See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/239388900

Load Tests on Residual Soil and Settlement Prediction on Shallow Foundation

Article  in  Journal of Geotechnical and Geoenvironmental Engineering · October 2001

DOI: 10.1061/(ASCE)1090-0241(2001)127:10(869)

CITATIONS READS
21 279

1 author:

António Viana Da Fonseca

University of Porto
284 PUBLICATIONS   2,556 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

LIQUEFACT: Assessment and mitigation of liquefaction potential across Europe: a holistic approach to protect structures / infrastructures for improved resilience to
earthquake-induced liquefaction disasters View project

Stabilization of Sedimentary Soil using Alkaline Cements Applied to the Deep Soil Mixing Technique View project

All content following this page was uploaded by António Viana Da Fonseca on 30 August 2022.

The user has requested enhancement of the downloaded file.

 LOAD TESTS ON RESIDUAL SOIL AND SETTLEMENT PREDICTION
ON SHALLOW FOUNDATION

By A. Viana da Fonseca1

ABSTRACT: The design of shallow foundations is often considered in semiempirical methodologies that are
based on linear and nonlinear models of behavior, mainly for settlement prediction purposes. In this paper, the
applicability of such criteria is discussed, by analyzing the results obtained at an experimental site on a fairly
homogeneous saprolitic soil derived from granite. This included a full-scale load test on a circular concrete
footing together with in situ and laboratory tests. The information obtained in terms of strength and stiffness
was integrated with the aim of refining some of the approaches based on the theory of elasticity. Emphasis was
especially given to semiempirical methodologies based on results of standard penetration tests, cone penetration
tests, plate loading tests, and triaxial tests on high-quality samples with the results from local instrumentation.
Some of the well-established methods were tested and some parametrical and methodological adaptations are
suggested that better fit the observed behavior.

INTRODUCTION and a corresponding geophone unit, also expanded into its

hole. In addition, block samples were taken from test pits and
Jean Kerisel (1985) pointed out that many a proud fortress
pieces carefully trimmed by hand to suit the sizes of oedom-
and many a great city have long since crumbled through being
eters and triaxial apparatus for these laboratory tests to be car-
badly founded. Those ancient structures still standing carry a
ried out. Details of this work have been given by Viana da
message, which is for us to decipher. Before the advent of
Fonseca (1996, 1998), Viana da Fonseco et al. (1994, 1997a,b,
science, there was considerable mystery associated with foun-
1998), and Viana da Fonseca and Cardoso (1999). From the
dations in an attempt to ensure that they would be successful.
results of all these tests, which are summarized in Table 1 and
Foundation ceremonies were performed, with senior persons
shown for the in situ tests in Fig. 1, an area of greatest ho-
called to approve the site and officiate, laying the foundation
mogeneity was chosen for the large loading test of a shallow
stone, so as to give the best chance for a successful building.
foundation.
The ceremony of laying the foundation stone by an important
person has persisted into modern times.
TESTS ON SHALLOW CIRCULAR FOUNDATIONS
But with the development of science, and particularly the
science and art of soil mechanics, methods have been devel- As part of the general testing program, plate loading tests
oped for the design of safe foundations. The strength and de- (PLT) were made with circular steel plates 30 and 60 cm in
formation characteristics of the ground are controlling factors, diameter. The main test was on a 1.2-m-diameter reinforced
and this study is concerned with founding on residual soils, concrete foundation 0.5 m thick, which was loaded from kent-
which cover large areas in various parts of the world. If one ledge by a hydraulic jack. The kentledge of 140 tons was pro-
considers the safe bearing capacity of a circular footing at vided by a water tank 11.2 m in diameter supported by four
shallow depth, one can find several methods that have been steel beams resting on concrete bases placed 4.6 m from the
developed by numerous authors, and it is the purpose of this center of the test area. The middle two beams were spaced 1.2
paper to give the results from loading tests and compare them m apart to give room for the hydraulic jack.
with the values that would have been given by several existing The layout for the main test is shown in Fig. 2 and views
design methods. of the site are given in Fig. 3. A separate beam, independently
supported and shown in Fig. 3(a), was used to hold dial gauges
SITE INVESTIGATION to measure the settlements of the concrete foundation. The
The greenfield site chosen for this study was to be for a layout of the measuring instruments is given in Fig. 4. Short
hospital, and the topsoil had already been removed. The bed- calibrated staffs, made from 2-cm-square steel tubes welded to
rock was granite, in situ weathered to a depth of 6 m to pro- a small base plate, were founded on small pads of mortar at
duce a saprolitic soil. Such soil is very common in the northern the positions shown, and a precise surveyor’s level was placed
part of Portugal where shallow foundations have been de- in a pit so that its collimation level was within the heights of
signed by methods developed for more traditional soils. The the staffs and positioned so that all the staffs could be observed
characteristics of the soil were measured by a range of in situ during the loading test.
tests including standard penetration tests (SPT), static cone Prior to the main loading test, and 30- and 60-cm-diameter
penetration tests (CPT), dynamic probing (DP), and the more steel plates were tested with a hydraulic jack reacting against
sophisticated Ménard pressuremeter (PMT), Marchetti dila- the two outer beams, as shown in Fig. 5. The settlements of
tometer (DMT), and self-boring pressuremeter (SBPT) tests. the plates were measured from the independently supported
In addition, cross-hole seismic tests (CH) were carried out in beam by use of Benkelmann beams. The main loading test
boreholes at 3 m centers, using standard apparatus with a drop- was carried out over a period of 15 days, using 35 increments
weight activator, expanded into the hole by compressed air, of load with each load maintained for 4 h. Full details of the
test are given in Table 2. A plot of settlement with load, shown
1
Asst. Prof., Faculdade de Engenharia, Universidade do Porto, Rua Dr. in Fig. 6, shows that at a maximum loading intensity of 1,000
Roberto Frias, 4050.123 Porto, Portugal. E-mail: viana@fe.up.pt kN/m2, settlement was about 100 mm. This loading on the
Note. Discussion open until March 1, 2002. To extend the closing date footing depressed the ground slightly: at a radius of 0.9 m
one month, a written request must be filed with the ASCE Manager of from the center of the footing the depression was about 10
Journals. The manuscript for this paper was submitted for review and
possible publication on January 2, 2001; revised May 7, 2001. This paper
mm. At a radius of 1.6 m it was about 4 mm and at 2.6 m it
is part of the Journal of Geotechnical and Geoenvironmental Engi- was only 1 mm.
neering, Vol. 127, No. 10, October, 2001. 䉷ASCE, ISSN 1090-0241/01/ The main purpose of this paper is to compare the measured
0010-0869–0883/$8.00 ⫹ $.50 per page. Paper No. 22054. behavior of the footing on saprolitic soil with the behavior that
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 869
 TABLE 1. Strength Parameters (␾⬘ and c⬘) Obtained from Laboratory Testing

