You are on page 1of 15

International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics and Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Determination of deformation modulus in a weak rock mass by using T


menard pressuremeter
İbrahim Ferid Öge
Department of Mining Engineering, Faculty of Engineering, Muğla Sıtkı Koçman University, Turkey

ARTICLE INFO ABSTRACT

Keywords: The deformation modulus of heavily jointed andesitic rock mass was investigated by using 41 Menard pres-
Rock mass deformation suremeter test results and geotechnical information obtained from seven geotechnical boreholes. The geo-
Deformation modulus technical borehole logs and laboratory test results provided by a mining company were used for the assessment
Nonlinear regression of rock mass characterization. The log and the test results were taken into account in the rock mass classification
Pressuremeter test
work. The well-known empirical equations were employed to predict the deformation modulus of the rock mass.
In-situ testing of rock
Discrete fracture network
The predictions and the pressuremeter test results were compared and their performance indicators were pre-
sented. Non-linear multiple regression methods were used for predictive modelling of the deformation modulus
of the rock mass by considering the available data. Rock Quality Designation (RQD), discontinuity condition
rating (Dc) of the Rock Mass Rating (RMR) system and uniaxial compressive strength of the intact rock (σci) were
taken into consideration as input parameters in four new prediction equations to be used for determination of
deformation modulus. One more equation was proposed by the addition of the depth as a predictive parameter.
The influence of the depth and vertical to horizontal stress ratio (k) on the deformation modulus was pre-
liminarily examined by using finite element modelling. Initially, the rock mass was assumed to be an isotropic
elastic-brittle-plastic medium. Later on, the rock mass was modelled as a discontinuum by imposing a discrete
fracture network (DFN). Keeping all of the mechanical properties constant, different depth and k parameters
were applied to the pressuremeter models. No influence of the depth and k on the deformation modulus was
observed for the isotropic medium while the depth contributed to an increase in the modulus in the anisotropic
discontinuum analysis. The numerical modelling findings constituted a basis for the inclusion of the depth
parameter into the new predictions.

1. Introduction several researchers questioned the reliability of the test findings and
commented on the high cost and labour intensiveness of the field
A clear understanding of the deformability properties of a rock mass tests.4,5 Relying on a single test method is considered to be misleading
is essential for an underground opening or rock surface structure de- since a considerable variation can be observed in a fairly uniform and
sign. The rock mass deformation modulus and the strength parameters good quality rock mass.6,7 Qualification of the testing personnel, the
are required as inputs to analyse the rock mass behaviour by using calibration and sensitivity of the measurement equipment, the delicacy
numerical models or other calculation methods. Still, the determination of the test setup and pre-test preparations can be accounted for other
of the global mechanical properties of a jointed rock mass remains one factors governing the reliability of the test results. During the inter-
of the most difficult tasks in the field of rock mechanics.1 The rock mass pretation of the in-situ test results, some simplifications can also be
deformability differentiates from the intact rock or laboratory scale made to reduce the number of unknown elastic constants such as
behaviour due to the embodiment of the discontinuities in the mass and treating an anisotropic rock mass as it is an isotropic medium.8
the groundwater influence. The difference is expected to be obvious Due to the abovementioned difficulties and discrepancies in field
when a high degree of fracturing or jointing is present in the rock testing, empirical approaches can also be used for the estimation of the
mass.2 Besides, the engineers generally pay more attention to assess the necessary deformability design parameters as well as taking the em-
strength properties than investigating the deformability parameters pirical estimates into account for the comparison or the verification of
which in turn lead to unpleasant consequences.3 Although field testing the field test results.9,10 Rock mass classification systems such as, the
is an opportunity to assess the rock mass deformability properties, Rock Quality Designation (RQD), Rock Mass Rating (RMR), Q-system

E-mail address: feridoge@mu.edu.tr.

https://doi.org/10.1016/j.ijrmms.2018.10.009

1365-1609/ © 2018 Elsevier Ltd. All rights reserved.


İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

(Q) and Geological Strength Index (GSI), constitute a basis for the gradually and corresponding volumetric increments are recorded
empirical estimation of deformation modulus.11–14 In case the empirical during the test. At the beginning of the volume and stress relationship,
approaches are used for the estimation of the modulus, a comparison the tightening and interlocking of the relaxed zone due to the drilling
between several approaches satisfy an increased reliability of the esti- and elastic reloading of the unloaded borehole walls can be observed.
mates. However, the empirical approaches may exhibit a great varia- After the interlocking of the rock mass is ensured during the initial
tion compared to each other. For underground structures, detailed phase of the pressure application, the pseudo-elastic zone is attained
ahead and behind the tunnel face measurements and tunnel con- and the deformation modulus is calculated along this ideally linear
vergence provide an opportunity to back-calculate the deformation phase. Following the pseudo-elastic zone, the plastic phase is reached.
modulus.15,16 While discussing the importance of field testing and This behaviour is discussed by using a numerical modelling technique
empirical estimation of the deformation modulus, it must be noticed in Section 3 including some visuals. The interpretation of the de-
that the rock mass is generally assumed to be an isotropic medium formation modulus is carried out by employing the solutions for the
which may not be necessarily true for all rock masses.17 Agharazi infinitely long cylindrical cavity in an elastic or elastic-perfectly plastic
et al.18 investigated the deformation mechanism of jointed rock based medium.21 The test scheme can be arranged in a way to measure the
on plate loading tests and concluded that the major joint set orientation loading or unloading modulus but then, the unloading curve enables
had a substantial influence on the deformation. The interpretation of the engineer to examine the elastic displacements only. The deforma-
structurally controlled deformability test results was suggested to be tion modulus is determined by taking the loading curve into account
undertaken by some detailed numerical analysis. Plate jacking and hence, both the permanent and elastic displacements are included in
large flatjack test results were compared to each other in another the interpretation.25 The typical equation used in the interpretation of
study.19 The two methods suffered from different factors such as the the deformation modulus is given:
blast damage zone or the boundary conditions of the test volume,
Ep = Erm = (1 + ) ×2V×( P/ V) (1)
however, reliable measurements were acquired.
In this study, in order to determine design parameters for founda- where Ep stands for a deformation modulus measured by the pres-
tion construction of a mine shaft headframe, seven boreholes were suremeter where the Poisson's ratio is assumed to be 0.33. It can be
drilled from which forty-one Menard pressuremeter test results were regarded as deformation modulus of the rock mass, Erm. V is the sum of
obtained. The andesitic, heavily fractured rock mass was also in- the initial volume of the measuring cell and the average additional
vestigated by preparation and evaluation of the borehole logs and la- volume change of the probe, and ΔV and ΔP are the change of volume
boratory test results. The available data were compared with sixteen and the change in pressure during the pseudoelastic phase, respec-
pre-existing empirical equations. Some equations were also proposed tively.26 Another concern for the test is the borehole wall disturbance
using the parameters RQD, intact rock strength and some discontinuity which is more pronounced for the soil formations. An appropriate
ratings together with taking the depth into account. drilling technique must be employed not to cause an excessive dis-
turbance as well as establishing a borehole diameter compatible with
2. Background the Menard pressuremeter probe. The impact of drilling disturbance on
the measurement results was discussed in Isik et al.27 In this study, the
The technical terms namely; rock mass modulus, rock mass elastic rock mass consists of a rock material having a variable strength which is
modulus (tangent modulus of elasticity), deformation modulus (secant mostly greater than 10 MPa and the mass is heavily jointed. Considering
modulus), re-loading, unloading (recovery) modulus parameters are the rock mass conditions, diamond drilling with triple core barrels was
used for rock mass deformability which can be derived from a stress- decided to be an appropriate option to ensure low disturbance.
strain curve obtained from a test result.17,20 Depending on the testing
method and the rock engineering problem, the appropriate modulus 2.2. Empirical estimation of deformation modulus
can be used for the interpretation of the test result. Among them, the
deformation modulus is accepted as a representative parameter gov- In case the field testing is unavailable or for the verification of the
erning the mechanical behaviour of the rock masses9 and it covers both field test results, the empirical equations are utilized for the estimation
recoverable and irrecoverable behaviour. Poisson's ratio is another of the rock mass deformability. The researchers proposed relationships
elastic parameter. It is basically the ratio of elastic lateral strain to axial between the measured rock mass modulus and the rock mass classifi-
strain which is out of the scope of this study. cation methods such as RQD, RMR, Q, GSI, Rock Mass Index (RMi),
Rock Mass Quality Rating (RMQR).4,28 Together with the rock mass
2.1. In-situ testing of deformation modulus classification ratings, the laboratory scale uniaxial compressive strength
(σci) and the modulus of elasticity (Ei) are included as input parameters
Back analysis of field deformation measurements by the employ- in several empirical estimations. The empirical approaches may exhibit
ment of a rigorous numerical analysis is a reliable source to assess the variable results even for the same rock mass. Even though the same
rock mass deformability parameters in case of its availability. The plate database is taken into account, different fitting functions were used to
loading, plate jacking, Goodman jack,5,8 dilatometer and pressuremeter propose different empirical equations by several researchers.9 Other
tests21,22 can be accounted for the field tests aiming to measure the reasons for the variation among the empirical equations can be attrib-
deformation modulus of the rock masses, especially for the poor ones. uted to the database of each independent study, the employed testing
Borehole expansion tests, such as dilatometer and pressuremeter was method, biased estimation of the rock mass classification rating and/or
promoted by Galera et al.22 to be used in the rock masses. Menard the geological nature of the rock mass of concern. The empirical
pressuremeter is a device for testing the rock mass deformability which equations used in this research for comparison purpose are given in
is carried out inside the borehole.23,24 The testing equipment has been Table 1.
improved in terms of several technological improvements such as the The empirical equations which accept RMR, Q or GSI as an input
material quality, increased sensitivity of the pressure cells and mea- value were selected due to the availability of the ratings, here. The
surement gauges but the main working principle of it remained the majority of the equations were proposed by using high-quality well-
same. Menard type pressuremeter used in this study consisting of a publicized in-situ data from Serafim and Pereira,29 Bieniawski40 and
probe with three inflatable cells (one measuring, and two guarding Stephens and Banks.41 Most of the equations listed in Table 1 cover all
cells) with a high-pressure setup to accommodate the rock mass stress- rock mass conditions, however, the one proposed by Serafim and Per-
strain relationship. Initially, the pressuremeter probe is lowered to the eira29 is valid if RMR < 50. σci must be smaller than 100 MPa to use the
testing depth in a previously drilled borehole. The pressure is increased equation by Hoek et al.35 D parameter is related to the blast or

