You are on page 1of 8

ARTICLE IN PRESS

International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377

Contents lists available at ScienceDirect

International Journal of Rock Mechanics & Mining Sciences


journal homepage: www.elsevier.com/locate/ijrmms

Estimation of rock mass deformation modulus using variations in transmissivity and RQD with depth
Xiao-Wei Jiang , Li Wan, Xu-Sheng Wang, Xiong Wu, Xin Zhang
School of Water Resources and Environment, China University of Geosciences, Beijing 100083, China

a r t i c l e in f o
Article history: Received 9 December 2008 Received in revised form 24 April 2009 Accepted 12 May 2009 Available online 7 June 2009 Keywords: Transmissivity RQD Normal stiffness Deformation modulus Packer test

a b s t r a c t
Discontinuity normal stiffness and deformation modulus of large scale rock masses are very difcult to determine. A method for estimation of discontinuity normal stiffness based on the decrease in transmissivity with depth has been proposed by the authors in a former paper. In the current study, the method is further developed by accounting for the changes in both discontinuity aperture and frequency with depth, which are key factors that cause the transmissivity to decrease with depth. The discontinuity frequency can be estimated from RQD measurements, which are readily available in most geotechnical investigations. The transmissivity data from packer tests are usually available in geotechnical investigations for hydropower plants. For a rock mass in a dam site mainly controlled by lithostatic stress, based on transmissivity and RQD data at different depths, the change in discontinuity aperture with depth can be linked to the change in aperture with stress, which denes the normal stiffness of discontinuities. In the case study, the discontinuity normal stiffness is successfully estimated by using transmissivity and RQD data, and the result shows that the normal stiffness increases with stress (depth) and the rate of normal stiffness versus stress (depth) decreases with stress (depth), which is consistent with experimental studies. The estimated normal stiffness has been utilized to calculate the rock mass deformation modulus using an equivalent model. The result of deformation modulus by the proposed method is close to that obtained by using in situ measurements, as well as by using empirical models relating RQD to deformation modulus. & 2009 Elsevier Ltd. All rights reserved.

1. Introduction The deformation modulus is the most representative parameter describing the pre-failure mechanical behavior of any engineering material. In rock masses, the existence of discontinuities makes the mechanical properties of rock masses differ greatly from that of intact rocks. Therefore, laboratory tests on small specimens cannot successfully predict the deformation of rock masses. Moreover, even in situ measurements of the deformability (such as plating loading, radial jacking, etc.), which usually involve difcult test procedures, are still not able to obtain parameters that can represent large scale rock masses. The determination of the deformation modulus of large scale rock masses remains one of the most difcult tasks in the eld of rock mechanics [1]. Although rock masses are composed of intact rocks and discontinuities, the deformation of rock masses occurs mainly in the discontinuities, especially when subjected to low stress state [2,3]. Experiments have also shown that the permeability of rock

Corresponding author. Tel.: +86 15801304208.

E-mail address: xiaow.jiang@gmail.com (X.-W. Jiang). 1365-1609/$ - see front matter & 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijrmms.2009.05.004

masses is primarily dependent on the discontinuities. Both hydraulic and mechanical properties of rock masses are related to the geometry of discontinuities [4], which suggests a new way to estimate mechanical properties from hydraulic information. Rutqvist [5] utilized hydraulic jacking test to determine normal stiffness of discontinuities in hard rocks. Jiang et al. [6] have estimated the stress-dependent discontinuity normal stiffness of large scale rock masses using the permeability data from packer test in a wide range of depths. In [6], it was assumed that the decrease in permeability with depth, a phenomenon which had been observed by numerous researchers throughout the world [710], was related to the decrease in aperture with depth, and can be linked to the change in normal stiffness with depth. The normal stiffness, which describes the relationship between discontinuity aperture and normal stress, was estimated based on the rate of decrease in permeability with depth. Then an equivalent continuum model was utilized to calculate the deformation modulus of the rock mass, which is consistent with in situ measurements. Unfortunately, in [6], the normal stiffness was linearly dependent on depth (stress). As pointed out in their paper, this result is not completely realistic due to the lack of information on change in discontinuity spacing (frequency) with depth. In fact, for a specic

ARTICLE IN PRESS
X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377 1371

site, discontinuity aperture would generally decrease with depth and discontinuity spacing would usually increase with depth [7], and the two factors would simultaneously contribute to the decrease in permeability with depth. If the change in discontinuity spacing (frequency) with depth is available, the method used in [6] could be improved. Rock quality designation (RQD), which was introduced by Deere [11], has been widely applied in the eld of rock mechanics. Many engineers have frequently determined the soundness of rock masses through only the RQD combined with their own experiences [12]. Numerous researchers have established empirical models to estimate mechanical properties from RQD [13,14], RMR [1518], or Q [19,20]. Although it is more sensible to estimate mechanical properties from RMR and Q due to the fact that RQD is only one of the components that affect mechanical properties of rock masses, RQD is still widely employed because, in many cases, RQD rather than RMR or Q is available [14]. Priest and Hudson [21] show that discontinuity frequency is correlated with RQD, which implies a method for estimation of fracture frequency (spacing) from RQD. The aim of the present paper is to further develop the method proposed in [6] by incorporating RQD data.

