ELECTRICAL RESISTIVITY CHANGES IN TUFFS
by
Carolyn Alexandria Morrow
S.B., Massachusetts Institute of Technology
(1978)
SUBMITTED IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE
DEGREE OF
MASTER OF SCIENCE
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September, 1979
Signature
of
Author......
......
.
........
....
..
Department of Ea th and Planetary Sciences,
September, 1979
Certified by..
. .
. .
. .
. .
. .
. .
........
,..
.. ..
Thesis Supervisor
Accepted by . . . . . . . . . . . . . . . . . . . . .
. . .
Chairman, Depar tment Committee on Graduate Students
3)H CHNLGY
MIT LIBRAR- i
ABSTRACT
ELECTRICAL RESISTIVITY CHANGES IN TUFFS
by
Carolyn Alexandria Morrow
Submitted to the Department of Earth and Planetary
Sciences on August 28 , 1979 in partial fulfillment of the
requirements for the degree of Master of Science.
Samples of northern California tuffs were stressed
while simultaneously measuring electrical resistance changes
to investigate a phenomenon observed by Yanazaki on similar
rocks from Japan. Resistance decreased substantially at low
strain values for partially saturated samples.
Strain was
amplified between 103 and 105 by the associated change in
electrical resistance.
The process was repeatable and recoverable in the tuffs
unlike the behavior of other rock types. The principal
factors involved were porosity, Young's modulus and degree
of saturation.
A method is described
electrically amplifying tuffs
first step in locating a field
could be used as an earthquake
Thesis Supervisor:
to quickly sort
out
the
from those that are not, as a
site where this phenomenon
monitoring technique.
William F. Brace
Professor of Geology
TABLE OF CONTENTS
LIST OF FIGURES
.
.
.
.
.
LIST OF TABLES.
.
.
.
.
.
ACKNOWLEDGMENTS
.
.
.
.
.
CHAPTER I
.
.
.
.
.
.
.
0
.
.
.
0
0
.
0
0
.
0
.
4
0
0
7
0
.
11
.
0
.
Introduction
II
Rocks Studied.
Sampling.
.
.
0
.
0
.
0
.
.
.
.
.
0
0
.
.
.
.
.
.
.
.
11
.
.
.
.
.
.
.
.
.
.
.
.
13
.
.
.
.
.
.
.
.
.
.
.
17
Rock Descrip tions
III
Experimental Pr ocedure
IV
Observations .
Stress-Strai
.
n
0
.. .
Behavior.
.
0
0
.
0
. . . .
..
.
.
Electrical C hanges with Stress. .
Discussion .
V
. . . . 20
.
.
.
.
20
.
.
.
.
23
.
36
Amplificatio n
Factor.
Recoverabili ty.
VI
.
.
.
.
.
.
.
Montana .uffs.
.
.
.
.
.
36
.
.
.
.
38
Conclusions. . . .
40
APPENDIX A
Studies of Neva da and Montana Tuffs. . .
41
APPENDIX B
The Chunk Test . 0. .o .0 . .0 .
55
APPENDIX C
Stress-Strain and Electrical Behavior
e
Selected California Tuffs
.
.
.
.
.
.
65
Resistance Measuring Techniques. .
.
.
.
.
.
74
D2
Choice of Electrode Material . .
.
.
.
.
.
86
D3
Frequency Effects.
0
.
.
.
89
.
0
.
.
93
APPENDIX Dl
REFERENCES.
.
.
.
0
~
~
~
0
0
~
-3-
.
.
.
. *
~
0
0
~
~
.
.
.
.
.
~
~
.
0
.
LIST OF FIGURES
CHAPTER III
Figure 3.1
Experimental apparatus.
.
.
.
.
.
.
.
.
.
.
.
.
.
19
.
.
21
.
.
.
22
.
.
.
26
.
.
27
.
CHAPTER IV
Figure 4.1
Typical stress-strain curves,
GPT and
Figure 4.2
DPR.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
28
.
.
.
.
.
.
.
.
.
.
29
.
.
.
.
.
.
.
.
30
.
31
.
32
Porosity of tuffs versus
.
.
Maximum amplification of tuff samples
versus
Figure 4.9
.
Relation of Young's modulus to
saturation at maximum amplification
peak saturation.
Figure 4.8
.
Amplification factor versus saturation
(peak saturation).
Figure 4.7
.
.
.
.
of the California tuffs.
Figure 4.6
.
Relative change in resistivity with
strain, SIL.
Figure 4.5
.
Change in electrical resistance with
stress of the Grizzly Peak tuff
and Berea sandstone.
Figure 4.4
.
Young's modulus of tuff samples
versus porosity.
Figure 4.3
.
porosity.
.
.
.
.
.
.
.
.
.
Maximum amplification of tuffs versus
peak
saturation.
.
.
.
.
.
.
.
.
.
Appendix A
Figure Al
Stress-strain curves of Nevada/Montana
tuffs and the Pottsville dnd Berea
sandstone.
Figure A2
Figure A3
.
.
.
.
.
.
.
.
.
46
Change in resistance with sta ass:
Tunnel Beds tuff and BereE sandstone
.
47
.
.
48
. . .
.
49
.
.
.
Change in resistance with stress:
Pottsville, Navajo and Mixed Company
(Kayenta) sandstones .
Figure A4
.
.
..
.
.
.
Relative change in resistivity with
strain, Butte Lapilli tuff . .
-4-
Figure A5
Relative change in resistivity with
strain, Berea sandstone.
.
.
.
.
.
.
0
50
Appendix B
Figure Bl
Sample configuration of the chunk test.
.
60
Figure B2
Chunk test experimental apparatus
Figure B3
Change in resistance with stress:
SHR, DPR, RLS, DRY .
Figure B4
.
.
.
.
.
.
.
60
.
.
.
..
61
.
.
.
.
62
Change in resistance with stress:
MWT, SZT,
Figure B5
.
.
GPT.
.
.
.
.
.
.
.
.
Change in resistance with stress:
SJB, SIL, RTT, COT .
.
.
.
.
.
.
.
.
.
63
.
.
.
.
.
.
66
.
.
.
.
67
.
.
.
.
68
.
.
.
69
.
.
.
70
Appendix C
Figure Cl
Stress-strain relation for the
California tuffs
Figure C2
GPT.
SHR.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
*
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Relative change in resistivity with
SZT.
.
.
.
.
......
....
.
71
Relative change in resistivity with
strain, DPR.
Figure C8
.
Relative change in resistivity with
strain,
Figure C7
.
Relative change in resistivity with
strain,
Figure C6
SHR.
SZT, RTT.
strain,
Figure C5
.
Change in resistance with stress:
DPR,
Figure C4
.
Change in resistance with stress:
GPT, SIL,
Figure C3
.
.
.
.
.
.
.
.
.
.
.
.
.
72
.
.
.
73
. .
.
.
88
.
92
.
Relative change in resistivity with
strain,
RTT.
.
.
.
.
.
.
.
*
.
.
Appendix D
Figure Dl
Stress effect of various electrodes,
20 kilohm resistor at 10 Hz. .
Figure D2
Resistance fall-off with frequency for
the California tuffs at varying
saturations.
.
.
.
.
.
.
.
.
.
.
.
LIST OF TABLES
Table 2.1
Rock Descriptions of Selected California
Tuffs.
Table 4.1
4.2
.
.
.
Table A2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. *
. *
. *
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Rock Descriptions and Locations of the
Montana/Nevada Tuffs and Sandstones.
.
33
.
.
35
.
.
44
.
51
.
53
.
58
.
64
Strain Amplification of the Montana/Nevada
Tuffs.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
Table A3
Strain Amplification of Sandstones. .
.
Table Bl
Location and Hand Specimen Descriptions
of the California Tuffs.
Table B2
15
Physical Properties of the California
Tuffs. . .
Table Al
.
Strain Amplification of the California
Tuffs.
-Table
.
.
.
.
.
.
.
.
.
Relative Ordering of Electrical Properties
of California Tuffs.
-6-
.
.
.
.
.
.
.
.
.
.
ACKNOWLEDGMENTS
I would like to thank William F.
and
advice throughout this work.
