Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE.

For personal use only; all rights reserved.

Loading Tests: Interpretation and Prediction of their Results

Luciano Décourt1
1
Luciano Décourt Engenheiros Consultores Ltda. Av. Brig. Faria Lima, 1616 / 2nd floor – 01451-001
São Paulo, SP. – Brazil, decourt@decourt.com.br

ABSTRACT: This paper aims at providing engineers with understandable load


(stress) movement curves whenever load tests are available. The idea is to allow
correct interpretations of test results. When they are not available, the idea is to
predict what would have been their results on basis of conventional field tests, the
Standard Penetration Test, SPT, the Cone Penetration Test, CPT or the Menard
Pressuremeter, PMT.

KEY WORDS: loading tests, shaft friction capacity, stiffness, stiffness plot, toe
capacity.

INTRODUCTION

Foundation design usually involves two steps: Initially, the bearing capacity or the
failure load is determined. Then, a safety factor is adopted and an allowable load is
determined by diving the capacity with the safety factor. In the majority of the cases,
this is all that is done. Only special cases also include an evaluation of load
movements of the foundations under their allowable loads.

The conventional way to compute bearing capacity makes use of Terzaghi type
formulas, the so-called triple N formulas. When calculating settlement, the
conventional way is to employ linear elastic theory or empirical formulas, all of
which apply relationships between load (stress) and movement.

In order to check the validity of capacity prediction, only one method is fully
reliable to carry out load tests on prototypes. However, theoretical bearing capacity
relations, which are usually based on rigid plastic theories, yield ultimate values,
regardless of movement. Load tests, while being the practical way to check computed
values, on the other hand, do not define ultimate values. Only pairs of loads (or
stresses) and movements are given. The ultimate value, i.e., the capacity, has to be
defined from the test data rather arbitrarily, which requires the use of one of the
numerous failure and/or extrapolation criteria that have been proposed.

452

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 453

THE CONVENTIONAL BEARING CAPACITY

Terzaghi (1942) proposed a definition of the bearing capacity of piles in load tests
as the load (Qu)c corresponding to a head pile deformation of 10% of the pile
diameter.De Beer (1988) resumes the problem of establishing limit values for bored
piles, defining physical (Qu)u and conventional failure (Qu)c.
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

For piles in clays, the load corresponding to a movement of 10% of the pile
diameter was also used by DeBeer (1988) to define the bearing capacity. For piles in
sands, the capacity was defined as the load corresponding to a movement of 25% or
30% of the pile diameter.

EXTRAPOLATION OF LOAD TESTS RESULTS

It is not usual to carry out load tests up to deformations corresponding to, at least,
10% of the pile diameter (or 10% of the equivalent width of foundation in the case of
footings). The most well known reference as far as methods for extrapolation of load
test results are concerned is Fellenius (1980). An updated version of this paper, very
import but nevertheless much less known, is Fellenius (2001). The methods
mentioned in Fellenius’s papers are presented below. (After Fellenius - 1980, 2001)

FELLENIUS (1980) FELLENIUS (2001)


Davisson Davisson
De Beer De Beer
Brinch Hansen 80% Brinch Hansen 80%
Brinch Hansen 90% ---
Chin Chin-Kondner
Mazurkiewicz ---
Fuller & Roy ---
Butler & Roy ---
Van der Veen ---
--- Décourt

When comparing Fellenius (2001) with Fellenius (1980) one observes that five
methods have been discarded and four maintained. Besides, a new method was
included, Décourt (1996a,1999).

THE DÉCOURT STIFFNESS METHOD

Décourt (1996a,1999) proposed a method for extrapolation of load test results. One
of the most important features of this method is the recognition that some foundations
may fail but others will never fail. This procedure makes use of the so-called
Stiffness Plot and is known in Brazil as the Stiffness Method. Using load test results,
the stiffness is computed (stiffness is the load or stress divided by the corresponding
movement) and plotted against the applied load. The physical failure (Qu)u is defined
as the load corresponding to zero stiffness. Zero stiffness corresponds to infinite

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
454 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

deformation and it cannot be reached in practice. So-defined, it can be stated that


physical failure has never been reached in a single load test, ever carried out in the
world.

