Sie sind auf Seite 1von 9

Int. J. Rock Mech. Min. Sci. & Geomech. Abstr. Vol. 29, No. 2, pp.

133-141, 1992 0148-9062 92 $5.00 + 0.00


Printed in Great Britain. All rights reserved Copyright t~ 1992 Pergamon Press Ltd

Interpretation of Plate Load Test Data


W. J. BOYLEt
The International Society for Rock Mechanics (ISRM) has a suggested
method for determining in situ the deformability of rock using a plate test.
Previous studies using a method of data interpretation that is similar to that
recommended by the I S R M show that there is a potential for significant error
when interpreting plate load test data. In particular, the determined defor-
mation modulus of a rock mass may be larger than Young's modulus of the
intact rock, even where the rock mass is jointed and should intuitively have a
deformation modulus that is less than that of the intact rock. Simulations of
plate load tests in jointed and unjointed rocks were made using a finite element
code to assess some of the effects of jointing and excavation geometry on the
determination of the deformation modulus. The results of the simulations are
also used to demonstrate a modification of the I S R M suggested method of
interpreting plate load test data that should help in interpreting the sometimes
apparently counterintuitive results. The proposed modification uses least
squares to calculate Young's modulus of the rock mass and also allows
Poisson's ratio to be calculated, which is in contrast to the I S R M suggested
method that requires Poission's ratio to be known a priori.

INTRODUCTION deformation modulus E for stressed rocks around under-


ground openings.
The International Society for Rock Mechanics (ISP, M)
The suggested ISRM method for determining in situ
has a suggested method for determining in situ the
the deformation modulus of a rock mass requires a priori
deformability of rock using a plate test [1]. This test is
knowledge of Poisson's ratio, /z of the rock mass (the
commonly conducted for foundation studies for large
symbols used by the ISRM in the suggested method are
dams and has also been suggested for use at the pro-
used in this paper). It has been shown [4] that simply
posed high-level nuclear waste repository at Yucca
guessing # is not only difficult because of its apparent
Mountain, Nevada [2]. Earlier studies using a method of
range for rocks, but can also lead to an inaccurate
data interpretation that appears identical to that rec-
determination of E. Provided # is known, the suggested
ommended by the ISRM show that there is a potential
ISRM method allows one to calculate E for any
for significant error [3]. In particular, the deformation
measured displacement, or alternatively, an average
modulus of a rock mass is found to be larger than
value of E can be calculated for the rock between any
Young's modulus of the intact rock, even where the rock
two points where displacements are measured. A modifi-
mass is jointed and should intuitively have a defor-
cation of the suggested method is shown in this paper
mation modulus that is smaller than that of the intact
that utilizes least squares and allows a single value of E
rock [4]. Some of the uncertainty and error involved in
to be calculated for any set of points for which displace-
the determination of the deformation modulus arises
ments are measured. Additionally, the proposed modifi-
because of the differences between a plate load test and
cation does not require an a priori knowledge of ~, but
the conditions assumed for the interpretation of the
allows # to be determined from the same measured
measured displacements of the test. Some of the assump-
displacements used to calculate E.
tions required for the suggested ISRM method are that a
uniform load over a circular area is applied to a homo-
geneous, isotropic, semi-infinite half space, for which
PROBLEM DESCRIPTION
Poisson's ratio is known before the plate load test is
even conducted. Because rock masses are not homo- The simulation problem is that of a plate load test in
geneous and isotropic, a finite element code was used to an underground excavation. The suggested ISRM data
simulate plate load tests in underground excavations in reduction method assumes that a circular plate is used
stressed, jointed and unjointed rocks. The simulation to apply the load. To produce a circular load, the
results are used to assess some of the effects ofjointing axisymmetric feature of the finite element code was used,
and excavation geometry on the determination of the which produces an underground excavation that is a
right-circular cylinder. It is assumed that the excavation
t R E / S P E C Inc., P.O. Box 725, Rapid City, SD 57709, U.S.A. and test are taking place in the proposed nuclear waste
133
134 BOYLE: PLATE LOAD TEST INTERPRF-7~\FION