239
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Table 1
The empirical equations used for comparison purposes in this study.
Proposed by Empirical equation Proposed by Empirical equation

29 36
Serafim and Pereira RMR 10 ) Gökceoglu et al. Erm = 0.0736e(0.0755RMR) (GPa)
Erm = 10( 40 (GPa)
Nicholson and Bieniawski30 Erm
= 0.009e RMR /22.82+0.000028RMR2 Gökceoglu et al.36 Erm = 0.1451e(0.0654GSI) (GPa)
Ei
31 37
Barton Erm = 10Q(1/3) (GPa) Zhang and Einstein Erm
= 10 (0.0186RQD 1.91)
Ei
Jasarevic and Kovacevic32 Erm = e(4.407 + 0:081 × RMR) (GPa) Sonmez et al.38 Erm
= (s a) 0.4
Ei
Aydan et al.33 Erm = 0.0097RMR(3.54) Galera22 Erm
= e(RMR 100)/36
Ei
34 39
Read el at. Sonmez et al.
Erm = 0.1 ( ) (GPa)
RMR 3
10
Erm
Ei
= 10
(RMR 100)(100 RMR)
4000
exp ( RMR
100 )
35 17
Hoek et al. Simp. Hoek and Diederichs
( ) 10( )
D ci GSI 10
Erm = 1 40 (GPa) 1 D/2
2 100 Erm = 100 (GPa)
1+e ( 75+ 25D11 GSI
)
Barton13 Hoek and Diederichs17
( )
1/3
ci
Erm = 10 Q (GPa) Erm 1 D/2
100 = 0.02+
Ei
1+e ( 60+15D11 GSI
)
(D: Disturbance factor17, σci: intact rock uniaxial compressive strength, Ei: intact rock modulus of elasticity, s and a are Generalised Hoek brown Parameters.

excavation damage which is equal to zero for this study. 5 of the still, the parameter E is the same. If a rock mass is of concern, the de-
equations are exponential equation while 3 of them are the functions in finition of the problem should be: Structurally the same rock mass
the form of the power of 10. The equation proposed by Zhang and having the same discontinuity pattern, orientation, mechanical prop-
Einstein37 gives a lower bound estimate with the input parameters of erties with constant intact rock mechanical properties, is subjected to
RQD and Ei. Mostly, the well-known equations are chosen for the deformability testing at different depth intervals. Does confining pres-
comparison purposes. Additionally, during the preparation stage of the sure stemmed from the depth have an effect on rock mass deformation
study, the equations proposed by Jasarevic and Kovacevic,32 Gökceoglu modulus? It means the orientation, amount, the stiffness-strength
et al.,36 Zhang and Einstein37 was observed to give comparable results function of the discontinuities, the rock material strength and elastic
with the field test results available for this study and they were included modulus are kept constant and subjected to deformability testing at
in the research. different depths. The strength of the rock mass is expected to be in-
creased in parallel with the increasing confining pressure (excluding the
3. A preliminary analysis of the influence of the depth on critical state). However, the strength function remains constant (i.e.
pressuremeter measurements when Mohr-Coulomb failure envelope is taken into account for the
problem, cohesion and the friction angle are constant). From a similar
Verman et al.42 relying on their correlations, concluded that the point of view, the deformation magnitude is affected by the confining
deformation modulus of the rock mass increases in parallel with the pressure but the parameters E and ν remain constant. In the reality,
depth increase. They added that the stress dependency of the de- when the same rock mass is considered for a shallow and deep case
formation modulus due to the increasing confining pressure is more there will be significant differences. The discontinuities will be more
likely to be pronounced in weaker rock masses and almost absent in tightly interlocked for a deep case and loose for a shallow case. If clay
strong, brittle rock masses. Isik et al.27 concluded their study that some infill or bands are present, a high consolidation and compaction are
degree of confining pressure dependency can be expected on the de- expected at greater depth while less degree of consolidation and com-
formation modulus. They discussed that the Hoek-Brown failure cri- paction (thus a greater void ratio) can be observed at shallow depth.
terion takes confining pressure into account. It means, the strength is Then, it is possible to say the deformation modulus or modulus of
confining pressure sensitive, which is a generally accepted truth. elasticity of the rock mass can exhibit an increasing trend with the
However, the failure criteria parameters still remain the same. They depth. If this case is accepted to be true, then the mechanical properties
also suggested additional investigations to be carried out for the de- of the discontinuities should have better properties which in turn led to
termination of the confining stress dependency of the deformation an increment of the deformation modulus. In terms of rock mass clas-
modulus of heavily fractured rock masses and for the comparison of sification systems, it is correct to say the parameters RMR and Q of the
pressuremeter tests with other in-situ tests. Palmström and Singh5 rock mass can increase at a greater depth which in turn can exhibit
pointed out that the in situ deformation modulus is not constant, but greater deformation modulus. Taking the abovementioned explanations
depends on the stress conditions, being generally higher in areas sub- into consideration, confinement dependency of the deformation mod-
jected to high rock stresses than in rock masses under low stresses. ulus cannot be expressed if all mechanical properties belonging to a
Additionally, they explained that the stress dependency may also be specific rock mass are constant at varied depth.
true due to a better rock mass quality where a higher stress is present. In It is commonly asserted that the findings in the literature are in-
Agan's43 study, the relationship between the deformation modulus and sufficient about the effect of confining pressure (or depth) on the de-
the depth was investigated. The study was carried out in a heavily formation modulus by the previously mentioned authors. In order to
disturbed chert rock mass by using Menard Pressuremeter at a depth analyse the problem, it is necessary to find a rock mass having the same
range of 5–30 m. physical, mechanical and structural quality and properties located at
Considering the basic principles of rock mechanics, in an isotropic different depth intervals, however, it may not be always possible.
elastic medium, the confining stress affects the deformation amount Hence, the finite element modelling can be employed to preliminarily
and the strength. The effect of confining pressure on the strength can be investigate the effect of confining pressure on the parameter Erm.
basically explained by Mohr-Coulomb failure criteria and the elastic Initially, the rock mass was modelled to be an isotropic elastic-
deformation is governed by the modulus of elasticity (E) and Poisson's brittle-plastic material. Later on, Discrete Fracture Networks (DFN), a
ratio ( ) according to the Hooke's law. The deformation is affected by joint network modelling technique was employed.44 In DFN analyses,
the confining pressure but this effect is attributed to the Poisson's ratio, the rock mass can be modelled as an anisotropic discontinuum.