plane, can be written as Kf gb 12f n


2

(2)

where Kf is the hydraulic conductivity of a single discontinuity, b is the aperture of the discontinuity, g is acceleration due to gravity, n is kinematic viscosity of the uid, and f is the roughness factor. Transmissivity, T, instead of hydraulic conductivity, is a more widely applied parameter to show the hydraulic property of rock masses. For a parallel set of discontinuities, the transmissivity parallel to the discontinuities is dened as T gb lL 12f n
3

(3)

where L is the length of a section of rock mass (with a parallel set of discontinuities) in the direction perpendicular to the planes of discontinuities. In Eq. (3), the permeability of intact rocks, which is usually several orders of magnitude smaller than that of rock masses, is ignored. When the L 1, transmissivity equals hydraulic conductivity of a section of rock mass. It is clear that the transmissivity is dependent on the aperture, frequency, orientation, and roughness of discontinuities. 2.3. RQD

2. Denitions 2.1. Discontinuity spacing and discontinuity frequency Spacing and frequency are two parameters that are frequently used to characterize discontinuities. Spacing is the distance between adjacent discontinuity intersections with the measuring scanline, whereas frequency is the reciprocal of spacing. The spacing of discontinuities determines the sizes of blocks making up the rock mass. In a rock mass, discontinuity spacings are usually not a constant. Instead, they take a wide range of values. Many researchers reported that the negative exponential distribution could be employed to t the discontinuity spacing histogram based on numerous measurements of discontinuity spacing [2123]. Therefore, the frequency, f(x), of a given discontinuity spacing, x, can be given by the following function: f x l explx (1) RQD is dened as the percentage of a scanline consisting of spacing values greater than or equal to 100 mm. Theoretically, RQD is also related to the orientation of the scanline, and thus the orientation of discontinuities. In engineering applications, RQD is usually calculated as the percentage of the borehole core in a drill run consisting of intact lengths of rock greater than or equal to 100 mm, which can be mathematically expressed as RQD 100
n X xi i 1

(4)

where xi are the lengths of individual pieces of core in a drill run having lengths Z100 mm and X is the total length of the drill run. Based on Eq. (1), it was found by Priest and Hudson [21] that the estimate of RQD could be obtained from discontinuity spacing measurements made on core using the following: RQD 1001 0:1le0:1l (5)

where l 1=x is the mean discontinuity frequency of a large discontinuity population and x is the mean discontinuity spacing. Note that in this distribution, both the mean and the standard deviation equal 1/l. 2.2. Permeability Permeability is dened as the ability of rocks to transmit uids when a pressure gradient exists. The permeability of a rock and a rock mass usually differs greatly, since the ow of uids mainly takes place in the discontinuities. The cubic law is fundamental to study groundwater ow in fractured media. It can be derived from the NavierStokes equation, and can also be obtained from ow rate through a smooth parallel-walled channel under laminar ow [24]. However, it is only valid at low stress, when the discontinuities are still open. Gale et al. [25] proposed that when the surfaces of a discontinuity are not in contact, existing laboratory data support the use of the cubic law, with appropriate corrections for roughness. According to the cubic law, after correction for roughness, the hydraulic conductivity of a single discontinuity, parallel to its

For values of l ranging from 6 to 16/m, a good approximation to measured RQD values was found to be given by the linear relation RQD 3:68l 110:4 (6)

Eqs. (5) and (6) could be utilized to convert between RQD and discontinuity frequency (spacing). 2.4. Normal stiffness The normal stiffness, which is dened as the rate of change in normal stress with respect to discontinuity closure, is one of the key parameters to estimate normal deformation characteristics of rock masses. With increasing normal stress, the opening of a joint would decrease. Normal stressdisplacement relationship is highly non-linear and has been empirically described by several models. According to Goodman [3], normal stiffness, Kn, is dened in an elastic manner as Kn ds0n ds0 n dun db (7)

ARTICLE IN PRESS
1372 X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377

where un is the normal displacement (positive for closure), b is the aperture of the discontinuity, and s0n is the effective normal stress. 2.5. Deformation modulus Under normal stress, the deformation modulus of materials, E, is dened as the ratio of normal stress, s, to corresponding normal strain, e E ds d (8)

Theoretically, to calculate deformation modulus of rock masses, the deformation of the intact rocks and the deformation of the discontinuities should be considered simultaneously. However, the total deformation is mainly caused by the deformation of discontinuities, especially at low stress conditions [2]. The equivalent continuum models [26,27] clearly show the effect of the fracture stiffness on the rock mass deformation modulus. Kulhawys equivalent continuum model [26] is frequently utilized to calculate the rock mass deformation modulus 1 1 1 Emn Er K n x (9)

Fig. 1. A schematic map showing the packer test.

