A STUDY ON ESTIMATING HYDRAULIC CHARACTERISTICS OF ROCK SPECIMENS USING ELASTIC WAVE

Ayumi MIYATA1, Yuzo OHNISHI1, Satoshi NISHIYAMA1, Takao YANO1 and Manabu TAKAHASHI2 Graduate School of Engineering, Kyoto University, Kyoto, Japan (e-mail of corresponding author: miyata@geotech.kuciv.kyoto-u.ac.jp)
2 1

Research Center for Deep Geological Environments, AIST

We develop the method to estimate permeability distribution in rock mass from elastic wave velocity. This method is based on dynamic poroelastic theory such as Biot and BISQ theory. These theories indicate that the elastic wave velocity through elastic solid saturated with pore water depends on its frequency and formulated with permeability with the characteristic frequency. Using these theories, we can estimate hydraulic characteristics from the characteristic frequency. In order to verify the applicability of these theories to rock mass, we performed laboratory tests. We measured P-wave velocity of rock specimens (shale, tuff, sandstone and granite) for various frequencies from 30 kHz to 1100 kHz. From these results, it was found that elastic wave dispersion for sedimentary rocks can be described by Biot theory and that for crystalline rocks can be described by BISQ theory. Keywords: elastic wave; rock specimens; permeability; Biot theory; BISQ theory

1.

Introduction

Recently, underground space has been utilized for the purpose to dispose such as geological disposal of radioactive wastes and CO2. In order to evaluate safety of geological disposal, it is important to know how ground water flows through rock masses. Various investigations (for example, permeability tests in borehole or geological explosion, etc) have been conducted in order to estimate hydraulic characteristics in rock mass. However, domain of borehole tests is limited and hydrogical charactaristics can not be gained by geological explosion. We develop the method to estimate permeability distribution in rock masses from elastic wave velocity. It is new geophysical explosion technique in which permeability distribution in rock mass can be evaluated directly and extensively. This method is based on dynamic poroelastic theory such as Biot and BISQ theory. These theories indicate that the elastic wave velocity through elastic solid saturated with pore water depends on its frequency and formulated with permeability with the characteristic frequency. We calculate P-wave velocity dispersion from the data measured for various frequencies among boreholes. Then we can estimate rock permeability. However, there has been no study that tried to apply poroelastic theories such as Biot and BISQ theory to rock mass. In this study, we performed the laboratory tests for some kind of rock specimens (shale, tuff, sandstone and granite) and verify the applicability of these theories to them. 2. Basic Theory

When the elastic wave propagates in the rock saturated with water, elastic wave velocity depends on frequency because of the interaction of the solid skeleton and fluid. This phenomenon can be explained by the poroelastic theory. There are two theories in the poroelastic theory about propagation of elastic wave.

2.1. Biot Theory Biot1)2) suggested the theory of propagation of elastic wave in a fluid-saturated porous media. In this theory, it is described that elastic wave velocity has a dependency on its frequency. In Biot model, interaction between solid and fluid is affected by viscous friction and inertial coupling. It is assumed that solid and fluid move only in the direction of elastic wave propagation (x-direction) . In Biot theory, dependency of elastic wave velocity on frequency is described by the interaction between solid and fluid. In low frequency, solid and fluid move together because viscous force is relatively higher than inertial force. It means that elastic wave is propagated by solid-fluid aggregate, so elastic wave velocity becomes low. In high frequency, solid and fluid move separately because inertial force is relatively higher than viscous force. It means that elastic wave propagated by solid, so seismic velocity becomes high. The mathematical implementation of Biot theory is based on the equation of solid-fluid dynamic interaction. From the stress-strain relationship in poloelastic media, Biot derived dynamic equation for solid and fluid when elastic wave propagates. In this equation, it is assumed that relative motion between solid and fluid is Poiseuille flow.
2u 2U x (u x U x ) x = 1 2x + a +b 2 x t t t 2u x 2U x (u x U x ) s = 2 a b 2 2 x t t t
b = 2 k

(1) (2) (3)

x : the stress acting on the solid pert of the rock in the x-direction, :porosity, 1 :density of solid, 2 :density of fluid, a :mass coupling coefficient, b : viscous coupling coefficient, :fluid viscosity, k :permeability , u x,U x :the displacements of the solid part and pore fluid in the x-direction, s = P , P : pore pressure
Then seismic velocity is derived in the form that depends on frequency. Biot considered characteristic frequency f c which is characterized as a transition domain from low velocity to high frequency. f c is described in the form that depends on permeability.

