1 INTRODUCTION

The main purpose oI this paper is to present a state

oI the art evaluation oI the rock mass deIormability
and to present the results oI comparisons between
the modulus oI deIormation obtained Irom dilatome-
ter and pressuremeter tests and the geomechanical
quality oI the rock mass, using the RMR classiIica-
tion. In addition, intact rock properties such as un-
iaxial compressive strength and Young`s modulus
are discussed.
2 SOME CONSIDERATIONS OF THE SCALE
EFFECTS IN ROCK MASSES
One oI the considerations oI scale eIIects in rock
masses was by Hoek and Brown (1980) where the
strength oI the rock mass was estimated by means oI
the value oI the RMR.
Evaluation of the deformation modulus of rock masses using RMR.
Comparison with dilatometer tests.

JOSE M. GALERA,
Geocontrol, S.A. - UPM, Madrid, Spain
M. ALVAREZ,
Geocontrol, S.A., Madrid, Spain
Z. T.BIENIAWSKI,
Bieniawski Design Enterprises, Ari:ona, USA

ABSTRACT: This paper presents the result oI comparisons between the modulus oI deIormation obtained
Irom dilatometer tests and the geomechanical quality oI the rock mass using the RMR classiIication and the
basic intact rock properties such as the uniaxial compressive strength and Young`s modulus.
The Iirst step was to compare the dilatometer modulus with RQD and RMR. Subsequently, it has been de-
cided to use the RMR without considering the lithology, as the diIIerences where Iound insigniIicant.
The second step was to scrutinize the data, excluding those with the Iollowing limitations: Weathering grade
IV and dilatometer modulus 0.5 GPa. Also in those cases in which Em 10 GPa, 15 points were added to
the value oI RMR because an undrained modulus was being considered.
Excluding any data with anomalous ratios, the Iinal database consists oI 436 cases in which known values oI
Em, RMR,
i
c
and E` are considered reliable.
With this database several correlations were investigated to estimate rock mass deIormability improving on
the existing criteria oI Bieniawski (1978), SeraIin-Pereira (1983), Nicholson-Bieniawski (1990) and Hoek
(1995). The results were presented at ISP5 Int. Symp. (Galera et al, 2005). A new relation between RMR and
E
m
/E
i
is recommended:


representing a useIul tool Ior the estimation oI rock mass deIormation modulus.
Finally in the paper, the new relations are proven using data Irom two main civil works.
Also, the ISRM organized a work group Ior the in-
vestigation oI the scale eIIects in rock masses con-
cerning strength, deIormability, joint properties,
permeability, and even in situ stresses. The results oI
these studies were presented in two Workshops, at
Loen (1990) and Lisbon (1993).
In particular, concerning the scale eIIect in rock
mass deIormability, Pinto de Cunha and Muralha
(1990) showed the eIIect oI the volume involved in
the test oI the deIormation modulus measured.
Figure 1 shows this phenomenon, where LAB are
laboratory tests, BHD are Borehole Dilatometer
Tests and LFJ are Large Flat Jack Tests. Two diIIer-
ent ideas can be derived Irom this Iigure. One is that
the bigger the volume involved in the test, the lower
the modulus. The second is that the bigger the vo-
lume, the smaller the variability oI the results.

'Figure 1 DeIormability modulus vs. tested volume (Pinto da
Cunha & Muralha, 1990).
3 EMPIRICAL EVALUATION FROM
GEOMECHANICAL CLASSIFICATIONS
It is clear that in situ methods are the best approach
to predict deIormability oI rock masses. However, in
situ tests are relatively expensive and not always
provide reliable results due to several reasons.
Rock mass deIormation modulus estimation by cor-
relations with geomechanical classiIications ap-
peared as a traditional tool in rock mechanics since
Bieniawski (1978) and his RMR index.
Subsequent correlations have included RQD (Gard-
ner, 1987; Kayabasi et al., 2003; and Zhang and
Einstein, 2004), Q system (Barton, 1983; Grimstad
and Barton, 1993), and RMR (SeraIim and Pereira,
1983; Nicholson and Bieniawski, 1990; and more
recently, Hoek et al., 1995).
Currently, three diIIerent correlations using Q, RQD
and RMR are used:
a) Q and rock mass deformation modulus
Barton (1983) and Grimstad and Barton (1993) pro-
vided a study with several geophysical borehole
measurements obtaining the Iollowing relations:


