ChapterI

EFFECT OF RATE OF LOADING ON STRENGTH

AND YOUNG'S MODULUS OF ELASTICITY
OF ROCK

by Richard L. Stowe and Donnie L. Ainsworth

The static, rapid, and shock-loadingresponseof rock is o.f interest to

many in the field of rock mechanics. For example, the effectsof loading
rates on strength and stress-straincharacteristicsof rock are quite impor-
tant in defining input loading for use in the design of underground pro.-
teerive installations. Responsedata on rock subjected to shock loading
over a wide range of pressuresare also vital in the field of underground
nuclear testing. In addition, material response is needed in design of
shock-absorbingliners.
A number of investigators have reported that ultimate compressire
strength of rock increaseswith increased rates of loading? 6 Wuerker a
stated that modulus of elasticity, obtained from sonic determinations,
increased in somerocks by a factor as high as 4.5. It appears that only
a limited amount of work has been done on comparing static modulusand
modulus obtained from impact testing2 Phillips 4 showed that with
increased rates of loading the deformation decreased. More recent work
by Atehley and Furr 5 showed that when moderate-strength concrete is
stressedwith increasedloading rates, the axial strain at failure increased.
One would expectthis to occurif we think in terms of purely elasticmate-
rials. For example, if E is constant and maximum strength O'max, in-
creases,the axial strain would also increase. However, in rock testing
thus far we have worked almost exclusively with nonhomogeneous and
inelastic rocks. From this fact alone the increase in axial strain with
increasedload application is noteworthy.

Richard L. Stowe and Donnie L. Ainsworth are, respectively, Geologist, Concrete

and RockPropertiesSection,and Physicist,EngineeringPhysicsSection,Engineering
Mechanics Branch, Concrete Div., USAE Waterways Experiment Station, Jackson,
Miss.
4 BASIC AND APPLIED ROCK MECHANICS

Many techniques have been devised for studying shock response of

rock.7,s Some are based on rigid mechanical considerations and do not
considerwave propagation aspectsat high loading rates. Plate-impact or
explosive-loadingtechniqueswhich measurethe shock-propagationveloc-
ity and the free-surface velocity of the sample are successfultechniques.
These techniques are characterized by the introduction of a known wave
form into the material and observing'the wave form after propagating
through a known distance. These techniquesare used primarily in the
hydrodynamic region. In low-pressure experiments (< 100 kilobars),
the specimen is not in the hydrodynamic region, but rather in a state of
uniaxial strain. The shock-loading technique described here is within
the region of uniaxial strain.

EXPERIMENTAL TECHNIQUES

Loading Rates
The loading rates usedduring the static confined(triaxial) and uneon-
fined (tensile and compressirestrength) testingwere lessthan 5 x 10a psi
per see, specifically, 1, 50, 500, and 2250 psi per see. Rates between
5 x 10• and 1 • 10* psi per see are defined as rapid-loading rates, while
rates greater than 1 • 10* psi per see are defined as shock-loadingrates.

Techniques
Static uneonfinedand confinedtestswere conductedusing a 440,000-1b
universal testing machine and a 10,000-psi-capacitytriaxial chamber.
The static tests,tensile splitting, compressirestrength, and triaxial shear
tests were conducted in accordancewith the standard procedures,•i.e.,
CRD-C 77-61, CRD-C 19-65, and CRD-C 93-64, respectively,with the
exceptionof lo.adingrates. CRD-C 93-64 was further modified slightly
to accommodatewave velocity equipment for purposes of determining
the effectsof axial and lateral pressureson compressionalwave velocities
parallel to the coreaxis. Confiningpressuresof 250, 1000, 4000, and 5000
psi were utilized during triaxial testing.
The method of triaxial testing as modified with compressionalwave
velocity equipment involved the use of a cylindrical specimenencasedin a
flexible membraneand placed in a triaxial chamber,subjectedto a con-
stant lateral fluid pressure,then loaded axially to failure.
Fig. 1 is a sketchshowingthe triaxial chamberand the accessories used
inside the chamberfor velocity determinations.End plates (housingthe
transducers) and bearing plates were used to allow for a size reduction
to the NX-size samples. Aluminum end plates and bearing plates were
 ]•FFECT OF RATE OF •LOADING ON STRENGTH

Loading
Head
I
Chamber Loading Ram
Transducer

End Plate

Bearing Plate

Membrane Sample

Bearing Plate
End Plate

Transducer

Base Plate

Lead Outl Chamber Bearing Plate

Fig. 1--Triaxial chamber witIs transducers.

used becausethe impedance of aluminum is quite closeto that of most

denserock. The travel time through the end plates and the bearing plates
is measuredprior to testing; this travel time is subtracted from the travel
time through the plates and rock sample. The equation V:d/t was used
to obtain the compressirevelocity where V is the velocity, d is the length
of the spedmen in feet, and t is the pulse travel time in milliseconds
through the sample.
The rapid tests were accomplishedusing both a drop tower and a hy-
draulically operated 200-kip loader. The drop tower had a capability of
3840 ft-lb of energy with a maximum striking velocity of 22.7 fps pro-
dueed by a falling massweighing 384 lb guided by two cylindrical steel
columns. The masswas remotely triggered and allowed to fall free from
a predetermined height. Friction brakes built into the falling masspre-
vented any rebound of the massafter impact.
The 200-kip loader consistsof a large hydraulic actuator and a rigid
support system as shown in Fig. 2. The actuator is pressurized with a
low-volume, high-pressuremultiplier. The actuator has three pressure
chambers,abovethe piston, below the piston, and betweenthe rupture
disks. The rupture disks act as a rapid-openingvalve. The machineis
pressurizedby slowbuildup of pressureaboveand belowthe piston while
a slightpreloadon the specimenis maintained. Concurrently,pressureis
built up in the volumebetweenthe two rupture disks; the pressurebe-
tween the rupture disks is maintained at exactly one-half the pressure
belowthe ram, thereby enablinghalf the total pressurebelowthe piston
6 BASICAND APPLIEDROCKMECHANICS

