You are on page 1of 11

Issues in Evaluating Capacity of Rock Socket Foundations

F. H. Kulhawy
School of Civil and Environmental Engineering, Cornell University, Ithaca, NY, USA
fhk1@cornell.edu
W. A. Prakoso
Department of Civil Engineering, University of Indonesia, Depok, Indonesia
wprakoso@eng.ui.ac.id

Abstract: Drilled foundations often are socketed into rock to increase the capacity. However, procedures to quantify the socket side
and tip resistance vary considerably. This paper reviews methods to predict socket capacity and critically assesses them. One method
for side resistance is recommended, and several approaches are suggested to assess tip resistance, depending on the degree of geologic
data available. Statistics for the methods are given, where available, and design and construction implications are noted.

1 INTRODUCTION 2 ROCK MASS ENGINEERING PROPERTIES

Drilled shafts (bored piles) are a common foundation selection The capacity of foundations in rock is a function of the rock mass
for all types of structures. When the structure loads are relatively strength, which often is estimated, at least partially, from the
large or where the soil is of relatively poor quality, the shafts intact rock strength, which in turn often is estimated from the
often are drilled through the soil to the underlying rock mass. intact rock index properties. However, several key issues need to
These shafts could be founded or seated on the rock mass be addressed during property evaluation, as described below. If
surface, or they could be drilled into the rock mass to form a rock these testing issues are not addressed properly, then the subse-
socket, as shown in Fig. 1. The applied butt load or stress is quent capacity predictions are likely to be in error.
supported by the socket through both tip and side resistances,
assuming for illustration that the soil is non-contributory. How
the loads are distributed between the tip and side is a function of 2.1 Intact Rock: Effect of Testing Parameters
the loading magnitude, problem geometry, properties of the rock Fig. 2 illustrates a very important testing issue, which is the
mass and shaft concrete, ultimate bearing capacity of the tip, side influence of sample diameter on the resulting uniaxial compres-
resistance of the socket, and butt displacement. Discussion of all sive strength (qu). Most standards specify a sample diameter on
of these issues is well beyond the scope of this paper. the order of 50 to 54 mm. As can be seen, non-standard samples
Herein, the basics of socket capacity are addressed. Key rock tend to give strengths that decrease with increasing sample
mass property issues are discussed first. Then methods are given diameter. Similar trends were noted with other strength measures
to calculate the socket capacity. The side resistance can be quan- as well (Prakoso 2002).
tified well, and various approaches are described to assess the tip Note that, in this figure and others herein, a wide variety of
resistance. Some simple observations are made regarding dis- igneous, sedimentary, and metamorphic (primarily non-foliated)
placement limits. The paper concludes with general observations rock types are included. These data were collected from the lit-
on construction and field acceptance criteria. erature but were included only if both high-quality foundation
load tests and physical property tests were conducted.
In Fig. 2, the solid line represents the regression for the entire
data population. When these data were examined by separate
rock type family, it was found that each rock type gave results
that were very similar to the entire population. For example, the
dashed line represents the results for all carbonate rock types. As
can be seen, the results are very similar.
Fig. 3 illustrates the importance of testing at the field water
content. As can be seen, the saturated qu is only about 0.79 of the
dry value. Although not shown, comparable data for the point
load index (IS) give a value of 0.84, while data for the Brazilian
tensile strength (qt-Brazilian) give a value of 0.89. Therefore, testing
samples that have been allowed to dry clearly will overestimate
the actual in-situ strength.
Fig. 1. Illustrative rock socket.
2.0 4000
I. Intrusive I. Intrusive
I. Pyroclastic I. Extrusive
S. Clastic I. Pyroclastic
S. Chemical S. Clastic
1.5 M. Non-Foliated 3000 S. Chemical
Strength Ratio, SRqu

M. Foliated
M. Non-Foliated

Mean qu / pa
1.0 2000
qu / pa = 143.1 exp (0.048 R)
m = 78, r2 = 0.57

0.5 SRqu = [50 / Bsample] 0.25 1000

0.0 0
0 50 100 150 200 0 20 40 60 80

Diameter, Bsample (mm) Mean R

Fig. 2. Effect of sample size on uniaxial compressive strength qu Fig. 4. Relationship between qu and R for Bsample = 50 - 58 mm
(Kulhawy & Prakoso 2001). (Kulhawy & Prakoso 2003; Prakoso & Kulhawy 2004b).

300 ×106
2.02000
I. Intrusive I. Intrusive
I. Extrusive I. Extrusive
S. Clastic I. Pyroclastic
S. Chemical S. Clastic
M. Non-Foliated ×106
Mean qu-saturated (MPa)

1.51500 S. Chemical
200 M. Foliated
M. Non-Foliated
Mean Et-50 / pa

×106
1.0 1000

100 qu-sat = 0.79 qu-dry


m = 67, r2 = 0.92
S.D. / pa = 165 106
0.5×500

Et-50 / pa = 5280 (qu / pa)0.62


m = 100, r2 = 0.57
0 0
0 100 200 300 0 1000 2000 3000 4000 5000

Mean qu-dry (MPa) Mean qu / pa

Fig. 3. Effect of water content on qu (Kulhawy & Prakoso 2001). Fig. 5. Relationship between Et-50 and qu.