( ) 5 . 3 V 10 Q
p
= with Vp in km/s, and concluding
LogQ 25 ) GPa ( E =
(1)
although in other projects LogQ 10 E = was Iound
more suitable.
b) RQD and rock mass deformation modulus
Gardner (1987) proposed the Iollowing expression,
i E m
E E = (2)
where
E
0.0231RQD-1.32 (0.15). This method
was used by the AASHTO (American Association oI
State Highway and Transportation OIIicials).
More recently Zhang and Einstein (2004) recom-
mended the Iollowing relations:

E
m
/E
i
0.210
0.0186RQD-1.91
(Lower bound)
E
m
/E
i
1.810
0.0186RQD-1.91
(Upper bound)
E
m
/E
i
10
0.0186RQD-1.91
(Mean)

These expressions are shown in Figure 2. Note the
large scatter.
c) RMR and rock mass deformation modulus
The Iirst correlation between RMR and rock mass
deIormation modulus was proposed by Bieniawski
(1978), as
100 RMR 2 ) GPa ( E
m
= (For RMR50) (3)

'Figure 2. Relationships between RQD and Em/Er (Zhang and
Einstein; 2004).

Later, SeraIim-Pereira (1983) proposed the more
known expression,
( )
40
10 RMR
m
10 ) GPa ( E

= (4)
Figure 3 shows graphically both expressions and
their comparison.


'Figure 3 Correlation between the in-situ modulus oI deIorma-
tion and RMR (Bieniawski, 1989).

Nicholson and Bieniawski (1990) derived the Iol-
lowing relation considering not only RMR but also
the Young`s modulus oI the intact rock E
i
:

+ =
82 . 22
RMR
2
i
m
9 . 0 RMR 0028 . 0
100
1
E
E
(5)
More recently Hoek et al. (1995) suggested a correc-
tion to the SeraIim-Pereira expression, using a Iactor
oI
100
) MPa (
i
c

, and interchanging GSI (Geological

Strength Index) with RMR, as Iollows
( )
40
10 GSI i
c
m
10
100
) MPa (
) GPa ( E

=
(6)

Figure 4 shows graphically this Hoek et al. (1995)
relation.


'Figure 4 Proposed relationships between GSI or RMR with
the intact rock strength (ci) and in situ modulus oI deIorma-
tion Em (Hoek et al., 1995).

Finally, Hoek and Diecrich (2006) suggested the Iol-
lowing equation:

( )
( ) ( )

= =
+ 11 / GSi D 25 75
m
e 1
2 / D 1
000 . 100 MPa E (7)
and also.

( )
( ) ( )

+ +

+ =
11 / GSi D 25 75 e 1
2 / D 1
02 , 0 Ei MPa Em (8)
considering the value oI the intact modulus.

The use oI RMR and not GSI is strongly recom-
mended because GSI introduces more empirism in a
classiIication that itselI is empirical, as was stated in
a recent review by Palmstrm (2003) who warned as
Iollows 'Jisual determination of GSI parameters
represents the return to qualitv descriptions instead
of advancing quantitative input data as in RMR, Q
and RMi svstems. GSI was found mainlv useful for
weaker rock masses with RMR20.
As GSI is used Ior estimating input parameters
(strength), is is only an empirical relation and has
nothing to do with rock engineering classiIication.
4 NEW CORRELATIONS BETWEEN RMR AND
ROCK MASS DEFORMATION MODULUS
4.1 Database
The inIormation presented here is derived partially
Irom bibliography (Bieniawski, 1978; SeraIim-
Pereira, 1983; and Labrie et al. (2004)) but mainly
Irom pressuremeter and dilatometers measurements
made by Geocontrol during the last decade.
The amount oI available data classiIied by its lithol-
ogy is the Iollowing:
- igneous rocks: 270
- metamorphic rocks: 108
- detritic sedimentary rocks: 175
- carbonate sedimentary rocks: 101
- bibliography: 48
This represents 702 data in which the E
m
Irom in situ
tests, RMR and RQD are known.
In 123 oI these data also the uniaxial compressive
strength (
i
c
) and Young`s modulus oI the intact
rock (E
i
) are also known.
Figures 5a, b, c and d show the available data classi-
Iied by the lithology. This classiIication is based on
the ISRM and Goodman lithological classiIications
oI rock masses.


