Professional Documents
Culture Documents
Geot 2007 00108
Geot 2007 00108
00108]
Wind, wave and current forces cause significant overturn- Les forces du vent, des vagues et des courants détermi-
ing loads to be transmitted to foundation systems of nent la transmission de charges de basculement significa-
fixed-bottom offshore structures, while buoyancy forces, tives sur des systèmes de fondations de structures
in conjunction with environmental loading, impart direct offshore à fond fixe, tandis que des forces de ascension-
tensile loads to foundations of floating facilities. Shallow nelles, conjointement avec des charges environnementales,
skirted foundations are a potentially attractive foundation appliquent des efforts de tension directs sur les fonda-
solution when overturning or tension loading is signifi- tions d’installations flottantes. Les fondations à jupe peu
cant, as suctions mobilised within the soil plug provide profondes sont des solutions potentiellement attrayantes
tensile resistance. While passive suctions are maintained, pour les fondations, en présence de charges de bascule-
undrained reverse end bearing will govern failure provid- ment et de traction significatives, car les forces d’aspira-
ing enhanced uplift resistance compared with vented tion mobilisées au sein du bouchon de sol offrent une
pull-out of the foundation unit (with or without the soil résistance à la traction. Tant que les forces d’aspiration
plug), governed by skirt/soil friction. Currently uncer- passives se maintiennent, la résistance inverse à la pointe
tainty exists over the limiting ratio of skirt depth to non drainée déterminera la rupture, en présentant une
foundation diameter to mobilise undrained reverse end résistance supérieure au soulèvement par rapport à la
bearing and the duration over which passive suctions can remontée aérée des fondations (avec ou sans bouchon de
be sustained. This paper reports results from beam sol), sous l’effet de la friction jupe/sol. On relève actuelle-
centrifuge tests investigating the response of shallow ment une certaine incertitude concernant le ratio limite
skirted foundations in lightly over-consolidated clay to de la profondeur de la jupe sur le diamètre des fonda-
concentric transient and sustained uplift. tions, pour mobiliser la résistance inverse à la pointe non
drainée et la durée de la période au cours de laquelle ces
forces d’aspiration passives peuvent être soutenues. La
présente communication présente des résultats de tests
centrifuges sur poutre effectués pour examiner la réac-
tion de fondations à jupe peu profondes, dans de l’argile
KEYWORDS: bearing capacity; footings/foundations; model consolidée par précompression, aux remontées concentri-
tests; offshore engineering ques transitoires et continues.
525
526 GOURVENEC, ACOSTA-MARTINEZ AND RANDOLPH
the bearing capacity factor for a surface foundation by a lines acknowledge that temporary suctions caused by dy-
depth factor, dc . Classical bearing capacity theory incorpo- namic loads may allow greater capacities to be mobilised,
rates empirical and semi-empirical depth factors based on because the effect is temporary they advise it should not be
undrained uniaxial compressive failure of a smooth-sided accounted for in design unless substantiated by appropriate
circular foundation (Skempton, 1951; Brinch Hansen, 1970; analysis or experimentation.
Vesic, 1975) although these are widely applied across a The results reported in this paper provide quantitative data
range of foundation geometry and interface roughness. These concerning the response of shallow skirted foundations under
historical factors have recently been challenged by analytical transient and sustained uplift as a function of embedment
and numerical studies accounting explicitly for foundation ratio, stress level and duration of consolidation prior to
geometry, soil strength profile and foundation/soil interface uplift, and the magnitude and duration of applied uplift. A
conditions (Martin & Randolph, 2001; Houlsby & Martin, selection of preliminary results from some of the tests has
2003; Salgado et al., 2004; Edwards et al., 2005). been previously reported by Gourvenec et al. (2007).
Despite refinement of undrained bearing capacity factors,
it is unclear whether similar factors are appropriate in
tension and compression. Experimental studies have investi- EXPERIMENTAL SET-UP
gated undrained uplift capacity, although typically of deeper Apparatus
skirted foundations (d/D . 2) and often at 1 g (e.g. The tests were carried out in the beam centrifuge at the
Fuglsang & Steensen-Bach, 1991; Steensen-Bach, 1992; Centre for Offshore Foundation Systems at the University of
Puech et al., 1993; Clukey & Morrison, 1993; Rao et al., Western Australia (COFS-UWA), which has a nominal
1997; Watson et al., 2000; Randolph & House, 2002; Luke radius of 1.8 m (Randolph et al., 1991). The strongbox has
et al., 2005). Small-scale laboratory-floor tests are limited dimensions 0.650 m by 0.390 m in plan and 0.325 m deep
by unrealistically low soil stresses relative to the shear with a maximum payload of 200 kg at 200 g. The tests
strength at foundation level for prototype conditions, a sig- reported in this paper were carried out at 167 g providing a
nificant limitation, particularly for problems involving uplift. testing area in excess of 7000 m2 .
