You are on page 1of 7

INVESTIGATION OF DESIGN ALTERNATIVES FOR

A PILED RAFT CASE HISTORY

By R. P. Cunha,1 H. G. Poulos,2 and J. C. Small3

ABSTRACT: This paper extends the design philosophy for piled rafts by exploring the factors that control the
design of a published case history where the piled raft was instrumented. The case history studied differs from
most published case histories in that concentrated column loads, rather than average ‘‘spread’’ loads, were
considered in the foundation design. An extensive series of backanalyses was initially carried out with this case
history, in order to calibrate the numerical program adopted herein. These analyses were followed by a parametric
analysis for the evaluation of different design alternatives. These alternatives adopted distinctive pile character-
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

istics (number, location, and length), which were varied for different raft thicknesses, yielding 26 different cases
Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.

for cross comparison and comparison with the ‘‘reference’’ case history. Each of the solutions was also assessed
in terms of relative costs against the reference case, allowing the establishment of conclusions of practical interest
for those involved in the design of piled raft structures.

INTRODUCTION history (Yamashita et al. 1994) in which the final raft defor-
Conventional pile groups are commonly designed by adopt- mations and pile loads were known. Hence, a ‘‘Class C’’ back-
ing a relatively high factor of safety for the piles. These piles analysis is initially carried out to calibrate the numerical tool
are placed in such a manner that they will sustain the entire adopted herein, furnishing ‘‘reference’’ initial values of contact
design load of the superstructure. Although the connection pressures, displacements, pile loads, and factors of safety for
‘‘cap’’ (often a raft) is in close contact with the soil, its con- comparison with the behavior computed from other design al-
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

tribution to the total bearing capacity and general pile group ternatives. The reference design solution and the parametric
behavior is seldom considered. alternatives are also compared in terms of relative costs, yield-
Nevertheless, in the past decade several papers have been ing conclusions of practical interest to those currently involved
published with emphasis on what are now called ‘‘piled-rafts,’’ with the design of both conventional deep foundations and
i.e., pile groups in which the raft connecting the pile heads piled rafts.
(positively) contributes to the overall foundation behavior
(Poulos 1998; Ta and Small 1996; Horikoshi and Randolph ADOPTED CASE HISTORY
1997; Mandolini and Viggiani 1997; Sinha and Poulos 1997; The case history adopted for the present series of back-
Cunha and Sales 1998; Cunha et al. 2000a,b). The Interna- analyses and parametric analyses is that described by Yama-
tional Society for Soil Mechanics and Foundation Engineering shita et al. (1994), and later examined further by Yamashita et
(ISSMFE) also focused the activities of one of its technical al. (1998). These authors described the design of a raft foun-
committees on the study of piled raft cases, gathering valuable dation on a stiff clay deposit, which was constructed to support
information on methods of analysis that is summarized in a standard five-story residential building in one of the suburbs
O’Neill et al. (1996) and Van Impe and Lungu (1996). of Tokyo. The 23 ⫻ 24 m rectangular raft was instrumented
On the other hand, the majority of the aforementioned pa- with earth and pore pressure cells, while some of the piles
pers have considered evenly distributed superstructure loads were also instrumented. Leveling points were also installed
rather than column loads acting on the piled rafts. The latter completely around the raft.
load type is more commonly found in practice, especially for The main characteristics of this case history and its design
typical urban structures. Hence, there is still a lack of under- are given as follows:
standing of the behavior of piled rafts under concentrated load-
ing conditions, as well as of the influence of several design • A design proposal was made to use ‘‘settlement reducing’’
variables on the piled raft performance. piles (Burland et al. 1977) to reduce both total and dif-
This paper explores the design of piled rafts, outlining the ferential settlements. They adopted composite 16 m bored
influence of the major external variables that affect their design piles of 0.8 m in diameter, placed at each of the 20 col-
under concentrated column loads. There are several external umn positions. These piles were founded in a 3-m-thick
variables that influence the design, but the most important ones medium-to-dense sand stratum, behaving as end bearing
are those related to the pile characteristics (number, length, piles with an individual capacity of 2,850 kN.
and disposition), raft characteristics (thickness), and the foun- • The pile loads and the raft settlements were carefully mea-
dation’s overall cost. These variables are incorporated into a sured during the building construction, providing an abun-
parametric analysis performed with the basic data (soil con- dance of data that were used by Yamashita et al. (1994)
ditions, load pattern, etc.) from an existing instrumented case in a Class C prediction. The pile loads were measured
1
when the 12 strain-gauged piles were cast in situ, while
Assoc. Prof., Dept. of Civ. and Envir. Engrg., Univ. of Brası́lia, the settlements were measured with an optical level after
70910-900, Brası́lia/DF, Brazil.
2
Prof., Dept. of Civ. Engrg., Univ. of Sydney, NSW 2006, Sydney,
the raft was cast. At the time of completion of the build-
Australia. ing, the maximum settlements had reached values as high
3
Assoc. Prof., Dept. of Civ. Engrg., Univ. of Sydney, NSW 2006, Syd- as 20 mm, whereas the total pile load carried by the piles
ney, Australia. was on the order of 50% of the design superstructure load.
Note. Discussion open until January 1, 2002. To extend the closing
date one month, a written request must be filed with the ASCE Manager A similar Class C prediction was carried out by the present
of Journals. The manuscript for this paper was submitted for review and
possible publication on April 19, 2000; revised March 12, 2001. This
authors with Yamashita et al.’s experimental data. A hybrid
paper is part of the Journal of Geotechnical and Geoenvironmental finite and boundary element program was adopted, employing
Engineering, Vol. 127, No. 8, August, 2001. 䉷ASCE, ISSN 1090-0241/ similar (but slightly different) geotechnical parameters to those
01/0008-0635–0641/$8.00 ⫹ $.50 per page. Paper No. 22359. of Yamashita et al. (1994). The general characteristics of the
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001 / 635
numerical tool employed here, and the adopted parameters, are TABLE 1. Typical Interaction Factors
discussed next. Pile spacing/diameter Interaction factor