California
Bureau
University
of
of
and
Mines
Brace for his support
Charles Chesterman of the
Garniss
Curtis
of
the
California at Berkeley aided in the location
and collection of samples.
Michael Coln was very helpful in
improving the electrical measuring techniques, and in typing
this thesis.
-7-
CHAPTER I
Introduction
In
1965
Yamazaki
called
attention
to
dramatic
electrical resistivity changes thought to be caused by earth
tides.
the
Resistivity variations a thousand times larger
than
tidal strain were recorded in a cave near Tokyo, Japan.
Although the tidal origin is still disputed
Yamazaki
subsequently
confirmed
in
the laboratory [1965,
1966] the unusual electrical sensitivity to
particular
tuffaceous
were first made.
rocks
changes
rocks
Others
with
where
have
stress
Madden, [1978],
strain
of
the field observations
shown
that
resistivity
a
small
fraction
of
(Parkhomenko [1967], Brace and
Orange [19631, for example), but generally the changes
but
the
of
those
noted
by
were
Yamazaki,
particularly in porous rocks [Brace 1974].
The
electrical
effects
reported
by
Yamazaki
were
limited to tuffaceous rocks, partially saturated and at very
low stresses.
crustal
Because
deformation
of
studies
the
or
potential
to
application
to
earthquake prediction
(Rikitake and Yamazaki
[1969]),
extend
only to other rocks, but to a wider
his
work,
not
range of conditions.
effect
than
lapilli
sandstones be similar?
it
seemed
worthwhile
Tuffaceous sandstone showed
tuff
What
[Yamazaki
is
-.8-
the
a
to
weaker
1966].
Would other
optimum
mineralogy,
porosity, saturation and stress level?
Do appropriate rocks
exist near active fault zones in the United States?
they
to
Even if
do not, the mechanism of this fascinating effect needs
be
explained.
saturated
highly
How
can
porous
resistivity
rock
in
a
partially
change so dramatically in a
reversible way?
To explore such questions, a laboratory experiment
designed
which
would
attain
the
saturation and would include rock
studied
by Yamazaki.
conditions
types
of
similar
was
stress,
to
those
Since the field site was located in a
tuff cave with electrodes affixed to the floor and walls, an
unconfined
compression
stress state.
electrical
best approximated the in situ
In the compression tests, stress, strain
resistance
were
maximum stress of 6 MPa.
permanent
test
damage
simultaneously
This stress was
measured
chosen
to
to a
avoid
to the weak tuffs and sardstones studied.
A low stress cycle would ensure repeatable elastic
and
and
behavior
perhaps simulate tidal earth loads and certain tectonic
stresses.
Research was
These
were:
a)
carried
a
study
out
in
four
c'stinct
of the electrica l properties of
sandstones and tuffs from Nevada and Montana to
further
investigate
phases.
Yamazaki's
results were very promising, but the
-9-
findings.
details
verify
and
Preliminary
were
vaguely
understood.
Nevertheless,
the
next
step
was
b) sample
collection of tuffs near active California faults,
major
goal
was
since
the location of sites in the United States
where tuffs could be used in earthquake prediction.
a
a
Then c)
rapid approximate method was devised to find which of the
California rocks had more pronounced electrical
and
d)
several
properties,
of these tuffs were investigated in detail
under a more complete range of saturations to answer some of
the basic questions mentioned above.
Due to the more comprehensive nature of the
study,
these results are presented as the principal text of
this thesis.
tuffs
California
are
The work on sandstones and the
described in Appendix A.
where these data
California
comparison
are
rocks,
of
rock
included
with
particularly
types.
These
referenced.
-
1-0-
Nevada/Montana
There are a few places
the
for
spots
results
of
the
completeness
and
are
noted
and
CHAPTER II
Rocks Studied
Sampling
maps
state
of
With the help
and
geologists,
local
tuff sites were located in norther California.
several
(a) volcanics
principal locations were
the
Pinnacles
to the San Andreas fault, (b)
adjacent
Monument,
National
of
The
the Berkeley Hills volcanics, next to the Hayward fault, and
(c)
Somona and Cotati valleys, which are cut by
Napa,
the
In many of these areas, a number
several lesser faults.
distinctive
deposits
are
exposed, each of which was
Sampling was conducted in the dry
sampled.
fresh
from
tuff
outcrops,
surface
of
season
(July),
with care taken to preserve
natural water content.
With the large number of samples involved, (from over a
sites),
dozen
each locality.
out
it
was impossible to do a detailed study of
Therefore a quick test was devised
the more electrically sensitive rocks.
the "chunk test", in reference to
chunks
of
rocks
used
rather
the
but
properties.
quickly
order
sort
This was called
hand
specimen
than machined samples.
intent of the test was not to produce accurate
data,
to
size
The
quantitative
the rocks in terms of electrical
Details are described in Appendix B.
-1-1-
From
the
results
of
the
chunk
tests,
six tuffs from the Berkeley
Hills and Napa valley were chosen for
more
thorough
study
(Table
2.1).
table.
Detailed results of other sandstones can be found in
A
sandstone (Berea) is also included in the
Appendix A.
-12-
Rock Descriptions
Tuffs are the products of explosive volcanic eruptions.
Ash,
crystals,
porous
rock.
glass and rock fragments consolidate into a
Carossi
[1960]
gives
a
comprehensive
description of tuff petrology and vitroclastic texture.
The six tuffs listed in Table 2.1 represent
general
catagories of tuffs.
composed mostly of
tuffaceous
rock
matrix.
.crystal tuffs,
phenocrysts.
which
welded
together
by
a
and DPR fall into the catagory of
contain
Finally,
three
SZT and RTT are lithic tuffs,
fragments
GPT
the
SHR
a
and
large
SIL
percentage
are
vitric
of
tuffs,
composed mostly of fine-grained glassy groundmass.
In the descriptions of
section),
the
Table
2.1
(as
seen
in
thin
amount of breakdown of the unstable glass to
clay minerals and zeolites was estimated, as
Lhese
samples
were not analyzed by X-ray techniques.
Study of fresh
electron
microscope
fracture
surfaces
revealed
structures, reflecting tuff
a
type.
wide
The
with
the
scanning
of
internal
range
glassy
specimens,
with little recrystallization of the groundmass had numerous
very small pores (0.5 microns) caused by gas
was
the most notable example of this type.
-13-
bubbles.
SIL
With increasing
recrystallization, pores and
larger,
passageways
evidently
became
to a maximum of around 50 microns, although precise
determination of pore size, glass content and mineral phases
was difficult from fracture surface micrographs.
An attempt
was made to study pore geometry from polished surfaces using
-the
ion
[1974].
thinning
described in Sprunt and Brace
technique
These results
were
largely
unsuccessful
tuffs were very soft and did not thin uniformly.
-14-
as
the
Table 2.1
Sample
Rock Descriptions of Selected
California Tuffs
Location
Porosity
(%)
Young's
Mod., GPa
DPR
Deer Park Rd.
St. Helena,
Napa County
17.9
3.5
20% plagioclase and
sanidine phenocrysts
80% fine grain
reworked groundmass,
welded banding and
flow textures
SHR
St. Helena
Road.,
Sonoma Co.
37.2
2.2
10% phenocrysts
including:
8% plagioclase
2% fractured quartz
X % lithic fragments
90% groundmass:
30% glassy lenses
& shard structures
30% recrystallized
radial zeolites
30% fine grained
groundmass
SZT
Siesta
17.1
Formation
zeolite tuff
Berkeley Hills
3.3
80% basaltic lithic
fragments
5% plagioclase- very
fractured
15% zeolitized matrix
RTT
Round Top Hill
Berkeley Hills
12.5
60% basalt fragments
5% crystalline quartz
X % zeolitized
plagioclase
35% reworked glass,
clay and zeolites
GPT
Grizzly Peak
10.7
Berkeley Hills
10.0
30% pnenocrysts:
-.ngul .r fragments,
varie~1le size,
inclu es:
20% plagioclase
5% ,uartz
few % biotite
10% lithic fragments
60% groundmass,
including iron rich
montmorillonite,
isolated areas of
banded texture
3.7
-15-
Description
Sample
SIL
Location
Description
35.6
2.4
65% glass
30% pumice fragments
few % quartz and
plagioclase
phenocrysts
Ohio,
17.2
W. Virginia
6.5
Orthoquartzite:
95% quartz & chert
5% feldspar
Silverado
Trail, Napa
County
Berea
sandstone
Porosity* Young's
Mod., GPa
(%)
*Porosities were determined using the immersion method
described in Brace, Orange and Madden [1965].