would have been predicted by the various design methods that ES = NP /␣P (1)
have been developed by several researchers in various parts
of the world. where N60 = SPT number of blows, allowing for the method
of dropping the driving weight, assuming a 60% efficiency;
and the subscript P refers to Parry.
FOOTING SETTLEMENT EVALUATION USING If the pressure is expressed in kilopascals, the footing di-
SPT RESULTS mensions in meters, and the settlements in millimeters, the
Terzaghi and Peck (1967) parameter ␣P will have the variation range of ␣P = 0.2–0.3. It
was found that over the zone affected by the load test, the
This method for estimating settlements is based on the re- variation of N60 with depth could be expressed as
sults from SPT. This crude test was developed from a method
used by the Reymond Piling Co., in 1912, to obtain samples N60 = 5.16 ⫹ 3.27 ⭈ z(m) (2)
of the soil at the base of their bored piles. A thick walled 5- This method, developed by Parry, agrees reasonably with
cm-diameter steel tube was hammered into the ground to ob- the test results, although only for the lower stress levels, be-
tain the sample, and it was realized that the energy required coming strongly nonconservative for stress levels above the
to cause penetration gave an indication of the strength of the ‘‘elastic’’ threshold. The limitation of this method is mainly
ground. The test was very easy to do because little special governed by the linearity of the depth of the influence zone
apparatus was needed in addition to the heavy well-boring in relation to the dimension of the loaded area, and does not
equipment already being used on the site. Through the years agree with the most recent works (Jardine et al. 1986; Burland
many tests were carried out, but researchers were somewhat 1989; Tatsuoka and Shibuya 1992). There is a risk therefore,
unhappy about the nonscientific nature of the test and were of extrapolating to larger loaded areas (which is usually the
doing their best to have it replaced by a penetration test, such more common case), of overestimating the settlements by cal-
as the Dutch cone test, when Terzaghi and Peck published their culating them on the basis of an average value of N60 over the
semiempirical method for estimating settlement in granular depth of 2B, particularly in grounds that exhibit increasing
materials. But as is recognized today, the predictions are very stiffness with depth.
conservative, and in the present case, the observed settlement
would have been predicted under a load of one-half to one- Burland and Burbidge (1985)
fourth of that actually applied.
This proposed method for settlement calculation uses an av-
erage value of N60 determined over a depth below the footing
Parry (1978) base ( NBB ) through the following expression:
The method of Parry (1978) is based on the classic expres- B 0.7
sion of the theory of elasticity for the calculation of settle- s = ␣BB ⭈ ⭈ qs (3)
NBB 1.4
ments. Being the deformability modulus taken as a function
of an average value of N60 determined in the depth 2B below with ␣BB varying between 0.93 and 3.09, with 1.71 being the
the footing base ( NP ) most probable value. In the expression, B denotes footing
870 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001
 FIG. 1. In Situ Tests: (a) SPT and CPT; (b) CH and DMT

width, qs the average contact pressure, and the subscript BB Table 3 gives values of ␣P and ␣BB that best agree with typical
stands for Burland and Burbidge. values of the experimental settlements that were measured.
This method, using ␣BB = 1.71, was found to be grossly From these values of ␣P and ␣BB, one can conclude that
conservative, giving rise for average service stress conditions
to ratios of 2 to 3 between predicted values and those ob- 1. The approach of limiting the settlement to 25 mm pro-
served. The discrepancy is more accentuated for the 60-cm- duces a reasonable consistency of values in the two
diameter plate than for the 1.2-m-diameter footing, although cases. This is a consequence of assuming a factor 0.7 for
over the lower stress range they are very similar. the minimum size of the loaded area B.
With the purpose of best adapting the methodologies to suit 2. For the same approach, the method of Parry gives values
the experimental results, the values of ␣P and ␣BB were cal- with slight variations for ␣P , resulting in a much greater
culated to obtain convergence for the two, following typical reduction of B (factors of 0.32–0.44) than the maximum
values of settlements: proposed by the present author (0.30).
3. The approach using the elastic threshold level (s/B =
0.75%) which is considered more realistic, confirms the
1. s/B = 0.75%—a level corresponding to a certain elastic good results from the Parry method for loading over rel-
threshold atively small areas; it being noticed, however, that there
2. s = 25 mm—a limit value in accordance with the Ter- is an increasing value of ␣ with increasing foundation
zaghi and Peck proposal size from the plate to the footing, indicating the begin-
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 871
 FIG. 2. Layout for Main Test

ning of a breakdown in the assumption of direct propor- FIG. 3. Views of Main Test: (a) Support Beams; (b) Weight Tank
tionality with B.
4. It should be noted that for this approach, the values ob- different load conditions, mainly on sandy soils (of different
tained for ␣BB were very low (0.63 for the footing, and origins), and proposed the following expression:
0.50 for the plate of 60 cm in diameter) compared with
the initial proposals (0.93–3.09). q nq ⭈ B nB
s=f (4)
N nN
This last verification shows that the soils at the test site
possess a more pronounced structural stiffness than those an- The terms were obtained by multiple regression, with pri-
alyzed by Burland and Burbidge. It should be noticed, how- ority for the dependent variables as a function of the relative
ever, that Burland and Burbidge did not include residual soils influence of each one. It should be noticed that the expression
in their analysis. From the work of Rocha Filho (1986), the of Burland and Burbidge (1985) constitutes a particular case
application of the author’s proposals to the results of loading of this more general one, with f = 0.93–3.09, nq = 1, nB = 0.7,
tests on shallow foundations and plates with diameters from and nN = 1.4.
0.40 to 1.60 m, carried out on residual soils from gneiss at the It is interesting to observe that the parameters proposed by
university campus of Pontifı́cia Universidade Católica in Rio Burland and Burbridge are reasonably similar to these indi-
de Janeiro, resulted in ratios of calculated to observed settle- cated by Anagnastopoulos et al., as representative of all sets
ments between 1.5 and 2.5. The ratios obtained in this study of studied cases. As a compensation, it is to be noted that there
were even larger (2.7–3.4). is a high dispersion of the parameters corresponding to those
Another reason for lack of agreement with the Burland and several classes of stiffness and of the size of the loaded sur-
Burbidge method when applied to residual soils may lie in the face.
fact that the considered influence zone should be smaller than Considering Anagnastopoulos et al. proposals that most
that suggested by them, due to the higher rate of stiffness adapt to the residual soil where the present study was con-
increase with depth caused by the simultaneous increase of ducted, the calculated load-settlement curves for the 1.2-m-
confinement stress as the degree of weathering decreases. diameter footing and the 60-cm-diameter plate are compared
It can be considered that the proposal of Burland and Bur- with observed curves in Fig. 7.
bidge (1985) will be applicable to larger foundations resting Analyzing the results expressed in Fig. 7, it can be con-
on residual soils (for example, B = 3–4 m), readopting the cluded that
average value of coefficient ␣BB. For the case of s/B = 0.75,
one suggests, conservatively, the value 1. On the whole, these proposals are strongly conservative,
overestimating the settlements, as can be seen by com-
␣BB(original) 1.71
␣BB(saprolitic soil from granite) = = = 0.85 paring the theoretical with the observed settlements in
2 2 Figs. 7(a and c).
2. The suggestion that B < 3 m [Fig. 7(c)] does not ade-
Anagnastopoulos et al. (1991)
quately explain the present results, both in absolute terms
Anagnastopoulos et al. statistically processed 150 cases of as well as in the curvature of the respective load-settle-
shallow foundations with several sizes and subjected them to ment plots. This last fact results, naturally, in the value
872 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001
 FIG. 4. Layout of Instruments to Measure Foundation Settlements and Movements of Ground