2.2 Intact Rock: Index Property and Strength Correlations 2.3 Intact Rock: Strength and Modulus Correlation
For small projects, and for general correlation studies, various The intact rock Young’s modulus often is represented by the tan-
quick and simple index tests have been used to estimate the intact gent Young’s modulus at 50% of the uniaxial compressive
rock uniaxial compressive strength (qu), including the Schmidt L- strength (Et-50). This value commonly is estimated from the uni-
hammer rebound hardness (R), Shore scleroscope hardness (Sh), axial compressive strength (qu), as shown in Fig. 5. The regres-
and point load index (IS). Many correlations among these sion equation in Fig. 5 is comparable to, and a bit less than, the
parameters have been proposed. Fig. 4 shows our relationship typical modulus correlation for concrete, which is given by E =
between qu and R, as normalized by the atmospheric stress, pa. 5000 (fc’ in MPa)0.5. Note that Et-50 / qu varies from about 500 or
Although not shown, the Sh and IS correlations are as follows: more at low strength to about 200 at very high strength.
qu (MPa) = 7.57 exp (0.064 Sh) [m = 30, r2 = 0.77] and qu = 23.3
IS [m = 43, r2 = 0.77]. Other useful correlations are summarized
2.4 Intact Rock: Weathering
by Prakoso (2002). It should be noted that these correlations are
not deterministic; there is always a transformation uncertainty The deleterious effect of weathering on intact rock properties is
associated with them. Using any of these correlations implies that well-recognized, but quantifying this effect is more difficult. Fig.
the COV of qu will be larger than that of R, Sh, or IS. 6 shows various mean correlations between the properties of
unweathered rock and rock weathered to varying degrees. These
Mean Property Ratio (Weathered / Unweathered)
1.0
SRblock = (σ1-σ3)f-block / (σ1-σ3)f-intact (1)

0.8 in which (σ1-σ3)f-block and (σ1-σ3)f-intact = deviator stresses at


failure of the “rock mass” and “intact rock”, respectively. The
results from several sets of tests, with typical confining stress
0.6
(σ3) for foundations (σ3 < 1 MPa), are plotted versus θ in Fig. 7,
and they can be fitted by the following:

θ < 40° SRblock = - 0.02 θ + 0.9 (2a)


0.4 γ 40° < θ < 60° SRblock = 0.1 (2b)
R θ > 60° SRblock = 0.02 θ - 1.1 (2c)
qu
0.2 qt-Brazilian The S.D. of Eqs. 2a and 2b is 0.09, while that of Eq. 2c is 0.10. It
Is can be inferred that the variability of rock mass strength, as rep-
Et-50 resented by the COV, is maximum for θ between 40° and 60°.
0.0 The effect of number of discontinuities, typically assessed by
Slight Moderate High using different block sizes, also was evaluated using the strength
ratio (SRblock) approach. The results from blocks with three dif-
Weathering Conditions ferent discontinuity orientations (vertical only, horizontal only,
and vertical and horizontal) were considered. Only three numbers
Fig. 6. Effect of weathering on rock properties (Kulhawy & of discontinuities were available for evaluation. As shown in Fig.
Prakoso 2001, 2003). 8, SRblock decreases only slightly with increasing number of dis-
continuities. However, the variability of SRblock appears to in-
degrees are somewhat subjective, as given by the source authors. crease with increasing number of discontinuities. In addition, for
This database is dominated by igneous intrusive rocks, followed vertical and horizontal discontinuities, the effect of discontinuity
by igneous extrusive and sedimentary clastic rocks. However, the orientation appears to be minimal.
general rock type is not expected to have a significant effect on These test results also were used to estimate the variability of
the overall results. rock mass strength, which is defined as the deviator stress at fail-
All properties decrease with increasing weathering. The unit ure [(σ1-σ3)f]. The results were separated based on the disconti-
weight (γ) decreases only a modest amount, to about 90% of the nuity angles (θ = 0-25° and 70-90°, and θ = 25-70°) to consider
unweathered value. However, all other properties decrease sub- the effect of θ on the strength. The results are plotted versus the
stantially, with the R value decreasing to about 40% of the confining stress (σ3) in Fig. 9 and show that the COV of (σ1-σ3)f
unweathered value. The strength and modulus values decrease
even more, to about 15 to 25% of the unweathered values. These 1.0
decreases should be addressed in engineering evaluations. 1: SRblock = -0.02 θ + 0.9; S.D. = 0.09
SRblock = (σ1 - σ3)f-block / (σ1 - σ3)f-intact

Furthermore, these decreases are not deterministic; there is 2: SRblock = 0.1; S.D. = 0.09
always some uncertainty in the data. The standard deviation 3: SRblock = 0.02 θ - 1.1; S.D. = 0.10
0.8
(S.D.) for γ is about 0.04 for all degrees of weathering, but the *

S.D. of the other properties is about 0.11. Therefore, the coeffi-


cient of variation (COV = S.D. / mean) increases substantially as
0.6
the weathering increases, from about 15% for slightly weathered
to 50% or more for highly weathered. These substantial varia- 3
tions need to be addressed cautiously.
0.4
1
2.5 Rock Mass: Strength of Artificial Rock Blocks
0.2 *
In bearing capacity calculations, it is necessary to estimate the
rock mass strength, which is difficult to do because of the need to 2
assess the in-situ rock mass structure. Instead, researchers have
conducted artificial rock block tests to estimate the effect of rock 0.0
mass structure on its strength (e.g., Brown 1970; Brown & Trol- 0 30o 60o 90o
lope 1970; Ladanyi & Archambault 1972; Einstein & Hirschfeld
1973; Kulatilake et al. 1997; Yang et al. 1998). Herein, these test Discontinuity Angle to Horizontal Plane, θ
results were compiled and re-analyzed to evaluate the rock mass
strength relative to the intact rock strength and to estimate the Einstein et al. (1969, 1973), σ3 = 0
variability of the rock mass strength. Ladanyi & Archambault (1972), σ3 = 0.35 MPa
In general, the rock mass strength is dependent on the primary Ladanyi & Archambault (1972), σ3 = 0.70 MPa
discontinuity orientation. However, the results of artificial block Kulatilake et al. (1997), σ3 = 0, symm.
tests are not always in agreement with theoretical solutions. Kulatilake et al. (1997), σ3 = 0, asymm.
Therefore, a simplified approach was adopted in evaluating the Yang et al. (1998), σ3 = 0
effect of discontinuity angle relative to a horizontal plane (θ) on
the strength. This effect is represented by a strength ratio Fig. 7. Effect of discontinuity angle on rock block strength
(SRblock), which is given by: (Prakoso & Kulhawy 2004b).
1.5 80
Discontinuity Angle to
SRblock = (σ1 - σ3)f-block / (σ1 - σ3)f-intact

