A comparison of short-circuited streaming potentials in Westerly granite from changes in the rock's volume, shape, saturation, and fracture under unconfined uniaxial compression

A Comparison of Short-Circuited Streaming

Potentials in Westerly Granite From Changes in

the Rock's Volume, Shape, Saturation, and

Fracture Under Unconfined Uniaxial

Compression

by

Elizabeth Annah Jensen

Submitted to the Department of Earth, Atmosphere, and Planetary

Science

in partial fulfillment of the requirements for the degree of

Master of Science in Geosy

at the

MASSACHUSETTS INSTITUTE OF

stems

TECHNOLOG1999

TECHNOLOG x

@ Massachusetts Institute

May 1999

of Technology 1999.

LW49-w

All rights reserved.

... .... .. ... . . .... ... . . ...

Department of Earth, Atmosphere, and Planetary Science

April 23, 1999

trtified by

Frank Dale Morgan

Professor

Thesis Supervisor

rtified by...

Chris J. Marone

Professor

Thesis Supervisor

A ccepted by ...

Ronald G. Prinn

Department Head

Author

A Comparison of Short-Circuited Streaming Potentials in

Westerly Granite From Changes in the Rock's Volume,

Shape, Saturation, and Fracture Under Unconfined

Uniaxial Compression

by

Elizabeth Annah Jensen

Submitted to the Department of Earth, Atmosphere, and Planetary Science on April 23, 1999, in partial fulfillment of the

requirements for the degree of Master of Science in Geosystems

Abstract

An experiment was designed to gain some insight into the phenomena of electrical charge in the uniaxially stressed environment of a sample of Westerly granite. Vary-ing properties of size, shape, saturation, and stress rate of the Westerly granite, a tentative set of physical relationships were measured. For instance, there is a strong suggestion that the amount of acoustic energy released when the sample fractures is proportionately related to the amount of charge that moves into or out of the sample, the net charge, over the period of the experiment. Increasing the concentration of the saturating solution increased the amount of charge that moved into or out of the sample. With all the samples that were dry, more charge move in than out over the period of the experiment. The few samples that were alike in every way before and after they fractured had a difference in charge and acoustic energy by a magnitude of their measured net charge and acoustic energy.

Thesis Supervisor: Frank Dale Morgan Title: Professor

Thesis Supervisor: Chris J. Marone Title: Professor

Acknowledgments

Deborah Jensen, Edward Jensen, and Patricia Chapela for their unparalleled emo-tional, financial, and editing support;

Zhenya Zhu for his careful and patient instruction in the experimental setup pro-cedure;

Dale Morgan for strategic guidance and research support;

Chris Marone for allowing me to use (and occasionally break) all the equipment I needed in his lab;

Steve Karner and Karen Mair for their assistance in helping me to avoid damage to the biax and in processing load data;

Karyn Green, Toby Kessler, and Robert Fleming for technical support and track-ing down obscure references;

Kate Lehetola and Mary Krasovec for helping me keep a sane mind in a sound body;

Youshun Sun for all the support and guidance a former Geosystems student could give;

Philip Reppert for filling in the gaps of my knowledge while I was writing my thesis;

and most especially, Rebecca Saltzer, for giving me both good advice and infor-mation about where to find good advice on the Geosystems program elements.

Biographical Note

Education

Massachusetts Institute of Technology 1998-1999 SM Geosystems Texas A&M University 1993-1998 BS Geophysics

Texas A&M University Honors

F. Vilas, E. A. Jensen, and L. A. McFadden. Extracting Spectral Information about 253 Mathilde Using the NEAR Photometry. Icarus, 129(2):440+, 1997.

F. Vilas, E. A. Jensen, D. L. Domingue, L. A. McFadden, and C. R. Coombs. An Unusual Photometric Signature Detected on Lunar Complex Crater Rims at the South Pole. In Lunar and Planetary Science XXIX, Lunar and Planetary Institute, Houston (CD-ROM).

E. A. Jensen. Satellite Observations of Oceanic Shelf-Slope Exchange. University Graduate Research Fellows Thesis, Texas A&M University, 1997.

Contents

1 Introduction 6

1.1 Background Description of Sample Material . ... 6 1.2 Characteristics of Westerly Granite in Electrolytic Solutions .... . 8 1.3 Compression of Core Samples ... ... 10

2 Methods and Materials 17

2.1 Sample Selection and Preparation . ... . . . . . 17 2.2 Collection of Data ... . . ... .. 20

3 Results 22

4 Discussion 29

4.1 Qualitative Differentiation ... ... 32 4.2 Other Electrical Sources . ... ... 32 4.3 The Piezoelectric Problem ... ... 33

5 Conclusions 35

5.1 Future Investigations ... ... 36

A Calculation of the Resistance of Saturated Westerly Granite 37

B Sample Preparation 39

C Streaming Potentials 41

Chapter 1

Introduction

Before the Ms (surface wave magnitude) 7.1 Loma Prieta, CA earthquake in 1989, Fraser-Smith, Bernardi, McGill, Ladd, Helliwell, and Villard (1990) [6] recorded sev-eral anomalous magnetic (EM) signals. Peaks were detected on September 16-17, followed by a rise in background noise on October 6 and an EM spike on October 17 approximately 3 hours before the quake. The regional stress prior to the earthquake changed by less than 1 pbar[18]. Park, Johnston, Madden, Morgan, and Morri-son (1993) review the history of looking for a possible link between generation of anomalous EM signals and a subsequent earthquake. The study of how electricity is generated and transfered in a rock is required. The objective of this research is to study the generation of electrical charge in Westerly granite before and during uniaxial fracture.

