly, butcan be quantify by observing the disconti-nuity trace lengths (intersection line of thediscontinuity and the surface of exposure. Itcannot be measured in boreholes.

4. Joint wall roughness and waviness: mea-sured respectively in millimetres or centi-metres, and in decimetres to metres. Theseare crucial parameters because theygovern the process of dilatancy and there-fore the joint shear strength. Although diffi-cult to measure, every effort must be madeto estimate them (cf. para. 3.3.2 ShearStrength Parameters).

5. Joint wall weathering: This is an impor-tant parameter especially when the discon-tinuity walls are in direct rock to rock intact,because it governs the deformability andthe possibility of dilatancy affecting theshear strength. The degree of alterationcan be assessed directly in the field fromthe description of the weathering pro-ducts, their thickness and the Schmidthammer test (Appendix 9).

6. Joint width or aperture: Size of the gapbetween the joint walls measured perpen-dicularly to the joint plane.

7. Infill: The nature of the material fillingthe discontinuity or coating the walls mustbe characterised, along with its thicknessand its mechanical properties.

8. Presence of water: Presence of dampspots or water flow.

3.2 – CHARACTERISATIONOF JOINT SYSTEMS

3.2.1 – Directional joint set patternsDiscontinuities are not arbitrarily orien-ted in rock, they frequently occur assets, in numbers determined by thegeological and mechanical processes

Photo 6Sangatte grey chalk

Figure 4 – Attitude of a disontinuty in space



occurring at the time of rock formation andthe tectonic history of the rock body. Thedistribution of a population of discontinui-ties into directional sets is investigatedfrom the azimuth and dip parameters butother geometric parameters such as aper-ture and persistence may also be relevantin identifying joint sets.

The very concept of joint sets however is asimplification of reality which may some-times be excessive or unjustified. This typeof analysis must therefore be conductedwith prudence, making use of the geolo-gist's experience. In all cases, it is neces-sary to work according to joint type, distin-guishing for example between beddingplanes, foliation or schistosity, joints properand faults.

The most frequently used analyticalapproach is the stereographic projection(see Appendix 7 for more details). On thebasis of field observations and stereogra-phic plots, the pattern of joint sets can bedescribed with the help of Table 10.

3.2.2 – Statistical analysisof geometrical parametersfor each joint set

Once the directional joint sets (if any) havebeen identified, statistical analysis of eachset can proceed, by means of distributionhistograms of geometrical parameterssuch as strike, persistence and spacing.One can then calculate the means andstandard deviations for each parameter,and a distribution function may be fitted.

3.2.2.1 – Joint set orientation

The orientation of each joint set must beconsidered in relation to the direction oftunnel driving. The dip angle β and theangle δ between the axis of heading A andthe direction of the dip vector ap for each

joint set dictate the tunnellingmethod, as described in Table 11.The stereogram in figure 5 is a gra-phical representation of theseclasses, with explicit examples infigure 6.

3.2.2.2 – Joint spacing in each joint set

The histogram of joint spacing ei in a jointset can be readily obtained from the dis-tances di between discontinuities in thesame set intersecting the survey lines, withreference to the angle θ between the sur-vey line and the normal to the mean planeof the joint set (ei = di.cosθ). The next stepis to calculate the mean value ES and stan-dard deviation of the spacing, and variousmodal values if they emerge clearly fromthe histogram.

In bedded rock masses, spacing is equal tobed thickness.

Joint spacing classes are listed in Table 12.

3.2.2.3 – Joint set persistence

Joint persistence must be analysed withcaution. It can only be estimated from

CLASSE DESCRIPTION OF JOINT SET NUMBERS

Few discontinuities or occasional random discontinuities

One main set

One main set plus random discontinuities

Two main sets

Two main sets plus random discontinuities

Three or more main sets

Three or more main sets and scattered discontinuities

Abundant discontinuities in no discernible pattern

N 1

N 5

N 2a

b

N 3a

b

N 4a

b

Table 10 – Classes and descriptions for joint set numbers

CLASS

ORIENTATION OF DISCONTINUITIESDESCRIPTION OF TUNNELLING

CONDITIONS

Indeterminate

0° to 30°

30° to 65°

65° to 90°

0° to 20°

20° to 90°

20° to 90°

20° to 60°

60° to 90°

With subhorizontal bedsOR 1

OR 3

OR 2a

b

Table 11 – Joint orientation classes with respect to tunnel axis

Angle δ between direction ofdip vector αp and axis

of heading Dip β

Cross-bed (a) with dip

(b) against dip

OR 4a

bWith strike

(a) moderate dip

(b) steep dip

Intermédiate conditions

Figure 6 – Sketches of selected OR orientation classes from Table 11and stereogram data from figure 5

Stereogram and plot of plane pole for the joint set considered(upper hemisphere) at left

Explanatory block diagram at right

Sectors in the stereogram correspond tothe geometrical locus of the poles of jointplanes oriented according to OR classes

from Table 11. See also sketches in figure 6

Figure 5 – Joint orientation. Stereogram and polar plot for one joint set



measurements of 2D trace lengths on sur-faces (outcrops, adit walls), but these dataare often biased. Furthermore, 2D tracedata can only be extended to actual 3Djoint persistence with the aid of a model.

Appendix 8 reviews recent developmentsin these two areas.

3.2.3 – Overall joint density indexes

The spatial density of a population of dis-continuities is defined as the mean jointarea per unit volume of rock. This definitionencompasses both joint spacing and jointpersistence and cannot be determineddirectly. It is usually only possible to esti-mate joint density by counting lengths ofmassive rock blocks between discontinui-ties encountered along a survey line inboreholes, outcrops or adit walls.

Joint data (without any morphogeneticbreakdown) is reduced for interpretationby means of various indexes and statisticalmanipulations.

3.2.3.1 – Rock Quality Designation

RQD was defined by D. Deere (1963) asthe cumulative length of intact core pieceslonger than 4 inches divided by the totallength of the core run not more than 1.50mlong, expressed as a percentage. Core sizemust be least NX (21/8 inches or 54,7 mm),drilled with a double-tube core barrel, withsubstantially 100% recovery.

In order not to penalise ground not yiel-ding exactly 100% recovery and systema-tise RQD records for the purposes of com-parison, AFTES recommends calculatingRQD from 1-metre core runs:

RQD, expressed as a percentage, is thetotal length of pieces of intact core longerthan 10 cm divided by core run of 1-metrelength.

In order to keep RQD meaningful, the fol-lowing coring conditions are critical:

• Core diameter greater than 50mm inmassive and poorly jointed rock mass; inrock mass of a kind which is inherently join-ted or incorporates planes of weakness(schists, mudstones, marls, bedded limes-tones, etc.), larger borehole sizes arerecommended, such as 85mm.

• Length of intact core pieces measuredalong the core centreline.

• Core recovery index R (percentage of reco-vered core lengths per core run legth) to be90-100%; the RQD index for R values lowerthan 90% is not considered meaningful.

• Only natural discontinuities must beconsidered; breakage caused by drilling,core handling and placing cores in theboxes should be ignored (this is particularlyimportant in inclined boreholes which aremore susceptible to core breakage);doubtful cases should be considered as

natural joints and included in the RQD cal-culation as a conservative measure.

• Discontinuities substantially parallel to thecore centreline to be ignored in the count.

• RQD must be determined promptly afterthe core is recovered in order to forestallany subsequent changes in the materialdue to swelling, stress release or drying;the recommendation at the beginning ofsection 3 above to systematically photo-graph all cores is critically important here.Photographs taken at the time of the RQDcount allow any changes to the cores overtime to be identified.

RQD classes and assessments of the ove-rall quality of the rock (see para. 4.1.1) thatcan be derived from the D. Deereapproach are presented in Table 13.

3.2.3.2 – Interval between discontinui-ties (ID index)

The ID index is defined as the mean of theintact rock length between successive dis-continuities along a survey line whoselength and orientation must be recorded.The counts must be made in several direc-tions, carefully chosen with reference to thecharacteristic joint strikes and tunnel orien-tation (this is also valid for the RQDapproach).

The joint density classes obtained from theID index are listed in Table 14

ES 1 > 200 Very widely spaced discontinuities

ES 2 60 à 200 Widely spaced discontinuities

ES 3 20 à 60 Moderately spaced discontinuities

ES 4 6 à 20 Closely spaced discontinuities

ES 5 < 6 Very closely spaced discontinuities

CLASS SPACING (cm) DESCRIPTION

Table 12 – Joint spacing classes within same joint set

RQD 1 90 to 100 Excellent

RQD 2 75 to 90 Good

RQD 3 50 to 75 Fair

RQD 4 25 to 50 Poor

RQD 5 0 to 25 Very poor

RQD % DESCRIPTION(After D. Deere)

CLASS

ID 1 > 200 Very low density

ID 2 60-200 Low density

ID 3 20-60 Moderate density

ID 4 6-20 High density

ID 5 <6 Very high density

CLASS ID INDEX (CM) DESCRIPTION

Table 14 – Joint density classes as determined along survey line

Table 13 – Overall rock quality classes by the RQD method* *after D. Deere



It is also recommended (figure 7)

• To draw the histograms of lengths foreach survey line and calculate standarddeviation S and coefficient of variation CV

CV = S/ID

• To plot the cumulative frequency curve ofcore lengths (equivalent to a grain size dis-tribution curve) in order to have a complete

image of the joint density. This curve canbe used to quantify the mean joint indexrepresented by the median, and dispersionrepresented for example by the 25% and75% quartiles.

The joint frequencies FD can also be calcu-lated (Table 15); it is the number of jointsper metre of drilled core, the inverse valueof the ID index.

3.2.3.3 – Remarks on use of indexes

It must not be forgotten that the determina-tion of the above indexes from boreholecores requires the coring conditions stipula-ted in para. 3.2.3.1 always to be adhered to.

Attention is also drawn to the fact thatRQD, the most commonly measured para-meter, is a quality index for the rock massbut yields very little information on jointdensity. For example, an RQD of 100 mayresult from a core run not encountering adiscontinuity or exhibiting a discontinuityevery 11cm. AFTES therefore recommendsusing the ID index to characterise jointdensity.

It is also strongly recommended to repre-sent the variations of these indexes alongthe survey line in the form of diagrams. Forborehole measurements, it is essential toaccompany the complete core diagramshowing core piece lengths against depthwith the FD diagram (expressed in m-1)and the RQD diagram (figure 8). This arran-gement facilitates analysis of the rock massdiscontinuities, and the different methodsbecome mutually complementary.Homogeneous zones can be identified, orcontrasts between zones with more or lesshigh density of discontinuities. The discon-tinuity data can also be compared withdata from downhole logging performed inthe same borehole.

It might also be useful to quantify thesevariables by a moving mean process (e.g.RQD and FD calculated in one-metre stepsin a 4m window moved down the borehole

Figure 7 – Histogram and cumulative frequency curve for lengths off pieces of core (after C. Louis 1974)

FD 1 FD < 1 Very low density

FD 2 1 < FD < 2 Low density

FD 3 2 < FD < 5 Moderate density

FD 4 5 < FD < 15 High density

FD 5 FD > 15 Very high density

CLASS ID INDEX (CM) DESCRIPTION

Table 15 – Joint frequency classes as determined along survey line

Figure 8 – Characterisation of thediscontinuity network in a limes-tone rock mass from verticalcored borehole (Serratrice &Durville 1997)



in 2m increments) or modify the values ofthe parameters used to calculate indexesin specific zones in the borehole such ascrushed zones which can conventionally beconsidered as consisting of 1cm fragments.

3.3 – MECHANICAL PARAMETERSOF DISCONTINUITIESWe are interested here in discontinuitieswithout infilling material, otherwise mecha-nical behaviour would be governed by thebehaviour of the infilling material, whichshould then be studied in itself.

The mechanical characteristics of disconti-nuities are usually obtained from laboratorytests; in situ tests are much less commonbecause they are more difficult to performand are more costly. Uniaxial compressiontests and shear tests with normal load(French standard NF P94-424) are perfor-med to characterise the behaviour of thediscontinuities.

The following leading parameters charac-terising the mechanical behaviour of dis-continuities can be obtained by analysingthese laboratory test results :

• deformation parameters: normal and tan-gential stiffness,

• shear strength defined by peak and resi-dual friction angles and apparent cohesion,

• a geometric parameter, dilatancy, a mea-sure of the deformation in the normaldirection accompanying tangential defor-mation during shear.

3.3.1 – Deformation parameters

3.3.1.1 – Normal stiffness

Uniaxial compression tests on joints orien-ted perpendicular to the direction of loadapplication always trace a hyperbolic curveof normal stress σn versus normal displace-ment Un with an asymptote representingthe maximum limit of joint closure Umax.Total joint opening can be obtained for anon zero tensile stress a, if there are rockbridges across, or filling material in, thejoint (Figure 9).

The slope of this curve gives the normalstiffness Kn, defined as

Kn = δσn/δUn

The value of Kn is dependent on the normalstress and can be expressed in terms ofparameters α, Umax and Kni (initial normalstiffness) characterising the mechanicalbehaviour of the joint subjected to uniaxialcompressive load, determined by fitting acurve on the test results, although not com-monly done.

3.3.1.2 – Tangential stiffness

Similarly, the shear test is used to definethe tangential stiffness Ks as the slope ofthe curve of tangential stress τ vs tan-gential displacement Us before failure(figure 9):

Ks = δτ/δUs

3.3.2 – Shear strength parametersThe behaviour of a discontinuity during ashear test (French standard XP P 94-424) isgoverned by the nature of the joint wallsbut more importantly, by their surfaceconditions. In particular, joint wall rough-ness, interlocking and weathering play aprimordial role.

In the ideal case of a planar and smoothdiscontinuity, i.e. with no asperities, shearbehaviour is entirely controlled by wall fric-tion. Shear strength is usually expressed bythe Coulomb criterion:

τ = σn x tan φb

in which φb is the friction angle for a planarjoint or basic friction angle, chiefly depen-dent on the petrographic composition anddegree of weathering of the joint walls.

Natural discontinuities generally have wallswhich are very irregular with abundantasperities of varied shape and size, repre-senting different scales of roughness. Theirshear behaviour reveals three fundamentalparameters (figure 10):

• Peak shear strength, defined by the maxi-mum shear stress τp, at which the asperitiesshear.

• Residual shear strength τr, characteristicof the friction of the joint walls which comeinto contact with each other after the aspe-rities have sheared.

Figure 9 – Joint deformability parameters: normal stiffness Kn and tangential stiffness Ks

Figure 10 – Shear behaviour of a natural discontinuitya) tangential stress τ vs tangential strain Usb) Normal strain Un vs tangential strain Us

• Dilatancy represented by the displace-ment of the joint walls in the direction nor-mal to the joint plane. It is characterised bythe dilatancy angle i (angle of slope of thedilatancy curve of normal displacement Un

vs tangential displacement Us). This anglereaches a maximum value ip at the inflec-tion point on the dilatancy curve. This pointcorresponds to the peak shear stength τp

reflecting, for a given level of normal stress,the shearing of the sharpest asperities.Beyond this point, dilatancy continues witha lower angle, governed by the inclinationof the stronger asperities with a wider baseand flatter angles.

Compared to a planar smooth joint, dila-tancy leads to an increase in peak strength.It is dependent on joint wall roughness andweathering, and also how the walls inter-lock and the direction of shear.

The joint shear failure criterion is represen-ted by two curves, characterising the peakand residual strength (figure 11).

