Sie sind auf Seite 1von 6

PRACTICAL CONSIDERATIONS IN THE USE OF ROCK MASS

CLASSIFICATION IN MINING
Douglas Milne1 and John Hadjigeorgiou2

ABSTRACT
Rock mass classification systems for rock engineering have gained widespread acceptance since their
introduction in the mid-60's. Their widespread use can be attributed to the ease with which they employ
simple measurements and observations to provide a quantitative index of rock mass quality. It is then
possible to link this index to specific engineering objectives such as predicted stand-up time, maximum size
of stable excavation, choice of support etc.
Improvements in rock mass classification systems and their application have been based on more
quantitative field measuring techniques, improvements to the weightings assigned to the different
parameters, and better links to design. The widespread use of classification systems for design has resulted
in a significant increase in case histories linking classification to rock mass behaviour, under a wide variety
of conditions. Consequently this has contributed to the statistical validity of design recommendations. More
recently there has been a trend to link different classification systems to avoid collecting the required field
data to quantify the additional classification values. This practice cannot be justified as it can result in
serious errors.
INTRODUCTION
Rock mass characterization is an integral part of any design process in mining, even though most
classification systems now in use have their origin in underground civil engineering applications. Due to
differences in engineering philosophies between civil and mining, it is not surprising that there have been
problems transferring experience from one field to the other. Project duration, scale, access, even tolerance
of risk, can contribute to developing different engineering strategies for civil and mining applications.
In recent years there has been a tendency to use the results of rock mass classification systems for a
variety of purposes. In general the basic aims are:
a) to identify zones of material of similar geomechanical characteristics
b) to provide an indication of predicted stability of excavations of a given size
c) to aid in the selection of an appropriate support strategy
d) to provide an indication of in situ rock mass strength, modulus of elasticity etc.
Assuming that rock mass classification systems can accomplish these objectives, it would appear that
there is ample justification for allocating the appropriate resources to collect and analyze the necessary input
data. There is evidence, however, that rock mass classification is not always done consistently and
geomechanical databases are not updated on a regular basis.
It is the aim of this paper to raise questions and encourage discussion on the wisdom of employing rock
mass classification systems. Based on the authors' experience the following questions are put forward:
a) How reliable and representative are classification systems?
b) Can and should we improve the existing classification systems?
c) Can we justify using "empirical formulae" to convert from one system to another?
d) Does the continued use of existing systems hinder the development of more rational approaches to
mine design?

Dr. Douglas Milne, Geological Engineering Program, University of Saskatchewan, 114 Science Place, Saskatoon,
SK.. S7N 5E2 CANADA.
2
Dr. John Hadjigeorgiou, Dep. de mines et mtallurgie, Universit Laval, Quebec, Que. G1K 7P4 CANADA

CLASSIFICATION AND DESIGN


The results of rock mass classification are routinely used to estimate rock mass behaviour as well as
provide guidelines for underground mine design. Documented case studies can provide an extraordinary
amount of knowledge and experience that can be used for design purposes. There has been a tendency to
also use the resulting empirical ratings to determine the constitutive laws as well as failure criteria now
employed with the majority of numerical models. The more popular systems in mining are RMR by
Bieniawski (1976), the Q system by Barton et. al., (1974) and MRMR proposed by Laubscher (1984).
Given the extent that classification systems have been integrated in all elements of design, it would
appear that there is a very strong case for assigning more resources to conduct regular rock mass
characterization campaigns. Mapping and routine interpretation of the results can provide valuable insight
into the behaviour of the rock mass and the performance of any design. In several mines it is now common
for the results of rock mass classification to be routinely integrated into different mine design packages. This
allows for quick electronic retrieval and the creation of maps zoning the mine by rock mass quality. It can
easily be demonstrated that a difference in the interpretation of the rock mass quality can have important
safety and economic consequences on the stability, size and support requirements of an excavation.
COMPARING EXISTING CLASSIFICATION SYSTEMS
One of the main objectives of classification systems is to minimize ambiguity in quantifying geological
properties of a rock mass. Each system selects what it considers the most significant properties of a rock
mass to which a rating is assigned. The addition or multiplication of these individual ratings provides a
unique value of the quality of the rock mass. Milne et al. (1998) have shown how the weight assigned to
different parameters has changed over the years and how this can have an adverse influence on the integrity
of prior work, especially empirically derived design recommendations.
Table 1 summarizes the comparative weight assigned to different rock mass properties in several
classification systems used in mining. The Q, Q' (a variation of Q used in mining where the Stress Reduction
Factor is set to 1.0) RMR76, RMR89 and MRMR classification systems are shown in Table 1. For
comparative purposes, the range for each parameter is compared to the total possible range for the
classification system. For the Q system, the percentage change is based on a logarithmic scale.
Table 1 : Comparative Weight Assigned to Different Rock Mass Properties
Classification system
Q
Q
RMR76
RMR89
Basic range in values
0.0006 to 1140
0.00003 to 2133
8 to 100
8 to 100
Unconfined Compressive Strength (UCS)
0
0
0 to 15
0
0
16
Block Size
Range
0.5 to 200
0.5 to 200
8 to 50
% of total range
41
33
46
Discontinuity Surface Friction
Range
0.025 to 5.3
0.025 to 5.3
0 to 25
% of total range
38
30
27
Groundwater
Range
0.05 to 1.0
0.05 to 1.0
0 to 10
% of total range
21
17
11
Stress or UCS/Stress Ratio
Range
0
0.5 to 20
0
% of total range
0
20
0
1
Groundwater is included with joint condition in the MRMR system.
2
The stress correction factor in the MRMR system has not been included.
Range
% of total range