of nq being substantially smaller than 1; this seems and the best adjustment of the curves was sought to a
strange considering the normal behavior of the stress- bearing pressure qs = 400 kPa.
strain laws. It seems, therefore, to be an unreal exercise 2. A linear answer was admitted (nq = 1) and the conver-
to obtain correlations considering nq factors <1. gence for values of the settlement defined for s/B =
0.75%.
With the purpose of obtaining the parameters that give the best
Fig. 8 shows a comparison between the curves obtained for
agreement with the experimental results, it was decided to as-
those two hypotheses, assuming the following:
sume the premises of Burland and Burbidge (1985) for the
influence depth z = B 0.75 and for the factor nN = 1.40. With 1. Excellent agreement between the theoretical and the ex-
these factors fixed, multiple regressive analyses were carried perimental results were obtained by using the value nq =
out assuming the following variants: 1.23. The resulting nonlinearity causes the value of the
constant f to be reduced to 0.18, which is much lower
1. The nq exponent was admitted higher than 1, which cor- than the value of 0.60 usually defined in linear elasticity.
responds to considering the nonlinearity of the answer, 2. The imposition of a linear relationship, as shown by Fig.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 873
 FOOTING SETTLEMENT EVALUATION USING
CPT RESULTS
Schmertmann et al. (1978)
The semiempirical method of Schmertmann (1970) assumes
a simplified distribution of the influence factor for the vertical
strains under the footing, whose values are computed on the
basis of a deformation modulus (secant) variable with depth,
which can be correlated with the CPT cone resistance
Es = ␣qc (5)

where Es represents Young’s modulus related to Schmertmann;

and qc represents static cone resistance.
Schmertmann et al. (1978) found ␣ = 2.5 for axisymmetric
conditions in granular soils. It should be remarked that, in spite
of using a constant modulus for a given depth, Schmertmann’s
method provides a nonlinear pressure-settlement curve, since
the strains depend on the ratio between the incremental pres-
sure and the initial effective vertical stress at the foundation
level.
This method was applied to the footing load test by adjust-
ing the ␣ parameter in order to fit the observed curve. As
shown in Fig. 9, the best agreement was achieved with values
of 4.0–4.5. These relatively high values should probably be
attributed to the influence of the cemented structure of the
saprolitic soil, being situated in the range referred to by Rob-
ertson and Campanella (1988) as applicable for old, overcon-
solidated sands.

Robertson Method
This method is based on the results of CPT carried out under
FIG. 5. 30- and 60-cm-Diameter Plate Tests: (a) Position; (b) Photo of the area to be loaded, and it incorporates factors related to the
Test degree of stress induced by the foundation and the effects of
the stress-strain history (including the natural structure of the
ground).
8(b), always implies a subjective approach; the values of Fig. 10 represents the normalized values of the shear mod-
the constant ␣BB obtained for the hypotheses associated ulus for very small strains G0 obtained from cross-hole testing,
with the Burland and Burbidge (1985) proposal, are sim- as a function of the normalized cone resistance qc1 defined by
ilar to the one now deduced ( f = 0.60), depending on Robertson (1991). From an analytical interpretation of these
the value of B being equally large. plotted results, the following is obtained:

From this discussion, it may be concluded that the sensitiv- 冉 冊

G0 /qc
qc1 Porto saprolitic soil
⬵ 1.8 ⫺ 2.0 冉 冊
G0 /qc
qc1 noncemented sands
ity analysis seems to offer a good method for prediction, by
applying exponents to the factors that condition the develop- revealing that the natural cemented structure of these residual
ment of settlements of shallow foundations under service soils generally induces higher values of the ratio between the
loads, seems to offer a good method for prediction. This, how- elastic or ‘‘pseudoelastic’’ stiffness and the strength, than those
ever, needs future confirmation by other experimental studies, ratios corresponding to transported soils, normally or overcon-
in particular, those including larger sizes of loaded area. As a solidated. This tendency is especially notorious for low con-
summary of the results obtained so far, it can be stated that finement stress levels, revealing a relative independence of the
low strain levels shear modulus (<10⫺6 –10⫺4) in relation to
the at-rest stress states (Tatsuoka and Shibuya 1992).
1. The depths, over which the values of NSPT should be One should also note that the results obtained for the nor-
averaged, can be accepted in the same terms, as stated malized point resistance oscillate among values of 100–300,
by Burland and Burbidge (1985). the value with larger incidence being of the order of qc1 = 150.
2. The nonlinear exponents, in relation to the applied load This corresponds, in soils of the sedimentary type, to dense
nq should be assumed to be greater than unity (the de- sands. Fig. 11 shows a curve that establishes the relationship
duced resulting value was around 1.23), causing rela- between Es /qc with the load level qser /qult obtained from the
tively low constants, e.g., for safety, f = 0.20. results of the footing load test. Also shown in the same illus-
3. Although these comparative analyses of the test results tration are similar curves indicated by Robertson (1991) for
indicate nB values close to one, smaller values should be dense sands, and those proposed by Stroud (1988) for over-
adopted [for example, nB = 0.7, as proposed by Burland consolidated (aged) sands. The curves for dense sand and those
and Burbidge (1985)] when designing shallow founda- obtained from test results shown in Figs. 11(a and b) are the
tions with dimensions generally larger than 2 to 3 m. same, but to a different scale, in order to permit a comparison
This method reflects the reducing dependence of settle- with that of Stroud.
ments on increasing values of B (Bjerrum and Eggestad From consideration of the three curves shown in Fig. 11, it
1963). can be concluded that
874 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001
 TABLE 2. Full Details of Loading Tests