repository at Yucca Mountain, Nevada, and therefore, model has a constant stress boundary condition equal to
the material properties and in situ stresses are similar that caused by tee overburden.
to those that might be encountered at Yucca Mountain
and have been used in other calculations for Yucca Initial conJiti~,ls
Mountain [5]. The init!:d, pre-excavation stress state simulates the
effect of body forces caused b v gravitational acceleration
Geometry of rock with a unit mass 7 of 2320 kg/m 3. The initial
The axisymmetric region of analysis is shown in Fig. 1, horizontal stresses are taken to be one-half the vertical
a vertical cross-section through the excavation and stress. The displacement field associated with the initial
surrounding rock. The floor of the excavation is at stress state is taken to be zero. The boundary and initial
elevation 0, which is 307.6 m below the ground surface. conditions are shown in Fig. 2.
The excavated room has a radius of 5 m and a height of
~aterial pro/~erties
5 m. The loaded plates are located on the floor and roof
at the centre of the room and have a radius of 0.5 m, a Four materials are used in the simulations: (1) intact
size suggested by the ISRM [1]. The ratio of the radius rock; (2) jointed rock; (3) partially excavated rock; and
of the plate to the radius of the room is 1/10, which is (4) air. The intact rock is an elastic material with a
smaller than the 1/1.5 to 1/2 suggested by the ISRM [1] Young's modulus E of 30.4 GPa, a Poisson's ratio of
to minimize edge effects of the room. The top and 0.24, and a unit mass of 2320 kg/m 3.
bottom of the region of analysis are chosen to minimize Blanford and Key [6] suggest that making the joints
the influence of boundary conditions. The horizontal quasi-discrete and separated by strips of intact elastic
lines beneath the excavation in Fig. 1 represent the rock is a useful method of observing the effects of joints
locations of the joints at 1, 3 and 5 m beneath the floor. when using finite element codes. The jointed rock in the
model is restricted to 0. l-m-thick plates. The jointed
Boundary conditions rock has a rock matrix with the same properties as the
intact. The spacing of the joints is 0.1 m and therefore,
The vertical boundary surface of the model is assumed
only one joint is in each joint plate. The mechanical
to not have any radial displacements, but is free to move
properties of the joints are based on the joint empirical
vertically. The bottom of the model is fixed vertically but
model (JEM) [7], which is based on the model formu-
is allowed to move radially. The top boundary of the
lated by Barton [8]. The parameters of the JEM used in

OVERBURDENPRESSURE • 4.38 MPa


192.6m
TO G R ~ tVvi
SURFACE
I

307.6m
TO GROUND
T
~ ~ B O D HORIZONTALIN $1TU STRESS
SURFACE O"H • 0.5 (7V

110rn
Y FORCE
I (-CJ• 9.8 m/= 2 )
I
5m --,,,

3 --'kn
- 3,,
...2---- lm
3m TRACI'ION-FREE SURFACE
c___5m AFTER EXCAVATION

110nl

AXIS
OF
SYMMETRY

\\\\\\

Fig. 2. Boundary and initial conditions o f analysis region for plate load
Fig. 1. Geometry o f analysis region for plate load tests, tests.
BOYLE: PLATE LOAD TEST INTERPRETATION 135

the model are the following: the joint wall compressive


strength (JCS0) is 171 MPa; the joint roughness co-
efficient (JRC) is 9; the length of the joint used in
laboratory testing is 0.1 m; the residual friction angle of -5.00
the joint is 28.4~; a JEM parameter that relates peak
shear displacement to the length of a joint and JRC is I! ..... ,Jo,,,'rs
0.002; a JEM parameter that relates JRC, JCS0 and the -10.00
normal stress to the dilantancy of a joint is 0.5; the ,,<
reciprocal of the initial joint normal stiffness is
15.0pm/MPa; and the reciprocal of the normal half- -15.00

closing stress is 0.5/MPa,


The excavation is simulated by decreasing the modu-
lus of the elements as they are excavated. The partially- -20.00
excavated rock is identical to the intact rock, except that
Young's modulus is 15.2 GPa, one-half of the intact -25.00 I ] t . t
rock. The fully-excavated room is filled with with "air" 0.00 tOO 2.00 3.00 4.00 500
that has a Young's modulus of 30.4 kPa, a Poisson's DlSi:'t.AC:~.~rr BE,,EATH PLATE On x 10"4)
ratio of 0.24 and a unit of weight of 1 N/m 3. Fig. 3. Displacements from simulated plate load tests and the ISRM
suggested methods.