Horizontal Plane, θ
0o - 25o 25o-70o
70o - 90o
60 Einstein & Hirschfeld (1973)

COV of (σ1 - σ3)f (%)


Brown & Trollope (1970),
1.0 Brown (1970)
Ladanyi & Archambault (1972)

40

0.5
Vertical Discontinuities
Horizontal Discontinuities 20
Vertical & Horizontal Discontinuities
No. Discontinuity = 2: Mean = 0.98; S.D. = 0.12
No. Discontinuity = 4: Mean = 0.92; S.D. = 0.10
No. Discontinuity = 8: Mean = 0.92; S.D. = 0.17
0.0 0
0 2 4 6 8 10 0 4 8 12 16

Number of Discontinuities Confining Stress, σ3 (MPa)

Fig. 8. Effect of number of discontinuities on rock block strength Fig. 9. Effect of confining stress on COV of rock block strength
(Prakoso & Kulhawy 2004b). (Prakoso & Kulhawy 2004b).

decreases with increasing σ3 but, for typical σ3 values for founda- 10000
tions (σ3 < 1 MPa), the COV range still is wide, from 10 to 75%. log10(Em / qu) = 2.73 - 0.49 log10(qu / pa)
In addition, the prior group of discontinuity angles (θ = 0-25° m = 71, r2 = 0.48, S.D. log(Em / qu) = 0.26
and 70-90°) tends to yield a lower COV of (σ1-σ3)f.
1000
Modulus Ratio, Em / qu

2.6 Rock Mass: Modulus


A realistic rock mass Young’s modulus (Em) is required in any
foundation displacement analysis, but typically it is obtained by 100
conducting field load tests, which are rather expensive. Alterna-
tively, Em can be estimated from the intact rock uniaxial
compres-sive strength (qu) or the intact rock modulus (Et-50).
A modulus ratio can be defined as follows: 10
Mudstone
Shale
Modulus Ratio = Em / qu (3) Sandstone
Others
Using mainly the data base developed by Rowe and Armitage 1
(1984), this ratio is plotted versus qu in Fig. 10. Note that the 1 10 100 1000
exclusion of some data in Fig. 10 was based on a further detailed
statistical analysis (Prakoso 2002). Uniaxial Compressive Strength, qu / pa
Em also can be estimated from Et-50 using a modulus reduction
factor defined as follows: Fig. 10. Rock Mass Young’s Modulus from Load Tests (Filled
symbols excluded from statistical analysis).
αE = Em / Et-50 (4)
10
In this form, αE is a lumped parameter that includes the intact Mean = 0.32
S.D. = 0.26
rock properties and the discontinuity frequencies and properties. 8
No. Observations

m = 27
Using the data base developed by Heuze (1980), the distribution
of αE is shown in Fig. 11. 6
In addition, geomechanical models have been proposed to
Log-Normal
estimate Em. The orthogonal model proposed by Kulhawy (1978) 4
Distribution
incorporates key physical properties of the intact rock and the
rock discontinuities, as well as the mean discontinuity spacing, as 2
given by:
0
−1 0.0
0 0.2 0.4 0.6 0.8 1.0
⎛ ⎞
⎜ 1 1 ⎟
Em =⎜ + ⎟⎟ αE = Em / Et-50
⎜ Er Sj Kn
⎝ ⎠ (5)
Fig. 11. Distribution of αE from plate bearing tests.
0.8
Simulation Results
Negative Exponential
0.6 Log-Normal (COV = 50%)
Log-Normal (COV = 100%)
Kulhawy (1978) QL2
0.4 Final linear region

Transition region

Load
0.2 QL1
Er / Kn (m) = 0.5
Mean αE

0.8 Initial linear region


Simulation Results
Negative Exponential
0.6 Log-Normal (COV = 100%)
Kulhawy (1978)

