Professional Documents
Culture Documents
(Asce) 1090-0241 (2001) 127 1
(Asce) 1090-0241 (2001) 127 1
Michael W. O’Neill
It is my pleasure to introduce Dr. Michael W. O’Neill as the shaft foundations. His joint work with Professor Lymon Reese,
thirty-fourth Karl Terzaghi Lecturer. Dr. O’Neill’s talk is en- Construction and Design of Drilled Shafts, Second Edition,
titled ‘‘Side Resistance in Piles and Drilled Shafts.’’ This lec- was developed for the Federal Highway Administration. This
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 01/16/23. Copyright ASCE. For personal use only; all rights reserved.
ture deals with a subject on which Dr. O’Neill has become volume represents the state of the art and state of the practice
recognized as a leading expert. in drilled-shaft foundations. Of tremendous importance to rou-
Mike O’Neill was born in San Antonio, Texas, in 1940 and tine use of drilled-shaft foundations is the work for which
was educated at the University of Texas at Austin. He com- O’Neill is currently serving as Principal Investigator. This
pleted his BSCE in 1963 and proceeded directly to graduate FHWA Pool Fund Study addresses the important issue of the
school, where he was awarded a Master of Science in Civil use of nondestructive geophysical testing techniques for eval-
Engineering (Soil Mechanics) in 1964. After receipt of his uating the significance of minor defects in constructed drilled-
MSCE, he entered the U.S. Army as a Second Lieutenant and shaft foundations.
served as Environmental Engineer at Fort Bliss, Texas. He rose In addition to drilled shafts, O’Neill has conducted signifi-
to the rank of Captain and was discharged from military ser- cant research on continuous flight auger piles, driven open-
vice in 1967. At that time he returned to the University of ended piles, design of offshore piles, resistance factors for
Texas at Austin, where he proceeded to obtain a PhD in Civil driven piles, and many other related deep-foundation topics.
Engineering (Soil Mechanics) in 1970. For his research efforts he has received the ASCE Huber
After completing his graduate studies, Dr. O’Neill worked Research Prize, the ASCE State of the Art in Civil Engineering
for one year as a Research Associate at the Center for Highway Award, the ADSC Outstanding Service Award, and the
Research at the University of Texas. In 1971 he moved to ASCE Texas Section Hawley Award, and was cited by Texas
Houston and joined Southwestern Laboratories as the Manager DOT for one of the Top Ten Innovative Research Projects in
of Geotechnical Services, where he served until 1974. 2000.
In 1974 O’Neill joined the faculty of the University of In addition to his research, Dr. O’Neill is often called on to
Houston as Assistant Professor of Civil Engineering. In 1978 consult on large, important deep-foundation projects. His re-
he became an Associate Professor, and in 1984 Professor of cent projects include the Woodrow Wilson Replacement
Civil Engineering. He served as Chairman of the Department Bridge across the Potomac River outside Washington, D.C.;
of Civil and Environmental Engineering, John and Rebecca the H-3 Viaducts in Oahu, Hawaii; the TH-36 Bridge over the
Moores Professor, and Cullen Distinguished Professor. He cur- St. Croix River, in Stillwater, Minnesota; and the Fred Hart-
rently serves as Director of the National Geotechnical Exper- mann Bridge over the Houston Ship Channel, in Texas.
imentation Site at the University of Houston. For his many accomplishments and contributions to the state
Professor O’Neill has devoted most of his research career of the art in deep foundations, ASCE and the Geo-Institute
to advancing the state of the art of design and construction of have proudly selected Dr. Michael W. O’Neill as the thirty-
deep foundations, with much of this effort devoted to drilled fourth Karl Terzaghi Lecturer.
ABSTRACT: More than 20 years have passed since a Terzaghi Lecture focused on the topic of deep foundations.
However, considerable research has been performed, and experience gained, in this subject area in the intervening
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 01/16/23. Copyright ASCE. For personal use only; all rights reserved.
period. The objective of this paper is to update the earlier references on deep foundations by summarizing results
of important recent research on a few aspects of the topic of side resistance, most notably (1) driven piles in
saturated clay, (2) driven piles in siliceous sand loaded in compression and uplift, (3) drilled shafts in clay, and
(4) drilled shafts in soft rock. It is concluded that, while simple design relations are available for topic (1), much
is still to be learned. Under topic (2), the case is made that loading the pile in compression and uplift produces
different values of unit side-shearing resistance. Regarding topics (3) and (4), the effects of details related to
construction—such as stress relief, moisture migration from the concrete to the geomaterial, borehole roughness,
and borehole smear—are shown to be significant. The final point made is that the design of deep foundations
is a complex matter that should be addressed in a design context by engineers who are experienced in the
observation of pile behavior, theoretical modeling, and the appropriate use of design methods.