METHOD OF ANALYSIS
point of the floor beneath the centre of the plate rises
The analyses were performed using axisymmetric
2.2 mm, and without the joints this point moves upwards
conditions with the large deformation, finite-strain,
1.4 ram. For all other points in the rock mass vertically
finite-element program SPECTROM-31 [6] and the
beneath the centre of the plate down to 110 m beneath
JEM. A mesh with 528 eight-noded elements and 1697
the plate, the upward displacement is greater if the joints
nodes was used.
are present.
The initial step in the solution is to establish the in situ
In the jointed and unjointed rock simulations, the
stresses before room excavation. The end of this initial-
displacements caused by the plate load test are smaller
ization was set to time, t = 0, and all displacements are
than those caused by the excavation. As with the
set to zero. The room to be excavated consisted of 100
displacements caused by the excavation, the vertical
elements. In a number of excavation steps, a group of
displacement (in this case, downward) of the point of the
these elements would be converted successively from
floor beneath the centre of the plate during the test is
intact rock to partially-excavated rock and then to "air".
greater with the joints present, 0.55mm, than the
The excavation sequence was progressively outward
displacement without the joints, 0.49 mm. However, in
from the centre of the room.
contrast to the displacements caused by the excavation,
The plate load test is simulated by applying a linearly
it is not true that the presence of joints causes greater
increasing pressure to those elements in the roof and
vertical displacements of all the other points in the
floor of the room that are above and beneath the loading
rock mass beneath the centre of the plate. Figure 3
plates until the final pressure of 14MPa is reached.
shows the displacements of the points beneath the
A final pressure of 14 MPa has been selected because it
centre of the plate for the jointed and unjointed
is in the achievable range [9] of plate load tests.
simulations down to a depth of 20 m. Also shown are the
displacements that are calculated using the equation
suggested by the ISRM for interpreting plate load test
RESULTS OF THE SIMULATION
data, using the values of Young's modulus, Poisson's
The suggested ISRM method of interpreting the ratio, plate size and plate pressure used for the finite-
results of plate load tests to determine the deformability element simulations.
of rock in situ requires that the downward displacements Although only the vertical displacements are required
of points vertically beneath the centre of the plate be for the determination of the in situ deformability of rock,
known. SPECTROM-31 calculates the displacements at the horizontal displacements of the jointed and un-
all nodal points in the region of interest, and while it is jointed models are of interest. Figure 4 shows contours
realized that displacements 110 m beneath the plate are of the ratio between the horizontal displacements at
not measured in practice, the displacements calculated at nodel points in the jointed simulation and the horizontal
depth are still useful in discussing the suggested ISRM displacements in the unjointed simulation at the end of
method and the alternative method for interpreting plate excavation before any load is applied. Figure 5 shows
load displacements. contours of the ratio between the horizontal displace-
For simulations of jointed and unjointed rock, all the ments at nodal points produced by the maximum plate
points beneath the centre of the plate (except the one load in jointed rock and the horizontal displacements
that was fixed 1 I0 m beneath the floor) move upward produced during the plate load simulation in unjointed
following excavation. With the three joints present, the rock.
RMMS 29/2--D
136 BOYLE: PLATE LOAD TEST INTERPRETATION

Om O.Sm 5m
E
0
O
-~:~///~.~j
0 "-,%o
-1
-2
-3
-4
-5
-6
0 1.5 3.5 5.0 6.5 8.0 11.0 15.0 21.0

DISTANCE (m)
Fig. 4. Ratio of horizontal displacementsbeneath the excavation before loading for jointed and unjointed simulations.

DISCUSSION OF THE RESULTS OF THE test, some displacements, when used with the ISRM
SIMULATIONS suggested equation, will show that both jointed and
unjointed rock masses have a greater deformation
Figure 3 shows that at depths between 0 and approx. modulus than the Young's modulus of the intact rock
3 m beneath the floor, the vertical displacements beneath and the jointed rock mass will have the greatest Young's
the centre of the plate for the jointed simulation are modulus. Such apparently counterintuitive results have
greater than those for the unjointed simulation, which been reported [4], even when it is known that there is a
are greater than those calculated using the equation fractured, disturbed zone caused by blasting that should
suggested by the ISRM. Between 0 and 2.5m, the cause a lower deformation modulus. The apparently too
displacements of the unjointed simulation are closer to large values of deformation modulus for a jointed rock
those of the ISRM equation than they are to the mass have been attributed to the geometric differences
displacements of the jointed simulation. However, at a between an underground excavation and the model
depth of approx. 3 m, this relation is reversed and the (infinite half-space) used to develop the ISRM suggested
ISRM equation predicts the greatest vertical displace- equation [3] or a combination of a confining stress
ment, the jointed simulation predicts the least, and the dependence of Young's modulus and geometric differ-
unjointed simulation is closer to the jointed simulation ences [4]. It is important to recognize that the ISRM
than to the ISRM equation. Therefore in a plate load suggested method of interpreting plate load test data