K (m/d) 0.0001 0.001 0.01 0 0.1 1 0 0 20

RQD (%) 40 60 80 100

where Emn is the equivalent rock mass deformation modulus and Er is the deformation modulus of the intact rock.
40 Depth (m) 40 Depth (m)

3. The study area A pumped storage power plant is to be built in Eastern Shandong Province, China. In such a power plant, two reservoirs should be built. When the demand for electricity is low, water is pumped up to the top reservoir. When there is a sudden demand for power, water rushes down the tunnels to drive the turbines to generate electricity. The water is then collected in the bottom reservoir, ready to be pumped back up later. The bedrock in the study area is composed of monzonitic granite, quartz monzonite, and quartz syenite. Due to their similarity in engineering properties, they are all considered as a granitic rock in this study. Furthermore, dykes, which are mainly composed of lamprophyre, develop in the deep part (the depth is greater than 85 m below surface) of the granite rock mass. Based on the results of hydraulic fracturing, which was employed to measure the stress eld of the study area, the existing stress is mainly lithostatic stress.

80

80

120

120

Fig. 2. The scatter plots of hydraulic conductivity and RQD versus depth: (a) hydraulic conductivitydepth and (b) RQDdepth.

4. Relationship between transmissivity and RQD in the study area 4.1. The transmissivity data and RQD data Numerous boreholes had been drilled in the study area. In each of the wells, RQD data, which were calculated using Eq. (4), are available. In most wells, hydraulic conductivity data obtained from packer test are also available. The packer test, also known as the injection test, or Lugeon test, is the most elementary and prevalent method of characterizing in situ permeability of fractured media [28]. It is a relatively inexpensive method and can determine variations in permeability with depth and also in different strata. In China, the packer test is indispensable before construction of hydropower stations, and is required by the national standards. As shown in Fig. 1, water is injected into a test section, which is isolated by a packer, and spread into the rock mass under external pressure. The ow rate can be used to represent the permeability

of the rock mass near the test section. Depending on the application of the test, the length of test sections ranges from tens of centimeters to tens of meters. In this way, an individual fracture, a group of fractures, or an entire rock formation can be isolated by packers [29]. Packer tests can be repeated at different depths to create proles of aquifer properties along boreholes. According to the current code of packer test in China [30], during the packer test, ve injection pressure steps (0.3, 0.6, 1, 0.6, and 0.3 MPa) are used. Compared with the pressure used in the hydraulic fracturing reopening test (the minimum horizontal stress is 10 or 50 MPa) [31], it is sensible to assume that no new fractures had been induced and the existing fractures had not been widened during the packer test in the study area. In addition, we assume that no turbulence happens during the packer test, which is commonly assumed in hydraulic well tests, thus the cubic law applies to the available permeability data. The hydraulic conductivity, K, of a test section in the packer test, as proposed by the code of packer test in China [30], can be approximated by K Q L ln 2pLDh rw (10)

where Q is inow rate, L is the thickness of a test section, Dh is the injection pressure with a unit of m, and rw is the radius of the borehole. In this study, the hydraulic conductivity and RQD data in the site for the top reservoir are employed. The hydraulic conductivity

ARTICLE IN PRESS
X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377 1373

K (m/d) 0.0001 0 20 40 Depth (m) 60 80 100 120 Depth (m) 0.001 0.01 0.1 1 0 20 40 60 80 100 120 0 25

RQD (%) 50 75 100

Fig. 3. The plots of hydraulic conductivity and RQD versus depth (with a resolution of 1 m): (a) hydraulic conductivitydepth and (b) RQDdepth.

data in each test section are shown in Fig. 2a and the RQD data in each drill run are shown in Fig. 2b. As shown in Fig. 2a, a decrease in permeability with depth (when the depth is less than 88 m) is shown, although the depth dependency is obscured by a large permeability spatial variation (up to four orders of magnitude). However, when the depth is greater than 88 m, the permeability in some sections is signicantly larger. This results from the high density of discontinuities caused by the development of lamprophyres. In Fig. 2b, which shows that the scatter is large, the trend of variation in RQD with depth is not clear.

K = 0.0208exp(-0.0144*RQD) R2 = 0.1054

1 0.1 K (m/d) 0.01 0.001 0.0001 0 20 40 RQD (%)


Fig. 4. The scatter plot of hydraulic conductivity versus RQD.