240
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

3.1. Isotropic elastic-brittle-plastic rock mass assumption similar to the face effect of a tunnel, should have a supporting effect
which in turn led to a greater measurement of the modulus. The
Axisymmetry is widely used to simulate three-dimensional problem modelling work covered the pseudo-elastic and plastic region which
by using a two-dimensional model. A similar model was used by Isik starts at the origin of the plot while the membrane expansion region
et al.27 to investigate the size of the testing volume of the field tests and was not included in the modelling but the typical complete path was
the consequences of using different testing length to borehole diameter given in Fig. 1.
ratios for a Menard pressuremeter. A 20 m deep borehole having a In the isotropic elastic-brittle-plastic rock mass, it was found that
diameter of 69 mm was modelled similarly to the actual testing borings the modulus of deformation is insensitive to the confining pressure. The
in this research. Three different rock masses were considered in iso- stress level at which the plastic region begins (plastic limit) is depen-
tropic rock mass assumption. Each of them was modelled with two dent on the confining pressure. When the strength parameters (of Gen.
different vertical to horizontal stress ratio, k = 1 and 3. Principal hor- Hoek–Brown failure criterion) like σci, mb, s and a are constant, in-
izontal stresses were assumed to be equal to the each other. The rock creasing depth (confining) influences the extent of the pseudo-elastic
mass A was modelled with Erm = 88.7 MPa (GSIpeak = GSIres = 14, σci curve length and plasticity initiation but not the parameters itself.
= 12.85 MPa, mi = 25), for rock mass B: Erm = 542 MPa (GSIpeak = 35, The modulus varies depending on the stress level along the pre-
GSIres = 27, σci = 20.97 MPa, mi = 25) and for rock mass C: Erm viously given P-V curve in Fig. 1. In fact, the deformation modulus is
= 1811 MPa (GSIpeak = 34, GSIres = 27, σci = 68.4 MPa, mi = 25). The typically calculated along the pseudo-elastic zone as given in Fig. 1
strength and Erm parameters are chosen based on 3 different test loca- which is expected to be linear wise. However, the identification of the
tions from the study. 19–20 m deep pressuremeter testing was modelled pseudo-elastic zone can be difficult and not clearly distinguishable. In
in this study but it may not always be possible to carry out a pres- Fig. 2, the same findings were interpreted in an alternative way by
suremeter test at a depth of 100 m however, the rock mass B was also using the simulated pressuremeter data. Tangent modulus was calcu-
simulated to be located at a depth of 100 m with two different k values. lated for gradually increasing pressure intervals and the internal pres-
Throughout the modelling stages, the field stresses were initialized sure and corresponding tangent modulus values are presented in Fig. 2.
then, the material is removed from the borehole (drilling stage). Later The vertical segments of the curves in Fig. 2 represent the Erm values
on, an internal pressure was gradually increased similar to a typical which are also called as a pseudo-elastic region. The extent of the
pressuremeter test. The displacements corresponding to the pressure pseudo-elastic region depends on the confining stress but the parameter
stages were recorded and the volume change in the borehole test in- Erm is not. For the rock mass with Erm = 542 MPa, two different k values
terval was calculated. The pressure (P) versus volumetric increase (V) and two different depths (20 and 100 m) were modelled and the Erm
plots are given in Fig. 1 and a P-V plot is a typical raw output of the remained constant.
Menard Pressuremeter test. The volume of fluid injected into the
pressuremeter cell was calculated by taking the radial borehole de- 3.2. Anisotropic and discontinuous rock mass assumption
formation acquired by the numerical modelling. The P-V plots enable
the calculation of the deformation modulus and Eq. (1), which is pro- Knowing that the rock mass cannot always be represented by an
vided in Section 2.1 can be used. Actually, the volume increment is isotropic and elastic-brittle-plastic material model, a fractured rock
measured by using flowmeters during a test. mass model was also constructed. The jointed rock mass was modelled
In Fig. 1, the input Erm parameters, k ratios and depths are given in by employing DFN technique which introduces a discontinuity pattern
the legend. In the plot, Erm parameters calculated by using the typical and integrates it with finite element mesh.44 The intact material
pressuremeter. As an observation, calculated/true Erm ratio was found properties and the joint properties can be imposed to the model which
to be 1.07. The difference can be attributed to the mesh geometry and can be regarded as a discontinuum. Initially, Barton-Bandis45 failure
the point where the radial displacement was acquired. The model was criterion was used to predict discontinuity strength. Barton Bandis
constructed in the form of a borehole drilled to a depth of 20 m and the parameters, namely; JRC, JCS, and φr were selected as 5, 15 MPa, and
test was carried out at the bottom of the hole. The borehole bottom, 30°, respectively. The equivalent Mohr-Coulomb parameters, cohesion

Fig. 1. Pressuremeter results obtained from the continuum numerical modelling.

241
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 2. Modulus variation with respect to the applied pressure for continuum modelling.