ROOM F
~- PLATE

Orn O.Sm 5m
E
"" 0
uJ
,~ -1
~l -2
-1"-
I-
<c -3
-4
111 -5
I,-
1.5 3.5 5.0 6.5 8.0 11.0 15.0 21.0

DISTANCE (m)
Fig. 5. Ratio of horizontal displacements beneath the excavation at the maximum plate load for jointed and unjointed
simulations.
BOYLE: PLATE LOAD TEST INTERPRETATION 137

can give anomalously large values of the deformation Interpretation of the results using the ISRM method
modulus, because in the Site Characterization Plan for The equation suggested by the ISRM [1] to interpret
Yucca Mountain [2], it is stated that plate load tests will plate load test data is:
be used to determine a lower bound for the rock mass
deformation modulus. E = 2q(l - #") [(a" + :~)l : _ :]
The results shown in Fig. 3 can be used to help w..
understand the effect of the differences of geometry q:(l
between an excavated room and an infinite half-space. + / a ) [ - ' ( a Z + z : ) - t " " - 1], (1)
W.
The surface of the half-space is assumed to be horizontal where
before the plate load is applied, while the floor of the
excavated room is actually bowed upward because of the z = the vertical depth beneath the centre of the plate,
excavation induced deformation. The simulations of E-- the value of Young's modulus to be calculated at
this study, for the first 3 m of depth beneath the depth z,
plate, produce greater vertical displacements than those wz = the measured vertical displacement at a depth z,
predicted for the half-space. Perhaps in a restricted a = the radius of the plate,
area near the plate it is easier to push back the bowed q -- the value of the uniform pressure applied to the
floor than it is to deform a horizontal half-space. plate,
Deeper than 3 m, the displacements underneath the /~ -- the value of Poisson's ratio.
excavation are less than those for the half-space because
One drawback of using the suggested method is that
of the stiffness imparted by the rock surrounding the
Poisson's ratio must be known before E can be calcu-
excavation.
lated. Should one assume an appropriate value of/~ for
The difference in vertical displacements for the jointed
the rock mass, or instead use a laboratory value?
and unjointed simulations cannot be attributed to the
Another drawback to equation (1) is that it is a
excavation and plate geometry, which is the same in both
quadratic in terms of/~. That is, for given values of w:,
simulations, but must be caused by the presence of the
q, a and z, two different values of/~ can give the same
joints. At less than 3 m depth, the vertical displacements
value of E. For example, Fig. 6 shows the variation of
of the jointed simulation are greater than those of the
E with/~ given that z is 3 m, a is 0.5 m, q is 14 MPa, and
unjointed simulation, as is intuitive, because the joints
w: is 0.045 mm. Although the range of E for this example
are more deformable normal to the joints than the
is only about 5% of the maximum value, the range for
unjointed rock is. The crossover at about 3 m, where
is 100% of its theoretical range. This result shows that
the vertical displacements of the unjointed simulation
the ISRM suggested method for determining E is of no
become greater than those of the jointed simulation, also
use in helping to determine/~, another parameter that
cannot be attributed to differences in geometry, but must
affects the deformability of the rock mass.
also be attributed to the joints. The results of Figs 4
The ISRM suggested method has a drawback in that
and 5 show that the presence of the joints cause both the
it has the potential to determine a different value of E
horizontal displacements produced by the excavation
for every displacement measured. Figure 7 shows the
and the horizontal displacements produced by the plate
variation of Young's modulus with depth for the first
load test to be much greater near the uppermost joints,
20 m beneath the plate for both the jointed and un-
particularly at 1 and 3 m at the end of excavation, and
jointed simulations, using the ISRM suggested equation
at 1 m for the plate load test. These joints are more
and 0.24 for/~, the value used in the simulations. While
deformable in shear than the intact rock is, and the
for some plate load tests it is possible to ascribe the
shear displacements along the joints cause a stress
variation of E with depth to confining stress effects on
redistribution such that work is done in moving the rock
E, that is not possible for these simulations because E
horizontally and less energy is available for deforming
and/~ are kept constant in the simulations. The variation
the rock vertically beneath the plate.

