Vous êtes sur la page 1sur 13

753

Field P–y curves in weathered rock


Kook Hwan Cho, Mohammed A. Gabr, Shane Clark, and Roy H. Borden

Abstract: In weathered and decomposed rock profiles, the lack of an acceptable analysis procedure for estimating lat-
eral load–displacement response of drilled shafts is compounded by the unavailability of weathered material properties,
including the material’s lateral subgrade reaction modulus. Such deficiency often leads to the overdesign of the drilled
shaft foundation. Six field tests were conducted on drilled shafts to investigate the shape and magnitude of P–y curves
in weathered rock material at three locations in North Carolina. The tested shafts were instrumented using dial gages,
strain gages, and continuous vertical inclinometers. The measured load versus deflection data are used to study the
stiffness response of weathered rock. Measured lateral responses are compared with the results estimated based on a
“weak rock” model and a stiff clay model. The comparison shows that Reese’s weak rock model overestimated the
resistances of the tested shafts while the stiff clay model consistently underestimated the measured shaft resistances.
The measured and computed results are analyzed and discussed.
Key words: drilled shaft, weathered rock P–y curve, subgrade modulus, ultimate resistances in weathered rock, verifica-
tion tests.
Résumé : Dans les profils de roches décomposées et altérées, l’absence d’une procédure d’analyse pour évaluer la
réaction en déplacement sous une charge latérale de puits forés est compliquée par l’ignorance des propriétés du maté-
riau altéré, incluant le module de réaction latérale du matériau. Une telle carence conduit souvent à une conception
trop sécuritaire de la fondation du puits foré. Six essais sur le terrain ont été réalisés sur des puits forés pour étudier la
forme et la grandeur des courbes P–y pour le matériau rocheux altéré sur trois localisations en Caroline du Nord. Les
puits testés ont été instrumentés au moyen de jauges à cadran, de jauges de déformation, et d’inclinomètres continus
verticaux. Les données de la charge mesurée en fonction de la déformation sont utilisées pour étudier la réponse en ri-
gidité de la roche altérée. Les réponses latérales mesurées sont comparées aux résultats estimés basés sur un modèle de
« roche molle » et un modèle d’argile raide. La comparaison montre que le modèle de roche molle de Reese a sures-
timé les résistances des puits testés alors que le modèle d’argile raide a sous-estimé de façon systématique les résistan-
ces des puits mesurées. On analyse et discute les résultats mesurés et calculés.
Mots-clés : puits foré, courbe P–y de roche altérée, module de réaction, résistances dans la roche altérée, essais de vé-
rification.
[Traduit par la Rédaction] Cho et al. 764

Introduction tensive weathering of the parent rock. With depth, the soils
grade into less weathered material and more evidence of the
In locations where geologic discontinuities have resulted parent rock features are retained. Quantitative definitions of
in relatively soft soils overlying massive hard rock, the ge- the soil–rock interface have been addressed in the literature.
ometry of the soil–rock boundary can be reasonably defined Coates (1970) recommended that the rock quality designa-
with existing subsurface exploratory techniques. In areas of tion (RQD) value could be used to estimate the depth of
weathered and decomposed rock profiles, however, defini- sound rock. RQD values less than 25% designate very poor
tion of the soil–rock boundary is a recurring challenge for rock quality that could be classified as soil for engineering
engineers and contractors. In this situation, the subsurface
purposes. Peck (1976) stated that the distinction between
conditions typically consist of surface soils derived from ex-
rock-like and soil-like material in transition zones is usually
unpredictable.
Received 22 June 2005. Accepted 15 September 2006. In these types of transitional subsurface profiles, the defi-
Published on the NRC Research Press Web site at cgj.nrc.ca nition of soil parameters needed for the analysis and design
on 31 July 2007. of laterally loaded drilled shafts poses a great challenge. The
K.H. Cho,1 M.A. Gabr,2 S. Clark,3 and R.H. Borden. lack of an acceptable analysis procedure in such profiles is
Department of Civil Engineering, North Carolina State compounded by the unavailability of means for evaluating the
University, Raleigh, NC 27695-7908, USA. weathered material properties, including the lateral modulus.
1
Such deficiencies often lead to the overdesign (conservative)
Present address: Seoul National University of Technology, of the shaft foundation.
172 Gongneung2-dong, Nowon-gu, Seoul, 139-743, Republic
of Korea. Previous work on displacement-based analysis of drilled
2
Corresponding author (e-mail: gabr@eos.ncsu.edu). shafts in weathered rock profiles is scarce. Notable studies
3
Present address: Western Regional Design Group, 5253 Z recently reported in the literature include work by Zhang et
Max Boulevard, Harrisburg, NC 28075, USA. al. (2000), Reese (1997), and Digioia and Rojas-Gonzalez