was calculated as 214 kPa and friction angle was 30.1°. Several mod- depth and k values and it is presented in Fig. 4.
elling trials were carried out and after the calibration of the models, The comparison of pressuremeter curves for 20 and 100 m deep
basic discontinuity properties were introduced as normal stiffness: models with a varied k ratio is presented in Fig. 4. The necessary in-
100 GPa/m, Shear stiffness: 10 GPa/m, cohesion: 100 kPa, and friction ternal pressure to reach a particular volumetric expansion is increased
angle: 35° for peak strength. The residual cohesion: 30 kPa and friction by the increasing confining pressure for the plastic region. The tan-
angle: 30° were introduced to the model. The discontinuity behaves as gential modulus throughout the pressure increments are plotted and the
an elastic-brittle-plastic material due to being an implicit finite element parameter Erm can be identified precisely (Fig. 5).
model with DFN. For different values of k, the models resulted in the same Erm value
Gen. Hoek-Brown parameters for intact rock are σci = 20.97 MPa mi for a particular depth. When the depth is increased, the parameter Erm
= mb = 25 s = 1 and a = 0.5. The intact rock was modelled to have a increased, too. According to the modelling results, the parameter Erm
GSI = 100 at peak and GSI = 25 at residual state then the strength depends on the vertical stress magnitude, not the horizontal stress. The
parameters for the residual state are mb,res = 1.717, sres = 0.00024, and confining pressure has an influence on the strength and deformation
ares = 0.531. The discontinuity parameters and constructed pattern rate in plastic stage. If the confining pressure is greater, the deformation
were aimed to represent a rock mass quality having a GSI value around amount at a particular internal stress level is smaller.
45. Based on the given parameters, Erm is estimated to be 1550 MPa Some of the modelling results were not plotted in Fig. 4 and Fig. 5
according to17 however, Erm is not used as an input parameter. The such as the 100 m deep testing case with k = 3 resulted in a measured
problem was modelled in plain-strain as the DFN cannot be imposed to Erm of ~477 MPa which is considerably smaller than the typical results
an axisymmetric model. as it was shown in Fig. 4. Similarly, 200 m deep testing with k = 1
The Baecher joint network model46 was chosen to construct the which resulted in Erm= 1820 MPa. Even though the depth is increased,
joint pattern which is a flexible algorithm that can generate intricate a slight decrease in Erm was observed compared to the 100 m deep k = 1
joint networks. In this model, joints are assumed to have finite trace case. For k = 2 and 200 m deep case, the result was Erm = 915 MPa.
lengths, which follow some statistical distribution. 1.45 m wide model This unexpected trend was observed due to the plastic zone develop-
boundary, 69 mm borehole, and the joint pattern are presented in ment around the borehole perimeter before the application of internal
Fig. 3(a). A stereographic projection of the discontinuities in the study pressure (after the borehole drilling). Even though the plastic zone
area is provided in Fig. 3(b). The two dominant joint sets have dip and around the borehole was thin, it caused a greater deformation due to
dip direction of 75°/215° ( ± 5°) and 48°/358° ( ± 10°). The acute angle the plasticity and a low Erm was calculated. Another model was con-
between the strikes of the joint sets is varying between 30° and 40°. structed at a depth of 4 m with k = 1 and k = 3. The Erm values were
Some exemplary modelling results are also presented in Fig. 3(c) and 1824 MPa for the former and 1758 MPa for the latter. The case at a
Fig. 3(d). In the former one total displacement contour with displace- depth of 4 m resulted in similar results with 20 m deep case and this
ment trajectories and in the latter one major principal stress contour are case was not reported in Figs. 4 and 5.
shown. Considering the modelled discontinuity pattern, the rock mass is
The discontinuity orientation and density imposed to the model heavily fractured and expected not to sustain great stress differentials
were impressed from the visual impression of the actual rock mass thus, the plots in Figs. 4 and 5 present only four cases, up to a maximum
(Section 4.2) existing in the study area even though the model is not an depth of 100 m. The numerical analyses presented here are constructed
exact representation of the rock mass. The aim here is to investigate the by simplifying the problem as less as possible. The major discrepancy
effect of confining pressure on the pressuremeter test results in a sim- can be the stiffness function for the discontinuities which is chosen to
plified manner. As it can be inferred from Fig. 3(c,d), the rock mass be linear in this modelling work. The failure criterion for the dis-
discontinuity pattern in the model leads to an anisotropic behaviour. continuities is also linear. Rigorous approaches and modelling techni-
However, a pressuremeter is not capable of measuring the deformation ques can also be employed for the discontinuum analysis. A wide range
modulus in different axes. Hence, it can only provide an overall value of discontinuity patterns, rigorous constitutive models and failure cri-
for the rock mass. The P-V plot was constructed for two different testing teria for the discontinuities and rock material can be constructed and

242
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 3. (a) Modelled discontinuity pattern and the borehole, (b) stereographic projection of the discontinuities in the study area, (c) a sample result showing total
displacement contour, slipped joints (with red) and displacement trajectories, (d) major principal stress contour (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article).

Fig. 4. Pressuremeter results obtained from the discontinuum numerical modelling.

243
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 5. Modulus variation with respect to the applied pressure for discontinuum modelling.

Fig. 6. (a) Yielded discontinuities at a depth of 20 m and (b) at a depth of 100 m at 1 MPa internal pressure.

the rock mass can be subjected to a varied range of confining pressures. In the continuum analyses, Erm was introduced to the models and
According to the analyses which were carried out by employing the the depth had no influence on the deformation modulus measurement.
DFM technique with anisotropy, the depth has an influence on Erm. This In the analyses with DFN, the discontinuities were introduced to the
effect can be quantified as a ratio, ΔErm/ΔH = 1.6 MPa/m for a depth model and the rock material between the discontinuities is still a finite
range from 20 m to 100 m. One must remember that it is a case-specific continuum. Then, the discontinuities remain as an influencing factor for
value and cannot be generalised. The discontinuity pattern and density, the deformation modulus. The findings are compatible with Verman
mechanical properties were kept constant while the depth and vertical et al.42 as the researchers suggested that the influence of depth is more
to horizontal stress ratio were changed. When the pressuremeter test pronounced for fractured rock masses. Relying on the findings pre-
was modelled and the obtained deformations were interpreted to find sented above, the depth is chosen to be a predictive parameter in
Erm, an increased modulus was observed in discontinuous media, unlike Section 5.
the isotropic, elastic-brittle-plastic rock mass assumption. Conversely,
the isotropic rock mass assumption concluded to be insensitive to the 4. Characteristics of the rock mass and measurement results
depth or confining pressure. In the model runs with DFM technique, the
discontinuities were subjected to greater normal stress due to the in- The study area is located at Soma Coal Basin in Manisa and İzmir,
creasing depth. The observed modulus increment can be attributed to Turkey. A state-owned open cast mine is under operation in the
the increment in the normal stress acting on the discontinuities. In the northern region of the Soma Coal Basin where the coal seam lies at a
deep cases, the discontinuities have less tendency to yield under the shallow depth. In the vicinity of the study area, underground coal mines
same pressure applied inside the borehole or at least the yielded dis- are under operation at a depth range of 150–400 m.47,48 New under-
continuity length is shorter. This finding governs the variation of the ground coal mines owned by the government and private companies
modulus with respect to the depth and it is illustrated in Fig. 6 and a with greater mining depth from 700 m to 1200 m are in planning or
comparison of yielded discontinuity lengths in 20 m and 100 m deep development stages and at an approximate distance of 5 km from the
cases under the same internal pressure (1 MPa) are shown. In Fig. 6(a), operating mines.49 The study area is located within the boundary of
the confining pressure is low and yielded discontinuities are highlighted new mining licenses dominated by the Soma formation.
in meters. In Fig. 6(b), the yielded discontinuities are very limited and In order to assess the foundation design parameters of a service shaft
only a few cm long yielded section is visible. headframe, sixty pressuremeter tests were carried out in seven