Piston

Rupture Disk No. 1

Ram

Ram Load Cell

Specimen

Pedestal Load Cell

Pedestal

Foundation

2•ig. 2---200-kip loader.

to besupportedby thefirstrupturediskandtheremaininghalf of this

totalpressure
to besupported by thesecond
rupturedisk.Ontriggering
themachine,therupturedisksburstandmovetheloadingramontothe
specimenwhich is positionedbelowthe ram.
Shock
loading
is obtained
in specimens
usinga system
involving
con-
tinuousmeasurements of boththevelocityof propagationof a shock
wave
andparticlevelocitiesin rockandothernonferromagnetie materialsfrom
the passage of a shockwaveproduced by a fiat-plateimpact.A stress-
strainrelationshipis deduced from the waveand particlemotiondata.
Thesystem hasanupperlimitof approximately 40kilobars(kb).
The experimental methodis primarilybasedon a technique usedby
FrasierandKarpovlOfor directlymeasuring theparticlevelocityin a
material. In this techniquea fine wire is embeddedin the material and
placed
in a magnetic
field.Anysubsequent
movement
of a portionof the
wire whichcutsthe magnetic-flux
linesproduces
an electromotive
force
(e) proportional
to theinstantaneous
velocity
asshown
in thefollowing
equation:

e-- l?lv (1)

 EFFECT OF RATE OF •JOADING ON STRENGTH

SCALESI MAGNETIC
POLE
I
• I

Fig. 3--Gas-operated gun and target configurationat muzzle end of gun barrel.

wherefi is the magneticfield strength,1 is the length of wire cutting the

flux lines,and v is the velocity of the wire (particle velocity).
Severalwires are placedin a test specimenand the voltagesmonitored
on oscilloscopes and recordedon photographs.The velocity of propaga-
tion can be determined from the initial rises of the voltage signals from
the induction wires.
The specimenis mountedat the end of the compressed-air-gun barrel
in a magneticfield (Fig. 3) and the barrel evacuatedto approximately
15/,. The stresswaveswill be producedby impacting the specimenswith
an aluminum projectile (Fig. 4). Impact velocitiescan be obtained up
to 1000 fps.

Fig. 4--Schematic of experimental setup.

 BASIC AND APPLIED ROCK MECHANICS

Thefinalstatebehindthe stresswaveis determined

by the application
of the conservationequations. Conservationof mass and momentum
acrossthe shockfront requiresthat:
poUs----pl
( Us- Upl) (conservation of mass) (2)
Po+poU82--P•+p•(U• - Up1)
2 conservation ofmomentum) (3)
•vherepois initial density,Usis shockvelocity,p• is densitybehindthe
shockwave,Up•isparticlevelocitybehindtheshock wave,andPois initia!
pressure,P•-poUsUp•+Po-the pressurebehind the shockwave. The
strain correspondingto P• is

c•-1- P_t
p•
= U•_2
Us
(5)
Theseequations ar• basedonthe assumptionthat an equilibriumstateof
stressis reachedbehindthe shockwave. In thissimpleform,they are
validonlyin theregionof uniaxialstrain.Theseequations correspondto
the Hugoniot or the locusof shocked-endstatesreached.

Rock Specimens
The rocksstudiedall camefrom the U.S. AtomicEnergy Commission
's
NevadaTest Site (NTS) at Mercury,Nev. The granodiorite(granite)
camefrom the PILE DRIVER experiment,the basaltcamefrom Buck-
board Mesa, and the tuff came from the RED HOT-DEEP WELL ex-
periments.The graniteis a light-gray,dense,coarse-grained, unweath-
eredrockwith phenocrysts of orthoclasefeldsparpresent. The basaltis
a light-gray,dense,fine-grained, unweathered rockcomposed of plagio-
clasefeldsparand lesseramountsof pyroxene,olivine,and magnetite.
The tuff is a light greenish-yellow
to brownish-red,generallypoorlyin-
durate rock composedof volcanic ash.

Sample Preparation
Therockcoresusedfor staticand rapid testingwereNX (2•-in.-diam)
in size.Thecoreswerecutto havea length-to-diameter(L/D) ratioof 2
usinga diamondsaw. After cutting to propersize,the coreendswere
surface-ground
andlappedto obtainplaneendsurfaces.Theendsurfaces
were within 0.001 in. in planehess,were parallel to each other within
0.006 in., and the ends were perpendicular to the sides xvithin 0.5ø.
The samplesfor air-gun tests (Fig. 5) xverepreparedfrom NX cores
sawedinto disksand groundplane on a milling machineto a tolerance
of 0.01mmacross the 50-ramdiam. Thesediskswerethenhand-lapped to
a smoothfinish,and the thicknessmeasured.The aluminummounting
platewasapproximatelyL--in.thick,fiat, and parallelto within 0.001in.,
and had a hand-lappedand polishedsurface.
 EFFECT OF RATE OF LOADING ON STRENGTH 9

SOLDER
TERMINALS
•1• • GRANITE

PLEXIGLAS•
ALIGNMENT
JIG
wIRES

Fig. 5--Typical test sample for impact st.dies.