'Figure 5a Igneous rocks 'Figure 5b Metamorphic rocks.













'Figure 5c Sedimentary carbonate rocks 'Figure 5d Sedimentary detritic rocks.
'Figure 5. Database according to the lithology.

Figure 6 shows all the data jointed in the same graph
and it can be observed that the diIIerences due to the
lithology are insigniIicant.
0
10000
20000
30000
40000
50000
60000
70000
80000
0 20 40 60 80 100
RMR
E
m

(
M
P
a
)
ROCAS CARBONATADAS
ROCAS GNEAS
ROCAS DETRTICAS
ROCAS METAMRFICAS

'Figure 6 Relation Em (MPa) vs.RMR according to the
lithologies.

The Iirst objective has been to compare the
pressuremeter and dilatometer results, which
represents the rock mass modulus Em, with RQD
and RMR. In Figure 7 a and b both comparison are
shown.
It is evident that RMR provides a better trend oI the
data, since RQD is only one oI the Iive major com-
ponents oI the RMR classiIication.
This Iigure clearly shows that RMR is more reliable
to estimate the deIormation modulus than RQD
alone by providing a lesser scatter oI data.
4.2 Analvsis of the data
The second step has been to scrutinize the data, ex-
cluding those with the Iollowing limitations:
- Weathering grade bigger or equal than IV.
- Pressuremeter or Dilatometer modulus lesser or
equal than 0.5 GPa.
The reason Ior this Iilter is to remove data with a
'soil behaviour in which the application oI RMR
classiIication is inappropiate as not constituting a
'conventional rock mass.
Also in those cases in which E
m
10 GPa, 15 points
were added to the value oI RMR because a drained
modulus was considered.
Celada et al. (1995) analyzed the relation between
drained and undrained modulus as:
( )
( )( ) n E Kw 2 1 1 2
n E Kw 1 3
E
E
u
+ +
+
= (9)
where Kw is the balk modulus oI the water and n is
the porosity. From this relation the Iollowing can be
concluded:
- II E is bigger than 10 GPa, E
u
/E 1 and no
signiIicant diIIerence exists between both mo-
dulus.
- II E is smaller than 10 GPa and with a drained
Poisson`s ratio oI 0.3, E
u
/E 1.15, so the un-
drained modulus is around 15 higher than
drained modulus.



0
5000
10000
15000
20000
25000
0 20 40 60 80 100
RMR
E
m

(
M
P
a
)
0
5000
10000
15000
20000
25000
0 20 40 60 80 100
RMR
E
m

(
M
P
a
)
0
20000
40000
60000
80000
0 20 40 60 80 100
RMR
E
m

(
M
P
a
)
0
5000
10000
15000
20000
25000
0 20 40 60 80 100
RMR
E
m

(
M
P
a
)
0
10000
20000
30000
40000
50000
60000
70000
80000
0 20 40 60 80 100
RMR
E
m

(
M
P
a
)

Figure 7a Modulus oI deIormation vs. RMR.
0
10000
20000
30000
40000
50000
60000
70000
80000
0 20 40 60 80 100
RQD
E
m

(
M
P
a
)

Figure 7b Modulus oI deIormation vs. RQD.


Finally, the third step has been to perIorm a sensitiv-
ity and quality analysis oI data, using the Iollowing
criteria:
- comparison E
i
vs.
i
c

- comparison E
i
vs. E
m
, and
- comparison E
m
/E
i
vs. RMR.
Excluding any data with anomalous ratios, the Iinal
database consists oI:
- 427 cases in which E
m
and RMR are considered
reliable.
- 98 cases in which E
m
, E
i
,
i
c
and RMR are con-
sidered reliable.
4.3 Discussion
With these data, several correlations have been in-
vestigated to estimate rock mass deIormability by
improving on the existing relationships described in
section 3.
Experience shows that with the current correlation
usually the deIormation modulus Em estimated is
higher than the modulus measured by means oI bo-
rehole expansion tests such as pressuremeters and
dilatometers.
Two new diIIerent types oI relations are proposed:
- without considering Ei values
- including E
i
values
In the Iirst case also the values oI
i
c
are in-
cluded using this expression:
( ) 24 / 100 RMR i
c m
e