For the case of skirted foundations, low soil stresses allow a Two shallow skirted foundation models were fabricated
tension crack to develop at skirt tip level leading to pull-out with skirt depth to foundation diameter ratios d/D ¼ 0.15
of the soil plug rather than enabling reverse end bearing to and 0.3. In each case, the base plate was fabricated from
be mobilised. aluminium to minimise the weight of the foundation and the
Considering previous centrifuge studies investigating tran- skirts and stiffeners were fabricated from stainless steel to
sient uplift of skirted foundations in clay, some report prevent buckling. The models comprised a circular plate
reverse end bearing resistance very similar to that in com- equipped with a peripheral skirt and internal cruciform
pression, while others report a reduced tensile resistance. stiffener (to a level flush with the peripheral skirt). The skirt
Puech et al. (1993) report centrifuge tests of skirted mud- thickness to foundation diameter ratio of t/D ¼ 0.01 was
mats for temporary foundations (d/D ¼ 0.18) and propose chosen as a compromise between typical ratios of ,0.005
that for shallow skirted foundations in soft sediments in for skirted foundations employed in the field (Randolph et
significant water depths, reverse end bearing can be ex- al., 2005; Bye et al., 1995), and practical fabrication limits
pected. Watson et al. (2000) report centrifuge tests of for the models. The foundation models were left unfinished
shallow skirted foundations (d/D ¼ 0.5) in normally consoli- (i.e. not painted) and therefore would be expected to be
dated kaolin in which tensile and compression capacities partially rough, with an interface friction angle of about 188
were indistinguishable. In contrast Clukey & Morrison (Chen & Randolph, 2007).
(1993) report centrifuge tests of skirted foundations with The four compartments formed by the internal stiffener
d/D ¼ 2, in normally consolidated clay, in which bearing were connected by a small hole at the cross-over to allow
capacity factors in tension of about 80% that in compression for drainage during installation through a single vent in the
were observed. Bearing capacity in undrained uplift may be base plate. Details of the foundation geometry at model and
less than that in compression, or the reduced end bearing prototype scale are summarised in Table 1.
may be attributable to partial drainage and dissipation of
negative excess pore pressures during swelling of the soil in
uplift. Further investigation is required before firm conclu- Instrumentation
sions can be drawn. The models were equipped with total pressure and pore
The pull-out capacity of longer skirted caissons under pressure measurement instrumentation to monitor conditions
sustained loading has received some attention (e.g. Clukey under the base plate, along the skirt and at foundation level.
et al., 1995; Randolph & House, 2002; Clukey & Phillips, A schematic of the models indicating the position of the
2002; Clukey et al. 2004), although in all cases d/D > 2. instrumentation is shown in Fig. 1.
Both numerical analysis and centrifuge model tests have Two miniature pore pressure transducers (PPT) were en-
shown that sustained loads may be maintained for years for cased in a housing attached to the cruciform stiffener to
caisson foundations with large d/D. For example, Clukey et provide pore pressure measurements at skirt tip level. The
al. (2004) considered suction caissons with length to dia- incorporation of transducers into the model has some short-
meter aspect ratios between 3.9 and 7 in Gulf of Mexico
soft clay and report negligible reduction in capacity for
holding times of up to 100 days, while sustained loads Table 1. Foundation geometry
between 65% and 75% of the undrained capacity were held
indefinitely with only minimal displacements. Model: m Prototype: m
Sustained pull-out of shorter skirted foundations has not
been considered previously, although it is arguably more Plate diameter D 0.12 20
critical with shorter drainage path lengths leading to more Plate thickness 0.005 0.835
rapid dissipation of suctions. Skirt depth d 0.018 3.0
0.036 6.0
In general there is uncertainty over, and no formal gui- Skirt thickness t 0.0012 0.20
dance regarding, the timescale for which tensile stresses can Stiffener thickness 0.0009 0.15
be sustained beneath a skirted foundation. Although guide-
EXPERIMENTAL STUDY OF UPLIFT RESISTANCE OF SHALLOW SKIRTED FOUNDATIONS IN CLAY 527
gradual loss of uplift capacity and to detect differences in
Stiffener
pore water pressure dissipation under eccentric uplift,
TPT-lid planned in a future set of tests.
Diametrically opposed pairs of total pressure transducers
(TPTs) were set into the external face of the peripheral skirt
to monitor variations of total radial stress acting against the
skirt. The external face of the TPT was set flush with the
external face of the skirt. A single pair of TPTs was
Drainage PPT-lid mounted on the shorter skirted model (d/D ¼ 0.15) and two
valve
levels (on the same diametric section) were mounted on the
longer skirted model (d/D ¼ 0.3). The upper level of TPTs
20 m on the longer skirted model was positioned at the same
distance below the base plate as the TPTs on the shorter
(a) skirted model.
A PPT and TPT were set into the underside of the
0·83 foundation base plate. The PPT was positioned to identify
See detail (c) excess pore pressures within the skirted cavity during instal-
Drainage Inter-cell PPT-lidTPT-lid lation in order to ascertain if drainage was able to take place
valve drainage sufficiently quickly. The TPT acted as a backup for pore
(b) pressure measurements prior to touchdown of the base plate,
to assess when contact between the soil and base plate
occurred during installation and if separation occurred dur-
Model d/D ⫽ 0·15 Model d/D ⫽ 0·30 ing transient or sustained uplift.