1.1

Background Description of Sample Material

Granite rocks are igneous in origin, comprising of interconnected crystals of quartz, feldspar, mica and other minerals of various concentrations. Because granite is formed in the crust, the uplifting and release of pressure compounded with exposure to the atmosphere causes the rock to crack, with cracks even developing between the crystals themselves (Figure 1-1). The core samples used in this study were Westerly granite, which comes from Westerly, Rhode Island. It is composed of 27.5% quartz, 35.4%

Figure 1-1: Wong, Fredrich, and Gwanmesia (1989) show a view of what the crack surface areas look like. This is also typical in that the cracks often occur on grain boundaries.

microcline, 31.4% plagioclase (with 17% anorthite), and 4.9% biotite [1]. The dis-tribution of minerals in Westerly granite is virtually homogenous, which makes it a popular specimen for rock mechanics experiments.

There is also extensive information about the microstructure of Westerly granite. The cracks between the grains occupy about 0.9% of the space. Labeled as a low porosity rock, Westerly granite also does not have much connectivity between the cracks, thus making it a low permeability rock as well [1]. Wong, Fredrich, and Gwanmesia (1989) [21] found that Westerly granite has crack apertures that vary from 0.7 to 0.002 micrometers in width. The crack surface area per volume (Sv)

was measured to be approximately 7.98 mm2/mm 3. The distribution of cracks in this range corresponded to a fractal dimension of 2.84 with the smallest apertures

1.2

Characteristics of Westerly Granite in

Elec-trolytic Solutions

When Westerly granite is saturated, the connected pore space is filled with fluid. When the fluid is an aqueous electrolytic solution, some of the ions adsorb to the grain surfaces around the cracks and some of the ions on the surfaces of the cracks dissolve into the solution. Because most of the molecules that make up granite consist

of SiO₂ 1

Revil and Glover (1997) [19] argue that there are five chemical species that occur on the surface of Si02 mineral lattices (where > refers to the mineral lattice): >SiOH,

> SiO-, > SiOH-, and >SiONa (in the presence of high pH values).

The adsorbtion of the ions make a net negative electrical charge on the grain surfaces around the cracks. Different solutions react to the static electrical charge in different ways.

Figure 1-2 of Morgan, Williams, and Madden (1989) [14] illustrates the adaptation of the current view of how ionic charge distribution on the surface of the cracks with respect to SiO2 surfaces depending on the nature of the aqueous soltion with respect

to the composition of the rock. The negative electrical charge on the surface of the rock is neutralized by positive ions in the electrolytic saturating solution. Because the ions have diameters, only a limited number can reach the charged surface. As a result, it takes more than one layer in most cases for the negative potential to be neutralized. The negative charge is at first linearly reduced by several layers of positive ions, then it decays exponentially with distance from the Inner Helmholtz 1_{It is observed that the solution used to saturate Westerly granite changes in composition after}

saturation. The solution conductivity measurements were made before and after the Westerly was saturated with tap water were different; the conductivity decreased from 200 pS to 8 pS. Other ions are probably involved in the development of the double layer [15]:

1. quartz is essentially pure SiO2

2. microcline KA1Si₃08

3. plagioclase (17% CaA12Si208, 83% is NaAISi30s)

$ S O(STANCE

7

•

7<

Figure 1-2: "S" corresponds to the slipping plane. The fixed layer where the electrical potential decreases linearly is inside the slipping plane. The diffuse layer where the electrical potential decreases exponentially is outside the slipping plane. Morgan, Williams, and Madden (1989).

Layer. The plane where this behavior changes corresponds to the Outer Helmholtz Layer. Within a short distance outside of the Outer Helmholtz Layer, ions can move. This is labeled the slipping plane. The region outside of the slipping plane where potential electrical energy is still decaying is labeled the diffuse layer. The region inside the slipping plane is labeled the fixed layer. The zeta potential is defined as the electrical potential that exists at the slipping plane.

This situation becomes a little more complex as the smallest crack apertures of .002 micrometers occur more frequently. The diffuse layer in some solutions is greater than half this value. In this situation, the double layer will overlap in a manner described by Hunter [8]. The repulsion between the double layers on either side of the crack creates an osmotic pressure. In a confining pressure situation, this adds to the pore pressure of the system.

Movement of ions outside of the slipping plane has been studied for several decades. When a pressure is applied, a physical current displaces ions in the diffuse layer. This is called a streaming or convective electrical current. The physical flow of ions which had kept the surface charge neutral begins to concentrate in some areas and dilute in others. This causes a current of ions, labeled the conduction current, to be induced

in the opposite direction. When a steady state is achieved, ions physically flow in both directions. An electrical potential field forms that is labeled the streaming potential. This is quantitatively described by the Helmholtz-Shmoluchowski equation

(derivation from streaming current in a capillary shown in (Appendix D):

A(P) c A(V) aov P=physical pressure

V=voltage in the streaming potential ( =zeta potential

E =dielectic constant

a =conductivity of the electrolytic solution v =viscosity of the liquid

1.3

Compression of Core Samples

As the rock sample is compressed uniaxially, those cracks that are easiest to close are closed. As the compression continues to increase past 50% of the fracture strength of the rock, new cracks open and old cracks dilate to wider apertures [1] (Figure 1-3). The dilating effect tends to follow a particular orientation that develops into a fracture in the rock [12] (Figure 1-4). Scholz [20] (Figure 1-5) explained that the fracture mode for unconfined samples is an elongated fracture in the direction of compression extending through the sample. Jaegar [10] (Figure 1-6) showed that the stronger platens above and below the sample affect how the uniaxial pressure is distributed in the sample. This distribution tends to follow two cones with their bases on the platens, and their tips touching in the middle of the sample. It is not surprising that the acoustic activity in Lockner's samples was first localized in the middle of the sample. Furthermore, the rupturing highlighted the side of the pressure cone as the sample approached failure. This is probably due to the high pressure gradient on the side of the cones in unconfined samples. The inside of the cone has the pressure load, while the outside experiences only atmospheric pressure. The samples in this

r

i

II

are closing Axial Cracks in

>

100O

Percent of fracture stress

Figure 1-3: Brace and Orange (1968) measured the changes in resistivity of Westerly granite under uniaxial compression. The results they found was an increase in resis-tivity followed by a decrease in resisresis-tivity. This was due to the closing and widening of cracks in the process of stressing the rock.

experiment exhibited behavior that was either conical or a combination of conical and elongated.