4.2. The transmissivityRQD correlation and variations in transmissivity and RQD with depth The original hydraulic conductivity data and RQD data shown in Fig. 2 are different in resolution. To establish the relationship between hydraulic conductivity and RQD, the two sets of data are discretized into smaller sections (with a length of 1 m), the depths of which are recorded as integral values. In this way, at each depth of a well, a value of hydraulic conductivity (or transmissivity) and a value of RQD can be obtained. The data after discretization are shown in Fig. 3. Fig. 3a is similar to Fig. 2a. In Fig. 3b, it is clear that most of the RQD values fall in the right part. And when the depth is less than 100 m, a weak trend of increase in RQD with depth can be identied. When the depth exceeds 100 m, the RQD values become more scattered, which is the result of the high density of discontinuities caused by the intrusion of lamprophyres. The decrease in hydraulic conductivity and the increase in RQD with depth indicate that there might be a correlation between hydraulic conductivity and RQD. The relationship between hydraulic conductivity and RQD can be established by using regression analysis. Here the negative exponential model is utilized. Such a model had also been employed by Liao [32] and El-Naqa [33]. The plot of hydraulic conductivity versus RQD and the result from regression analysis are shown in Fig. 4. Fig. 4 shows that an increase in RQD would result in a decrease in hydraulic conductivity. However, the coefcient of determination (denoted as R2) is very low. There is signicant scatter in the hydraulic conductivity values even for sub-sections with nearly identical RQD values. Therefore, it is difcult to predict the hydraulic conductivity of a specic position based on RQD data only.

60

80

100

Based on the data shown in Fig. 3, the average values (arithmetic mean) of hydraulic conductivity (after log transformation) and RQD at each depth are calculated (Fig. 5). When the depth ranges between 10 and 100 m, there is a general trend of increase in RQD with depth and a general trend of decrease in hydraulic conductivity with depth. When the depth is below 100 m, the trends are opposite due to the intrusion of lamprophyres. The relationship between mean hydraulic conductivity and mean RQD has been established using the negative exponential model. When the depth ranges between 11 and 100 m, R2 is 0.7431, and when the depth ranges from 11 to 88 m, R2 is 0.7809. In the following discussions on the relationship between hydraulic conductivity and RQD, the depth range 1188 m is utilized. The depth of 88 m is consistent with Fig. 2a. The mean hydraulic conductivitymean RQD correlation is shown in Fig. 6, and the large R2 value evidently indicates that the mean value of hydraulic conductivity could be estimated using the mean value of RQD obtained from borehole data. Such a conclusion had also been reported by El-Naqa [33]. 4.3. Analysis of the phenomenon Although the same model that relates RQD to hydraulic conductivity is utilized, the equations in Figs. 4 and 6 differ greatly. The hydraulic conductivityRQD correlation shown in

ARTICLE IN PRESS
1374 X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377

log K (m/d) -2.8 0 -2.4 -2 -1.6 -1.2

5. Estimation of deformation modulus using the decrease in transmissivity and the increase in RQD with depth 5.1. The theory for estimation of normal stiffness using hydraulic tests The theory proposed by Jiang et al. [6] is to be further developed here. A model for the relationship between transmissivity and stress had been proposed by Swan [34]:

40

K RQD

  0 3 sn T T r 1 A ln 0

sr

(11)

Depth (m)

80

where T is the transmissivity corresponding to a normal stress of s0n , Tr is the transmissivity at the reference normal stress s0n , and pA is a constant factor. Based on Eqs. (3) and (11), the derivative of 3 T with respect to b can be obtained as p s d3T 3 g lL (12) db 12f n In a test section, the discontinuity frequency is a constant, i.e., transmissivity only changes with aperture, which is stress dependent. By combining Eqs. (7) and (12), we obtain s p d3T 1 3 g lL (13) ds0n 12f n K n For a rock mass in a region mainly controlled by lithostatic stress, the effective normal stress of a discontinuity can be determined as follows. Suppose that the discontinuity has a dip angle of b, subject to the horizontal stress s1 and the vertical stress s3, the effective normal stress on the discontinuity can be expresses as [6]   M1 M1 s0n (14) cos 2b ge h 2 2 where h is the depth, ge is the effective specic weight that depending on saturation to calculate effective stress instead of total stress, and M is the ratio of horizontal stress to vertical stress. p Combination of Eqs. (13) and (14) leads to the derivative of 3 T with respect to h: p p rp d3T d 3 T ds0n g 3 3 lL dh ds0n dh 12f n   1 M1 M1 (15) cos 2b ge Kn 2 2 To calculate normal stiffness of discontinuities, Eq. (15) can be rearranged into the following form: rp  g 3 M1 M1 1 l Kn 3 cos 2b ge p (16) 12f n 2 2 d 3 T =L=dh The rst term in the right side of Eq. (16) is a constant, the second term can be easily obtained by using the correlation between RQD and discontinuity frequency, and the fourth term can be acquired easily from a certain amount of packer test data at different depths by using regression analysis. When the packer test is conducted on a parallel set of discontinuities, the result from Eq. (16) is the average normal stiffness. Therefore, for a site where conductance of ow is controlled by a set of discontinuities with a steady dip angle, even with complicated stress conditions, this method is suitable. For a site where the horizontal stress approximates to the vertical stress, i.e., l 1, Eq. (16) can be simplied as rp g 3 1 Kn 3 lge p (17) 12f n d 3 T =L=dh

120

50

60

70 80 RQD (%)

90

100

Fig. 5. The variations in mean hydraulic conductivity and mean RQD with depth.