Residual strength of discontinuities is notgreatly influenced by scale effect and thefailure criterion is readily obtained by labo-ratory testing in the form of a standardCoulomb law characterised by a residualfriction angle φr which differs from the basicfriction angle φb by no more than a fewdegrees, and a residual cohesion, usuallyminimal or zero, which is always consideredto be 0:

τr = σn x tan φ

The peak shear strength curve has a pro-gressive shape reflecting the non-linearrelationship between shear strength τ andnormal stress σn. The curve is steep at lownormal stresses, reflecting the influence ofthe sharpest asperities, which are the causeof severe dilatancy. As normal stressincreases, more and more asperities fail,dilatancy becomes less and the (τ, σn) curveflattens and progressively becomes a

straight line. For a limited normal stressrange, this curve can be approximated by astraight line:

τpeak = Ca + σn x tan φpeak

Ca is an apparent cohesion which does notexpress an intrinsic property of the jointwall material but the influence of irregulari-ties in the walls on shear behaviour.

At very low normal stresses, apparentcohesion Ca is close to zero, and φpeak isclose to φr + ip.

At high normal stresses, apparent cohesionCa is high and peak friction angle φpeak

tends towards φr.

In practice, laboratory tests are not easy tointerpret and determination of peak jointstrength characteristics involves many diffi-culties arising from scatter in the data andscale effect.

Using experimental data, Barton (1973)proposed a semi-empirical failure criterionin which peak strength depends on a dila-tancy angle i allowing for the joint wallroughness (JRC), joint strength (JCS) andnormal stress applied to the joint:

τpeak = σn x tan (φb + i)

= σn + tan [φb + JRC x log10 (JCS/σn)]

in which φb is the basic friction angle whichdiffers from the residual friction angle φr bya few degrees.

JRC is the Joint Roughness Coefficient, adimensionless coefficient relating to jointwall roughness and size. It can be estima-ted by comparing joint roughness profilesin the direction of shear with Barton's stan-dard profiles, ranked in ascending orderfrom 0 for a flat smooth discontinuity to 20for a wavy rough discontinuity (figure 12).JRC also varies with the joint deformabilitywall displacement: the more asperities aresheared, the lower the value of JRC.

RCS is the Joint Compressive Strength, fre-quently estimated indirectly in situ with theSchmidt hammer (Appendix 9).

σn is the normal stress applied to the dis-continuity.

At very low normal stresses (JCS/σn ≥100),the equation gives unrealistic values andBarton suggests using the simplified form:

τ = σn x tan 70°

However, the determination of a represen-tative value of JRC for three-dimensionaljoint wall roughness is not always a simplematter, even on laboratory size samples.

It should be recognised that this approachis confined to discontinuities with thin or noinfilling material. When there is a sufficientthickness of infilling material for shearing tooccur wholly within the infilling material,shear characteristics will be those of theinfilling material, which must be investiga-ted specifically.

3.3.3 – Hydraulic parametersWhile the mechanical behaviour of rockjoints is mainly controlled by joint wall com-position, weathering, roughness and nor-mal stress, other external factors affectbehaviour: thickness, composition andmoisture content of infilling material, pre-sence of water in joints likely to inducepore pressures modifying normal stress,and boundary conditions affecting themagnitude of displacements.

Fracture fluid flow is a highly complex sub-ject. Experiments have shown it not isotro-pic but occurs preferentially along channelswhose geometry depends, of course, onthe aperture of the discontinuity, but alsoon wall roughness and surface of contactbetween walls, applied normal and tan-gential stresses, and tangential joint displa-cements, plus of course the possible pre-sence of infilling material.

Various more or less simplified approachescan be used to estimate the flow rate Q ofa fluid circulating in a discontinuity (seeappendix for details). For a planar andsmooth discontinuities, the flow is gene-



Figure 11 – Failure criterion for a natural discontinuity

Figure 12 – Standard joint wall roughness profiles(after Barton & Choubey 1977)



rally assumed to be proportional to thecube of the discontinuty aperture(Appendix 10).

But even allowing for roughness, it mustnever be forgotten that measured flowsoften differ substantially from rates calcula-ted by these methods.

4 – CHARACTERISTICS OFROCK MASS

4.1 – IDENTIFICATIONPARAMETERS

4.1.1 – RQDRQD determined from jointing (see para.3.2.3.1) was originally considered as anindex of rock mass quality determined bycounting discontinuities in borehole cores.

If the discontinuities are all oriented moreor less uniformly in all three dimensions('isotropic' jointing), RQD can be taken asindependent of the direction of the bore-hole and can effectively be considered as aoverall index of the quality of the rockmass.

But if the distribution of discontinuities isstrongly polarised (as in finely beddedrocks, schists, slates, etc.), the value of theRQD index will differ widely with differentdirections of drilling (figure 13). The RQDfrom a single borehole therefore will giveonly a 'snapshot' of the jointing in a givendirection rather than a representation ofthe overall jointing of the rock mass.Because of this, AFTES recommends thatRQD should be determined from severalboreholes drilled in different directions tointersect all joint sets, especially thosewhich may be unfavourable for the plan-ned underground structure.

RQD, originally defined by its author onthe basis of counting discontinuitiesfound in borehole cores, can also bedetermined by counting them on expo-sed rock surfaces:

• Natural outcrops: the count proceedsalong one or more lines intersecting thenetwork of fractures such that the valuesobtained is truly representative of thehomogeneous blocks of rock as definedelsewhere.

• Quarry faces, trench sides, adit walls: inthese cases, cracking caused by the use

of explosives should be ignored as far aspossible.

If the rock mass displays strongly polari-sed discontinuities, the same reservationscan be made as in the case of boreholesas to the representativeness of the dis-continuities recorded and the RQD calcu-lated with reference to the direction ofthe survey line.

4.1.2 – Degree of alteration The degree of alteration of a rock mass isdescribed by breaking it down into altera-tion zones for the different geological for-mations present. A distinction is made bet-ween weathering proper and hydrothermalalteration occurring at depth (frequentlylinked to contemporary or more ancientvolcanism). Alteration of the rock mass as awhole is classified as the sum of the wea-thering of the rock matrix and of the majorjoints.

With weathering properly so called, des-criptive terms, conforming to thoserecommended by ISRM (AM = W), appearin Table 16. They apply predominantly tocrystalline rocks.

4.1.3 – Rock mass continuity index ICMUsing the same procedure as described inpara. 2.1.10 for the rock matrix, a rockmass continuity (or quality) index ICM canbe defined as the ratio between P wavevelocity as measured over a base length L(VpM) and the velocity measured on asample (Vp):

ICM = VpM/Vp

This concept of a rock mass continuityindex ICM makes it possible to estimate theimpact of the scale effect and deteriorationin the mechanical properties the rock masscompared to results from laboratorysamples (rock matrix).

Figure 13 – Influence of direction on the characterisationof discontinuities in finely bedded formations

AFTES DESCRIPTION OF THE ROCK MASSCLASSES

AM1a Sound rock

AM1b Poorly weathered rockWeathering confined to surfaces of main discontinuities; rock sound in the mass

AM2 Slightly weathered rockLittle weathering of rock in the mass but well developed in discontinuities

AM3 Moderately weathered rockWeathering clearly visible in whole rock mass but material not friable

AM4 Well weathered rockSevere weathering in the mass

AM5 Completely weathered rockTexture and large fractures still visible

AM6 Completely decomposed rockTexture and fractures unrecognisable

Residual soil - Undisturbed

Table 16 – Rock mass weathering classes and descriptions



Whatever the absolute value of Vp, if VpM isequal to Vp, this means that the rock mass,at the scale L at which VpM was measured,displays the same properties as the sampleand is unaffected by discontinuities or voidswhich would reduce P wave velocity.

If however, VpM is less than Vp, the lowervelocity in the rock mass than on the labo-ratory samples can be attributed to discon-tinuities and voids in the rock mass overbase length L at which VpM was measured.

Rock mass continuity classes, at the scaleof L over which VpM was measured, are lis-ted in Table 17.

Base length L is usually the same as stan-dard seismic refraction test base lengths(60m, 120m, 240m) but rock mass conti-nuity can be measured over shorter baselengths, based on borehole microseismicsor adit wall seismic tests.

Of course, base length L must always bestated alongside the relevant ICM value.

Note. ICM may be greater than 100%, forexample when the rock matrix containscracks or microcracks closed tight by theconfining pressure in the rock mass butwhich might open when coring releasesthese stresses. This is a not uncommonoccurrence in some schistose rocks.

4.2 – MECHANICAL PARAMETERS

4.2.1 – Rock mass deformability,rock mass deformation modulus EMas

Because of the discontinuities, rock massdeformation at prototype scale is generallymuch greater than for the intact rock matrixas determined from small laboratorysamples.

Depending on rock mass volume and theloads applied to it, deformability may beapprehended through two classes of in situinvestigations:

• 'Indirect' geophysical methods, basedprimarily on wave velocities.

• 'Direct' methods consisting of measuringdeformations on parts of the rock massunder changing states of stress. Thechanges may be brought about by specificload tests (in situ tests) or by construction ofthe tunnel (back analysis).

4.2.1.1 – Indirect (geophysical) methods

P (longitudinal compression) wave and S(transversal shear) wave travel time is mea-sured over a known distance between theseismic source and pick-ups. Emitter andpick-ups can be arranged in various ways:

• In separate boreholes (seismic cross holetest)

• Source in a borehole and pick-up atground level (seismic up-hole test)

• Pick-up in the borehole and source at sur-face (seismic downhole test).

All arrangements measure compressionwave Vp and shear wave Vs velocities toderive the 'dynamic' deformation modulusand 'dynamic' Poisson's ratio of the rockmass (Ed and νd respectively) through thefollowing equations:

Ed = ρ[Vp2(1 + νd) x (1 – 2νd)]/(1 - νd)

νd = [0.5 – (Vs/Vp)2]/[1 – (Vs/Vp)2]

(ρ is density, see para. 2.1.4).

Computing Ed and νd by geophysics meansmeasuring both Vp and Vs.

If only the compression wave velocity Vp

has been recorded, the Ed modulus can stillbe obtained by taking an assumed valuefor Vs (usually 0.25 or 0.30).

Other methods based on seismic velocitiesin the rock mass can yield estimates of thecorresponding moduli, through experi-mental correlations with past constructionsites (Schneider's 'Petite Sismique'method, SCARABEE method).

It must be realised that the term 'dyna-mic' here in fact refers to very small strains(10-7 < ε < 10-5) under very small loads.

4.2.1.2 – Direct measurement

The main in situ tests for measuring rockmass deformability are:

• Rigid plate loading test, quite wides-pread and routine. It characterises thedeformability of the rock mass throughdeformation modulus E determined fromthe tangent to the envelope curve of the'load-displacement' curves under increa-sing loading cycles (cf. Appendix 11).

With usual plate sizes (0.28m to 0.60m),this test yields rock mass deformabilityvalues at the scale of a few cubic metres ofrock, provided the pressure is high enoughto penetrate beyond the decompressedsurface zone.

• Borehole dilatometer test (French stan-dard P 94-443). The instrument consists ofa deformable cylindrical cell applying acontrolled radial pressure to the boreholewalls, and several strainmeters directlymeasuring the radial deformation of theborehole wall under the applied pressure.

The E modulus of thr rock mass is calcula-ted with the following equation in which, inthe absence of specific data, the Poisson'sratio n is frequently taken as 0.25:

∆σr/∆εr = E/2(1 + ν)

∆ε is the change in radial strain producedby the change in applied stress ∆σ on theborehole wall.

The strainmeters measuring the radialdeformation at the borehole wall must dis-play sufficient resolution to measure theusually high moduli encountered in rock.They are arranged in different directions atdifferent places along the length of the dila-tometer (3 pairs at 120° or 4 pairs at 90° indifferent models). This arrangement showsup any anisotropy in rock deformation.

The dilatometer can also be used for creeptests in which the applied pressure is heldconstant over time and time-dependentdisplacements are recorded.

• Borehole pressuremeter test (standard P94-110-1 & 2) also measures mass defor-mability by means of a deformable cylindri-cal cell applying an increasing pressure tothe borehole walls. The slope of the 'pres-suremeter curve' showing the change inpressuremeter cell volume vs applied pres-sure is used to calculate a shear modulus G(Menard pressuremeter modulus EM).

However, the characteristics of the instru-ment mean that the pressuremeter test isunsuitable for rock mass determinations,

ICM 1 > 90 % Very high continuity

ICM 2 90 % to 75 % High continuity

ICM 3 75 % to 50 % Moderate continuity

ICM 4 50 % to 25 % Low continuity

ICM 5 < 25 % Very low continuity

CLASS ICM DESCRIPTION

Table 17 – Rock mass continuity classes at L scale



even if it is still widely used. The rea-son is that, with moduli of a few hun-dred MPa (a value which is abun-dantly exceeded in rock), the modulifrom pressuremeter tests becomeincreasingly underestimated whereasactual moduli increase. The pressure-meter test must only be used forsoils and some 'soft' materials (chalk,marl) on the borderline between soiland rock. When dealing with theusual types of rock, the pressureme-ter should not be reckoned amongthe panoply of relevant testmethods.

4.2.1.3 – Measurement on actual structures and estimation of deformability by back analysis

These are probably the most effectivemethods for finding the large scale defor-mability of a rock mass and the anisotropyparameters governing it.

On actual structures (generally exploratoryadits driven prior to the full size construc-tion stage), the measurements most com-monly performed are the following:

• displacements of the adit wall (conver-gence)

• displacements of points within the rock(displacements relative to fixed or movingreference points) by means of boreholeextensometers around the adit

• angular changes between studs fixed tothe rock (inclinometers or deflectometers).

Back analysis, generally with 2D or 3Dcomputer models (finite element and simi-lar models) allows the engineer to workback from the known stress state to themost plausible rock moduli under theconditions of the completed structures. Inthe case of anisotropy in the moduli, whichusually accompanies anisotropic states ofstress, back analysis is more difficult.

When checking the design of very highhead water pressure tunnels, the concretelining subjected to the high water pressureis instrumented (to measure diameters andstresses). This type of 'chamber' test alsoprovides a check on rock watertightness.

4.2.1.4 – Classification of rock massdeformability

Rock mass deformability classes based onthe rock mass deformation modulus EMas

are listed in Table 18.

4.2.1.5 – Time-dependent effects – long term modulus

The construction of an underground struc-ture always causes deformations in the sur-rounding rock due to changes in the stressfield around the opening. In many cases,rock deformations are accompanied byeffects which appear over time and defor-mations increase asymptotically towardswhat is generally called the "long termstate."

As discussed in para. 2.2.2, there may beseveral causes of time dependent rockmass behaviour:

• Rheological behaviour specific to therock mass, viscoelastic or viscoelasticplastic

behaviour (as in certain rocks such asevaporates, marls, etc.).

• The elastic limit may be exceededwith plastic zones developing aroundthe opening.

• Time-dependent deformations maybe linked to consolidation processessubsequent to changes in flow pat-terns, with the opening acting as adrain, or to the original pore pressurepatterns gradually re-establishing afterthe disturbance caused by excavation.

These causes may be concomitant,and this aggravates the difficulties of cor-rectly interpreting the observed timedependent behaviour.

At present, the most widely used simpli-fying approach to the understanding oftime dependent rock mass behaviour is toconsider the rock mass deformation modu-lus as a decreasing time function:

EMas(t) = EMas0/[1 + Φ(t)]

in which

- EMas0 is the instantaneous rock mass defor-mation modulus

- EMas(t) is the rock mass deformationmodulus at time t under a load that has notchanged since time t = 0.

Φ(t) is a monotonous function increasingfrom Φ(0) = 0 to Φ(∞) = α.

In low- to moderate-strength rock, α = 1 isoften suggested even if it does not alwaysappear justified by experimental andother data.

In stronger rocks, α = 0.3 to 0.5 are also fre-quently proposed without any justificationfor the choice.

A less doubtful choice of these valueswould require long-term instrumental datafrom completed structures. At the presenttime, little feedback is available andrecords cover only a limited number ofyears.

4.2.2 – Rock mass limit strengthMechanical properties are strongly influen-ced by the geometrical dimensions of thevolumes of rock involved, in the generalsense of these properties becoming worseas the rock volume grows (scale effect). Insitu rock mechanics tests to failure such asshear tests and hydraulic fracturing involveconsiderable effort and still only concernlimited volumes of rock.