A 0.0015-in.-diam;vire xvithan effectivelengthof 0.25 in. (length per-

pendieular to the field) was placed in the center of each of four disks.
The wire was held in this position with a light application of cement
(Eastutah910). The diskswerethen cementedtogetherwith epoxy,and
a light lead was applied until tile epoxy cured. The stack of disks was
then cementedwith epoxyto the alu•ninummountingplate with a trigger
wire insertedat the aluminum-rockinterface. The Plexiglasjig shown
in Fig. 5 is for alignment purposesonly and doesnot separatethe reek
disks. The assembledspecimenwas measuredto ascertainthat the pla-
narity and the glue-layerthicknesswere within the prescribedlimits (i.e.,
0.0005 in. and not greater than tile diameter of the wire, respectively).
Method of Measurement

The reek corestestedstaticallyin unconfinedand confinedcompression

had six I i•-in.-longelectrical-resistance
strain gagesbondedto the core;
three gageswereplacedaxially 120ø apart. and three gageswere placed
circumferentially120ø apart. All gageswere locatedat the midpointof
the core and had a resistanceof 120.4_+0.2ohmsand a gage factor of
2.01_+19•,.The stress-straindata were recordedon a strip-chartoseillo-
graph.
The measurementof compressionalwave velocity within the triaxial
chamberis accomplishedusing a through-samplemethod. A pair of
barium titanate transducerswith a lower resonantfrequencyof 1 mega-
cycle per seeis coupledto either end of the samplewith a film of silicone
grease.A pressureimpulseis imparted to the samplefrom the expansion
of a transducercausedby a step in voltagebeing applied to the trans-
dueer. The incidenceof the transmittedpressureimpulseon the receiving
transducergeneratesa voltagesignal indicating this arrival. Thesesig-
10 BASIC AND APPLIED ROCK MECHANICS

Basal•

Granite

Fig. 6--Traces of compressional wave velocity. The top

trace is the time-mark generator trace with one large
division equal to 10 •sec. The bottom trace is the com-
pressional wave velocity signal.

nals are displayed on an oscilloscopeand comparedwith a signal from a

crystal-controlled time-mark generator for determining the transit time
through the sample. From this measurement of time and the known
transmissive path length, the compressional wave velocity can be com-
puted. Fig. 6 shows a typical compressionalwave velocity trace. The
input signal is on the left and the first arrival is towards the right.
For the drop-tower apparatus, a 200,000-lb-capacity SR-4-type load
cell, two single-sweepdual-beam oscilloscopes,and two Polaroid cameras
were used to record the stress-straintraces. A «-in. piece of celotex was
placed on top of the rock sample to mitigate the pulse. The tuff was tested
using the drop-tower apparatus. Fig. 7 is a sketch showing the typical
traces of the stress-strain pulses recorded during the drop-tower tests.
 EFF•C? OF RA?E OF LOAmNO ON S?R•NO?H 11

Trace of vertical
strain pulse

Strain = 1112 •im./in./

cm

Trace of load pulse

Load = 56,380 lb/c•n

Time = msec/cm

Trace ,of horizontal

strain pulse

Strain = 1112 •in./in./

cm

Trace of load pulse

Load = 56,380 lb/cm

Time = I msec/cm

Fig. 7--Traces of stress and strain.

During testing with the 200-kip loader, the load is measuredaboveand

below the test specimenby meansof strain-gage-type load cells. Accelera-
tions are measured above and below the specimen, using commercial
accelerometers. The outputs of all the sensing devicesare recorded simul-
taneously on a multichannel, magnetic tape recorder and later played
back using a light-beam galvanometeroscillograph.
To observethe particle and shockvelocitiesat various points in a rock
specimen, induction wires were mounted at four locations in the speci-
men, and the specimenwas placed in a magnetic field so that the effective
length of the wires was perpendicular to the magnetic lines of force
(Fig. 4).
The impact creates a shock wave which interacts with the induction
12 B•s•c •N•) Arr•m•) Rock: MECHANICS

wire as it propagates through the specimen, and the wire follows the
particle motion. Movement of the wire in the magnetic field generatesan
electromotive force (emf) by cutting lines of magnetic force. This emf
is proportional to the particle velocity in the specimen as shown by the
following equation:
E (6)
xvhereE is emf, volts; fi is •nagnetie-fieldstrength, webersper sq m; 1 is
effective length of wire, m; and U•, is particle velocity, m per see.
The magnetic-field strength fi was 775 x 10-* webers per sq m and the
wire length 1 was 6.35 x 10-:• m (0.25 in.) for all the tests. The shock-
propagation velocity can be determined from the time difference between
the particle velocity signalsand the known wire spacing.
The impedance mismateh between the wire and rock should not cause
a serious perturbation in the shock front since its transit time is small
compared to the rise of the shock front. Reflections within the wire,
which should have a frequency of about 40 MHz, should bring the wire
to equilibrium with the rock very rapidly. Thus, the assumption that the
induction wire follows the particle motion of the rock seemsto be justified.
Each induction wire was connectedto the input of a Hewlett-Packard
450A amplifier. The output signal from the amplifier was fed to the input
of a Type 545 Tektronix oscilloscopewith a Type CA preamplifier
through 100 ft of RG-SU cable. The cable was terminated in its character-
istic impedance (51 ohms) at the input to the oscilloscope.Another in-
duetion wire was positioned between the mounting plate and the rock
specimento trigger the oscilloscopes and the fidueial generator. The
fidueial signal was fed to the B input of the oscilloscopesand recorded on
each signal record. This provided a time reference which could be used to
compare all the signals.
Data Reduction

The stress-strain path followed by the material in each shot was de-
ducedfrom the measuredwave propagationand particle velocities. Each
continuousrecord of particle velocity was approximatedby straight-line
segments,which is equivalentto approximatingthe wave shapeby a series
of successiveincrements of pressure.
From the four induction wires placed in the material, signals similar
to thoseshownin Fig. 8 will be received.
A particle velocity U• of amplitude E• occurs at the time t•. The
magnitude in meters per second of U• at E1 is determined from the
equation