fc =

2k f
Solid and fluid move Separately

(4)

Solid and fluid move together

Low frequency

High frequency

Figure.1. Interaction between solid and fluid

2.2. BISQ Theory In BISQ theory3), it is assumed that solid skeleton of the rock deforms in the direction of wave propagation. The fluid, however, can flow not only parallel (x-direction), but also perpendicular (r-direction) to the direction of wave propagation. The fluid flow to x-direction called Biots flow and it to r-direction called squirt flow (Fig.2). Traditionally, both flows have been modeled separately. However in BISQ theory, the model that treats those two flows simultaneously is suggested (BISQ model). Consider a representative cylindrical volume of the rock with the axis in the x-direction which stands for the local flow radius. The radius of this cylinder is the average length that gives a squirt-flow effect identical to the cumulative effect of local flow in pores of various shapes and sizes. BISQ model assumed that pore pressure on the external surface of this cylinder does not change in time, so if pore is compressed or extended by the elastic wave propagation, pressure gradient appear inside of rock and squirt flow occur. In the BISQ model, the effect of squirt-flow appears in pore pressure. To use dynamic equation described by Biot, Dvorkin describes pore pressure as an averaged value with respect to r. The mathematical implementation of BISQ theory becomes same as Biot theory by substituting fluid pressure P in Biot theory with averaged fluid pressure respect to r. From the BISQ theory, seismic velocity is also derived in the form that depends on frequency and f c which is characterized as a transition domain from low velocity to high frequency is also described in the form that depends on permeability. (Fig.3)

fc =

Kfk 2L2

(5)

( K f :bulk modulus of the fluid)

Figure 2. Biots flow and Squirt flow

Figure 3. Relation between elastic wave velocity and frequency, and characteristic frequency

3.

Laboratory Tests 3.1. Experimental apparatus

Fig.4 shows the outline of experimental apparatus. We propagated the elastic wave with pulse generator, and measured the waveform with the oscilloscope.

3.2. Rock specimens The rock specimens used in this study were sandstone, tuff, shale, a kind of sedimentary rock and Inada-granite, a kind of crystalline rock. In order to create thermal crack, we put specimens of Inada granite under temperatures of 300 and 600. For comparison with rock specimens, we use stainless steel specimen, which has no porosity and no interaction with pore fluid. For the experiment, they were fabricated at pillar specimens with a diameter of 50mm and height of 100mm. Each specimen was saturated with water using the carbon dioxide replacing method. 3.3. Experimental procedure In order to prevent attenuation of the elastic wave by coupling, we applied echo jelly (for medical treatment) to the upper and under surface of specimen, and put the transducers. We propagate elastic wave for several frequencies from 30kHz to 1100kHz and read an initial shock position on the oscilloscope.
AMP
Pulse generator Oscillator

Specimen Digital oscilloscope Reciever

AMP

Fig4. The outline of measurement system

3.4. Results of experiments Fig.5 shows the results of the measurements. For all rock specimens, the velocity-frequency dispersion was observed as the velocity was greater at higher frequency, and this tendency was clearer than that for the specimen of stainless steel.
5900 5880 5860 5840 5820 5800 5780 5760 5740 5720 5700 0 2230 2220 2210 2200 2190 2180 2170 2160 2150 2140 10 3260 3240 3220 3200 3180 3160 3140 3120

Elastic wave velocity(m/s)

Elastic wave velocity(m/s)

(a)

(b)

200

400

600

800

1000

1200

10

100
Frequency(kHz)

1000

10000

Frequency(kHz)

Elastic wave velocity(m/s)

Elastic wave velocity(m/s)

(c)

2120 2100 2080 2060 2040 2020 2000 1980 1960 1940 1920

(d) (a)

100
Frequency(kHz)

1000

10000

10

100

1000
Frequency(kHz)

10000

Elastic wave velocity(m/s)

6000 5500 5000 4500 4000 3500 3000 2500 10

(e)
intact 300 600

100
Frequency(kHz)

1000

10000

Fig.5.Result of the experiment of (a) stainless steel, (b)sandstone, (c)shale,(d) tuff and (e)Inada granite (intact, 300 and 600)

4.