IXII^IIII

Perpendicular to the plane of fracture . Looking at plane of fracture face on \ ' '" i "" L I 24 ~r b

Figure 1-4: Lockner, Byerlee, Kuksenko, Ponomarev, and Sidorin (1991) measured acoustic emissions from uniaxially compressed Westerly granite sequentially over time. At first the acoustic waves are distributed through out the sample. Then they begin to concentrate on the plane of the developing fracture.

12 ... ... ... ... ...

•

4

il

Figure 1-5: Scholtz (1990) illustrates three modes of fracturing seen in rocks. Arrows indicate the direction of stress. (a) tensile fracturing. (b) triaxial fracturing. (c) uniaxial fracturing.

Figure 1-6: Jaegar and Cook (1969) illustrate how pressure is distributed in a rock under uniaxial compression. A saddle point occurs in the middle where pressure increases toward the platens and decreases on all other sides.

Chapter 2

Methods and Materials

2.1

Sample Selection and Preparation

All samples except one were cored from the same block of Westerly granite to ensure uniformity of chemical makeup and crystal distribution in the samples. Core samples were obtained from the Westerly granite by using a drill press, then slicing the column to various lengths using saws. Then the core samples were polished on the ends with a diamond drill while squeezing on the sides of the sample with a conical base to ensure that the ends were also parallel. The samples varied in size and ranged from

1870 mm3 to 37800 mm3.

The samples were subsequently wrapped with an insulated wire, size #18 19/30, attached to the lengthwise surface of the core by conductive glue, DuPont conductor composition 4929N, submitted to different atmospheric pressures and humidities, and soaked in different solutions before being submitted to uniaxial stress. The acoustic transducer marked the time range for recording the electrical signals produced during fracture of the core samples.

Figure 2-1 shows how the wires were placed onto the samples using 2 different methods. The experimental setup is drawn out in Figure 2-2 in terms of what the electrical circuit of the experiment consists of.

Sample

Method 2

This was the second method that was used to wrap the wires amund the sample. Conductive glue was rubbed on to the sample around the wires.

Wres

Mlethod I

This was the first method used for

wrapping the wires around the

sample. Conductive glue was brushed around the wires.

Figure 2-1: Methods used for wrapping wires around the samples

2~

'p

)

Wi res

Current

to wund

Impedance >> 1 mega-ohm

Figure 2-2: Experimental setup using GageScope as an Ammeter

2.2

Collection of Data

The electrical signals resulting from fracture of the samples were recorded with a GageScope, an IBM PC-software based oscilloscope. The current passed through the wire wrapped around each sample with a 1 mega-ohm impedance, which has to be less than the impedance of saturated Westerly granite (Appendix A) [2] for the Gagescope to work as an ammeter. The signal was recorded in terms of volts with a sample rate of 200 kilohertz. The system accepted an external trigger source from an acoustic transducer which measures pressure waves in terms of electrical volts. 5 milliseconds of time requires 30 kilobytes of memory. This sort of precision is necessary to get enough points into the electrical waves produced by the sample at the amplitudes before and after fracture. Maintaining this precision for the duration of the experiment would require 24.3 megabytes of memeory for 8.5 minutes. At least 4 files were produced for each run. The acoustic transducer signal was processed through an amplifier to

increase the voltage of the signal to positive or negative 5 volts. The voltage was recorded with the oscilloscope both before and after the trigger signal of positive 5 volts, so 512 points were recorded for each sample both before and after the signal making a total of 1024 points.

Piston

Plastic Film

Top Probe

Middle Probe ... Sample

Bottom Probe . ... ..

Plastic Film Acoustic Transducer

PlasticPlastic FilmFIm

Figure 2-3: This is the experimental setup used in fracturing the samples of Westerly granite. The plastic film was used to isolate the sample from the ground. Each of the probes have 1 MQ impedance.

Systems, Inc. following plans drawn up by E. Sholtz in 1993. The loadcell and LVDT recordings of displacement over time were recorded by Superscope II Macintosh program with the sampling rate at 1 hertz. The samples were lined up with a manual laser placed in the loadcell to ensure radial symmetry of the samples before they were fractured . Plastic film was placed above steel blocks on the sample and below the sample itself. The uniaxial system grounded the probes that connected the wires on the sample to the computer. The acoustic transducer produced electrical disturbances from pressure waves interacting with its bladder. The plastic film acted as an electrical insultator for the sample by preventing the electrical energy from passing straight to the ground rather than through the 1 mega-ohm probe (Figure 2-3).

Chapter 3

Results

An integration to calculate the number coulombs which had moved through the acous-tic transducer and the probes in any paracous-ticular direction was used 1. The' Acousacous-tic versus Net Charge (Figure 3-1) shows a rough increasing relationship between the acoustic total charge in coulombs and the net charge of coulombs flowing through the Westerly granite. This plot uses all the experimental data that has been collected by Morgan and Zhu (1998) and Jensen (this report). The increasing relationship is also in samples that had a net influx of positive electrical charge. An influx of positive electrical charge is negative, and these samples were not plotted in Figure 3-1.