1. For very low load levels, below around 10%, the non- tween the experimental curve for the saprolitic soil and
linearity of the relationship of Es /Qc with qser /qult is much the proposal of Stroud (1988) for overconsolidated
more accentuated for clean sands, even when overcon- sands.
solidated (aged), than for the granitic saprolitic soil. This
may be the consequence of a larger material stability, due It should be noticed, on the other hand, that the values of
to the cemented structure between particles. the qc1 (⬵150) indicated by Stroud (1988) for the test results,
2. For higher load levels, there is a good agreement be- are typical of the middle range for dense overconsolidated
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 875
 FIG. 6. Load Settlement Curve

TABLE 3. Values of ␣P and (Parry) and ␣BB (Burland and Burbidge)

from Experimental Results
s/B = 0.75% s = 25 mm
Footing Plate Footing Plate
(120 cm (60 cm (120 cm (60 cm
Parameter diameter) diameter) diameter) diameter)
␣P 0.32 0.30 0.44 0.39
␣BB 0.63 0.50 0.86 0.65

sands, with age. With shallow foundations on the saprolitic

soil derived from granite, the dependence of the secant de-
formability modulus on the load level in shallow foundations
on the saprolitic soil from granite, seems to represent, in a
consistent way for the highest load levels, the proposal of
Stroud—cited by Robertson (1991)—to use the value qc /N60
= 5.

SETTLEMENT EVALUATION BY INTERPRETATION OF

PLT TESTS
Ghionna et al. (1991)
The Ghionna et al. (1991) method considers the dependence
of the deformability modulus on the normalized stress-strain
levels. It uses the hyperbolic relationship to model the behav-
ior of the soils (Duncan and Chang 1970) and allows the ex-
trapolation of the results of load tests on foundations of dif-
ferent sizes and shapes for the evaluation of the settlement of
larger load surfaces and with different geometries, considering
an equivalent homogeneous mean.
In the expression of Ghionna et al. proposal for the evalu-
ation of settlement FIG. 7. Observed versus Calculated Load Settlement Curves for 1.2-
m-Diameter Footing and 0.6-m-Diameter Plate: (a) 0 < N ⱕ 10; (b) 10
1 qn ⭈ B.I.(1 ⫺ ␯ ) 2
< N ⱕ 30; (c) B ⱕ 3 m
s= ⭈ (6)
Ki qn ⭈ B.I.(1 ⫺ ␯ 2)
␴⬘oct n ⫺
␴⬘oct 1⫺n ⭈ Cf ⭈ Hi In the present case, the n parameter was deduced from the
the parameters have the usual meaning, but the following pa- common expression
rameters require special mention: E = K ⭈ (␴⬘oct 0)n (7)
1. qn = qs ⫺ 2/3␴⬘v 0 = qs, because in the present foundation produced from the similar analysis of different deformability
loading tests, ␴⬘v 0 at the loaded surface is zero. modulus taken from triaxial compression tests made using an-
2. Hi represents the depth of the load influence zone that, isotropic consolidation, referred to as CID and CAD tests, on
according to Ghionna et al., should be considered down samples taken from the zone of influence of the pilot load tests
to 2B from the footing base. (Table 4). Values for Eel were deduced from the initially linear
3. n is a suitable hyperbolic exponent. reload branch of an intermediate unload-reload cycle, those for
4. Ki and Cf represent hyperbolic parameters (the first, of Eur between vertices of that cycle, Eti obtained from the initial
stiffness, and the second, of strength) that will be deter- tangent by hyperbolic modeling, and Es25% and Es50% from se-
mined from the load tests. There is a large dispersion of cant values for 25% and 50% of the failure load, respectively.
the Ki values, due to the high stiffness sensibility. In the subsequent study, in order to analyze the influence of
876 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001
 FIG. 10. G0 /qc versus qc1 [Compared with Robertson (1991)]

FIG. 8. Determination of Values for f, nB , and nq That Give Best Agree-

ment with Experimental Results: (a) for Nonlinear Relation nq < 1; (b)
Linear Relation nq = 1 (ng Fixed at 1.40)

FIG. 9. Effect of Varying Values for ␣

FIG. 11. Es /qc versus qser /qult on Footing Load Test [Compared with
the parameter n, considering its variation range, the following Robertson (1991)]
two values were taken:

1. nel = 0.263—corresponding to very low loads within the

elastic threshold TABLE 4. Hyperbolic K and n Parameters for Different Triaxial
2. n25% = 0.55—corresponding to the mobilization of me- Testing Results
dium stress levels
Deformability modulus K n
Based on the results of the loading tests on the footing and Eel (linear elastic) 35,660 0.263
the 60-cm-diameter plate, the values obtained for Ki and Cf Eur (unload-reload) 19,637 0.250
are given in Table 5. It should be noted that the depth to the Eti,h (hyperbolic q = 70–95% of qf) 2,749 0.539
Es25% (secant for q = 25% of qf) 1,804 0.588
settlement center was taken as zI = B/2, as assumed in the Es50% (secant for q = 50% of qf) 1,517 0.504
method of Ghionna et al. The predicted results, using n =
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 877
TABLE 5. Ki and Cf Values Determined on Loading Tests (zI = B/2)
Footing
(120 cm Plate (60 cm Average
Diameter) Diameter) Values
Exponents Ki Cf Ki Cf Ki Ci
n = 0.263(Eel) 46.3 1.87 45.6 1.95 46.0 1.91
n = 0.588(Es25%) 14.6 1.83 15.6 1.88 14.9 1.86

FIG. 14. Influence of Cf on Application of Ghionna et al. (1991)

Method to Load Tests

0.263 with the corresponding values for Kj and Cf , are in very

good agreement with the observed results from the footing
load test, as shown in Fig. 12.
Some observations can be made in relation to the values
given in Table 5:

1. Although for each assumed value of n there is consid-

erable consistency between the values of Ki and Cf for
the two tests, the variability of Cf being the smaller, it
FIG. 12. Comparison of Predictions Given by Ghionna et al. (1991)
Method with Observed Values from Footing Load Test (n = 0.263) should be noted that small variations of this parameter
have a strong influence on the results.
2. The variation of Cf for different values of n is relatively
low.
3. Ki varies substantially with n, reflecting the constitutive
nature of this parameter.