U•:E•(2.03xlO •) (7)
 EFFECT OF RATE OF LOADING ON STRENGTH 13

e•(2 --SIGNAL
FROM
WIRE
1
(i)
1) •SIGNAi FROMWIRE2

• E1(2) • --SIGNAL
FROM
WIRE3

.• • i (3) SIGNAL
]_
iv
t,(3) _i
• I
••2(4) FROM
WIRE
4
]• tl(4) •

which is obtained by evaluating Eq. 6 for fi-775 x 10-* webers per sq m

and I=6.35 x 10-am. The propagation velocity Us of each increment of
particle velocity commonto each record is then determined from the time
difference measured between the records and the distance d between the
signal wires, i.e.
d
Us•(•-•)--
tl(2)--t1(1) (8)
The number of stress-strainpoints which can be computedfrom a single
shot is dependent on the number of straight-line segments selected to
approximate the continuousrecord of particle velocity.
Conservationof massand momentum,pressurebehind the shockwave,
and the strain corresponding to P, have been given in Eqs. 2-5. These
equations are based on the assumption that an equilibrium state of stress
is reached behind the shock wave. If the shock structure is stable with
pressure (the propagation velocity does not change significantly), the
peak steady-state value of particle velocity determined by Eq. 8 can be
used in Eqs. 4 and 5 to compute a point on the stress-strain curve for this
material. The points determined from these equations correspondto the
Hugoniot or the locusof shocked-endstatesreached.
For various combinations of material properties and shock-loading
pressures, two or more stress waves may develop in the material as the
input stress propagates away from the impact surface. Materials with
an elastoplastiemode of failure may give rise to a relaxing shockfront,
i.e., the shocktransition region is lengthenedxvith time.
The conservationequations may be successivelyapplied to determine
the final state behind the stresswave to yield the following equations for
n appro.ximations to a stresswave:

Pn----P•-I
+ po(U•n-U•,
n-•)
1__(U•n--
en_• U•, (9)
14 BASIC AND APPLIED ROCK MECHANICS

1--•--_ U•-•--•Y•,•'-x
' (10)
These equations are taken from Ref. 11, which also shows a com-
parison of a various number of n-line approximations. This indicates
that the calculation processis convergentto the actual ttugoniot, with
the accuracy depending on the number of straight-line segmentsused to
approximate the curved record.

PRESENTATION AND DISCUSSION OF RESULTS

Static

The averages of three tensile splitting tests for the three rocks are
given below. These results are used in conjunction with the triaxial shear
results on the plots of Mohr envelopes(Fig. 16).

Average Tensile Range in

Splitting Strength,
Rock Type Strength, Psi Psi
Granite 1700 380
Basalt 1900 300
Tuff 170 120

Twelve basalt sampleswere tested under triaxial conditions at confin-

ing pressures (•a) of 250, 1000, and 5000 psi. Four specimensat each
•a were tested at loading rates of 1, 50, 500, and 2250 psi per see. The
granite and tuff were tested at a loading rate of 50 psi per see at •a's up
to 4000 psi. Compressionalwave velocities (V•) were recorded for all
specimensat the 50 psi per see rate only. Table 1 shows the increase in
V•, from zero to maximum deviator stress,at each •3. The V• results for
the granite and the basalt indicate that very little consolidation and V•
change took place under relatively low confining pressures to 4000 and
5000 psi. Porosity determinations before triaxial testing confirmed this
effect; porosity of the granite and the basalt averaged 0.30 and 4.5%,
respectively.
The V• for the tuff specimensincreased 27% under a •a of 1500 psi.
The average porosity of 12 tuff specimenswas 20%. The increase in
¾• under combined stressesis primarily due to the consolidation of the
sample resulting in greater density and, hence, larger velocities. The
 EFFECT OF RATE OF LOADING ON STRENGTH 15

Table 1. Relation of Velocity to Confining Pressure

Factor by Which Velocities

Increased at Confining Pressures,Psi
Rock 250 500 1000 1500 4000 5000

Granite 1.09 1.10 1.04

Basalt 1.03 1.06 1.12
Tuff 1.23 1.21 1.27

axial deformationat failure was considered

in calculating¾•; how-
ever, this did not significantlyalter the ¾p results. Figs. 9-11 show
typical plotsof ¾prs. deviatorstressfor the three rocks.
Generally, the Up for the three rocks increasedsharply within one-
third of the maximumdeviator stressand then remainedrelatively con-
stant to failure.
Figs. 12-14 showthe relationship of deviator stressand axial strain.
Generally, maximum deviator stress and axial strain at failure increased
with increaseda3 and with rate of loadingas in the caseof the basalt.

Fig. 9--Compressional wave velocity vs. deviator stressfor basalt.

16 BASIC AND APPLIED ROCK MECHANICS

œ;-4

Fig. 10--Compressional wave velocity rs. deviator stress for granite.

, •ec•nen No, Conf in tng Lo•d •ng

Pressure t• •te

T-J4 !50_•
psi 50psi/sec

0 2 4
T-20
T-21
00•
ps•50
50
psi/sec
6
psi 50 •si/sec

Dev_ator Stress, 1000 psi

Fig. 11. Compressional wave velocity rs. deviator stress for tuff.
 ]•FFECT OF •:{ATE OF LOADING ON STRENGTH 17

03= 21Opsi O•= 1000

pi 73= 5000
ps 50

4O
ß /•''•-500

225• 50
00

•
o 20

lO
•OTE: Loadi •g rates
•re l, 50, ] )0, and
2250 psi/sec

o
0 4000 0 4000 8000 0 4000 8000 12,000
Axial Strain, •in./in.

Fig. 12--Increasein deviatorstressand axial strain with increasein loadingrate at

½sfor basalt.

2009 ,'•:)"• 6000

Axia • •train, • n, /in.

Fig. 13--Deviator stress rs. strain curves for granite.

18 BASIC AND APPLIED ROCK MECHANICS

•3 = 500 •si

,,o
3=1500
si
J/ o=lOOO

0 2000 4000 6000

Axial Strain, •in./in.