When plotting the Net Acoustic energy versus the Volume of all the different samples (Figure 3-2). There are three ranges of acoustic energy that appear. The

topmost range, Region 1, exists above 8X10- 1 Coulombs. Region 2 ranges from

1The data collected from the probes and the acoustic energy were received in terms of volts. To find the current at each point in time, the values were divided by the 1 mega-ohm resistance. For some, a smoothing function was applied. This took the form of a running average of 10 points. For these, the values were buffered on the ends by 10 points. In order to compare how much the values changed in the integration to total up the the moving charge, an average was taken of the beginning of each spectra. The average used the first 200 points for the Poland Spring data and the first 100 for the data collected by Morgan and Zhu (1998). This average was then used to set the zero level. The caculation of how much charge had moved added the electrical current multiplied by the change in time to the previous calculation of electrical current by change in time. Then the value from this calculation for each probe were summed together. This was the net flow of charge in the sample. If it was negative, in the case of all the dry samples, then this meant that more positive electrical charge had flowed into the sample. If it was positive, more positive electrical charge had flowed out of the sample. The value for the Relative Acoustic Energy was calculated as the root-mean-square of the acoustic signal.

10-'

E

U

o

10'

1 0

_{1 0}

₀

1 0-"

1 0

0

Acoustic Relative Energy

1.4E-10 -r 1.2E-10 IE-10 8E-11 U 6E-11 1A 4E-1 1 2E-11 2 I U

0.OOE+00 2.00E+04 4.00E+04 6.00E+04 8.00E+04 1.00E+05 1.20E+05 1.40E+05 1.60E+05 Yolume (mm^3)

Figure 3-2: The acoustic relative energy is unrelated to volume. With respect to Figure 3-1, the net charge is unrelated to volume.

5X10- 12 and 2X10- 11 Coulombs. Region 3 lies below. This strongly suggests that there is no relationship between the volume of the sample and the relative acoustic energy that the sample produces when it fractures. This in turn argues that there is no relationship between volume and the net electrical charge flux.

In Region 1, there are only three points. Two are 25mm diameter by 50.8mm length dry samples which were broken at a loading rate of 100 um/sec. The third is a 25mm diameter by 50.8mm length sample that was saturated in 200 uS tap water and broken at a loading rate of 100 um/sec. The most remarkable point to make in Table 3.1 is that the saturated sample produced as much as a magnitude more

coulombs of net charge than the dry samples.

To get an idea as to how much variation there is in samples that are identical experiments, three sets of data were compared. The samples used in the data set include:

Table 3.1: The three highest acoustic relative energy samples Acoustic Relative Energy Charge (coulombs) sample style

1.28E-10 -2.05E-09 dry

1.04E-10 -1.38E-09 dry

1.04E-10 1.13E-08 wet

1. The dry samples from Table 3.1

2. Two 25mm diameter by 50.8mm length Region 2 samples that were saturated with 200pS tap water and broken at a 100 pm/sec loading rate

3. Two 25mm diameter by 76.2mm length Region 3 samples that were saturated with 225pS Poland Springs water and broken at a 1 pm/sec loading rate It is important to note that the experiment was identical for these samples except in the way that they fractured. The Fracture Variation plot (Figure 3-3) demonstrates that a relationship may exist between the Acoustic energy produced during rupture

and the fracture type. Six samples were compared. Four Poland Springs saturated samples broken at 1 um/sec and two tap water samples broken at 100 um/sec were plotted against each other depending on which fracture type they followed (Figure 3-4). There are two main fracture types with some variations. The first fracture type is a fault extending from upper portion of one side of the sample to the lower portion of the other side. The second fracture type is a combination of the first type but with the development of a second fault on the other side that propagates through the sample at least part way. An example of other fracture types is pictured as well, but other fracture types did not occur in the samples used in the plot. For the samples which were broken at the 1 pm/sec loading rate, a strong relationship is suggested between the acoustic energy and the way the sample fracture. At the 100pm/sec loading rate, this trend does not appear to be present.

In Region 2, there are two samples which are identical in every way except in the conductivity of the solutions that they were saturated in. Shown in the Conductivity versus Net Charge plot (figure 3-5), the 450 and 820 pS samples are 25mm diameter by 50.8mm length, broken at the same 100 pm/sec loading rate, and produced similar

10

Acoustic Relative Energy

Figure 3-3: Fracture versus Stress Rate (Positive Charge)

Fracture Type 1 _{Fracture Type 2 Other types of} fractures

Figure 3-4: fractured.

These are the most common ways that the Westerly granite samples

S 100 um/sec 100 um/sec

U.

[

2.00e-9

2j 1.00e-9

E

a-SO.OOe+O

I

Solution Conductivity (uS)

Figure 3-5: Solution Conductivity versus Charge and Electrode Placement amount of Net Charge. They indicate a general increasing trend in the Net Charge produced and the conductivity of the saturating solution.

More variation may be in seen in how the electrical energy is drawn off of the sample. The conductivity samples varied somewhat in this respect. The 450 and 820 pS saturated samples had electrodes placed as seen in Figure 3-6(a). The Conductiv-ity versus Net Charge plot also shows the 590 AS saturated sample. The electrodes on this sample were place with 1 vertical as used in the 450 and 820 AS saturated samples and with 2 electrodes placed around the length of the sample which is some-what similar to Method 1 (Figure 3-6b). Although the 590 /S saturated sample lines up with the 450 and 820 AS saturated samples, it appears that the electrodes in this formation did not collect as much electrical charge. The 570 S saturated sample had electrodes placed on it almost identical to the wires in Method 1. The electrodes in

(b)

E(

C

/

El

Morgan and Zhu (1998) placed electrodes on the Westerly granite samples in this way

for the samples saturated in

450 and 820 microsiemens solutions.

Morgan and Zhu (1998) placed electrodes on the Westerly granite sample in this way for the sample saturated in the 590 microsiemens solution.

Figure 3-6: Variations in electrode placement

this formation detected even less charge (Figure 3-5). Method 2 for placing the wires onto the samples was meant to take advantage of the geometry of having a conduct-ing loop wrappconduct-ing around the sample, as well as cover much more of the sample with conducting wires to collect any local charges. The raw data plot from sample 101

(following the appendicies) shows that the current was moving in opposite directions through wires that were next to each other on the sample.

