Content uploaded by Aljoscha Ganal
Author content
All content in this area was uploaded by Aljoscha Ganal on Sep 22, 2022
Content may be subject to copyright.
Centrifuge tests on foundations under alternating loads
in overconsolidated clay
1Aljoscha Ganal
1
, 2SW Jacobsz, 1Oliver Reul
1Department of Geotechnical Engineering, University of Kassel, Germany, a.ganal@uni-kassel.de
2Department of Civil Engineering, University of Pretoria, South Africa, sw.jacobsz@up.ac.za
ABSTRACT
To investigate the time-dependent load-deformation behaviour of foundations under alternating loads in overconsoli-
dated clay, centrifuge tests on rafts and piled rafts in Kaolin clay are carried out. The foundations are subjected to
unloading and reloading processes as well as to groundwater lowering, simulating typical loading scenarios of struc-
tures in an urban environment. In the scope of this paper initial results are presented and are discussed in the light of
previous research on the distribution of pile resistances within piled rafts and the influence of the consolidation process.
Keywords: Centrifuge modelling, overconsolidated clay, piled raft, consolidation, pile resistances
1 INTRODUCTION
The prediction of the time-dependent load-settlement
behaviour of foundations in clayey soils is particularly
important when existing and new foundation elements
are to be integrated, which is becoming increasingly im-
portant, especially in inner-city areas (Butcher et al.
2006). In order to be able to exploit further optimization
potential by coupling existing foundations with new
buildings, the time-dependent deformation behaviour of
foundations caused by both consolidation processes and
the time-dependent material behaviour (creep) of soils
must therefore also be taken into account.
In addition to the time-dependent deformation behav-
iour of the soil and variable building loads resulting from
demolition and reconstruction phases, construction ac-
tivities in the surrounding area also influence the long-
term deformation behaviour of foundations. In particular,
the influence of neighbouring groundwater drawdowns
has to be considered in this context, resulting in alternat-
ing stresses due to changes in the uplift and the effective
stresses in the subsoil.
To investigate these topics, centrifuge tests on rafts
and piled rafts in overconsolidated Kaolin clay were car-
ried out. Further research activities in terms of this pro-
ject include the evaluation and back-analysis of available
long-term field measurements on high-rise buildings in
Frankfurt Clay (Franzen & Reul 2022).
This paper presents the results of the first centrifuge
test on a piled raft which was intended to enable optimi-
zation of the remaining investigation program.
1
Corresponding author. email: a.ganal@uni-kassel.de
2 CENTRIFUGE TEST
2.1 General remarks
The centrifuge test described below has been carried
out in the beam centrifuge at the University of Pretoria
with a platform radius of 3 m capable of spinning up to
1500 kg to an acceleration of 100g (Jacobsz et al. 2014).
The test program comprised centrifuge tests on verti-
cally loaded piled rafts and raft foundations under alter-
nating vertical loads and varying ground water (gw)
level, respectively. This paper focuses on the first test
carried out on a piled raft with a constant gw level, while
the results achieved for other configurations will be the
subject of future publications.
All tests have been executed at a nominal centrifugal
acceleration of 80g. The results presented throughout the
paper are at model scale.
2.2 Experimental setup
The general test layout is shown in Fig. 1 in a cross
section and a ground plan. The raft, a square, 150 mm
150 mm aluminium plate (thickness tr =16 mm) was
placed in a strong box with a ground area of 400 mm
600 mm. Applying the definition by Horikoshi & Ran-
dolph (1997) the raft-soil stiffness ratio can be estab-
lished to Krs 100 indicating an essentially rigid raft.
Beneath the raft 16 piles were installed in a 4 4 grid
(spacing of e = 4dp). The model piles comprised smooth
aluminium tubes (wall thickness tp = 1.25 mm) with a
diameter of dp = 10 mm and a length of Lp = 150 mm.
The pile heads fitted into recesses in the raft resulting in
a rigid connection in horizontal direction.