It was found that only two possibilities exist:


a) From some level of the test, the stiffness decreases linearly with increasing load,
as indicated in Figure 1. The ultimate pile capacity (Qu)u is defined as the load for
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

which the extrapolated stiffness line intersects with the horizontal axis (abscissa).
In this paper, for both sands and clays the conventional bearing capacity (Qu)c is
considered as being the load corresponding to a pile top movement, “s”, of 10%
of the pile diameter, “d”.
b) From some level of the test, the stiffness decreases non-linearly with increasing
load, approaching asymptotically a limit value as indicated in Figure 2.
30
values considered in the regression
STIFFNESS St (MN/mm x100)

values not considered in the regression


25 Q = 125.04 - 4.75St
R2 = 1
20 (Qu)c = 105.03
or (tf/mm)

(Qu)u = 125.04
15

10

5 ESOPT ll
PRE CAST CONCRETE PILE
0
0 50 100 150
LOAD, Q (MN x100) or (tf)

FIG. 1. Stiffness plot for a displacement pile. (After Décourt - 1999)

A footing foundation and a pile toe of a bored pile will never reach failure. Figure 2
shows a case of a footing foundation.

140
STIFFNESS St (kN/mm)

120 TEXAS A&M


1.0 X 1.0 FOOTING
100
80 log(Q) = 3.90 - 0.63logSt
2
60 R = 0.996
40
(Qu)c = 1,458.11kN
20
0
0 500 1000 1500 2000
LOAD Q (kN)

FIG. 2. Stiffness plot for a shallow foundation. (After Décourt - 1999)

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 455

When the increase of load with the decrease of stiffness is linear – Figure 1 - it is
very easy and reliable to define both, the physical failure (Qu)u (load for zero
stiffness) and the conventional failure (Qu)c (load for s = 0.1d), showing what may be
called a failure behavior. When this increase is non-linear – Figure 2 - it is quite clear
that only the conventional bearing capacity (Qu)c might have some meaning, showing
what may be called a non-failure behavior. The fundamental conclusion is that some
load test may show the full capacity to have been neared – Figure 1 - but others will
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

never reach the ultimate load – Figure 2.

Décourt (1999) upon analyzing results from more than 140 loading tests concluded
that generally speaking the foundations that may fail are:
- Floating piles, which are piles that derive their capacity exclusively, or perhaps
mostly from side friction (all types of piles).
- Displacement piles.

Figure 1 presents an example showing the stiffness curve from a load test on a
reinforced concrete displacement pile, showing clearly both the conventional (Qu)c
and the ultimate (Qu)u failure loads. Some very rare cases of foundations tested in
cemented soils may eventually show a different behavior as exemplified in Figure 3.
However, this almost perfect rigid-plastic behavior is rather an exception than a rule,
even for cemented soils and soft rocks.
stress q (MPa)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
0.0
10.0
20.0
s (mm)

30.0
40.0
50.0
60.0

FIG. 3. Load-movement curve for a footing on a cemented (lateritic) soil.


(After Décourt - 2003)

FUNDAMENTALS OF THE METHOD FOR LOAD TEST


INTERPRETATION

Shallow Foundations and Bases of Bored Piles

There is no doubt that load tests represent the most reliable way for determining the
load (stress)-settlement behavior of foundations. The results of a load test carried out
on a square shallow foundation with a width of 2.50m will be used as an example of
the application of the method, Décourt (1999).