The general application of this method requires a strict

adoption of representative values on the ground in question.
It was therefore necessary to use average values of n, Ki, and
Cf or, alternatively, those critically selected from the available
values obtained from the individual analysis of each test. The
results of a general analysis of the footing and plate tests,
taking into account the average values for n (as proposed by
the Ghionna et al.) are shown by Fig. 13. It can immediately
be concluded that a low value of n (=0.263) gives the best
simulation for the low stress levels. Good agreement is lost
for higher stress levels, giving rise to a nonconservative result
for the larger footing sizes. This tendency will limit extrapo-
lation for larger size footings.
It is clear that this tendency decreases its relevance (al-
though it can maintain it for smaller sizes) as, in the general
situation, the parameters Ki and Cf refer to the test over the
largest size of loaded area. Although it is conservative for
small footings, on the whole this approach seems to be ac-
ceptable.
To demonstrate the influence of the parameter Cf , predic-
tions were made using n = 0.263, the average value of Ki, and,
for each test, the respective value of Cf . Good agreement was
verified between the observed and calculated curves in the two
cases, as shown in Fig. 14. It should be noted, however, that
the strong sensitivity to the parameter Cf is particularly im-
portant for high load levels in relation to service conditions.

Modifications to Improve Prediction

The following modifications to the original method of
Ghionna et al. were considered:
FIG. 13. Application of Ghionna et al. (1991) Method to Footing and
60 cm Plate Load Test: (a) n = 0.263, Ki = 46.0, and Cf = 1.91; (b) n = 1. Depth of the point of maximum influence is zI = B in-
0.588, Ki = 14.9, and Cf = 1.86 stead of B/2
878 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001
 2. Exponential dependence of the dimension by a factor nB with the other methods under review, because that is consid-
≠1 ered convenient to maintain the proportionality in relation to
B presented by Ghionna et al.
The results obtained from the first modification, taking average
values of Ki and Cf , are plotted in Fig. 15, where it will be Conclusions about Application of Method to
seen that the agreement, although better, is not good. Residual Materials
The second modification requires alteration of the original
formulation in the following: The applicability of Ghionna et al.’s method to those of
residual soils can be inferred from
1 qn ⭈ B nB ⭈ I ⭈ (1 ⫺ ␯ 2)
s= ⭈ (8)
Ki qn ⭈ B nB ⭈ I ⭈ (1 ⫺ ␯ 2) 1. The model—which integrates results of both in situ load
(␴⬘oct)n ⫺
⬘ )1⫺n ⭈ Cf Hi
(␴ oct tests and laboratory tests—presents a good approach for
foundation settlement prediction; however, an in view of
An attempt can be made to improve the fit of the curves by the previous discussion, this should be applied exclu-
assuming average parameters of Ki and Cf , making qn = qs, sively for moderate load levels.
and introducing nB as a weighting factor related to the loading 2. With regard to the geometric assumptions, it is reason-
area. able to retain the direct proportionality in relation to the
As is expressed in Fig. 16, making n = 0.263 and consid- dimension of the loaded surfaces, while for the depth of
ering the value nB = 1.10 gives the tested best hypothesis. This the point of maximum influence, for evaluation of the
tendency for nB > 1 is contradictory to the one that was verified at-rest effective octahedral stress and that induced by the
loading process, it seems better to adopt zI = B, instead
of zI = B/ 2. This alternate proposal, however, requires
confirmation by a greater amount of experimental data,
especially for large loaded areas (more common in prac-
tical foundations), where the relative depth of influence
tends to decrease (Burland and Burbidge 1985)
3. Some ground heterogeneity has notorious consequences
in the model, perhaps worst in parameter Cf .
4. The dependence of the deformability modulus in relation
to the at-rest effective octahedral stress should be eval-
uated for the lowest stress-strain levels, which could be
achieved with triaxial tests on undisturbed samples plus
local instrumentation or, alternatively, by the use of seis-
mic refraction (cross-hole tests). If it proves impossible
to obtain adjusted values for the ground, it is suggested
that values for n ⱕ 0.5 should be adopted.
5. In chosing values for Ki and Cf , the average values ob-
tained in load tests with different loading areas should
be used, provided that the variation among them is not
high. When there is considerable variability of those pa-
rameters, the lowest values of Ki , and above all, of Cf ,
should be chosen to ensure a conservative result.

Wahls and Gupta (1994)

The method of Wahls and Gupta (1994) accurately considers
the nonlinear nature of the stress-strain relationships (␴-ε).
That is translated, first, as a basic formulation taking account
of the penetration testing parameters (NSPT or qc-CPT) and the
FIG. 15. Modifications to Original Methodology of Ghionna et al.
(1991), Considering Depth of Point of Maximum Influence is zI = B
resulting correlations among the small shear modulus (G0 =
Gmax), and second, as a law of variation of the secant or tangent
shear modulus with the distortion level proposed by others
[e.g., Seed et al. (1986)]. These laws apply only to the mate-
rials of the type from which they were developed.
As an alternative, it can be based on the back-analysis of
one or more load tests on plates or experimental footings, pref-
erably of different sizes to enable adjustment of the nonlinear
tendency, defining a variation law of Young’s modulus with
load level qsj in relation to failure.
Adapting the Wahls and Gupta method for this last alter-
native, and assuming a given load step qsj , applied to a foun-
dation on a layer i, of thickness ⌬hi , and with deformability
modulus Eij (i and j translate the dependence in relation to the
depth and to a certain load level, respectively), the vertical
deformation can be calculated by
Isi
FIG. 16. Application of Ghionna et al. (1991) Method by Considering ⌬sij = ⭈ qsj (9)
Dimension Factor nB ≠ 1 Eij

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 879

and the corresponding settlement by Back-Analysis of Footing Load Test

冘 冘
n n
Isi ⌬hi From the experimental results of the footing load test and
sj = ⌬sij ⭈ ⌬hi = qsj ⭈ (10) for s/B = 0.1%, a value of qsref (26 kPa) was obtained, and the
i=1 i=1 Eij
value of the reference modulus (for n = 0.5) deduced from

冘
where n = number of sublayers into which the ground is di- n
Isi
qsref
vided (the larger the number the higher the precision obtained) Eref = ⭈ ⭈ ⌬hi (14)
within the main settlement influence zone, which should ex- sref i=1 兹␴⬘m i

tend to such a depth that the shear stress increment does not
On the other hand, with the pair of values sj and qsj , cor-
exceed the value of initial shear stress, with depths, in prin-
responding to each loading step and obtained from the ex-
ciple, around 2B, for L/B ⱕ 3, and of 4B, for L/B > 3; L and
perimental curve, the respective equivalent modulus can be
B are the dimensions of rectangular footings.
calculated (for an increment from zero to qsj) through
The greater the number of divisions used, the greater will

冘
n
be the accuracy. Isi is the load coefficient for the layer i, de- qsj Isi
pendent on the size of the loaded area and the value of Pois- E0 j = ⭈ ⭈ ⌬hi (15)
son’s ratio.
sj i=1 兹␴⬘m i