INTERPRETATION OF SIMULATION RESULTS O. 42

The analyses of underground excavations in jointed u J"


and unjointed rocks described above give insights into
the effects of excavation geometry and joints on the .J
displacements measured in a plate load test. However, Q
the primary purpose of the simulations is to provide a 0 4O
set of vertical displacements that are respresentative of oO
those that would be measured during a plate load test.
z
In the following sections, these representative displace-
0
ments are used to show some difficulties of using the ~- 0
ISRM suggested method of interpreting such displace- P~ON'S RATIO /.J,
ments. In addition, an alternative method of interpreting Fig. 6. Variation of E as a function of/~ using the ISRM suggested
the data is presented. method.
138 BOYLE: PLATE LOAD TEST INTERPRETATION

o.oo I method for both the unjointed and jointed simulations


- - IS,CtM T H E O I ~ A L ALL P<:~NTS,
WR'H J(::l~'S
are much greater than the input value of 30.4GPa
i \ -- - - - :S~M Tl-'~q~l'~aL ALL P G ~ $ , used in the si:z--:lations. If t.':c:e a~er.~:., values are
NO ,;OINTS
-5.CG - used in further analyses, they ~vill unJerpredict the
\ displacements that will occur.
\ The averaging of the results calculated by the ISRM
\
~. -10.00~- method is also unsatisfactory because some data must be
I-- "" \ \ \ \
exclude.d in calculating the average. In the unjointed
simulation, the vertical displacements at 10t and 92 m
-15.00 depth beneath the plate are both zero. The value of E at
-,r"
I,--
(3. both these depths using the ISRM method is infinite and
therefore, any average including the value is infinite also.
-20.00 Therefore, if an average value of the deformation modu-
lus is desired, any measured displacement that is zero
cannot be used, even though it may be a valid, repeatable
-25.00 measurement. Deletion of such data is unsatisfying
0 50 100 150 200 250 because the deletion is made, not because there is some
YOUNG'S MODULUS, E GPa doubt about the measurement, but because the measure-
Fig. 7. Variation o f E with depth using the ISRMsuggested method. ment does not agree with the idealization used to develop
the suggested equation.

Interpretation of simulation results using an alternative


of E shown in Fig. 7 is due to the differences between the
method
conditions of the test and the conditions assumed in
developing the ISRM suggested equation. The variation There is an alternative method for interpreting the
is so large that the ISRM suggested method calculates measured displacements from a plate load test. The
that E at a depth of 20 m for the jointed and unjointed alternative method uses least squares and eliminates
excavations is 196 and 142 GPa, respectively. some of the drawbacks of the suggested method. The
It is possible to obtain an average value of E over an following section shows that the use of the alternative
interval by averaging the values of E calculated by the method does not require that/2 be known; although two
ISRM suggested method for each point in the interval. values of/2 may be associated ~i'2 z a value of E, only one
This has been done for 39 points in the first 20 m beneath of these values is the least square value of #; E is a
the plate for both the jointed and unjointed simulations, constant with depth over any interval of choice; the
and the results, shown in Fig. 8, are 45.2 GPa for the value of E for the jointed simulations is less than that of
jointed rock and 40.3 GPa for the unjointed rock. Such the unjointed simulations, as is intuitively expected; the
results are problematic because the jointed rock mass is values of E for the simulations are less than the input
found to have a greater modulus than the unjointed and are therefore conser-'ative with respect to possible
rock, even though the joints are less stiff. Furthermore, displacements; and measured displacements of zero need
the averages of the values as calculated by the ISRM not be deleted from a database. The derivation of the
alternative method and its application to the displace-
ments of the simulations are presented in the following
0.00 sections.
Derivation of the alternative. This method uses the
' ~ : t - - - ISRMTHEORETICAL,AU. P<:~'TS, same equation as the ISRM suggested method, except
NO ,,ON'rS
-5.00
that w. now represents the predicted displacement, yp, at
~d ISPJVl TNE(~IETICAL, AVERAGe,
Wll'H JOICr~ E • 45.,?.GP,, a depth zi beneath the centre of the plate. In the
k-
\
\ ',
"\--
~
~ ISRM~ A L , AVERAGE,
NO JOINTS,E =40.3 C-~a
simulations or in a plate load test, there are measured
~_ -lo.oo \
displacements Ym, at n points. A function • is defined as
-r-
l--
the sum of the squared differences between the predicted
<:
and measured displacements:
-15.00
...r..
I-- qb = Z (yp, - - )'m, )2 (2)
LLI i=l