-16-
CHAPTER III
Experimental Procedure
Rock samples were machined into cylindrical
cm
in
diameter
by
5
cm
long.
BLH
cores
strain
2.5
gauges
(FA-50-12-S6), oriented along the cylindrical axis, recorded
linear
strain.
An
oil pressure piston device was used to
stress the samples in unconfined uniaxial compression up
an
axial
3.1.
stress
Lead
attached
of 6 MPa as shown schematically in Figure
electrode
to
sheets,
1.3
mm -in
thickness
were
either end of the sample to form one side of a
Wheatstone bridge.
measurement
to
A
detailed
techniques
can
description
be
found
sample and electrodes were insulated
of
resistance
in Appendix D.
from
the
piston
The
and
load cell by a 0.1 mm layer of Pallflex teflon.
Electrical resistance measurements were taken at stress
levels
of
0,
increasing
frequent
0.2,
and
at
the
0.5,
1,
decreasing
lower
4,
and
stresses.
stresses
expected when the sample first
cases
2,
as
began
6
MPa for both
Sampling
the
to
was
more
most change was
strain.
In
all
a source voltage across the rock of 10 Hz AC was used
to minimize the frequency effects due to rock capacitance at
high resistance values.
-17-
The
Saturation
the
fluid
pore
sample
saturation,
saturation
measurements
was
used
35
tap
ohm-meter
water.
level was determined by averaging the weights of
both
each
after
and
before
run.
For
low
the variation was at most 1 percent, for higher
it
reached
were
5
accurate
resistance was known to
Stress
percent.
to
5
percent.
and
strain
The absolute
20 percent, and relative changes in
resistance were accurate to
5 percent.
-18-
piston
baLI
Figure 3.1
lead
teflon
Experimental Apparatus
CHAPTER IV
Observations
Stress-Strain Behavior
Stress-strain behavior in tuffs was highly dependent on
porosity,
previous
stress
history and time factors.
Some
samples showed a permanent change in the first stress cycle,
as
either a permanent strain or an increase in the modulus.
Succeeding cycles
linear
on
were
loading
all
and
nearly
unloading
samples were less recoverable and
identical
(Figure
also
and
fairly
4.la).
changed
Other
with
time
after a loading cycle,
as seen for instance with DPR (Figure
4.1b).
the initial
On the left
after
trace.
Then, immediately
the first stress cycle, the sample had become stiffer
(center trace).
days
is
later
The trace on the right
when the sample had relaxed.
was
taken
several
The stress-strain
curves for sandstone were similar in shape to those of
Young's
modulus
was
closely
related
to porosity (Figure
4.2); the more porous rocks tended to be less stiff.
-20-
DPR.
DPR
GPT
a.
Cn
U)
w
cr
I-
Cn
10~
LINEAR
4.la
Figure 4.1
STRAIN
4.lb
Typical stress-strain curves, GPT and DPR
1[30
RTT
GPT
0
0
DPI
z
SZT
0
SHR
30
20
POROSITY,
Figure 4.
40
%
Young's Modulus of tuff
samples vs. porosity
-22-
50
Electrical Changes with Stress
Grizzly Peak tuff
(GPT)
and
Berea
sandstone
showed
changes in resistance with stress that were typical of their
respective rock types (Figure
samples
can
be
4.3).
nearly
Recoverability
loaded,
for
found in the appendices.
consistently fell below loading
recovered
Data
varied
resistance
the
electrically
GPT
the sandstone hardly at all.
somewhat
of
other
Unloading curves
curves;
completely,
the
among
the
sandstone
tuffs.
Once
changed little with
stress, either upon unloading, or during
subsequent
cycles
(Figure 4.3).
The
sole
crystalline
they
observation
rocks
behave
[Brace
like
and
sandstone
Resistivity
rapidly
granite
45
at
of
effects
Orange, 1968]
when
decreased
percent
resistive
suggests that
partially
with
saturation.
stress
Upon
*in
saturated.
for
Westerly
unloading
and
subsequent loading, resistivity changes were quite small and
in the same direction as for saturated rocks,
that
is,
in
the reverse direction to the changes seen in the tuffs.
Returning
resistivity
to
predictably
typical value was
103 ohm-meters
the
porous
rocks
decreased
107 ohm-meters
at high values.
-23-
at
of
with
low
this
study,
saturation;
saturation,
a
and
Also, changes in resistance
(or
equivalently,
saturation.
resistivity)
Silverado
Trail
with
tuff
stress
depended
on
(SIL) -ias typical:
at
high saturation Ap/p changed little (Figure 4.4).
content
decreased
was
water
the slopes became steeper and the change
in resistance was larger.
trend
As
reversed,
At some value of
and
saturation
the
changes in resistance were small
again.
A convenient way to compare this effect among different
samples
is
to
define
factor, (Ap/p)strain
4
a quantity called the amplification
, namely the slope of the Ap/p
plot
at
a
Higher amplification factors indicate more
of 10~4.
sensitivity to earth strain.
The
values
measured
ranged
between 103 and 105.
Amplification factor depended on saturation as shown in
Figure
4.5.
The
amplification
saturation to a peak value and
saturation
is
significant
increased with decreasing
then
in
that
fell
off.
The
peak
it corresponds to the
highest sensitivity to strain.
As strain is related to Young's modulus through Hooke's.
Law,
peak
saturation
should
be dependent on the modulus.
This was
indeed
required
more conducting fluid to reach optimum sensitivity
(highest
amplification
the
case
(Figure
factor).
-24-
4.6).
Variations
Stiffer
in
rocks
Young's
modulus
in
the
tuffs
was
primarily
(Figure 4.2), therefore porosity and
related to porosity
peak
exhibit a strong correlation (Figure 4.7).
to
predict
the
water
contents
3aturation
must
This can be used
necessary
for
high
amplification, given porosity data.
Porosity was thus a controlling parameter for both
amplification
factor
and the peak saturation.
Less porous
rocks were better able to amplify strain (Figure
required
was
higher
stiffness
and
4.8),
water content to do so (Figure 4.9).
predominantly
result
a
its
effects
of
the
variations
in
been
inferred
but
This
rock
on cracks, as discussed later.
There was no constant volume of water in the rocks as
have
the
might
from the inverse slopes of Figures 4.8
and 4.9.
Table 4.1 lists
sample
at
several
the
amplification
saturations,
and
factors
graphs
from
which
these
compiled in Appendix C.
-25-
numbers
each
Table 4.2 contains a
summary of the important physical parameters of
The
for
were
each
derived
rock.
are
E
0
-c
w
SANDSTONE
BEREA
5%
105
Saturation
S234
STRESS,
5
6
MPa
Figure 4.3
Change in electrical resistance with stress
of the Grizzly Peak tuff and Berea sandstone
-26-
92
SILVERADO
TRAIL TUFF
10
8W
z
6 -
~
z
CD)
> 4
-
23
18 % SATURATION
90
0
Linear
Figure 4.4
30
20
10
strain, E ,
104
Relative'change in resistivity with strain, SIL
-27-
1*
'U
'
10
20
30
40
50
60
70
80
90
100
SATURATION, %
Figure 4.5
Amplification factor vs. saturation
-28-
100
RTT
GPT
0
DPR
C,)
SZT
z
SHR
0
20
PEAK
Figure 4.6
30
40
SATURATION,
50
60
*'a
Relation of Young's modulus to ;aturation
at maximum amplification (peak aturation)
-29-
70
40
30
20
10
10
20
30
PEAK
Figure 4.7
40
60
50
70
SATURATION
Porosity of tuffs vs.
-30-
peak saturation
80
RTT
GPT
z
0
jDPR
SIL
LL
S7T'
--
0.