=
(Kalamaras and Bieniawski,
1995) (10)
In all the cases the coeIIicient oI regression r
2
has
been calculated as Iollows:
( )
( )

=
2
i
i i 2
y y
' y y
1 r (11)
where y
i
is the value oI E
m
, y is the mean and y
i

are the adjusted values.
a) New correlations between RMR and rock mass
deformation modulus
Figure 8 shows all the new correlations considered
and also the SeraIim-Pereira expression.
i. considering
( ) 24 / 100 RMR i
c m
e

= it is de-
rived:
RMR 202 . 0 28 . 147 ) GPa ( E
i
c
m
m

=
(12)
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50 60 70 80 90 100
RMR
E
m

(
G
P
a
)
697 , 0
2
R
40
) 10 (
10
=

=
RMR
Em
742 , 0
2
R
18
) 10 (
=

=
RMR
e Em
765 , 0
2
R
24
) 100 (
202 , 0 280 , 147
=

= RMR e Em
RMR
800 , 0
2
R
) 50 (
2
) 50 (
) 50 ( 015 , 0 ) 50 ( 056 , 1 0876 , 0
0876 , 0
=
>
<
+ + =
=
RMR
RMR
RMR RMR RMR Em
RMR Em
N puntos: 427

'Figure 8. New correlations between RMR and rock mass de-
Iormation modulus E
m
.

The coeIIicient oI regression, r
2
, obtained is 0.765,
that is higher than the one obtained in the regression
oI the data Iollowing SeraIim-Pereira, namely, a r
2

0.697.
ii. The second relation is an improvement oI Se-
raIim-Pereira, as Iollows:
( ) 18 / 10 RMR
m
e E

= (13)
The coeIIicient oI regression r
2
0.742, that im-
proves by 10 the estimation oI E
m
.




iii. Finally, Iollowing the original estimation, a
threshold oI RMR 50 is derived:
50 RMR for RMR 0876 . 0 ) GPa ( E
m
= (14)
2
m
) 50 RMR ( 015 . 0 ) 50 RMR ( 056 . 1 RMR 0876 . 0 ) GPa ( E + + =
50 RMR for > (15)
This above correlation gives a coeIIicient oI regres-
sion r
2
0.8, that improves by more than 15 the
estimation oI SeraIim-Pereira.
b) New correlation between RMR and rock mass
deformation modulus including Ei
Figure 9 shows the relation
( ) 36 / 100 RMR
i m
e E E

= .
The coeIIicient oI regression, r
2
, is 0.656 which is
smaller than that given by the previous correlations
but it makes a more reliable estimation as Ei is con-
sidered and improves by almost 40 the estimation
due to Nicholson and Bieniawski (1990) which gives
r
2
0.472.
0,000
0,100
0,200
0,300
0,400
0,500
0,600
0,700
0,800
0,900
1,000
0 10 20 30 40 50 60 70 80 90 100
RMR
E
m
/
E
i

(
G
P
a
)
656 , 0
2
R
36
) 100 (
=

=
RMR
e Ei Em
N puntos: 98

'Figure 9 Correlation between RMR and rock mass deIorma-
tion modulus ratio including E
i
.
5 SUMMARY AND CONCLUSIONS
(1) The Borehole Expansion Tests, mostly Flexible
dilatometers, were Iound to be the best in situ test
Ior the determination oI the rock mass deIormation
modulus.
(2) The empirical Em RMR correlations present a
smaller scatter than the previous correlations Em
RQD.
(3) Several empirical correlations have been studied
to estimate rock mass deIormation modulus Em.
Most oI them provide an overestimation oI the value
Em.
(4) Considering 427 data collected Irom the pub-
lished literature and our own data, the best coeIIi-
cient oI regression is obtained considering a thre-
shold oI RMR 50. A linear regression is suggested
Ior values smaller or equal to 50, while a polynomial
expression is recommended Ior values oI RMR big-
ger than 50.
(5) A new relation between RMR and Em/Ei is rec-
ommended, considering 98 data. This expression is
( ) 36 / 100 RMR
i m
e E E