The dependence of the deformability modulus on depth can To define the relationship of nonlinear dependence of E0j
be related to the at-rest octahedral effective stress at the center with the strain level, translated for (s/Bj)/(s/Bref), or with shear
of the layer ␴⬘mij by means of stress levels, defined by qsj /qsref , a logarithmic scale was
adopted. In both cases, two influence depths were considered:
Eij = E0 j(␴⬘mij)n (11) zI = 2B and zI = 5B.
In Fig. 17 a comparison is made between the experimental
with n = 0.5, as suggested by Wahls and Gupta.
and simulated curves, with parameters determined in a suitable
Dependence in relation to the vertical stress can be ex-
way. For both influence depths the agreement is excellent, re-
pressed by the following relationship:
vealing the potentialities of the method to model the non-
E0 j = f (qsj /qsref)Eref (12) linearity of the load-settlement behavior of the experimental
footing. Concerning definition of the reference modulus, there
or, alternatively for are obvious potentialities for this formulation in the design of
shallow foundations, by means of a specific load test or by
E0 j = f ((s/B)j /(s/B)ref)Eref (13) preloading an experimental footing, using (s/B)ref = 0.1%.

where qsref , sref , and Eref represent, the load, settlement, and SETTLEMENT EVALUATION BY ADAPTATION OF
deformability modulus, respectively, corresponding to certain WAHLS AND GUPTA METHOD USING TRIAXIAL TEST
reference load steps (for example, s/B = 0.1%); and qsj and sj RESULTS ALONE
= load and settlement for a generic load level.
In the application of the method of Ghionna et al. (1991)
the triaxial test results were partially used to evaluate the value
of the exponent n relating the deformability modulus to the
octahedral effective stress. The formulation introduced by
Wahls and Gupta goes further, giving an opening for the direct
and exclusive use of triaxial test results in the prediction of
shallow foundation settlements. This new proposal has been
used to predict the load-settlement results obtaining from the
experimental footing.
Considering the method proposed by Wahls and Gupta
(1994), the results of triaxial tests can be used to determine
the values of Young’s modulus from the relevant depth and
shear stress level in the soil, using the classical equation
Etl = pa Kl (␴⬘ml /pa)nl (16)
where pa = atmospheric pressure; and Kl and nl = dimension-
less parameters.
Using the results of several triaxial tests with different con-
solidation stresses, and evaluating for each test, the tangent
modulus Etl for distinct shear stress levels, it is impossible to
compute for each of these levels, the corresponding values of
Kl and nl . This is reported in Table 6 where q and qf are the
current deviatoric stress and the deviatoric stress at failure,
respectively.
The relationship of Kl and nl to shear stress level now has
to be defined. Since nl exhibits a modest variation, a constant

TABLE 6. Values of Kl and nl As Function of Stress Level

q/qf
(%) 0 10 25 50 75
Kl 520 256 142 88 74
FIG. 17. Comparison of Experimental Curves and Simulated by Wahls nl 0.383 0.450 0.488 0.234 0.328
and Gupta Model: (a) zI = 5B; (b) zI = 2B

880 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001

 calculated assuming a Mohr-Coulomb law with the ␾⬘ and c⬘
parameters deduced from triaxial tests, and the value of q
given by

qij = ␴⬘vij ⫺ ␴⬘hij = 1.5(␴⬘vij ⫺ ␴⬘mij) (27)

The curve obtained by the application of this methodology

to the footing load test is compared in Fig. 19(a) with the
experimental curve. The analytical result is rather poor, as
could be easily predicted taking into account a linear back-
analysis (Viana da Fonseca et al. 1997a).
Although the linear elastic assumption for the calculation of
the incremental stresses may affect the analytical results (Jar-
dine et al. 1986), it is believed that the main reason for dis-
crepancy is the underestimation of stiffness by triaxial testing
(Viana da Fonseca 1996, 1998; Viana da Fonseca et al. 1997a).
Fig. 19(b) illustrates a comparison between the observed
settlements and the results of the application of the nonlinear
FIG. 18. Variation of Kl with Stress Level simplified procedure in which the deformation modulus is now
corrected by the factor G0 /Gel . The linear diagram obtained by
average value of nl = 0.375 is adopted. To express the depen- the application of the linear elastic analysis with a constant
dency of Kl on the stress level q/qf (computed from the results modulus equal to EtK 0 ⭈ G0 /Gel is shown.
of corresponding triaxial tests), the following function is as- The agreement between the observed results and the ana-
sumed: lytical curve obtained on the basis of the corrected modulus
is very good up to a value of twice the allowable pressure.
Kl = CK log(q/qf)l (17) Even the linear elastic analysis gives a good agreement up to
The value of CK was obtained from the slope of the curve, as loads close to the allowable pressure.
shown in Fig. 18. These parameters are subsequently used in
a simplified nonlinear elastic analysis of the footing load test.
The analysis requires the ground to be divided into several
layers (i = 1, n) each with a thickness of ⌬hi and the appli-
cation of a stepped ( j = 1, m) loading process. For the applied
load step ⌬qsj the vertical and mean effective stress increments
in each layer i, ⌬␴vij and ⌬␴mij , are calculated by using, for
instance, the Ivi and Imi factors proposed by Giroud (1972) for
rigid circular foundations
⌬␴vij = Ivi ⌬qsj (18)
⌬␴vij = (1 ⫹ ␯)Imi ⌬qsj (19)
Stresses at each stage are defined by
␴⬘vij = ␴⬘v( j⫺1)i ⫹ ⌬␴vij = ␴⬘v0i ⫹ ⌺j ⌬␴vij (20)
␴⬘mij = ␴⬘m( j⫺1)i ⫹ ⌬␴mij = ␴⬘m0i ⫹ ⌺j ⌬␴mij (21)
where
␴⬘m0 i = ␴⬘v0i(1 ⫹ 2K0i)/3 (22)
The corresponding vertical strain increments are calculated
by
⌬␴vij ⫺ Iei ⌬qsj /Etij (23)
in which
Iei = (1 ⫹ ␯)[Ivi ⫺ 3␯Imi] (24)
and Etij represents the tangent Young’s modulus. The settle-
ment increment ⌬si for each load step and for the set of all
sublayers (i = 1, n) can be defined by
⌬sj = ⌬qsj ⌺i [Iei ⌬hi /Etij] (25)
and the total settlement is given by
sj = ⌺j ⌬sj (26)
By means of this incremental process it is possible to select,
for each layer, the appropriate deformation modulus. A value
of Etij is adopted, for each increment, by considering the av- FIG. 19. Footing Load Test: Comparison of Experimental Curve with
erage of the values of ␴⬘m and q/qf corresponding to the current That Calculated by Nonlinear Analysis with Deformation Modulus De-
step ( j ) and to the previous step ( j ⫺ 1). The value of qf is duced: (a) Directly from Triaxial Tests; (b) Corrected by Factor G0 /Gel