Can. Geotech. J. 44: 753–764 (2007) doi:10.1139/T07-026 © 2007 NRC Canada


754 Can. Geotech. J. Vol. 44, 2007

(1994). Zhang et al. (2000) considered the nonlinear behavior Table 1. List of test sites and rock types.
of soil and rock by assuming both to be elastic – perfectly
Test site Rock type
plastic materials. Reese (1997) utilized the P–y concept for
the analysis of a single pile in “weak” rock profile. The Nash County Meta-argillite
method was termed “interim” principally because of the Caldwell County Gneiss
dearth of load test data available to validate the proposed de- Wilson County Crystalline granite
sign equations. Digioia and Rojas-Gonzalez (1994, p. 1575)
performed seven tests on drilled shafts supporting transmis-
sion towers in rock profiles and reported the applicability of nonlinear part of shaft load–deflection response. This was done
their design model (MFAD) in predicting the measured field with the knowledge that the loading frame had a FS of 1.25.
behavior. They concluded that “classical methods for pre- Test load increments were applied using a hydraulic jack,
dicting the load–deflection relationship for drilled shafts in which has an approximately 10 inch (254 mm) stroke. The
soil consistently over-predict drilled shaft deflection.” They hydraulic jack was fitted with a pressure gage. This gage was
also stated that additional research is necessary to assist the calibrated to indicate the magnitude of the applied load. In
designer with various rock types. From an engineering per- addition, a load cell was used to monitor the induced loads.
spective, the distinction between transitional material and The loading sequence consisted of applying the lateral load
rock is important to understand the behavior of the drilled in increments of 45–90 kN, followed by an unloading in
shaft foundation. The work presented herein includes results cycles. Each load was held until no further appreciable deflec-
of field load tests performed at three sites in North Carolina tion at the tops of the test shafts (defined as <0.127 mm/h) was
where residual soil profiles were encountered. The shafts measured.
were embedded mostly in weathered and decomposed rock Data collected during the field tests include the following:
materials. The results include measured displacement re- (i) load–deflection measurements at the top of the shaft;
sponse with depth as well as subgrade reaction values back- (ii) deflection versus depth profiles measured by continuous
calcuated from the field strain measurements. A comparison inclinometer probes; and (iii) strain with depth using vibrat-
of results with data from models reported in the literature is ing wire (VW) strain gages mounted on sister bars that were
presented and discussed. attached to the shaft reinforcement using steel ties. From the
measured strains, moments were calculated along the depth
of the shafts by piece-wise numerical integration, and then
Field load testing the soil reaction was back-calculated based on the estimated
moment distribution.
Six tests were performed in three different counties in
North Carolina. Two load tests were performed at each of
Instrumentation plan
the three test sites in Nash, Caldwell, and Wilson counties.
Table 1 presents the underlying rock types at each of the test Shaft strains and displacements were monitored during
sites. field testing with dial gages, strain gages, and slope incli-
At each site, two 0.762 m diameter shafts were constructed nometer probes. The positions of these measuring systems
approximately 7.6 m apart. Figure 1 shows the general layout are schematically illustrated in Fig. 1 for the dial gages
of the test shafts. The shafts were drilled using conventional and the strain gages. Each shaft was instrumented above
earth augers, preceded by the insertion (screwed in) of perma- ground with four dial gages to measure surface displace-
nent casing to the tip. The 12.7 mm thick permanent casings ment. A fixed reference beam was used for mounting the
were used to make the test shafts stiffer to induce displace- dial gages in accordance with section 5.1.1 of ASTM
ment around the tip area embedded in the transitional mate- D3966-90 (ASTM 1995). Two dial gages were used to
rial. A small gap between rock and casing existed in one case measure shaft vertical deflection. One dial gage was used
during construction when the permanent casing was installed to measure lateral movement parallel to the direction of
even though the casing was carefully installed by a drilled loading, and the other one was used to measure movement
shafts contractor. The no solid contact problem was inevita- perpendicular to loading direction.
ble, however, but after loads were applied to the shafts, solid VW strain gages were attached to the rebar cage along the
contact was obtained within 10 mm of lateral deformation as shaft length using sister bars (tied to the vertical shaft rein-
will be described later in the Results section. The steel rein- forcing bars as previously described). These strain gages
forcement cage, with the sister bars attached, was lowered were placed at the elevations shown in Fig. 1 for the Nash
into the hole with spacers to ensure proper positioning. The County tests. A CR-10 data logger was used to electroni-
drilled hole was filled with concrete. Figure 1 also shows the cally acquire strain and temperature data.
overlay of the test shafts and the load frame used to apply the Slope inclinometer probes were used to continuously
lateral load to both shafts simultaneously. The load frame was measure shaft lateral inclination as a function of depth.
attached to the constructed shafts at a vertical distance of Electrolytic vertical in-place inclinometer probes were in-
0.3 m above the excavated ground line. The maximum capac- serted into a plastic inclinometer casing installed during
ity of the load frame, including a FS of 1.25, was 979 kN for shaft construction. This plastic inclinometer casing was
the Nash County tests. This capacity was increased, through tied to the rebar cage prior to construction of the shafts.
structural frame modifications, to 1334 kN for the Caldwell The inclinometer instrument consisted of a continuous chain
and Wilson counties tests. However, during the last two tests of probes (sensors with wheels) that are attached to each
at Wilson County, higher load (1680 kN) than the maximum other at pivot points approximately 0.50 m apart. These
design load was applied to obtain as much as possible of the chains remained inside the casing throughout testing and
© 2007 NRC Canada
Cho et al. 755