Analytical Modeling
We obviously quickly reach a limit to the usefulness of field
observations alone. Further progress can be made through an-
alytical modeling, however. The first efforts at purely analyt-
ical solutions were sponsored by the oil industry to predict
side resistance of very long piles in clay, for which very little
empirical data existed in the 1970s. The initial research, which
resulted in a process termed the ‘‘general effective stress
method’’ (GESM) (Kirby et al. 1983), required knowledge of
the in situ stress state as well as several soil properties. Mod-
eling then progressed to the effect of cylindrical pile installa-
tion on the effective pressures in the soil mass around the pile,
the effective stresses after consolidation, and finally, the effect
of loading on fmax. Most subsequent analytical models have
gone through the same steps, with variations.
Installation was modeled in the GESM by the undrained
plane strain cavity expansion theory, symbolized in Fig. 7. The FIG. 7. Stresses Predicted by Expanding Cylindrical Cavity
mean effective stress in the plastic (gray) zone was equated to Theory in Undrained Clay Soil with Mohr-Coulomb Characteris-
the mean effective stress at the critical state at the void ratio tics
encountered in the free field, which was then reduced by a
factor of 0.6 to account for the extreme remolding that occurs
in driving displacement piles. The mean total stress came di-
rectly from the von Mises model depicted in Fig. 7 and was
dependent on the Young’s modulus and undrained shear
strength of the soil. We of course have uncertainties about
which measurements of su and Eu should be used. There are
obviously many possibilities, which points toward the need for
site- or formation-specific calibration of the model.
Radial consolidation was then modeled by considering the
framework of the soil in the plastic zone to be elastic, com-
pressible, and pointwise isotropic (Leifer et al. 1979). The
Young’s modulus of the soil varied in a radially nonlinear
manner from a small magnitude at the pile surface to its un-
disturbed value at the outer edge of the plastic zone. If a con-
stant, isotropic value of Young’s modulus in the plastic zone
is used during consolidation, we return to Terzaghi’s one-di-
mensional consolidation theory and are forced to conclude that
the change in effective normal stress on the face of the pile
during consolidation is equal to the initial excess pore water
pressure. Observations tell us that this does not happen—the FIG. 8. Predictions of fmax by GESM (after Esrig and Kirby
effective stress changes are not that large, so, as a minimum, 1979)
models were also developed, such as CAMFE (Randolph et driven displacement piles in saturated clays with various min-
al. 1979) and CASH (Matlock et al. 1982), which had some- eralogies and stress histories are summarized in Table 1. The
what different attributes than the GESM but which had the model appears to lead to reasonable results over a wide spec-
same objectives. I am personally more familiar with the trum of soils, even though it contains a degree of empiricism.
‘‘VECONS/AXIPLN’’ model, developed by Heydinger (Hey- A significant further advance came with the coupling of the
dinger 1982; Heydinger and O’Neill 1986). It is similar to the strain path method (Baligh 1985) and the MIT-E3 constitutive
GESM, but it empirically considers the soil in the plastic zone model (Whittle 1987) in the late 1980s (Azzouz et al. 1990).
to have anisotropic stiffness during consolidation, with a ratio The strain path method offers a better solution for strain
of tangential modulus to radial modulus of about 2. Otherwise, changes in soil around a pile as it penetrates by using the
the solution of Leifer, used in the GESM, was used to compute analogy of a viscous, irrotational fluid flowing past a fixed
the effective stresses at the end of consolidation. object—very different from a cylindrical cavity that magically
Fig. 9 shows the conditions after installation, which repre- expands. The MIT-E3 constitutive model tracks those strains
sent the initial conditions for consolidation modeling. The sub- by assuming that the soil is completely nonlinear. Irrecovera-
script ‘‘cs’’ indicates ‘‘critical state.’’ Fig. 10 is a typical result ble strains are generated when stresses are reversed during
from VECONS for the Houston overconsolidated clay site. For installation, as illustrated in Fig. 11. Such behavior is not rec-
reference, the ambient vertical and horizontal effective stresses ognized in the expanding cavity model. The stress-strain be-
( ⬘vo and ho
⬘ , respectively) are indicated on the right axis. Al- havior of the soil framework is then tracked as it consolidates
though the tangential effective stress ( ⬘) in the soil adjacent around the pile and during loading in a finite-element envi-
to the interface drops after driving, significant increases are ronment. There is no need to assume variations in Young’s
predicted in all three major principal effective stresses near the moduli, as with the previous models, and the model inputs can
interface during consolidation. The largest increase between all be developed from high-level laboratory tests.