El

>E3

EI1

Chapter 4

Discussion

The phenomenon of piezoelectricity in a quartz crystal is the release of electrical en-ergy when the crystal is compressed. There is 25.7% quartz in a Westerly granite; this is a source of some electricity. Quartz is a very hard mineral, 7 on the hardness scale. For this reason, many cracks in the Westerly granite develop between the quartz grains and the plagioclase grains and along the cleavage planes of the plagioclase. As the granite is loaded, these cracks are among those closed. As the faulting begins, it will tend to go around the quartz crystals; however, when a faulting plane propogates, enough inertia is built up that it will pass through the crystals and break them. The dangling bonds on the new cracks become a new electrical field for adsorbing ions. Electricity that is produced by breaking quartz minerals or through piezoelectric-ity [16][17] has two ways of passing through the rock to the wires surrounding it: through the grains themselves, and through any sort of fluid in the cracks. With the impedance of dry Westerly being greater than 1 mega-ohm (Appendix A), the preferred vector for the propogation of electrical energy is through any fluid in the cracks. Since the percentage of quartz in Westerly granite is constant, a voltage in-crease in the maximum amplitude of the amperes may be initially detected between a dry granite and a granite that is saturated with a conductive solution. However, as the conductance of the solution increases, the finite energy source of the electricity prevents it from contributing more than a constant value to the voltage readings.

i/

I

Figure 4-1: Matijevic (1974) illustrates the effect of increasing the concentration of the aqueous solution on the zeta potential. cl and c2 are solutions with increasing

molarities. x is the distance from the surface wall.

in length. Electrodes were glued onto the sample in the manner shown in Figure 3-6. When the results are plotted against the conductivity of the saturating solution, there is an increasing trend that becomes visible (Figure 3-5).

One aspect of the behavior of the system described by the Helmholtz-Shmoluchowski equation argues that the greater the molarity of a solution, the less the amperage of the streaming current should be. This is because the greater conductivity of the solution decreases the amount of charge in the diffuse layer by concentrating it in the fixed layer which decreases the zeta potential [8] (Figure 4-1). The streaming current depends on the zeta potential through a logarithmic relationship. The other variables which effect the streaming potential, geometry, conductivity of the solution, and viscosity, are relatively constant in comparison to the zeta potential.

If the piezoelectric effect were at work, the electrical relationship should be a flat line; if the streaming current effect were taking place, the line should be decreasing. Because this is not the behavior observed in Figure 3-5, neither of these phenomena are producing the increasing electricity with concentration.

4.1

Qualitative Differentiation

Three types of fractures were observed (Figure 3-4). The first type is a fracture extending through the sample to the other side. The second type was characterized by curved structures that did not always pass through the center of the sample. The third type consisted of the loading surface of the sample breaking through as if two type one fractures on either side met up at the top of the sample. There appears to be different acoustic energy generated with the different fracture types (Figure 3-3).

4.2

Other Electrical Sources

Another electrical potential field exists at the contact points between the wires and the metal in the conductive glue. This probably does produce some electricity. The purpose of the glass experiment was to determine what the magnitude of this would be. An amorphous glass sample cored from a block of Corning glass was broken using the setup described for the Poland Springs saturated sample. The glass sample of dimensions (19mm by 54mm) was loaded to a significantly higher pressure than the sample of Westerly granite of the same dimensions was because it took twice as long for the sample to rupture. When it broke, it produced a popping sound about four times before shattering. Table 4.1 shows that some electrical energy was detected after the sample ruptured. The postive results was surprising. Because of the positive detection of electrical energy, samples of the same magnitude of electrical energy are suspect.

Morgan and Zhu's (1998) data for the conductivity experiments confirms the con-cept that the measured electrical signal is not due to the wires alone because electrodes were place on the sample instead of glued wires. The signal that he measured could not have been produced by the electrical potential field between the glue and the wires because neither were present.

Much of Morgan and Zhu's (1998) data which measured the effects of loading rates and piezoelectric trends used wires rather than electrodes placed on the sample in the

manner shown in the description of sample zhu. In order to be able to use Morgan and Zhu's (1998) data in this thesis for comparing amplitudes and frequencies, it was necessary to use the same placement of wires on the sample. Some minor variations were used as shown in Figure 2-1. The geometry is similar to zhu's, yet more surface area of the sample is covered with glue and electrical wires. This was the best way to measure how the electrical signal varied between volumes of the sample that were wrapped by the wires. Rather than a wire measuring a circle in the volume, the volume itself was measured.

4.3

The Piezoelectric Problem

Ogawa, Oike, and Miura (1985)[17] and Nitsan (1977)[16] similarly measured elec-tricity produced from fracturing rocks. Although neither author discusses how their experimental samples were prepared, they measure little to no electrical activity in samples which lack piezoelectrical minerals. Enomoto and Hashimoto (1990) [5] used samples of different minerals for their fracturing tests. The electrical signals that Enomoto and Hashimoto measured depended on the moisture content of the rocks they were fracturing. They detected electrical energy in fracturing saturated rocks that lack piezoelectric minerals.