2·0 m 3·0 m
TPT 1·0 m Load and displacement application and measurement
2·5 m 6·0 m
TPT The models were attached by a rigid arm to a one-
1·5 m dimensional actuator allowing load or displacement control
t ⫽ 0·2 m
via a 3 kN load cell and a 25 mm stroke linear displacement
(c) transducer (LDT). Load or displacement control was
achieved with a software feedback loop using output from
the load cell or LDT from data recorded at a typical
frequency of 10 Hz during testing.
Drainage PPT-lid
valve
PPTs at skirt
tip level
Soil sample
(in housing) The tests were carried out in a lightly over-consolidated
kaolin clay prepared from slurry in a consolidation press at
1 g. The sample was incrementally loaded to a maximum
Stiffener TPT-lid vertical stress of 150 kPa over 11 days, held for a maximum
of 4 days to avoid viscous, creep and secondary compression
Inter-cell drainage effects, and subsequently incrementally unloaded to 40 kPa
over a further 4 days. The sample was then moved to the
Dimensions given for equivalent prototype foundation centrifuge and reconsolidated at 167 g with an average
(d) 0.17 m head of surface water, producing a total stress of
280 kPa on the soil surface serving as backpressure for
Fig. 1. Foundation geometry and location of model instrumenta- negative increments in pore pressure, sufficient to prevent
tion: (a) plan view—top; (b) elevation; (c) skirt details; (d) plan
view—base
cavitation. The surface water level was maintained constant
during the testing by way of a hose to compensate for
evaporation effects. Test results are reported in terms of
comings, particularly remoulding soil within the plug owing excess pressures (i.e. the hydrostatic pressure has been
to local failure around the housing during installation. It deducted from the measured pressure) such that the reported
was, however, considered more reliable to attach the transdu- results are independent of the head of surface water.
cers to the model than trying to pre-embed transducers One-dimensional consolidation tests in a standard oed-
within the body of the soil. The latter is a complex process ometer were carried out as part of this study to accompany
when preparing samples from slurry and moreover when the centrifuge tests. The samples were prepared from the
several tests are carried out at different sites and different same material used for the centrifuge sample and consoli-
elevations in the same strong box. The main purposes of dated to the same maximum stress level. The findings of the
PPTs at tip level were to provide information about the oedometer tests are summarised in Table 2. The mechanical
Stress: kPa
Table 3. Testing programme
20 Touch-down
Reference Test sequence (skirt tip)
40 Touch-down
Box 1; d/D ¼ 0.15 (base plate)
60
Tip penetration: m
0
# tc ¼ 0
B1T2_2 Sustained 0.3Vu 2 Touch-down
# tc ¼ 0
B1T2_3 Sustained 0.3Vu 4 Penetration rate:
B1T3_1 Sustained 0.6Vu vmodel ⫽ 0·1 mm/s
# tc ¼ 4 h (12.7 y) 6
B1T3_2 Sustained 0.3Vu 0 1 2 3 4 5
B1T4 (Consolidation stress , 2Q) Prototype time, tp: months
Undrained (to UC) (a)
B1T5_1 Undrained (to w/D ¼ 0.02) 40
# tc ¼ 16.0 h (50.9 y)
Sustained 0.3Vu
Stress: kPa
B1T5_2 45
# tc ¼ 5.5 h (17.5 y)
50
B1T5_3 Sustained 0.1Vu
55
Box 2; d/D ¼ 0.3
60
B2T1_1 Undrained (to w/D ¼ 0.02) 0 10 20 30 40 50
5·7
# tc ¼ 5 h (15.9 y)
Tip penetration: m
⫺20 0·3 Vu
(h ¼ hours, y ¼ years), Vu ¼ uplift resistance mobilised at w/D ¼
0.02, w/D ¼ relative uplift displacement, UC ¼ ultimate capacity, 0
consolidation stress , Q unless otherwise stated, Q ¼ installation 20 Reconsolidation
resistance 40
60
0 10 20 30 40 50 60
5·0
Tip penetration: m
was likely in the vicinity of the skirt tips and stiffener 5·1
(v ¼ vt/cv , 1, taking t as the average thickness of the tip 5·2
bearing area of the skirt and stiffener), as described above 5·3
for installation. 5·4 Tip level at end of reconsolidation
Six undrained uplift tests were carried out; three with the following preceding undrained uplift
5·5
shorter skirted foundation, d/D ¼ 0.15 (prefixed B1: B1T1,
0 10 20 30 40 50 60
B1T4 and B1T5) and three with the longer skirted founda-
Prototype time, tp: months
tion, d/D ¼ 0.3 (prefixed B2: B2T1, B2T2 and B2T3). Three
(c)
tests (B1T1, B1T5 and B2T1) were limited to a relative
vertical displacement w/D ¼ 0.02, in order to protect the site Fig. 3. Time histories for: (a) installation; (b) consolidation
to enable subsequent sustained load tests. The remaining following installation; (c) sustained uplift (test B2T1)
three undrained uplift tests (B1T4, B2T2 and B2T3) were
continued to larger displacement to investigate ultimate up-
lift capacity.