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
456 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

Step I: The Load Test Data


The results of the load test are presented, as measured, in Figure 4, which also
shows the extrapolation of the results according to the proposed method, Step II.
Q (MN)
0 1 2 3 4 5 6 7 8
0

50
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

MOVEMENT s (mm)

logQ = -0.07 + 0.39logs


100 2
R = 0.998

150 values considered in the regression


values not considered in the regression
200
(Qu)c = 7.17MN
250
(qu)c = (Qu)c/A = 7.17/6.25 = 1.15MPa
300

FIG. 4 - Load test data on a 2.50m x 2.50m shallow foundation and extrapolation
of the results. (After Décourt - 1999)

Step II: The Basic Regression


The basic regression may be established for load (Q) or stress (q) versus movement
(s). Generally speaking, it is postulated that the load or stress-movement curve may
be divided in two parts. The initial one corresponds to either, the pre-stressing branch
or in the case of cemented soils to stresses lower than the yield stress 'p. It is usually
an approximately straight line. For higher stresses, the stress-movement behavior is
represented by a parabolic curve. The latter turns out to be another straight line,
provided both, loads (stresses) and movements are plotted in a log scale.

The proposed procedure starts with statistic regressions. Initially, only the last three
points on the curve are considered.

Other regressions follow, considering successively the last four points, the last five
points, and so on. Two criteria for judging the quality of the correlations are
considered: the value of the coefficient of correlation R2 and the number of points
entering in the correlation. It is quite obvious that the higher the value of R2 and the
greater the number of data points entering in the correlation, the more reliable the
statistically generated curve will be.

In many cases, however, these two conditions are fulfilled by different


combinations of criteria. As of yet, no general criterion of preference has been
established, the decision to adopt one or another being a matter of judgment of the
engineer. Ideally, the value of R2 should be higher than 0.999 and not lower than
0.99. Moreover, in most of cases, the separation between the straight line (if existent)
and the main curve is easy to be visually established. The rebound curve also helps to

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 457

separate these two behaviors. The Stiffness Plot however is the most important tool to
help in identifying the limit of pre-stressing as will be explained later.

Once the equation of the curve is chosen, an estimate of the conventional failure
load (Qu)c or failure stress (qu)c can easily be made, as indicated in Figure 4. It is
enough to enter in the correlation with a settlement of 10% of the width of the
footing, in the example, 250mm. If the footing has a shape other than square, an
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

equivalent width (Beq) is defined, Beq A , “A” being its area. Values of (Qu)c of
7.17MN and of (qu)c of 1.15MPa are obtained, Figure 4.

Step III: The Normalized Plot


Once (qu)c is determined, the variation of q/(qu)c x s/Beq is established, as indicated
in Figure 5.

log (q/(qu)c) = C’ + C” log (s/Beq) (01)

Theorectically, C’ should be equal to C”. In most of the cases, they are not equal,
but pretty close together. A value C is then introduced, which is the average of C’
and C”.

C = (C’ + C”)/2 (02)

log (q/(qu)c) = C + C log (s/Beq) (03)

C is called Coefficient of Intrinsic Compressibility. In the given example, as


indicated in Figure 5, C = 0.39

log (q/(qu)c) = 0.39 + 0.39 log (s/Beq) R2 = 0.998 (04)

q/(q u)c
0 0.2 0.4 0.6 0.8 1
0
0.01

0.02
0.03

0.04
s/B eq

0.05 logq/(qu) c = 0.39 + 0.39logs/Beq


R2 = 0.998
0.06

0.07
0.08
values considered in the regression
0.09
values not considered in the regression
0.1

FIG. 5. Stresses x movements in a normalized plot. (After Décourt - 1999)

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
458 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

Step IV: The Stiffness Plot


Step four aims to check if the postulate of the non existence of geotechnical
physical failure (Qu)u (point of zero stiffness), is correct for the load test being
analyzed. It also helps in identifying if the soil has been pre-stressed or if it is
normally consolidated.