244
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

size due to the closely spaced discontinuities for the most of the core
sections. Totally, thirty-three point load strength index, fourteen uni-
axial compressive strength, three sets of triaxial compressive strength
tests which were carried out in accordance with the suggested ISRM
methods were available. The length/diameter ratio (L/D) for the com-
pression tests varied from 1.7 to 2.5. The use of relatively short core
length (L/D = 1.7–2) was necessary as the rock mass is heavily frac-
tured and in some of the pressuremeter test intervals, only short sam-
ples were available. The triaxial compressive strength test results were
also taken into account to estimate the uniaxial compressive strength if
σci was unavailable along a borehole measurement interval for de-
formation modulus. The intercept of σci at the σ1 - σ3 plot of the triaxial
testing was considered in such circumstances. The point load to uniaxial
compressive strength conversion factor was simply taken as 16 in ac-
Fig. 7. Test locations around the shaft location.
cordance with the site-specific experience. Another borehole had been
drilled for the geotechnical investigations for the shaft sinking opera-
boreholes (Fig. 7). Nineteen of the sixty test results were discarded tions and regular deformability testing was luckily carried out on the
which were located in a rockfill section and some of them had ques- samples obtained from this borehole. Those tests resulted in an average
tionable results (very low modulus for a rock mass) probably due to Ei/σci (modulus) ratio of 330, which is used for the estimation of the
oversized borehole diameter or drilling disturbance. Eventually, forty- modulus of elasticity of the core samples obtained from the pressure-
one pressuremeter test results and corresponding borehole core logs meter test boreholes.
were taken into account and all of them were located in the virgin rock The examination of the core-boxes enables the determination of the
mass. The maximum testing depth reaches a maximum of 24 m and the practical and commonly used RQD, discontinuity condition rating of
average interval between the test locations along a single borehole the RMR (Dc) and other parameters required for the calculation of
varies between 3 and 4 m. RMR, Q and Qc. As it is explained in Section 5, the deformation modulus
Due to the fractured and greatly varying nature of the rock mass, a predictions taking the parameters RMR and Q into account, gave con-
detailed site investigation was decided to be conducted since the service siderably higher results. During the determination of rock mass classi-
shaft of the main headframe is expected to serve throughout the life of fication ratings for the poor ground conditions, especially if RMR <
the mine and the foundation will be subjected to considerable static and 45, some difficulties can be encountered.50 That is why the ratings
dynamic loads. The shaft was sunk down to a depth of a few hundred RMR and Q were estimated by the previously proposed correlations
meters when the testing programme was initiated. which take GSI as an input. Then the correlations gave slightly low
The uppermost level of Deniş formation is defined as Tuff- classification ratings. The correlations were taken from the study which
Agglomerate Series (P4) which is dominated by volcanic rocks. Rocks was postulated by considering a database collected in the same mine
corresponding to P4 are composed of pyroclastic and lava that forms and the geology.51 Qc=Q(σci/100) is also calculated by considering the
the uppermost levels in the study area and its vicinity. All testing available laboratory test results. The GSI rating was calculated con-
programme is located within this zone. The rock mass consists of an- sidering the quantified approach proposed by Hoek et al.52 A modified
desite and weathered andesite having a variable colour from grey to version of a quantified GSI chart was also given in 52 and GSI value can
cream. The rock material exhibits a variable strength which is most be calculated as follows:
probably depending on the weathering and the mineralogical content.
The structural quality of the rock mass varies from disintegrated to very
GSI = (RQD (%)/2) + 1.5 Dc (2)
blocky. A photograph from the pre-sinking stage of the shaft give the The GSI chart is shown in Fig. 9 on which the field data is plotted.
visual illustration of the rock mass and three core-box photographs The data are mainly concentrated in the lower half of GSI chart. The
exhibit different rock mass qualities are presented in Fig. 8. acquired database consisting of forty-one rows is represented in the
Although the sections with relatively long solid core pieces are form of frequency histograms (Fig. 10). The distribution of the pres-
presented in Fig. 8, the RQD is low due to the poor core recovery. suremeter measurements (indicated by Ep), testing depth, uniaxial
Another problem was recovering the intact core pieces with adequate compressive strength, RQD, Dc, GSI, RMR, Q and Qc are presented.

Fig. 8. Heavily fractured andesitic rock mass in the shaft excavation and in the core boxes.

245
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 9. The field data is represented in Quantified GSI chart52.

The maximum deformation modulus in the database is 2150 MPa and it is depending on the size of the pressuremeter. Hence, the method
and the minimum value is 48 MPa. The majority of the measurements becomes appropriate for low RQD rock masses.27 In this study, the rock
vary between 500 and 1500 MPa. The minimum testing depth is 4 m mass is densely fractured and the use of Menard pressuremeter resulted
and testing depths are distributed almost evenly down to a depth of in measurements which are comparable with the dataset used by Hoek
24 m. The uniaxial compressive strength of the intact core pieces ex- and Diederichs17 which is presented in Fig. 11. Still, the use of pres-
hibits a strong variation. The RQD values lie in the range of suremeters in a rock mass with widely spaced joints should be avoided
0–57 however, almost half of the pressuremeter test intervals have zero as it cannot influence such a large volume with widely spaced joints.
RQD and the majority of the data has a rating smaller than 20. The field data and the previous studies seem to be in a good
Discontinuity condition rating indicates that the majority of the dis- agreement, however, the modulus range is large. The data collected in
continuities are in fair condition and none of them is in very good this study falls in the lower bound of the previous studies presented in
condition. The majority of the GSI values lie between 15 and 35 similar Fig. 11. The vertical axes have a wide range and this circumstance
to the RMR ratings. The maximum Q rating is 0.481 and most of the makes the prediction error not clearly visible. In terms of the order of
data points fall into a range of 0.001–0.01. The strength adjusted Qc magnitude, still, the commonly used empirical equations result in rea-
parameter has a minimum value of 0.0002 due to the having “σci/100” sonable predictions. In order to deeply investigate the problem, the
multiplier which is smaller than 1 for this case. From a statistical point field data is plotted against RQD, RMR, GSI, Q and Qc, (Fig. 12). De-
of view, taking the advantage of a normally distributed histograms formation modulus is represented by Ep as an indicator of pressuremeter
could be beneficial. However, in rock engineering, it may not always be measurement. In fact, Fig. 12 provides a closed look to the plots shown
possible to obtain such distribution of the collected data.39,53 in Fig. 11.
The pressuremeter test data is plotted on the graphs given in 17 for As it is illustrated in Fig. 12, the axes of the plots arranged to the
the comparison (Fig. 11). For the Erm vs. RMR plot, the commonly used available field data and low R2 values for the relationships between the
empirical equations for the estimation of the rock mass modulus are pressuremeter measurements and the rock mass classification ratings
shown as it is given in their study. Hoek and Diederichs17 opposed using are obtained. The scatter of the data becomes obvious when the poor
borehole expansion test results because of the scattered results in their rock mass in the study area is considered only, with adjusted scales for
dataset. In fact, the volume of the rock mass tested by a pressuremeter the available data range. Among the correlations, the rating Q provided
should be examined and its compatibility with discontinuity orientation a slightly better performance with an R2 value around 0.34.
and spacing must obviously be checked. The joint density must be high

246
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 10. Frequency histograms for the field data.

5. Comparison of empirical equations and proposal of new deformation modulus and considered correlation coefficient (r) for the
equations performance evaluation.
Kallu et al.56 mentioned that the correlation coefficient, F-statistics,
Statistical evaluation of the performance of the predictions can be and RMSE are overall metrics of the predictive performance and such
undertaken by using taking several parameters into account like it is parameters do not indicate the specific strengths or weaknesses of the
performed by several researchers. The correlation coefficient was used relationships. The ranges over which the relationships have the best and
in 27 for a study on the correlation of pressuremeter data and RMR/GSI. worst predictive performances can be evaluated by considering an
Chun et al.54 reported pressuremeter tests performed in Korean rock “error ratio, ER”. Calculating ER between predicted and measured va-
mass and gave a comparison of the commonly used empirical equations. lues is expressed to be a versatile method of quantifying the perfor-
All of the commonly used equations overestimated the test results and mance of a predictive model. Kallu et al.56 described ER as follows,
an empirical equation was suggested with an R2 value around 0.36. In
ER = (Y Y)/(Y) (3)
fact, the value seems to be low however, there were numerous test
results. It is quite understandable when there is a great number of data where Y’ represents predicted data and Y stands for measured data.
points and then, small R2 values can be considered for the comparison Hoek and Diederichs17 used the same term but a different calculation
of the performance of the prediction. Bashari et al.55 correlated the for the performance evaluation for their deformation modulus predic-
parameters UCS, GSI, density, RQD, joint spacing and, porosity with the tions. In order to avoid a confusion, the term is named as the predicted-

Fig. 11. Deformation modulus versus RMR and GSI with the data in this study (indicated by red)17 (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article).

247
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 12. Deformation modulus versus rock mass classification systems in this study.