NUMERICAL TOOLS 1 0.410


2 0.321
The Geotechnical Analysis of Raft with Piles program 3 0.268
(GARP6) (Poulos and Small 1998) was adopted to evaluate 5 0.200
7 0.157
the behavior of the reference rectangular piled raft subjected 10 0.113
to vertical loading. This program is based on a simplified form 15 0.068
of boundary element program in which the raft is represented 20 0.042
as a linear elastic plate and the soil can be modeled either as
an elastic-layered continuum or as a Winkler spring medium.
The piles are represented by elastic-plastic springs that can shear wave measurements similarly as presented by Ya-
interact with each other and the raft. Limiting values of contact mashita et al. (1994).
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

pressure (beneath the raft) and pile capacity (shaft friction plus • Water level: This was assumed to be 5.0 m below the
Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.

tip end bearing) can also be specified, and the raft is analyzed ground surface.
using the finite-element technique, rather than via finite dif- • Bearing capacity of pile: The estimated drained capacity
ferences. was 3,000 kN for the 16-m-long piles and 1,800 kN for
GARP6 also considers ‘‘interaction factors’’ between the the 8-m-long piles used in some of the parametric anal-
springs that represent the piles. Such factors are computed via yses.
the use of another well-established software program, DEFPIG • Bearing capacity of raft: The estimated drained capacity
(Deformation Analysis of Pile Groups) (Poulos 1990). It was of the isolated raft was 2,500 kN.
originally written for piles designed using the ‘‘conventional • Pile and raft finite mesh assemblage: A nonsymmetrical
approach,’’ by considering a group of identical elastic piles mesh (because of the number of piles and their location)
having axial and lateral stiffness that are constant with depth. was adopted, with 290 raft elements used in each of the
It also allows for the eventual slippage between the piles and cases.
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