MRMR
0 to 100

0 to 15
16

0 to 20
20

8 to 40
35

0 to 40
40

0 to 30
33

4 to 40
36

0 to 15
16

8 to 281
20

0
0

02
0

Table 1 shows there are important similarities between the different classification systems. This is
reflected in the choice and comparable weight assigned to each parameter. Closer scrutiny, however,
suggests that there are also important differences, any of which can render efforts to correlate between the
different systems invalid.
It is difficult to justify the seemingly arbitrary relative weightings assigned to rock mass properties in
classification schemes. A further concern is that in arriving at a single classification value, certain
parameters can compensate for each other. For example, in a given rock mass the addition of chlorite to a
clean joint surface (i.e. an increase in the joint alteration value) has the same effect on the final Q rating as
increasing from one joint set plus random joints, to three joint sets plus random joints. Similarly in the RMR
system, a reduction in UCS from 220 MPa to 150 MPa has the same effect on the final rating as going from
dry rock mass conditions to moist conditions. These seemingly arbitrary relationships demonstrate why
detractors of classification systems argue that representing the rock mass as a single number can mask
important characteristics of the rock mass.
Some of the differences between classification systems have been highlighted, however, there are still
frequent attempts made to justify linking different systems. A variety of relationships have been developed
to correlate some of the more commonly used classification systems. Correlations between Q and RMR are
frequently used and may appear statistically valid, however, it can be easily shown that they cannot apply in
all circumstances. Stress is absent in the RMR system while intact rock mass strength is only indirectly
assessed in the NGI system. Any of these parameters can be the dominant factor leading to instability for a
particular rock mass. Contrary to what may be the consensus, the authors feel that converting from RMR to
Q is a potential source of significant errors. Changing stress conditions encountered in a deep mine can
easily result in a range of Q values over two or three orders of magnitude. The corresponding RMR
classification would remain constant at changing depths. Under these conditions, any correlation would only
be accurate within a narrow range of depth and stress conditions.
APPROACHES FOR IMPROVING CLASSIFICATION SYSTEMS
There are three major elements in a rock mass classification system:
a) Identification of key parameters governing the behaviour of the rock mass.
b) Assigning a relative weighting to each of these parameters.
c) Quantifying these key parameters of the rock mass.
These three elements combine to produce classification values for a rock mass, which must then be linked to
a particular engineering process.
Many of the existing classification systems rely on descriptive terms such as fairly rough or soft joint
wall rock to describe rock mass conditions. Consistent engineering design cannot rely on subjective
descriptive terms such as this. As a result, several techniques have been developed to better quantify rock
mass properties such as joint surface conditions and intact block size and shape, some of which are described
in the next sections. Our ability to measure and quantify rock mass properties has increased significantly
since the development of most classification systems.
Input Data
The most promising area of rock mass characterization has been in developing techniques to improve and
standardize how we quantify rock mass properties. The key parameters of the majority of rock mass
classification systems can be related to the degree of fracturing of a rock mass and the joint surface
properties. Advances have been made in the areas of block size analysis, joint surface roughness and joint
alteration assessment.
Joint Roughness
It is now recognized that small-scale and large-scale roughness have an important role in defining the
behaviour of the rock mass. In the RMR system, joint roughness is part of "condition of discontinuities",
however, no distinction between small and large-scale roughness is made. The Q system follows a similar
subjective approach, although it identifies two scales of joint roughness as a distinct input parameter, Jr.
Classification systems have benefited from less subjective approaches for assessing joint surface
conditions. The Barton-Bandis failure criterion employs the Joint Compressive Strength, the basic angle of

friction and the non-dimensional Joint Roughness Coefficient (JRC) to predict the shear strength of
discontinuities, Barton & Bandis (1990). The JRC term is linked to the roughness profile of a joint surface.
Barton and Bandis (1990) report how the JRC term can also be linked to the Jr term in the Q classification
system (Table 2). A significant degree of judgement is required for estimating JRC values and another, more
quantifiable approach for determining Jr is also presented (Milne et al., 1991). The method is based on
measuring the maximum amplitude of a joint with respect to a standard length. Figure 1 demonstrates how a
small ruler and a 1m long stick can be used to quickly evaluate the relationship between joint amplitude (a)
and joint length (L). This approach provides a quick, easily conducted and repeatable field measurement and
the authors feel the benefits of this simplistic method outweigh the disadvantages. If sufficient
measurements are taken it can provide a statistically significant and representative measure of some of the
joint properties influencing strength. Table 2 shows the two approaches for estimating Jr.