Table 2
Equations proposed by nonlinear multiple regression modelling.
Equation number Deformation modulus prediction models (MPa) Limitations

Eq. (1) Erm = (RQD + 10)0.648124 × ci


1.06661 + Dc 2.23513 RQD(%)≤ 55
4 ≤Dc≤ 20
Eq. (2) 10 ≤ σci≤ 100 MPa
( )
RQD + 1 + Dc × 11.7478 1.1056
( )
0.251
Erm = 100 × ci
20 100
Eq. (3) (*) (*) 4 ≤H≤ 24 m
=( )
H 0.4797
Erm × (RQD + 10) 0.6496 × ( ci)1.038 + Dc 2.229
10
Eq. (4) (**) Erm = 1.25×(RQD + 5) 0.75 × (0.075× ci)2.25 + (2 × Dc )1.5 (**) Proposed for a lower bound prediction

measured ratio, PMR. PMR can be expressed as: gave similar results. However, one should remember that the depth is
included in Eq. (3) and in the plot the parameter H is taken as 10 m. The
PMR = Y / Y (4)
contribution of the depth is represented by (H/10)0.4797 in Eq. (3) and it
PMR is greater than 1 when predicted value is overestimated. When yields 0.64 for H= 4 m, 1 for 10 m and 1.55 for H= 25 m. 3 fold dif-
an underestimation is observed PMR value is calculated as 1/PMR. ference is calculated for 4 m and 25 m depth if RQD, σci, Dc are kept
Then, the PMR value satisfies the condition of always being greater fixed. In another explanation, the predicted Erm at a depth of 25 m is
than 1. The absolute value of PMR can also be used however it is not three times more than the Erm at a 4 m deep rock mass when the pre-
applied, here. dictive parameters (RQD, σci, Dc) are the same. In Section 3, the in-
Forty-one Menard pressuremeter measurements were compared to fluence of the depth was found to be ΔErm/ΔH= 1.6 MPa/m which
the empirical equations which are introduced in Section 2.2. For new leads to a difference of 33.6 MPa between the depths of 4 m and 25 m.
equations, depth, RQD, Dc and σci were chosen in order to propose In the numerical analyses, the simulated rock mass had Erm approxi-
simplistic and practical equations. The calculation of RQD is straight- mately equal to 1800 MPa and the increment due to the depth would be
forward; it gives the summation of the length of recovered solid core around 1.6 × 21 = 33.6 MPa. The rate of modulus increase was found
pieces greater than 10 cm11. Modulus of elasticity may not always be to be practically insignificant in the modelling results and as small as
available every time hence, it is disregarded and σci was chosen as a 1.87% for a depth interval between 4 m and 25 m. In contrast to the
predictive parameter. The structural quality of the rock mass is re- modelling results, the field studies pointed out a considerable influence
presented by RQD and Dc is chosen to represent the discontinuity on the modulus along the testing depth covered in this study (4–25 m).
condition. Since the linear multiple regression models did not exhibit a The remaining three equations do not take the depth into account.
satisfactory performance, nonlinear multiple regression techniques Eqs. (1), (2), and (4) are in the form of Erm = ƒ(RQD, σci, Dc). Con-
were employed and the proposed equations are given in Table 2. sidering Eqs. (1) and (4), if σci is 10 MPa, the contribution of the
For Eq. (3), depth (H) is in meters with a 4 m minimum and max- parameter Dc to Erm is dominant and RQD has only a slight influence on
imum of 24 m. In case of a utilization of the equations for a foundation the modulus. The increase of intact rock strength makes the influence of
engineering or a surface structure design, the parameter depth can play RQD more pronounceable. In fact, the contribution of the RQD to Erm is
a role in the estimation of the deformation modulus. Erm and σci are in also dependent on the Dc. In the presence of discontinuities having
MPa. Discontinuity condition rating of RMR (Dc) is unitless and nor- good and fair conditions (i.e. Dc= 10–20), the influence of RQD is
mally varies between 0 and 30. There is a limitation on the parameter greater, and smaller where low Dc values are present in the rock mass.
Dc which must be between 4 and 20 for the proposed equations. Thick In Eq. (4), the influence of Dc parameter on the modulus ceases if σci is
clay infills or bands are excluded in this study. The parameter RQD 100 MPa and RQD becomes the dominant parameter. It should be re-
normally varies between 0 and 100 however, because of the available minded that the lowest Dc values, such as 0–4 are not covered in this
database, it covers a range of 0–55%. All of the predictive parameters study and the presence of such zones may deviate the trend observed in
have a positive correlation with the deformation modulus contributing this study. Eq. (2) has similar characteristics with Eqs. (1) and (4) as it
in the favour of greater Erm when any of the parameters is increased. can be inferred from Fig. 13.
In order to evaluate the physical meaning of the equations, contour Apart from the previously explained equations, the dependence of
plots and 3D surfaces are constructed and they are shown in Fig. 13. Erm to the parameters RQD and Dc follows a slightly changing trend for
The proposed equations are plotted for different levels of the uniaxial Eq. (2). Increasing σci leads to a slight increase of the impact of Dc
compressive strength of the intact rock. Eq. (3) has four predictive parameter on Erm. For all intact strength values, the contribution of Dc
parameters and the depth is fixed to 10 m depth in order to construct to Erm is more pronounceable than RQD.
the contour and surface plots in 3D. The maximum and minimum deformation modulus predictions are
Considering Fig. 13, the predictions suggested by Eqs. (1) and (3) also calculated taking the limitations into consideration. The minimum

248
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 13. Contour and 3D surface plots of the proposed equations.

249
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Fig. 14. ER(%) for the previously developed empirical equations.

and maximum deformation modulus values for Eq. (1) are 74 MPa and Among the previously proposed empirical equations, the one sug-
2842 MPa, 148 MPa and 1930 MPa for Eq. (2), 53 MPa and 3581 MPa gested in 37 exhibits a slightly underestimating trend with respect to
for Eq. (3), 25 MPa and 2762 MPa for Eq. (4). The validity of the the measured data. It is not surprising since the equation is suggested
equations is expressed as: 0 ≤RQD(%)≤ 55, 4 ≤Dc≤ 20, 10 ≤ σci for lower bound estimates by the researchers. The remaining empirical
≤ 100 MPa and, 4 ≤H≤ 25 m. equations have a tendency to overestimate the deformation modulus.
After the evaluation of the physical meaning of the mathematical Summation and average of PMR and ER values, together with r, R2 and
expressions, statistical performances of the equations are discussed. A RMSE are given for each equation including the ones proposed in this
comparison of the predictions obtained from the empirical equations study, (Table 3). Σ PMR gives the summation of the measured to the
presented in Table 1 with the equations proposed in this study was estimated ratio for 41 data row. Similarly, average PMR gives an
made by previously explained parameters: r, R2, RMSE, ER and PMR. average value. However, for error evaluation, the summation and
The greater r and R2 values indicate a better correlation performance. average values of PMR may be misleading since the best performance is
Since they are generally low in this study, they are used for comparison attained when PMR= 1. For overestimation and underestimation, a
purposes. The smaller RMSE value indicates a better performance. correction can also be made however, ER parameter was used for the
Here, ER and PMR play the major role in the performance evaluation by quantification of the prediction error in terms of percentage. The ER
directly providing a value about the predicted and measured data. ER is indicates the absolute error percentage no matter it is an overestimation
expressed in terms of percentage and smaller the value, smaller the or underestimation. Table 3 is presented in a conditionally coloured
prediction error. PMR represents the ratio of predicted to measured form with the performance indicators. The parameters indicate better
values. It is better if it is close to unity. If the PMR value is greater than performance for the greener table cells.
1, the prediction overestimates Erm and if it is smaller than 1, an un- The equations proposed in this study perform well and slightly
derestimation is observed. better than the equations proposed by Hoek and Diederichs,17 Ni-
ER values are plotted for each data point. The plots belonging to the cholson and Bieniawski,30 Gökceoglu et al.36 and Zhang and Einstein.37
previously developed empirical equations (Fig. 14) and developed in The mentioned predictions have better performance among the existing
this study (Fig. 15) are provided in individual plots to increase the equations. In fact, Hoek and Diederich's17 equation has a flexibility
visibility. since there is an additional parameter D, the disturbance factor. The

Fig. 15. ER(%) for the multiple regression models developed in this study.