the surrounding soil, and it can take into account the effects
of soil nonhomogeneity along the length of the pile. The stress BACKANALYSES
distributions are computed from the theory of elasticity, more
specifically, from Mindlin’s (1936) solutions for an isotropic, Adopting the same raft dimensions and pile monitoring po-
homogeneous, linear elastic medium. The general character- sitions (horizontal lines A–D and vertical lines 1–5 in Fig. 3)
istics of all numerical analyses carried out herein (backanal- as those presented by Yamashita et al. (1994), a Class C back-
yses and parametric studies) are given below: analysis of the referenced design was carried out and is sum-
marized here.
• Pile characteristics: The elastic pile stiffness was consid- The given column loads were slightly modified in order to
ered as constant during GARP6 (linear elastic) analyses, directly compare the experimental and computed settlements.
and was obtained via DEFPIG analyses with the given This was done because the experimental settlements were ob-
pile length, diameter, and structural Young’s modulus. All tained after the raft was cast in situ, i.e., they do not account
DEFPIG and GARP6 analyses considered an average for the initial settlement of the raft due to its own weight.
Poisson’s ratio of 0.3 for the layered soil, and distinct These loads included the raft weight, which then had to be
values of Young’s modulus for each layer. subtracted from the total load (adopting ‘‘influence’’ areas un-
• Raft characteristics: The Young’s modulus of the raft was der each column load) to produce the ‘‘net’’ load that was used
assumed to be 25,000 MPa and its Poisson’s ratio was within GARP6 backanalyses. On the other hand, the possibil-
taken as 0.2. A constant raft thickness of 0.3 m was used ity of load transfer along the sides of the foundation beam was
for the backanalyses, also with values of 0.6 and 1.2 m not considered here, given the lack of information on the ge-
in the parametric analyses. The latter (1.2 m) value turns ometrical and construction details of this raft. This seems to
the raft from a ‘‘fully’’ flexible slab to a slab of inter- be an important aspect, since it led to lower loads at the pe-
mediate flexibility, according to the definition of stiffness ripheral piles in Yamashita et al.’s analyses, hence closer val-
given by Fraser and Wardle (1976). The base of the raft ues of computed and measured loads.
was in full contact with the underlying soil. The Young’s modulus of the soil (Es) was varied from 0.1
• Column loads on raft: The same loads as presented by to 0.3 of the maximum (low strain Emax) value, which was
Yamashita et al. (1994) were used. determined from the given shear wave velocities at the site.
• Interaction factors: These were obtained via the DEFPIG Fig. 1 presents a comparison between measured and com-
analysis, and used within GARP6 for pile/pile and pile/ puted settlements, while Fig. 2 presents a similar comparison
raft settlement interactions. These interactions were lim- in terms of pile loads. This was done for all 12 instrumented
ited to a horizontal spacing equal to the length of the pile piles, and the following observations can be given:
adopted in each particular case. In the case of piles with
dissimilar lengths (in some of the parametric analyses) the • As with the Yamashita et al. (1994) backanalyses, the case
simplified procedure put forward by Hewitt (1988) was with Es = 0.2Emax seems to be the one that gives the better
adopted in order to derive the interaction factors. As an agreement between measured and predicted values.
example, Table 1 presents typical values of the interaction • The analyses were quite sensitive to the adopted Es values,
factors for one of the parametric analyses. suggesting that the input soil variables can greatly influ-
• Raft/raft and raft/pile interactions: These were obtained ence the accuracy of the numerical analysis.
via the Boussinesq elastic equations by adopting an elastic • The peripheral piles had computed loads well above the
continuum model. measured values, because of load transfer at the edges as
• Soil underneath raft: The soil deformations were limited explained before.
to a total depth of 40 m, where a hard silt layer is located.
A constant drained Young’s modulus was considered for The case in which Es = 0.2Emax was adopted herein as the
each clay and sand layer, being derived from (low strain) reference design solution, for further comparison with the al-
636 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