Sustained uplift was applied through load control, by factor, 0.083 , T , 0.325. In selected cases, consecutive up-
imposing a selected proportion of the undrained uplift lift tests were carried out without an intermediate period of
resistance mobilised at the limiting displacement criterion reconsolidation in order to assess the effect of the degree of
w/D ¼ 0.02. consolidation on uplift resistance.
Sequences of tests were carried out at a single site to Reinstallation between consecutive tests was achieved
optimise the testing programme. Consecutive uplift tests through load control to the original installation load, noting
were carried out immediately following a previous uplift that in most cases the foundation did not return to the embed-
test, immediately following reinstallation or (more typically) ment depth prior to the preceding test due to swelling of the
following reinstallation and a period of reconsolidation soil during the initial uplift, therefore the embedded depth of
sufficient to achieve a relatively steady state. Periods of the foundation was slightly reduced in subsequent tests (as
intermediate consolidation varied from 12 years to 50 years seen in Fig. 3(c)). Reinstallation took place with the base plate
(as seen in Table 3), or in terms of the dimensionless time sealed since displacements were small (w/D < 0.02).
530 GOURVENEC, ACOSTA-MARTINEZ AND RANDOLPH
EXPERIMENTAL RESULTS predicted from the sum of tip bearing and skirt friction
Installation response through the simple relationship
Figure 4 shows installation resistance measured by the
Q ¼ Nc(tip) su(tip) þ ª9z Atip þ Æ su(av) Asurface (3)
load cell and excess pore pressures measured by the PPTs at
skirt tip level. The load cell measurement, Q, has been
divided by the plan area of the foundation, A, to give where Nc(tip) is a bearing capacity factor for tip resistance,
installation resistance in terms of stress. Touchdown of the su(tip) and su(av) are respectively the undrained shear strength
foundation base plate, at a relative embedment z/D is evident at tip level and averaged over the embedment depth and Æ is
from the rapid increase in the load cell and PPT readings. a friction ratio defining the limiting friction along the skirt/
One of the tests with the longer skirted foundation was soil interface as a proportion of the shear strength of the soil
loaded some way beyond touchdown to achieve a higher mass. Atip is the sum of the skirt and stiffener end bearing
consolidation pressure, indicated by the longer tail of the area and Asurface is the sum of the surface areas of the inner
curve in Fig. 4 (test B2T3, see Table 3). The final installa- and outer skirt and stiffener.
tion load was maintained during the subsequent consolida- For undrained conditions a bearing capacity factor for tip
tion stage and was re-established for periods of intermediate resistance Nc(tip) ¼ 7.5 is commonly adopted, based on analo-
reconsolidation between consecutive tests carried out at the gous conditions of a deeply embedded strip (Randolph et
same site. al., 2005; Andersen et al., 2005). Owing to the low rate of
Installation data for one of the tests on the shorter skirted installation in the centrifuge tests, v ,1, partial consolida-
foundation, test B1T4, is not shown as accidental dynamic tion in the vicinity of the skirt tips would be expected to
installation occurred after the thread connection of the lead to increased tip resistance. Randolph & Hope (2004)
model with the actuator sheared in-flight. As a result con- indicate doubling of resistance for a normalised velocity
solidation took place under the submerged weight of the v ¼ vt/cv ,1.
model, equal to approximately twice the consolidation pres- The undrained shear strength at skirt tip level su(tip) was
sure in the other tests with the shorter skirted foundation, 14.5 kPa for the tests with the shorter skirted foundation, d/
B1T1 and B1T5. D ¼ 0.15, in box 1 and 20.0 kPa for the tests with the longer
Installation of the foundations was accompanied by 1 or skirted foundation, d/D ¼ 0.3 in box 2 (Fig. 2). The average
2 mm plug heave occurring with the shorter and longer undrained shear strength over the depth of the skirts, su(av)
skirted foundations respectively (i.e. corresponding to em- was 11.5 kPa and 14.2 kPa for d/D ¼ 0.15 and 0.3 respec-
bedment of ,95% of the total skirt length). An exception tively, determined from the integral of equation (2)
was observed in one of the tests with the shorter skirted ð z¼d
foundation, test B1T5, in which 2 mm of plug heave was v90 að v9m = v90 Þ m dz
observed, corresponding to embedment of ,90% of the total su(av) ¼ z¼0 (4)
skirt length. The internal stiffeners and PPT casings repre- d
sent about 3.5% of the internal area of the foundation, while
the skirt represents about 4%. Assuming for these short Comparing the observed profile of the measured installation
skirts that the wall thickness is accommodated equally by resistance prior to touchdown of the base plate with equation
inward and outward flow of the soil (Chen & Randolph, (3) indicates an interface friction ratio Æ,0.3, similar to that
2007), a heave ratio [(portion of volume of caisson wall reported by Chen (2005) for partially drained conditions and
accommodated by flow into the plug + volume of stiffener a lightly over-consolidated kaolin clay.