I I SHR
04
20
30
40
50
POROSITY, %
Figure 4.8
Maximum amplification of tuff
samples vs. porosity
-31-
60
GPT
I
DPR
]1
±
SIL
z
SZT
0
LL
SHR
2
i104
X
10
20
PEAK
Figure 4.9
30
40
SATURATION,
5(
%
Maximum amplification of tuffs
vs, peak saturation
-32-
60
70
Table 4.1
Sample
GPT
SIL
SZT
Strain Amplification of the
California Tuffs
Saturation
(%)
Amplification
100
2.9x10 4
68
4.5x10 4
45
1.3x10 5
35
4.4x10 4
17
3.3x10 4
11
0
90
9.7x10 2
53
1.0x10 4
28
4.5x.0 4
18
5.5x10 4
11
6.0x10 3
8
8.8x10 3
3
too dry
100
3.7x10 3
89
6.5x10 3
74
".5x1^4
38
5.Oxl 4
23
0
-33-
(Ap/p) E
Sample
RTTa
RTTb
DPRa
SHR
Saturation
Amplification
100
1.7x10 5
83
4.2x10 4
66
1.4x10 5
47
2.3x10 5
35
3.4x10 4
82
5.0x10 4
68
1.1x105
53
2.4x10 4
100
2.1x10 3
75
1.5x10 4
31
8.0x10 4
11
6.4x10 4
7
5.5x10 4
93
2.6x10 2
3
72
1.lxlO
14
1.3x10 4
3
5.9x10
4
4.6x1J3
19
-34-
(Ap/p) _
Table 4.2
Sample
Porosity
(%)
Physical Properties of the
California Tuffs
Modulus
(GPa)
Peak Sat.
Max. Amplification
(%)
GPT
10.7
10.0
45
1.3x10 5
RTT
3.7
12.5
58 (av)
2.3x105
DPR
17.9
3.5
31
8.0x10 4
SIL
35.6
2.4
18
5.5x10 4
SHR
37.2
2.2
14 (av)
1.3x10 4
SZT
17.1
3.3
38
5.0x1.04
-35-
CHAPTER V
Discussion
rocks
The electrical behavior of the
Orange and Madden,
[1965], Brace
least in certain respects.
particularly
and
Orange,
for
surface conduction may
at
be
samples containing clays, but
mineral conduction is probably not significant.
in
[1968]),
Conduction appears to be largely
through water-filled pore space;
important,
above
to typical wet rocks studied elsewhere (Brace,
similar
was
described
An increase
stress lowers resistivity for partial saturation, as was
However,
the case for a granite (Brace and Orange, [1968]).
there
are
least two characteristics of the tuffs which
at
discussion.
are quite unusual, and require some
the
First
is
high amplification factor, and the second is electrical
recoverability.
Amplification Factor
Madden
has
shown
[1978]
that
amplification
factor
ranges from about 1600 (granodiorite of 0.4 percent porosity
at 7.5 MPa) through 800 (granite of 0.9 percent porosity) to
about
100 for porous sandstones.
was 103 to 10s
Amplification factor here
(Figure 4.5) and 10' in Yamazaki's study.
-36-
Saturation
factors
and
porosity
seemed
to
be
determining
for electrical amplification of strains.
Both high
and low saturation tended to reduce amplification.
saturation
high
there is only a small percentage increase in the
number of conduction paths before all the fluids
mobilized.
Amplification
would be low.
the
At
factors
under
have
been
these conditions
As saturation decreases, a larger fraction of
water is initially in the form of isolated pockets that
are subsequently joined.
amplification.
For
This situation
The
to
highest
still lower saturations, an increasing
fraction of the water pockets
applied.
leads
do
not
join
as
stress
is
percentage of newly interconnected cracks is
again low, and amplification is lower.
The stiffer rocks are not able to close cracks as
as
the
lower modulus samples for a given stress.
be concluded that amplification would be low in
However,
there
saturation)
not as effective.
make
small
this
overcomes
the
stiffer
cracks
Second, the lesser pore volume
changes
than
case.
in
pores.
rocks
(at
the fact that crack closure is
tends
to
interconnectivity more pronounced.
This is particularly true for rocks with a greater
of
It might
are two factors that counter this argument.
First, the higher water content in
peak
well
fraction
Less stiff rocks do not require as
much water for maximum sensitivity because the crack closure
-37-
due to the larger strain
is
great
enough
to
effectively
distribute the water.
Yamazaki [1966] found that for rocks of about the
high
porosity,
permeability.
amplification
was
associated
Based on other studies,
same
with
high
Brace, [1977],
high
permeability implies large pore diameter.
Unfortunately, it
was difficult to determine a meaningful pore diameter in all
cases
for these tuffs based on the SEM studies, so that his
observation could not be fully tested here.
conclusion
was
.pore size.
that
Clearly
amplification
there
is
The
tentative
did not correlate with
need
here
for
further
petrographic work.
Recoverability
In many ways the most puzzling behavior of the tuffs is
the
electrical
recoverability
other partially saturated rocks
degree.
during
show
a stress cycle.
this
Two parameters seemed important:
size.
Porosity is the dominant
tuffs
(35
percent)
did
factor;
effect
the
most
porous
Recoverability
improved as porosity decreased or equivalently,
-38-
any
porosity and pore
not recover well.
increased.
to
No
as
modulus
The more recoverable tuffs contained larger cracks
pores, and coarser mineral structures.
diameters 100 times
bubble
-sample,
voids
making
increased
it
stress is decreased.
poor
(SIL),
In samples with pore
electrical
insignificantly upon unloading.
increased
gas
smaller
electrical
more
the
and
resistance
Perhaps the fine
capillarity
within
the
difficult for water to retract as
Sandstones
recoverability,
also
showed
probably
consistantly
a result of the
difference in pore structures between the two rock types.
A third factor, and one which
assess,
must
in the tuffs.
phase,
the
be
difficult
to
be surface properties of the different phases
Perhaps recoverability requires a
nonwetting
such that water is dispelled from pores after stress
is released.
all
would
Unfortunately it was not possible to determine
phases
characteristics.
in
the
rocks,
This too is
studies.
-39-
a
let
alone
fertile
area
their surface
for
future
CHAPTER VI
Conclusions
1)
The rate of change of resistance with applied stress has
been shown to be very large, particularly for small strains,
in agreement with Yamazaki's results on certain
tuffs
from
Japan.
2)
Amplification
as
defined
at
a
strain
of
range between 103 and 105 for the California tuffs.
10
3)
factors
Porosity and saturation are the key factors for
optimum
amplification.
.4) Tuffs exhibit an electrical recoverability not
observed
in other rock types.
5)
Electrically sensitive tuffs exist
where
they
near
active
faults
could be used as stress monitors for earthquake
prediction.
6)
A preliminary
similar
to
the
investigation
chunk
test
of
would
electrical
seem
properties
appropriate
for
locating a field site, as a precise description of lithology
for a suitable rock is not available at this time.
7)
Further
permeability
studies
on
mineralogy,
pore
structure
and
would greatly improve the understanding of the
processes involved.
-40-
APPENDIX A
Studies of Nevada and Montana Tuffs
The
initial
phase
of
this
project
was
aimed
at
verifying the observations of Yamazaki and hopefully gaining
insight
some
into
the
involved.
processes
Yamazaki
demonstrated that the particular conditions of interest were
low stress
chosen
and
was
to
low
saturation.
consider
stresses
Therefore
only
up
the
approach
to
6 MPa and
saturation levels generally less than 25 percent.
Three samples from the Nevada Test Site
Montana
were
used
in
this
first
sample preparation and experimental
the
same
and
two
study (Table Al).
procedure
was
from
.The
exactly
as discussed previously for the California tuffs.
After testing these rocks, four sandstones were analyzed for
a comparison of rock types (Table Al).
-41-
Observations
Stress-strain curves for the tuffs and
and
the
Berea sandstone are shown in Figure Al.
indicated next to the curves.
elastic
Pottsville
Porosities are
These tuffs are more linearly
than many of the California rocks:
if any hysteresis in the curves.
there is little
Sandstones usually
retain
a permanent strain after the stress cycle.
The
effect
properties
is
of
stress
apparent
in
cycling
Figure
recoverd nearly completely upon
sandstone
did not.
on
A2.