FIG. 1. Backanalysis: Comparison of Settlements FIG. 2. Backanalysis: Comparison of Pile Loads

ternate designs. It gave the better agreement between measured


and computed results for this initial prediction, and also fur-
nished good values of maximum pile displacements, loads, raft
inclinations, contact pressures, and factors of safety. Fig. 3
presents the contours of vertical displacement along the raft
for this design, together with the adopted finite-element mesh
of the raft and Yamashita et al.’s monitoring positions.
It can be noted in this figure that the points that underwent
the higher settlements underneath the raft are those associated
with the column load positions (dots in the finite-element
mesh). These 20 column positions dictated the location of the
piles in the design by Yamashita et al. (1994). Thus, the ref-
erence case contains one 16-m-length pile underneath each of
the existing column load positions. This figure also depicts a
concentration of settlements for the central piles. This behavior
was associated with the relatively small thickness of the raft,
i.e., the 0.3 m thickness led to the raft being flexible when
FIG. 3. Raft Details and Contours of Vertical Displacement
compared with the stiffness of the piles, hence, most of the
column loads were absorbed by the piles.
• Raft thickness: The reference value was 0.3 m, with raft
PARAMETRIC ANALYSES thicknesses of 0.6 and 1.2 m.
• Number of piles: The reference case involved 20 piles.
The predicted reference values were compared with equiv-
Other cases considered were that of 10 piles (one pile
alent results from various design alternatives. These alterna-
under each column carrying a net load above 2.5 MN),
tives were simulated by carrying out a parametric analysis with
six piles (centrally located in the area of higher column
this same problem, i.e., by keeping some of the external var-
loads), and 0 piles (raft only).
iables constant (e.g., the column load and the soil conditions)
• Pile/raft configuration: The configuration was related to
while varying the others (number, disposition/length of the
the number of piles and their length, as indicated before.
piles, and the raft thickness).
Eight distinct configurations were used, as depicted in Fig.
The parametric analyses adopted the following variations:
4 by letters A–E.
• Pile length: The original length of 16 m was the reference
value while shorter 8-m-length piles were also analyzed. Therefore, 26 distinct parametric design alternatives were
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001 / 637
2.4-m excavation, and all material costs (grout, reinforced
bars, etc.). It also assumed that the piles were to be excavated
by using continuous flight auger equipment. No price provi-
sion was made for the support of the excavation.

DISCUSSION OF RESULTS
All the parametric results were plotted against the raft thick-
ness, as presented in Figs. 5–8. Fig. 5 presents the results for
maximum settlement, maximum differential settlement, and
minimum inclination parameter n (1/n is the inclination ratio,
or angular distortion, i.e., the differential displacement divided
by the distance between the settlement points). These values
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

are the ‘‘critical’’ values for each of the cases, and may not
Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.

refer to the same position underneath the raft in each condi-


tion, i.e., the position in which the maximum displacement was
observed in one particular case may not be the same for an-
other case. All of the investigated points were directly located
underneath the building columns. The following observations
can be drawn:

• By increasing the raft thickness (t) there is a tendency for


both the maximum displacements and the maximum dif-
ferential displacements to decrease. There is also a ten-
dency to increase the inclination parameter n (thus de-
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

creasing the ratio 1/n).


• For each raft thickness, by decreasing the number of piles
(from the most ‘‘piled’’ 20P-16 m(A1) case to the Raft
Only case), there is a tendency to increase both the max-
imum displacements and the maximum differential dis-
placements. An exception is the case of t = 1.2 m, where
the maximum displacement is almost constant and the

FIG. 4. Adopted Piled Raft Configurations

cross compared and analyzed against the values of the refer-


ence design solution of the previous item (which represents
Yamashita et al.’s case history). A nomenclature was devised
in order to plot the results for each of these cases, following
the code ‘‘**P-## m(config),’’ where the initial asterisks rep-
resent the number of piles, the following hatches denote the
pile length(s), and ‘‘config’’ denotes the piled raft configura-
tion. This nomenclature has been slightly changed for config-
urations D and E, by respectively adopting the codes ‘‘20P-8/
16m(D)’’ and ‘‘20P-8/16m(E).’’ The case with no piles was
denoted as ‘‘Raft Only,’’ and the reference case refers to the
alternative 20P-16m(A1) when the raft thickness is 0.3 m.
A comprehensive cost analysis was additionally carried out
with the design data (geometry of the raft/piles, soil charac-
teristics, etc.) in order to evaluate the relative costs of all al-
ternate designs, with respect to the reference cost. This anal-
ysis was done by an Australian foundation contractor, and
included the pile and raft construction, the backfilling behind FIG. 5. Displacement, Differential Displacement, and Inclination An-
the raft, the survey of the piles, any spoil removal due to the gle

638 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001


No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

FIG. 7. Pile Load Sharing and Factor of Safety


FIG. 6. Pile Load, Contact Pressure, and Average Contact Pressure

maximum differential displacement slightly decreases and


stays constant with the decrease of the number of piles.
This occurs because of the higher stiffness of the raft in
this condition, which tends to reduce and even suppress
any variation in settlement among the points underneath
the raft.