Table 4.1: Experimental

Data

sample volume mm3 Acoustic Energy sum of abs fracture sample style load rate (pm/sec)

91 4.83E+04 1.13017E-12 2.8953E-11 1 Poland Spring 1

101 6.12E+04 6.91819E-12 2.0946E-09 2 Poland Spring 1

102 6.12E+04 2.01123E-13 3.4867E-11 1 Poland Spring 1

110 1.50E+05 8.63303E-14 1.4322E-11 1 Poland Spring 1

120 1.50E+05 1.229E-12 1.6761E-10 2x Poland Spring 1

140 6.12E+04 1.78488E-13 2.1E-11 glass 1

450 9.97E+04 7.26319E-12 1.5311E-10 450 uS 100

570 9.97E+04 8.51587E-12 7.7034E-11 570 uS 100

590 9.97E+04 5.44736E-12 2.1256E-10 590 uS 100

820 9.97E+04 6.2775E-12 1.1445E-09 820 uS 100

zhu 1.50E+05 1.7149E-12 1.4451E-11 1 dry 100

splayed wires 1.50E+05 6.02616E-12 6.5764E-10 1 1

RD 9.97E+04 1.27559E-10 2.0495E-09 dry 100

RG 9.97E+04 1.03589E-10 1.3841E-09 1 or 2 dry 100

RI 9.97E+04 8.96146E-11 9.3363E-09 200 uS 100

RJ 9.97E+04 1.03589E-10 1.1272E-08 200 uS 100

RK 9.97E+04 6.05045E-13 2.8541E-09 1 200 uS 1

RL 9.97E+04 6.02523E-13 2.9075E-09 2 200 uS 100

Chapter 5

Conclusions

Both the piezoelectric effect and the streaming potential effects for the production of the energy fail to explain why the energy increases with molarity of the saturating solution in Morgan and Zhu's (1998) conductivity experiment. This leaves taking a closer look at the whole system and what assumptions have been made. The most significant assumption is that the layer inside the slipping plane is fixed. This is true in a steady state, but in the system of a rock under stress, this possibly fails. As cracks get smaller the repulsion between double layers increases. A lot of charge is accumulated in a small area. When the crack closes, much of this charge is pushed out into the adjoining cracks, and the charges unbalance the charge distribution in the adjoining decreasing crack widths. The movement of this energy is closer to the charge stored in the surface potential on the grain surfaces.

The acoustic energy is proportional to how much charge moves through the rock in any particular direction for the duration of the measurement. In general, the more acoustic energy the sample breaks with, the greater the amount of net charge has passeed through. This includes such effects as piezoelectricity. There is some deviation, however. The samples that were broken at a 100pm/sec loading rate were far more dependent on this relationship than samples that were broken at a 1 pm/sec loading rate. For the 1 pm/sec loading rate samples, there was a significant variation in the way that the samples fractured whenever they were being loaded. The fracture variation contributed to the increasing net charge with acoustic energy

relationship. Whenever this was taken into account, the samples depended much less on the acoustic energy in the amount of net charge moving through them.

The movement of electrical energy through the rock appears to be a localized effect but with a net magnitude and direction for the whole of the sample. The raw data from sample 101 shows this. Even though the middle and bottom probes literally had wires right next to each other on the sample, they each had current passing in opposite directions through the wire. Although it may only be coincidental, it is interesting to note that all of the dry samples produced negative net electrical charges.

5.1

Future Investigations

Placing multiple small electrodes around the sample in a similar run could provide a resolved view of just how localized the direction of charge movement is in the sample. There is even the possibility that observations may be made of how the charge is flowing with respect to how the rock is fracturing in this type of experiment. However, the method of data collection would need to be altered.

Continuously taking data through the 1 mega-ohm probes may provide some in-teresting data on how the system of the saturated sample behaves in a temporal way. Maintaining this precision for the duration of the experiment would take up 24.3 megabytes for 8.5 minutes. At least 4 files are produced for each run. There is no reason to assume that there are no significant electrical signals being produced with longer wavelengths.

Experiments with increasing the conductivity of the saturating solution should be performed. At greater magnitudes of conductivity, a problem may arise with the resistance of the rock decreasing into the the range of one mega-ohm. If this is the case, a new experimental setup would need to be considered.

The behavior of saturated rocks that lack piezoelectric minerals should be ex-plored. The experiments performed by Ogawa, et al(1985)[17], Nitsan (1977)[16], and Enomoto and Hashimoto (1990) [5] argue that this is an important direction for more research.

Appendix A

Calculation of the Resistance of

Saturated Westerly Granite

The porosity of Westerly granite is approximately 0.009.

The conductivity of the Poland Springs water was measured to be 225 pS. Morgan and Zhu's greatest solution conductivity was 820 pS. Using Archie's Law, the resis-tance of the rock before the experiment was a magnitude greater than the impedance on the probe. It is improtant to note that the porosity and the level of saturation are altered at the time of fracture when measurements are made [1].

I

V where

I = Amperes V = volts

The resistance of the soltion is

Rsolution =

where R= Q

1 R=p A where p = Qmeters 1 = meters A = meters2

According to Archie's Law

Prock = Psolution - 2

where

0 = porosity

So the resistance of the rock is calculated as

A 1

Rrock = Rsolution A-2

so

Rrock = 0.009-2

The resistance of a rock saturated in 820pS NaC1, Rrock = 1.2X107 , is about a magnitude greater than the impedance of the Gagescope probes.

Appendix B

Sample Preparation

The insulated wire wrap size #18 19/30 contains 19 fine wires that were split roughly in half and wrapped around sample zhu and glued with the conductive glue. The sample was baked in a vacuum for 30 minutes. One hour later in 22% humidity and 23.5 degrees Celsius, it was loaded and broken in the lab at a piston displacement rate of 100 microns/sec.

The splayed wire sample had 16 wires around the top, 18 around the middle, and 12 around the bottom. With generous amounts of glue, this sample sat in room humidity (greater than 22%) for several weeks before being broken in the lab at a rate of 1 micron/sec.

Samples 91, 101, 102, 110, and 120 were placed in a vacuum for two days. After this period of time, the samples were placed in 204 microsiemens solution of Poland Springs water for two days to saturate and moved to the lab while immersed in the solution. The wires for the samples were placed while still immersed in the solution. As the stress was being applied to the sample, the conductive glue was being applied to the wires as well. Compression of all samples was applied at the rate of 1 micron/sec. Each sample was removed from solution only when the time came for stress to be

applied.