Fig. 1. Test layout.
To apply the load, five brass plates were lowered sep-
arately onto the raft with the help of a linear actuator (Fig.
1e, f). To monitor the load actually applied to the raft,
the weight of the brass plates carried by the load frame
was measured by a load cell, i.e. the load P applied on
the raft is the difference between the weight of all brass
plates and the force measured by the load cell.
The soil stratigraphy in the centrifuge tests comprised
10 mm of sand (Sand 1) to ensure a proper contact be-
tween raft and soil, followed by 50 mm of overconsoli-
dated clay (Clay 1), 20 mm of sand (Sand 1) and over-
consolidated clay (Clay 2) extending to a depth of
220 mm. A 50 mm thick sand layer (Sand 3) at the base
of the strong box, separated from Clay 2 by means of a
thin geotextile, served as a drainage layer. The water ta-
ble was kept constant at 20 mm below ground level.
The instrumentation in the tests included four pie-
zometers in Clay 1 and Clay 2 to monitor the consolida-
tion process and three piezometers located in Sand 2,
Sand 3 and in the standpipe used to control the ground-
water level. It should be noted that the indicated piezom-
eter depths may differ from the final position, as they
have been installed into the slurry clay.
The vertical displacements of the raft were measured
by means of four LVDT displacement transducers. An-
other two displacement transducers were used to meas-
ure the surface displacements.
Eight piles were equipped with one pair of strain
gauges at the pile head and one pair of strain gauges at
the pile base to derive the axial load transferred to the
piles using the mean axial strain and the cross sec-
tional stiffness EA (Fig. 1c, d). In the scope of this paper,
pile loads at the pile head and pile base are termed pile
resistance R and pile base resistance Rb, respectively.
The pile shaft resistance Rs can be calculated as the dif-
ference between pile resistance and pile base resistance.
The piles equipped with strain gauges were chosen so
that at least two piles of each position (corner, edge, cen-
tre) have been instrumented (Fig. 1b).
2.3 Model preparation
In the preparation of the model, a drainage layer of
sand was placed at the base of the box with a thin geo-
textile on top. Clay layer 2 was consolidated from a Ka-
olin slurry with a water content of 100 %, initially using
weights gradually applied to 12 kPa. To produce an over-
consolidated sample, a stress of p = 246 kPa was subse-
quently applied in several steps to the surface using a hy-
draulic press. During the consolidation, both the pore
pressure and the settlement of the sample were moni-
tored. The load was increased once the excess pore pres-
sure had dissipated and the settlements completed. After
the consolidation process monitored by means of pie-
zometers was completed, Sand 2 was rained at an esti-
mated dry density of d = 1700 kg/m³. The procedure
was then repeated for layers Clay 1 and Sand 1. The
overconsolidation OCR ratio profile from this temporary
loading is shown in Fig. 2a for a model at 80g.
To ensure internal equalisation of water heads in the
different soil layers, vertical sand drains were installed
in the four corners of the box.
2.4 Soil properties
The strength characterisation of the sample was un-
dertaken immediately after the centrifuge test by means
of a cone penetrometer (Fig. 2b). The penetrometer test
was carried out at 80g after a short stop of the centrifuge
to remove the actuator. Assuming a correction factor of
Nkt = 15 the undrained shear strength averages approxi-
mately su 29 kPa for Clay 1 (below gw) and su 74 kPa
for Clay 2, respectively.
The properties of the Kaolin clay used to prepare the
clay layers are listed in Table 1. The properties of the
sand are summarized in Table 2.
Fig. 2. Soil profile.
Table 1. Properties of the Kaolin clay.
Parameter
Saturated unit weight
[kN/m³]
16.6
Plastic limit
PL
[%]
36
Liquid limit
LL
[%]
47
Percentage of particles d < 0.002 mm
-
[%]
31
Activity
IA
[-]
0.35
Critical friction angle
[°]
27
Compression index
Cc
[-]
0.129
Swelling index
Cs
[-]
0.030
Viscosity index
Iv
[-]
0.01
Table 2. Properties of the sand.