Fig. 1. Layout of the test shafts at the Nash County test site.

were used to collect data along the entire length of each large enough lateral displacement to define the lateral sub-
shaft. A signal cable extended up through the casing for grade modulus at the depth of the weathered rock layers,
each sensor, which was then connected to the CR-10 data while the long shaft would posses enough capacity to act as
acquisition system. The data acquisition system consisted a reaction member. At the Nash County test site, prior to
of an A/D (analog to digital) board and control system run construction of the shafts, the test area, measuring roughly
by a computer program developed by Slope Indicator Co. 10.67 m × 3.05 m, was excavated by removing 0.6–0.9 m of
soil. This excavation eliminated some of the overburden soil
and enabled the applied loads from the frame to be closer to
Test setting the weathered rock elevation, thereby inducing movement in
For each test site, both short and long shafts were con- the subsurface layer of interest. The overburden soils were
structed apart from one another, as schematically illustrated entirely removed for the Caldwell and Wilson counties tests.
in Fig. 1. The logic of pushing a short against a long shaft The length of each shaft and the number of strain gages in-
during testing was to ensure that the short shaft would incur stalled in each shaft are summarized in Table 2.
© 2007 NRC Canada
756 Can. Geotech. J. Vol. 44, 2007

Table 2. Summary of shaft length and number of strain gages.


Nash County test Caldwell County test Wilson County test
Short shaft Long shaft Short shaft Long shaft Short shaft Long shaft
Shaft length (m) 3.35 4.57 4.00 4.80 4.85 5.71
No. of strain gages 8.00 9.00 7.00 9.00 7.00 9.00

Fig. 2. Test area subsurface cross-section, Nash County site.