a
L

Figure 1 : Measuring joint amplitude to assess planarity and roughness (From Milne et al., 1991)
Table 2 : Field Measurement Approach for Obtaining Jr Values
Large scale roughness
Planar
Wavy or undulating
Stepped
(a/L < 10mm/m)
(a/L 20mm/m)
Jr = 0.5
Jr = 1.5
Jr = 2.0
JRC.2m = 5, JRC1m = 4
JRC.2m = 7, JRC1m = 6
JRC.2m = 11, JRC1m = 8
Smooth
Jr = 1.0
Jr = 2.0
Jr = 3.0
(a/L<2.5mm/10cm) JRC.2m =1.5, JRC1m = .9
JRC.2m = 11, JRC1m = 8
JRC.2m = 14, JRC1m = 9
Rough
Jr = 1.5
Jr = 3.0
Jr = 4.0
(a/L>2.5mm/10cm) JRC.2m = 2.5, JRC1m = 2.3
JRC.2m = 14, JRC1m = 9
JRC.2m = 20, JRC1m = 11
Jr
Small Scale
Roughness
Slickensided

Joint Alteration
It is well recognized that joint alteration can have an important role to play in characterizing a jointed
rock mass. In the most popular classification systems, alteration is described in only subjective terms
making accurate and repeatable assessments of this property difficult. Hadjigeorgiou et al. (1994)
investigated the use of a portable device, the Friction Index Tester (FIT), to assess small-scale surface
friction, and by extrapolation, the degree of joint alteration. The unit consists of a small smooth wheel,
which is pressed onto the joint surface at a measured normal load. A torque gauge applies a shear force on
the wheel / rock contact and causes movement at a measured torque. The shear and normal force can then be
related to an index of the frictional properties of the joint surface. In a laboratory program, the FIT provided
a satisfactory assessment of the friction angle for a series of minerals commonly found as infillings. More

recent studies (Murray, 1999) have not succeeded in demonstrating that the prototype could be implemented
as a routine testing device. The authors often rely on the following simple tests for their fieldwork, shown in
Table 3.
Table 3 : Field Measurement Approach for Estimating Ja (After Milne et al., 1992)
Description
Ja
Cannot be scratched with a knife (healed)
0.75
Can be scratched with a knife
1.0-1.5
Can be scratched with a fingernail and feels slippery when touched
2.0
Can be dented with a fingernail and feels slippery when touched
4.0
Influence of Block Size and Scale
The degree of fracturing in a rock mass has long been recognized as a major influence in underground
opening stability. The RQD system was one of the first to recognize this fact, however, the system has
significant drawbacks (Milne et al., 1998). RQD is effective at alerting the engineering to the presence of a
heavily fractured rock mass, however, it is insensitive to joint spacing in excess of 10 cm. Subsequent
classification systems have incorporated RQD, however, they scaled up the sensitivity to joint spacing past
the 10 cm limit. This was done by adding a joint spacing term to account for improved rock mass stability
due to increased joint spacing. The Q system only uses RQD to directly assess joint spacing. The Jn term,
which is based on the number of joint sets present in a rock mass, is an indirect measure of the degree of
jointing. It can be easily shown, however, that if three joint sets with equal spacing are present in a rock
mass, the ratio of RQD/Jn becomes constant beyond a joint spacing of 0.7m. The RMR system chose an
average joint spacing cutoff of 2m, beyond which opening stability is assumed to be insensitive to increased
joint spacing. It can be argued that increasing joint spacing beyond 1 to 2m does not significantly change the
stability of civil engineering openings with spans that are commonly less than 5 to 10m. This is an
inappropriate assessment of joint spacing for large mining openings such as production stopes and can have
important repercussion on design.
The assessment of joint spacing in classification systems is an attempt to indirectly define block size. The
RMR system attempts this with the use of RQD and spacing parameters. The NGI classification system uses
the term RQD/Jn. (Jn is an index of the number of joint sets). Based on data collected and analyzed from
underground hard rock mines Hadjigeorgiou et al. (1998) have demonstrated that these parameters may be
inadequate.
For underground mining the assessment of block size, based on the number of joint sets present, the
spacing of joints and RQD estimates, is often based on spot mapping rather than extensive detailed line
mapping. This approach developed because classification assessments of block size ignore much of the data
collected from detailed line mapping programs. This simplified approach has important limitations and can
result in non-representative results. If, however, a comprehensive mapping program is conducted, there is
little justification to discard the full data in favour of the simplified parameters used in classification.
Hadjigeorgiou et al. (1998), amongst others, have shown how mapping data can be used to generate realistic
three-dimensional joint systems. This information can then be used to develop more complete information,
such as block size distributions, that better capture the discontinuous nature of the rock mass. Based on field
work in several underground mines Hadjigeorgiou et al. (1998), have shown that three-dimensional joint
systems can provide a better estimate of block size than provided by the traditional systems. There is a case
to be made to eventually replace the actual block size, represented by RQD and spacing in the RMR system
and RQD/Jn in the NGI system, by a more representative value. Alternatively, however, it may make more
sense to use representative three-dimensional joint systems to provide direct input into more sophisticated
analysis tools such as numerical modelling packages.
CONCLUSION
" Despite its past contributions, the rock mass classification approach will be supplemented by other methods
in due course, as the correct mechanisms are identified and modelled directly." Hudson & Harrison (1997).
The basic question is whether efforts to improve the existing classification systems are hindering the
development of new techniques. The authors fully appreciate the difficulty in changing existing