250
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

Table 3
Comparison of the empirical equations in the literature and proposed in this study.

parameter D is generally chosen with respect to the blast damage or the influence of the depth is limited to 9% for a 100 m depth differential for
excavation method. This parameter can also be used for the upper and the model of concern in this research and not all rock masses ne-
lower bound values. The predictions proposed by Sonmez et al.39 and cessarily exhibit the same impact with the same order of magnitude.
Gökceoglu et al.36 having RMR as an input parameter, performed better Another finding in the numerical analyses is that the deformation
than the predictions by the same researchers with GSI as an input modulus is insensitive to k ratio at a constant depth. Not only are the
parameter. Among the new Eq., Eq. (4) has the best ER values and it is two of them, but all of the principal stresses are important for the stress
suggested for a lower bound prediction. Eq. (3) is a prediction with the dependency.
depth input and it can be used for the rock engineering problems at During the analyses, the 200 m deep numerical models resulted in
shallow depth. Eq. (1) is the suggested prediction for general use in inconsistent deformation modulus, which was slightly smaller than it
poor rock conditions. was found in a 100 m deep model. Especially when k > 1 at great
depth, the development of plastic zones after drilling the borehole and
6. Conclusion before the initiation of the testing is attributed to the decrease in the
modulus. Hence, the deformation modulus was governed by plasticity
The modulus of deformation measurements were obtained from rather than elasticity. The modelling case at a depth of 4 m was also
Menard pressuremeter testing in a poor rock mass. Numerous pre-ex- tried, again it resulted in similar deformation modulus values with 20 m
isting empirical equations and the pressuremeter test results were deep models. The absence of the depth sensitivity of the deformation
compared to each other. The performance of sixteen predictions was modulus at shallow depth can be explained by the discontinuity con-
evaluated, discussed and presented. A few of them performed well but stitutive model and failure criterion both of which are linear in this
the majority gave overestimated results. The equations commonly research. The use of a non-linear constitutive model and a non-linear
consider RMR, Q and GSI and some laboratory test results as predictive failure criterion can be used for a further rigorous modelling. A detailed
parameters. modelling work possibly with a parametric study may provide a valu-
Considering the practicality, the well-known RQD, discontinuity able data about stress dependency of the deformation modulus. A rig-
condition rating of the RMR system Dc and σci parameters were selected orous numerical modelling is promoted since it is really difficult to
to predict the deformation modulus for a heavily fractured andesitic carry out in-situ tests at different depths but in the rock masses which
rock mass. Eventually, four equations were proposed, one of which is are in identical condition and have the same mechanical properties.
aimed to give lower bound predictions (Eq. (4)). Eq. (1) performs better
than Eq. (2), both of which can be used for a general purpose. There are Acknowledgements
some limitations for the proposed equations: RQD < 50%, 4 < Dc <
20 and 10 < σci < 100 MPa. The equation with the inclusion of the The author thanks the management personnel of Polyak Eynez
depth (Eq. (3)) is valid for a depth range of 4–24 m without the Energy Mining A.Ş. and Fina Energy for granting permission to use
groundwater. The equations presented in this study can be used espe- necessary data. The author presents his gratitude to Geological
cially for comparison purposes where the similar conditions exist. In Engineers of Polyak Eynez, Feridun Emre Yağımlı, Ali Türkoğlu, and
such cases, the provided limitations must be considered. Yusuf Tonbaktepe, for providing extensive data on the geology of the
Another concern was the influence of the depth on the deformation study area, and their additional care during the data collection.
modulus. The preliminary analyses were carried out by using the finite
element analysis assuming the rock mass behaves as a continuum and References
discontinuum. In continuum approach, the rock mass modelled as an
isotropic, elastic-brittle-plastic rock mass while the other one contains 1. Cai M, Kaiser PK, Uno H, Tasaka Y, Minami M. Estimation of rock mass deformation
discontinuities in the elastic-brittle-plastic rock material. A range of modulus and strength of jointed hard rock masses using the GSI system. Int J Rock
Mech Min Sci. 2004;41(1):3–19.
depth and k values were imposed on the numerical models aiming to 2. Farmer IW, Kemeny JM. Deficiencies in rock test data. In: Rock Characterization:
simulate the pressuremeter test. The effect of the depth was not ob- ISRM Symposium, Eurock'92, Chester, UK, 14–17 September 1992 p. 298-303.
served in the isotropic continuum model. The simulated pressuremeter 3. Hammah R, Curran J, Yacoub T. The Influence of Young’s Modulus on Stress
Modelling Results. In: Golden Rocks 2006, Proceedings of the 41st US Symposium on
measurement was not affected by the confining pressure. In contrast to Rock Mechanics (USRMS); 2006.
the continuum modelling, the discontinuum modelling resulted in an 4. Palmstrøm A. Characterizing rock masses by the RMi for use in practical rock en-
increasing trend for the deformation modulus with respect to the in- gineering: Part 1: the development of the Rock Mass Index (RMi). Tunn Undergr Space
Technol. 1996;11:175–188.
creasing depth provided that the mechanical properties of the rock mass 5. Palmström A, Singh R. The deformation modulus of rock masses—comparisons be-
were kept constant. The discontinuities have greater tendency to de- tween in situ tests and indirect estimates. Tunn Undergr Space Technol.
form at a shallow depth than they have at a deep case which in turn 2001;16:115–131.
6. Bieniawski ZT. Engineering classification of rock masses. Trans S Afr Inst Civ Eng.
result in the stress dependency of the deformation modulus. Still, the

251
İ.F. Öge International Journal of Rock Mechanics and Mining Sciences 112 (2018) 238–252