Test site characteristics core was poor RQD rock, which can be classified as soft
weathered rock.
Nash County Hard rock core recoveries exceeded 95%, and RQD values
The test site is located in gently rolling terrain along for the lower 4.57 m exceeded 75%. The location of the tip of
the easterly edge of the Piedmont physiographic province. the long shaft was approximately at elevation 36 m, near the
Metamorphosed mudstone, siltstone, and sandstone of the East- bottom of the soft weathered rock zone. All weathered rock
ern slate belt underlie the area. The residual soils in this area core samples were inspected. Eight samples were chosen for
are mostly sandy silt (ML) and silty clay (CL), and are stiff to unconfined compression testing at NCDOT (North Carolina
hard. The water table is located approximately 2 m below the Department of Transportation) Materials and Tests Unit Soils
ground surface. Residual soils grade with depth from soft (soil- Laboratory. Table 3 summarizes the results of this testing.
like) to hard (rock-like) weathered rock, which is derived from
the underlying meta-argillite (refer to the cross-section in Fig. 2). Caldwell County
Two standard penetration test profiles were performed near Alluvium, sandy saprolite, weathered rock, and hard rock
the test area. The site’s alluvial soils consist of stiff silty and comprised the foundation materials encountered in the bor-
sandy clay and soft to very stiff sandy silt. Just below the al- ings at the Caldwell County test site. The test area is under-
luvial layers is a thin residual, stiff to hard, silty clay. The lain by a Cenozoic-age biotite gneiss and schist rock unit of
groundwater was present at the interface of the alluvial soil the Inner Piedmont belt. Core borings revealed that, locally,
and residual layers. Beneath the residual clay, a soft weath- much of the rock is granetiferous. Tan to brown, medium
ered meta-argillite rock grades into a hard weathered rock, dense silty to fine coarse sand (micaceous residual material)
and finally into a competent rock around a depth of 6.10 m. exists over the weathered rock layer. Figure 3 shows the
The residual soils were cored from a depth of 2.44 m to cross-section profile at the Caldwell County test site.
approximately 7.0 m below the surface. The core was “H” Two boring logs were performed in the vicinity of the test
size to increase the potential of sample recovery. Most of the shafts. Data from the boring logs at the locations of the long

© 2007 NRC Canada


Cho et al. 757

Fig. 3. Test area subsurface cross-section, Caldwell County site.

and short shafts, respectively, indicated profiles to be nearly Table 3. Nash County laboratory test results.
identical. No groundwater was encountered during
subsurface investigation. Beneath the residual soil is a Unconfined
weathered gneiss, which grades into a hard weathered rock Unit weight compressive strength,
and finally into a competent high RQD quality rock around Depth (m) (kN/m3) qu (kPa) RQD (%)
a depth of 10.7 m below the ground surface. The core size 4.58–4.76 24.6 33 095 <25
was an “H,” and RQD values ranged from 12% to approxi- 5.27–5.57 26.1 19 305 <25
mately 65%. 5.62–5.92 22.4 31 026 <25
All weathered rock core samples were inspected; however, 6.15–6.35 26.1 126 864 50
few samples were chosen for lab testing. Only two samples 6.64–6.80 25.5 48 263 50
were tested at the Materials and Test Unit Soils Laboratory 6.80–7.00 25.4 55 158 50
owing to poor rock quality. Table 4 summarizes the lab test- 10.42–10.61 27.0 154 443 85
ing results. 12.71–12.94 26.4 135 827 98
14.35–14.58 26.4 50 332 100
Wilson County
Tan brown fine to coarse sand, soft and hard weathered long shaft locations, respectively. Table 5 summarizes the
crystalline rock, and hard rock comprised the foundation lab testing results. The sample from the short shaft location
materials encountered at this site as shown in Fig. 4. Allu- has a higher RQD value (59%) than the relatively low RQD
vial material occurs to a variable extent at the site. Rock was of 13% estimated at the location of the long shaft. This ob-
cored at the two boring locations. Beneath the alluvium soil, servation is manifested in the measured load response to be
the weathered crystalline rock graded into a competent high- presented later in the paper. The groundwater was encoun-
RQD rock around a depth of 9.1 m. This test site provided a tered near the ground surface during the investigation.
subsurface profile for the short shaft that was different from
that for the long shaft. Load test results
The residual soil layer existed from the surface to approx-
imately 2.4 m deep. All weathered rock core samples were Nash County tests
inspected, and specific samples were chosen for lab testing. During testing, the short shaft experienced nearly 0.137 m
Two samples were evaluated, one each from the short and of displacement at an applied lateral load of 534 kN. After

© 2007 NRC Canada


758 Can. Geotech. J. Vol. 44, 2007

Table 4. Caldwell County laboratory test results.