The reference design yielded a maximum settlement and a


differential settlement of 24 and 15 mm, respectively, with an
inclination ratio of 1/800. These values were considered ac-
ceptable by Yamashita et al. (1994), and will be used here as
‘‘benchmarks’’ for comparison purposes. Therefore, the fol-
lowing comments can be made based on the adopted raft thick-
nesses (t): FIG. 8. Relative Costs of All Design Alternatives

• t = 0.3 m: The unpiled case yielded a maximum displace- denoted as 20P-8/16m(D), with 10 piles of 16 m and 10
ment and a maximum differential displacement that was of 8 m in length underneath the raft.
around two times the reference values, with an inclination • t = 0.6 m: All the alternatives yielded satisfactory values
ratio of 1/400. These results are close to those presented of the inclination parameter n (i.e., above 500), but only
by Yamashita et al. (1994), and substantiate their conclu- a few of them yielded maximum settlements below 25
sion that the unpiled raft could not be accepted as a design mm. By considering this value (i.e., 25 mm maximum
alternative. They reached this conclusion on the basis of settlement, n > 500) as the design criteria, then only the
the computed maximum settlement (which was above 25 alternatives 20P-16m(A1) and 20P-8/16m(D) would be
mm) and on the basis of the inclination ratio (which was acceptable. However, if the maximum design settlement
above 1/500). If such values are adopted as the limiting is increased to 30 mm, as before, then all the alternatives
design values, then Yamashita et al.’s design choice was for this raft thickness would be acceptable. The ‘‘opti-
the right one for this particular raft thickness, since in mum’’ design would then be related to other aspects, such
none of the other alternatives was the maximum settle- as the overall cost, and so forth.
ment below or equal to 25 mm. However, if the maximum • t = 1.2 m: All the alternatives yielded satisfactory values
design settlement limit is slightly increased to 30 mm then of both the inclination parameter n (above 500) and max-
another design alternative can be accepted. This is the one imum settlement (below 25 mm). Hence, all the solutions
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001 / 639
could be, in principle, adopted in practice. Nevertheless, Regardless of the previous comments, some of the piled raft
the optimum design would also have to take into consid- configurations are conservative, since some of them had fac-
eration the cost aspect. tors of safety for the pile group above 3 and overall factors of
safety above 6. This result clearly indicates that the main pur-
Fig. 6 presents the computed results for the maximum pile pose of inclusion of the piles was to reduce the differential
loads, the maximum contact pressures, and the average contact and the total settlements, rather than to increase the capacity
pressures. As before, these values are the critical values for of the foundation system.
each of the cases, and again, do not necessarily refer to the All the design alternatives were also compared in terms of
same position underneath the raft. The maximum contact pres- their relative costs with respect to the reference design solu-
sures were obtained by ‘‘filtering’’ the results for each of the tion. In the reference case, a cost of unity was adopted, and
finite elements underneath the raft, and refer to the contact Fig. 8 presents the relative costs of all alternatives. This figure
pressure at the soil/raft interface. The following observations shows that, for each raft thickness, decreasing the number of
can be drawn from this figure: piles (from the most piled 20P-16 m(A1) case to the Raft Only
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

case), decreases the associated cost of the foundation. More-


Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.