Tunnel Beds tuff
successive
loading,
This behavior was typical
other sandstones, as seen in Figure A3.
recoverability
electrical
Berea
The second cycle of Berea showed almost
no change in resistance.
of
the
varied
among
the
of
the
Although the degree
tuffs,
it
was
both
the
of
the
consistently poor in the sandstones.
If only the loading curves
sandstones
and
the
tuffs
(with
Pottsville sandstone), have high
low
strains
(Figures
are
the
exception
amplification
A4 and A5).
observations have been verified.
considered,
factors
Thus Yamazaki's unusual
The amplification
factors
as previously defined are compiled in Tables A2 and A3.
-42-
at
Discussion
Clearly rock type is
electrical
characteristics
an
important
of
these
parameter
samples.
in
the
After the
unrecoverable nature of the sandstones was observed, further
studies were concentrated solely on tuffs.
There were
factor
with
no
identifiable
porosity
The
restricted
earlier
ideal
for
amplification
had
not
yet
been
range of low saturations led to
some misleading initial conclusions.
forture
of
or saturation because the concept of
"optimum saturation" discussed
realized.
trends
Low saturation was
by
Yamazaki's very porous rocks, but it is
not a key factor when considering tuffs in general.
After
importance
the
of
work
on
Figure
rocks
revealed
the
saturation level, certain of the Montana and
Nevada tuffs were re-tested
from
California
4.7.
With
at
the
saturations
predicted
little surprise, these rocks were
consistent with the California samples and fell nicely on to
the linear plot of Figure 4.7.
-4-3-
Table Al
Sample
Location
Rock Descriptions
(%)
Modulus
(GPa)
17.6
4.1
Zeolitized ashfall tuff
20% phenocrysts:
5% quartz
15% plag & sanidine
5% lithic fragments
75% groundmass:
40% clinoptilolite
10% clay minerals
15% opal & silica
minerals
10% glass
8.5
40% phenocrysts:
15% quartz
15% plag & sanidine
3% Ti rich augite
5% opaques
2% biotite
5% lithic fragments.
55% groundmass: fine
grained zeolitized ash
& glass
phenocrysts:
10% quartz
10% plagioclase
X% opaques
X% Ti stained biotite
10% pumice fragments
70% fine grained
groundmass:
10% zeolites
10% clay
50% ash
Porosity
TBT
Nevada Test
Site, Tunnel
Beds area 20
ATT
Ammonia Tanks
tuff, Nevada
Test Site,
Silent Canyon
Caldera
RMT
14.2
Ranier Mesa
tuff, Sleeping
Butte Caldera
segment, Nevada
Test Site
8.5
BLT
Butte,
Montana
10.2
11.0
5.8
-44-
Description
phenocr-sts:
10% q 'artz
10% p agioclase
5% bi tite
75% groundmass:
20% plag lathes
X % chlorite
35% recrystallized
glass
20% pumice fragments
Porosity
(%)
Modulus
(GPa)
29.6
4.6
5% phenocrysts: quartz
95% groundmass:
pronounced vitroclastic
texture; zeolites and
clays radially lining
glass shard structures,
20% disseminated
opaques
Berea
sandstone
17.2
Ohio,
W. Virginia
6.5
orthoquartzite:
95% quartz & chert
5% feldspar
Mixed
Co.
(Kayenta)
source
unknown
22.0
3
arkosic sandstone:
52% quartz
21% orthoclase
20% calcite
7% microcline
opaque and lithic
fragments
Navajo
sandstone
source
unknown
16.4
8.5
99% quartz
1% oxides
Spring
city,
Tenn.
3.0
7.5
46% quartz
41% orthoclase
11% muscovite
2% oxides
Sample
FPT
Location
Frying Pan
Basin,
Montana
Pottsville
sandstone
-45-
Description
SANDSTONES
T UF F S
RM
U)
to
FP
TB
AT
76
14.2
.8
BL
110.2
w
Hr
U)
2
10-
0
LINEAR
Figure Al
STRAIN
Stress-strain curves of Nevada/Montana tuffs
and the Pottsville and Berea sandstones
17.2
108
10
E
0
1
z
Cn
w
6
10
BEREA
SANDSTONE
4.7% Saturation
-0
0
1
2
3
STRESS,
Figure A2
4
5
Mpa
Change in resistance with stress, TBT, Berea ss.
108
E 6
=10
0
z
4
I-
104
I
-Figure A3
2
3
STRESS
4
MPa
5
6
Change in resistance with stress: sandstones
-48-
BUTTE LAPILLI TUFF
2
0
H
z
-
wj
z
% SATURATION
cc
0
I
2
3
4
5
6
7
8
Linear strain , C,
Figure A4
Relative change in resistivity with strain, BLT
-49-
cc
(e)4
w
0
z
9% SATURATION
13
4027
40
I
2
3
Linear
Figure A5
4
5
6
7
8
strain, E , 10
Relative change in resistivity with strain,
Berea Sandstone
-50-
Table A2
Sample
BLT
RMT
TBT
Strain Amplification of the
Montana/Nevada Tuffs
(AP/P) -4
Amplification
Saturation
(%)
4
4.2
2.3xl0
5.0
2.5x10
5.2
1.6x10
11.2
8.1x10
12.6
2.1x10
29.6
3.0x10
49.1
2.8x10
2.5
5.0x10
4.7
0
5.1
5. 0x10
5.4
6.6x10
3
4
3
4
3
4
38.0
1.0x10
45.0
1.6x10
2.8
7.0x10
4.8
8.0x10
7.3
8.0x10
9.9
1.0x10
12.9
1.0x10
20.1
4.4x10
46.8
2.0x10
-51-
2
2
2
4
4
3
3
3
4
4
3
3
Sample
Amplification (Ap/p)
Saturation
(%)
FPT
ATT
2.1
2.2x10 3
3.8
1.2x10 3
6.7
1.7x10 3
15.2
3.2x10 4
17.1
2.4xl03
22.2
8.8x10 4
58.0
8.0x10 2
7.0
1.2x102
11.1
1.3x10 2
14.3
1.5x10 3
18.6
5.0x10 2
56.0
1.0x10 4
-52-
Table A3
Sample
Mixed Company
(Kayenta)
Strain Amplification of Sandstones
Saturation
(%)-
Amplification
3.8
9.2x10 2
7.6
4.0x10 2
21.8
1.0x10 4
34.3
1.6xl03
44.0
94.0
Navajo
Berea
2.5
. 3.1x10 3
5.6
2.4x10 4
12.3
3. 2x10 4
19.3
9 .2-:103
33.0
8.6x10 2
4.7
4.4x10 4
9.2
3.9x10 4
13.1
1. 2x10 4
26.9
1. 6x10 3
40.0
9.1x10 2
-53-
(Ap/p)
Sample
Pottsville
Amplification
Saturation
(%)
6.4
0
8.5
0
13.9
0
21.0
0
24.9
0
32.5
0
42.5
1.3x10 0
63.4
1.0x10
-54-
(Ap/p)
1
-4
APPENDIX B
The Chunk Test
devised
was
test
This
to
quickly
the
investigate
electrical properties of samples while still in a crude hand
as
the
the same manner as
A,
Appendix
sample size.
a
tests
performed
the
on
tuffs showed that resistance cycled in
Montana
and
chunk
However,
precisely known.
Nevada
of the irregular shapes were not
factors
geometry
qualitative,
Results were strictly
specimen configuration.
although
samples
cylindrical
the
described
in
to a lesser degree due to the larger
Therefore, the chunk test was indeed valid for
preliminary
sorting
by electrical properties.
Table Bl
with
their
Rock fragments were broken off into more or less
equal
lists the
samples
used
in
this
test
along
locations, and a brief hand specimen description.
Sample Configuration
shapes
about
5 cm long.
Two lead sheets, 0.02 cm thick by
1.9 cm square were conformed on to
sample,
opposite
sides
of
the
with copper wire soldered to each sheet, leading to
the resistance measuring
The
circuit.
lead
was
around the edges to the rock to avoid separation.
assemblage was
then. potted
in
-55-
Dow
Corning
epoxied
The whole
Sylgard
186
silicone elastomer with
extending
wires
the
out
of
the
cylindrical case of rubber (Figure Bl).