0.1 K (m/d)

K = 0.4892exp(-0.0543*RQD) R2 = 0.7809

0.01

0.001 50 60 70 80 RQD (%) 90 100

Fig. 6. The scatter plot of mean hydraulic conductivity versus mean RQD.

Fig. 4 is signicantly weaker than that shown in Fig. 6. This could be explained by the different factors inuencing RQD and hydraulic conductivity. As mentioned before, the RQD calculated from borehole core is only a measure of discontinuity frequency or spacing (although RQD may be affected by the drilling process, this is not considered in the current research), while the hydraulic conductivity of a test section is dependent on the aperture, frequency, orientation, roughness, and interconnectivity of discontinuities. As a result, it is hardly possible to estimate hydraulic conductivity at a specic position based on the corresponding RQD only. However, based on the strong correlation between mean hydraulic conductivity and mean RQD, when one needs to estimate the mean hydraulic conductivity at certain depths, it is probable by using the corresponding mean RQD. This indicates the strong inuence of discontinuity frequency on hydraulic conductivity. For a site with a relatively uniform lithology but without signicant tectonic events, a decrease in discontinuity frequency with depth would lead to an increase in mean RQD with depth, and would partly contribute to the decrease in mean hydraulic conductivity with depth.

ARTICLE IN PRESS
X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377 1375

RQD = 0.2875 h + 65.751 R2 = 0.8053

Kn (GPa/m) 20 10 20 30 40 50 60 70 80
Fig. 9. The values of calculated normal stiffness at different depths.

100 90 RQD (%)

60

100

140

180

70 60 50 10 30 50 Depth (m) 70 90

Fig. 7. The relationship between RQD and depth.

0.3 T(1/3) ((m2/d)(1/3)) 0.25 0.2 0.15 0.1 1

T(1/3) = -0.1116log h + 0.3695 R2 = 0.8826

1.2

1.4 1.6 log (depth) (m)

1.8

Fig. 8. The relationship between transmissivity and depth.

In this case, the method can provide the average normal stiffness of several sets of discontinuities. 5.2. Results of normal stiffness by incorporating the decrease in transmissivity with depth and the increase in RQD with depth Based on the discussions in Section 4, it is assumed that when the depth ranges between 11 and 88 m, the RQDdepth trend is caused by the change in discontinuity frequency with depth, and the transmissivitydepth trend is caused by the depth dependent discontinuity frequency as well as the stress aperture. The relationship between RQD and depth is tted using the linear regression model (Fig. 7). The equation is RQD 0:2875h 65:751 (18)

Considering the suitability of Eq. (6), h ranges from 11 to 80 m. Physical properties, r and u, of groundwater are assumed equal to that of water at standard condition. Witherspoon et al. [35] found out that f ranged from 1.04 to 1.65. Jiang et al. [6] had discussed that the choice of f would not signicantly affect the results. So, f is selected to be 1.5 here. The specic weight had been measured to be about 26.5 103 N/m3; however, since the water table is shallow in the study area, the effective specic weight should be employed. As a result, ge is chosen to be 16.5 103 N/m3. The result of average normal stiffness versus depth is shown in Fig. 9. Compared with the plot of fracture normal stiffness versus normal stress for 13 different rock samples collected and compiled in [36], which shows that the slope (normal stiffness versus stress) decreases with stress, our result shown in Fig. 9 is realistic. This is a great advantage over the results in the former research [6], which showed that the normal stiffness is linearly proportional to depth. The reason is that, in [6], the change in discontinuity spacing (frequency) with depth was not considered. However, in the current study, the increase in discontinuity spacing with depth, or the decrease in discontinuity frequency with depth, has been accounted for by using RQD data, which is correlated with discontinuity spacing (frequency). 5.3. Estimation of rock mass deformation modulus from normal stiffness using the equivalent models As mentioned in Section 2.5, the deformation modulus of rock masses can be estimated by using deformation modulus of intact rocks, which can be easily obtained from indoor measurements on specimens, and normal stiffness of discontinuities, which has been calculated in Section 5.2. A few measurements of deformation modulus of intact rocks, which were sampled in zones with different degree of weathering, have been carried out. The mean of deformation modulus of intact rocks in the highly weathered zone is 1.6 GPa, the mean of deformation modulus of intact rocks in the moderately weathered zone is 18.5 GPa, and the mean of deformation modulus of intact rocks in the weakly weathered zone is 44.5 GPa. Unfortunately, the change in deformation modulus of intact rocks with depth is not available since the depths of rock samples are not known to the authors. Numerous researchers have reported that deformation modulus, or compressibility, which is dened as the reciprocal of deformation modulus, of rocks is stress dependent [3741]. Research has also revealed that deformation modulus of intact rocks is correlated with the degree of weathering [42]. As pointed out in [41], it is considered that the deformation modulus of intact rocks increases exponentially with depth. Since the range of depth is 1180 m, and the range of deformation modulus is 1.644.5 GPa, a synthetic line showing the exponential increase in deformation modulus of intact rocks with depth is provided in Fig. 10a.