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 881

CONCLUSIONS termine stiffness. The results can be used to determine an ap-
propriate stiffness degradation law with respect to stress-strain
An attempt has been made in this paper to integrate the level.
information, in terms of stiffness and strength, obtained from The method proposed by Wahls and Gupta showed great
the experimental data, to assess methods used for the design promise. By back-analyzing the more significant load test
of shallow foundations, and in this case, considering the be- (footing of 120 cm diameter), defining a variation law of Eij
havior of residual soils. Some conclusions can be drawn about with (s/B)j , and admitting dependence on the use of two influ-
the methods discussed for predicting settlements, based on the ence depths, the predicted and observed load settlement were
results from the loading tests. Interpretation of the footing load practically coincident.
test was developed primary by means of SPT and CPT results. Notwithstanding the care taken in sampling and testing of
The proposal of Terzaghi and Peck (1967), in the reviewed the soil, the structure of the specimens may have been sub-
version introduced by Bowles (1988), determined the load re- stantially affected (Viana da Fonseca et al. 1997a), and this
quired to produce a settlement of 25 mm, which was found to problem was dealt with by introducing the ratio G0 /Gel (max-
be 2–4 times smaller than that obtained experimentally. imum shear modulus from cross-hole tests and from the linear
The proposal of Parry, considering ␣ = 0.3, predicted the branches of triaxial tests, respectively), which reached the very
test results fairly well, although only for the preliminary load high value of about three.
levels (below the ‘‘elastic’’ threshold). The correction of the stiffness parameters by means of that
The Burland and Burbidge (1985) equation based on SPT factor resulted in a very good agreement between calculated
results led to an overestimation of the observed settlement by and observed results. This methodology, combining results
a factor of 2 or 3. If the semiempirical method of Schmert- from laboratory and in situ tests, in a simple and apparently
mann (1970) and Schmertmann et al. (1978) is applied to- favorable way, is an approach that has been already explored
gether with the CPT results, the footing settlement is accu- by other authors. From this it can be seen that the method of
rately reproduced provided that the value of ␣ = E/qc is taken Wahls and Gupta (1994) is recommended to predict the settle-
in the range of 4.0–4.5. Both methods identify this saprolitic ments of shallow footings on the residual soils of northern
soil in the global typology of cemented or overconsolidated Portugal.
granular soils.
Sensibility analysis of the factors that condition the expres- ACKNOWLEDGMENT
sions for settlement evaluation, was accomplished starting
from the empiric formulation of Anagnastopoulos et al. (1991), The writer wants to express his most sincere gratitude to Dr. Arthur
having verified that the original proposals for sands overesti- Penman, a dear friend and a most distinguished geotechnical engineer,
who has spent so much of his time revising the English and giving so
mated the settlements by a factor greater than two. The pre- many precious suggestions.
sented study derived parameters that adjusted the predictions
to be in reasonable agreement with the test results. REFERENCES
The Robertson (1991) method, based on the results of CPT
tests, was also analyzed. It has the advantage of incorporating, Anagnastopoulos, A. G., Papadopoulos, B. P., and Kawadas, M. J. (1991).
‘‘Direct estimation of settlements on sand, based on SPT results.’’
in the proposed correlations, the load level q/qult induced by Proc., 10th Eur. Conf. Soil Mech. Found. Engrg., Vol. 1, Balkema,
the foundation and the effects of the stress-strain history. The Rotterdam, The Netherlands, 293–296.
study provided verification that Bjerrum, L., and Eggestad, A. (1963). ‘‘Interpretation of loading test on
sand.’’ Proc., 1st Eur. Conf. Soil Mech. Found. Engrg., Vol. 1, 199–
1. The nonlinearity of the relationship qs /qsult for very low 203 (Reprinted in NGI Publ. 58, 23–27).
load levels (<10%), is much less accentuated for the sap- Bowles, J. (1988). Foundation analysis and design, 4th Ed., McGraw-
Hill, New York.
rolitic soil than for clean sands, even when overconsol- Burland, J. B. (1989). ‘‘Ninth Laurits Bjerrum Memorial Lecture: ‘Small
idated; this can be a consequence of a high structural is beautiful.’ The stiffness of soils at small strains.’’ Can. Geotech. J.,
stability of the saprolitic soil. Ottawa, 26(4), 499–516.
2. For higher load levels, there is a good agreement be- Burland, J. B., and Burbidge, M. C. (1985). ‘‘Settlement of foundations
tween the experimental curve and the predictions of on sand and gravel.’’ Proc., Inst. Civ. Engrg., London, 78, 1325–1381.
Stroud (1988) for sands, when the value qc /N60 = 5, as Duncan, J. M., and Chang, C. H. (1970). ‘‘Non linear analysis of stress
and strain in soils.’’ J. Soil Mech. and Found. Div., ASCE, 96(5), 1630–
established in the present case. 1653.
Ghionna, V. N., Manassero, M., and Peisino, V. (1991). ‘‘Settlement of
The prediction of settlements through the interpretation of large shallow foundations on partially cemented gravely sand deposit
load tests was made by the methods of Ghionna et al. (1991) using PLT data. Deformation of soils and displacements of structures.’’
and Wahls and Gupta (1994). With reference to the former, a Proc., 10th Eur. Conf. Soil Mech. Found. Engrg., Vol. 1, Balkema,
relationship was established between the deformability mod- Rotterdam, The Netherlands, 1417–1422.
Giroud, J. P. (1972). Tables pour le calcul des fondations, Dunod, Paris
ulus Eel , Eur , Eti, Es25%, or Es50%; deduced from triaxial tests; (in French).
and the mean effective stress at rest. In general, it was con- Jardine, R. J., Potts, D. M., Fourie, A. B., and Burland, J. B. (1986).
cluded that the model could define parameters for the design ‘‘Studies of the influence of non-linear stress-strain characteristics in
of foundations in residual soils, but that it should be used soil-structure interaction.’’ Géotechnique, London, 36(3), 377–396.
exclusively for moderate levels of the applied load. Kerisel, J. (1985). ‘‘The history of geotechnical engineering up until
Before obtaining these results, it had been assumed that 1700.’’ Golden Jubilee Book, Proc., 11th Int. Conf. Soil Mech. and
Found. Engrg., Balkema, Rotterdam, The Netherland, 3–93.
there was a direct proportionality between settlement and the Parry, R. H. G. (1978). ‘‘Estimating foundation settlements in sand from
size of the loaded area. The tests have indicated that for the plate bearing tests.’’ Géotechnique, London, 28(1), 107–118.
depth to the point of maximum influence for determining both Robertson, P. K. (1991). ‘‘Vertical and horizontal deformation of foun-
the at-rest mean effective stresses and those induced by load- dations and embankments.’’ Geotech. Spec. Publ. No. 27, ASCE, New
ing, it seems better to use zI = B instead of zI = B/2. The York, 764–778.
relationship between E and ␴⬘oct(␴⬘m) should be determined for Robertson, P. K., and Campanella, R. G. (1988). ‘‘Guidelines for geo-
technical design using CPT and CPTU data.’’ Rep. No. FAWA-PA-87-
very low stress-strain levels. It is suggested that in the triaxial 014-84-24, Federal Highway Administration, Washington, D.C.
tests, strain gauges should be of the type that are attached to Rocha Filho, P. (1986). ‘‘Discussion on ‘Settlement of foundations on
the sample itself to obtain a true value of stiffness or, as an sand and gravel,’ by J. B. Burland and M. C. Burbidge.’’ Proc., Inst.
alternative, seismic ‘‘cross-hole’’ tests should be used to de- Civ. Engrg., London, 79, 1633–1635.