The tuff was the only rock tested that showed a decrease in deviator
stresswith increasedconfining pressures. The rock was tested at natural
moisture contentof approximately 21% and in the undrained state. The
pore pressurebuildup due to confining pressureand axial loading prob-
ably causedthe pore pressureto break down someof the rock structure,
thereby causinglower strengthsat increasedconfiningpressure. Mogi
found a similar condition. In additionaltriaxial testing,the effectof pore
pressureshouldbe accountedfor in terms of effectivestresses(a• =a-
where rr• is effectivestresses.a is total normal stress,and • is pore pres-
sure. Young's modulus also decreasedwith increased rra's.
Basalt was testedin triaxial compression at loadingrates of 1, 50, 500,
and 2250 psi per see. Fig. 12 showsthat at •a's of 250 and 1000 psi, the
maximumdeviatorstress(•) increaseswith increasedloadingrate, with
the exceptionof the specimensloaded at 50 psi per see. At 5000 psi
the a increasesthrough the full range of loading rates used. Total axial
strain at failure increasesin all caseswith an increasein loading rate at
eachof the rr:•'susedexceptat 250 psi rr:•and at 50 psi per seeloadingrate.
Plotting • at maximum loading rate (•'n,•x)over a at minimum loading
rate (•'n,•n)for eachrr.•usedfor basaltresultsin a linear relation, with a
positiveslopeas shownin Fig. 15. This is to say that as rraincreases,the
effectof •' on strength becomesmore significant.
Generally,the straight-linerelationshipdescribedby Mohr's criterion,
 EFFECT OF RATE OF I•OADING ON STRENGTH 19

Fig. 15. Effect of • strength at various confining pressures.

T-C-Fm• tan 4, where T is the shearingstress,C is the cohesion,a, is

normal stress,and 4 is the angle of internal friction, fits mostof the lower
stresscirclesfor the granite and basalt. However, a curvilinear analysis
may fit all circles better. There was no envelopedrawn for the results of
the tuff due to the decreasein deviator stresswith increasing aa.
Fig. 16 showsthat the 4 for the basalt and granite are very nearly the
same at a loading rate of 50 psi per sec. The 4 for basalt appears to
increase bet•veenthe loading rate of I and 2250 psi per sec, which could

J !o,.,;J 2t,,l.;,JJ ; ,Dr,f, 2•, ,o, L., J Z

N,r a• btress, •s

Fig. 16--Mohr envelopesat various loading rates for basalt and granite.
20 BASIC AND APPLIED ROCK MECHANICS

Table 2. Effect of Loading Rate on ½ and C for Basalt

Loading Rate, ½ at Tangent of

Psi per See 1000 Psi a3 Circle C, Psi
I 50 ø 3250
50 53 ø 3800
500 50 ø 3900
2250 55 ø 3900

indicatethat shearstrengthas well as strengthand deformationincreases

under combinedstressesand under increasingloading rates. This merits
further investigation.Table2 givesthe effectof loadingrate on •kand C
under combined stresses for basalt.
All but twoof thebasaltspecimens testedin triaxial compression
failed
in shear;shearanglesweremeasured from the horizontal.The observed
anglesof failure, Table3, further substantiatethe suggestion
mentioned
previously thatloadingratesmayhavemoreeffectontheshearproperties
of basaltthan do confiningpressuresup to 5000 psi.
Static and Rapid Compression
Three uneonfined eompressivestrength tests were conductedfor each
of the three rocks at a loading rate of 50 psi per see. The basalt, assumed
to be more homogeneous than the granite or tuff, was further tested at
higher rates of loadingup to 1.60X 107psi per sec. Sevengranite and two
tuff specimenswere tested at approximately 1.00 x 107 and 4.41 x 10* psi
per see, respectively. Young's modulus of elasticity (E) and Poisson's
ratio (•) were calculated for each specimen. The calculated E is a
tangent value taken at one-half the ulthnate eompressivestrength, i.e.,
E--Aa/Ae•, where Art is the change in axial stress and Ae• is the change
in axial strain. Poisson's ratio is also calculated at one-half the ultimate
strength, i.e., t•--ed/ea,where • is the axial strain and • is the diametral
strain. The average uneonfined eompressivestrengths, modulus of elas-
ticity, and Poisson'sratio for the static and rapid loading rates used are
given in Table 4.

Table 3. Effect of Loading Rates on ObservedFailure Anglesfor Basalt

Loading
Rates, Confining
Pressures,
Psi
Psi per See 250 1000 5000
I • 61 61
50 60 63 63
5OO 71 75 7O
2250 -- 74 72
EFFECT OF RATE OF I•OADING ON STRENGTI-I 21

XXXXX X XXX
22 BAsic AND APPLIED ROCK MECHANICS

It can be seenfrom Table 4 that the strength and modulus of the basalt
are not greatly affected by small increasesin loading rates. The ratios of
strength and modulus of specimenstested at 500 and 1 psi per sec are
1.04 and 1.10, respectively. However, the ratios of rapid to static
(F,./Fs) strength and modulus (E,./Es) are somewhat different; the
strength ratio is 1.48 and the modulus 1.17. In calculating E•./Es, the
toodull obtained from specimens tested at 2.06 x 10• psi per sec were
omitted for the following reasons. These specimenswere tested at the end

25

lJ

Axial St•-aœn, perccitt

_Fig.17--Static and rapid stress-straincurvesfor basalt.