0.4 Er / Kn (m) = 0.1

Displacement
0.2 Er / Kn (m) = 1.0
Fig. 13. Generalized load-displacement behavior.
0
0 20 40 60 80 100 (absolute and percent of diameter), graphical constructions, and
mathematical functions. These also reflect a mix of what actually
Mean RQD (%) are both ultimate limit state and serviceability limit state criteria.
Our detailed studies (e.g., Hirany & Kulhawy 1988; Prakoso
Fig. 12. Relationship between Rock Mass Modulus and RQD. 2002) indicate that a consistent and reasonable method for
defining the "interpreted failure load" is to use QL2, which is the
in which Er = intact rock modulus (typically given by Et-50), Sj = load at L2. Similarly, QL1 is the load at L1, which represents the
discontinuity spacing, and Kn = discontinuity normal stiffness. "elastic limit". The L1 and L2 points are determined graphically
This Em also can be correlated to Er = Et-50, as in Eq. 4, to define from a plot at a scale similar to that of Fig. 13. As can be seen,
the modulus reduction factor αE. QL2 always follows the nonlinearity, sometimes represents the
The discontinuity spacing (Sj) is not obtained routinely in actual curve peak where there is little or no dilation, and can be
foundation practice, but the Rock Quality Designation (RQD) is evaluated from virtually all quality test data.
used commonly to characterize the rock mass. Kulhawy (1978) Once the "capacity" is defined, then the tip and side resis-
used a simple geometric model to relate Si to RQD; others have tances can be evaluated based on measurements made in com-
used statistical or random number-generated relationships. In any pression tests of full sockets. In uplift tests, and in compression
case, as shown in Fig.12, the effect of different Sj-RQD relation- tests with a void or frangible material beneath the tip, the evalua-
ships is minor, compared to that of the properties of the rock and tion is straightforward and only requires consideration of the
discontinuities. shaft weight.
Most often, the tip and side resistances then are compared to
one of the simpler rock material indices, such as the uniaxial
3 ROCK SOCKET CAPACITY compressive strength (qu). The qu tests should all be done in
accordance with proper test procedures, such as those given by
3.1 Generalized Socket Behavior ASTM, ISRM , or others. Estimating qu from simpler tests such
as the point load index, Schmidt hammer, or others, may be inap-
Fig. 13 depicts the generalized load-displacement behavior of
propriate, as shown by the variability in the correlations shown
drilled shafts under axial load. This general pattern holds in both
previously. Strictly speaking, any comparison also should be
soil (Hirany & Kulhawy 1988) and rock (Carter & Kulhawy
with the average qu over the depth of the socket.
1988), as shown in many load tests that were carefully conducted
Most studies conducted to date have not met these criteria,
and well-documented. There is essentially a linear response from
based on the documentation presented or stated. This statement is
the origin to L1, followed by a nonlinear transition region to L2,
not intended to fault the authors, who undoubtedly presented the
after which there is a final linear region. In rock masses, these
best information they could. It is intended to point out that we are
regions correspond to initial linear elastic behavior, followed by
frequently dealing with imperfect and sometimes poor data, and
bond breakage and progressive slip, and then full frictional slip
therefore our expectations should be tempered accordingly.
with dilation. The same general pattern holds for both compres-
sion and uplift tests, although the relative sizes and importance of
the regions differ somewhat. In all cases, the occurrence of a 3.2 Calculation Model
clearly defined peak to the curve is infrequent.
With nonlinear curves such as these, there is always a major In general, foundation capacity is a function of the tip resistance
question about how to define the foundation "capacity" for sub- (Qt), side resistance (Qs), and foundation weight (W). By force
sequent design use. Examination of the literature (Hirany & equilibrium, the compression capacity (Qc) is given by:
Kulhawy 1988) reveals at least 41 different methods used for the
interpretation of axial load tests, including displacement limits Qc = Qtc + Qsc – W (6)
in which the subscript c refers to compression. The uplift capac- 10
ity (Qu) is given by: Nc* = 3.38, S.D. = 1.20

Qu = Qtu + Qsu + W (7) 8

Bearing Capacity Factor, Nc*


in which the subscript u refers to uplift.
In most design cases, only limited information is available on 6
the rock mass properties and in-situ conditions, and consequently
the use of theoretical solutions is difficult. More often, the only
rock strength property available is the intact rock uniaxial com-
4
pressive strength (qu), and therefore the foundation resistances
typically are related to qu. In simplified fashion, the tip (or base)
resistances of circular footings and drilled shafts in compression
can be estimated by: 2

Qtc = 0.25 π B2 Nc* qu (8)


0
in which B = foundation diameter and Nc* = empirical tip resis- 0 0.4 0.6 0.8 1.0 1.2
tance factor. Information on the tip resistance in uplift is very
limited, so this resistance is not discussed herein. Socket Diameter, B (m)
The side resistance of drilled shaft foundations socketed in
rock involves a complex interaction among the adhesion, friction, Fig. 14. Drilled socket tip resistance factor (Prakoso & Kulhawy
dilatancy, and normal stress effects along the socket wall. These 2002).
effects are difficult to measure or estimate, and therefore they
often are lumped into an average unit side resistance (f). Using 8
Mean = 3.38
this simplification, the side resistance (Qs) can be estimated by: Log-Normal COV = 35.4%
Distribution m = 14
No. Observations

6
Qs = π B D f (9)

in which B = foundation diameter and D = foundation socket 4


depth. It is assumed commonly that f can be related directly to
the intact rock uniaxial compressive strength (qu), and therefore 2
the side resistance is given by:

Qs = π B D αr qu (10) 0
0 1 2 3 4 5 6
in which αr = empirical side resistance factor = f / qu.
Tip Resistance Factor, Nc*

3.3 Tip Resistance Fig. 15. Distribution of drilled socket tip resistance factor
The tip resistance for tests conducted on the socket tip and on (Prakoso & Kulhawy 2002).
complete sockets was evaluated to assess the range of the tip
resistance factor (Nc*). The data base developed for this study data and noted that there is an approximate lower bound to side
included 9 sites with 14 field load tests conducted in several rock resistance that is given by:
types, mainly in fine-grained sedimentary rocks, and 2 centrifuge
laboratory tests. All of the load tests had qu, and all were con- f / pa = 0.63 (qu / pa)0.50 (11)
ducted on straight-sided rock sockets. Axial compressive loading
was applied in all cases, with 7 tests performed on socket tips and in which f = average unit side resistance. To link the f / pa format
7 tests performed on complete sockets. For the complete sockets, with the αr format, the equations are given by:
the tip resistance was determined from the reported tip and side
resistance load distribution. log10 αr = A – B log10 (qu / pa) (12a)
The tip resistance factor (Nc*) was evaluated using Eq. 8, and f / pa = 10A (qu / pa)B (12b)
these values are plotted versus the corresponding socket diameter
(B) in Fig. 14, in which Nc* appears to be independent of B. The After examination of these data, they also made two important
mean of the tip resistance factor (mNc*) and its COV (COVNc*) design check recommendations. First, values of f in excess of
are given by mNc* = 3.38 and COVNc* = 35.4%. The distribution 0.15 qu, over the full range of expected values, should be used
of Nc* is shown in Fig. 15, and it resembles a log-normal prob- only when they are demonstrated to be reasonable by a load test,
ability distribution with the same mNc* and COVNc*. Furthermore, local experience, or adequate in-situ testing. And second, after
Nc* appears to be independent of the rock type. obtaining the design value of f, typically from Eq. 9, and apply-
ing a factor of safety to this value, a check should be made
against the concrete bond value of 0.05 f'c. The lower value
3.4 Side Resistance should be used unless load test data show otherwise.
More recently, Prakoso (2002) re-examined the data available
Carter & Kulhawy (1988) reviewed the Rowe & Armitage (1984)
and attempted to evaluate them in a more consistent manner.
First, the only data used were those that had load-displacement
curves to failure so that the "interpreted failure load" could be 1
determined for all the data and therefore the "capacities" were
evaluated in a consistent manner. However, it was not possible to
reevaluate the qu data to ensure consistency in test conduct and