Unconfined compressive
Depth (m) Unit weight (kN/m3) strength, qu (kPa) RQD (%)
Long shaft 9.70–9.20 26.67 59 128 30
Short shaft 9.92–10.05 27.01 61 578 27

Fig. 4. Test area subsurface cross-section, Wilson County site.

failure of the short shaft (defined by excessive displacement the short and long shafts are shown in Fig. 7. The load–
under the applied load), a concrete block was installed be- displacement response for the long shaft did not exhibit a
hind it to add extra resistance, and loading of the long shaft yielding response under the applied maximum load. In the
was continued to a maximum load of 979 kN (the limit ca- case of the short shaft, the onset of yield was at a load of
pacity of the testing frame). 1088 kN. Both shafts did not reach their ultimate lateral
The top displacements of the short and long shafts are resistance under the maximum applied load of 1334 kN.
shown in Fig. 5. Based on the measured responses, the short Initial large displacements were observed for both shafts,
shaft was near its capacity under a lateral load of 534 kN. The as shown in Fig. 7. The presence of poor contact between
long shaft, however, did not reach its ultimate resistance un- the augured holes and the shafts was evident by the “con-
der an applied lateral load of 979 kN. Lateral displacement cave” shape of the initial part of the load–deflection curves.
for the long shaft under 534 kN was measured to be 0.017 m Displacement corresponding to no solid contact is readily
(nearly one order of magnitude less than the 0.137 m mea- identified from the shaft’s load–deflection curve.
sured for the short shaft). The data obtained from the continuous inclinometer mea-
Based on the inclinometer-measured deflection profile, surement system in each shaft during loading are shown in
the short shaft behaved as a rigid body with a linear dis- Figs. 8a and 8b. Similar to results from the Nash County
placement profile along the shaft’s full length (Fig. 6a). The tests, the short shaft behaved as a rigid body with a linear
long shaft behaved as a “restrained tip” shaft under the ap- displacement profile along the shaft’s full length (Fig. 8a).
plied lateral load, as indicated by the nonlinear displace- The long shaft behaved as an element with a partially re-
ments profile along its length (Fig. 6b). strained tip, as indicated by the nonlinear displacement pro-
file along its length (Fig. 8b).
Caldwell County tests
During testing, the short shaft experienced 0.089 m of dis- Wilson County tests
placement, and the long shaft was deflected 0.023 m at a Under a maximum applied load of 1681 kN, the short
maximum applied load of 1334 kN. Top displacements of shaft experienced 0.034 m of top lateral displacement while

© 2007 NRC Canada


Cho et al. 759

Table 5. Wilson County laboratory test results. Fig. 6. Deflection profile from slope inclinometer data, Nash
County.
Unconfined
Unit compressive
Depth weight strength, RQD
(m) (kN/m3) qu (kPa) (%)
Long shaft 3.2–4.7 26.67 57 578 13
Short shaft 3.0–4.5 27.01 62 567 59

Fig. 5. Top displacements measured from dial gages, Nash County.

the long shaft experienced a higher deflection magnitude of


0.055 m. Given the relatively small displacement achieved
for both shafts, it was decided during testing to continue ap-
plying the load past the test frame maximum capacity
(knowing that the test frame was designed with a FS = 1.5).
The top displacements of the short and long shafts are
shown in Fig. 9. The long shaft exhibited a nonlinear deflec-
tion response at the early stage of loading. Top deflection of
the short shaft, however, showed a nearly linear response
with loading. The stiffer response measured for the short
shaft, as compared to the long shaft, is mainly due to the dif-
ferent geological conditions at each location. The short shaft using the equation Ec = (E1I1 + E2I2 + E3I3)/Ic, where Ec is
location has rock with an RQD value of 59% as compared to the elastic modulus of the composite material; Ic is the mo-
13% at the location of the long shaft (see Table 4). Accord- ment of inertia of the composite material; E1 is the elastic
ing to the inclinometer-measured deflection profile, both modulus of the concrete; I1 is the moment of inertia of the
shafts exhibited a nonlinear displacement profile along their concrete; E2 is the elastic modulus of the steel casing; I2 is
length, as shown in Figs. 10a and 10b. the moment of inertia of the steel casing; E3 is the elastic
modulus of the steel rebar; I3 is the moment of inertia of the
Back-calculated P–y curves steel rebar. Therefore, the steel casing was included in the
calculations of the EI value of the shafts. The lateral deflec-
Strain measurements from the VW strain gages were re- tion (y), measured in metres, was obtained from the incli-
corded using a CR-10 data logger for the Caldwell and Wil- nometer data.
son counties tests and a readout box for the Nash County Figures 11 and 12 show the P–y curves calculated from
tests (strain gage numbers and locations were as marked on the strain gages for the long and short shafts, respectively, at
Fig. 1). From the measured strains, moments were calculated the Nash County test site. Best-fit P–y curves are plotted us-
along the length of the test shafts by piece-wise numerical ing the regression curve fitting procedure.
integration. The soil reaction, P, calculated in kN/m, was de- Figures 13 and 14 show the back-calculated P–y curves
termined using the calculated moment and the composite EI for the Caldwell County tests. As observed from the top de-
(elastic modulus × moment of inertia) of the reinforced con- flection measurement, initial deflection behavior is evident
crete with a casing of 859 MN·m2. The E value of the com- owing to no solid contact between shafts and surrounding
posite material (concrete and steel casing) was calculated material. The back-calculated P–y curves from the short