The relationship between transmissivity (with a resolution of 1 m) and depth is also tted using regression analysis based on the Swan model [34] (Fig. 8). The equation is in the form of p 3 (19) T 1 A1 log h A2 where T1 is the transmissivity data with a resolution of 1 m, A1 is estimated to be 0.1116 (m2/d)1/3, and A2 is estimated to be 0.3695 (m2/d)1/3. When Eq. (6) is applied, Eq. (17) can be expressed in the following form: r g p ln 10 h 3 30 RQD=3:68ge p (20) Kn 3 12f n d 3 T =L=dlog h Substituting Eqs. (18) and (19) into Eq. (20) yields the following empirical equation for calculation of normal stiffness in the study area: rp g 3 ln 10 h (21) 12:13 0:078hge Kn 3 12f n A1

Depth (m)

80

ARTICLE IN PRESS
1376 X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377

Er (GPa) 0 10 20 Depth (m) 30 40 50 60 70 80 Emn (GPa) 0 10 20 Depth (m) 30 40 50 60 70 80


Fig. 10. The values of deformation modulus at different depths: (a) the deformation modulus of intact rocks versus depth based on an assumed model and (b) the calculated deformation modulus of the rock mass versus depth using Kulhawys [26] equivalent model.

Em (GPa)

10

20

30

40

50
10 20 30 40 50 60 70 80

10

20

30

40

50

mean lower bound upper bound Emn in Fig. 10b

10

15

20

Fig. 11. The results of rock mass deformation modulus versus depth from the two different methods.

By using the deformation modulus of intact rocks (shown in Fig. 10a), normal stiffness (shown in Fig. 9), and discontinuity frequency, which are all depth-dependent, the average deformation modulus of the rock mass at each depth can be calculated by using Eq. (9) (Fig. 10b). Fig. 10b shows that the deformation modulus of the rock mass increases with depth, which is in accordance with the results of others [43,44]. Moreover, three sets of in situ measurements of rock mass deformation modulus had been performed in a tunnel, whose depth is generally greater than 50 m. The results of in situ deformation modulus are 12.83, 17.99, and 19.24 GPa, respectively. Compared with the results shown in Fig. 10b, the in situ measurements are close enough to the estimated deformation modulus in the depth range of 5080 m. 5.4. Verication of results by deformation modulus estimated from RQD using the empirical models Since in situ tests are usually expensive and time-consuming, it is quite common that only limited in situ measurements of deformation modulus are carried out. Therefore, validation of the estimated deformation modulus, which is depth dependent, using such a small number of in situ measurements is not convincing enough. Here, the empirical model proposed by Zhang and Einstein [14], which relates RQD with deformation modulus, is utilized to compare with the results from the method proposed in this paper. Based on the RQDEm/Er data reported in [13,15,45], Zhang and Einstein [14] recommended the following equation for estimation of deformation modulus using RQD: lower bound : Em =Er 0:2 100:0186RQD1:91 upper bound : Em =Er 1:8 100:0186RQD1:91 mean : Em =Er 10
0:0186RQD1:91

Based on Eq. (22), the deformation modulus of the rock mass at each depth can be estimated by using the deformation modulus of intact rocks, which is shown in Fig. 10a, and the RQD values, which are calculated from Eq. (18). The results are shown in Fig. 11. For convenience of comparison, the deformation modulus obtained in Section 5.3 is also shown. As shown in Fig. 11, the result estimated using the method proposed by the authors in the current study is not exactly the same as that obtained using the empirical equation derived in [14]. Fortunately, both of them obey the fact the deformation modulus of a rock mass should increase with depth due to factors such as changes in stress, weathering degree, and discontinuity aperture and spacing (frequency) with depth. Furthermore, in the depth range studied, the values obtained by the two methods are close enough. It is necessary to mention that this comparison also has limitations. In Zhang and Einsteins empirical model [14], RQD (correlated with discontinuity frequency) is the only fact that has been taken into account. In the authors model, changes in RQD (correlated with discontinuity frequency) and permeability (dependent on discontinuity frequency and aperture) with depth are both employed. Even if the increase in RQD (or discontinuity spacing) with depth is not signicant at this particular site, the depth dependency of deformation modulus could still be revealed by the authors model. This situation is similar to Jiang et al.s former model [6], which did not consider the change in discontinuity spacing with depth. However, if the increase in RQD (or discontinuity spacing) with depth is not signicant, the value of Em/Er in Eq. (22) would approximate to a constant. A better way to validate the model proposed by the authors would be to use geophysical methods, for example, run some sonic logs in the boreholes and compare with the observed trends in sonic velocity with depth. However, this is beyond the scope of research in the current paper. 6. Conclusions In the current study, changes in discontinuity aperture and frequency are considered simultaneously to estimate discontinuity normal stiffness from hydraulic tests. Transmissivity, which is