 7EFFECT OF ]•ATE OF LOADING ON STRENGTH 23

of the investigation in order to fill in 32 decadesof loading rates. At this

time only foil-backed strain gages were available. The foil gages were
bonded to the basalt with epoxy, while the paper-backed gages used on
all other specimens were bonded with Eastman 910 adhesive; the gage
length of the foil gages was 50% less than the paper gages. It appears
that the different gages,different adhesives,and gage lengths affected the
strain in relationship to stress for these three specimens,and, hence, the
modulusof elasticity appears lower than expected.
The results in Table 4 show an F,./F., for granite and tuff of 1.27 and
1.70, respectively; E,./E., is 1.28 for the granite and 1.02 for tuff. The very
slight increase in modulus with increased rates of loading for the tuff
suggeststhat this parameter is not as sensitive to increased loading in
soft rock as it is in brittle rock. Watsrein 6 found the opposite to be the
casebetween weak and strong concrete.
Figs. 17-19 give average static and rapid stress-strain curves for the
basalt, granite, and tuff. The stressand strain data points are averages
of from three to sevenvalueseach. In Fig. 20 the rates of loading are
plotted to a logarithmic scaleagainst the ultimate strength.
As shownin Fig. 20, the strength increaseswith an increasein the
logarithmic rate of loading. The general trend of the resultsin this paper
compare quite well with other investigators? The results of Kumar
et al.• are only different from the resultsin this chapter in that their rates

• L2

Static
50
ps
•/$ec
Rapid lO7 p;i $ec

0.i 0.2 o.3

Axial Strain, percent

Fig. 18--Static and rapid stress-strain c•rves for granite.

24 BASIC AND APPLIED ROCK MECHANICS

3 2 '• q. • q,
Axial 5tra{ •, n.:ccc•nt

Fig. Z9--Static and rapid stress-strain curves for t•ff.

 EFFECT OF RATE OF LOADING ON •TRENGTH 25

[] SLyess I'o•nt

i 3 ) 3 o:) 7H 3 1 r) 2J 30

I•'ig.20--Elferr of loadingrate on •dtimatecompressire

strength.

of loadingare higher. Fig. 20 indicatesthat the rocksthey usedwere

inherentlystrongerby about6000psi. Thiscouldaccount for thehigher
ultimatestrength.Also,this difference couldbe due to test equipment
and the size (L/D) of specimensused.
Additionaltestingof theweakermaterials,tuff andconcrete,in uncon-
fined compression at loadingrates greaterthan 107 psi per secwould
clarify a suggestion
put forth by AtchleyandFurr • that, "The trendof
the curvesindicatesthat as the rate of stressincreases, the dynamic
strength approaches a constant value." This statement refers to concrete
with nominalstrengthsup to 5000psi. The resultspresentedfor the tuff
appearto be in accordwith this statement.However,under shockload-
ing, the tuff wasstressedto 1.6x 104psi withoutfailingat an average
loadingrate of 1.09x 10"psipersec.If thetestgeometry is neglected,
it
26 BASIC AND APPLIED ROCK MECHANICS

can be stated that the weaker materials would increase in strength with
loading rates greater than 10• psi per seeinstead of remaining constant as
Atehley and Furr suggest. The maximum stresslevel for tuff is plotted
in Fig. 20. A stress level for basalt of 4.7 kb is also plotted in Fig. 20,
which appears to agree with the basalt curve presented by Kumar et al.1
if it were projected. These two stresspoints are plotted to indicate that
compressire strength does increase with increased loading regardless of
test geometry.
Total axial strain (Fig. 21) observed at failure in the rapid tests was
greater than the correspondingstrain in the static tests. Diametral strain
decreasedslightly with increasedrates of loading for the basalt (Fig. 22).
As mentionedpreviously, the Young's modulusof elasticity for the basalt
tested at 2.06 x l0 s psi per seewas suspectedto be low. For this reason,
these values are not included in the plot of Young's modulus rs. loading
rate (Fig. 23). The method of least squareswas usedto obtain a straight-
line equation of best fit for the data in Figs. 21-23.

Shoek Loading
The shock-loading work was accomplishedas part of a different re-
search program from that in which the static and rapid work was done.

(o.52!
>)(x)

..

1 l) 10 • 10 • 10 10

• , psi/see

Fig. 21--• rs. total axial strain at failure for basalt.

 EFFECT OF RATE OF LOADING ON 8TRENGTI-I 27

• y = •.67,0• (-0.16'•2)
(^)

L adinf% r. te. o$•sec (Log Sc• le)

Fig. 22--•' vs. total diametral strain at failure for basalt.

J.

--

--

(Log S'-a:e)

Fig. 23--Loading rate vs. modulusol• elasticity for basalt.

28 BASIC AND APPLIED ROCK •¾[ECHANICS

Therefore, a few of the figures in this portion of this chapter include data
for quartzite which will not be discussed.
Fig. 24a is a plot of pressure rs. particle velocity computed from a
single impact test on granite. Three curves were computed from the four
records obtained for the particle velocity wires, which were separated by
¬ in. The scatter in the data is probably causedby local inhomogeneities
within the specimen. Fig. 24b shows a comparison of stress-strain data
for granite obtained from two separate tests. The maximum stress
reached in Shot I was 3.2 kb and in Shot 2, 13.8. As can be seenfrom the
figure, the computed stress-straincurvesfroin thesetwo tests are identical
within experimental error. The modulus,which can be computed from
thesedata points, is a constrainedmodulus.
Fig. 25a shows a plot of pressure rs. particle velocity, or the shock
impedance, for all materials tested. Each curve was computed from a
single experiment. Fig. 25b shows the stress-strain relation for all the
materials tested. The maximum stressonly represents the stress reached
in the respectivetests and doesnot indicate elastic failure. Under condi-
tions of one-dimensionalstrain thesematerials, with the exceptionof tuff,
would not be expectedto fail in compressionat these pressure levels. For

DD

12

SINGLE SHOT
LEGEND
DATA COMPUTED fROM
LEGE
0 SHOT 2

D
SHO
1
O SIGNAL WIRES 1 AND 2
A SIGNAL WIRES 1 AND 3
i"l SIGNAL WIRES I AND 4

o
2 4 6 @ lO 0 0 4 0 8 1.2 1.6 2.0

PARTICLE VELOCITY, 10-2 MM/•ZSEC STRAIN. PCT

: SHOCKPRESSUREVERSUSPARTICLE VELOCITY b STRESS VERSUS STRAIN

l•ig. 24--Test results for granite.