Photo 7 – Rigid plate loading test.Volcanic agglomerate,Takamaka, Reunion Is.

DM 1 > 30 Very low deformability

DM 2 10 to 30 Low deformability

DM 3 3 to 10 Moderate deformability

DM 4 1 to 3 Fair deformability

DM 5 0.1 to 1 High deformability

DM 6 < 0.1 Extremely high deformability

CLASS ROCK MASS DEFORMATIONMODULUS EMas (GPa)

DESCRIPTION

Table 18 – Rock mass deformability classes



There are not really any tests for characte-rising the mechanical strength of a rockmass. Only an empirical approach basedon feedback from past construction jobs ispossible. This leads to modifying the crite-ria for samples (see para. 2.24) by downs-caling the characteristic parameters whenextrapolating from the intact rock matrix tothe large scale rock mass (see 5.3.4).

4.3 – HYDROGEOLOGICALCONDITIONS

Ground water is the cause of many difficul-ties encountered in underground enginee-ring:

• Flowing water slows down excavationwork

• Water pressures may destabilise the tun-nel walls or lead to fearsome squeezingground into the tunnel

• Dewatering may have severe environ-mental consequences: flow depletion fromsprings and wells, subsidence due togroundwater lowering.

Basically, flow rate Q through a section S ofthe rock mass is related to permeability Kand hydraulic gradient i by Darcy's law:

Q = K.S.i

Flow is associated with body forces propor-tional to the hydraulic gradient i.

Characterisation of the hydrogeologicalconditions in a rock mass therefore pro-ceeds in three steps:

1. Identify aquifers and how they function.

2. Measure hydraulic head H on the tunnel.

3. Measure rock mass permeability KM.

In practice, these three steps do not neces-sarily proceed in the order describedbecause it may not be possible to eluci-date the existence of certain aquifers andtheir functioning until exploratory workshave been undertaken. It is also importantto realise that such investigations mustcover the whole aquifer system affected bythe tunnel, not only the part through whichthe tunnel passes, as is the case with inves-tigations for mechanical parameters.

4.3.1 – Identification of aquifersThe extent of the aquifer system liable tobe affected by the underground structureis determined with the aid of the geologi-cal model mentioned in section 1 above,identifying the main individual aquifers if

applicable. The hydrogeological functio-ning of the system is then tentativelymodelled, with rough estimates, for eachaquifer crossed by the project, of

a) The type of permeability concerned,which can be classified under five hea-dings:

1. Granular material (sand and gravel);

2. Jointed rock (granite, gneiss, basalt, etc.)with water circulating only through the dis-continuities;

3. Double porosity ground in which watercirculates both through discontinuities andthe porous rock matrix (chalk, sandstone)or weathered rock matrix (severely weathe-red granite);

4. Karstic rock (limestone, gypsum) inwhich most of the water circulates throughrandomly distributed voids of various sizes;

5. Fault zones with breccia infill frequentlyacting as drains within fractured rock-masses.

b) Boundary conditions, i.e.

- sources (rainfall, infiltration, river, lake,sea, etc.)

- flow rate at point of outflow

- watertight boundaries.

This first step involves a desk search andinspection of a project survey plot; it usesdata from the geological model anddemands considerable engineering expe-rience. It should not overlook local sourcesof information from groundwater users,amateur kart environment, etc. In karsticsettings, it is important to know whetherthe karsts are active or fossil karsts more orless filled in.

4.3.2 – Measurement of initial piezometric conditionsKnowledge of the pre-construction piezo-metry in the rock mass at the project site isa critically important factor in good design.The project may be a very long tunnel andpre-construction piezometry must bedetermined along its whole length. Itseems unnecessary to stress the risks ari-sing from ignorance of the initial piezome-tric conditions (even over a short length oftunnel) both during construction and sub-sequent operation of the structure.

Where more than one aquifer and differentpressure heads are suspected and in zoneswhere the relief may be an influencing fac-tor (hillsides, valley bottoms, etc.) local pie-

zometric testing may be indicated, usingpressure cells; open well piezometersshould not be used.

Piezometry is usually subject to seasonalfluctuations and it is strongly recommen-ded that piezometric monitoring of eachaquifer identified should commence at thevery earliest stage of the design process. Itwill often be necessary to have severalyears' records before being sure of theamplitude of the piezometric fluctuationsto be expected and designed for. In mostcases, only continuous records will show upsometimes short-lived transients whichmay have a serious impact on the project(karstic aquifers, tidal river reaches, etc.).The designer must also assess the risk ofthe water table rising, especially in urbanareas, due to local abstractions beingunexpectedly interrupted. Lastly, know-ledge of changes in the peizometry andflow rates leads indirectly to certain aquiferhydrodynamic parameters.

Hydraulic head classes are listed in Table 19.

As a general rule, every cored boreholesunk to investigate a tunnel site should befitted out as a piezometer, in view of theimportance of this parameter.

4.3.3 – Measurement of rock masspermeability KM

Determining the permeability of the rockmass calls for interpretation of the resultsof hydraulic tests, which must be chosento suit the type of aquifer concerned.Available tests are as follows.

• Localised tests in boreholes:

- either steady-state tests where permeabi-lity is moderate to high, such as the stan-dard Lefranc test for soils and Lugeon testin jointed rock (water is injected at 1 MPa

H 0 Lower than invert Zero head

H 1 < 5 Low head

H 2 5 to 20 Moderate head

H 3 20 to 100 High head

H 4 > 100 Very high head

CLASS

INITIAL HYDRAULIC HEAD H

(in metres above tun-nel invert

DESCRIPTION

Table 19 – Hydraulic head classes



pressure); the Lugeon test measures appa-rent permeability and supplements thejointing data obtained from cores andmechanical behaviour;

- or transient-state tests, causing instanta-neous changes in the open (slug test) orclosed (pulse test) test interval, or a combi-nation of them (drill stem test); these morecomplex tests are suitable for low permea-bility conditions (K < 10-7 m/s).

• Larger-scale tests, such as:

- pumping-out tests with measurement offar-field water table drawdown in suitablylocated boreholes;

- measurement of tunnel drainage (the tun-nel may be divided up into separatelengths for this purpose).

When conducting these tests, it is importantto ensure that the disturbance induced bythe test is of the same order of magnitude asthe disturbance that will be caused by thetunnel, so as not to disrupt the environment.In all cases, upscaling measured data to thewhole aquifer calls for great caution.

Rock mass permeability classes are listed inTable 20.

Permeability in jointed rock is very oftenanisotropic: it may typically be ten timesgreater parallel to the bedding or cleavagethan in the perpendicular direction.Anisotropy may also depend on the orien-tation of the principal stresses. Equivalentpermeability for the whole rock mass isgiven by a tensor; for classification pur-poses, the highest permeability coefficientis used, stating the direction in which itapplies; the anisotropy ratio Kmax/Kmin isalso used.

When running borehole tests, it is recom-mended to proceed in contiguous 5m to10m stages and plot a permeability logwhich can be usefully compared to thejointing log.

Other useful data that can be loggedconcerns:

• hydraulic head, by measuring pressuresbetween packers,

• borehole inflow/outflow, by continuousmeasurement with a miniature flowmeter.

When dealing with deep-lying mine wor-kings, tunnels, mined storage chambersand other deep structures, understandingthe hydrogeology is complicated by thepossibility of exploratory drilling causingunwanted interconnections between sepa-rate aquifers. This calls for more sophistica-ted techniques as used in the mining andoil industries, such as those described inthe AFTES Working Group 24Recommendations. They accurately locateaquifers through the use of fluid logs andmultilayer tests, and make it possible totest them selectively with packers andprobes.

4.3.4 – GasMethane (CH4), nitrogen (N2), hydrogensulphide (H2S), carbon monoxide anddioxide (CO and CO2), radon 220 or 222(Rn) and other gases may be present in thefree state or dissolved in the ground waterwithin certain sedimentary formations (car-bonaceous, carbonate, argillaceous andsaline rocks) or igneous formations (e.g.granite). When such formations host anunderground opening, the gases theycontain tend to migrate towards the exca-vation, creating a risk of explosion, poiso-ning, suffocation or disease (cancer andother occupational illnesses), not onlyduring excavation, but equally during theservice life of the structure. While such risksare more the province of health and safety,they must be addressed at the tunnel andventilation system design stage and musttherefore be accorded special attentionwhen proceeding with the geotechnicalcharacterisation of the ground to host thestructure when such gases may be presentwithin it (coal measures for example).

4.3.5 – Other parametersIn addition to the H and K parameters foreach aquifer identified, other parametersmay be of use in characterising the rockmass, in particular:

• Storage coefficient S, representing thecapacity of the rock mass to store water.This must be investigated for modellingtransient flow conditions; it can be derivedfrom pumping-out test data and may rangefrom 10-5 (10 cm3 of water released whenthe hydraulic head in a 1 m3 volume ofsaturated ground is lowered by 1m) to 0.1to 0.15 in clean sands.

• Groundwater temperature, pH and che-mistry (and sometimes isotopes). Theseparameters serve more as indicators ofwhere the water comes from and helpunderstand the hydrogeological functio-ning of the aquifer system; and they makeit possible to assess how aggressive thewater will be for tunnel support and linings.

4.4 – INITIAL STATE OFSTRESS IN ROCK MASS

The initial stress state is a determining fac-tor in the response of the rock mass toexcavation: the convergence pattern on atunnel section, the location and extent ofzones where the limit strength of the rockmay be reached during tunnel driving, areall strongly dependent on the initial stressstate, and it is vital to consider it at thedesign stage.

4.4.1 – Initial state of stress andapproximationsComputation and modelling in the designstage make it possible to investigate andanalyse the impact of the initial state ofstress.

The state of stress is represented everyw-here by a tensor whose principal compo-nents are the σ1 (major), σ2 (intermediate)and σ3 (minor) stresses.

In the absence of data, it is commonly assu-med that the vertical is the principal direc-tion and that the vertical stress is equal tothe "weight of overlying ground," i.e.

σv = γ x z

These two assumptions are valid in subho-rizontal sedimentary formations but are notgenerally valid in mountain areas where therelief and tectonics introduce considerabledistortions, especially under mountainsidesand at valley bottoms.

K 1 < 10-8 Low permeability

K 2 10-8 à 10-6 Moderate permeability

K 3 10-6 à 10-4 High permeability

K 4 > 10-4 Very high permeability

K 5 Pratically infinite Karst permeability

CLASSROCK MASS

PERMEABILITYKM (m/s)

Table 20 – Rock mass permeability classes

DESCRIPTION

Photo 8 – Water flow into Saint Guillaume II tunnel fromGrange Pellorce fault, France



In sedimentary basins, a third assumption isfrequently made, that the horizontal princi-pal stresses σh and σH are identical andequal to a constant fraction of σv:

σh = σH = K0 x σv

This assumption may be very far from rea-lity since the horizontal stresses are rarelyisotropic and K0 frequently ranges from 0.5to 2 or more 5.

4.4.2 – Characterisation of stresstensorWhen designing any underground struc-ture, it is important to try to determine theinitial state of stress. Because of the difficul-ties involved in this, a step-by-step proce-dure is usually followed, based firstly onindirect approaches and then, if possible,on in situ test data which can be checkedduring construction by specific observa-tions (Table 21).

In the project planning stage when anapproximate estimate of stresses is suffi-cient, the designer focuses on indirect ana-lyses, using published information andconclusions that can be drawn from thegeological history and local topography ofthe project area. This first step should pos-tulate a stress range to be expected.

The estimated stress range and its impacton design may subsequently justify perfor-ming in situ field tests. If severe horizontalanisotropy is considered possible at thisstage, data on the azimuth of sH may leadto the orientation of the underground ope-ning being optimised (provided such free-dom is possible, in view of the purpose ofthe structure).

At the detailed site investigation stage, theavailable methods for stress measurementare always considered, not without reason,expensive, difficult to perform and interpretand above all extrapolate. On top of this,one can never directly "measure" a state ofstress; at best, one measures fluid pressuresconsidered as equivalent to the normalcomponent acting on a given surface, orstrains caused by stress changes. Despitethese difficulties, it is important to have ins-trumental data, especially when the presu-med mean stress is of the same order ofmagnitude as rock strength (cf. para. 4.4.4).

Lastly, once construction has commenced,the stress assumptions derived from thefield tests have to be viewed alongside theactual response of the tunnel walls anddata from monitoring instruments.

If the initial state of stress is found to be adetermining factor in project design and

feasibility, it may be decided to undertakefield testing at an earlier project planningstage.

4.4.3 – Commentary on field testmethodsWhen measuring in situ stresses, it is stron-gly recommended to plan an abundantnumber of mutually complementary tests,not only because of metrological difficul-ties, but more importantly because of theoften very sudden local variations in thestress tensor; these variations may be dueto lithological heterogeneities or the proxi-mity of geological discontinuities, fracturezones or even a free surface (favouringstress release). Such conditions make extra-polation of test data even more problema-tical and it is vital to check the data duringconstruction work.

The various test methods available havebeen described in Tunnels et OuvragesSouterrains No. 123 (Briglia et al. 1994).Only a few general recommendations willbe given now, focusing on boreholemethods.

a) Methods based on stress release, byovercoring, undercoring, or cutting a slotwith a borehole slotter, usually in boreholesthat are differently oriented; data on rockdeformability is needed. The CSIRO,USBM and other instruments involved arequite difficult to use properly and aremainly suitable for poorly jointed rock, inwhich they yield purely local information(on a decimetric scale); they can only beused in the elastic range.

b) The hydraulic fracturing method mea-sures the normal component of the stressacting on a discontinuity, by means of anelevated water pressure on a section ofborehole. It has the advantage of involvinga volume of rock several cubic metres insize, being feasible at great depths (inexcess of 1000m) and needing no assump-tions on rock behaviour. There are twovariants:

• "Standard" hydraulic fracturing, creatingartificial fractures perpendicular to theminor principal stress s3 to determine itsmagnitude and direction.

• The hydraulic test on pre-existing frac-tures (HTPF test) extends the scope tonatural fractures with different orientations.Provided enough tests are performed,HTPF is one of the most reliable means ofdetermining the complete stress tensor.

PROJECTSTAGE OBJECTIVES METHODS AND MEANS

*As an initial approximation, the P and T axes of focal mechanisms can be taken as s1 and s3

Table 21 – Stress estimation methods at different project stages

Project

planning

Design

Construction

• Regional tectonic regime (compres-sive, strike-slip, extensional)

• Orientation of major principalstresses

• Local disruptions due to Quater-nary geological processes (palaeo-relief, glaciations, erosion, etc.)

• Influence of relief on state of stress

• Estimate of σH/σV ratio• Orientation of major horizontal

stress σH

• Determination of complete tensorif possible

Validation of design assumptionsbased on exploratory works

Published data:• Stress maps• Mechanisms at regional earthquake

hypocentres*• Geological and topographic maps• Geotechnical reports on existing

structures• Palaeogeographic assumptions

Data from deep boreholes• Ovalisation of bore• Disking in cores

Borehole stress measurements• Overcoring and borehole slotter• Hydraulic fracturing and HTPF

Flat jack tests in adits

• Observation of tunnel walls

• Interpretation of strain data

5 Strictly speaking, coefficient K0, more commonly used in soil mechanics, is the effective stress ratio



If an exploratory adit is available, stressescan be measured at different points in theadit wall by the flat jack method. It is onlysuitable in rock having few or no joints thathas been substantially unaffected by exca-vation. The only assumption needed is thatrock response will be reversible, but calcu-lation of the complete tensor relies on amodel of the rock mass around the adit.Tests must be made over several straightsections of the adit.