(22a) (22b) (22c)

Depth (m)

ARTICLE IN PRESS
X.-W. Jiang et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 13701377 1377

dependent on both aperture and frequency of discontinuities, is obtained from packer test. Discontinuity frequency is calculated from RQD, which is obtained from borehole data. Due to their interrelationship, the normal stiffness is successfully determined by using the decrease in transmissivity with depth and the increase in RQD with depth. Finally, the deformation modulus of the rock mass is estimated and compared with that obtained from in situ measurements and empirical equations. The following conclusions can be drawn. (1) In a study area mainly controlled by lithostatic stress, a decrease in permeability with depth and an increase in RQD with depth can be identied. It is difcult to predict the permeability in a position based on RQD only; however, the mean permeability can be estimated by the mean RQD. Based on the relationship between RQD and discontinuity frequency, the average increase in RQD with depth can be employed to derive the average decrease in discontinuity frequency with depth. (2) The discontinuity normal stiffness can be estimated by the decrease in permeability and the decrease in discontinuity frequency with depth, which can be obtained from packer test and RQD data, respectively. The calculated normal stiffness, which increases nonlinearly with depth, is consistent with results from experimental studies. (3) The normal stiffness has been utilized to calculate deformation modulus of the rock mass, and the result is close enough to the deformation modulus obtained from in situ measurements as well as from empirical equations. However, the methods currently employed for comparison have limitations. Better ways are needed to validate the results of deformation modulus of large scale rock masses.

Acknowledgments The nancial support by the National Natural Science Foundation of China under contracts (40528003) and (50639090) are greatly acknowledged. The authors are also grateful to two anonymous reviewers, whose comments were helpful in improving this paper. References
[1] Gokceoglu C, Yesilnacar E, Sonmez H, Kayabasi A. A neuro-fuzzy model for modulus of deformation of jointed rock masses. Comput Geotech 2004;31:37583. [2] Bandis S, Lumsden AC, Barton NR. Fundamentals of rock joint deformation. Int J Rock Mech Min Sci Geomech Abstr 1983;20:24968. [3] Goodman RE. Method of geological engineering in discontinuous rocks. New York: West Publishing; 1976. [4] Chen DW. Coupled stiffness-permeability analysis of a single rough surfaced fracture by the three dimensional boundary element method. PhD dissertation, University of California, Berkeley; 1990. [5] Rutqvist J. Determination of hydraulic normal stiffness of fractures in hard rock from well testing. Int J Rock Mech Min Sci Geomech Abstr 1995;32(5):51323. [6] Jiang XW, Wan L, Wang XS, Liang SH, Hu BX. Estimation of fracture normal stiffness using a transmissivitydepth correlation. Int J Rock Mech Min Sci 2008; doi:10.1016/j.ijrmms.2008.03.007. [7] Snow DT. Rock fracture spacings, openings, and porosities. J Soil Mech Found Div ASCE 1968;94(SM1):7391. [8] Louis C. Rock hydraulics. In: Muller L, editor. Rock mechanics. Vienna: Springer; 1974. p. 299387. [9] Stober I, Bucher K. Hydraulic properties of the crystalline basement. Hydrogeol J 2007;15:21324. [10] Jiang XW, Wan L, Song G, Wang XS. The permeability of basalts and shear zones. Geotechnical Investigation Survey 2008;231:259 [in Chinese with English abstract].