1973;15:335–344. 33. Aydan Ö, Ulusay R, Kawamoto T. Assessment of rock mass strength for underground
7. Ozgenoglu A, Mehmetoglu AG, Karpuz C. Comparison of rock mass deformabilities excavations. Int J Rock Mech Min Sci. 1997;34(3–4) [18-e1].
determined using large flat jack technique and plate jacking test at two dam sites. In: 34. Read SA, Perrin ND, Richards LR. Applicability of the Hoek-Brown failure criterion to
Proceedings of the 6th ISRM Congress International Society for Rock Mechanics; 1987. New Zealand greywacke rocks. In: Proceedings of 9th ISRM Congress; 1999.
8. Ünal E. Determination of in situ deformation modulus: new approaches for plate- 35. Hoek E, Carranza-Torres C, Corkum B. Hoek-Brown failure criterion-2002 edition. In:
loading tests. Int J Rock Mech Min Sci. 1997;34(6):897–915. Proceedings of NARMS-Tac. 2002. p. 267-273.
9. Shen J, Karakus M, Xu C. A comparative study for empirical equations in estimating 36. Gokceoglu C, Sonmez H, Kayabasi A. Predicting the deformation moduli of rock
deformation modulus of rock masses. Tunn Undergr Space Tech. 2012;32:245–250. masses. Int J Rock Mech Min Sci. 2003;40(5):701–710.
10. Karaman K, Cihangir F, Kesimal A. A comparative assessment of rock mass de- 37. Zhang L, Einstein HH. Using RQD to estimate the deformation modulus of rock
formation modulus. Int J l Min Sci Tech. 2015;25(5):735–740. masses. Int J Rock Mech Min Sci. 2004;41:337–341.
11. Deere DU, Hendron AJ, Patton FD, Cording EJ. Design of surface and near-surface 38. Sonmez H, Gokceoglu C, Ulusay R. Indirect determination of the modulus of de-
construction in rock. In: Proceedings of the 8th US symposium on rock mechanics formation of rock masses based on the GSI system. Int J Rock Mech Min Sci.
(USRMS); 1966. 2004;41:849–857.
12. Bieniawski ZT. Engineering Rock Mass Classifications: A Complete Manual for Engineers 39. Sonmez H, Gokceoglu C, Nefeslioglu HA, Kayabasi A. Estimation of rock modulus: for
and Geologists in Mining, Civil, and Petroleum Engineering. John Wiley & Sons; 1989. intact rocks with an artificial neural network and for rock masses with a new em-
13. Barton N. Some new Q-value correlations to assist in site characterisation and tunnel pirical equation. Int J Rock Mech Min Sci. 2006;43:224–235.
design. Int J Rock Mech Min Sci. 2002;39(2):185–216. 40. Bieniawski ZT. Determining rock mass deformability—experience from case his-
14. Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Rock Mech Min Sci. tories. Int J Rock Mech Min Sci Geomech Abstr. 1978;15:237–247.
1997;34(8):1165–1186. 41. Stephens RE, Banks DC. Moduli for deformation studies of the foundation and
15. Tanimoto C Contribution to discussion in proceedings of the paris conference on abutments of the Portugues Dam—Puerto Rico. In: Proceedings of the 30th US
analysis of tunnel stability by the convergence–confinement method (Published in Symposium on Rock Mechanics as a Guide for Efficient Utilization of Natural
Underground Space 4 (4)). Resources: Morgantown; 1989:31–38.
16. Aksoy CO, Geniş M, Aldaş GU, Özacar V, Özer SC, Yılmaz Ö. A comparative study of 42. Verman M, Singh B, Viladkar MN, Jethwa JL. Effect of tunnel depth on modulus of
the determination of rock mass deformation modulus by using different empirical deformation of rock mass. Rock Mech Rock Eng. 1997;30(3):121–127.
approaches. Eng Geo. 2012;131:19–28. 43. Agan C. Determination of the deformation modulus of dispersible-intercalated-
17. Hoek E, Diederichs MS. Empirical estimation of rock mass modulus. Int J Rock Mech jointed cherts using the Menard pressuremeter test. Int J Rock Mech Min Sci.
Min Sci. 2006;43(2):203–215. 2014;65:20–28.
18. Agharazi A, Tannant DD, Martin CD. Characterizing rock mass deformation me- 44. Rocscience Inc. Phase2 Software v.8.0; 2011.
chanisms during plate load tests at the Bakhtiary dam project. Int J Rock Mech Min 45. Barton NR, Bandis SC. Effects of block size on the shear behaviour of jointed rock. In:
Sci. 2012;49 [1-1]. Proceedings of 23rd U.S. Symposium on Rock Mechanics, Berkeley; 1982:739-760.
19. Kavur B, Cvitanović NŠ, Hrženjak P. Comparison between plate jacking and large flat 46. Baecher GB, Lanney NA. Trace length biases in joint surveys. In: Proceedings of the
jack test results of rock mass deformation modulus. Int J Rock Mech Min Sci. 19th US Symposium on Rock Mechanics (USRMS); 1978.
2015;73:102–114. 47. Basarir H, Oge IF, Aydin O. Prediction of the stresses around main and tail gates
20. Jaeger C. Rock Mechanics and Engineering. Cambridge University Press; 1979. during top coal caving by 3D numerical analysis. Int J Rock Mech Min Sci.
21. Mair RJ, Wood DM. Pressuremeter Testing. Ciria; 1989. 2015;76:88–97.
22. Galera JM, Alvarez M, Bieniawski ZT. Evaluation of the deformation modulus of rock 48. Aksoy CO, Kucuk K, Uyar GG. Long-term time-dependent consolidation analysis by
masses: comparison of pressuremeter and dilatometer tests with RMR prediction. In: numerical modelling to determine subsidence effect area induced by longwall top
Proceedings of the ISP5-PRESSIO. Paris; 2005:1–25. coal caving method. Int J Oil, Gas Coal Tech. 2016;12(1):18–37.
23. Ménard L. An Apparatus for Measuring the Strength of Soils in Place (M.Sc. Thesis). 49. Öge İ.F, Çırak M. Relating rock mass properties with Lugeon value using multiple
Urbana: University of Illinois; 1956. regression and nonlinear tools in an underground mine site. Bull Eng Geol Environ.
24. Ménard L. Influence de l′amplitude et de l′histoire d′un champ de contraintes sur le 2017:1–4. https://doi.org/10.1007/s10064-017-1179-0.
tassement d′un sol de foundation. VICSMFE, Paris. 1961:249–253. 50. Özkan İ, Erdem B, Ceylanoğlu A. Characterization of jointed rock masses for geo-
25. Goodman RE. Introduction to Rock Mechanics. New York: Wiley; 1989. technical classifications utilized in mine shaft stability analyses. Int J Rock Mech Min
26. Baguelin F, Jezequel JF, Le Mehaute A. Etude des pressions interstitielles devel- Sci. 2015;73:28–41.
oppees lors de lessai pressiometrique. In: Proceedings of the 8th International 51. Öge IF, Yağımlı FE, Türkoğlu A. Soma Kömür Havzasında Pliyosen Kaya Birimlerinde
Conference on Soil Mechanics and Foundation Engineering; 1973:19–24. RMR, Q-Sistemi ve GSI Değerlerinin İlişkilendirilmesi - Relating RMR, Q-System and
27. Isik NS, Ulusay R, Doyuran V. Deformation modulus of heavily jointed–sheared and GSI Ratings in Pliocene Formations of Soma Coal Basin. Turkey 25. International
blocky greywackes by pressuremeter tests: numerical, experimental and empirical Mining Congress, IMCET. Antalya; 2017. p. 170-174. (in Turkish).
assessments. Eng Geo. 2008;101(3):269–282. 52. Hoek E, Carter TG, Diederichs MS. Quantification of the geological strength index
28. Aydan Ö, Ulusay R, Tokashiki N. A new rock mass quality rating system: rock mass chart. In: Proceedings of the 47th US Rock Mechanics/Geomechanics Symposium;
quality rating (RMQR) and its application to the estimation of geomechanical 2013.
characteristics of rock masses. Rock Mech Rock Eng. 2014;47(4):1255–1276. 53. Kayabasi A, Yesiloglu-Gultekin N, Gokceoglu C. Use of non-linear prediction tools to
29. Serafim JL, Pereira JP. Consideration of the geomechanical classification of assess rock mass permeability using various discontinuity parameters. Eng Geo.
Bieniawski. In: Proceedings International Symposium on Engineering Geology and 2015;185:1–9.
Underground Construction (Lisbon); 1983:33–44. 54. Chun BS, Lee YJ, Seo DD, Lim BS. Correlation deformation modulus by PMT with
30. Nicholson GA, Bieniawski ZT. A nonlinear deformation modulus based on rock mass RMR and rock mass condition. Tunn Undergr Space Tech. 2006;21(3):231–232.
classification. Int J Min Geol Eng. 1990;8:181–202. 55. Bashari A, Beiki M, Talebinejad A. Estimation of deformation modulus of rock masses
31. Barton N. The influence of joint properties in modelling jointed rock masses. In: by using fuzzy clustering-based modeling. Int J Rock Mech Min Sci.
Proceedings of Keynote Lecture of 8th Congress of ISRM, Tokyo, vol. 3. Rotterdam: 2011;48:1224–1234.
Balkema, 1995. 56. Kallu RR, Keffeler ER, Watters RJ, Warren SN. Development of a highly portable
32. Jasarevic I, Kovacevic MS. Analyzing applicability of existing classification for hard plate loading device and in situ modulus measurements in weak rock masses. Rock
carbonate rock in Mediterranean area. In: Proceedings of EUROCK’96, Turin, Italy, Mech Rock Eng. 2016;49:443–454.
September 1996, 1996:811–818.

252

You might also like