© 2007 NRC Canada


760 Can. Geotech. J. Vol. 44, 2007

Fig. 7. Top displacements measured from dial gages, Caldwell Fig. 9. Top displacements measured from dial gages, Wilson
County. County.

Fig. 8. Deflection profile from slope inclinometer data, Caldwell Fig. 10. Deflection profile from slope inclinometer data, Wilson
County. County.

© 2007 NRC Canada


Cho et al. 761

Fig. 11. Back-calculated P–y curves for the soft weathered rock, Fig. 13. Back-calculated P–y curves for the soft weathered rock,
short shaft at the Nash County site. short shaft at the Caldwell County site.

Fig. 12. Back-calculated P–y curves for the soft weathered rock, Fig. 14. Back-calculated P–y curves for the soft weathered rock,
long shaft at the Nash County site. long shaft at the Caldwell County site.

shaft indicated a full nonlinear range as shown in Fig. 13. Verification of back-calculated P–y curves
However, the P–y curves in Fig. 14 only show the linear
part of the P–y response, since the maximum applied load Using back-calculated P–y curves from the field data pre-
yielded relatively small deflections for the long shaft. sented in Figs. 11–16, analyses were performed using the
Figures 15 and 16 show the back-calculated P–y curves computer program BMCOL 76 (Matlock et al. 1981) to esti-
for the long and short shafts, respectively, from the Wilson mate the shaft-top deflection and to compare these computed
County tests, as previously mentioned. The resistance val- values with the field measurements obtained from dial gages
ues calculated for the short shaft are higher than those from mounted at the top of the test shafts. The objective of this
the long shaft owing to the difference in geological profiles exercise is to discern the adequacy of the P–y curves back-
at each shaft location (RQD = 13% at the long shaft, and calculated from field data as computed and measured
RQD = 59% at the short shaft). At the Wilson County test responses are compared. As shown in Figs. 17 and 18, the
site, most of the back-calculated P–y curves from the short calculated shaft-top deflection responses determined from
shaft show a nearly linear response. For the longer shaft, BMCOL 76 are in good agreement with the measured data
only P–y curves within the top 1.2 m of the profile show a for both shafts, with the exception of the Caldwell County
nonlinear response. Below this depth, the P–y curves also test data. In the Caldwell County comparison, the back-
plot essentially as a linear function, since the test load did calculated P–y curves from the short shaft used as input data
not produce enough deflection to reach a nonlinear range. for BMCOL 76 analysis were adjusted by removing data
For clarity of presentation, Figs. 15 and 16 have different points corresponding to no solid contact, but the measured
scales for the load and deflection axes. load–deflection response at the shaft top was not modified.

© 2007 NRC Canada


762 Can. Geotech. J. Vol. 44, 2007

Fig. 15. Back-calculated P–y curves for the soft weathered rock, Fig. 18. Verifying back-calculated P–y curves for long shafts.
short shaft at the Wilson County site.

Fig. 16. Back-calculated P–y curves for the soft weathered rock, Fig. 19. Measured versus estimated top deflection comparisons.
long shaft at the Wilson County site.

Fig. 17. Verifying back-calculated P–y curves for short shafts.

© 2007 NRC Canada


Cho et al. 763

Table 6. Summary of field load tests data.