driving and end-of-consolidation occurred in the tangential ef- This model predicts  of 0.10 for normally consolidated
fective stress because more volume change occurs near the Boston Blue Clay to 0.21 for lightly overconsolidated BBC,
interface where the greatest pore pressure dissipation takes which are values generally consistent with the trend reported
place, leading to arching in the tangential direction. Here the
radial effective stress against the pile wall, r⬘ at r/Ro = 1, is
about 2.5 ⬘ho, which is consistent with observations made near
the middepths of the test piles described in Fig. 3 (Kc /Ko).
After determining these stress relations, fmax and correspond-
Complicating Factors
While our ability to predict fmax has greatly improved since
the 1970s, at least at the highest level of analytical effort, it
is important to realize that the design value of unit side resis-
tance fult may not be equal to fmax, the peak value, because of
deflection softening and progressive failure as a compressible
pile is loaded. These effects can be suitably modeled using the
‘‘t-z’’ method of Coyle and Reese (1966) or the ‘‘3-inverse’’
method of Murff (1980).
A more subtle explanation of why fult is not always equal to
fmax along the pile may concern the installation process (Aurora
et al. 1980). Fig. 13 summarizes the results of several com-
pression loading tests on closed-toe steel pipe piles in uniform,
medium stiff clays that were compacted wet of optimum in a
calibration chamber. The Ip of the clay soils varied from 23 to
37, and su (remolded) was 10–21 kPa in CKoU triaxial com-
FIG. 12. Range of Measurements and Predictions of fmax via
pression with v⬘ = 21 kPa. The pile was driven to a shallow
Strain Path/MIT E-3 Model for Driven Pipe Piles Loaded in Uplift depth, load tested to plunging failure, driven deeper, load
at NGI Haga Test Facility (after Whittle 1991) tested to plunging failure, and so forth. Whatever the depth of
penetration, fmax in the upper 30% of the pile was lower than
in the lower 70%. This effect could possibly be attributed to
in Fig. 6 for soils identified as normally consolidated. It also extreme destructuring of the soil at the pile-soil interface due
reveals that radial consolidation of soil with a residual vertical to large accumulated shear deformations; to reduced effective
shear stress resulting from the driving operation will produce stresses at the interface from effects of lateral expansion of
a higher value of fmax than without a vertical shear stress if the pile upon the passage of longitudinal stress waves during
subsequent loading does not reverse the shear stress direction. impact driving and the inability of the soil to rebound laterally
Whittle (1991), using this model, analyzed a series of 16 back against the pile; to the lateral motion of the pile due to
uplift tests performed on identical pipe piles (which were flexural waves during driving; or to a combination of factors.
jacked into place, not driven) at the NGI Haga test facility, in These data suggest that Rs, the shaft resistance, is about 15%
highly overconsolidated, crustal soil, considering residual less than one would obtain by integrating fmax over the perim-
stresses. A comparison of the measurements of peak unit side eter surface of the pile in this plastic clay.
resistance (including residual stresses) versus depth for the 16
piles and the range of predictions from the MIT-E3/Strain Path Unsolved Mysteries
model for the central parts of the piles are shown in Fig. 12.