4.4.4 – Classification of stressstatesIn the project planning stage, the designercan make his first assessment of the gene-ral stability of the unsupported opening onthe basis of the σc/σ0 ratio, with σc repre-senting the uniaxial compressive strengthof the intact rock matrix (the most readilyaccessible parameter at this stage) and σ0

being the value of the major stress in aplane normal to the tunnel.

On the basis of the σc/σ0 ratio, one can pro-ceed to the general qualification of the ini-tial state of stress with respect to its conse-quences on the planned structure.

Table 22 lists stress state classes based onthe σc/σ0 ratio.

It is absolutely essential to estimate thestresses by means of field tests when thestructure can be expected to lie in class CN2or CN3 conditions (assuming σ0 = γx z).

4.5 – TEMPERATURE

It is only necessary to predict natural tem-peratures in the rock mass for some under-ground structures which are temperaturecritical during their service life (mined sto-rage) or need special cooling systems to beprovided during and after construction(deep tunnels and storage facilities).

The estimate is frequently based on amean value of the local geothermal gra-dient Γ, which usually ranges from 2° to 4°per hundred metres depth, although it maytake other values in tectonic zones and vol-canic ground.

4.5.1 – Geothermal parametersThe most important thermal property of arock mass is its conductivity λ (in W/m.°C).Values for this parameter for samples of dif-ferent rocks are confined to the range 1.5to 5.5 W/m.°C (cf. Appendix 12). Becauseof the low porosity of rock, the presence ofwater and air has very little impact on the

thermal properties of the rock mass underpure conduction conditions (where there isno flowing fluid) and it is usually taken thatrock conductivity at depth is close to thevalue measured on samples.

In contrast, the effect of anisotropy mustbe considered because conductivity maybe up to twice as high in the directionparallel to the cleavage than in the perpen-dicular direction.

A rock mass is naturally supplied with heat

• from underneath, due to the geothermalflux Φ, which can be considered constant atthe scale of civil engineering construction.There is little precise local date on this point.It usually ranges from 50 to 80 mW/m2. Inpure conduction, we have

Φ = Γ x λ

• from the surface: heat from sunlightvaries with time of day and season but suchchanges do not affect rock temperaturebeyond about twenty metres’ depth; fortemperature modelling purposes, sunlightgoverns surface temperature (on whichabundant meteorological data is available).

Other very localised sources of heat maybe chemical reactions, radioactivity, hotwater from deep underground, or coldwater from the surface.

4.5.2 – Methods for estimatingtemperatures for undergroundstructuresAt the project planning stage, the designeris usually confined to a desk search forregional geothermal data (geothermal fluxmap of France, Mechler 1982; isogradientmaps, Gable 1980) and temperature datafrom oil wells and geothermal boreholes.Geothermal springs should be identifiedwhere they exist, along with nearby volca-nic ground (“hot” anomalies), and the pos-sibility of there being aquifers near theunderground structure site supplied fromsurface water (“cold” anomalies).

At the design stage, a more detailed searchis made for anomalies by systematicallymeasuring water temperature in explora-tory boreholes (temperature-conductivitylogs).

During construction, temperature measure-ments as tunnelling proceeds allow theparameters determined in the previousphase to be checked and validated. Anythermal anomalies detected in boreholesdrilled ahead of the face are excellent war-ning signs that the drive is nearing zones offlowing underground water and geologicaldiscontinuities.

With very large projects, temperaturemodelling with thermal models or suitablymodified hydraulic models (Goy 1996) ispossible after precise determination of sur-face temperature, structural geology, rockconductivities and local geothermal flux(pure conduction models are valid if thereis no significant water flow, Fabre 2001).

5 – USE OF ROCK MASSCHARACTERISATION FORUNDERGROUND STRUCTURESTABILITY ANALYSIS ANDCONSTRUCTION

5.1 – CHARACTERISTICVALUES AND PARAMETERSFOR DESIGN

5.1.1 – Individualisation of sub-unitsAs mentioned in para. 1.2.2, the first step isto divide the rock mass up into sub-unitsexhibiting substantial uniformity in theircharacterisation parameters. A sub-unitmay correspond to a section of the tunnel(for linear structures) or part of the under-ground structure (for non-linear structures).

CN 1 > 4 WEAKRock matrix satisfactorily strong but support may be neededbecause of jointing

CN 2 2 à 4 MODERATEFailure or plastic zones possible at tunnel walls

CN 3 < 2 STRONGRock matrix strength manifestly insufficient

CLASS σc/σ0 RATIO DESCRIPTION OF STRESS STATES

Table 22 – Preliminary classification of relative state of stress around a tunnel



Individualisation of a sub-unit generallystarts with the lithology and geologicalstructure as seen on the tentative geologi-cal cross section; however, the lithologymay admit of sub-divisions by reason ofvariations in

• the mechanical properties of the material(with the same petrographic signature),

• alteration and weathering,

• hydrogeology,

• rock cover,

• joint density, etc.

In practice, this operation of dividing theunderground structure up into uniformsub-units usually involves several iterationsbefore the resulting breakdown emergesas entirely logical and coherent. As it is notpossible to set boundaries between classesof values for the different parameters usedas the most pertinent criteria for individua-lisation, a trial and error process is gene-rally necessary.

5.1.2 – Geotechnical characterisationof sub-unitsConsider the measurements and tests for agiven sub-unit.

Once site investigations have been made,the raw data consists of measured valuesfrom tests performed by the geotechnicaloperator.

Depending on the geotechnical terms ofreference (cf. French standard P 94-500),the next step is for the geotechnical opera-tor and/or designer to proceed with a criti-cal analysis of these measured values. Thiscritical analysis eliminates any anomalousdata and validates the significant values tobe retained.

The significant values relate to a type oftest and a in-situ test:

• They may be 'local' values associatedwith a type of test or measurement at agiven sampling or in-situ test;

• They may on the contrary be 'global'values, associated with a type of test ormeasurement, on a wider geographic area(such as geophysical test data forexample).

It may be useful to make a sort of 'consoli-dation' of the reliability of the significantvalues by using simple but proven correla-tions such as E/Vp, σc/Vp, σc/γd, σc/σtb, etc.to detect any anomalies.

Lastly, a quick statistical analysis should berun if the number of data points available is

at least of the order of 10. This analysis,with number of data points, maximum andminimum values, mean and standarddeviation, makes it possible to assess scat-ter and the possible existence of severalhomogeneous populations, as may hap-pen with an anisotropic rock or a materialconsisting of thin alternating layers of diffe-rent lithologies, e.g. marl-limestone.

For each uniform sub-unit, these manipula-tions can be used to determine a 'characte-ristic' value for each parameter. The charac-teristic value of a parameter represents areasonably cautious value, not a maximumor minimum (cf. AFTES GT 29Recommendations on the use of generaldesign standards and rules for reinforcedand plain concrete tunnel linings).

For any one parameter, it may be useful toset one or more characteristic values accor-ding to the intended use. For example, alow characteristic value for the uniaxialcompressive strength would be needed forthe stability analysis and a higher value forthe drillability assessment.

5.1.2.1 – Rock matrix

The description for non quantitative para-meters (common name, petrography andmineralogy, alteration of the rock matrixmaterial) must be precise and, wherenecessary, supplemented with quantitativevalues (mineral contents, methylene bluetest value, etc.).

For most physical identification parameters(density ρ, ρd, ρs or volume weight γ, γd, γs,moisture content w, porosity n, wave velo-city Vp and continuity index IC) and forhardness DU and abrasiveness AIN or ABR,the arithmetic mean of the values obtainedis the starting choice for the characteristicvalue, provided that the statistical analysisconfirms that the population is normal.

The choice is not valid in the followingcases:

• Discovery of more than one population ofvalues (marl limestone, schists, etc.)

• Variation linked to the spatial distributionof samples

• Severe scatter militating for caution infavour of a lower characteristic value thanthe mean minus standard deviation.

With parameters for mechanical behaviour,a distinction must be made between thefollowing:

• Young's modulus E and Poisson's ratio nwhich admit of the same approach as forthe physical identification parameters,

• Swelling parameters σg and Cg, for whichthe pair of characteristic values must lie atthe upper end of the measured values,

• Uniaxial compressive strength σc, tensilestrength stb and Franklin index Is, for whichthe characteristic value may be

- either the 5% fractile when focusing onthe stability of the structure,

- or the 95% fractile when dealing withdrillability.

When dealing with stability, caution dic-tates giving consideration to the casewhere rock strength might locally be lower,whereas with drillability, attention focuseson the higher values of σc in order that tun-nel driving aspects are not underestima-ted. In all situations, a reasonably prudentchoice of characteristic value must be theoutcome of a thinking process that must beexplicit and substantiated on the basis forexample of a statistical analysis, site speci-fics, reference to completed structures insimilar settings, etc.

• Cohesion C and friction angle φ, for whichthe analysis preceding determination ofthe pair of characteristic values C and φmust include for all the circles determinedon the same material, with no distinctionbetween one test and another.

5.1.2.2 – Discontinuities

Before determining characteristic valuesqualifying discontinuities, the designermust address the following points:

• Joint density may itself be an individuali-sation criterion for the sub-unit, and onemust always ask whether it is better to takea mean value or consider two sub-units ins-tead of only one.

• Values found for the overall indexes(RQD, ID, FD) may be closely dependenton the orientation of the survey line.

When determining a characteristic value toquantify discontinuities, the arithmeticmean of overall index values (RQD, ID, FD)measured in the same direction is satisfac-tory provided scatter is reasonable. Themost appropriate survey line directionmust be chosen with reference to tunnelalignment and this is the direction alongwhich the characteristic value is determi-ned.

For parameters describing discontinuitiesin the same joint set, a distinction may bemade between

• strike OR, whose characteristic value canbe taken as the most frequently occurringdata point on the Schmidt density chart,



• joint spacing ES, whose characteristicvalue can be taken as the arithmetic meanin simple cases displaying little scatter; inother cases, the consequences of variablespacings must be examined in detail,

• joint persistence, whose value can beapproximated by observations on outcropsbut whose characteristic value can only bethe result of engineering judgement andexperience,

• joint wall roughness, waviness and wea-thering, joint width, infilling and water ifany, for which, once again, engineeringjudgement only can arrive at an averagecharacterisation for the whole joint set.

5.1.2.3 – Rock mass

The rock mass may be characterised eitherdirectly from the results of appropriate insitu tests, or indirectly with the aid of empi-rical classifications and correlations, relyingmainly on the characteristic values determi-ned from laboratory samples of the intactrock material as well as all other sources ofdata (geophysics, borehole tests and mea-surements, etc.).

Because of the time and cost involved, insitu tests are usually kept for a relativelylate stage in the design process once theproject layout has been more or less finali-sed. In the earlier stages, indirect methodsas discussed in para. 5.2 – GeotechnicalClassifications and 5.3 - Correlations aremostly used.

in situ tests proper in boreholes, shafts andadits aim primarily at determining rock massdeformability, in situ state of stress andhydrogeological conditions. Knowledge ofthe rock mass will be more or less extensiveand precise, depending on the resourcesassigned and the size of the project. Therecan be no doubt that an exploratory aditdriven over part or all of the alignment ofthe permanent structure will yield more,and more precise, data than many bore-holes and shafts. The designer will alsohave at his disposal several different waysof arriving directly or indirectly at any givenparameter and comparing the variousvalues obtained, assessing scatter andscale effect.

For example, the rock mass deformationmodulus can be approached in various ways:

• borehole dilatometer tests

• plate loading tests on adit or shaft walls

• dynamic moduli derived from wave velo-cities obtained by seismic methods

• analysis of displacements measured inthe exploratory adit.

Determination of the characteristic valuethen becomes a "reasoned" choice amongdifferent significant values from the varioustests and measurements. The choice musthowever take into account

• scale effect, illustrated by variations in thefield modulus with the rock volume consi-dered, and by the empirical rule

EL > ED > EP > EG

in which EL is the laboratory modulus, ED isthe dilatometer value, EP is the plate loa-ding test value, and EG is derived from aditwall displacements;

• strain range: compared to other modulusvalues and means of measuring them, thedynamic modulus is based on very smallstrains.

A mean permeability value can only be esti-mated in so far as local values exhibit littlescatter, to ensure there is a good probabilityof their belonging to the same sub-unit. Ifthis is the case, the mean is calculated onthe log K values (log normal distribution). Ifthere is significant scatter in the data orsignificant differences between two or moretest results, the designer should ask whe-ther it would not be more appropriate toconsider several units with their own speci-fic hydrogeological characters.

By the same logic, where several in situstress measurements are available, it mightbe preferable to select the stress state thatagrees best with drilling records (disking,ovalisation, wall failure).

Generally speaking, the approach to obtai-ning characteristic values is to find and ana-lyse the greatest number of cross-checksbetween data from different sources inorder to arrive at a considered judgementas to the correct characteristic value.

In the absence of any direct means ofdetermination, when significant valueshave not yet become available (as is oftenthe case with rock mass deformability andnearly always with rock mass limit strength),a possible approach is to refer to a similar,completed, structure and/or rely on classifi-cation systems, always provided that theproject falls within their category of validity.

5.2 – GEOTECHNICAL CLASSIFICA-TIONS

5.2.1 – GeneralVarious authors have proposed classifica-tion systems such as Protodiakonov 1909,

Terzaghi 1946, Lauffer 1958, Deere 1964,Wicham 1972, Bieniawski 1973, andBarton, Lien & Lude 1974. The Bieniawskiand Barton systems are by far the mostwidely used.

Geotechnical classification systems arebased on an empirical rock mass "qualityscore" drawn from values determined forcertain design-critical parameters. Theparameters involved vary slightly from onesystem to another but are basically

• rock matrix strength

• joint density

• mechanical behaviour of discontinuities

• hydrogeological conditions

• state of stress (partially).

The scoring process produces a final valueobtained through a simple calculation, whichalso differs from one system to another.

In 1978, at the time of first writing theseRecommendations on the description ofrock masses, AFTES adopted a restrictiveposition towards these systems, arguingthat quantifying rock mass quality bymeans of a single score was too reductio-nist and did not reflect the complexity ofthe real world.

Now these systems are widely used toderive, via the various correlations propo-sed, mechanical parameters for rockmasses (modulus, Hoek & Brown coeffi-cients, etc.) which can be used as designinput (see para. 5.3.1). It is neverthelessextremely important to remain scepticalabout the simplifying assumptions inherentin these systems and the choice of data onwhich they are based.

Using a classification system for any parti-cular project presupposes that the desi-gner has first assured himself that the pro-ject is truly compatible with the systemused (cf. para. 5.2.4).

Furthermore, a classification system mustnever be considered as a substitute for siteinvestigations or be an excuse for cuttingdown on efforts to arrive at the geotechni-cal characterisation of the rock mass. Noneof these classification systems are univer-sally applicable.

5.2.2 – Bieniawski's Rock MassRatingThe Rock Mass Rating (RMR) has beendeveloped by Bieniawski since 1973 toprovide a quantitative estimate of the pro-perties of the rock mass and support



necessary for stability. This approach wasinitially based on records from more than300 tunnels, most of them lying at mode-rate depth in sedimentary rock. The data-base was based primarily on South Africanexperience but has since been considera-bly enriched from many examples roundthe world. After the first version had beenwidely circulated in 1976, Bieniawski mademany changes to the parameters for esti-mating RMR. The current version, descri-bed here, is the RMR89 (Bieniawski 1989).

The RMR index is the sum of five scoresquantifying five characteristic rock massparameters and an adjustment factordependent on azimuth and dip of the dis-continuities. The RMR has been calculatedto span the range 0-100.

The five ratings A1 to A5 and rating adjust-ment B (cf. Appendix 13) are defined asfollows:

- A1: Strength of rock matrix: Range ofvalues 0 to 15 based on uniaxial compres-sive strength or point load strength Is.

- A2: Drill core quality: Range of values 3 to20 for rock core quality, from RQD.

- A3: Spacing of discontinuities: Range ofvalues 5 to 20 (lowest ratings for each jointset).