Parameter
Mean grain size (mm)
d50
[mm]
0.283
Effective grain size (mm)
d10
[mm]
0.138
Critical friction angle (°)
c
[°]
33
Coefficient of uniformity
CU
[-]
2.32
2.5 Testing procedure
After the sample had been completed, the piled raft
was jacked into the soil as a group at 1g using a press. It
is assumed that the behaviour of ‘pre-jacked’ piles is
comparable to bored piles in the field (Li et al. 2010).
Water was only added to the model shortly before
starting the centrifuge to prevent excessive swelling of
the Kaolin. The water was added to the model via a
standpipe connected to the strong box. To keep the gw
level constant throughout the test, water was added con-
tinuously to the standpipe, with an overflow ensuring the
desired height.
In the test presented in this paper the gw level was
held constant while the foundation was loaded by means
of the brass plates lowered subsequently onto the raft.
The loading scheme included a temporary unloading and
reloading sequence and two consolidation phases where
the load was held constant over a longer time period as
summarized in Table 3.
3 RESULTS
3.1 General remarks
Before starting the loading sequence, the sample was
allowed to consolidate under its weight with the devel-
opment of settlements and the dissipation of excess pro-
cess monitored. During this process, the centrifuge had
to be stopped and restarted twice unscheduled which is
believed to have increased the settlements of the founda-
tion significantly. In the remainder of this paper, all re-
sults presented refer to the start of the main loading se-
quence, i.e. to the end of the consolidation process.
Fig. 3 shows the development of the load on the foun-
dation due to the plates successively lowered onto the
raft. In the first two loading phases the load decreased
shortly after the lowering of the respective brass plates
due to problems in the load application. From loading
phase 3 on the brass plates had been adjusted to achieve
a correct load transfer to the raft. During the loading in
phase 8, the full weight of a fifth brass plate could not be
fully applied.
Table 3. Loading scheme.
Phase
Time t [min]
Load p [kPa]
1
loading
20.5
65
2
loading
20.5
115
3
loading
20.5
185
4
loading
20.5
250
5
consolidation
328.5
250
6
unloading
20.5
185
7
reloading
20.5
250
8
loading
20.5
275
9
consolidation/creep
821.2
275
Fig. 3. Variation of load with time.
Fig. 4. Variation of settlements with time.
3.2 Settlements
The variation of settlements with time is plotted in
Fig. 4 for all six displacement transducers. While the
sensors S5 and S6 located on the soil surface show more
or less identical settlements, the other sensors placed on
the raft indicate relatively large differential settlements
which might have been caused by the load plates not be-
ing lowered exactly vertically.
Although the load increments amount to p =68 kPa
in each loading phase, the increase in settlement became
significantly larger as the load increased. This may be
due to the overconsolidation ratio of the soil approaching
OCR = 1 during the loading process. It is also interesting
to note that during the unloading phase only a small
amount of heave of the raft can be observed.
3.3 Pore pressures
Fig. 5 shows the variation of pore pressures with time
for the piezometers T2, T5 and T7 in the sand and the
standpipe, respectively, and piezometers T1 and T6 in
Clay 2. T3 and T4 in Clay 1 proved not functional.
The decrease of pore pressure at T5 and T7, which
shows the hydrostatic water pressure at their respective
depths, was caused by an inadequate water supply to the
standpipe which was fixed after 480 min, resulting in an
illustrated increase in pore pressure.
The several loading phases applied is evident from
the buildup of excess pore pressures in the clay (T1, T6).
Fig. 5. Variation of pore pressure with time.
Fig. 6. Variation of pile resistances with time.
Fig. 7. Variation of total resistance, raft resistance and sum of all
pile resistances with time.