Predictions for the heads and toes of the piles are omitted Although the preceding discussion suggests that the problem
because the model that was used did not include end effects. of side resistance in driven piles in clay is essentially a ‘‘solved
Also shown in Fig. 12 are the predictions made by using Mey- problem,’’ there is evidence to the contrary. Consider the case
erhof’s version of the  method, which in this case greatly of a series of uplift load tests conducted on a steel pipe pile
overpredicts the side resistance. The use of the completely at West Delta Block 58A offshore from the Mississippi River
analytical model, in this case, clearly provided more accurate delta (Audibert and Hamilton 1998; Bogard and Matlock
predictions. The strain path model has recently been improved 1998a,b). The test pile was 0.76 m in diameter and about 70
to include the effects of a free surface (Whittle 1999), and m long, and the soil was a normally consolidated—perhaps
several similar models that include pore fluid dynamics and slightly underconsolidated—plastic clay. The pile was tested
variable yield surfaces have also been proposed (Wathugala to failure at intervals over a period of about two-and-a-half
and Desai 1989). years after it was initially driven. Furthermore, effective nor-
8 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2001
FIG. 14. Measured Values of [fmax and Kc versus Depth for Test Pile at West Delta 58A (Compiled from Bogard and Matlock 1998a,b)
mal stresses were measured along the face of the pile during
each load test.
Plotted in Fig. 14 are measured relationships between Kc (at
pile-soil slip) and depth for each load test. Kc clearly decreased
with time. On the other hand, the side resistance generally
increased each time the pile was loaded, despite the fact that
radial effective stresses at the pile-soil interface were decreas-
ing. Either the theory of effective stress did not apply, the
measurements of effective stresses against the pile wall (or
measurements of shear stress) were not appropriate, or there
is some other explanation for this unusual behavior. Jardine
and Saldivar (1999) offer several plausible reasons for the re-
ported behavior. Another is offered below.
The pile was extracted and examined after the final load
test. A layer of hard clay about 75 mm thick was attached to
the side of the pile. When the attached soil was dissected,
evidence was found of several microshear surfaces, at varying
distances from the face of the pile. These apparently repre- FIG. 15. Interpreted Relations of ␣ to su /OCR from ‘‘Universal’’
sented portions of the shearing surfaces that were produced by Database (Modified from Kolk and van der Velde 1996)
the successive loadings. The soil nearest the pile apparently
had hardened appreciably with time, forcing the failure surface
to migrate farther and farther into the soil mass. Indeed, if this model. But such modeling is justified only on projects near
happened, the effective stress measurements at the interface the outposts of our experience. For most projects, we have to
became irrelevant because failure was occurring elsewhere use something more tractable, considering the resources we
along a radially expanding shear surface, which produced have at our disposal. This often means reverting to a correla-
higher and higher overall resistances. tive solution, such as the ␣ method.
X-rays of the removed soil revealed zones of very dense The database for side resistance of piles in clay originally
soil, mostly near the interface. Despite being hard, the soil near developed for the API (Dennis and Olson 1983; Olson 1984),
the visible microshear surfaces had higher moisture contents and added to subsequently, has been catalogued intelligently
than the masses of soil between them. The increased density and includes correlations between undrained shear strength
was associated with the precipitation of iron. Where the iron and fmax, the so-called ␣ factor.
came from is a mystery, but it is quite possible that it was After analyzing this database, Semple and Rigden (1984)
removed from the steel pile and deposited in the adjacent soil proposed that ␣ be related to the ratio of su to ambient vertical
through the action of anaerobic, sulfate-reducing bacteria that effective stress, ⬘vo, which is an indirect measure of OCR.
were present in the organic soil. The iron may have become Semple and Rigden’s proposed relation is shown in Fig. 15,
a cementing agent, forcing the outward migration of the failure which also shows the data points in the database, compiled
surface. This is probably good news if only we knew how to mostly from load tests on steel pipe piles. The GESM, for
predict the phenomenon. Much still needs to be done in un- example, predicts that ␣( fmax /su) decreases as OCR increases,
derstanding chemical and biochemical interactions between so Semple and Rigden’s interpretation has a theoretical basis.