Experimental Procedure
Potted samples were placed in a beaker of
inside
water
a
15
Sylgard
or
cm diameter argon gas pressure vessel.
corrosion
Kerosene is prefered to prevent
However,
kerosene
K-Kote
then be coated with Kenyon
the
vessel.
kerosene and the samples must
in
swells
of
expansion,
avoid
to
a
process which adds a few days to the sample preparation.
Aesistances were measured at natural water
method
2
of
Appendix
D,
content
by
while increasing and decreasing
hydrostatic pressure to a maximum of 5 MPa.
The
set-up
is
schematically illustrated in Figure B2.
Data
Figures B3 through B5 show
to
response
tuffs.
stress
cycling
the
electrical
resistance
on a number of the California
Samples which exhibited little or no change are
included.
not
The unloading curves consistently fell below the
loading curves as with the cored rocks.
-56-.
Samples are ordered in
change
in
Table
B2.
terms
of
relative
resistance
Those with an asterisk were the ones
chosen for more detailed study.
Some very
sensitive
were not selected because of their friability.
-57-
rocks
Table Bl
Sample
Location and Hand Specimen Descriptions
of the California Tuffs
Locati on
Hand Specimen Description
PIN
Pinnacles N ational
Monument
Dense green tuff, numerous lithic
fragments and phenocrysts
SJB
San Juan Grade Rd.
(Salinas Rd .),
San Juan Ba utista
Conglomerate in tuffaceous matrix,
pebbles up to 0.75 cm in diameter
COT
Stony Point
Quarry, Cot ati
Fine grained powdery grey clay in
0.5 cm thick bed
CMV
Coomsville Rd.,
Napa
Plum colored matrix, lithic
fragments up to 1 cm, plagioclase
lathes visible
MWT
Montecello Rd.,
Napa
Welded grey pumice fragments,
light and porous
SIL
Silverado irail,
St. Helena
White rhyolitic tuff, fragments of
white pumice up to 1 cm, lithic
fragments (0.5 cm), plagioclase
lathes
PET
Petrified Forest
Rd., Calist oga
Rhyolitic tuff similar to
Silverado, more weathered & friable
SIHR
St. Helena Rd.,
St. Helena
Coarse, grey, hard nd very
crystalline tuff, e: tensive and
well exposed in roa" cuts
DPR
Deer Park Rd.,
Sunset Poir t
.Very
dense fine grained purple and
grey matrix with large pores; no
noticeable lithic fragments
-58-
Sample
Location
Hand Specimen Description
RLS
Robert Louis
Stevenson State
Park, Calistoga
Very hard grey tuff weathers to
yellow-brown; lithic fragments up
to 0.5 cm; fractured
GPT
Little Grizzly
Peak,
Berkeley Hills
Large massive outcrop, coarse and
dense lithic fragments up to 2 cm,
highly fractured
RTT
Round Top Hill,
Berkeley Hills
Basalt tuff; black, fractured and
weathered into boulders; part of a
cinder cone 400 m in diameter
SZT
Siesta cinder cone
Siesta Valley
Brown lithic tuff filled with
veins of zeolite
DRY
Grizzly Peak Rd.,
Berkeley Hills
Buff colored airfall tuff,
weathered and friable laver a few
meters thick
-59-
copper wire
lead sheet
sample
rubber
Figure Bl
Sample configuration for the chunk test
F.igure B2
Chunk test experimental apparatus
-60-
0
E
(n
C0
w'
I
2
3
STRESS
Figure B3
4
5
MPo
Change in resistance with stress;
SHR, DPR, RLS, DRY.
-61-
106
tn
E
.c:
0
w
.
10
O-
I
Figure B4
2
3
STRESS
4
5
MPo
Change in resistance with stress;
MWT, SZT, GPT.
-62-
6
10
(n
E
0
w
z
(n
(n
w
1
2
3
STRESS
Figure B5
4
5
MPa
Change in resistance with st--ess;
SJB, SIL, RTT, COT.
-63-
6
Table B2
Relative Ordering of Electrical Properties
of California Tuffs
Good
Fair
Poor
No Effect
COT
SZT*
SHR*
CMV
DRY
RLS
SJB
PET
GPT*
DPR*
MWT
PIN
RTT*
SIL*
* Studied in detail
-64-
APPENDIX C
Stress-Strain and Electrical Behavior of
Selected California Tuffs
This section contains the complete set of data
California
tuffs
that
for
the
six
tuffs
the
not included in chapter 4, as
were
they are similar to the examples shown there.
curves
on
are
Stress-strain
illustrated in Figure Cl.
Figures C2 and C3 show the change in resistance with
for each sample at an arbitrary saturation.
stress
The curves were
chosen to show the most typical electrical recoverability of
that
sample.
Figures
C4
through
C8
are
plots
relative change in resistivity with strain and were
from the stress-strain and resistance data.
-65-
of the
derived
RTT
GPT
DPR
SIL
SHR
SZT
0.
(n)
(14
U)4
2
10-3
0
LINEAR
Figure Cl
STRAIN
Stress-strain relation for the California tuffs
108
07
GPT
35%
Saturation
SIL
18 % Saturation
SHR
10 % Saturation
106
05
I
Figure C2
2
3
STRESS
4
MPa
5
Change in resistance with stress;
GPT, SIL, SHR.
-67-
6
z
106
Saturatior
DPR
10%
SZT
74 % Saturatio
RTT
90% Saturati
105
104
1
Figure C3
2
3
STRESS
4
MPa
5
Change in resistance with stress;
DPR, SZT, RTT
6
$20
Q-
C: 15
10
5
45 %
z8
SATURATION
35
35
2
Figure C4
4
c , >0
strain,
Linear
6
8
Relative change in resistivity with strain, GPT
-69-
ST. HELENA ROAD (SHR)
4
-
z
W
z
26 %
SATURATION
10
73
0
20
10
Linear strain, E
Figure C5
,
30
10
Relative change in resistivity with strain,
-70-
SHR
12
10
m
CC
z 6
(D
z
-
-
2-
38%
20
010
Linear strain,
Figure C6
SATURATION
30
C4
Relative change in resistivity with strain, SZT
-71-
12
DEER
PARK ROAD
(DPR)
10
8
(D)
z6
-J
w
4
31
2
7%
SATURATION
75
95
0
Figure C7
2
4
6
8
strain,
Linear
10
E
12
14
16
18
104
Relative change in resistivity with strain, DPR
-72-
50
40
m30
z
z
r20
m
66
10
100%
SATURATION
47
I0'
583
I
Figure C8
3
2
Linear
5
4
10
strain, e,
6
7
8
9
Relative change in resistivity with strain, RTT
-73-
APPENDIX Dl
Resistance Measuring Techniques
There
are
resistance
of
accuracy.
Most
several
a
rock,
of
techniques
each
these
for
with
methods
measuring
the
trade-offs on ease and
involve
matching
the
voltage across a variable resistance decade with that across
the rock sample.
at
10
Hz
In all cases, the source voltage was
kept
AC, to minimize any frequency effects that could
occur at higher frequencies.
See Appendix
D2
for
a
nore
complete description of frequency effects.
Method- 1.
R
rock
V.
in
M
ti/
Rbox
V
B
M VA
The particular circuit used for the early work
Nevada
and
Montana
tuffs
Orange and Madden, [1965].
vacuum
tube
volt
meter
was
VB are
model 40011).
circuit diagram, it follows that V
V.
in
measured
on
Rox
R rock + Rbox
a
From the above
=VB and the system is
voltage divider where
VA
the
the same as that in Brace,
VA and
(HP
on
a
is known and a particular VA/V.
Since Rb
then Rrock
can be calculated.
is chosen,
To simplify the calculation,
assume that Rbox is much less than Rrock , by
appropriately
adjusting the voltage ratio.
Now
Rrock = Rbox
V.
VA
The ratio VA /Vin is determined to make the
in
resistance
fall
a reasonable range on the decade box and to introduce no
significant
errors.
Rrock = 100R box
For
(Turn
example,
the
with Vin /VA
voltmeter
range
orders of magnitude to measure VA , then
adjust
lo0
down
by two
the
decade
until the meter needle comes to the same point as it did for
VB)-
A
1%
theoretical
Rexact rock = 9 9 Rbox
negligable.
is
possible
*
error
The
is
introduced
difference
is
because
considered
By turning the voltage down in this manner,
to
measure
it
rock resistances that are greater
than the resolution of the variable resistance decade.