and internal protuberances)/volume previously occupied by Installation resistance of 18 and 36 kPa was measured just
soil] of around 5% would be expected. The lower embed- before contact of the base plate for the foundations with
ment achieved in test B1T5 may reflect more localised heave embedment ratios d/D ¼ 0.15 and 0.30 respectively.
in the vicinity of the PPT casings. Excess pore pressures measured with the PPTs at skirt tip
Considering the installation resistance defined by the load level are also shown in Fig. 4. Readings from the two
cell data, good repeatability was observed during installa- sensors during each test were consistent, although it is
tion. The slightly higher installation resistance for the longer noteworthy that they were around four times greater than the
skirted foundation for relative embedment z/D < 0.15 corre- average installation resistance indicated by the load cell. It
sponds to the higher undrained shear strength in box 2 (see is likely that the measured pore pressures relate to a local
Fig. 2). bearing failure around the housing of the PPTs.
The installation resistance of a skirted foundation can be Changes in total vertical stress and excess pore water
pressures during installation, measured by the TPT and PPT
located on the underside of the base plate, are shown in Fig.
Installation resistance (Q/A) and excess pore pressure (∆u): kPa
5. Excess pore water pressures were not developed beneath
0 20 40 60 80 100 120 140
0·00 0
the base plate during penetration of the skirts, indicating that
drainage provision in the model was adequate. Both sensors
indicated touchdown of the baseplate with a rapid increase
Normalised embedment, z/D
0·05 1
Prototype embedment, z: m
PPTs at skirt tip in pressure. The PPT always recorded contact with the soil
0·10 2
surface slightly after the TPT as the PPT was slightly
Load cell recessed (0.001 m) into the underside of the base plate while
0·15 3 the TPT was flush with the underside of the base plate (see
Fig. 1(b)). Termination of the PPT and TPT curves in Fig. 5
0·20 4 corresponds to the same point in time, with the higher final
End of B2T1 &
contact stress recorded by the TPT owing to the additional
0·25 B2T2 5 increment of embedment.
Figure 6 shows the total radial stress measured during
0·30 6 installation by the TPTs located along the skirt. The changes
in gradient correspond to the change in surrounding media
Fig. 4. Installation resistance and excess pore pressure at skirt from water to soil adjacent to the mid-point of the sensor.
tip level during installation Similar readings were observed in all tests and the similarity
EXPERIMENTAL STUDY OF UPLIFT RESISTANCE OF SHALLOW SKIRTED FOUNDATIONS IN CLAY 531
Change in total vertical stress (∆σv) and excess pore
pressure (∆u): kPa tests with the shorter skirted foundation (prefixed B1: B1T1,
0 20 40 60 80 100 B1T4 and B1T5) and three tests with the longer skirted
0·00 0 foundation (prefixed B2: B2T1, B2T2 and B2T3) are shown.
The inset shows a detailed view of the load–displacement
response up to a relative displacement w/D ¼ 0.02, the
Normalised embedment, z/D
0·05 1
Prototype embedment, z: m
prescribed limiting displacement criterion for tests that were
0·10 d/D ⫽ 0·15 TPT-lid 2 not continued to mobilisation of ultimate capacity.
The shorter skirted foundation, d/D ¼ 0.15, appeared to be
0·15 PPT-lid 3 close to ultimate capacity by a relative displacement
w/D ¼ 0.02 (tests B1T1, B1T4 and B1T5), while by contrast
0·20 4 the longer skirted foundation was still developing resistance
d/D ⫽ 0·30 at w/D ¼ 0.02 (tests B2T1, B2T2 and B2T3), reaching
0·25 TPT-lid 5 ultimate capacity at a relative displacement of around
w/D ¼ 0.1.
0·30 PPT-lid 6 Greater uplift resistance was achieved with the longer
skirted foundation as would be expected. Following consoli-
Fig. 5. Changes in total stress and excess pore pressure beneath dation at a stress level close to the installation resistance, a
top cap during installation doubling of the skirt depth to diameter ratio from d/D ¼
0.15 to 0.3 led to a 150% increase in uplift resistance
Change in total radial stress, ∆σt: kPa mobilised at a relative displacement w/D ¼ 0.02 (comparing
⫺5 0 5 10 15 20 25 30 35 40 45 50 B1T1 and B1T5 with B2T1 and B2T2) or a 250% increase
0·00
d/D ⫽ 0.15
0 comparing ultimate capacity (B1T1 and B2T2).