= (16)
representing a useIul tool Ior estimation oI the rock
mass deIormation modulus.
Considering that rock mass strength
( ) 24 / 100 RMR i
c m
e

=
and equation (14), it results in
the Iollowing expression:
3
2
i
c
m
i
m
E
E

= (17)
providing another useIul relationship Ior rock mass
characterization.
(6) Nevertheless, the presented correlations should
be used realizing that some Iactors are ignored such
as directional eIIect oI jointing.
6 BIBLIOGRAPHY
Barton N (1983) Application oI Q-system, index tests to esti-
mate shear strength and deIormability oI rock masses. Pro-
ceedings oI the International Symposium on Engineering
Geology and Underground Construction, vol. I (II), Lisbon,
Portugal, 1983, 51-70.
Bieniawski Z.T. (1978) Determining rock mass deIormability:
experience Irom cases histories. Int J. Rock Mech and Min.
Sc., vol. 15, 237-247.
Bieniawski Z.T. (1989) Engineering rock mass classiIication: a
complete manual Ior engineers and geologists in mining,
civil and petroleum engineering. John Wiley & Sons, 251
pages.
Celada B., Galera J.M., Varona P. (1995) Development oI a
new calibration and interpretation procedure oI pressureme-
ter test to obtain elastic parameter. The pressuremeter and
its new avenues. Ed Balkema, 265-272.
Celada B. Galera J.M., Varona P, Rodriguez A. (1995) Deter-
minacion del modulo de elasticidad de Iormaciones arcillo-
sas proIundas. Enresa. Publicacion Tecnica num. 01/95.
Celada B., Galera J.M., Rodriguez A. (1996) Ensayos de de-
Iormabilidad para la caracterizacion del terreno in situ. In-
geopres, n 36, 22-27.
Gardner W.S. (1987) Design oI drilled piers in the Atlantic
Piedmont. Smith RE, editor. Foundations and excavations
in decomposed rock oI the piedmont province, GSP ASCE,
no. 9, 1987, 62-86.
Grimstad E., Barton N. (1993) Updating the Q-system Ior
NMT. Proc. Int. Symp. Sprayed Concrete, Fagernes, 21
pages.
Kalamaras G.S. and Bieniawski Z.T. (1995) A rock mass
strength concept. ISRM International Congress oI Rock
Mechanics. Tokyo, Japan.
Kayabasi A, Gokceouglu C, Ercanoglu M (2003) Estimating
the deIormation modulus oI rock masses: a comparative
study. Int. J. Rock Mech. Min. Sci., 2003, 40:55-63.
Hoek E, Kaiser P.K., Bawden W.F. (1995) Support oI under-
ground Excavations in Hard Rock. Ed. Balkema, 215 pages.
Hoek E, Diederichs, M.S. (2006) Empirical estimation oI rock
mass modulus. Int. Journal oI Rock Mechanics & Mining
Sciences, vol.43, 203-215.

Labrie D. et al. (2004) Measurement oI in situ deIormability in
hard rock. Proceedings ISC-2 on Geotechnical and Geo-
physical Site Characterization, Porto, Portugal. Ed. Mil-
press, 963-970.
Nicholson G. and Bieniawski Z.T. (1990) A non-linear deIor-
mation modulus based on rock mass classiIication. Int. J.
Mining & Geol. Eng., vol. 8, 181-202.
Palmstrm A., Singh R. (2001) The deIormation modulus oI
rock masses Comparison oI in situ tests and indirect esti-
mates. Tunnelling and Underground Space Tech. Vol. 16,
115-131.
Palmstrm A., Stille H. (2003) ClassiIication as a tool in Rock
Engineering. Tunnelling and Underground Space Tech.,
vol. 18, no. 4, 331-345.
SeraIim J.L., Pereira J.P. (1983) Considerations on the geome-
chanical classiIication oI Bieniawski. Proc. Int. Symp on
Eng. Geol. and Underground Construction, vol. I (II), Lis-
bon, Portugal, 1983, 33-44.
Zhang L., Einstein H.H. (2004) Using RQD to estimate the de-
Iormation modulus oI rock masses. Int. Journal oI Rock
Mechanics & Mining Sciences, vol. 41, 337-341.