Nash County Caldwell County Wilson County
Short shaft Long shaft Short shaft Long shaft Short shaft Long shaft
Length (m) 3.35 4.57 4.00 4.80 4.85 5.71
Max. load (kN) 534 979 1334 1334 1681 1681
Shaft-top deflection (m) 0.135 0.036 0.089 0.023 0.034 0.055
RQD (%) <25 <25 <30 <30 ≈60 ≈15
Rock type Meta-argillite Gneiss Crystalline granite

In the case of Caldwell County short shaft, it seems that The overestimated response yielded by the weak rock
10–15 mm of lateral movement was needed before lateral re- model may be partly due to the recommended distribution of
sistance was mobilized by the surrounding materials. At dis- the soil mudolus, kh, with depth (Reese 1997) in which kh
placement levels <50 mm, measured and computed was multiplied by a factor kir = 100 + 400x/3b; where x is
responses were relatively comparable. the depth and b is the shaft diameter. At the ground surface
this factor is equal to 100 and increases to over 500 at 2.5 m
below the surface. Such an increase in modulus created a
P–y curve model comparison stiff layer near the shaft top, which consequently led to the
rather stiff load–deflection response shown in Fig. 19. On
Lateral resistance analyses for drilled shafts embedded in
the other hand, at relatively small displacements (correspond-
weathered rock are regularly performed based on P–y curve
ing to a load less than 200 kN), the stiff clay model slightly
models. Among a few P–y models available in the literature,
overpredicts the measured response at the low load levels.
weak rock (Reese 1997) and stiff clay (Reese and Welch
This is a manifestation of the selected value of subgrade
1975) models can be used to design drilled shafts embedded
coefficient coupled with the low value of shear strength,
in weathered rock. Reese’s weak rock model was considered
which leads to softer P–y curves. Using the stiff clay model,
to be “interim” owing to the lack of load tests data to vali-
and as the displacement level increases, the computed load
date it, as stated by Reese (1997). An adjusted stiff clay
response was, however, underestimated because the shear
model was recommended for use in weathered rock by Gabr
strength value based on the stiff clay criterion was low. At
(1993) in the absence of a specific P–y model for weathered
relatively large displacement levels, the stiff clay model un-
rock at that time.
derestimated the shaft resistance, but such estimation is a
Figure 19 shows lateral pile-top deflection comparisons
manifestation of the lower input shear strength value used in
between measured values from field tests and values esti-
the stiff clay model.
mated from LPILE (Reese and Wang 1997) using Reese’s
weak rock model (Reese 1997) and the stiff clay model It is clear that selection of shear strength and modulus
(Reese and Welch 1975). For the input values of Reese’s properties greatly impact the computed lateral response re-
weak rock model, a shear strength of 30.0 MPa was used, gardless of which P–y approach is taken (stiff clay versus
which is half of the unconfined compressive strength based weak rock). While the values and approach recommended by
on laboratory test results reported by Reese (1997). A value Reese (1997) led to good estimation of the load test data re-
of 0.0005 for ε50 and a value of 600 MN/m3 for the subgrade ported by Nyman (1980), this approach did not yield good
coefficient were also used in the analysis following LPILE results for the drilled shafts tested in the Piedmont weath-
program manual recommendations (Reese and Wang 1997). ered profile. To obtain shaft-top deflection values similar to
The input data for the stiff clay model are 192 kPa for the the measured values, one possible combination for input to
shear strength, 0.004 for ε50, and 543 MN/m3 for the subgrade Reese’s weak rock model is 600 kN/m3 for the subgrade
coefficient. These properties for the stiff clay model are as coefficient, which is 0.1% of the recommended value; and
recommended by LPILE program manuals (Reese and Wang 0.005 for ε 50, which is 10 times the recommended value
1997), and explain the low value of cohesion. It should be based on shear strength. For the stiff clay model, one possi-
mentioned that the back-calculated subgrade coefficient ble combination of input parameters is 300 kPa for shear
from field tests is approximately 40 MN/m3 near the sur- strength, which is about twice the recommended value nor-
face and increases to approximately 80 MN/m3 at a depth mally used for stiff clay but 0.1% of the shear strength
of around 2.5 m. The 40 MN/m3 is about 7% of the value value; and 543 MN/m3 back-calculated from the load test
recommended by Reese and Wang (1997). The range of un- data.
confined compressive strength obtained from the field was
19–154 MPa.
As shown in Fig. 19, the computed shaft-top deflection Summary and conclusions
from the weak rock model, using design properties reported
in the literature, overestimates the measured response of the Full-scale lateral load tests were performed on six 0.762 m
shafts. Although the coefficient of subgrade reaction was diameter drilled shafts embedded in weathered rock at
comparable in both cases (600 MN/m3 for the weak rock three different sites. Site characterizations were conducted
model versus 543 MN/m3 for the stiff clay model), the weak and fully instrumented drilled shafts were constructed to
rock model consistently yielded stiffer response. obtain data and to develop P–y curves for these profiles.