Examination of applicability of Biot and BISQ model

We examined if the observed velocity dispersion curves can be explained by Biot and BISQ model. We measured properties of specimens which are necessary to estimate dispersion relationships by Biot and BISQ models (Table 1). We compared the permeability measured by permeability test with the permeability estimated by the theories. Permeability was measured directly by falling head permeability test for Inada granite specimen. Then we consider the possibility to estimate hydraulic characteristics by elastic wave dispersion. We group the rock specimens in sedimentary rocks and crystalline rocks and provide further insights into application of the theories. It is presupposed that fractured media is treated in BISQ model. Therefore we make a study on application of Biot model for sedimentary rocks and application of Biot and BISQ model for crystalline rocks.
Table 1. Physical properties on fitting to result of the experiment

Sample Ks(N/m2) E(N/m2) G(N/m2) Sandstone 1.10E10 1.43E10 5.14E09 0.322 0.2417 Shale 8.00E09 4.59E09 1.67E09 0.375 0.3792 Tuff 8.00E09 1.55E09 5.40E09 0.440 0.5553 Inada granite (intact) 6.50E10 3.42E10 1.32E10 0.300 0.0047 Inada granite (300) 6.50E10 3.13E10 1.25E10 0.253 0.0071 Inada granite (600) 6.50E10 7.10E09 2.87E09 0.237 0.0550 E:elasitic coefficient, Ks:elasitic coefficient (solid), G:shear moduli:Poissons ratio,:porosity 4.1. Consideration of sedimentary rocks In the case of sandstone, tuff and shale, we compared the Biot theoretical solution with measured velocity. Biot theoretical solution was obtained with measured physical properties (Table 1). We found that variation of velocity fitted into the theoretical solution (Fig.6). Then assuming that characteristic frequency was in the area estimated from the result of measurement, we evaluate the domain of permeability coefficient estimated by Biot theory. The area of elastic wave velocity dispersion was estimated by setting the upper and lower limit of elastic wave velocity. In the case of shale, the measured permeability coefficient was contained in the domain of the estimated one. As to tuff and sandstone, the measured permeability coefficient was not contained in the domain of the estimated one. But the estimated permeability coefficient

was close in value to the measured one. Fig.7 shows the example of estimating the area of elastic wave velocity dispersion. Table 2 shows estimated and measured permeability coefficient.
3280 Elastic wave velocitym/s Elastic wave velocitym/s 3260 3240 3220 3200 3180 3160 3140 3120 3 10 10
4

2240

2100 Elastic wave velocitym/s

(a)

2230 2220 2210 2200 2190 2180 2170 2160 2150

(b)

2080 2060 2040 2020 2000 1980 1960

(c)

Measured results Biot theoretical solution

10
5

10 10 FrequencyHz

10

10

2140 3 10

10

10

10 10 FrequencyHz

10

10

1940 3 10

10

10

10 10 FrequencyHz

10

10

Fig.6. Comparison with measured result and Biot theoretical solution of (a) sandstone, (b)shale and (c)tuff

2120 Elastic wave veloitym/s 2100 2080 2060 2040 2020 2000 1980 1960 1940 1920 4 10 10
5

Measured results The upper and lower limit of elastic wave velocity. The estimated domain of characteristic frequency

10 FreauencyHz

10

10

Fig.7. Estimation of the domain of characteristic frequency (Tuff) Tabel 2. Estimated and measured permeability coefficient.

characteristic frequency(Hz) Estimated permeability coefficient(m/s) Measured permeability coefficient(m/s)

Sandstone 1.00e05 5.00e06 8.67e-04 1.73e-07

Shale 1.00e05 1.00e07 3.77e-06 3.77e-08

Tuff 1.00e05 5.92e-06 2.00e07 2.96e-08

1.16e-07

2.61e-08

1.0e-07~e-08

4.2. Consideration of crystalline rocks 4.2.1. Application of Biot theory

In the case of 3 kind of Inada granite (intact, 300 and 600), the Biot theoretical solution was far from the measured elastic wave velocity. (Fig.8) From the results, it was found that measured variation of elastic wave is not explained by the Biot theory. Therefore Biot theory cannot apply crystalline rocks.