Rearrangement of equation (1) to solve for w: and


-20.00
substitution of that for the predicted displacements in
equation (2) gives:
-25.00
0 50 100 150 200 250 q~ = ,=,Z{q [(a'-+ z~)"2- z~][2(1-/22)
YOUNG'S MODULUS, E GPa
Fig. 8. Variation of E with depth and the average values o f E using +(1 +l~)z,(a2+z~)-~:]-ym, (3)
the ISRM suggested method.
BOYLE: PLATE LOAD TEST INTERPRETATION 139

which, when expanded, gives:


+ 2(1 + 2/a + ,uz) ,~,~ [z4(a 2 + z2)-t]t
¢ = ~ q" a 2 + 2-T, - 2zi(a 2 + ..~)1,'21
i~l
- 2 q~ {(-- 1 + p + 2~,") ,~l
E-- ~ (ym Zi)
x [4(1 - 2g 2 + ~4)
+ 4(1 + ~ - ~z _/.~3)zi(aS + =~)-i 2 + 2(I - ~ : ) .211,21,
[.vm(a:+-i,
iml
+-~(1 +2~ + ~")(a: +-~) -~]
(6)
- 2ym,q [(a s + z~) ~'2 -- zd[2(1 -- ~")
dO q2 {
+ zi(l +/z)(a 2 + z~) -'/21 + y 2 }, (4) d'--~= ~5 16na2(-~ + #3)

Equation (4) is more conveniently expressed as:


e{
• = ~-5 4na:(1 - 2~ z + p~)
+8(-1-2#+3~2+4# 3) ~ z~
i-!

- 3 2 ( - g + ~3) ~ [zi(a 2 + z2)t/~]


i=!
+8(-~ -~z+#3+~4) ~ z~
is!
+ 4a2(1 - 2/~ - 3W') ~ [z,(a s + z~) -In]
i-I
- 80 - 2~ 2 + u~) Y'. [z,(a 2 + z~)~/~l
is|
+ 4(1 - 5. - 6u 2) ~ [z~(a' + z,~)-'/s]
+ 4a2(1 + ~ _ ~z _ ~3) ~ [zi(a ~ + z~)-~a] i=l

+ 2a~(1 + . ) ~ [z~(a'+ zD-'l


i-I
+ 2(3 + 2~ - 5u ~ - 4~ ~) ~ [z~(a ~ + z,~)- ~:~1

+ a~(1 + 2u + u') ~ [z,~(a~+ z~)-'l


i=|
+4(I +p),.,~ [z4(a2+z~)-']}
i-!
E (1 + 4~) ~ (Ym:,)
i=l
+2(1 +2u + . s ) ,-,~ [z.(a. + ~ ) - , ] }

-2~ q{ (-1 +u +2~') (y~.,z,)