- A4: Condition of discontinuities: Range ofvalues 0 to 30 (joint persistence, width(separation), roughness, infill (gouge) andwall rock weathering).

- A5: Groundwater: Range of values 0 to 15(inflow rate and/or pressure).

- B: Adjustment for joint orientation: Rangeof values –12 to 0, for strike and dip of dis-continuities with respect to tunnel align-ment.

The basic Rock Mass Rating (RMRbasic) cha-racterising the rock mass is simply the sumof ratings A1 to A5 (B = 0). In undergroundengineering work, the standard RMR (orRMR89) is written as

RMR89 = A1 + A2 + A3 + A4+ A5 + B

Basically therefore, RMR is a rating assi-gned to the rock mass ranging from 0 to100, more than 70% depending on discon-tinuities and only 15% on rock matrix pro-perties and 15% on hydrogeology. Therating completely ignores the state ofstress in the rock mass at the tunnel site.

This should theoretically limit the use of theRMR only to strong rock whose response isgoverned by the discontinuities. This

would automatically exclude rock strengthclasses RC6 to RC7 and stress class CN3.

The rating system produces five rock massclasses (Appendix 13) and five correspon-ding support classes, and this is nowadaysinadequate to cover the variety and pro-gress encountered in excavation and sup-port techniques.

5.2.3 – Barton's Q indexThe Q index is the central parameter in asystem developed by the NorwegianTechnical Institute in 1974 based on datafrom more than 200 completed tunnels,mostly situated in the crystallineScandinavian Shield with high horizontalstresses (Barton et al. 1974). The systemwas revised in 1993 to include data frommore than 1000 tunnel case histories(Grimstad & Barton 1993).

The Q system method provides a quantita-tive estimate of support needed for tunnelstability on the basis of the following infor-mation:

• Largest dimension (diameter) of the plan-ned opening

• Planned use of the completed structure(implicitly, acceptable level of risk)

• Rock mass Q index.

The Q index is a total score from 0.001 to1000 (this is the theoretical range, reducedin most practical cases to 0.005-50), calcu-lated from (Appendix 14)

• Rock Quality Designation (Deere 1964)

• Joint set number Jn

• Joint roughness number Jr (concerns themost unfavourable discontinuities)

• Joint alteration number Ja (concerns themost weathered discontinuities and infillmaterial)

• Joint water reduction factor Jw (flow rateand pressure)

• Stress reduction factor SRF.

The Barton Q index is written

Q = (RQD/Jn) x (Jr/Ja) x (Jw/SRF)

In other words, the Q index is the productof three factors for• the potential size of rock blocks• the geomechanical quality of the contactsurfaces between blocks• the initial state of the rock mass as regardswater and stresses (Barton's "active stress").

Calculating the range of variation of Q, firstwith the most unfavourable values, thenwith the most favourable values, may pro-duce very large differences if the calcula-tions are done for sub-units displaying verydifferent characteristics.

Table 23 recapitulates the ranges of varia-tion of the different parameters to assesstheir relative weight in the final Q indexvalue.

The weight of the SRF factor in the thirdterm Jw/SRF is particularly high, which isthe unique feature of the Q index, whichrefers to :• the possibility of sheared, brecciated orvery clayey zones;• the level of stress in brittle rocks;• potential creep and swelling stresses indeformable rocks.

The Q index is thus strongly dependent onnon-intrinsic rock properties, especially thestate of stress in the rock mass. The formu-lation of the Q index does however havethe drawback of not directly reflecting thecharacteristic parameter of the mechanicalstrength of the rock material.

5.2.4 – Summary and precautionsThe growing popularity of classificationsystems in France is probably due to :• their apparent simplicity of use;• their very widespread use outside France,especially by French engineers workingabroad;

RQDJn

JrJa

JwSRF

1020

0,520

0,0520 (6)

1000,5

40,75

10,5

1040

827

2040

PARAMETERS MOST UNFAVOURABLECONDITIONS

MOST FAVOURABLECONDITIONS

RANGE(highest ratio)

Table 23 – Ranges of variation of parameters used in calculating the Barton Q index



• the convenience of using a rating systemfor making comparisons between designpredictions and actual conditions encoun-tered during construction at different sites;

• the possibility of amending scores in thelight of conditions encountered duringconstruction;

• the possibility of using correlations tofind the quantitative data need for designanalyses.

In underground engineering, the ultimatepurpose of these classification systems isto design tunnel support; this approachhas been tried and found satisfactory inmuch drill and blast tunnelling. But thesesystems are not always suitable with otherexcavation methods (road headers, tunnelboring machines).

Generally speaking, the RMR and Q sys-tems are unsuitable for soft rock (R6 to R7).Table 24 summarises the features and limi-tations of these two systems.

In addition to the general and specific limi-tations discussed above and in Table 24, itmust also be stressed that classification

systems must be used with the followingprecautions:

• Do not use only one system.

• Explain in detail how the scores were cal-culated; most importantly, identify the jointsets considered at each step.

• Examine the sensitivity of the RMR or Qindex to changes in the values of the para-meters and present results as envelopevalues for the final rating.

• Do not use the ratings as a "rule-of-thumb recipe," but be critical and vigilantas to the proper field of application.

• Remember that classification systems areempirical and reflect certain tunnelling andsupport practices, and these practices maychange.

5.3 – CORRELATIONS

Warning: It must never be forgotten that trea-ting a jointed rock mass as a continuum mate-rial is in itself a considerable simplification.Secondly, highly anisotropic rock masses dis-play special behaviour not covered by the

usual classification systems. Correlations "incascade" must never be used.

5.3.1 – GeneralIn view of the difficulties of making directtests of deformability and (even more so)limit strength at rock mass scale, manyauthors have sought to start from actualcase histories to establish empirical rela-tionships linking these parameters to rockmatrix characteristics and rock mass join-ting.

These relationships have been establishedfor particular contexts and must be usedwith great caution; they must always, as faras possible, by set side-by-side with in situfield test results.

5.3.2 – Estimating rock massdeformabilityThe rock mass deformation modulus EMas isone of the critical parameters for modellingstresses and strains around an under-ground opening. There are several meansof measuring this parameter at a volumescale of up to a few cubic metres (see para.4.2.1.2) and estimating it at a larger scale(but for very small load values) by seismictests (para. 4.2.1.1). As already stated inpara. 5.1.2.3, the scale effect is very impor-tant here.

Many schemes have been proposed forindirectly estimating the rock mass defor-mation modulus. The more important onesare tabulated in Appendix 15, along withauthor references.

These schemes may directly combine para-meters for the rock matrix (E, σc) and rockmass (RQD) or be derived indirectly via theRMR and/or Q index.

Figure 14 shows a few examples of theempirical relationship between E moduliand RMR and Q index.

5.3.3 – Hoek's GSI indexThe Geological Strength Index is notdirectly a classification system, it is an inter-mediate step to determining the mechani-cal properties of a rock mass, using theempirical formulae proposed by Hoek &Brown (see below).

GSI was introduced in 1995 by Hoek(Strength of Rock and Rock Masses, ISRMNews Journal, 1994, vol. 2). It derives fromvariants of the RMR and/or Q index, desi-gnated RMR' and Q' respectively.

Overall characterisa-tion of rock mass

• Jointing patterns well descri-bed except for anisotropic rock(schist, slate, etc.)

• Mechanical properties of dis-continuities well described

• Natural stresses described

Assessment ofmechanical characte-ristics at scale ofwhole rock mass

• Empirical correlations bet-ween RMR and deformabilityand strength parameters

• Empirical correlations betweenQ and physical and mechanicalparameters (P longitudinal wavevelocities, deformability)

Use for project

• Allowance for orientation ofdiscontinuities with respect toaxis structure

• Quick means of setting lengthof pull

• Stand-up time (conservativeapproach)

• No use in deciding excavationmethod

PARAMETERS RMR Q system

Table 24 – Comparison between RMR and Q system in underground engineering applications

• Must be used with great caution, especially for strength parameters: avoid correlations 'in cascade' of the type Q ⇒ RMR ⇒ (m, s) ⇒ (C, ϕ)

• Not relevant to orientation ofdiscontinuities with respect tocentreline

• Quick means of stipulatingsupport needed at roof, side-walls and intersections but givesfalse impression of accuracy insetting bolt lengths

• Use in design stage and formonitoring tunnel driving

• Allows for changes in supporttechniques

6

6 Much higher values, up to 400, have been suggested by Barton for very deep underground openings where there is a risk of sudden violent decompression



RMR' is calculated like the basic RMR butusing only the first four criteria (Strength,RQD, Spacing of Discontinuities andCondition of Discontinuities), systemati-cally taking the fifth groundwater value as15 (it is rock behaviour under "completelydry" conditions that is considered) and therating adjustment for joint orientation as 0.

By a similar process, Q' ignores the thirdratio for the fifth and sixth parameters(water and 'active stress').

Q' = (RQD/Jn) x (Jr/Ja)

GSI is determined from RMR'89 values asfollows:

• for RMR'89 > 23

GSI = RMR'89 – 5

• for RMR'89 < 23

GSI = 9(Log Q' + 44).

5.3.4 – Estimating rock mass limitstrengthEstimating rock mass limit strength atunderground structurescale calls for finejudgement. As mentioned above (para.4.2.2) no in situ test – except a full size testto failure, which is unfeasible for obviousreasons – is capable of yielding useableresults. The only possible approach is todownscale, on empirical evidence, the pro-perties of the intact rock matrix with refe-rence to rock mass jointing. The most signi-ficant research in this area is due to Hoek

and Brown (brought together in Hoek,Kaiser & Bawden 1977) who suggestextending the parabolic failure criterionproposed for the rock matrix (para. 2.2.5.2)to the rock mass, suitably modified to thefollowing generalised form:

σ1 = σ3 + σci [mb . σ3/σci + s]a (1)

in which a, s and mb are characteristicconstants for the rock mass.

Except in highly weathered rock with prac-tically no remaining cohesion, the valuegenerally adopted for a is

a = 1/2 (parabolic criterion).

Values for the constants can be derivedfrom equations containing Hoek's GSI (seepara. 5.3.3).

mb = mie[ ( GSI x 100)/28]

For GSI > 25

s = e[(GSI x 100)/9]

a = 0.5

For GSI < 25

s = 0

a = 0.65 x (GSI/200)

The relationship between mb and mi, therock matrix characteristic constant close tothe brittleness index FR (see para. 2.2.5.2)is very important for calculating numericalvalues in equation (1). From Hoek &Brown's compilation, the ratio mb/mi mayrange from low values (<0.1) for jointed

low-friction rock masses to 0.4 to 0.6 forhard rock containing few, very rough-wal-led joints.

There are also formulae proposed byvarious authors (Hoek 1990) to derive fric-tion angle and cohesion values from coeffi-cients s and mb. They are obtained bylinearising the parabolic criterion andreplacing it by a tangent or secant over aset stress range. Despite the attractionsthis may appear to offer, making it possibleto extend standard soil mechanics calcula-tions based on the linear Mohr-Coulombcriterion to rock masses7 , it must be stres-sed that this criterion is generally irrelevantto the characterisation of rock mass beha-viour. It may produce quite suspect resultsand must always be used with great cau-tion and scepticism.

Lastly, this approach must never be usedwhen dealing with rock masses whose dis-continuities are strongly polarised (thinlybedded rock, schist, slate, etc.): assigningan isotropic failure criterion derived fromthe RMR to a situation where actual beha-viour is strongly anisotropic can only leadto results that have little relationship withreality.

Photo 9 – Anisotropic heterogeneous rock mass –Mequinenza region, Spain

7 The success that this has had with many sometimes unsuspecting users is due to the ease it offers them when having deal with rock mass problems. Butsuch popularity cannot in any way be advanced as an argument for the scientific merits of the approach.

Anisotropic heterogeneous rock mass – Mequinenza region, Spain

Figure 14 – Estimating rock mass deformation modulus EMas from RMR and Q values (Hoek, Kaiser & Bawden 1997)



5.4 – PRESENTATION OF ROCKMASS CHARACTERISATIONDATA

5.4.1 – Basics and generalremarksThe variety of geologies and geotechnicalconditions which may be encountered inunderground engineering and the specificfeatures of each project make it difficult toenvisage any single format for presentingsynthetic data suitable for all possiblecases. The general presentational schemesuggested here will have to be amendedas required for each individual project.

Data presentation may differ at differentstages of project development (projectplanning, preliminary design, final design,etc.) and according to the quantity andreliability of the data available.

When summarising data characterising therock mass, one must bear in mind the ulti-mate destination of the data (generaldesign, construction method, design ana-lyses, etc.) and the fact that some (or all)items may have contractual relevance,depending on the stage reached in theproject implementation process: the formand content of the summary presentationmust consider this point.

5.4.2 – Example of data presenta-tion in tabular formIn order to remain consistent with the hie-rarchy of the classes for the various parame-ters examined in these Recommendations,the homogeneous sub-units determinedfrom the field investigations and studiesshould be ordered into geotechnicalclasses from E1, representing the globallyhighest geotechnical classes, to EN, repre-senting the poorest geotechnical proper-ties (see figure 15). It must be stated whe-ther the values for these parameters aresignificant values or characteristic values (cf.para. 5.1.2 above).

Naturally, this is a recommendation whichone should try to follow, sometimes with alittle difficulty in that certain parametersexamined may act in contrary directions,making it more problematical to arrive at aperfect hierarchy.

The parameters listed in the example sum-mary Table 25 must be considered as the

'basic minimum' to be collected for anyunderground project. Other parametersmay be added as required for any specificproject (e.g. swelling pressure σg and swel-ling index Cg for potentially swellingrocks, etc.).

For each homogeneous sub-unit, the sum-mary table may include the AFTES classesfor rock matrix, discontinuities and rockmass parameters. One should not howeverbe tempted to have so many items as tomake what is a 'summary' table difficult toread, because the point of a summary is tobe clear and easily understood.

NOTE. Figures shown in the E4 column ofTable 25 and details in the Orientation linefor discontinuities are not taken from anactual project and are shown simply forillustrative purposes.

5.4.3 – Synoptic presentation ofrock mass characterisation dataand cross-referencing to geologicalprofile

5.4.3.1 – Longitudinal profile

As an aid for using the parameters charac-terising the homogeneous geotechnicalsub-units for the various uses to be madeof the data (project design, constructionmethods, etc.), the summary table isusually shown alongside the geologicalprofile which, for any given project, is the

basic document bridging the gap betweengeologists and geotechnical engineers,and civil engineers. The most commonmethod is to have a single drawing with thegeological profile at the top and the break-down into sub-units below, along with thesummary table. The geological profileshould be drawn to true scale so as not togive an inaccurate picture of the structureswith distorted contacts and dips.

5.4.3.2 – Cross sections and horizontalsection at project depth

The complexity of the geological and geo-technical context may make it extremelyuseful or essential to accompany the geo-logical longitudinal profile in some placeswith cross sections. This is particularly thecase when structural details of significantimportance for the project are strongly ske-wed in relation to the tunnel. Again, undersome conditions when the geological lon-gitudinal profile alone is unable to give atrue picture, a horizontal geological profileat project depth should be added for abetter understanding of the geologicalcontext and potential risks: this situationarises for example when the tunnel align-ment is expected to draw near to, withoutnecessarily intercepting, a major structuralfeatures (fault, unconformable contact,permeability interface, etc.) running nearand parallel to the tunnel.