Sample 140 consisted of cored glass from a block of Corning amorphous glass. The sample had always been in 22% humidity. Wires were placed on the sample, and the

Table B.1: Experimental methods used for preparing samples sample Wire Method Prep Procedure

91 2 3 101 2 3 102 2 3 110 2 3 120 2 3 140 2 3 450 Electrodes 1 1 570 Electrodes 4 1 590 Electrodes 2 1 820 Electrodes 1 1 zhu 1 1 splayed wires 2 2 RD 1 1 RG 1 1 RI 1 1 RJ 1 1 RK 1 1 RL 1 1 RM 1 1

Appendix C

Streaming Potentials

I, = --- 7rr2 p

The pressure p that causes movement of ions on the slipping plane causes an electrical or streaming current I,. I, then induces a conductive current Ic in the opposite direction.

Figure C-1 illustrates the geometry of a capillary with radius, r, and length, 1. Fluid, of viscosity v, in the capillary moves under an induced pressure, p. Adsorbtion of ions on the surface of the capillary produces a double layer with a slipping plane conductivity of ao and a zeta potential C. The conductivity with respect to the distance r towards the slipping plane is described by a,.

7r r2 Vuo

le =

1

The balance of electrical charge between the currents causes the streaming poten-tial V.

I, + Ic = 0

V _

p Hunter [8].

r

Capillary Surface

Figure C-1: Diagram of capillary geometry with P as the direction of a pressure gradient

T

Appendix D

Experiment Data

Integration Sample RM Fracture 1 D/L=25/50.8 mm

Integration 2e-12 I Oe+O -2e-12 -4e-12 -6e-12 -8e-12 ' -u.003 -0.1 0.0 Time (sec)

Sample "zhu" Fracture 1

-0.002 -0.001 -0.000 0.001 Time (sec) Top Probe Middle Probe Bottom Probe 0.2 0.1 D/L=25/76.2 mm 0.002 Top Probe Middle Probe Bottom Probe 0.003 le-9 Oe+O -le-9 -2e-9 ' -U. 2 r III I I I I I

-e

Integration Sample 101

Fracture 2

D/L

0 1 e-9

E

=19/57 mm

Integration Sample 102 Fracture 1 D/L=19/57 mm

- 2e-11

E _{Top Probe}

0

o le-ll

0

&

c 0 -le-ll S) Bottom Probe -2e-1 1 -u.003-0.002-0.001-0.000 0.001 0.002 0.003 Time (sec) Sample 110 Fracture 1 D/L=25/76.2 mm Top Probe Middle Probe Bottom Probe -u.003-0.002-0.001-0.000 0.001 0.002 0.003 Time (sec) ition

Integra

Sample 120 Fracture 21 003-0.002-0.001-0.000 0.001 Time (sec) ( D/L=25 0.002 0.0

ion

Sample 140

Fracture ?

D/L=19/54 mm

Fracture ?

/50.8 mm

ation

Integr

-u.003-Integra

tion

Integration -0 le-ll E . Oe+O o -le-11 o -2e-1 1 -3e-1 1 cc S-4e-11 " -5e-11 -6e-11 -u.002

Sample

570

Fracture ?

D/L=25/50.8

mm

Integration

Sample

590

Fracture ?

Fracture ?

D/L=25/50.8

mm

DIL=25/50.8

mm

Sample RD

Fracture ?

D/L=25/

tion

D/L=25/50.8

mm

50.8 mm

ation

Integr

E

Sample RJ

Sample RK

Fracture ?

D/L=25/50.8

mm

Fracture 1

Integr

E

.

5e-9

j Oe+O

z

D/L=251

50.8 mm

Sample RL

ation

tion

Integra

E

2 Oe+O

z

-ation

Integr

0 0 O I-0 0 o

O

- 0.07 0 > 0.06 .0 0.05 0 CL 0.04 o 0.03 0.02 -u.003 -0.002 -0.001 -0.000 0.001 S0.02 S0.01 0 & 0.00 .-0.01 2 -0.02 -u.003 -0.002 -0.001 -0.000 0.001 (A m 0.00 0 0 Q-0.01

E

M

D/L=19/54 mm

Sample 110 Fracture 1 D/L=25/76.2mm 0.04 .0 0.03 0 0.02 I-0.01 -u.003 -0.002 -0.001 -0.000 0.001 0.002 0.003 0.03 0.02 S-0.01 0.00 'a-.01 2 -0.02 -u.003 -0.002 -0.001 -0.000 0.001 0.002 0.003 0.01 S0.0 0 , -0.01 -u.003 -0.002 -0.001 -0.000 0.001 0.002 0.003 o 0.2 S0.1 Fn 0.0 -n -0.1 0 S-0.2 -u.003 -0.002 -0.00L-n ngn 001 0.002 0.003 I

me

Sample 120

Fracture 2X

D/L=25/76.2 mm

N

I

(secl

0.022 0.021 0.020 0.019 0.018

1

Fracture ?

(secT

Sample 450

Fracture ?

D/L=25/50.8

mm

--

iMY

\r

Sample

570

Sample

590

-0.001

Ti-0.

0oe

0.001

I I I I I I -0.001 -0.000 0.001 -0.001 -0.000 0.001 -0.001 -0.000 0.001 -0.001 T-0.000 Time (sec) -u.002 -1 -u.002 -u.002 0.002 0.002 0.002 6 4 2 0 -2 -4 -6 -U. I I I I I I 0.002 | Sample 820 Fracture ? D/L=25/50.8 mm 002 0.001

Sample RD Fracture ? D/L=25/50.8 mm .2 . 0 0 . 0 0 I-l 0 ._

a

-I

0.0

-1

Sample RI

Fracture ?