5550 Elastic wave velosctym/s Elastic wave velocitym/s 5500 5450 5400 5350 5300 5250 5200 4 10 10
5 6

5450

3900 Elastic wave velocitym/s

(a)

5400 5350 5300 5250 5200 5150 5100 5050

(b)

3800 3700 3600 3500 3400 3300 3200

(c)

Mesured result Biot theoretical solution

10 10 FrequencyHz
7

10

10

5000 4 10

10

10 10 FrequencyHz

10

10

3100 3 10

10

10 10 FrequencyHz

10

10

Fig.8.Comparison with measured result and Biot theoretical solution of Inada granite ((a)intact, (b)300 and (c)600)

4.2.2.

Application of BISQ theory

In the same way, we compared the BISQ theoretical solution with measured elastic wave velocity. In all cases, BISQ theoretical solution was compliant in terms of variation of velocity (Fig.9).There was possibility to apply BISQ theory to the crystalline rocks. Allowing for the fact mentioned above, we tried to estimate the permeability coefficient by BISQ theory. Then we should consider Squirt flow length, unique parameter in BISQ theory. We used the size from rock inner structure as the value which is equivalent to Squirt flow length (Maximum void size, Average void size, Mode value of void size and Grain size) and compared the estimated permeability with measured one. In consequence, in the case of Inada granite of the condition of intact and heat-treated at a temperature of 300, considering the maximum void size as Squirt flow length, permeability coefficient was well estimated. In the case of Inada granite heat- treated at a temperature of 600, considering average grain size as Squirt flow length, the estimate permeability was well estimated, too. Fig.10 shows that the result of comparison of measured and estimated permeability coefficient Therefore, it seems reasonable to conclude that deciding the value of Squirt flow length in view of rock structure and connectedness among the crack, it is possible to estimate the permeability using BISQ theory.
5800 Elastic wave velositym/s Elastic wave velositym/s 5700 5600 5500 5400 5300 5200 3 10 5700 4200 Elastic wave velositym/s

(a)

5600 5500 5400 5300 5200 5100 5000

(b)

4000 3800 3600 3400 3200 3000 3 10

(c)

:Measured result BISQ theoretical solution

10
4 5 6 7 8 9 10 11

10

10 10 10 FrequencyHz

10

10

10

4900 3 10

10

10

10 10 10 FrequencyHz

10

10

10

10

10

10 10 10 FrequencyHz

10

10

10

Fig.9.Comparison with measured result and BISQ theoretical solution of Inada granite ((a)intact, (c)600)

(b)300 and

1.00E-04 1.00E-06

intact

300

600

Permeability(m/s)

1.00E-08 1.00E-10 1.00E-12 1.00E-14 1.00E-16 1.00E-18

Grain size Maximum void size

Average void size

Mode value of void size Falling head permeability test results

Fig.10. Relationship between measured and estimated permeability coefficient

5.

Conclusions and Further issues

In this study, we performed the laboratory test for the purpose to develop the method of estimating hydraulic characteristics of rock masses by elastic wave dispersion. From the result of laboratory tests, the velocity-frequency dispersion was observed as the velocity is greater at higher frequency for all rock specimens. Comparing measurement results with theoretical solutions, it was found that elastic wave dispersion for sedimentary rocks can be described by Biot theory and that for crystalline rocks can be described by BISQ theory, namely applicable theory was different by rock structure. For sedimentary rocks, we can guess the area of characteristic frequency from measured elastic wave dispersion and calculate the permeability coefficient with Biot theory. Since the calculated value was close to the measurement value, we may say that it is possible to estimate the hydraulic characteristics from elastic wave dispersion. On the other hand, for crystalline rocks, deciding the value of Squirt flow length in consideration of void structure, we found that it is possible to estimate the permeability coefficient in a similar way. Furthermore, it is necessary to observe the inner structure of test specimens and to take the condition of crack heated into account. We will estimate hydraulic characteristics more exactly in field test by using elastic wave. It is a further study how the value of Squirt flow length in field test will be evaluated.

References 1) Biot, M.A.: Theory of propagation of elastic waves in a fluid-saturated porous solid. I. Low-frequency range, J. Acoust. Soc. Am., Vol. 28, pp. 168-178, 1956. 2) Biot, M.A.: Theory of propagation of elastic waves in a fluid-saturated porous solid. II. Higher frequency range, J. Acoust. Soc. Am., Vol. 28, pp. 179-191, 1956. 3) Dvorkin, J. and Nur, A: Dynamic poroelasticity: a unified model with the squirt and the Biot mechanism, Geophysics, Vol. 58, pp. 524-533, 1993.