 EFFECT OF RATE OF LOADING ON STRENGTH 29

LEGEND
•
LEGEND

GRANITE p GRANITE
QUARTZITE
BASALT BASALT

QUARTZITE
/
PLEXIGLAS PLEXIGLAS
TUFF TUFF

/
/

______._..Q.____-------O------'---
2 3 4 5 6 7 0 02 04 06 08 10 12 •4
PARTICLE V•LOC(TY 10-• MM/•SEC STRAIN PCT
SHOCKP•ESSUREVERSUS•ARTICLEVELOCITY b STRESS VERSUS STRAIN

_Fig.25--Test res•dts for all materials tested.

comparativepurposesthe static,rapid, and shock-loading stress-strain

dataareplottedin Fig. 26 to indicatethe increase
in stress,strain,and
modulusfor the full rangeof loadingratespresented in this chapter.
The shockresponse during loadingobserved for basaltand granite
is essentially
linear. Thestructure
of theshock
waveis notsignificantly
altered over a propagationdistanceof i in. exceptfor randomscatter
betweensetsof particle velocity transducers.This scatter is believedto
becaused by localinhomogeneities in therocksamples.
The tuff specimen did not exhibit a stable shock structure. The width
of the shocktransitionincreases
continuouslywith distance propagated.
Thiscouldbe a resultof localyielding,sheareffects,or viscosityeffects;
however, the data are too limited to determine the cause. Since there is a
muchgreaterrelaxationof the shockfront in tuff, the approximation
methodof analyzing the data is lessaccuratethan that for the other mate-
rials. Thismethodof analyzingthe data can,however,be adjustedto
make the error comparable with that associatedwith the data from the
other materials.

Plexiglaswasusedas a reference
materialto verify the testmethod.
30 BASIC AND APPLIED I•0CK MECHANICS

! ¸

(2) .,- •

, .2 .' 0.6 q.• '. ) •.2 1.4

A•.i•I St•-a•n, 'ercent

ß . ) •o o

Fig. 26--Static, zapid, and shock-loadingstzess-stza•ncurves.

As shown in Fig. 27a, these data correlate closely with the data points
reported by Wagner et al.; •3 however, this comparisonrequires extrap-
olation of their data from approximately 6 kb to the pressure (approxi-
mately 2 kb) achievedin the present tests. The Plexiglas data also seem
to correlate well with low-pressure stress-strain data from compressive
 EFFECT OF RATE OF LOADING ON STRENGTH 31

?
7

IN DIAMETER BY 0 500 INCH LONG

SYM•DL STRAIN
RATE,IN/IN/SEC
0.005 --

0 0070
o o •o
• • •o
ß 1210 •0
•RoJg'r- GENERAL DATA -
WES I•,TA

WES
DATA,
PLEXIGLAS
SPECIMEN
2)25
INCHES
IN DIAMETER BY ) 25 IHCHES
m

0 2 4 8 8 I0 o ;• 4 8 8 io 12
STRAIN, PCT STRAIN ,• PCT
o. PLEXIGLAS b. LUCITE

,ooc
14

12 ooc
I

• io ioc

io

4
I.C /•// n LEGEND
- /
/'/ I•
o' IITRI DATA

LEGEND
o.
SRI DATA--
WES DATA
•], WES
DATA
(2SHOTS)
o.o -•r
'• • I IllIll I I I Illll I I illIll I I •lllll I I [ I IIIII
0o 0.5 1.0 I .S 2.0 0.001 0.01 0.1 I I0 I00
STRAIN • PCT STRAIN ,• PCT
c. QUARTZITE d. GRANITE

Fig. 27--Comparison with results of previous studies.

32 BASIC AND APPLIED ROCK MECI-IANICS

strain-ratetestson lucite presentedby Maiden and Green14 (Fig. 27b).

The shock responseof these two materials is approximately the same;
therefore, a comparisonof the data would seemjustified althoughthe data
from Maiden were determined under conditions of uniaxial stress and
our data correspondto a uniaxial strain geometry.
Very little low-pressuredata are available on granite; however, some
stress-straindata are presented by Dennen • on a log-log plot for pres-
sures both above and below our data points. As shown in Fig. 27d, our
data compare favorably with those from Dennen's work.
Data that would be directly comparable were not available on basalt
and tuff below 50 kb. A rough comparisonof granite and tuff data with
data from unconfined static and dynamic tests • shows the constrained
modulus determined is these tests to be somewhat higher in each case.

CONCLUSIONS

On the basis of results obtained from the static, rapid, and shock-
loading testsof the basalt, granite, and tuff, the following conclusionsare
drawn:

1) Both maximum deviator stress and axial strain at failure increase

with increasingloading rates up to 2250 psi per secat confining pressures
of 250, 1000, and 5000 psi. Shear stressalso increasesat loading rates of
1, 50, 500, and 2250 psi per secup to 5000-psi confining pressure. For the
loading rates and confiningpressuresinvestigated, confiningpressurehas
considerably more influence on deviator stress than does loading rate.
Loading rates up to 2250 psi per sec affect the observedangle of shear
more than do confining pressuresup to 5000 psi.
2) The influence of loading rates on the deviator strength of basalt
becomesmore significant at higher confiningpressures.
3) The straight-line relationship described by Mohr's criterion de-
scribes the triaxial failure of basalt and granite only at the lower con-
fining pressures. Von Mises-Hencky's yield criterion does not describe
the failure of basalt and granite as well as does Mohr's criterion.
4) The compressionalwave velocity of the three rocks is affected by
increasesin both the applied axial stressand confiningpressure. Veloci-
ties recorded in the direction of maximum applied stressincrease sharply
within about one-third of the maximum deviator stressand then generally
level off until failure.
5) The unconfined compressire strength of the three rocks studied
increaseswith an increase in the logarithmic rates of loading. The ratio
of rapid to static strength F•./F• for the basalt was 1.48, for granite 1.27,
 ]•FFECT OF RATE OF LOADING ON STRENGTH 33