I I I -0.1 0.0 0.1 0.2 -0.1 0.0 0.1 0.2 -0.1 0.0 0.1 0.2 2 -0 I I I 0.0 Time (sec) 23 -u.2 2 1 0--1 --u.2 2 --u.2 2 1 0 -1 -2 -3 -4 -U 0.1 0.2 Sample RJ Fracture ? D/L=25/50.8 mm .2 -0.1

Sample RK

D/L=25/50.8

mm

a.

-u

-.... - n

D/L=25/50.8 mm -0.1 0.0 0.1 0.1 -0.1 0.0 0.0 0.2 0.1 0.2 0.1 0.2 1 I 0 Ti 0.0 Time (sec) -u.23 -u.2 _1 -u. -u .2 -1 -u.2 6 4 2 0 -2 -4 -6 -U -0.1 0.1 0.2 __I 1 _'rl Sample RL Fracture 2 .2

Sample RM

Sample "zhu" Fracture 1 - 0.03 > 0.02 e 0.01 0 0.00 "-0.01 o -0.02 -0.03 -u.0 003 -0.002 -0.001 -0.000 D/L=25/76.2 mm 0.002 0.003 0.001 0.002 0.003 o 0.00 a -0.0 1 0 L -0.02 E o -0.03 0 M -0.04 --u.003 -0.002 -0.001 -0.000 0.001 0.002 0.003 -0.002 -0.001 -0.000 0.001 Time (sec) 03 -0.002 -0.001 -0.000 0.001 o 0.04 0.02 o 0.00 -0.02 S-0.04 2 -0.06 -u. 1'* • - - : =" ' "-- --: ' -- -- = ': :--- ---- : - ' - - = ' ' = L -I o 2 1 r 00 0 .- -2 M -3 0 o -4 -u. I I I I I I 0.002 0.003 003

Bibliography

[1] W. F. Brace and A. S. Orange. Electrical resistivity changes in saturated rocks during fracture and frictional sliding. Journal of Geophysical Research, 73(4):1433-1445, 1968.

[2] W. F. Brace, J. B. Walsh, and W. T. Frangos. Permeability of granite under high pressure. Journal of Geophysical Research, 73(6):2225-2236, March 1968.

[3] G. O. Cress, B. T. Brady, and G. A. Rowell. Sources of electromagnetic radiation from fracture of rock samples in the laboratory. Geophysics Research Letters, 14:331, 1987.

[4] J. T. Dickinson, M. K. Park, E. E. Donaldson, and L. C. Jensen. Fracto-emission accompanying adhesive failure. Journal of Vacuume Sciece and Technology, 20(3):436-439, 1982.

[5] Y. Enomoto and H. Hashimoto. Emission of charged particles from indentation fracture of rocks. Nature, 346:641-643, August 1990.

[6] A. C. Fraser-Smith, A. Bernardi, P. R. McGill, M. E. Ladd, R. A. Helliwell, and Jr. O. G. Villard. Low-frequency magnetic field measurements near the epicenter of the ms 7.1 loma prieta earthquake. Geophysical Research Letters, 17(9):1465-1468, August 1990.

[7] Gage Applied Sciences Inc., Montreal, Quebec, Canada. Gagescope Technical Reference and User's Guide, seventh edition, 1996.

[8] Robert J. Hunter. Zeta Potential in Colloid Science. Academic Press, San Diego, California, 1981c1988.

[9] T. Ishido and H. Mizutani. Relationship between fracture strength of rocks and zeta-potential. Techtonophysics, 67:13-23, 1990.

[10] J. C. Jaegar and N. G. W. Cook. Fundamentals of Rock Mechanics. Chapman and Hall, New York, New York, third edition, 1969c1979.

[11] Ayao Kitahara and Akira Watanabe. Electrical Phenomena at Interfaces, vol-ume 15 of Surfactant Science Series. Marcel Dekker, Inc., New York, New York, 1984c.

[12] D. A. Lockner, J. D. Byerlee, V. Kuksenko, A. Ponomarev, and A[] Sidorin. Quasi-static fault growth and shear fracture energy in granite. Nature, 350:39-42, March 1991.

[13] Egon Matijevic. Surface and Colloid Science, volume 7 of Electrokinetic Phe-nomena. John Wiley & Sons, New York, New York, 1974c.

[14] F. D. Morgan, E. R. Williams, and T. R. Madden. Streaming potential prop-erties of westerly granite with applications. Journal of Geophysical Research, 94:12449-12461, September 1988.

[15] William D. Nesse. Introduction to Optical Mineralogy. Oxford University Press, New York, New York, second edition, 1986c1991.

[16] U. Nitsan. Electromagnetic emission accompanying fracture of quartz-bearing rocks. Geophysical Research Letters, 4(8):333-336, August 1977.

[17] T. Ogawa, K. Oike, and T. Miura. Electromagnetic radiations from rocks. Journal of Geophysical Research, 90:6245-6249, June 1985.

[18] S. K. Park, M. J. S. Johnston, T. R. Madden, F. D. Morgan, and H. F. Mor-rison. Electromagnetic precursors to earthquakes in the ulf band; a review of observations and mechanisms. Reviews of Geophysics, 31(2):117-132, 1993.

[19] A. Revil and P. W. J. Glover. Theory of ionic-surface electrical conduction in porous media. Geophysical Review B Journal, 55(3):1757-1773, January 1997. [20] Christopher H. Scholz. The Mechanics of Earthquakes and Faulting. Cambridge

University Press, Cambridge, Massachusetts, 1990c.

[21] T. Wong, tics and quartzite: Journal of

J. T. Fredrich, and G. D. Gwanmesia. Crack aperture statis-pore space fractal geometry of westerly granite and rutland Implications for an elastic contact model of rock compressibility. Geophysical Research, 94:10267-10278, August 1989.