• By increasing the raft thickness there is a tendency to over, by increasing the raft thickness, there is a tendency to
decrease both the maximum pile load and the maximum increase the overall price of the design solution. For instance,
contact pressure. The increase in raft thickness, however, by increasing the raft thickness from 0.3 to 0.6 m, there will
has little or no influence on the average pressure under- be an increase of around 26% on the average cost of the 0.3-
neath the raft, although it influences the (maximum) lo- m solutions.
calized values of contact pressures. By also taking account The present exercise has demonstrated, so far, that distinct
of the results from Fig. 5 (displacement), one may con- design solutions can be adopted depending on the design lim-
clude that the maximum pile load is directly proportional itations and assumed tolerances. Hence, if it is assumed that
to the maximum raft displacement. the maximum inclination ratio of any point of the raft must
• For each raft thickness, by reducing the number of piles, be lower than 1/500, the maximum factor of safety of the
there is a slight tendency for both the maximum and the isolated pile group must be lower than 3, the minimum overall
average contact pressures to increase. However, the values
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

safety factor must be higher than 3, and the relative cost must
of average contact pressure at the soil/raft interface are be lower than or equal to 1, then few solutions will be accepted
well below the long-term drained capacity of the surface depending on the adopted value for the maximum settlement
clay, indicating that all the alternative solutions could be underneath the raft.
accepted in practice. If all these criteria are met, and this maximum settlement is
• The maximum pile load, for a specific raft thickness, is limited to 25 mm, then only Yamashita et al.’s reference so-
mainly dependent on both the number and length of the lution could be adopted, i.e., only the piled raft solution with
piles in the raft. the 0.3-m-thick raft on top of 20 piles of 16-m length. On the
• For all the cases (with the exception of one pile in the other hand, if all of the criteria are met and it is assumed that
10P-8m(B2) configuration, t = 0.3 m), the piles have not the maximum settlement could increase to 30 mm, then several
reached their ultimate capacity. The maximum pile loads, other solutions could be adopted, as presented in Table 2. This
for each respective pile length, were well below the fail- table also indicates the relative cost of such alternatives and
ure values. These values were also lower than the struc- the percentage of cost saving in relation to the reference case.
tural failure value of the composite pile, estimated to be It is clearly noted that a higher cost saving may be obtained
around 4,500 kN. depending on the adopted solution. In fact, the most econom-
ical solution in this case is a raft only, without any piles. Nev-
Fig. 7 presents the results of the pile load sharing and the ertheless, it should be pointed out that, although all solutions
factors of safety. It is noted that the load carried by the piles of Table 2 meet the predefined criteria, they yielded slightly
continuously decreases with an increase in raft thickness, vary- different results of settlement, pile load, raft pressure, and so
ing from maximum values of about 60% for t = 0.3 m to forth.
around 40% for t = 1.2 m. This means that by increasing the It is finally emphasized that all aforementioned comments
raft thickness more load is absorbed by the raft. It is also are valid only for the conditions analyzed herein, and may not
observed that the decrease of this load share is proportional to reflect the optimum design solutions under different soil con-
the decrease of the number of piles in each of the configura- ditions, raft geometries, and/or load patterns. Moreover, the
tions (for a particular raft thickness). As expected, the lower presented costs are current, and may vary in the future.
the number of piles for a particular raft thickness or configu- Nevertheless, this exercise highlights the fact that, whenever
ration, the lower will be the percentage of the load carried by possible, several alternatives should be considered for the op-
the piles. timum design of a piled raft foundation under a particular set
However, the total capacity of the pile group (isolated with- of soil, load, and pile/superstructure conditions. Local stan-
out raft) will also change by varying the number of piles in dards and practice are also important for this ‘‘optimizing’’
each of the adopted configurations. Therefore, the overall ca- design process, since they give the benchmarks and the lim-
pacity (pile group plus raft) will also change, since it is con- iting criteria one should apply to filter the distinct alternatives.
sidered to be the lesser value of (1) the sum of the pile group
and the raft capacities; or (2) the piled raft ‘‘block’’ failure TABLE 2. Costs of Cases That Satisfy All Criteria
mode capacity. The changing values of factor of safety from
Raft Percentage of
this figure reflect the aforementioned variation of the geo-
Design thickness Relative cost saving
technical capacity of the pile group, as well as the variation
alternative (m) cost (%)
of the overall (piles plus raft) geotechnical capacity of the
piled raft. Furthermore, there is not a clear trend for the var- 20P-16m(A1)a 0.3 1.0 0
iation of the overall factor of safety with the increase in the 20P-8/16m(D) 0.3 0.93 7
10P-8m(B2) 0.6 0.96 4
raft thickness, but by decreasing the number of piles in each 6P-16m(C1) 0.6 0.97 3
of the piled raft configurations (from 20P-16m(A1) to Raft Raft Only 0.6 0.76 24
Only) there is a slight decrease in this latter factor. This ob- a
Reference case.
servation is valid for all raft thicknesses studied herein.
640 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001
CONCLUSIONS of foundations and structures.’’ Proc., 9th Int. Conf. on Soil Mech. and
Found. Engrg., ISSMFE, Japan, Vol. 2, 495–546.
This paper has explored and extended the design philosophy Cunha, R. P., and Sales, M. M. (1998). ‘‘Field load tests of piled footings
for piled raft foundations, by considering a series of design founded on a tropical porous clay.’’ Proc., 3rd Int. Geotech. Seminar
analyses related to a published case history. This case was Deep Found. on Bored and Auger Piles, ABMS, Brazil, Vol. 1, 433–
438.
considered because it involved discrete column loads, rather
Cunha, R. P., Small, J. C., and Poulos, H. G. (2000a). ‘‘Class C analysis
than a uniformly distributed load, and is more representative of a piled raft case history in Gothenburg, Sweden.’’ Proc., Year 2000
of real conditions in urban construction. Geotechnics—Geotech. Engrg. Conf., AIT, Thailand, Vol. 1, 271–280.
The design alternatives adopted pile characteristics (number, Cunha, R. P., Small, J. C., and Poulos, H. G. (2000b). ‘‘Parametric anal-
location, and length) which were varied against the raft thick- ysis of a piled raft case history in Uppsala, Sweden.’’ Proc., 4th Sem-
ness, yielding 26 different cases for cross comparison and con- inar of Spec. Found. and Geotechnics, ABMS, Brazil, Vol. 2, 381–
trast against the ‘‘reference’’ design described by Yamashita et 390.
Fraser, R. A., and Wardle, L. J. (1976). ‘‘Numerical analysis of rectan-
al. (1994). The reference case was fully instrumented, allowing gular rafts on layered foundations.’’ Géotechnique, London, 26(4),
the settlement and load for some of the piles to be used to
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