Figure 15Principle of segmentationinto homogeneous geo-technical sub-units anddata hierarchy



Tabl

eau

25 :T

able

au d

e sy

nthè

se d

es p

aram

ètre

s de

car

acté

risat

ion

des

sous

-ens

embl

es g

éote

chni

ques

hom

ogèn

es



APPENDIXAPPENDIX

PagesPages

CONTENTSCONTENTS

APPENDIX 1RECOMMENDED ROCK DESIGNATIONS AND PRINCIPAL GROUPS - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 34

APPENDIX 2DENSITY AND THEORICAL P WAVE VELOCITIES VP IN MINERALS - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 34

APPENDIX 3ORDER OF MAGNITUDE THEORETICAL P WAVE VELOCITIES VP* INSOME ROCKS ASSUMED TO BE SOUND AND NON POROUS- - - - - - - - - - - - - - - - - - - - - - - 35

APPENDIX 4SWELLING TESTS - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 354.1 – Huder-Amberg test - - - - - - - - - - - - - - - - - - - - - - - - - - 35

4.1.1 – Test procedure - - - - - - - - - - - - - - - - - - - - - - - - - - 354.1.2 – Interpretation- - - - - - - - - - - - - - - - - - - - - - - - - - - 36

4.2 – ISRM tests - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 364.2.1 – Axial swelling pressure at constant volume - - - - - 364.2.2 – Axial swelling pressure versus axial strain - - - - - - 36

APPENDIX 5MEANING OF PARAMETERS IN HOEK & BROWN - - - - - - - 37

APPENDIX 6CHART DEFINING DRILLING RATE INDEX AND CHARACTERISTICS OF S 20 and sj tests- - - - - - - - - - - - - - - 376.1 – Laboratory tests- - - - - - - - - - - - - - - - - - - - - - - - - - - - - 376.2 – Interpretation - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 38

APPENDIX 7TEREOGRAPHIC PROJECTION - - - - - - - - - - - - - - - - - - - - - 38

APPENDIX 8set persistence and 3 D GEOMETRICAL MODELING OF FRACTURE NETWORKS - - - - - - - - - - - - - - - - - - - - - - - - - - 39

APPENDIX 9SCHMIDT HAMMER - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 40

APPENDIX 10HYDRAULIC PARAMETERS - - - - - - - - - - - - - - - - - - - - - - - - 41

APPENDIX 11RIGID PLATE LOADING - - - - - - - - - - - - - - - - - - - - - - - - - - - 41

APPENDIX 12THERMAL ROCK PARAMETERS- - - - - - - - - - - - - - - - - - - - - 42

APPENDIX 13 BIENIAWSKI’S RMR CLASSIFICATION OF ROCK MASSES - 43

APPENDIX 14BARTON’S Q INDEX - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 45

APPENDIX 15empirical formulae for evaluating rock mass deformation moduli - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 47

REFERENCES - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 4716.1 – General - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 4716.2 – Rock matrix characterisation parameters- - - - - - - - - - 4716.3 –Characterisation parameters of discontinuities - - - - - - 4716.4 –Set persistence of discontinuities and 3D geometrical modelling of networks of discontinuities(Appendix 8) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 4816.5 – Schmidt hammer test (Appendix 9)- - - - - - - - - - - - - - 4816.6 – Hydraulic parameters (Appendix 10) - - - - - - - - - - - - - 4816.7 – Rock mass characterisation parameters- - - - - - - - - - - 48

16.7.1 – Indirect measurements - - - - - - - - - - - - - - - - - - - 4816.7.2 – Rigid plate loading test- - - - - - - - - - - - - - - - - - - 4816.7.3 – Stresses - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 4816.7.4 – Thermal rock parameters (appendix 12) - - - - - - 4916.7.5 – Hydrogeological conditions - - - - - - - - - - - - - - - 49

16.8 – Use of rock mass characterisation for design and construction of underground structures - - - - - - - - - - - - - - - 4916.9 – AFNOR standards - - - - - - - - - - - - - - - - - - - - - - - - - - 49


Caracterisation of rock masses useful for the design and the construction of underground structuresA

PPE

ND

IX

APPENDIX 1

RECOMMENDED ROCK DESIGNATIONS AND PRINCIPAL GROUPS

IGNEOUS ROCKS

METAMORPHIC ROCKS

SEDIMENTARY ROCKS

Granites

Diorites

Basalts & Gabbros

Massive metamorphic rocks

Schistose metamorphic rocks

Carbonate rocks

Detrital rocks

Saline rocks

Carbonaceous rocks

Granite, granodiorite, syenite, microgranite, rhyolite, rhyodacite, tra-chyte, etc.

Diorite, quartzitic diorite, microdiorite, andesite, dacite, trachyandesite,etc.

Gabbro, dolerite, peridotite, serpentine, basalt, pozzolana, etc.

Gneiss, amphibolite, cornelian, quatzite, marble, leptynite, etc

Schist, micaschist, slate, calcschist, etc.

Limestone, chalk, dolomite, cargneule, travertine, marl, etc.

Sandstone, arkose, claystone, pelite, conglomerate, etc.

Rock salt, gypsum, anhydrite, potash, etc

Coal, lignite, etc.

Names in italics are extrusive or volcanic equivalents

APPENDIX 2

MASSE VOLUMIQUE ET VITESSE THEORIQUE VP DES ONDES P DANS LES MINERAUX

Minerals Density ρs (g/cm3) Vp (m/s)*

Amphiboles 2.98 - 3.20 7 200

Augite 3.2 - 3.4 7 200

Biotite 2.90 5 130

Calcite 2.71 6 660

Dolomite 2.87 7 900

Magnetite 5.17 - 5.18 7 410

Muscovite 2.83 5 810

Oligoclase 2.64 - 2.67 6 260

Olivine 3.25 - 3.40 8 400

Orthose 2.57 5 690

Quartz 2.65 6 050

*from French standard P 18-556



PPE

ND

IX

TUNNELS ET OUVRAGES SOUTERRAINS - N° 177 - MAI/JUIN 2003

APPENDIX 3

ORDER OF MAGNITUDE THEORETICAL P WAVE VELOCITIES VP* INSOME ROCKS ASSUMED TO BE SOUND AND NON POROUS

Rock Vp* (m/s)

Granites and rhyolites 6 000

Diorites 6 500

Gneiss 6 000

Amphibolites 6 500

Calcaires 6 500

Silica rocks 6 000* A utiliser au besoin dans la détermination de l’indice de continuité Ic

APPENDIX 4

SWELLING TESTS

4.1 – Huder-Amberg test

4.1.1 – Test procedure

A – First step: Reconfinement of sample

The sample of height h is accurately dres-sed to fit snugly in the test cell. It is placedin the oedometer between porous stones.The top stone is in contact with the pistonapplying a pressure opposing all (or partof) the increase in sample height ∆h. Thetest starts with only the weight of the pis-ton exerting a very low stress σm of theorder of 0.025 MPa, considered as the ori-gin on the semi-log paper on which thetest is plotted.

Any imperfections (decompression, micro-cracking) from the sampling or sample pre-paration process are corrected for by:

• applying load (a) to produce a stress σD

equivalent to the in situ stress;

• then decreasing this load (b) to stress σm;

• increasing the load (c) to σD (if the samplewas unaffected by sampling, the points forthe two σD load conditions coincide).

B - Second step:Wetting

The sample is them brought into contactwith water through the bottom porousstone, with the top porous stone allowingair in the sample to escape.

Wetting causes the material to expand,causing by an increase in sample height ∆hunder stress σD. The change in height ∆hD isrecorded over time until there is no further

change and ∆hD remainsconstant.

On figure 1, this changein sample thickness at

constant pressureσD appears as the

straight seg-ment D’-D,

with point

D representing the state of the sampleafter stabilisation of volume expansion.

C – Third step: Decremental unloading

From point D, the load on the sample isinstantaneously reduced from pressure σD

to σE; the immediate state of the samplecan now be represented by point E’. Theeffect of load reduction is to allow swellingto proceed in the form of a further volumeexpansion causing a change in sampleheight ∆hE. As before, the change is recor-ded to stabilisation, at which point, thestate of the sample can be represented bypoint E, the change in height of the sampledue to unloading from stress σD to stress σE

is represented by straight line E’ – E.

This is repeated several times, reducing thestress s each time and, each time, waiting

for stabilisation of the height increase ∆h.

Note. For consistency with ISRMrecommendations (see para. 1.4), the

unloading should follow a geome-trical function:

σi+1 = 0.5 σi = 0.25 σi-1

and limit load decre-ments to a recommen-ded minimum value of25 kPa.

Appendix 4 – Figure 1 – Huder-Amberg swelling test



PPE

ND

IX

4.1.2 – InterpretationThe test is interpreted in a semi-log plot inwhich the sample volume increase ∆h isrepresented by the height change of thesample in the oedometer expressed as apercentage of the initial height:

(∆h/h) x 100

The changes in pressure on the sample arerepresented by the axial pressure on thepiston, expressed in log σi.

A – Huder-Amberg law

Experience shows that, in this semi-logplot, points D, E, F, etc. representing thestate of the sample after stabilisation of theheight increase at each stage of loadreduction plot on a straight line. From this,a very simple relationship can be foundbetween the stress change ∆σ and samplethickness change ∆h.

The thickness change ∆h when unloadingthe sample from stress σi to stress σJ can bewritten

∆h/h = Cg x log (σj/σJ)

in which the swelling index Cg is a constantgoverned by the intrinsic nature of thematerial.

B – Huder-Amberg meaning of ‘swellingpressure’

The swelling pressure as understood byHuder & Amberg is defined as the stress σG

beyond which wetting ceases to cause fur-ther sample thickness change. This value isgiven by the intersection of the Huder-Amberg straight line S with the extensionof the reloading curve c.

Note. The precision of this determination isimproved when stress σD on first wetting(point D’) is very close to σG (this reduces theapproximation of the extrapolation).

4.2 – ISRM tests(see References, Appendix 16.2)

4.2.1 – Axial swelling pressure atconstant volume

The sample of height h is inserted in theoedometer as in the Huder-Amberg test.After wetting, the axial pressure on thesample is controlled to oppose any heightchange ∆h in order to keep the samplevolume constant. The test is continueduntil reaching the maximum pressure nee-ded to achieve this.

4.2.2 – Axial swelling pressureversus axial strainThe test procedure is exactly the same asfor the Huder-Amberg test except for theway in which the test data is plotted.

Test results are plotted as a curve of per-cent thickness change strictly due to swel-ling vs applied axial pressure.

‘Thickness change strictly due to swelling’means the total change resulting from thechange in pressure minus pseudo-elasticstrain corresponding to the same loadreduction.

APPENDIX 5

MEANING OF PARAMETERS IN HOEK & BROWN

Hoek & Brown (1980) wrote the intact rock(rock matrix) parabolic failure criterion thus:

σ1 = σ3 + σci (mi x σ3/σci + 1)1/2

and introduced constant mi, thereaftercommonly known as the Hoek & Browncoefficient.

The standard form of the parabolic crite-rion giving components σn and τ of thestress on the failure face versus σci and σti

(rock matrix compressive and tensilestrengths) is

τ = σti [(1 + σci/σti)1/2 - 1] x (1 + σn/σti)1/2

From this, the relationship between theHoek & Brown coefficient mi, and σci and σti is

σti = σci/2 x [mi – (mi2 + 4)1/2 ]

This is equivalent to

FR1 = σci/σti = 2/[mi – (mi2 + 4)1/2 ]

= 1/2 x [mi + (mi2 + 4)1/2 ]

From the compilation of mi values (Hoek,Kaiser & Bawden 1995), it is found that thiscoefficient ranges from 4 (for some clayeyrocks) to more than 30 (igneous rocks andsome metamorphic rocks). In practice the-refore, the following equation can be used:

mi = FR

This has the advantage of giving physicalmeaning to coefficient mi and allows it tobe estimated from standard laboratorytests. Uniaxial compression and Braziliantest results readily lend themselves to sta-tistical analysis. The observation in para.2.2 in the main text on anisotropyobviously applies.

Finally, the parabolic criterion can be writ-ten as

σ1 - σ3 = σci (FR x σ3/σci + 1)1/2

1 FR = brittleness index (see para. 2.2.4.3.



PPE

ND

IXThe Drilling Rate Index (DRI) was originallydeveloped in the sixties at the NorwegianInstitute of Technology at Trondheim toassess the drill hammer . Since the eighties,it has mainly been used to predict the per-formance of rock tunnelling boringmachines ( TBM )(see References,Appendix 16.2).

6.1 – Laboratory tests

DRI is calculated from the results of twoNorwegian standard laboratory tests:

• The S20 brittleness (fragmentation) test(figure 1)

This test estimates the resistance of therock to crushing under repeated blows, asin the French dynamic fragmentation test.

• The Siever J-value (SJ) penetration test(figure 2)

This test estimates rock resistance to pene-tration, as in the Cerchar-Ineris hardnesstest.

6.2 – Interpretation

DRI is obtained from S20 and SJ using thechart shown in figure 3. It ranges from 20to 90. A high DRI value indicates easierpenetration of the TBM cutters.

APPENDIX 6

CHART DEFINING DRILLING RATE INDEX AND CHARACTERISTICS OF S 20 AND SJ TESTS

Appendix 6 – Figure 1 – Brittleness test Appendix 6 – Figure 2 – Siever test apparatus

Appendix 6 – Figure 3Chart for calculating Drilling Rate Index

BRITTLENESS VALUE S20

DR

ILLIN

G R

AT

E I

ND

EX

DR

I Example:

Rock = gneiss of South Africa

S20 = 45

SJ = 6.0

From index A4.3, DRI = 43

More DRI value is ligh,more the penetration of TBM is high



PPE

ND

IX

The stereographic projection is the mostwidely-used method of plotting disconti-nuities. Various techniques are availablebut two only are in routine use:

• Equal angle projection to study relation-ships between discontinuities (Wülff net)

• Equal angle projection to measure spa-tial distributions (Schmidt net).

Each joint plane is plotted as the projectionof its pole (intersection of the normal to theplane with the upper or lower hemisphere

of the reference sphere) or the projectionof its trace (intersection of the plane withthe upper or lower hemisphere of thesphere).

Two types of plots are routinely used toanalyse the pattern of discontinuities intosets:

• The cluster of azimuths of the dip vector:this consists of grouping observations inangular sectors of the dip vector azimuth,with the absolute or relative number of

observations represented on the radialaxis. This type of plot ignores dip anglesand is only meaningful for finding directio-nal sets of discontinuities with similar dipangles.

• Density stereograms on the Schmidt net:densities are obtained by counting thenumber of poles within the target 1% ofthe diagram area. The count is done on acounting net: count by circles centred on aregular grid or Dimitrijevic count byellipses centred on an irregular grid. Thetrace of the joint number or densitycontours bounds the zones of pole concen-trations and may identify main joint sets.

APPENDIX 7

TEREOGRAPHIC PROJECTION

Représentation d’un plan (αp,β) à l’aide de la projection stéréographique (hémisphère supérieur)

Canevas de Wülff et de Schmidt

Appendix 7 – Figure 1 – Cyclographic plot and polar plot of a plane of a discontinuty by stereographic projection

Rose diagram of azimuth dip vector (example)

Density stereogram on Schmidt net (lower hemisphere)with isovalues curves

Appendix 7 – Figure 2 – Analysis of patterns of discontinuities (examples)



PPE

ND

IXThe persistence of discontinuities can onlybe estimated from measurements on expo-sed surfaces (outcrops, adit walls) and mustbe analysed with caution.

Persistence is only accessible through two-dimensional data – trace lengths – whosemeasurement is affected by multiple geo-metrical bias factors. The first is that thesurvey line preferentially intercepts the lon-ger discontinuities, the second is that thelonger discontinuities extend beyond thesurvey surface, introducing a truncation inthe measurements, the last is that somelimiting value is generally imposed on shortdiscontinuities, which also has a truncatingeffect. Data must be corrected for any rigo-rous estimate of this parameter (Priest &Hudson 1981, Pahl 1981).

Over the last thirty years, many authors

have developed three-dimensional geo-metrical models of networks of discontinui-ties. Among the more recent ones, the ran-dom disk model is one of the most widelyused (Baecher 1977, Long 1985, Cacas1989, Xu 1992). For a given joint set, eachdiscontinuity is represented by a disk ofzero thickness, defined by the position ofits centre, its radius and orientation, eachparameter being drawn stochastically fromits own distribution law, characteristic ofthe set.