- 4 ~ ~ [ym,(a~ + z]) ''s]
isl

-- [Ym, z s
i(a z+ z~)-':2]} (7)
+2(1-~:) ~ [y~,(a'+z~):'l isl

The predicted displacements most closely match the


- (1 +/1) i-I
~" [Ym"~(a2 + Z2)-1/21} measured displacements when ¢, is minimized. By equat-
ing (6) and (7) to zero and solving them, the values of
E and # that minimize ¢5 are calculated. The equations
+ ~ s
Ym i
(5) are solved by first rearranging equation (6) to find E in
i-!
terms of #. This expression for E is substituted into
All the parameters of equation (5) are known, except E equation (7) to give an equation in which the only
and ~. Differentiating equation (5) with respect to E and unknown is #. This equation cannot be solved explicitly,
# gives the following two equations: but can be solved numerically for the best fit value
d~ q" { for p. After this value of p is calculated, it is used in
d--E= E3 4na:(l -- 2//2 +/.t 4) equation (6) to calculate E.
Application of the ahernative method. As they were
+ 8(--it - /a2 +/z3 +/z4) ~ -~'2 used for the ISRM suggested method, the calculated
i--I displacements of the simulations are used as the
measured displacements for the alternative method.
- 8(1 - 2/,/2 +/t'*) ~. [z/(a s + Z~)''2 ]
i~l
Thirty-nine displacements in the first 20 m beneath the
plate for the jointed and unjointed simulations are used
+ 4a2(1 +/z --/~s _ #3) ~ [zi(a 2 + z2)-1/2 ] with the alternative method and the results, shown in
~.. i s l Fig. 9, are moduli of 21.4 GPa for the jointed simu-
lations and 27.5 GPa for the unjointed simulations. This
+ 2(3 + 2U - 5U2 - 4 # 3) ~ [z~(a2+z~) -'/~1 alternative method can be used to calculate a constant
i=t
value of E for any set or subset of displacement measure-
+a2(1 + 2 # + U s) ~ [z~(a2 + z2) -|] ments. This is an advantage over the ISRM suggested
i=l method because it is known that the value of E used in
140 BOYLE: PLATE LOAD TEST INTERPRETATION

0.00 2B
' , ) WIDI,g~I'$
I~'~, I - -- - ~ ~-¢l~s-nCAt. J~. ~ q T ~ ,
I ~ ~ NO , 1 ( ~ ' ~
-5.00 ",s~ 26 ' :T~-CUTJOINTS
, I ~. \ vaT. aoN'r=, e • ,5.z Q,'. cD
. il~ \ .... ~ r~t.o~,'r~ ,~,~ Lu"
' I \ NO ,,¢I~S. E • 40,,,1GPt
~,I ", ~ - - - LF-.'L~ SOIJ'~F~ I='rT,NO JO@ffs.

-10.00 ', I \ ~ - - - ~..~T ~I~I~ES FIT, ~ $0lqTtk

o
))
-
\
\
~Z 22
-15.00 rrH JOINTS
il
,I \ x \\ \
r-,
;[ ". 20
-20.00
I
0.5

) t .... P(XSSON'8 RATIO ~z


-25.00
0 50 100 150 200 250 Fig. 10. Variation o f E as a function of/~ using the alternative method.

YOUNG'SMODULUS, E GPa
Fig. 9. Values of E using the alternative methods and the ISRM range for the unjointed simulations is 24.0-27.5 GPa.
sugg~tedmethod. The best fit values of E associated with the input value
of 0.24 for g are 22.6 GPa for the jointed simulation
the simulations is a constant, while the ISRM method (21.4 GPa for the unrestricted best fit) and 27.0 GPa for
calculates values of E that are dependent on the depth the unjointed simulation (27.5 GPa for the unrestricted
as shown in Fig. 9. The alternative method does calcu- best fit).
late a constant value for E, and although it is not the Another advantage of the alternative method over
same value of E as input to the simulations, one should the ISRM suggested method is that, at least for the
not expect to calculate the exact same value of E used simulations described in this paper, the alternative
in the simulations, given the geometric and material method calculates a lower value of E for a jointed rock
property differences between the simulations and the mass than for an unjointed rock mass, which is what is
idealization used to predict displacements. intuitively expected, but not necessarily achieved, with
An advantage of the alternative method is that not the ISRM method.
only is no prior knowledge of/~ required to calculate E, The alternative method has another advantage over
but a least squares estimate of p is calculated also. The the ISRM method in that for the simulations described
alternative method gives a value of 0.39 for the jointed in this paper (and probably many others), the alternative
rock mass and 0.10 for the unjointed rock mass. These method calculates conservative values of E with respect
values are about equidistant from the value of 0.24 used to displacements, while the ISRM method is unconserva-
in the simulations, but they do agree with the results tive. That is, for the jointed and unjointed simulations
shown in Figs 4 and 5 that the jointed rock mass has described in this paper, the input value of E is 30.4 GPa,
greater horizontal displacements than the unjointed rock and for the first 20 m beneath the plate, the alternative
mass. method calculates 21.4 and 27.5 GPa, respectively, and
One of the drawbacks of the ISRM method is that the averages of the values calculated by the ISRM
the equation can produce a single value of E for two method are 45.2 and 40.3 GPa, respectively.
different values of p. Because the same quadratic The alternative method also does not require the
equation is the basis for the alternative method, deletion of any displacements because they are equal to
different values of p still can produce the same value of zero. For example, using 51 displacements in the 110 m
E. However, with the alternative method, only the beneath the plate, the three displacements at 92, 101 and
value of/~ associated with the best fit value of E is 110 m (the displacement at l l 0 m is zero because the
also the best fit value of/~. If one does not wish to accept node is pinned in the analyses but it is used in the least
the best fit value of/~, calculated by equation (7), but squares fit anyway) are zero in the unjointed simulation,
instead wants to assign it a value based on other and the calculated values of E and/z are still 27.5 GPa
information, equation (6) can be used to calculate E, and 0.10, the same values using the 39 displacements in
using the assigned value of/~. This is similar to fixing the the first 20 m beneath the plate. The result that E and p
intercept in a straight line fit and solving for the best fit calculated by the alternative method are the same for
slope, rather than solving for the best fit intercept and the unjointed simulation whether the data are taken
slope. This was done for the 39 displacements in the over 20 or 110 m, indicates that the large displacements
first 20 m of depth beneath the plate of the unjointed and close to plate have the most effect in determining E and
jointed simulations and the results are shown in Fig. 10. #. This is in contrast to the ISRM method, in which an
For the full theoretical range for p, the range of E averaging of the calculated values of E puts the most
for the jointed simulations is 19.9-23.0 GPa and the weight on the very small displacements farthest from the
BOYLE: PLATE LOAD TEST INTERPRETATION 141