The conventional bearing capacity (Qu)c is determined from the Stiffness Plot
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

shown in Figure 6. In the example, (Qu)c is 7.16MN and (qu)c is 1.15MPa. It is clear
from this figure that for loads lower than about 1.2MN, the soil behavior is different,
indicating that up to this level the soil behaves as pre-stressed and/or cemented. It is
also quite clear from this figure that the physical failure will never be reached.

0.7

0.6 values considered in the regression


STIFFNESS St (MN/mm)

0.5 values not considered in the regression

0.4 logQ = -0.12 - 0.63logSt


R2 = 0.995
0.3

0.2 (Qu)c = 7.16MN


(qu)c = 1.15MPa
0.1

0
0 1 2 3 4 5 6 7 8
Q (MN)

FIG. 6. Stiffness Plot for the load test presented in Figure 4.


(After Décourt - 1999)

Shaft Friction Capacity

Bored piles, as any other type of pile, may show shaft friction failure. In
conventional load tests (not instrumented) it is not an easy task to separate shaft
friction from end bearing. However, there is a type of load test, developed
independently more than 20 years ago, by da Silva (1983), in Brazil, and Osterberg
(1984), in USA, which allows the development of pure side friction. A hydraulic jack
called Expancell in Brazil and Osterberg cell, O-cell, in USA is placed close to the tip
of the pile, and the jack pushed simultaneous upward to establish the side resistance
and downward to establish the load-movement for the pile toe.

In most such load tests, carried out under the Author’s supervision, a slightly belled
base is made just below the cell level in order to ensure a complete elimination of the
shaft resistance in the part of the pile below the cell. Some authors question the
validity of the shaft bearing capacity as determined by this test, arguing that the
loading is in the upward direction and the usual loading is in the downward direction.
However, the differences, in the Author’s opinion, are of second order of magnitude.
Both tests are compression tests. Besides, the octahedral stresses, which are

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 459

fundamental for the shaft bearing capacity, are rigorously equal, for both types of test,
at least at the beginning of the loading.

Very recently the author carried out a research program analyzing a large number of
such tests. The results were partially presented in Décourt (2006a). A very important
load test will now be used to exemplify this method. It is the test, carried out by
Loadtest International Inc. in South Korea which is presently a world record, Figures
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

7 and 8.

350
(Qs)u = 164.81MN
300.0mm
300
(Qs)c = 159.76MN
values considered in the regression
MOVEMENT s (mm

250
values not considered in the regression
200
d = 3.0m
150 (Qs)10 / (Qs)c = 0.53MN
Q = 164.81 - 9.49St
R2 = 0.981
100

50
(Qs)10 = 84.60MN
0 10.0mm
0 50 100 150
Q (MN)

FIG. 7. Load test data and the extrapolated curve, statistically established.

60
values considered in the regression
STIFFNESS St (MN/mm

50
values not considered in the regression
40 Q = 164.81 - 9.49St
R2 = 0.981
30
(Qs)10 = 84.60MN
20 (Qs)c = 159.76MN
(Qs)u = 164.81MN
10

0
0 50 100 150 200
Q (MN)

FIG. 8. Stiffness Plot for the load test shown in Figure 7.

The method allows the visualization of the failure process and a mathematically
correct estimate of both, conventional - (Qs)c and physical - (Qs)u failure loads.
Moreover, it also allows the identification of some problems that occurred during the

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
460 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

test. In the Author’s opinion, the statistically determined correlation represents the
load-settlement relationships even better than the measured values themselves.
In analyzing Figures 7 and 8, it becomes clear that for the three largest (last) loads the
actual movements should have been larger than the measured values. For those
familiar with carrying out loading tests, this is a quite common situation, because of
the difficulties in maintaining the applied loads when the deformations are
substantially increasing.
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

One important point to be emphasized is that the value of Qs does not reach an
ultimate value for displacements of about 10mm as assumed by many authors. On the
contrary, the shaft friction load keeps on increasing with deformations. Besides, it is
quite clear that the assessment of both, the conventional (Qs)c and the ultimate (Qs)u
capacities can be easily made, even in cases where the loading has been prematurely
interrupted. In the mentioned example, if the maximum test load had been 92.81MN
with a corresponding movement of 12.70mm and provided the stiffness straight line
could be properly defined, the linear regression would yield to Q = 146.96 – 7.26St,
R2 = 0.923. The computed values of (Qs)c and (Qs)u, would have been 143.49 and
146.96 respectively a 10% difference from the actual values shown in Figures 7 and
8.