Estimating disk radii, i.e. the true extensionof the discontinuities, raises problems. Inaddition, passing from 2D continuity to 3Dpersistence is not simple and requireshypotheses on

• The geometrical shape of the disconti-nuities: in the disk model, it is possible to

write mathematical relationships betweenthe 2D distribution of trace lengths and 3Ddistribution of disk radii (Warburton 1980).

• The persistence distribution law: in thisway, the parameters in this law can bedetermined on the basis of trace lengthdistribution parameters.

From this, one can obtain a 3D model ofthe network of discontinuities and study itsconnectivity to analyse the fluid flowthrough the network or the mechanicalbehaviour of the assembly of rock blockscreated by the fractures, from fracturebehaviour and, sometimes, the rock matrixbehaviour (figure 1).

More complex models also introduce aranking of the joint sets (Heliot 1988) or usegeostatics (Billaux 1990).

APPENDIX 8

SET PERSISTENCE AND 3 D GEOMETRICAL MODELING OF FRACTURE NETWORKS

Appendix 8 – Figure 1 – Example of geometricalmodelling of network of discontinuities with SIMBLOC program (ENSP-CGI, after Xu 1991)

MODELLEDDISCONTINUITIES

CONNECTEDDISCONTINUITIES



PPE

ND

IX

The Schmidt hammer is an apparatus for mea-suring the rebound as a mass strikes a surface,powered by a compression spring, with knownenergy. Rebound is measured as an index from0 to 100.

Various types of hammer are available with dif-ferent impact energies. In rock mechanics, themost widely-used model is the Type L hammer,for which correlations between the reboundindex and uniaxial compressive strength havebeen established by Miller in 1965 (seeReferences, Appendix 16.5) (figure 1).

The measurement of the uniaxial compressivestrength of a joint wall is governed by a numberof procedural parameters (Barton & Choubey1977):

• Hammer orientation: rebound is minimumwhen the apparatus is held vertically down-wards, and maximum when it is held verticallyupwards. Readings must be corrected when theapparatus is used in other directions. In allcases, it must be held perpendicular to the testsurface.

• Sample size: the sample must be largeenough for the impact not to cause it to move.Small samples must always be fixed on a rigidbase.

• Number of measurements: experiencesshows that at least ten readings must be madeat different points on a representative sample orper square metre area. The rebound value to beused is the average of the five highest readings,the lowest readings being considered as unre-presentative because the samples are assumedto have moved or the grains to have crushed.

APPENDIX 9

SCHMIDT HAMMER

Appendix 9 – Figure 1 – Relationship between Schmidt hammer rebound and uniaxial compressive strength (after Deere & Miller)



PPE

ND

IX

APPENDIX 11

RIGID PLATE LOADING

The test consists of applying load to therock by jacking a rigid plate against it. Thetest is performed in an adit (2-2.5m section)with the reaction being provided by theopposite rock face, allowing two measure-ments to be made simultaneously. It isusually performed either horizontally orvertically, although other set-ups may bejustified in strongly anisotropic rock.

The test apparatus consists of three parts(figure 1):

∑ A 2000-3000 kN hydraulic cylinder is ali-gned with a stanchion of variable lengthbetween two rigid plates. They must besmall (280mm diameter) in order to apply asufficiently high force to neutralise themost decompressed zone at the adit wall,

affected by the excavation process. A balljoint makes up for any misalignment andlack of parallelism between the two faces.

∑ A two-speed pump and two 1% classpressure gauges.

∑ A rigid reference frame fixed beyond thezone of influence, carrying displacementgauges (C4, C5, etc.) to measure the dis-placement of the plate and surroundingarea. The displacement gauges may bedial gauges or have an analogue or digitaloutput, accurate to one-hundredth of amillimetre.

One or two extensometers may be instal-led in small boreholes under the rigid bea-ring plate to identify decompressed sur-face zones.

The plate must bear on a flat rock surface,free from any loose fragments caused byblasting, and dressed by bush hammering.Cement dressing should be kept to a mini-mum, never more than a few millimetresthick. If measurements are to be made withboth plates, the bearing surfaces of therock must be strictly parallel.

Places where the frame is anchored mustalso be surface-dressed to ensure thatrods, supposed to be fixed, are not set onrock spalls, not intimately attached to thesurrounding rock.

As in the dilatometer test, the rigid plateloading test proceeds by loading/unloa-ding cycles to increasing maxima, maintai-ning load constant for short times in each

APPENDIX 10

HYDRAULIC PARAMETERS

Fluid flow through discontinuities is ahighly complex problem. Experimentalwork (Gentier 1986) has shown it is not iso-tropic but follows channels whose geome-try, while of course dependent on jointwidth, is also governed by wall roughnessand their contact surfaces, by applied nor-mal and tangential stress levels, and bytangential movements of the walls, withoutforgetting the possible presence of fillingmaterial.

Detailed determination of the network ofchannels and changes thereto due to shearmovements is therefore essential formodelling the hydraulic behaviour of a dis-continuity, but this is as yet still at theresearch stage.

Various more or less simplified approachesare however available for estimating frac-ture flow rate Q with the equation

Q = Va = Kf Jf a (for laminar flow)

in which

Q is the flow rate through the discontinuityper unit width

V is the mean fluid flow velocity in the dis-continuity

Kf is the hydraulic conductivity of the dis-continuity

J f is the orthogonal projection of thehydraulic gradient on the fracture plane

a is the width or aperture of the disconti-nuity.

With a smooth-walled planar discontinuity,hydraulic conductivity Kf(LT-1) dependssolely on its physical aperture and, by ana-logy with flow between flat plates, is written

Kf = (a2g)/12ν

in which

a is the physical aperture of the disconti-nuity (L)

g is the acceleration due to gravity (LT-2)

ν is the kinematic viscosity of the fluid (L2T-1)

whence flow Q is proportional to the cubeof the aperture (cubic law).

In a natural discontinuity, the geometry ofthe voids between the walls is not constant

and depends on wall roughness. Varioussemi-empirical formulae include for rough-ness with the coefficient

ra/2a

called relative hydraulic roughness, inwhich ra represents the difference betweenthe highest peak and the lowest trough injoint wall geometry:

Kf = (a2g/12ν) [1/{1 + B(ra/2a)1.5}]

with, for example, B = 8.8 (Louis 1969), B = 20.5 (de Quadros 1982).

With ra≤ 2a < 0.033, the effect of rough-ness is negligible and the cubic law forparallel plate flow can be used.

Barton (1985) established a correlation bet-ween relative hydraulic roughness and theJoint Roughness Coefficient JRC. This ledhim to introduce the concept of hydraulicaperture A of a discontinuity, related tophysical aperture a and JRC as

A = JRC2.5/(A/a)2

The dimensions of a and A are millimetres.



Appendix 11 – Figure 1 – Rigid plate loading test apparatus set-up

Appendix 11 – Figure 2 – Curves from plate loading test and derived moduli

cycle to read the gauges and for a longertime at each cycle maximum to detect anycreep that might occur. The test may beentirely automated, especially for creepmeasurements.

Test data is interpreted with theBoussinesq equation for a circular rigidplate of diameter f and a stress s applied to

a semi-infinite, homogeneous, isotropic,elastic material with modulus of elasticity Eand Poisson's ratio ν. The displacement ofthe plate ∆d at each stress interval conside-red is

∆d = π/4 x (1 - ν2) x ∆σ x φ/E

If ν is not known, it is usually assigned avalue of 0.25.

The stress strain curves for successivecycles are plotted for each test (figure 2)and so-called deformation global moduliare found, along with higher reversiblemoduli. A global deformation modulus isalso determined, which corresponds to themean slope of the tangent to the cyclecurves.

APPENDIX 12

THERMAL ROCK PARAMETERS

Rock Conductivity // Conductivity _λ1 (W/m.°C) λ2 (W/m.°C)

Granites 2.7 – 3.5 2.7 – 3.5

Basalt 2.2 2.2

Gneiss 3 – 4 2V6 - 3

Crystalline schists 3.3 – 4.7 2.5 - 3

Quartzites 5-5.6 5

Anhydrite 5.5 5.5

Sandstone 2.7 2.7

Limestones 2.5-3.6 1.9-3.6

Dolomite 4.2 4.2

Clay 1.9 1.9

Rock salt 5.7 5.7

Appendix 12 – Table 1 Thermal conductivities of selected rocks (afterMechler 1982, Handbook of Chemistry and Physics,1987, Goy 1996)

The lower the thermal conductivity value,the more the rock acts as a heat insulant

APPE

ND

IX


PPE

ND

IX

App

endi

x 13

� T

able

1 (

cont

�d)

� D

efin

itio

n of

RM

R (

afte

r B

ieni

awsk

i 198

9)

APP

END

IX

13

BIEN

IAW

SKI’S

RM

R CL

ASS

IFIC

ATI

ON

OF

ROCK

MA

SSES


PPE

ND

IX

Appe

ndix

13

– Ta

ble

1 (c

ont’d

) –

Def

initi

on o

f RM

R (a

fter

Bien

iaw

ski 1

989)

APP

END

IX 1

3

BIEN

IAW

SKI’S

RM

R CL

ASS

IFIC

ATI

ON

OF

ROCK

MA

SSES



PPE

ND

IX

APPENDIX 14

BARTON’S Q INDEX

DESCRIPTION VALUE NOTES

ROCK QUALITY DESIGNATION RQDVery poor 0-25 i) Where RQD is reported or measuredPoor 25-50 as < 10 (including 0) a nominalFair 50 - 75 value of 10 is used to evaluate QGood 75 - 90 ii) RQD intervals of 5, i.e. 100, 95,90 etc. are suffi-ciently accurateExcellent 90 -100

JOINT SET NUMBER JnMassive, no or few joints 0.5 – 1.0One joint set 2One joint set plus random 3Two joint sets 4Two joint sets plus random 6Three joint sets 9 i) For intersections use (3.0 x Jn)Three joint sets plus random 12Four or more joint sets, random, heavily jointed, “sugar cube”, etc 15 ii) For portals use (2.0 x Jn)Crushed rock, earthlike 20

JOINT ROUGHNESS NUMBER Jra) Rock wall contactb) Rock wall contact before 10 cm shear

Discontinuous joints 4Rough and irregular, undulating 3Smooth undulating 2Slickensided undulating 1.5Rough or irregular, planar 1.5Smooth, planar 1.0Slickensided, planar 0.5 i)Add 1.0 if the mean spacing of the relevant joint set

is greater than 3c) No rock wall contact when sheared

Zones containing clay linerals thick enough to 1.0 ii) Jr = 0.5 can be used for planar, slickensided jointsprevent rock wall contact (nominal) having lineations, Sandy, gravely or crushed zone thick enough o 1.0 Provided that the lineations are prevent rock wall contact (nominal) oriented for minimum strength.

JOINT ALTERATION NUMBER Ja _r degrees (approx.)

a) Rock wall contact

Tightly healed, hard, non-softening, impermeable filling 0.75 i) Values of _r, the residual angle friction

Unaltered joint walls, surface staining only 1.0 25 - 35 Are intended as an approximate

Slightly altered joint walls, non-softening mineral 2.0 25 - 30 Guide to the mineralogical properties coatings, sandy particles, clay-free, disintegrated of the rock, etc.Silty, or sandy-clay coatings, small clay fraction 3.0 20 - 25 Alteration products, if present(non-softening)

Softening or low-friction clay mineral coatings, i.e. 4.0 8 - 16kaolinite, mica. Also chlorite, talc, gypsum and graphite etc., and small quantities of swelling clays (Discontinuous coatings 1 – 2 mm or less in thickness)


PPE

ND

IX

DESCRIPTION VALUE NOTESJOINT ALTERATION NUMBER Ja _r degrees (approx.)

b) Rock wall contact before 10 cm shear Sandy particles, clay free, disintegrating rock, etc. 4.0 25 -30 Stronly over-consolidated, non-softening clay minerals 6 16 - 24fillings (continuous <5 mm thick) Medium or low over-consolidation, softening clay 8 12 - 16mineral fillings (continuous <5 mm thick)Swelling clay fillings, i.e. montmorillonite, 8.0 –12.0 6 -12(continuous <5 mm thick). Values of Ja depend on percent of swelling clay-size particles, and access to water

c) No rock wallcontact when shearedZones or bands of disintegrated or crushed 6.0Rock and clay 8.0Zones or bands of silty-or sandy clay, small clay fraction, 5.0non-softeningThick continuous zones or bands of clay 10.0 – 13.0

JOINT WATER REDUCTION Jw Approx.water pressure (kgf/cm2)

Dry excavation or minir inflow i.e. < 5 l/m locally 1.0 < 1.0Medium inflow or pressure, occasional outwash of joint fillings 0.66 1.0 – 2.5 i) Factors are crude estimates;Large inflow or high pressure in competent rock with 0.5 2.5 – 10.0 increase Jw if drainageinstalledunfilled jointsLarge inflow or high pressure 0.33 2.5 – 10.0Exceptionally high inflow or pressure at blasting, 0.2 – 0.1 > 10 ii) Special problems caused decaying with timeExceptionally high inflow or pressure 0.1 – 0.05 > 10 By ice formation are not

considered

STRESS REDUCTION FACTOR SRFa) Weakness zones intersecting excavation, which may cause loosening of rock mass when tunnel is excavated

Multiple occurrences of weakness zones containing 10.0 i)Reduce these values of SRF by 25-50% if the relevant shearclay or chemically disintegrated rock, very loose zons only influence but do not intersect the excavationsurrounding rock and depth)Single weakness zones containing clay, or chemically disintegrated rock (excavation depth < 50 m) 5.0Single weakness zones containing clay, or chemically disintegrated rock (excavation depth > 50 m) 2.5Multiple shear zones in competent rock (clay free), loose surrounding rock (any depth) 7.5Single shear zone in competent rock (clay free) (depth of excavation <50 m) 5.0Single shear zone in competent rock (clay free) (depth of excavation >50 m) 2.5Loose open joints, heavily jointed or “sugar cube”, (any depth) 5

b) Competent rock, rock stress problems σc / σ1 σt / σ1 SFRLow stress, near surface > 200 > 13 2,5 i) For strongly anisotropic virgin stress field

(if measured): whenMedium stress 200 - 10 13 –0.66 1.0 5<σ1 / σ3 <10 reduce σc to 0.8σcHigh stress, very tight structure (usually favourable to stability, may be unfavourable to wall stability) 10 - 5 0.66 – 0.33 0.5-2 and σt to 0,8σt.When σ1 / σ3 >10Mild rockburst (massive rock) 5-2.5 0.33-0.16 5-10 Reduce σc and σt to 0.6σc and 0.6 σtHeavy rockburst (massive rock) < 2.5 < 0.16 10-20

c) Squeezing rock plastic flow og incompetent rock under influence of high rock pressure

Mild squeezing rock pressure 5 – 10 ii) Few case recordsHeavy squeezing rock pressure 10 -20 avalaible where depth of crown below

d) Swelling rock, chemical swelling activity surface is less than span width. Suggestdepending on presence of water

Mild swelling rock pressure 5 – 10 SRF increase from Heavy swelling rock pressure 10 -15 2.5 to 5 for such cases



PPE

ND

IX

Many empirical relationships have beenproposed by various authors to estimatethe rock mass deformation modulus fromthe characteristics obtained from labora-tory samples and other parameters. Table1 lists the more important correlations withRMR, RQD and Q index used in rock massclassification systems.

AFTES expresses no opinion on the rele-vance or validity of these correlations inrespect of the use to which they are put.Users should refer to the writings of therelevant authors.