piles and soil. Here, fmax is the average unit side resistance at the load inter-
preted to correspond to failure of the pile; therefore, it may
Practical Design Guidelines include effects such as are shown in Fig. 13. Fig. 15 does not
take into explicit account the differences in the effects of shear
The ideal approach to design would be to model in detail and normal stress reversals during installation, nonlinear
every pile that we install in a clay profile with a model such stress-strain behavior, degree of anisotropy, rate of loading,
as VECONS, Strain Path/MIT-E3, or some other appropriate and other factors that theoretical models demonstrated are im-
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2001 / 9
Observations
tests at a normally consolidated silty clay site called ‘‘Pentre,’’
in the U.K., where ␣ was very low, especially for a pile driven Some foundation engineers, mostly through observations of
in two stages, with a long time hold between stages. The clay universal databases, have concluded that the magnitude of unit
at Pentre is characterized as normally consolidated, and it is side resistance is independent of the direction of loading. Oth-
silty (Gibbs et al. 1993). Kolk and van der Velde gave the ers cite evidence to the contrary. For example, when the pile
two-stage pile a low weight when constructing the design is loaded in uplift, there is a Poisson’s effect in the pile that
curve because the pile had been driven such that there was will likely result in a reduction in radial effective stress at the
considerable pore pressure dissipation between stages. All tests interface. Uplift loading also produces a net reduction in ver-
on plugged (displacement) piles in the silty clay at Pentre, tical effective stress in the soil around the pile, which re-
however, exhibited lower ␣ factors that have been observed at duces the soil’s shear strength. Consider Fig. 16, which shows
sites where the clays were more plastic (Ip > 20). Although the measurements of vertical effective stress changes in the soil
phenomenon is not clearly understood, at least some clays with near a relatively rigid, open-toed, steel pipe pile that was
high silt contents appear to have very high unloading moduli, driven into dense sand confined in a calibration chamber and
such that when they are thrust slightly away from the pile loaded axially (O’Neill and Raines 1991). The large, gray cir-
during the passage of the stress wave, they do not rebound cles indicate the beginnings and ends of load tests, first in
back against the pile as strongly as plastic clays (Karlsrud compression and then in uplift. The vertically oriented earth
1999). Organics in the soil may have a similar effect. There- pressure cell clearly indicates an increase in vertical effective
fore, it appears appropriate to reduce the ␣ factor for silty clay, stress as the pile is loaded in compression and a decrease as
clayey silt, and organic soil sites to values below those on all it is loaded in uplift, both because of load transfer from the
of the design curves in Fig. 15. The task of the foundation pile. In Fig. 16, the term ‘‘fa’’ is the average unit side shear
designer then shifts to determining the stress history of the transferred from the pile to the soil above the level at which
deposit, its silt (and possibly organics) content, and its un- the measurements were made. At the depth of the measure-
drained shear strength. This task must be performed accurately, ment point, the increase in ⬘v is less than 10% of the average
because ␣ changes rapidly as su /⬘vo increases above about side resistance above that depth; however, the increase or de-
0.25.
These correlative methods apply to piles with length (pen-
etration)-to-diameter (L/B) ratios of about 50 or less. Each
recognizes that lengthening the pile reduces fmax to fult due to
progressive failure. How this reduction is made varies from
method to method. A 30% reduction in ␣ is proposed by Sem-
ple and Rigden for piles with L/B = 120 (relative to ␣ for
L/B ⱕ 50, Fig. 15). Kolk and van der Velde recommend the
use of (5), involving relative penetration (L/B) and implied
OCR, for piles with L/B > 50:
fult = [0.5(L/B)]⫺0.2[su /⬘vo]⫺0.3 (5)
It should be clearly understood that the method used to
quantify undrained shear strength is not arbitrary. Use of su
data should follow the same order of priority as was used by
the developers of the database, which is (1) UU triaxial com-
pression tests, (2) unconfined compression tests, (3) field vane
shear tests, and (4) results of any other undrained tests that
are available. This suggests that success in design is related to
using tests for estimating su that are as high up on the priority
ladder as possible. For the first two means of obtaining un-
drained shear strength, the tested soil samples should be clas-
sified as ‘‘undisturbed’’ (taken with thin-walled samplers with
area ratios of 10% of less than inside clearance ratios of 1 to
3%).
where
= pile(tan ␦) 冉 冊冉 冊
L
B
Gsoil avg
Epile
(7)
FIG. 18. Effect of Borehole Excavation on Low-Strain Shear
The evidence is compelling that when one designs a pile in Modulus for Drilled Shaft at Houston Site (Overconsolidated
uncemented, siliceous sand that will be loaded in uplift, seri- Clay) (Modified after Kalinski and Stoke 1998)
Analytical Modeling
TABLE 2. Statistical Summaries of ␣ and  for Drilled Shafts
Researchers have not pursued the analysis of the effect of in Cohesive Soil from ‘‘Universal’’ Data Bases
installation on side resistance in drilled shafts with the rigor
Coefficient of
that the analysis of driven displacement piles have been pur- Mean measured Rs / variation (COV) Number
sued. Since the relation of fmax to su depends on many factors Method mean computed Rs in parameter of tests
that are construction dependent, some of which were just dem- (1) (2) (3) (4)
onstrated, we are left with correlative methods for routine de-
␣ 1.07 0.36 43
sign, most of which are again based on the familiar ␣ factor  1.08 0.37 140
concept.