613–630.
‘‘calibrate’’ the numerical program used herein. A variety of
Downloaded from ascelibrary.org by UNIVERSIDADE DE BRASILIA on 07/13/12. For personal use only.

Hewitt, C. M. (1988). ‘‘Cyclic response of offshore pile groups.’’ PhD


alternative design solutions were examined in order to study thesis, University of Sydney, Sydney, Australia.
which combinations of parameters led to designs that were less Mandolini, A., and Viggiani, C. (1997). ‘‘Settlement of piled founda-
expensive than the reference solution and also satisfied one tions.’’ Géotechnique, London, 47(4), 791–816.
Mindlin, R. D. (1936). ‘‘Force at a point in the interior of a semi-infinite
particular performance criteria.
solid.’’ Physics, 7.
The reference design adopted by Yamashita et al. (1994) O’Neill, M. W., Caputo, V., De Cock, F., Hartikainen, J., and Mets, M.
involves the use of a pile below each column location, in con- (1996). ‘‘Case histories of pile-supported rafts.’’ Rep., University of
junction with a thin raft. The present analyses have demon- Houston, Houston.
strated that this solution fully satisfied the design requirements. Poulos, H. G. (1990). ‘‘DEFPIG user’s guide.’’ Ctr. for Geotech. Res.,
However, they have also demonstrated that by assessing the University of Sydney, Sydney, Australia.
performance of various combinations of raft thickness and Poulos, H. G. (1998). ‘‘The pile-enhanced raft—An economical foun-
dation system.’’ Keynote Lecture, Proc., 11th Brazilian Congr. on Soil
number and length of piles, it may be possible to develop a
J. Geotech. Geoenviron. Eng. 2001.127:635-641.