d/D
0·15 The magnitude of the consolidation stress appeared to have
Normalised embedment, z/D
TPTupper
Prototype embedment, z: m
⫺8 B1T1 0·3 ⫺1·6 was taken to be isotropic, but both homogeneous and hetero-
B1T4 ⫺1·2
geneous soil strength profiles were considered. For the hetero-
⫺7 0·2
B1T4
⫺6
⫺0·8
B2T2 ⫺1·2 geneous conditions, the undrained shear strength, su , was
⫺5
⫺0·4
B2T1
B2T3
0·1
defined as varying linearly with depth in terms of a hetero-
0·0 0·0 geneity coefficient kD/sum , where k is the gradient of the
⫺4 ⫺0·8
⫺160⫺120 ⫺80 ⫺40 0 40 80 120 undrained shear strength profile, D is the diameter of the
⫺3 foundation and sum is the undrained shear strength at the
B2T3 Uplift d/D
⫺2 B1T5 0·15 ⫺0·4 mudline. For the conditions in this study, the heterogeneity
B1T1 0·30
⫺1 End of B2T1 B2T3
coefficient lies in the region of 4 to 5 (taking sum , 8 kPa, k
0 0·0 in the range , 2 kPa/m close to the surface, reducing to ,
⫺200 ⫺150 ⫺100 ⫺50 0 50 100 1.5 kPa/m for 0.15 , z/D , 0.3 and D ¼ 20 m), giving Nc ,
Uplift resistance, Vu/A: kPa 8 to 8.5 for both the foundation embedment ratios considered
(a similar Nc is to be expected given the relatively shallow
Fig. 7. Undrained uplift resistance embedment ratio of both foundations). Assuming uniform
532 GOURVENEC, ACOSTA-MARTINEZ AND RANDOLPH
undrained soil strength with depth (i.e. kD/sum ¼ 0), a lower the applied load carried by friction along the skirts, which is
bearing capacity factor, Nc , 6.5, is predicted, although as up to 30% in some tests. The point of separation between
for the heterogenous condition, a similar value is predicted the soil plug and the underside of the base plate was clear
over the range of foundation embedment ratios considered in in Test B1T1 with no further increase in pressure observed
this study. The theoretical bearing capacity predictions may
be expected to lie on the low side given the assumption of
frictionless skirts in contrast to the partial roughness of the
skirts of the physical models.
Comparing the bearing capacity factors from the measured
uplift resistance from the centrifuge tests with the theoretical
predictions indicates the ultimate measured capacity of the
shorter skirted foundation, d/D ¼ 0.15, was less than 50% of
the theoretical end bearing capacity, while the longer skirted
foundation, d/D ¼ 0.3, mobilised full (theoretical) reverse
end bearing capacity.
Recent tests at COFS-UWA indicated a bearing capacity
factor in compression Nc , 13 for d/D ¼ 0.3 under similar
conditions as this series of tests (Acosta-Martinez et al.,
2008). Comparing with the bearing capacity factor in uplift
from this study, Nc , 8.9, indicates undrained uplift capacity
of approximately 70% of the compression capacity was
mobilised (similar to observations reported by Clukey &
Morrison (1993)). It is noteworthy that the experimental
bearing capacity in compression is around 60% greater than
the theoretical prediction (Houlsby & Martin, 2003), high-
lighting the degree of influence of the soil strength profile
(a)
and skirt interface roughness on bearing capacity. Experi-
mental results are not available for the shorter skirted
foundation, but if a similar bearing capacity factor as ob-
served for the longer skirted foundation is assumed (as
indicated by the theoretical bearing capacity factors), ulti-
mate resistance measured in uplift with the shorter skirted
foundation was possibly as low as 30% of the compressive
end bearing capacity.
Examination of the soil surface after testing indicated local
shear governed failure of the shorter skirted foundation, d/D
¼ 0.15, with cracking observed near the skirt, , 0.1D–0.15D
from the perimeter, and concentrated on the side of the PPT
housings. Conversely, a reverse end bearing mechanism was
evident for the longer skirted foundation, d/D ¼ 0.3, with the
soil plug proud of the surface and a circumferential crack ,
0.3D–0.4D from the footprint. Photographs of the observed
mechanisms are shown in Fig. 8.
Figure 9 shows excess pore pressures measured by the
PPTs at skirt tip level plotted against relative uplift displace-
(b)
ment. The load paths begin at a state of almost zero excess
pore pressure, indicating the duration of consolidation was Fig. 8. Failure modes observed in undrained uplift: (a) local
sufficient for full dissipation of excess pore pressures devel- shear mechanism; d/D 0.15 (test B1T1); (b) reverse end
oped during installation (Fig. 4). During uplift the two bearing mechanism; d/D 0.30 (test B2T1)
sensors recorded consistent values in each test other than in
B2T3 where some instability was observed in one of the
sensors. The PPT readings suggested higher uplift resistance ⫺10 ⫺2·0
⫺2·0 ⫺0·4
Prototype uplift displacement, w: m
B2T1
than indicated by the load cell readings, and as noted with ⫺9 B3T4
B3T4 ⫺1·5 ⫺0·3
Uplift displacement, w/D: %
relate to a local failure around the housing and not the B2T3 ⫺0·5 ⫺0·1
⫺1·2
⫺6
global behaviour of the foundation. (The PPTs at skirt tip B1T1
⫺5 0·0 0·0
level were damaged during the accidental dynamic installa- ⫺400 ⫺300 ⫺200 ⫺100 0
tion and no data were available for test B1T4.) ⫺4 ∆u from pairs of ⫺0·8
Figure 10 shows excess pore pressures and change in total ⫺3 B1T1 PPTs at skirt tip
B2T1
vertical stress from the PPTs and TPTs located beneath the ⫺2 ⫺0·4
d/D B1T5
base plate during undrained uplift. During all tests a gradual ⫺1 0·15
drawing in of water into the cavity occurred from the start 0
0·30
0·0
of the test and the PPT and TPT readings are similar.