Side Resistance Factor, αr


averaging over the shaft depth. An initial assessment of addi- 0
0.1
tional Asia data (e.g., Ng et al. 2001) indicates that they fall in
the data range as above.
Fig. 16 shows the results for all of the data, including multiple
tests at the same site and results for (a) shafts in natural and man- 0
0.01
made rocks, (b) grouted piles in natural rocks, and (c) rock an-
chors in natural rocks. The regression line is given by:
log10 αr = 0.24 - 0.67 log10(qu / pa)
f / pa = 2.00 (qu / pa)0.69 (13) 0.001
0 m = 52, r2 = 0.69, S.D. = 0.30
I. Intrusive S. Clastic (coarse)
I. Extrusive S. Chemical
Fig. 17 shows the results of the data averaged per test site. The I. Pyroclastic M. Non-Foliated
regression line is given by: S. Clastic (fine) Man-Made
0.0001
0
f / pa = 1.74 (qu / pa)0.67 (14) 1 10 100 1000 10000

Uniaxial Compressive Strength, qu / pa


Careful examination of these results indicates that the rock an-
chor data are clustered in the lower portions of the figure, espe-
cially in the lower right. Setting these data aside gives the results Fig. 17. αr vs. qu for all data, averaged per site (Kulhawy et al.
for drilled shafts and grouted piles as shown in Fig. 18 by the 2005).
solid line. The regression line corresponds to:
1
0.50
f / pa = 0.98 (qu / pa) (15)

which can be conveniently rounded to


Side Resistance Factor, αr

0
0.1
0.50
f / pa = (qu / pa) (16)

The lower bound 10A value of 0.63 that was cited previously
0
0.01
actually represents the lower bound for 90% of the data in Fig.
18. To capture 100% of the data, the absolute lower bound would
be about 0.5. It should be noted in Fig. 18 that the regression is log10 αr = - 0.01 - 0.50 log10(qu / pa)
altered significantly when the rock anchor data are included. m = 41, r2 = 0.51, S.D. = 0.31
0.001
0 I. Intrusive S. Clastic (coarse)
Clearly these data constitute a separate population. I. Extrusive S. Chemical
In addition to the general relationships described above, there I. Pyroclastic M. Non-Foliated
S. Clastic (fine) Man-Made
have been a number of studies that have focused exclusively on Regression Line for Data with Rock Anchors

1 0.0001
0
1 10 100 1000 10000

Uniaxial Compressive Strength, qu / pa


Side Resistance Factor, αr

0
0.1
Fig. 18. αr vs. qu for drilled shafts and grouted piles, averaged per
site (Kulhawy et al. 2005).

localized rock units, such as the chalks of southern England and


0
0.01
the limerocks of Florida. These studies are of local importance
and are too specialized to be discussed herein. When these are
addressed, they should be considered within the broad framework
log10 αr = 0.30 - 0.69 log10(qu / pa) described above.
0.001
0 m = 104, r2 = 0.72, S.D. = 0.29
I. Intrusive S. Clastic (coarse)
I. Extrusive S. Chemical
I. Pyroclastic M. Non-Foliated 3.5 Effect of Socket Roughening
S. Clastic (fine) Man-Made
0.0001
0 The values of αr for roughened drilled foundations are plotted
1 10 100 1000 10000 versus their corresponding qu/pa in Fig. 19. The αr decreases with
increasing qu/pa, and the regression equation is given by:
Uniaxial Compressive Strength, qu / pa
f / pa = 1.91 (qu / pa)0.46 (17)
Fig. 16. Non-roughened side resistance of drilled foundations
(Kulhawy et al. 2005). The side resistance factor (αr) differs with socket roughening.
1 100
I. Intrusive S. Clastic (coarse)

Side Resistance, f / pa
I. Extrusive S. Chemical
80 I. Pyroclastic M. Non-Foliated
S. Clastic (fine) Man-Made
0.4
Side Resistance Factor, αr