12 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2001
FIG. 23. Predicted Side Resistance-Movement Relations for Drilled Shaft Socket with Various Interface Conditions (O’Neill and Has-
san 1994)
Studies such as this demonstrate the critical nature of the fmax /pa = ⍀(qu /2pa)0.5 (11)
construction details and show how closely these details are tied
to design assumptions. Designs can be more efficient, for ex- where, for qu > 2.5 pa, ⍀ = 1 for sockets judged to be smooth,
ample, if grooving due to natural drilling operations or with 2 for sockets with average roughness, and 3 for sockets judged
the use of an artificial tool can be assured to produce an in- to be rough, including artificially roughened sockets. O’Neill
terface pattern similar to that shown in Fig. 21 and if control et al. (1996) review procedures for judging whether a socket
of construction operations ensure that smear does not accu- is rough. Caution should be used in applying these results to
mulate at the interface. very large diameter sockets, which were largely missing from
For drilled shafts in soft rock, particularly IGMs that are the database from which (11) was derived.
formed from weathering of sound rock, the effects of discon-
CLOSING REMARKS
tinuities on fmax must be considered because they affect the
percentage of the interface that is in contact with sound rock Computing pile resistances, even in the simple monotonic
and they limit the magnitude of the normal stresses produced axial loading case considered here, requires a keen knowledge
by interface dilation. O’Neill (1999) described a case history of the effects of fundamental geotechnical and construction
of very chaotic soft rock in which these issues were dealt with. phenomenon and of the past performance of piles in geologic
The use of sinusoidal roughness patterns of relatively long formations similar to that for which the pile is to be designed.
wave length (e.g., Fig. 21) tend to predict more settlement of While much is understood, much still remains to be learned,
the drilled shaft than actually occurs. Improvements in mod- particularly concerning the effects of construction on the be-
eling the effects of interface roughness for design purposes havior of individual piles. These effects include such factors
have been incorporated in a relatively new model as the installation method (impact driving, jacking, vibrodriv-
(‘‘ROCKET’’) in which asperity shapes have been assigned ing, rate of driving), lateral movements of piles during driving,
using fractal theory to match, in general, standard joint rough- and shear drag in layered soils in driven piles. They also in-
ness patterns observed in rock (Seidel 1998). This model in- clude borehole roughness, water content of fluid concrete, time
dicates that interface dilation has a very strong effect on fmax, required for excavation (stress relief ), impact of the details of
which, in sound, unsmeared rock, decreases dramatically as drilling slurry, and the effect of drilling tools and practices on
the radius of the socket increases, partially because radial development of rock smear. Effects of construction procedures
strain in the rock due to dilation is inversely proportional to for pile groups, such as order and rapidity of pile installation
the socket diameter in an elastic system. For example, in a (which were not subjects of this paper) are equally important
sound, unsmeared rock with qu of 3 MPa, ROCKET indicates in the larger subject area of deep foundation systems. Inade-
a reduction in fmax from 0.72 to 0.30 MPa when the socket quately quantified effects of construction lead to the inevitable
diameter is increased from 350 to 2,000 mm (Baycan 1996). conclusion reached by Terzaghi that site-specific load testing
In most rock formations the effect of socket diameter will not must remain an integral part of the design process for driven
be this great, because of softening the rock asperities during piles and drilled shafts.
drilling. However, this phenomenon must be considered when From the perspective of engineering practice, however, in
deciding upon the scale at which to test a rock socket to arrive today’s bottom-line-driven environment, many owners view
at design parameters for a specific project. the need to acquire adequate and appropriate subsurface in-
formation, understand the relevance of load test databases to
Partial Design Guidelines the job at hand, load test piles at specific sites where war-
ranted, and apply available analytical methods as being too
When a local load test database does not exist from which complex, too expensive. Thus, we are told to use a simplified
to determine side resistance and when resources do not permit design method and get on with being ‘‘productive.’’
advance modeling, (11), developed from a database analysis Use of information such as is covered here, and the expe-
by Kulhawy and Phoon (1993), can be used with an appro- rience that is implied to use it correctly, requires time and
priate safety margin: effort by individuals who are dedicated to excellence in foun-
14 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2001