⫺450 ⫺350 ⫺250 ⫺150 ⫺50 50
Some excess pore pressure was present under the base ⫺400 ⫺300 ⫺200 ⫺100 0
plate at the start of uplift, particularly when consolidation
Excess pore pressure, ∆u: kPa
had taken place at higher stress levels. Comparing the TPT
measurements beneath the base plate with the load cell Fig. 9. Pore pressure generation at skirt tip during undrained
readings (Fig. 7) provides an indication of the proportion of uplift
EXPERIMENTAL STUDY OF UPLIFT RESISTANCE OF SHALLOW SKIRTED FOUNDATIONS IN CLAY 533
⫺8 ⫺1·6 ⫺3·0
G1 V/Vu ⬃ 0·6
1·6
⫺0·3
2_3
⫺6 ⫺1·2 B1 2
B2T3 1·2 B2T2 5_
2
B1T2_1
⫺2·0
B1T
T
T2_
⫺0·2
B1
_1
⫺5 0·8
B1
B1T3
⫺0·1
0·4 B1T4 ⫺1·5 V/Vu ⬃ 0·3 F1
⫺4 ⫺0·8
B2T2 0·0 B1T1 0·0 A C
⫺3 ⫺100 ⫺80 ⫺60 ⫺40 ⫺20 0 20 40 ⫺1·0 B
Vu: uplift capacity at w/D ⫽ 0·02
⫺2 B1T1 d/D ⫺0·4 D
B1T5 ⫺0·5 (see Fig. 7)
0·15 D B1T5_3
B2T1 0·30
⫺1
PPT-lid
0·0 V/Vu ⬃ 0·1
TPT-lid
0 0·0
0 40 80 120 160 200 240 280 320 360
⫺140 ⫺120 ⫺100 ⫺80 ⫺60 ⫺40 ⫺20 0 20 40
Prototype time, tp: days
Change in total vertical stress (∆σv) and excess pore (a)
presssure (∆u): kPa
⫺3·0
Fig. 10. Change in total stress and excess pore pressure beneath
G2
base plate during undrained uplift ⫺2·5
1_3
with continued uplift. A gradual increase in pressure was
2
T3_
B2T
observed until the end of the other tests. ⫺1·5
B2
Figure 11 shows examples of readings from the TPTs A E B2T3_1
located along the skirt during undrained uplift from tests ⫺1·0
_2
with both foundations. In each case, the negative excess pore
F2 B2T1
pressure generated during transient uplift gives rise to a ⫺0·5
Vu: uplift capacity at w/D ⫽ 0·02
temporary increase of total radial stress. A similar magni- V/Vu ⬃ 0·30
(see Fig. 7)
tude of radial stress was recorded by TPTs located at the 0·0
same level in the two models and similarity of readings from 0 40 80 120 160 200 240 280 320 360
each diametric pair of TPTs confirmed verticality of the Prototype time, tp: days
foundation during uplift. (b)
Fig. 12. Time histories of sustained uplift tests: (a) d/D 0.15;
(b) d/D 0.3
Sustained uplift response: load and displacement time
histories uplift capacity. For the longer skirted foundation, d/D ¼ 0.3,
Time histories for each sustained load test are presented the undrained uplift resistance Vu mobilised at a relative
in Fig. 12 in terms of relative uplift displacement against displacement w/D ¼ 0.02 corresponded to approximately
prototype time for a year following application of the uplift 75% of the ultimate undrained uplift capacity. Expressed
load. with respect to the ultimate undrained uplift capacity, the
Eleven sustained load tests were carried out; seven with proportional uplift loads labelled as 0.3Vu and 0.6Vu in Fig.
the shorter skirted foundation, d/D ¼ 0.15 (Fig. 12(a)) and 12 correspond to 0.22VuULT and 0.45VuULT respectively (a
four with the longer skirted foundation, d/D ¼ 0.3 (Fig. 0.1Vu load was not investigated for the foundation with
12(b)). The position in a sequence of the sustained uplift test embedment ratio d/D ¼ 0.3).
is indicated by the number following the underscore in the The displacement response to a sustained load was essen-
test label, for example B1T2_1 is the first sustained test in tially bilinear over the range of displacement considered,
the series at that site, followed by tests B1T2_2 and B1T2_3 with immediate displacement followed by time-dependent
respectively at the same site. displacement at a constant rate.