APPENDIX 15

EMPIRICAL FORMULAE FOR EVALUATING ROCK MASS DEFORMATION MODULI

APPENDIX 16

REFERENCES

ROCK MASS DEFORMATION MODULUS AUTHORSEM (MPa)

2*(RMR – 50) ou 1.7* (RQD –60) Cording et al (1971)

25 Ln Q ou 10 Ln Q Fujita (1977)

0,7*(RMR/100)2 *Ei Barton (1980)

RMR/10 + (RMR3/105) Hoek & Brown (1982)

10(RMR-10)/40 Serafim & Pereira (1983)

0,5*(RQD/100)2 *Ei Bieniawski (1989)

10*exp[(RMR-10)/40] Grimstad & Barton (1993)

0,07*RQD+0.05 σc+55*Ei Hönisch (1993)

1000 *[σc/100]0,5*10(GSI-10)/40 (σc<100 MPa) Hoek & Brown (1997)

Appendix 15 – Table 1 – Empirical relationships proposed by various authors for assessing rock mass deformation modulus

• Ei Modulus of elasticity of rock measured on laboratorysamples

• σc Uniaxial compressive strength of rock measured on labora-tory samples

• RQD Rock Quality Designation (Deere 1967)

• Q Quality factor (Barton 1980)

• RMR Rock Mass Rating (Bieniawski 1989)

• RMR = 50 + 15 log10 Q (Barton 1995)

• GSIGeological Strength Index

• GSI = RMR89 – 5 (Hoek & Brown 1994, 1995); GSI = 9 Ln Q + 44(Bieniawski 1989)

16.1 - BIBLIOGRAPHIE GENERALEComité Français de Mécanique des Roches (2000) - "Manuel deMécanique des Roches – Tome 1 – Fondement " – Coordonné parHOMAND F. & DUFFAUT P. – Presse de l’Ecole des Mines – Paris, 268 p.

BOUVARD-LECOANET A., COLOMBET G., ESTEULLE G. (1988) -" Ouvrages souterrains-Conception-Réalisation-Entretien " - Pressesde l’Ecole Nationale des Ponts et Chaussées, 262p.

HOEK E. & BROWN ET. (1980) - "Underground excavations in rock"– London : Int. Min. Metall. 1980.

PANET M. & FOURMAINTRAUX D. (1976) – "La Mécanique desroches appliquée aux ouvrages de génie civil ", Association Amicaledes Ingénieurs Anciens élèves de l’E.N.P.C., Paris, 236 p.

16.2 - ROCK MATRIX CHARACTERISATIONPARAMETERSBROCH et FRANKLIN, Int. J. of Rock Mech. And Min.Sc., 1972,pp.669 – 697.

HUDER J. & AMBERG G. 1970 - "Quellung in Mergel, Opalinustonund Anhydrit", Schweizerische Bauzeitung n° 43, p. 975 – 980.

I.S.R.M. (1999) - "Suggested methods for laboratory testing of swel-ling rocks " - International Journal of Rock Mechanics and MiningSciences, 36, p. 291 – 306.

I.S.R.M. (1985) - "Suggested methods for determining point loadstrength " - International Journal of Rock Mechanics and MiningSciences, 22 (2), 51-60.

MOVINKEL T. & JOHANNESSEN O. (1986) - "Geological parametersfor hard rock tunnel boring." - Tunnels & Tunneling, april 86, pp. 45-48.

16.3 - CHARACTERISATION PARAMETERS OFDISCONTINUITIESBARTON N. & CHOUBEY V. (1977) - "The shear strength of RockJoints in Theory and Practice" - Rock Mech. Engng. Geol. 10,pp. 1-54.

DEERE D.U. (1963) - "Technical description of rock cores for enginee-ring purposes" - Rock Mech. Engng. Geol. 1, pp.16-22.



PPE

ND

IX

LOUIS C. (1974) - "Reconnaissance des massifs rocheux par sondageset classifications géotechniques des roches" - Sols et Fondations,108, N°319, pp. 97-122.

SERRATRICE J.F. & DURVILLE J.L (1997) - "Description des roches etdes massifs rocheux : exploitation de deux bases de données" -Bulletin des Laboratoires des Ponts et Chaussées, 211, pp. 73-87.

16.4 - SET PERSISTENCE OF DISCONTINUITIESAND 3D GEOMETRICAL MODELLING OFNETWORKS OF DISCONTINUITIES(APPENDIX 8)BAECHER G.B., LANNEY N.A. & EINSTEIN H.H. (1977) - "Statisticaldistribution of rock properties and sampling" - Proc. 18th U.S. Symp.on Rock Mech., Colorado, pp. 5c1.1-8.

BILLAUX D. (1990) - "Hydrogéologie des milieux fracturés : géomé-trie, connectivité et comportement hydraulique" - Thèse de docto-rat, CIG, Ecole des Mines de Paris.

CACAS M.P. (1989) - "Développement d'un modèle tridimensionnelstochastique discret pour la simulation de l'écoulement et des trans-ferts de masse et de chaleur en milieu fracturé" - Thèse de doctorat,CIG, Ecole des Mines de Paris.

HELIOT D. (1988) - "Conception et réalisation d’un outil intégré demodélisation des massifs rocheux fracturés en blocs" - Thèse de doc-torat INPL, Nancy.

HUDSON J.A. & PRIEST S.D. (1979) - "Discontinuities and rock massgeometry" - Int. J. Mech. Min. Sci. & Geomech. Abstr., Vol. 16, pp.339-362.

LONG J.C.S, GILMOUR P. & WISERSPOON P.A. (1985) - "Amodel for steady fluid flow in random three dimensional networksof disk-shaped fractures" - Water resources research, 21, N°8,pp. 1105-1115.

PAHL P.J. (1981) - "Estimating the mean length of discontinuity traces"- Int. J. Mech. Min. Sci. & Geomech. Abstr., Vol. 18, pp. 221-228.

WARBURTON P.M. (1980) - "Stereological interpretation of joint tracedata, influence of joint shape and implication for geological surveys" -Int. J. Mech. Min. Sci. & Geomech. Abstr., Vol. 17, pp. 305-316.

XU J. (1991) - "Simulation statistique de discontinuités et évaluation dela blocométrie de massif rocheux, application à l’analyse de l’écoulementet de la stabilité" - Thèse de doctorat, CGI, Ecole des Mines de Paris.

16.5 - SCHMIDT HAMMER TEST (APPENDIX 9)BARTON N. & CHOUBEY V. (1977) - "The shear strength of RockJoints in Theory and Practice" - Rock Mech. Engng. Geol. 10,pp. 1-54.

HOEK E. & BRAY J.W. (1974) - "Rock slope Engineering" - TheInstitution of Mining and Metallurgy, London, revised third edition1981.

I.S.R.M. (1978) - "International Society for Rock Mechanics

Commission on Standardization of laboratory and field tests : Suggestedmethods for the description of discontinuities in rock masses" - Int. J.Mech. Min. Sci. & Geomech. Abstr., Vol. 15, pp. 319-368.

MILLER R.P. (1965) - "Engineering classification and index propertiesfor intact rock" - Ph.D. Thesis, University of Illinois.

16.6 - HYDRAULIC PARAMETERS (APPENDIX 10)BARTON N., BANDIS S. & BAKHTAR K.(1985) - "Strength, deforma-tion and conductivity coupling of rock fractures" - Int. J. Mech. Min.Sci. & Geomech. Abstr., Vol. 22, pp. 121-140.

BARTON.& DE QUADROS E.F., (1997) - "Joint aperture and rough-ness in the prediction of flow and groutability of rock masses" - Int. J.Mech. Min. Sci. & Geomech. Abstr., Vol. 34, No. 3-4, PaperNo.252.

GENTIER S. (1986) - "Morphologie et comportement hydroméca-nique d'une fracture naturelle dans le granite sous contrainte normale;étude expérimentale et théorique" - Documents du BRGM b34,Orléans, 597 p.

LOUIS C. (1969) - "Etudes des écoulements d'eau dans les roches fis-surées et leur influence sur la stabilité des massifs rocheux" - BRGM,Bulletin de la Direction des Etudes et Recherches. Série A.(3), pp.5-132.

DE QUADROS E.F. (1982) - "Determinacão das caracteristicas+ defluxo de agua em fracturas de rochas" - Dissert. de Maestrado.Dept. of Civil Eng., Polytech. School, University of São Paulo.

16.7 - ROCK MASS CHARACTERISATIONPARAMETERS16.7.1 - Indirect measurements

BOUVARD A., HUGONIN J. & SCHNEIDER, B. (1994) - "SCARA-BEE, - Méthode de reconnaissance des massifs rocheux – Applicationaux ouvrages souterrains" – Tunnels et ouvrages souterrains n°123Mai/Juin 1994.

CARRERE A., RIVET, J. & SCHNEIDER B. - "La petite sismique" –Géologues n° 92.

16.7.2 - Rigid plate loading test

MAZENOT P. - "L’essai à la plaque – Congrès des Grands Barrages –Edimbourg – 1964 – Q.28 – Rapport 15 – Groupe de travail". –Ann.ITBTP, 1965.

16.7.3 - Stresses

AMADEI B. & STEPHANSON O. (1997) – “Rock Stress and itsMeasurement” – Chapman and Hall, London, 490 p.

BRIGLIA P., BURLET D. & PIRAUD J. (1994) – “La mesure descontraintes naturelles appliquée au Génie civil ” – Tunnel etOuvrages Souterrains, n°123, mai-juin 1994, pp. 149-152.

BURLET D. (1991) – “ Détermination du champ des contraintes régio-nal à partir de tests hydrauliques en forage ”. Thèse de doctorat,Univ. Paris VII.

APPENDIX 16

REFERENCES



PPE

ND

IX

Comité Français de Mécanique des Roches(2000) - "Manuel de Mécanique des Roches– Tome 1 – Fondement- Chap. 7 – Lescontraintes dans les massifs rocheux et leurdétermination " –Coordonné parHOMAND, F. et DUFFAUT, P.– Presse del’Ecole des Mines – Paris, 268 p.

FABRE D., MAYEUR B. & SIRIEYS,P. (2002) -"Paramètres caractérisant l’état decontraintes naturel dans les massifsrocheux". - Symposium "Paramètres de cal-cul géotechnique" – Magnan (ed), Pressesde l’E.N.P.C., Paris p. 359-368.

16.7.4 - Thermal rock parameters(appendix 12) FABRE D. (2001) - "Prediction of tem-perature for deep tunnel projects". Pro.Int. Symp.Geonics – Temperature andits influence on Geomaterials – Ostrava,pp. 67 –75.

GABLE R. (1986) - "Température, gradient etflux de chaleur terrestre. Mesures, interpréta-tion" – Document du BRGM, n° 104, 187 p.

GOY L. (1996) - "Mesure et modélisationdes températures dans les massifs rocheux.Application au tunnel profond Maurienne-Ambin." – Thèse de doctoratGéomécanique, Univ. Grenoble, 202 p.

MECHLER P. (1982) - "Les méthodes de lagéophysique" – Dunod éd. Paris 2000

HODGMAM Ch.D. (1987) - "Handbook ofChemistry and Physics" – 30ème édition -Chemical Rubber Publishing Company,Cleveland , Ohio, USA, 2651 p.

FABRE D., GOY L., MENARD G. & BURLETD. (1996) - "Température et contraintes dansles massifs rocheux : cas du projet de tunnelMaurienne-Ambin" – Tunnels et ouvragessouterrains, n° 134, p.85 – 92.

16.7.5 - Hydrogeological conditions

BORDET Cl. "L'eau dans les massifs rocheuxfissurés ; observations dans les travaux sou-terrains". Conf. au CERES, Liège, 14décembre 1970.

DUFFAUT P. "Les tunnels en terrain aqui-fère" In : La pratique des sols et fondations,coord. G. FILLIAT, Ed. du Moniteur, 1981,pp. 808-810.

VANDENBEUSCH M. & WOJTKOWIAK F."La prise en compte de l'eau et la prévision

de l'exhaure en mines et carrières" Mines etCarrières - Industrie Minérale. Février 1992.

LOMBARDI G. "Hydrogeologische Aspektevon Tunnelprojekten" Felsbau. Vol. 12, n° 6,1994.

GAUDIN B. "Maîtrise de l'eau dans les tun-nels en construction et en service". Conf. àl'Ecole des Ponts et Chaussées, 27 mars1997.

16.8 - USE OF ROCK MASSCHARACTERISATION FORDESIGN AND CONSTRUC-TION OF UNDERGROUNDSTRUCTURESFascicule Génie civil Dossier pilote desTunnels – Génie Civil (Juillet1998)

Section 1 : IntroductionSection 2 : Géologie – Hydrogéologie –GéotechniqueSection 3 : Conception et dimensionnement

Centre d’Etudes des Tunnels.

HOEK E. & BROWN ET. – "Practical esti-mates of rock mass strength" – Int.J.RockMech. Mn-Sci, Vol 34, pp.1165-1186, 1997

HOEK E., KAISER P.K. & BAWDEN W.F. -"Support of underground excavations inhard rock" – Rotterdam, Balkema. 1995

Evaluation de la résistance limite des mas-sifs rocheux

BALMER G. "A general analytical solutionfor Mohr's enveloppe" - Am. Soc. Test. Mat.52, 1260-1271

16.9 - AFNOR STANDARDS

NF P 94-066 (Décembre 1992) –"Coefficient de fragmentabilité des maté-riaux rocheux" – AFNOR 1992

NF P 94-067 (Décembre 1992) – "Coefficientde dégradabilité des matériaux rocheux" –AFNOR 1992

NF P 94-410-1 (Mai 2001) – "Essais pourdéterminer les propriétés physiques desroches – Partie 1 : Détermination de la teneuren eau pondérale – Méthode par étuvage" –AFNOR 2001

NF P 94-410-2 (Mai 2001) – "Essais pourdéterminer les propriétés physiques desroches – Partie 2 : Détermination de la massevolumique – Méthode géométrique et parimmersion dans l’eau" – AFNOR 2001

NF P 94-410-3 (Mai 2001) – "Essais pourdéterminer les propriétés physiques desroches – Partie 3 : Détermination de la poro-sité" – AFNOR 2001

NF P 94-411 (Avril 2002) – "Déterminationde la vitesse de propagation des ondes ultra-sonores en laboratoire – Méthode par trans-parence" – AFNOR 2002

NF P 94-420 (Décembre 2000) –"Détermination de la résistance à la com-pression uniaxiale" – AFNOR 2000

NF P 94-422 (Janvier 2001) – "Déterminationde la résistance à la traction – Méthode indi-recte – Essai brésilien" – AFNOR 2001

NF P 94-423 (Mars 2002) – " Déterminationde la résistance à la compression triaxiale" –AFNOR 2002

NF P 94-425 (Avril 2002) – " Déterminationdu module de Young et du coefficient dePoisson" – AFNOR 2002

NF P 94-430-1 (Octobre 2000) – "Déterminationdu pouvoir abrasif d’une roche – Partie 1 :Essai de rayure avec une pointe" – AFNOR2000

NF P 94-430-2 (Octobre 2000) – "Déterminationdu pouvoir abrasif d’une roche – Partie 1 :Essai avec outil en rotation" – AFNOR 2000

NF P 94-429 (en cours d’établissement) –"Détermination de la résistance à la com-pression d’une roche entre pointes – EssaiFranklin" – AFNOR 2000

P 18-572 (Décembre 1990) – " Essai d’usuremicro-Deval" – AFNOR 1990

P 18-573 (Décembre 1990) – " Essai LosAngeles" – AFNOR 1990

XP P 94-443-1 (Février 2002) –"Déformabilité – Essai dilatométrique enforage – Partie 1 : Essai avec cycles" –AFNOR 2002

XP P 94-443-2 (Février 2002) –"Déformabilité – Essai dilatométrique enforage – Partie 1 : Essai de fluage après lepremier cycle" – AFNOR 2002

APPENDIX 16

REFERENCES

GT1R1A_Caracterisation of Rock Masses.pdf

Documents

Transcript of GT1R1A_Caracterisation of Rock Masses.pdf