Defining D as the angle that the straight line makes with the load axis, and (Qs)10 as
the shaft friction for a movement of 10,0mm an attempt was made to correlate tgD
with the ratio r, (Qs)10/(Qs)c. The values of tgD and (Qs)10/(Qs)c of 40 load tests were
plotted and are shown in Figure 9. The conclusion is that the stiffer the pile-soil
system is, (higher values of tgD) the higher this ratio, r, will be. The ratio r is
therefore a measure of the error made when (Qs)10 is assumed to be the ultimate
value. In another words, it may represent an additional shaft resistance as compared
to that usually assumed, (Qs)10.

5
4.5
4
3.5 (Qs)10/(Qs)c = 0.94 + 0.07logtg
R2 = 0 937
3
tg

2.5
(Qs)10/(Qs)c = 0.95
2
(Qs)10/(Qs)c = 1.01 + 0.44logtg
1.5
R2 = 0.955
1 (Qs)10/(Qs)c = 0.88
0.5
0
0 0.2 0.4 0.6 0.8 1
r = (Qs)10/(Qs)c
FIG. 9. (Qs)10/(Qs)c as a function of tgD.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 461

UNIQUENESS OF THE NORMALIZED LOAD (STRESS) MOVEMENT


CURVE

For Gibson soils, the shear strength, i.e. the penetration resistance of the SPT (NSPT)
and the point resistance of the CPT (qc) increase with octahedral stresses. Because for
normally consolidated soils, the octahedral stresses increase with depth, it may be
assumed that other characteristics also increase with depth.
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

The bearing capacity theories state that for sands, the capacity, (qu)c, is a function of
the sand density, the dimension of the footings (B) and the vertical effective stress,
q = Jz

1
(q u ) c JBNJ  qN q (05)
2

It is often said, which by the way is quite obvious, that (qu)c increases with the
density of the sand, the width of the foundation (B) and the vertical effective stress
(q).

As already mentioned, to take into account size and depth effects, a representative
shear strength, which in practice is assessed via penetration tests, SPT or CPT for
example, should be considered. Normally not only one, but many SPTs or CPTs are
carried out for a given job. The strength of the bearing layer is then assessed taking
into account the results of as many tests as possible carried out nearby and
establishing statistical regression of their results with depth. To reduce the effect of
the natural scatter of the data, a suggestion is made that average values for each depth
be considered.

A statistical regression of these average values with depth is then established.


According to Décourt (1999), the characteristic value of the penetration resistance is
the value corresponding to a depth of 0.7 Beq below the level of the foundation. So, if
a 2.50m width square footing is placed at a depth of 1.10m, the characteristic depth
(zc) will be:

zc = 0.7 x 2.50 + 1.10 = 2.85m (06)

Décourt (1999) suggested that in a normalized plot, the stress-movement curve is


unique, regardless of the density of the sand, the size and depth of the foundation.
The stresses are normalized by the value corresponding to the conventional bearing
capacity (qu)c (q for s = 0,1d) and the movements by the width (or diameter) of the
foundation. All analyses so far carried out by the author have confirmed this
suggestion, Décourt (1999).