and for tuff 1.70. The F,./F8 of tuff is closeto the Fr/F8 of weak concrete
reported by several other investigators.½.7
6) For basalt the axial strain at failure increaseswith an increasein
the logarithmic rate of loading, while diametral strain at failure
decreases.
7) Young's modulus of elasticity increaseswith an increase in the
logarithmicrate of loadingup to about2 x •10TMpsi per sec. This is based
on a comparisonof the constrainedmoduluscalculatedfrom the shock-
loading data and unconstrainedmodulusof all three rocks.
8) Uniaxial stress-strain or Hugoniot equation of state data can be
determinedby the methoddescribedin this chapter for solid materials in
the pressurerange from 0 to 40 kb. The velocity of propagationcan be
the application of the conservationof massand momentumequations.
The data obtained from these equations correspondto the Itugoniot or
the locus of shocked-endstates reached. This method of obtaining con-
tinuous records of propagation velocity and particle velocity does not
dependon a prior knowledgeof the unloadingcharacteristicsof a mate-
rial as doesthe free-surfacetechnique. The shockresponseof the basalt
and granite was found to be linear. The tuff specimen exhibited an
unstable shock structure.

ACKNOWLEDGMENTS

The authorsare particularly grateful to JamesB. Eskridge, JamesR.

Itossley, Leo F. Ingram, Kenneth L. Saucier, and Billy R. Sullivan of
the Waterways Experiment Station for their suggestionsand hard work
during this research,and to Bryant Mather for his helpful criticismanal
review of this manuscript.
The work was done under the auspicesof the Defense Atomic Support
Agency.

REFERENCES

1. Kumar, A., Hausser, F. E., and Dorr, J. E., The Effect of Stress Rate and Tem-
perature on the Strength of Roelos,publisher unknown.
2. Stowe, lt. L., "Dynamic Properties of Granodiorite, Project PILE DRIYER, •
(Abstract only), Engineering Mechanics Division Specialty Conference Proceed-
ings, ASCE, Oct. 12-14, 1966.
3. Wucrker, R. G., •Influence of Stress ]late and Other Factors on the Strength and
Elastic Properties of ]locks," Quarterly of the ColoradoSchoolof Mines• vol. 54,
No. 3, July 1959.
34 BASIC AND APPLIED ]:•OCK MECHANICS

4. Phillips, D. W., "Tectonics of Mining," Colliery Engineering, Part II, ¾o1. 25,
No. 294, 1948.
5. Atchley, B. L., and Furr, H. L., "Strength and Energy Absorption Capabilities
of Plain Concreteunder Dynamic and Static Loading," ACI Proceedings,Vol.
64, No. 11, Nov. 1967, pp. 745-776.
6. Watsrein, D., "Effect of Straining I•ate on the Compressive Strength and Elastic
Properties of Concrete,' ' ACI Proceedi•igs,Vol. 49, No. 8, Apr. 1953, pp. 729-744.
7. Munson, D. E., and Barker, 14. M., "I)ynamically Determined Pressure-Volume
Relationships for Aluminum, Copper, and Lead," Jot•rnal of Applied Physics,
Vol. 37, No. 4, Mar. 15, 1966, pp. 1652-1660.
8. Karnes, C. H., "Stress-Strain Characteristics of Materials at High Strain Rates;
Part IV: Experimental and Theoretical Analysis of Plastic Impacts on Short
Cylinders," June 21, 1960, University of Texas, Austin.
9. U.S. Army Engineer Waterways Experiment Station, CE, Handbook for Con-
crete and Ceme•t, August 1949, Vicksburg, Miss. (with quarterly supplements).
10. Frasier, J. T., and Karpov, B.C., "The Transient Response of Wax Targets
Subjected to Hypervelocity Impacts," paper presented at Society for Experi-
mental Stress Analysis, Spring Meeting, Denver, Colo., May 5-7, 1965.
11. Dennen, R. S., '•Synthesis of Rock Hugoniots," DASA 1652, March 1965, IIT
Research Institute, Chicago, Ilk
12. Mogi, K., "Deformation and Fracture of Rocks under ConfiningPressure, (1)
Compression Tests on Dry Rock Sample," B•dletin of the Earthq•take Research
Instit2tte, Vol. 42, Pt. 3, September 1964.
13. Wagner, M. H., Waldorf, W. F., .lr., and Louie, N. A., "Determination of
Itugoniot Equations-of-State for Polymers and Reentry Vehicle Materials and
Investigations of Fracture Phenomena," AFSWC TDR-62-66, Vol. 1, August
1962, Aerojet-General Corp., Downey, Calif.
14. Maiden, C. J., and Green, S. J., "Compressire Strain-Rate Tests on Six Selected
Materials at Strain Rates from 10-:" to 10• In./In./Sec," Jo•trnal of Applied
Mechanics, Vol. 33, No. 3, Sep. 1966, pp. 496-504.
15. Dcnnen, R. S., "Hugoniot Determination for Granite and Other Geological
Samples," IIT-A6040-1, September 1963, lit Research Institute, Chicago, Ill.
16. Stowe, R. L., "Static and Dynamic Strength and Stress-Strain Properties of
Rocks of Four Types," U.S. Army Engineer Waterways Experiment Station,
CE, Vicksburg, Miss., in preparation.
17. Duvall, G. E., "Propagation of Plane Shock Waves in a Stress-Relaxing
Medium," Stress HZavesin Anelastic Solids (A SymposD•m), H. Kolsky and W.
Prager, eds., Springer-Verlag, Berlin, 1964, pp. 20-32.
18. Het•nyi, M., Handbook of Experimental Stress Analysis, 5th Printing, John
Wiley, Ne•v York, 1963, p. 442.