© 2007 NRC Canada


764 Can. Geotech. J. Vol. 44, 2007

Table 6 shows a summary of the test results and character- References


istics of the test sites.
ASTM. 1995. Standard method of testing piles under lateral loads
Deflection characteristics with depth were back-calculated
(ASTM D3966-90). In 1995 Annual Book of ASTM Standards,
from inclinometer data. Using the continuous inclinometer
Vol. 04.08. American Society for Testing and Materials
probes, the deflection profile along each shaft was directly (ASTM), Philadelphia, Penn.
measured without the need for complicated back-calculations. Coates, D.F. 1970. Rock mechanics principles. 3rd ed. Department of
The lateral soil reaction was, however, back-calculated from Energy Mines and Resources. Mine Branch Monograph. 874 pp.
strain measurements with depth. Digioia, A.M., Jr., and Rojas-Gonzalez, L.F. 1994. Rock socket
The measured shaft-top deflections from field load tests transmission line foundation performance. IEEE Transactions on
were compared with the estimated values obtained using Power Delivery, 9(3): 1570–1576.
two different P–y models: (a) Reese’s weak rock model Gabr, M. 1993. Discussion of “Analysis of laterally loaded shafts
and (b) stiff clay model. The weak rock model overesti- in rock” by John P. Carter and Fred H. Kulhawy. Journal of
mated the resistance of the test shafts mainly because of Geotechnical Engineering, ASCE, 119(12): 2015–2018.
the selection of the analyses parameters in terms of increas- Matlock, H., Bogard, D., and Lam, I.P. 1981. BMCOL 76. A com-
ing the kh value by a factor of 100 at the ground surface puter program for the analysis of beam-columns under static axial
and by a factor of over 500 at a depth of 2.5 m. The stiff and lateral loading. Documented at Ertec, Inc. Calif.
clay model, with the recommended stiff clay parameters, Nyman, K.J. 1980. Field load tests of instrumented drilled shafts in
underestimated the lateral load–displacement response and coral limestone. M.S. thesis, Department of Civil Engineering,
it is considered to be conservative for use in weathered University of Texas at Austin, Tex.
rock profiles. It is clear that the selection of shear strength Peck, R.B. 1976. Rock foundations for structures. In Proceeding of
and modulus properties greatly impacts the computed lat- ASCE Specialty Conference on Rock Engineering for Founda-
eral response regardless of which P–y approach is taken tions and Slopes, Boulder, Colo., 15–18 August 1976. New
(stiff clay versus weak rock). While the values and ap- York. Vol. II, pp. 1–21.
Reese, L.C. 1997. Analysis of laterally loaded piles in weak rock.
proach recommended by Reese (1997) led to good estima-
Journal of Geotechnical and Geoenvironmental Engineering,
tion of the load test data reported by Nyman (1980) in coral
ASCE, 123(11): 1010–1017.
limestone, this approach did not yield reasonable lateral Reese, L.C., and Wang, S.T. 1997. LPILE [computer program].
displacement estimations for the drilled shafts tested in Ensoft, Inc., Austin, Tex.
Piedmont weathered profiles. The authors advocate that Reese, L.C., and Welch, R.C. 1975. Lateral loading of deep foun-
since the lateral resistance of weathered rock under loading dations in stiff clay. Journal of the Geotechnical Engineering
is a function of its shear and compressive strengths, the Division, ASCE, 101(7): 633–649.
level of the groundwater table, and joint spacing and joint Zhang, L., Ernst, H., and Einstein, H.H. 2000. Nonlinear analysis
condition, a P–y curve procedure should take the geologi- of laterally loaded rock-socketed shafts. Journal of Geotechnical
cal components into consideration. and Geoenvironmental Engineering, ASCE, 126(11): 955–968.

© 2007 NRC Canada

Vous aimerez peut-être aussi