To check once more the validity of such proposal, a comparison is now made
between conventional stress-settlement curves and normalized ones. Décourt (2006b)
analyzed the data provided by Lee and Salgado (2005). In Figure 10 the load-

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
462 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

settlement curves of the tests carried out on three footings with 1.0, 2.0 and 3.0m in
diameter and relative densities of 50%, 70% and 90% are shown. It is observed the
obvious increase of q with B and Dr.

In Figure 11 the same data are presented. However, in this case the stresses have
been normalized by the values of (qu)c and the settlements by the footing diameters.
The results of the nine load tests merge approximately in a single curve.
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

q (MPa)
0 1 2 3
0 q/q uc
B=1m
0 0.5 1
50 B=2m 0
B=3m 0.01
100 0.02
MOVEMENT s (mm)

0.03
Dr=50%
150
0.04
Dr=70%
s/B

0.05
200
0.06 Dr=90%

0.07
250 logq/q uc = 0.47 + 0.47logs/B
0.08 2
R = 0.983

300 0.09
Dr=50% Dr=70% Dr=90 0.1
%
350

FIG.10. q (MPa) x s (mm). Footings FIG. 11. Data from Figure 10


with various diameters, B, on sands in a normalized plot. (After
with various relative densities, Dr. Décourt – 2006b)
(After Décourt – 2006b)

INTERPRETATION OF CONVENTIONAL LOAD TESTS ON PILES

Conventional load test on piles are much more complex to interpret than those with
Expancells (O-cells) because the load is transferred to the soil simultaneously by
shaft friction and end bearing. During the first stages of the loading, the transfer is
mostly by shaft friction. For large deformations the opposite happens.

For rigid piles, the transition between these two behaviors may be assumed to occur
for settlements of about 2% of the pile diameter. Using the MBM as proposed by
Décourt (1993), estimates of both, Qs and (Qu)c may be easily made.

Brierley et al (1979) proposed an interesting method for assessing side shear Qs,
which is based on the displacements of the toe of the pile. However, considering that
these displacements frequently are not measured, and that the differences between toe
and top displacements are expected to be very small for bored piles of common

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 463

lengths (10.0/15.0m), it was postulated that the use of top displacements would also
allow reasonable estimates of Qs. This new version of the Brierley Method was
designated the Modified Brierley Method (MBM), Décourt (2003).

Figure 12 presents the basis for the interpretation of such tests.

Following Décourt (1993) a conservative estimate (lower bound limit) of Qs is


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

900.42kN, Figure 12. According to the method that now is being proposed, as shown
in Figures 13 and 14, an upper limit of Qs may be obtained using the Stiffness Plot.
The straight line mainly represents the linear correlation obtained for small
settlements (typically less than 2% of the pile diameter for rigid piles).
Qs = 1,247.43 – 2.41St provides estimates of (Qs)c and (Qs)u, as shown in Figures 13
and 14. The curve that follows the straight line mainly represents an end bearing
behavior. An assessment of (Qu)c may be obtained using log x log correlations
obtained for settlements higher than approximately 2% of the pile diameter, logQ
= 3.48 – 0.24logSt, Figures 13 and 15.
Q (kN)
0 200 400 600 800 1000 1200 1400
0
900.42kN
2%
5

10 4%
Q = 900.42 + 13.10s
SETTLEMENT s (mm)

15 2
R = 0.999
(1,031 < Q < 1,178.2)
20
Qs = 900.42kN
25
10% (Qu)c = 1,261.51kN
30

35 Q = 1,012.87 + 8.29s
values considered in the regressions 2
R =1
40
(1,251.9 < Q < 1,325.5)
45

FIG. 12. Load test data and estimates of Qs and (Qu)c using the MBM, Décourt
(1993). Also shown the six points used in the regression of Figure 13.