Mech. and Geotech. Engrg., ABMS, Brazil, Vol. 4, 27–43.


foundation design that also performs satisfactorily, and is Poulos, H. G., and Small, J. C. (1998). ‘‘GARP6 user’s manual.’’ Coffey
somewhat less expensive than the reference design. Partners International Pty. Ltd, Sydney, Australia.
Although the results are restricted to the conditions of the Poulos, H. G., Small, J. C., Ta, L. D., Sinha, J., and Chen, L. (1997).
analyses, they do highlight the fact that, whenever possible, ‘‘Comparison of some methods for analysis of piled rafts.’’ Proc., 14th
an ‘‘optimized’’ parametric procedure should be considered in Int. Conf. on Soil Mech. and Found. Engrg., ISSMFE, Germany, Vol.
practice for the design of both piled rafts and standard deep 2, 1119–1129.
Randolph, M. F., and Clancy, P. (1993). ‘‘Efficient design of piled rafts.’’
foundations. It allows some insight into the most relevant var-
Proc., 2nd Int. Geotech. Seminar Deep Found. on Bored and Auger
iables that affect the behavior of the foundation, as well as the Piles, U. Ghent, Belgium, Vol. 1, 119–130.
establishment of the ‘‘optimum’’ design characteristics valid Sales, M. M., Cunha, R. P., Farias, M. M., Small, J. C., and Poulos, H.
for the particular soil, load, and superstructure conditions. Lo- G. (2000). ‘‘Comparisons of some programs analysing classical piled
cal standards and practice also need to be considered through- raft problems.’’ Proc., Year 2000 Geotechnics—Geotech. Engrg. Conf.,
out the design process, since they can be used as the limiting AIT, Thailand, Vol. 1, 317–326.
criteria upon which one can base the distinct alternatives. Sales, M. M., Cunha, R. P., and Jardim, N. A. (1999). ‘‘Analysis of piled
footing tests on a tropical porous clay.’’ Proc., 11th PanAmerican Conf.
on Soil Mech. and Geotech. Engrg., ABMS, Brazil, Vol. 3, 1395–1402.
ACKNOWLEDGMENTS Sinha, J., and Poulos, H. G. (1997). ‘‘Piled raft foundation systems in
This research was made possible through a joint technical cooperation swelling and shrinking soils.’’ Proc., 14th Int. Conf. on Soil Mech. and
program from both the Universities of Sydney and Brası́lia. This program Found. Engrg., ISSMFE, Germany, Vol. 2, 1141–1144.
has resulted in the first writer’s sabbatical leave at the former university. Ta, L. D., and Small, J. C. (1996). ‘‘Analysis of piled raft systems in
Therefore, the first writer would like to express his gratitude to the Bra- layered soils.’’ Int. J. Numer. and Analytical Methods in Geomech., 20,
zilian Council for Research and Development-CNPq, for the scholarship 57–72.
he received throughout his period in Australia. Van Impe, W. F., and Lungu, I. (1996). ‘‘Technical report on settlement
The writers also acknowledge the fruitful discussions with some of prediction methods for piled raft foundation.’’ Rep., University of
the staff members of the University of Sydney, in particular, Dr. Tim Ghent, Ghent, Belgium.
Hull, Prof. John Carter, and Prof. S. Reid. Likewise, the aid provided by Yamashita, K., Kakurai, M., and Yamada, T. (1994). ‘‘Investigation of a
J. Adamiec from Wagstaff Piling Inc., in calculating the foundation costs piled raft foundation on stiff clay.’’ Proc., 13th Int. Conf. on Soil Mech.
is most appreciated. and Found. Engrg., ISSMFE, India, Vol. 2, 543–546.
Yamashita, K., Yamada, T., and Kakurai, M. (1998). ‘‘Simplified method
REFERENCES for analysing piled raft foundations.’’ Proc., 3rd Int. Geotech. Seminar
Deep Found. on Bored and Auger Piles, U. Ghent, Belgium, Vol. 1,
Burland, J. B., Broms, B. B., and de Mello, V. F. B. (1977). ‘‘Behaviour 457–464.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 2001 / 641

You might also like