classification systems because of the extensive database behind them, which makes them so useful. The
authors have the following recommendations:
a) Continue work on improving how we quantify the input parameters for the different classification
systems.
b) Record all parameters of the most popular systems as opposed to recording a single rating.
c) Continue work in defining and quantifying structural and strength characteristics in a way that can be
introduced directly into different models.
Changes to improve classification systems are difficult to introduce. The major advantage of older, wellestablished, classification systems is the extensive database of case histories that make the classification
systems such valuable tools. Any changes to existing classification systems can invalidate these empirical
databases, significantly reducing their value to industry. Many of the design charts, which were developed
with the older versions of the RMR system (RMR76), have not been changed to correspond to the new
weighting systems for the more recent RMR schemes (Bieniawski, 1976 & Bieniawski, 1989). The new
RMR systems should not be applied to empirical design graphs developed with the older versions of RMR.
The changes to RMR may have improved the weightings system employed to account for rock mass
properties, but they have also resulted in confusion and errors in applying empirical design guidelines.
REFERENCES
Barton, N., R. Lien R. & J. Lunde (1974). "Engineering Classification of Jointed Rock Masses for the Design
of Tunnel Support". Rock Mechanics, 6, pp. 189-236.
Barton and Bandis, (1990). "Review of predictive capabilities of JRC-JCS model in engineering practice".
Proceedings of the International Symposium on Rock Joints, Loen, Norway, pp. 603-610.
Bieniawski, Z.T. (1976). "Rock Mass Classification of Jointed Rock Masses". Exploration for Rock
Engineering, Z.T. Bieniawski Ed. Balkema, Johannesburg, pp. 97-106.
Bieniawski, Z.T. (1989). Engineering Rock Mass Classifications. John Wiley & Sons, New York, p. 251.
Deere, D.U., A.J. Hendron Jr., F.D. Patton & E.J. Cording 1967. "Design of Surface and Near Surface
Construction in Rock". Failure and Breakage of Rock. C. Fairhurst ed. Society of Mining Engineers of
AIME, New York, pp. 237-302.
Murray, C., (1999). FIT Test Results, Internal Report, University of Saskatchewan, Geological Engineering.
Hadjigeorgiou J. Germain P. & Y. Potvin (1994). "In-situ estimation of Rock Joint Alteration". Canadian
tunnelling, pp. 155-162.
Hadjigeorgiou J., M. Grenon & J.F. Lessard (1998). "Defining In-Situ Block Size". CIM Bulletin, Vol. 91,
No. 1020, p. 72-75.
Hudson & Harrison (1997). "Engineering rock mechanics". Pergamon, Oxford, p. 444.
Laubscher, D.H. 1984. "Design aspects and effectiveness of support systems in different mining conditions".
Trans. Instn. Min. Metall. 93, pp. A70-A81.
Milne, D., Germain, P., Grant, D. and Noble, P. (1991). "Field observation for the standardization of the
NGI classification system for underground mine design". International Congress on Rock Mechanics, A.A.
Balkema, Aachen, Germany.
Milne, D., Germain, P. and Potvin, Y. (1992). "Measurement of rock mass properties for mine design".
Preceedings of the ISRM-Eurock Symposium on Rock Characterization, A.A. Balkema, Chester, England.
Milne, D., Hadjigeorgiou, J., Pakalnis, R., 1998. "Rock mass characterization for underground hard rock
mines". Tunnelling and Underground Space Technology, Oct. 1998, vol. 13, no 4, pp 383-391.

Das könnte Ihnen auch gefallen