60
0.05 FSlim (fc' / pa) FSlim = 3

40 fc' / pa = 400
2
3
0
0.1 20 fc' / pa = 200 2

0
0.04 1 10 100 1000 10000
log10 αr = 0.28 - 0.46 log10(qu / pa)
m = 43, r2 = 0.67, S.D. = 0.16
S. Clastic
Uniaxial Compressive Strength, qu / pa
S. Chemical
Man-Made Fig. 20. Socket side resistance versus concrete bond strength
0.01
0 (Kulhawy et al. 2005).
1 10 100 1000 10000
Table 1. Comparison of Side Resistance and Concrete Bond
Uniaxial Compressive Strength, qu / pa Strength (Kulhawy et al. 2005).
Rock Socket % socket fallow > 0.05 f'c
Fig. 19. Roughened side resistance of drilled foundations. FSlim = 2 FSlim = 3
Non-Roughened
For non-roughened sockets, αr is lower than that for roughened f'c / pa = 200 16% 4%
sockets. The overall trend of both data sets and the regression f'c / pa = 400 2% 0%
lines is similar. Both regression lines are close for lower qu, sug-
gesting that the nominal side resistances are about the same. Roughened
f'c / pa = 200 40% 14%
f'c / pa = 400 9% 0%
3.6 Rock Socket Side Resistance and Concrete Bond Strength
Concrete bond strength = 0.05 FSlim (f'c / pa)
Carter & Kulhawy (1988) recommended a design check to com-
pare the allowable side resistance of the rock socket (f / FS) to variation (φr = φj) on Ncs is shown in Fig. 21. As φr increases, the
the concrete bond strength, given by 0.05 f'c. The lower value maximum and minimum Ncs increase, θ1 for the minimum Ncs
would control, unless field testing showed otherwise. By using decreases, and the ratio of maximum to minimum Ncs increases
typical safety factors of 2 and 3, the ultimate side resistance can slightly. Also, the range of θ1 affecting Ncs decreases, and the
be compared with the factored concrete bond strength, as given shape of the line of Ncs changes with increasing φr.
in Fig. 20. Typical ranges of concrete strength, f'c / pa = 200 - The effect of discontinuity cohesion, given by cj / cr, is
400, were used for comparison. shown in Fig. 22. As cj / cr increases, the minimum Ncs increases,
Fig. 20 shows that most side resistances are below the lower and the range of θ1 affecting Ncs decreases. The effect of
concrete strength and factor of safety. All of these cases showed discontinuity friction angle (φr ≥ φj) is shown in Fig. 23. As φj
acceptable behavior when the bond strength of the concrete was increases, the minimum Ncs increases, and θ1 for the minimum
exceeded. Clearly the concrete behaves better when it is confined Ncs changes. Also, the range of θ1 affecting Ncs decreases, and the
in a socket and reinforced than when it is unconfined and unre- shape of the line of Ncs changes significantly with increasing φj.
inforced. The percentages are given in Table 1, which shows that Ncs for rock masses with two discontinuity sets is given in Fig.
there are more cases of sockets exceeding the concrete bond 24. As shown, Ncs is influenced significantly by θ1 and the angle
strength with lower concrete strength and factor of safety. Again, between the discontinuity sets (Δθ). The minimum Ncs and θ1 for
all of these cases showed acceptable behavior when the bond this minimum differ for different Δθ values. Also, the range of θ1
strength of the concrete was exceeded. affecting Ncs varies, and the shape of the line of Ncs changes sig-
nificantly with different Δθ values. The results for both one and
two discontinuity sets suggest that the strength of both the rock
4 ROCK MASS CONDITIONS & ROCK FOUNDATION material and the discontinuities, and the number and orientation
BEARING CAPACITY: THEORETICAL RELATIONSHIPS of the discontinuity sets, all have significant effects on Ncs.
Prakoso & Kulhawy (2006) provide an update of the Kulhawy
Prakoso & Kulhawy (2004a) proposed a lower bound bearing & Goodman model for foundations on rock with vertical discon-
capacity model, coupled with a simple discontinuity strength tinuities. For a rock mass with open vertical discontinuities,
model, for strip footings on jointed rock masses. The strength of where the discontinuity spacing (Sj) is less than or equal to the
both the rock material (φr and cr) and the discontinuities (φj and foundation width (B), the likely failure mode is uniaxial com-
cj), and the number and orientation of the discontinuity sets pression of rock columns. The ultimate capacity based on the
(θ1…n), are considered explicitly in the model. The lower bound Mohr-Coulomb failure criterion then is given by:
bearing capacity factor (Ncs) for strip footings on rock masses
with a single discontinuity set is given in Figs. 21 through 24. qult = qu = 2 c tan (45 + φ / 2) (18)
The uniaxial compressive strength (qu) of the rock material, nor-
malized by cr, also is given. In all figures, Ncs is related to the in which qu = uniaxial compressive strength, c = cohesion, and φ
discontinuity orientation angle (θ1). The effect of friction angle = friction angle. The qu, c, and φ are rock mass properties.
60 30
cj / cr = 0.3 φr = 40o; cj / cr = 0.3
50 θ1 θ1
50o
Bearing Capacity Factor, Ncs

Bearing Capacity Factor, Ncs


40o
40 20
35o
45o
30
40o
20 10
35o
30o

10 25o qu /cr
φr = φj = 30o
φj = 20o
0 0
0 30 60 90 0 30 60 90

Discontinuity Angle to Horizontal Plane, θ1 (o) Discontinuity Angle to Horizontal Plane, θ1 (o)

Fig. 21. Lower bound bearing capacity of strip footings on Fig. 23. Effect of discontinuity friction angle on lower bound
jointed rock masses - one discontinuity set (Prakoso & Kulhawy bearing capacity (Prakoso & Kulhawy 2004a).
2004a).
30
30 φr = φj = 40o; cj / cr = 0.3

θ1
Bearing Capacity Factor, Ncs

θ2 = θ1 + Δθ
Bearing Capacity Factor, Ncs

0.9
20
20 0.7

0.5
θ1
10
0.3 φr = φj = 40o
10
Δθ = 30o qu /cr
cj / cr = 0.1 qu /cr Δθ = 60o
Δθ = 90o
0
0 0 30 60 90
0 30 60 90
Discontinuity Angle to Horizontal Plane, θ1 (o)
Discontinuity Angle to Horizontal Plane, θ1 (o)
Fig. 24. Lower bound bearing capacity of strip footings on
Fig. 22. Effect of discontinuity cohesion on lower bound bearing jointed rock masses - two discontinuity sets (Prakoso & Kulhawy
capacity (Prakoso & Kulhawy 2004a). 2004a).