The advantage of this technique is that it is quick and
requires no calculation.
In order to find t'e resistance of
the rock, one need only add two orders of
the
resistance read off the decade box.
mi nitude
to
Most of the Nevada
and Montana tuff samples were measured in this way.
-75-
on
Problems:
Since the signal to be measured on the meter
has been reduced substantially, there is a potential problem
In fact
of noise dominant errors.
signal
to
resistance of the rock will have a
(as
much
as
factor
a
situation, if V /V.
A
in
1/1000,
then
box
gives
better
be
the
In
10).
the
error
large
noticeably
noise dominant
is chosen to be 1/10 instead
1/100
of
the signals are larger and the resistance
or
Rrock must
of
1000,
V /Vin =
enough that the measured
high
is
ratio
noise
if
resolution.
used,
The
exact
there
otherwise
is
formula
a
10%
for
error:
Rrock 9Rbox'
On the practical side, this choice of VA/Vin is not
as
quick, as it requires more mental gymnastics to calculate 9R
than 10R.
However, it is more accurate.
If the resistance of the decade, which has a maximum of
1
megohm, approaches the input impedance of the meter, then
the effective resistance is reduced:
1
Reffective
R
effective
1
+
=1
Rbox
meter
= Rbox . Rmeter
Rbox + Rmeter
The effective value is substituted into the R
VA
7in
Rbox
R rock + R box
VA
becomes
in
-
R
values:
Reffective
+ R
rock
effective
The HP 400H voltmeter has an input impedance of roughly
2.5
megohms.
taken
into
This
account
impedance
when
is rather low and should be
dealing
with
dry
rocks
whose
resistance can be as high as 10 megohms.
Method 2.
+
+
M
in
-
box
R rock
_
technique
of
RF
Method
voltages
as
on R rock
method 1.
the
same
until
Now
necessary
resistance
is
the same voltage is obtained as through the
Rrock=Rbox*
in
matching
A voltage is chosen with the switch
Then with the switch on Rbox, the
adjusted
rock.
2. involves
this
An
circuit
additional
resistor
RF is
otherwise Vmeter always equals
Vin*
Choice of RF:
sensitivity
of
the
To find a suitable
meter
value
of
RF
,
is analyzed with respect to the
power transfer between the circuit and the load (rock).
Joule's Law,
2
2
____R
pR +Rr
the
V
Rr
(1 + RF / Rr
r
-77-
2
By
According to this equation, the power in the load is zero if
the
of
resistance
large.
rock is either very small, or very
the
some
Thus there must be
power is a maximum.
dP rock
_
V
R
dR
- rock
2
optimum
value
where
the
Differentiate and equate to zero:
2R /R 2
F
r
+
3
~
l+RR
rock (1 + RF/Rr)
2R
F = 1+
Rrock
RF
=
V
2 R2 (lRR
RF
(1 + RF/Rr)
R
F
R
rock
0
-
rock
Therefore, a value of RF should be chosen that is close
for maximu. sensitivity~ of
value of R
rock
This
the meter (greater accuracy 'in matching voltages).
to
expected
the
method was used in the "chunk tests" on the California tuffs
mainly because it is fast, and a switching box (R =1 megohm)
was readily available courtecy of D. Johnston.
A problem arises when Rrock is greater than the maximum
Rbox,
If the vnltage range of
mainly low saturation rocks.
the meter
is
switched
down
as
in
method
simpiifying assumptions can not be made.
-78-
1,
the
same
V.
(
V
(
in
in
Rk
rock
Rrock + RF
Ml
Rb
box
Rbox + R F
M
if V Ml
=
R rock
M2
t
0VM2' then
Rbox
= 100
Rrock +KRF
Rbox + RF
RrockRbox + RrockRF = 100RboxRrock + 100 RboxRF
Rrock (RF - 99Rbox
100 RboxRF
R1OR
R100boxRF- -9 R
rock
RF -99Rbox
Since RF = 106 ohms,
then
This
is
true
if
Rmeter is
much
greater
than
Rrock'
Otherwise the same problem with a low impedance meter exists
because the meter is in parallel with the rock and the box.
-79-
box effective
1
meter
ox
Ret1
mete
Rrock equation in box
'rock effective
As can be seen from the above discussion, the switching
box
method
is no longer quick when trying to measaure high
impedance rocks.
Method 3.
V.
in
VB
Bridge method:
RA and RB are fixed resistors.
adjusted until the bridge is balanced, i.e.
zero volts.
the meter reads
Then VA=VB'
VA
V.
in
RA
RA + Rbox
RB
R B + Rrock
RA
+ R
) = R(R
B A + Rbox
R((RB +Rrock
rock
RA
-80-
box
Rbox is
VB
V.
in
The ratio of RB to RA is
the meter.
For
Rrocke
Rbx max =
1
instance,
if
megohm,
then
chosen co scale Rbox to
of
RB and
sensitivity to the null.
Now resistance
make RB/RA=10.
made.
The
absolute
RA are set to gain the maximum meter
RB is chosen to be around the same
(See sensitivity derivation in
order of magnitude as R rock.
Then RA~ 100 kilohm for a typical low saturation
method 2).
This
rock.
and
megohm
1
Rrock max =
measurements greater than Rbox can be
magnitudes
on
depend
way
any
Note that the response does not in
value
is optimum for the decade box, since it
has maximum resolution in the middle ranges.
the
Because the resistance of
on
depending
the
rock
can
mineralogy and degree of saturation, say
between 1 kilohm and 10 megohm, it may be best to
sets
resistors
of
appropriate
values
widely
vary
for
of
optimum sensitivity.
RA
This
.
have
two
Each set has
modification
is
not
tremendously important to the workings of the circUit unless
there is reason to be concerned with sensitivity.
RB = 10 kilohm for 1 kilohin<Rrock<100 kilohm
and
RB = 1 megohm for 100 kilohm<Rrock<1 megohm
The bridge method has a
other
two methods.
definite
advantage
over
the
Since the meter does not enter into any
calculations, there are no errors due to low input impedance
or noise.
-81-
The source for the
A problem common to all techniques:
resistance
10 Hz.
other
measuring
circuit has an alternating current of
The strain gauge and corresponding
hand
runs
on
DC,
changes in the DC voltage.
rock
the
is designed to detect small
Since the AC and DC currents are
AC current appears in the strain signal.
the
on
the sample, there is a coupling in which the
in
juxtaposed
and
circuit
and
between
The epoxy
gauge is not always enough to ensure
strain
proper insulation, particularly with saturated samples.
The
result is a 10 Hz vibration of the pen along the strain axis
as
much
as
2
cm
wide,
(leading
to
a
fuzzy
mighty
stress-strain curve!).
There are two solutions:
1) Disengage the pen while making a resistance
This
tends
to
slow
down
thirteen measurements to be
twenty-six
times
to
flick
measurement.
the experiment, since there are
made
on
the pen.
each
run,
The chance for human
error arises since one sometines forgets to engage
after
therefore
the
pen
a measurement, and part of the stress-strain curve is
lost.
2) Build a low pass filter into the chart input.
detailed
description
A somewhat
of the filter is in order as it has a
significant effect on the chart response.
-82-
FILTER
A standard low pass filter has response
(s)
VCiu
in (s)
V
Log
(
)
inin
v
1 + sRC
0
s= complex frequency 2Trf
R= resistance
C= capacitance
-
2
Log s
some
Above
db/decade.
by
20
cutoff
s=l/RC,
the
response
drops
(i.e.
V 0 /Vin goes
as l/s) .
Since it is not
advantageous to vary s, for reasons stated in Appendix D3, R
and
C
are chosen such that 10 Hz is well above the cutoff,
On the other hand,
respond
sluggishly
if
the cutoff is too low,
since
the
pen
will
the filter causes a time delayed
response which goes as e-t/T
-83-
input
chart
output
Thus
if
the
instantaneously),
strain
the
is
pen
will
true value of strain and the
slightly incorrect.
where
10
(say
not quite catch up to the
stress-strain
curve
will
Hz
There are tight constraints
is
and Tmin ,
response curve,
quickly
be
This is a real problem because 10 Hz is
quite a low frequency.
on Tmax,
changed
just
a
imposed
at the break point of the
value
that
will
not
cause
noticEble delay in the pen.