3.4 Pile resistances
In-situ measurements (e.g. Sommer et al. 1985) and
numerical simulations (e.g. Reul 2004) indicate that the
pile resistance, i.e. the pile head load, of a pile in a piled
raft strongly depends on the position of this pile in the
pile group. Under working load conditions the pile re-
sistance can be expected to increase from centre piles to
edge piles to corner piles which is confirmed in Fig. 6 by
the results of the centrifuge tests. Fig. 6 also separates
the weighted average pile resistance Ravg into the contri-
bution of the pile shaft resistance Rs,avg and the pile base
resistance Rb,avg. As can be expected for piles in a rela-
tively homogenous soil, the shaft resistance is responsi-
ble for the major share of the pile resistance, amounting
to 81% in phase 1 and 79 % in phase 4.
The variation of total resistance, raft resistance and
sum of all pile resistances Rpile with time is documented
in Fig. 7. For this plot, the raft resistance Rraft was calcu-
lated as the difference between the total resistance Rtot,
which is equal to the measured load applied on the foun-
dation (i.e. the dead weight of the raft not taken into ac-
count), and the sum of all pile resistances Rpile which
has been calculated from the weighted average of the
measured pile resistances.
The load share between piles and raft is quantified by
the piled raft coefficient :
(1)
The measurements confirm previous findings (e.g.
Reul 2004) that the load share between piles and raft is
not constant for a certain piled raft configuration, but
generally decreases with increasing load level. During
the centrifuge test the piled raft coefficient decreased
from pr = 0.83 (phase 1) to pr = 0.69 (phase 4). It is
interesting to note that during a loading phase, i.e. after
the load had been applied and held constant for a period
of time, the piled raft coefficient increased. For example,
in phase 4, the piled raft coefficient increased from pr =
0.67 immediately after the load had been applied within
approximately 20 min to the value of pr = 0.69 men-
tioned above. This phenomenon is a result of the consol-
idation process as predicted based on numerical simula-
tions (Reul 2002).
4 CONCLUSIONS
The centrifuge test data presented in the scope of this
paper shows good agreement with previous research on
the distribution of pile resistances within piled rafts and
the influence of the consolidation process on the bearing
behaviour of piled rafts. In addition, valuable infor-
mation was obtained for process improvement for fur-
ther tests planned within the scope of this research pro-
ject. Future tests will especially focus on the effect of
unloading-reloading sequences and also investigate the
influence of changes in the groundwater level.
ACKNOWLEDGEMENTS
The overall project is funded by the Deutsche For-
schungsgemeinschaft (DFG, German Research Founda-
tion) - RE 3881/4-1.
REFERENCES
Butcher A.P., Powell J.J.M. & Skinner H.D. 2006. Reuse of foun-
dations for urban sites. IHS BRE Press, Bracknell.
Franzen A. & Reul O. 2022. Numerical investigation of the long-
term settlement behaviour of piled rafts in over-consolidated
clay. Proc. 20th ICSMGE, Sydney (accepted for publication)
Horikoshi K. & Randolph M.F. 1997. On the definition of raft-soil
stiffness ratio. Géotechnique 47, 5, 1055–1061
Jacobsz S.W., Kearsley E. & Kock J.H.L. 2014. The geotechnical
centrifuge facility at the University of Pretoria. Proc. 8th Int.
Conf. on Physical Modelling in Geotechnics, 169-174
Li Z., Haigh S.K. & Bolton M.D. 2010. The response of pile
groups under cyclic lateral loads. Int. Journal of Physical
Modelling in Geotechnics 10: 47–57.
Reul, O. 2002. Study of the influence of the consolidation process
on the calculated bearing behaviour of a piled raft. Proc. 5th
NUMGE, Paris, 383-388
Reul, O. 2004. Numerical study of the bearing behaviour of piled
rafts. Int. Journal of Geomechanics, Vol. 4, Issue 2, 59-68
Sommer, H., Wittmann, P. & Ripper, P. 1985. Piled raft founda-
tion of a tall building in Frankfurt clay. Proc. 11th ICSMFE, San
Francisco, 4, 2253–2257