For a rock mass with vertical discontinuities spaced wider in which Sj = spacing between vertical discontinuities and Nφ =
than the foundation width (B), the likely failure mode is splitting bearing capacity factor given by:
of the rock mass. Bishnoi (1968) proposed the following format
to evaluate this failure mode: Nφ = tan2 (45 + φ / 2) (21)

qult ≈ J c Ncr (19) As the spacing between a pair of vertical discontinuities (Sj)
increases, the failure mode changes, from splitting of the rock
in which J = correction factor, c = intact rock cohesion, and Ncr = mass to general shear failure. For general shear, the modified
bearing capacity factor. The J factor is later. The bearing solution proposed by Bell (1915) can be used:
capacity factor (Ncr) is given by Goodman (1980):
qult = c Nc ζcs ζcd + 0.5 B γ Nγ ζγs ζγd + q Nq ζqs ζqd (22)
2 Nφ 0.5⎡ ⎛ Sj ⎞
(1−1 N φ ) ⎤
N cr = ⎢ Nφ ⎜ ⎟ − 1⎥ (20) in which c = rock mass cohesion, B = foundation diameter or
Nφ − 1 ⎢ ⎜B⎟ ⎥
⎣ ⎝ ⎠ ⎦ width, γ = rock mass effective unit weight, q = γD = overburden
stress at tip, D = foundation depth, and Nc, Nγ, and Nq = bearing
capacity factors. The factors ζcs, ζcd, ζγs, ζγd, ζqs, and ζqd are 1000
modifiers for square or circular foundations; the second subscript
“s” denotes the shape factor and the second subscript “d” denotes
the depth factor. For shallow foundations, D is very small, and

Bearing Capacity Factor, Ncr


60o
the 0.5 B γ Nγ term normally is very small compared to the c Nc
term, and therefore Eq. 22 often is simplified as: 100 Ncr Equation 50o
qult = c Nc ζcs (23a)
qult = 2 c [tan(45 + φ / 2) + tan3(45 + φ / 2)] ζcs (23b) 40o
Uniaxial
Compression 30o
The modifying factor ζcs is given by:
10 20o
ζcs = 1 + Nφ1.5 / [2 (Nφ + 1)] (24) φ = 10o
Bell Solution
The results of Eqs. 18-23 are shown in Fig. 25 for a range of rock
Limit for Sj > B
mass friction angles. Note that, as the results are given in terms
of Ncr, for the uniaxial compression failure mode, Ncr is given by: 1
0
0.1 1 10 100
Ncr = 2 tan (45 + φ / 2) (25)
Discontinuity Spacing, Sj / B
For the general wedge failure mode, Ncr is given by:
Fig. 25. Capacity factor for vertical open discontinuities (Prakoso
Ncr = Nc ζcs (26) & Kulhawy 2006).

The changes in failure modes in Fig. 25 are identified by the 2.0


dashed lines.
The correction factor (J), based on Bishnoi (1968), also was
updated. Contrary to Bishnoi’s suggestions, the results show no
apparent trend, as can be seen in Fig. 26. The mean and COV of J 1.5
Correction Factor, J

are 1.14 and 33.3 percent, respectively.