plate. A disadvantage to the alternative method is that Report DOE/RW.0199, Once of Civilian Radioactive Waste
is requires more than one measured displacement. Management, Washington, DC (1988).
3. Misterek D. L., 51ebir E. J. and Montgomery J. 5. Field testing and
instrumentation of rock. Proc. 76th ,4am' Mtg of the Am. Soc. for
Testing and Ma~.er:als, ASTM 5TP 5~, Philadelphia, PA (1974).
CONCLUSIONS 4. Dodds D. J. and Schroeder W. L. Factors bearing on the
interpretation of in situ testing results. Proc. Rapid Excavation and
Plate load test results have been described [3, 4] that Tunneling Conremzon, Vol. 1, Chap. 31. AIME, San Francisco,
produce variable and counterintuitive values of the in (1974).
situ deformation modulus. Analyses of the ISRM 5. Lahreche D. A. Elastic analysis of the preliminary benchmark
problem (PDM No. 71-032) gith 5PECTROM-31. Prepared by
suggested method of interpretation indicates that some RE/SPEC Inc., Albuquerque, NM, TLM(ALO).0031, for Sandia
of the problems are probably caused by the method of National Laboratories, Albuquerque, NM (1988).
interpretation itself. An alternative method is derived, 6. Blanford M. L. and Key S. W. An example of continuum versus
quasi-discrete model of a jointed rock mass Proc. 2nd Int. Conf.
and its advantages over the present method are dis- on Constitutire Laws for Engineering Materials: Theory and Design,
cussed. University of Arizona, Tucson (1986).
7. Key S. W. and Labreche D. A. SPECTROM-31, A Finite Element
Computer Program for the Large Deformation, Static and Quasi-
Accepted for publication 30 September 1991. static Response of Planar and Axlsymmetric Solids, Vol. 1: User's
Manual with a Descriptwn of Mathematical Models and Numerical
Methods. Prepared by RE/SPEC Inc., Albuquerque, NM, RSI-
REFERENCES 0210 (1988).
8. Barton N. Modelling rock joint behavior from/n situ Block tests:
1. Brown E. T. (Ed.). Rock Characterization Testing & Monitoring, implications for nuclear waste repository design. ONWI-308,
ISRM Suggested Methods. Published for the Commission on Prepared by Terra Tek Engineering, Salt Lake City, LIT, for Office
Testing Methods International Society for Rock Mechanics. of Nuclear Waste Isolation, Battelle Memorial Institute, Colum-
Pergamon Press, Oxford (1981). bus, OH (HQS.880517.2252) 0982).
2. U.S. Department of Energy. Site Characterization Plan, Yucca 9. Jaeger J. C. and Cook N. (3. W. Fundamentals of Rock Mechanics,
Mountain Site, Nevada Research and Development Area, Nevada. 2nd Edn. Wiley, New York (1977).

Das könnte Ihnen auch gefallen