The solution is to make sure that
pumped
up
or
released
delayed response.
rate
too
fast,
the
to
This introduces the
stress
not
allow time for the
question
dependance of the system as a whole.
experiment is not to include strain
is
rate
of
strain
The point of the
as
a
parameter.
However, running a relatively slow test for response reasons
goes hand in hand with the fact
allowed
for
the
The pore water must
rock
that
there
must
be
time
to equilibrate to the new stresses.
redistribute,
permeability.
-8.4-
a
factor
dependant
on
Note that if
frequency,
the
experiment
were
run
at
a
then "sluggishness" could be avoided altogether.
When the amplitude of the noise is low, a smaller
can
be
higher
used
to
alleviate
capacitor
the response problem.
higher frequencies, say 10 k1z, the pen will not
At even
even
have
time to respond to the noise.
Both
solutions
disadvantages.
One
measured with DC.
ions
through
to
the
noise
their
This would require the actual movement of
the sample.
If it is assumed that the charge
the
water
phase
rather
by mineral conduction, then a DC current would tend to
electrolyze the water.
be
have
might wonder why the resistance is not
is being transmitted primarily by
than
problem
At one of the electrodes there would
dissociated water forming oxygen and hydrogen gas.
is far from a desirable situation,
especially
since
This
fixed
partial saturation is an important factor in the experiment.
-85-
APPENDIX D2
Choice of Electrode Material
The ideal electrode material for
measurina
resistance
should have the following qualities:
1) Be a good conductor.
2) Maintain a good electrical contact with the sample
3) Have no stress effect (change in measured resistance with
applied stress) .
4) Have no frequency effect, or at least no frequency effect
within a given working range.
with
Four different electrode combinations were tested
a 20 kilohm resistor in the following configuration:
Piston
+-Teflon
_
a)
4Metal
'r'-Wet paper
20K
b) lead sheet + wet layer
Westerly
Granite
c) copper sheet
20K__ -Paper
+-Metal
Base
d) copper sheet + wet layer
Teflon
The wet layer of paper is
electrical
contact
to
included
the
inconsistent, because as the
increases.
lead sheet
rock.
paper
to
-ensure better
Results
dries,
the
tend
to
be
resistance
The copper was annealed before testing to reduce
-8.6-
strain hardening effects, and enable the
better
to
stress
cycled
permanent,
rock
the
many
to
conform
However, as the copper is
surfaces.
times,
sheet
strain
can
hardening
The electrical conduction
even after annealing.
because
properties of the copper lattice will vary slightly
of
distorted
become
This effect
grains and dislocation pile-ups.
will also cause the material to become less pliable.
There
dependance
appeared
to
be
no
significant
frequency
with the various electrode combinations, however
both the wet and dry copper in the stress test showed higher
resistance values than the correct value of 20 kilohms + 200
ohms (Figure Dl).
The
material.
dry
lead
sheet
chosen
as
the
electrode
Lead is able to conform better to the surface of
the sample than the copper
stress.
was
This
is
sheet,
particularly
due
to
important
its
when
sandstones, which can have a very course surface.
no apparent stress effect with the dry lead.
-87-
low
yield
testing
There was
'
9"
9'
9'
9'
WE
T
2.10
COPPER
9'
9'
.066I
z
II
00
9'
cc
.4
00
I
It
9'
j
9'
DRY COPPER/
I
"I
A
I
2.00
/
DRY LEAD
9'
'9',
40
50
S T R E S S , MPa
Figure Dl
Stress effect of various electrodes, 20K resistor at
10 HZ
APPENDIX D3
Freguency Effects
When dealing with high impedance
effects
the
become
frequency
saturated
rock
important.
rocks,
spectrum.
Consider
a
dry
between two copper wires.
characteristics
order
This is particularly true over
or
of a capacitor.
partially
Because the rock
is so much less conductive than the wires, it
the
second
has
some
of
A model of the system
could be represented as shown below.
.dV--
C
i=C
R
i=V/R
dt
Since the current through the capacitor i=dV/dT C, then
at
high
through
frequencies,
C
increases
correspondingly
less
dV/dT will be large.
with
higher
frequencies,
current through R.
smaller voltage across R, and hence a
true
value.
As the current
there
is
The meter reads a
smaller
R
than
the
The actual response can be found by analyzing
ZR1 1 ZC
-89-
Z
= R
jfC
ZC
1
jf
jC
.1C
jfC
+
=
1
R
jfR
Log Z
Z
R
1 + jfCR
Log f
When the frequency is small, Z=R, and when the frequency
large
Z falls off as 1/w.
is
There will be a family of curves
for varying resistance.
The accuracy of standard
film
resistors
at
ohms
107
breaks down at 100 Hz, therefore a frequency of 10-100 Hz is
suitable to eliminate the capacitance effects of
the
fixed
resistors in the electrical resistance bridcre of method 3.
The plot in Figure
varying
high
saturation
internal
D2
was
obtained
with no axial stress.
Since
capacitance.
the
using
rocks
of
The tuffs have a
this
of
point
experiment is not to investigate the frequency dependance of
resistance
frequency
samples.
in
as
the
tuffs,
it
is
necessary
to
run
the
low as possible to measure the high impedance
Above 10
ohms
. (at 10 Hz),
the
resistance
are already on the sloping part of the response curve.
they represent a minimrum resistance, and are a
-90-
values
Thus
contributing
factor to the scatter in the data.
Frecuencies of less than
10 Hz are not desirable due to the electrolyzing problems of
a DC-like current.
-91-
RTT (77)
GPT (45)
SZT (75)
E
0
z
SZT (86% Sot.)
I:
10-
sHR (58)
SIL (65)
S HR '(34)
lei00
102
O',
FREQUENCY
Figure D2
04105
103
(Hz)
Resistance fall-off with frequency for the
California tuffs at varying saturations
REFERENCES
Brace, W.F., Electrical resistivity of sandstone
Final report to Defence Nuclear Agency, contract no.
DNA-001-74-C-0057, 40p, 1974
Brace, W.F., Permeability from resistivity and pore shape,
J. Geophys. Res., 82(23), 3343, 1977
Brace, W.F., and A.S. Orange, Electrical resistivity changes
in saturated rocks during fracture and frictional
sliding, J. Geophys. Res., 73(4), 1433, 1968
Brace, W.F., A.S. Orange, and T.M. Madden, The effect of
pressure on the electrical resistivity of water saturated
crystalline rocks, J. Geophys. Res., 70(22), 5669, 1965
Carozzi, A., Microscopic Sedimentary Petrology
John Wiler & Sons, Inc. New York, 1960
Madden, T.M., Electrical measurements as stress-strain
monitors, U.S.G.S. Office of Earthquake Studies.
Proceedings of conference VII: Stress and strain
measurements related to earthquake prediction. Open
file report 79-370; Menlo Park, California, 1978
Parkhomenko, E.I., Electrical Properties of Rocks
Plenum Press, New York 314p, 1967
Rik'itake, T. and Y. Yamazaki, Electrical conductivity of
strained rocks: The fifth paper. Residual strains
associated with large earthquakes as observed by a
resistivity variometer.
Bul. Earthrauake Res. Inst. 47, 99, 1969
Sprunt, E.S., and W.F. Brace, Direct observation of microcavities in crystalline rocks, Rock Mechanics and Mining
371974
Sciences and Geomechanics Abstracts, 11(4),
Stesky, R.M., and W.F. Brace, Electrical conductivity of
serpentinized rocks to 6 kilobars.
J. Geophys. Res., 78(32), 1973
Yamazaki, Y., Electrical conductivity of strained rocks:
The first paper. Laboratory experiments on sedimentary
rocks. Bul. Earthquake Res. Inst. 43, 783, 1965
Yamazaki, Y., Electrical conductivity of strained rocks:
The second paper. Further experiments on sedimentary
rocks. Bul. Earthquake Res. Inst. 44, 1553, 1966
-93-