882 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001

Schmertmann, J. H. (1970). ‘‘Static cone to compute static settlement Eel = deformability modulus deduced from initially linear
over sand.’’ J. Soil Mech. and Found. Div., ASCE, 96(3), 1011–1043. reload branch in triaxial tests;
Schmertmann, J. H., Hartman, J. P., and Brown, P. R. (1978). ‘‘Improved Esn% = secant deformability modulus for deviatoric stresses
strain influence factor diagram.’’ J. Geotech. Engrg. Div., ASCE, of n% of failure load;
104(8), 1131–1135.
Seed, H. B., Wong, R. T., Idriss, I. M., and Tokimatsu, K. (1986). ‘‘Mod-
Eti = initial tangent deformability modulus deduced by
uli and damping factors for dynamic analyses of cohesionless soils.’’ hyperbolic modeling;
J. Geotech. Engrg., ASCE, 112(11), 1016–1032. Eur = deformability modulus taken between vertices of in-
Stroud, M. A. (1988). ‘‘The standard penetration test—Its application and termediate unload-reload cycle;
interpretation.’’ Proc., Geotech. Conf. Penetration Testing in U.K., Tho- G = shear modulus;
mas Telford, London, 24–49. G0 = Gmax = small strain (maximum) shear modulus;
Tatsuoka, F., and Shibuya, S. (1992). ‘‘Deformation characteristics of Hi = load influence zone for ground under footing base;
soils and rocks from field and laboratory tests.’’ Proc., 9th Asian Re- I = shape and geometry factors for settlement solutions,
gional CSMFE, Vol. 2, Balkema, Rotterdam, The Netherlands, 101– considering homogeneous space;
170.
K = stiffness (hyperbolic) parameter;
Terzaghi, K., and Peck, R. B. (1967). Soil mechanics in engineering prac-
tice, 2nd Ed., Wiley, New York. K0 = coefficient of earth pressure at rest;
Viana da Fonseca, A. (1996). ‘‘Geomechanics in residual soils from Porto m = number of load steps;
granite. Criteria for the design of shallow foundations.’’ PhD thesis, n = stress dependency exponent; number of sublayers
University of Porto, Porto, Portugal (in Portuguese). into which ground is divided;
Viana da Fonseca, A. (1998). ‘‘Identifying the reserve of strength and pa = atmospheric pressure (101.3 kPa);
stiffness characteristics due to cemented structure of a saprolitic soil p⬘ = mean effective stress;
from granite.’’ Proc., 2nd Int. Symp. on Hard Soils—Soft Rocks, Vol. q = deviatoric stress (␴1 –␴3);
1, Balkema, Rotterdam, The Netherlands, 361–372. qc = cone resistance on CPT;
Viana da Fonseca, A., and Cardoso, A. S. (1999). ‘‘Elastic analysis of qc1 = normalized cone resistance on CPT;
surface load tests on a saprolitic soil from granite.’’ Proc., 11th Pan-
qf = failure deviatoric stress;
Am. Conf. on Soil Mech. and Geotech. Engrg., Associaçáo Brasileite
de Melànica des Solos e Engenha ria Geotéinica, Säo Paulo, Brasil. qs (qser) = average contact pressure on base of footing (under
Viana da Fonseca, A., Matos Fernandes,¯ M., and Cardoso, A. S. (1997a). service);
‘‘Interpretation of a footing load test on a saprolitic soil from granite.’’ qsj = load level;
Géotechnique, London, 47(3), 633–651. qsref = reference load level (for specific criterion);
Viana da Fonseca, A., Matos Fernandes, M., and Cardoso, A. S. (1997b). qult = ultimate load or bearing capacity load of shallow
‘‘Correlations between SPT, CPT and Cross-Hole testing results over foundation;
the granite residual soil of Porto.’’ Proc., 14th Int. Conf. Soil Mech. s = settlement (vertical);
Found. Engrg., Vol. 1, Balkema, Rotterdam, The Netherlands, 619– sref = reference settlement for specific load or deformation
622. level;
Viana da Fonseca, A., Matos Fernandes, M., and Cardoso, A. S. (1998).
‘‘Characterization of a saprolitic soil from Porto granite by in situ test-
z = depth (from ground surface or base of loading area);
ing.’’ Proc., 1st Int. Conf. on Site Characterization—ICS’98, Vol. 2, zI = settlement center depth;
Balkema, Rotterdam, The Netherlands, 1381–1388. ␺ = angle of dilatancy;
Viana da Fonseca, A., Matos Fernandes, M., Cardoso, A. S., and Martins, ⌬hi = thickness of sublayer;
J. B. (1994). ‘‘Portuguese experience on geotechnical characterization ⌬qsj = applied loading step;
of residual soils from granite.’’ Proc., 13th Int. Conf. Soil Mech. Found. ⌬si = deformation (vertical), partial settlement of sub-
Engrg., Vol. 1, Balkema, Rotterdam, The Netherlands, 377–380. layer;
Wahls, H. E., and Gupta, M. (1994). ‘‘Settlement of shallow foundations ε = strain;
on sand.’’ Vertical and horizontal deformations of foundations and em- ␾⬘ = angle of shearing resistance (‘‘friction angle’’) in
bankments, Geotech. Spec. Publ. No. 40, ASCE, New York, 190–206. terms of effective stresses;
␾⬘cv = angle of shearing resistance in critical state;
NOTATION ␾⬘p = peak angle of shearing resistance;
␯ = Poisson’s ratio;
The following symbols are used in this paper: ␴, ␴⬘ = total and effective stress;
␴c = consolidation or confinement stress;
B = smallest size (width) of shallow foundation (foot- ␴f = failure stress;
ing); ␴h = horizontal stress;
Cf = strength dependency (hyperbolic) parameter; ␴v = ␴z = vertical stress;
c⬘ = effective cohesion; ␶ = shear stress; and
E = deformability (Young’s) modulus; ␶f = failure shear stress.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 2001 / 883

View publication stats