5 DEFORMATION OF ROCK SOCKETS 1.0

Carter and Kulhawy (1988) suggested that the displacement of a


rigid shear socket under uplift loading can be evaluated by:
0.5

⎛ ζ ⎞ ⎛ 1 ⎞ Igneous Intrusive
yu = ⎜ ⎟ ⎜ ⎟ F Sedimentary Chemical
⎜2 π⎟ ⎜ Gm D ⎟ u Concrete
⎝ ⎠ ⎝ ⎠ (27) 0.0
0 5 10 15 20
in which ζ = ln[5 (1-ν) D/B], ν = rock mass Poisson’s ratio, Gm =
rock mass shear modulus, D = shaft depth, and Fu = applied uplift Thickness of Rock Layer, H/B
load. This equation also is used for a single anchor in uplift.
Carter and Kulhawy (1988) also suggested that the elastic dis- Fig. 26. Correction factor J (Prakoso & Kulhawy 2006).
placement of a rigid complete socket can be evaluated by:
These equations are valid up until L1, at which displacements
Fc typically are on the order of 10-15 mm. Beyond L1, nonlinearity
yc = and load transfer must be addressed.
⎛ Eb B ⎞ ⎛ π ⎞ ⎛ Em D ⎞
⎜ ⎟+⎜ ⎟ ⎜ ⎟
⎜ 1 − ν 2 ⎟ ⎜⎝ ζ ⎟⎠ ⎜ 1+ ν ⎟
⎝ b ⎠ ⎝ ⎠
(28) 6 CONSTRUCTION AND FIELD ACCEPTANCE
CRITERIA
in which Fc = applied compressive load, Eb = rock mass Young’s
modulus below the tip, νb = rock mass Poisson’s ratio below the
In general, when constructing a rock socket, it is necessary to
tip, and B = socket diameter. Note that, because the typical val-
ensure that the rock mass is of sufficient quality to carry the load
ues of ν and νb are low and Eb = Em is commonly assumed, Eq.
without adverse behavior. To achieve this goal, it is common to
28 can be simplified as follows:
set exploration and/or construction criteria that must be met.
First and foremost, there must be sufficient exploration data to
Fc E m
yc ≈ define the rock materials present and to delineate the rock mass
⎛π⎞ ⎛ D B⎞ structure and discontinuities. These data should be of sufficient
1 + ⎜⎜ ⎟⎟ ⎜ ⎟
⎝ ζ ⎠ ⎝ 1+ ν ⎠ depth beneath the tips of the rock sockets to define the rock mass
(29) well enough so that the bearing capacity and settlement can be
computed with some confidence. In particular, it is necessary to
define the layering and/or soft seams that can be present in many First, several methods are presented to estimate key rock material
types of stratified rock masses and the voids that can be present and rock mass properties needed for socket design. Then one
in certain types of volcanic rocks and in the carbonate rock fam- method for evaluating side resistance is presented, after a
ily. If these features are not defined with some confidence during detailed evaluation. Several possible approaches are suggested to
exploration, then it usually will be necessary to do so during con- assess the tip resistance, depending on the degree of geologic
struction. data available. Where available, statistics are given for the
Second, the socket must be constructed to give a nominally properties and the methods. Some design and construction
"clean" socket. The tip should be cleaned out as best as possible implications are noted as well. Detailed load-settlement-load
using conventional clean-out tools. Only in extraordinary cases transfer evaluations are beyond the scope of this paper.
should any special procedures be used. If side resistance is being
considered in the design, then the sides of the socket must be
clean as well, again using conventional tools. There is no need to REFERENCES
resort to special procedures for removing any light drilling muds,
because they will be displaced by proper tremie placement of Bell, A.L. 1915. Lateral Pressure & Resistance of Clay, & Sup-
high-slump concrete. However, in some softer or weathered porting Power of Clay Foundations. Minutes of Proc. of In-
rocks, for example compaction clay-shales, softening of the stitution of Civil Engineers: 199: 233-336.
socket side may occur. Special clean-out procedures and socket Bishnoi, B.L. 1968. Bearing Capacity of Closely Jointed Rock.
roughening or grooving may be considered in these cases. PhD Thesis. Atlanta: Georgia Institute of Technology.
Third is the issue of socket use and design. Where there is Carter, J.P. & Kulhawy, F.H. 1988. Analysis & Design of Drilled
rock surface uncertainty and therefore a need to ensure a quality Shaft Foundations Socketed into Rock. Report EL-5918. Palo
bearing surface, "seating" sockets can be considered. These types Alto: Electric Power Research Institute.
of sockets minimally penetrate the rock surface, usually to a Goodman, R.E. 1980. Introduction to Rock Mechanics. New
depth less than one socket diameter, and provide little, if any, York: Wiley
side resistance. For these sockets, only tip clean-out is necessary. Heuze, F.E. 1980. Scale Effects in Determination of Rock Mass
When the sockets are deeper, they will be "load-carrying" sock- Strength & Deformability. Rock Mechanics, 12(3-4): 176-92.
ets. These sockets can be used and designed in several ways. If Hirany, A. & Kulhawy, F.H. 1988. Conduct & Interpretation of
"tip resistance only" sockets are designed, then there is no ration- Load Tests on Drilled Shafts. Report EL-5915. Palo Alto:
ale to prescribe acceptance criteria for the rock along the socket Electric Power Research Institute.
sides. If "side resistance only" sockets are designed, then there is Kulhawy, F.H. 1978. Geomechanical Model for Rock Founda-
no rationale to prescribe acceptance criteria for the rock quality tion Settlement. J. Geotech. Eng. Div., ASCE, 104(2): 211-27.
beneath the socket tips. However, if both side and tip resistances Kulhawy, F.H. & Prakoso, W.A. 2001. Foundations in Carbonate
are included in the design, then acceptance criteria for both are Rocks & Karst. Foundations & Ground Improvement (GSP
appropriate. For the side, criteria sometimes are suggested that 113), Ed. T.L.Brandon. Reston: ASCE: 1-15.
relate to the percent of soil surface area or number of seams pre- Kulhawy, F.H. & Prakoso, W.A. 2003. Variability of Rock Index
sent along the surface of the socket. These may or may not be Properties. Proc. Soil & Rock America, Ed. P.J.Culligan et al.
realistic, depending on the actual geologic details. For the tip, Cambridge (MA): 2765-70.
criteria sometimes are suggested for probe holes drilled beneath Kulhawy, F.H., Prakoso, W.A., & Akbas, S.O. 2005. Evaluation
the tip to determine the frequency and thickness of soil seams in of Capacity of Rock Foundation Sockets. Proc. 40th U.S.
the rock within a depth beneath the tip equal to a shaft diameter Symp. Rock Mechanics, Ed. G.Chen et al. Anchorage: paper
or sometimes more. These types of criteria are based on settle- 05-767 CDROM.
ment limitations and must be evaluated as such, considering Ng, C.W.W., Yau, T.L., Li, J.H.M., & Tang, W.H. 2001. Side
stress distribution models for sockets in layered media. Guide- Resistance of Large Diameter Bored Piles Socketed into De-
lines are warranted to adjust these criteria if deepening is needed composed Rocks. J. Geotech. Eng., ASCE, 127(8): 642-57.
because the acceptance criteria are not met. In this case, more Prakoso, W.A. 2002. Reliability-Based Design of Foundations on
load is transferred through the socket side as the socket deepens. Rock for Transmission Line & Similar Structures. PhD
How the field acceptance criteria are implemented is an issue Thesis. Ithaca: Cornell University.
of potentially significant economic concern, in the same category Prakoso, W.A. & Kulhawy, F.H. 2002. Uncertainty in Capacity
as the evaluation of rock characteristics. If the rock characteris- Models for Foundations in Rock. Proc. 5th North Amer. Rock
tics are defined well during the investigation, then there should Mechanics Symp., Ed. R.Hammah et al. Toronto: 1241-48.
be few, if any, surprises during construction. If the investigation Prakoso, W.A. & Kulhawy, F.H. 2004a. Bearing Capacity of
is minimal, the opposite is likely. To minimize surprises, some Strip Footings on Jointed Rock Masses. J. Geotech. Eng.,
even note that probe holes be drilled prior to construction at each ASCE, 130(12): 1347-49.
shaft location to establish the final shaft depth before construc- Prakoso, W.A. & Kulhawy, F.H. 2004b. Variability of Rock
tion. This important economic issue needs to be assessed care- Mass Engineering Properties. Proc. 15th SE Asian Geotech.
fully and delineated clearly prior to construction. Conf.(1), Ed. S.Sambhandharaksa et al. Bangkok: 97-100.
Prakoso, W.A. & Kulhawy, F.H. 2006. Capacity of Foundations
on Discontinuous Rock. Proc. 41st U.S. Symp. Rock Mechan-
7 CONCLUDING COMMENTS ics, Ed. D.P.Yale et al. Golden: paper 06-972 CDROM.
Rowe, R.K. & Armitage, H.H. 1984. Design of Piles Socketed
Drilled foundations often are socketed into rock to increase the into Weak Rock, Report GEOT-11-84, London: University of
capacity. However, procedures to quantify the socket side and Western Ontario.
tip resistance vary considerably. In this paper, a critical
assessment is made of some key aspects of rock socket behavior.

You might also like