[11] Deere DU. Technical description of rock cores for engineering purposes. Rock Mech Eng Geol 1963;1:1822. [12] Choi SY, Park HD. Variation of rock quality designation (RQD) with scanline orientation and length: a case study in Korea. Int J Rock Mech Min Sci 2004;41:20721. [13] Coon RF, Merritt AH. Predicting in situ modulus of deformation using rock quality indices. In: Determination of the in situ modulus of deformation of rock, ASTM STP 477. Philadelphia: ASTM; 1970. p. 15473. [14] Zhang L, Einstein H. Using RQD to estimate the deformation modulus of rock masses. Int J Rock Mech Min Sci 2004;41(2):33741. [15] Bieniawski ZT. Determining rock mass deformabilityexperience from case histories. Int J Rock Mech Min Sci Geomech Abstr 1978;15:23748. [16] Nicholson GA, Bieniawski ZT. A nonlinear deformation modulus based on rock mass classication. Int J Min Geol Eng 1990;8:181202. [17] Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Rock Mech Min Sci 1997;34(8):116586. [18] Chun BS, Ryu WR, Sagong M, Do JN. Indirect estimation of the rock deformation modulus based on polynomial and multiple regression analyses of the RMR system. Int J Rock Mech Min Sci 2009;46:64958. [19] Barton N. Application of Q-system, index tests to estimate shear strength and deformability of rock masses. In: Proceedings of international symposium on engineering geology underground construction, vol. 1(II). Lisbon, 1983. p. 5170. [20] Barton N. Some new Q value correlations to assist in site characterisation and tunnel design. Int J Rock Mech Min Sci 2002;39:185216. [21] Priest SD, Hudson JA. Discontinuity spacings in rock. Int J Rock Mech Min Sci Geomech Abstr 1976;13:13548. [22] Wallis PF, King MS. Discontinuity spacing in a crystalline rock. Int J Rock Mech Min Sci Geomech Abstr 1980;17:636. [23] Poulton MM, Mojtabai N, Farmer IW. Scale invariant behavior of massive and fragmented rock. Int J Rock Mech Min Sci Geomech Abstr 1990;27: 21921. [24] Snow DT. Anisotropic permeability of fractured media. Water Resour Res 1969;5(6):127389. [25] Gale JE, Rouleau A, Atkison LC. Hydraulic properties of fractures. In: Proceedings of 17th congress of the International Association of Hydrogeology. Tuscon, 1985, p. 116. [26] Kulhawy FH. Geomechanical model for rock foundation settlement. J Geotech Eng ASCE 1978;104:21127. [27] Gerrard CM. Elastic models of rock masses having one, two and three sets of joints. Int J Rock Mech Min Sci Geomech Abstr 1982;19:1523. [28] Neuman SP. Trends, prospects and challenges in quantifying ow and transport through fractured rocks. Hydrogeol J 2005;13(1):12447. [29] National Research Council. Rock fracture and uid ow. Washington DC: National Academy Press; 1996. [30] National Standards of PRC. Code of water pressure test in borehole for hydropower and water resources engineering. DL/T 53312005. [31] Rutqvist J, Stephansson O, Tsang CF. Uncertainty in estimate of maximum principal stress from hydraulic fracturing due to the presence of the induced fracture. Int J Rock Mech Min Sci 2000;37:10720. [32] Liao YX. A quantitative study on the zonation of weathered rock masses. Beijing: Earthquake Publishing House; 1994 [in Chinese]. [33] El-Naqa A. The hydraulic conductivity of the fractures intersecting Cambrian sandstone rock masses, central Jordan. Environ Geol 2001;40:97382. [34] Swan G. Determination of stiffness and other joint properties from roughness measurements. Rock Mech Rock Eng 1983;18(1):1938. [35] Witherspoon PA, Wang JSY, Iwai K, Gale JE. Validity of cubic law for uid ow in a deformable rock fracture. Water Resour Res 1980;16:101624. [36] Pyrak-Nolte LJ, Morris JP. Single fractures under normal stress: the relation between fracture specic stiffness and uid ow. Int J Rock Mech Min Sci 2000;37:24562. [37] Adams LH, Williamson ED. The compressibility of minerals and rocks at high pressures. J Franklin Inst 1923;195:475592. [38] Fatt I. Compressibility of sandstones at low to moderate pressures. AAPG Bull 1958;42(8):192457. [39] Zimmerman RW. Compressibility of sandstones. Amsterdam: Elsevier; 1991. [40] Jaeger JC, Cook NGW, Zimmerman RW. Fundamentals of rock mechanics. Oxford: Blackwell; 2007. [41] Liu HH, Rutqvist J, Berryman JG. On the relationship between stress and elastic strain for porous and fractured rock. Int J Rock Mech Mining Sci 2009;46(2):28996. [42] Cai MF. Rock mechanics and engineering. Beijing: Science Press; 2002 [in Chinese]. [43] Verman M, Singh B, Viladkar MN, Jethwa JL. Effect of tunnel depth on modulus of deformation of rock mass. Rock Mech Rock Eng 1997;30(3):1217. [44] Asef MR, Reddish DJ. The impact of conning stress on the rock mass deformation modulus. Geotechnique 2002;52(4):23541. [45] Ebisu S, Aydan O, Komura S, Kawamoto T. Comparative study on various rock mass characterization methods for surface structures. In: Proceedings of Eurock92. London: Thomas Telford; 1992. p. 2038.

You might also like