Sustained loads of 10%, 30% and 60% of the undrained The magnitude of immediate displacement increased with
uplift resistance, Vu , mobilised at a relative displacement w/ increased magnitude of load, as would be expected. A
D ¼ 0.02, were considered. For the shorter skirted founda- reduction in immediate displacements may usually be ex-
tion, d/D ¼ 0.15, this corresponded to the ultimate undrained pected with increasing consolidation stress as a result of
increasing stiffness. For uplift following consolidation the
⫺2·0 ⫺0·4 relationship is not straightforward, however, as the magni-
Prototype uplift displacement, w: m
B1T5_1 tude of the stress increment during the load path reversal
B2T1_1
Uplift displacement, w/D: %
⫺1·6 TPTd/D⫽0·15 increases with higher consolidation stress. This may account
⫺0·3 for the large immediate displacements observed in some of
⫺1·2 the tests carried out from the higher consolidation stress
TPTd/D⫽0·3
TPTd/D⫽0·3
⫺0·2
levels, particularly for the tests involving higher levels of
(upper level)
(lower level) uplift load for which the stress increment was large
⫺0·8
(B2T3_2). Tests without a period of consolidation prior to
⫺0·1 uplift did not exhibit immediate displacements (B1T2_2,
⫺0·4 d/D B1T2_3 and B1T5_4.)
0·15
0·30 Repeatability of the tests was confirmed by comparison of
0·0 0·0 the time–displacement response of tests in which conditions
0 10 20 30 40 50 were identical, for example, tests B1T2_1 and B1T3_1 (A in
Total radial stress, σr: kPa Fig. 12(a)), were both carried out with the shorter skirted
foundation, d/D ¼ 0.15, subjected to an uplift load of 0.6Vu
Fig. 11. Radial stresses on skirts during undrained uplift and in both cases were the first test in a sequence (_1).
534 GOURVENEC, ACOSTA-MARTINEZ AND RANDOLPH
Effect of consolidation. The beneficial effect of prior under the higher load of 0.6Vu (G1 and G2 in Figs 12(a) and
consolidation on sustained uplift capacity is evident compar- (b) respectively, that is comparing tests B1T2_1 or B1T3_1
ing the time–displacement response from tests B1T2_2 or with B2T1_3).
B1T2_3 and B1T3_2, (B in Fig. 12(a)), which were both
carried out with the shorter skirted foundation, d/D ¼ 0.15
and subjected to an uplift load of 0.3Vu. Tests B1T2_2 and Pore pressure and radial stress response. Figures 13, 14 and
B1T2_3 immediately followed previous load tests (i.e. 15 respectively show time histories for the year following
without an intermediate period of consolidation) whereas application of the uplift load of the change in vertical stress
test B1T3_2 followed a 12 year period of consolidation and excess pore pressure measured under the base plate, the
(T ¼ 0.083). Displacement occurred around three times more excess pore pressure measured at skirt tip level, and the
slowly when a period of consolidation had preceded uplift. change in radial stress for two selected tests, B1T2_2 and
Limited additional benefit was observed with continued B2T1_2, for d/D ¼ 0.15 and 0.3 respectively with V ¼ 0.3Vu
increase in the period of consolidation prior to uplift for
T . 0.083. This is evident by comparing tests B1T3_2 (as
⫺40
Effect of magnitude of uplift load. The effect of the magnitude
of the uplift load on the rate of displacement was more 0 40 80 120 160 200 240 280 320 360
significant for the foundation with the smaller embedment Prototype time, tp: days
ratio, as would be expected. Considering the shorter skirted
foundation, d/D ¼ 0.15, under proportional loads of 10%, 30% Fig. 13. Change in total stress and excess pore pressure beneath
and 60% of the undrained uplift resistance Vu (D in Fig. 12(a), base plate during sustained uplift
i.e. tests B1T5_3 with B1T5_2 or B1T3_2 and B1T2_1 or
B1T3_1), a 15-fold increase in the rate of displacement is 20
observed with a 20% increase in load, from 0.1Vu to 0.3Vu , and
10 ∆u from pairs of
a further 15-fold increase in the rate of displacement with a
Excess pore pressure ∆utip: kPa
30
25
Effect of embedment ratio. The effect of embedment ratio on
sustained uplift capacity is evident by comparing the response 20 d/D ⫽ 0·30
d/D ⫽ 0·15 B2T1_2
of the two foundations subjected to the same proportional 15 B1T2_2 TPTs
load, V/Vu. For example, tests B1T3_2 or B1T5_2 (F1 in Fig. (upper level)
12(a)) carried out on the shorter skirted foundation, d/ 10
D ¼ 0.15, with B2T1_2 (F2 in Fig. 12(b)) carried out on the 5
longer skirted foundation, d/D ¼ 0.3, in both cases subjected
0
to an uplift load of 0.3Vu. In this case a doubling of the 0 40 80 120 160 200 240 280 320 360
embedment ratio led to a three-fold reduction in the rate of Prototype time, tp: days
displacement. A 50% greater reduction in the rate of
displacement was observed between the two foundations Fig. 15. Radial stress during sustained uplift
EXPERIMENTAL STUDY OF UPLIFT RESISTANCE OF SHALLOW SKIRTED FOUNDATIONS IN CLAY 535