These two statistically determined equations may be used to represent the full load-
settlement curve, as shown in Figure 15.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
464 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

1000
900
values considered in the regression

STIFFNESS St (kN/mm)
800
values considered in the regression
700
values not considered in the regression
600
500 Q = 1,247.43 - 2.41St
R2 = 0.996
400
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

logQ = 3.48 - 0.24logSt


300 R2 = 0.996
200 (Qu)c = 1,262.61kN

100
0
0 200 400 600 800 1000 1200 1400
Q (kN)
FIG. 13. Stiffness Plot and statistical regressions

Qs (kN)
0 200 400 600 800 1000 1200 1400
0
d = 300mm
5 2%
(Qs)10 = 1,005.18kN
10

15 4%
s (mm)

Q = 1,247.43 - 2.41St
20 R2 = 0.996 (Qs)10 / (Qs)c =
0.87kN
25

30 10%
(Qs)c = 1,154.67kN
35

40

FIG. 14. Development of Qs with movement(s)


Q (kN)
0 200 400 600 800 1000 1200 1400
0
d = 300mm
5 2%

10 4%
15
values considered in the regression

20 values considered in the regression

25 values not considered in the regression


10% (Qs)c = 1,154.67kN
30
Q = 1,247.43 - 2.41St logQ = 3.48 - 0.24logSt
35
R2 = 0.996 R2 = 0.996
40 (Qu)c = 1,262.61kN

45

FIG. 15. Curves Q x s and Qs x s put together with estimates of (Qu)c and (Qs)c.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 465

PREDICTION OF LOAD TEST RESULTS

Correlations of Failure Stresses with the Penetration Resistance of the SPT

The prediction of the load test curve depends on the knowledge of the failure stress
in conventional load tests.
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

Considering that many different types of piles exist, the Author suggests computing
the bearing capacity for a reference concrete displacement pile with a round section.
For other types of pile, coefficients D and E are applied to adjust, respectively, the
end bearing and the shaft friction. For the reference pile, (qu)c is given by:

(qu)c (kPa) = KN60 (07)

For other piles:

(qu)c (kPa) = DKN60 (08)

All information presented in Tables I, II and III were taken from the text book
Foundations, Theory and Practice, Décourt (1996b).

The values of K are given in Table I, and the values of D in Table II.

TABLE I. Values of the coefficient K for the reference pile as a function of the
soil type, (qu)c (kPa) = KN60
Soil Type Values of K (kPa)
Sands 333

Intermediate Soils 175

Clays 100

It must be emphasized that, following De Beer (1988), for bored piles in sands, the
failure stress, (qu)c is supposed to correspond to a movement of 30% of the pile
diameter.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
466 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

TABLE II. Values of the coefficient D as a function of pile and soil types
Pile Type
Bored Bored with Continuous Root Injected under
Bentonite Slurry Flight Auger High Pressure
Soil Type
Sands 0.50 0.50 0.30* 0.50* 1.0*
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

Intermediate 0.60 0.60 0.30* 0.60* 1.0*


Soils

Clays 0.85 0.85 0.30* 0.85* 1.0*


*values are no more than indications due to shortage of available data.

The unit shaft friction of the reference pile was found to be independent of the soil
type.

(qs)c kN/m2 = 10 (N60/3.6 + 1) (09)

For piles other than the reference pile, a correction factor E is introduced, Table III.

(qs)c = 10E(N60/3.6 + 1) (10)

TABLE III. Values of the coefficient E as a function of pile and soil types
Pile Type
Injected
Bored with Continuous Flight
Bored Root under High
Bentonite Slurry Auger
Soil Type Pressure
Sands 0.50 0.60* 1.0* 1.50* 3.0*

Intermediate
0.65 0.75* 1.0* 1.50* 3.0*
Soils

Clays 0.80 0.90* 1.0* 1.50* 3.0*


*values are no more than indications due to shortage of available data.

For shallow foundation (qu)c has been correlated with N60, Décourt (1995), Table
IV.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 467

TABLE IV. Conventional bearing capacity of spread footings, (qu)c (kPa) = KN60
Soil Type Values of K (kPa)
Sands 95

Intermediate Soils 80
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

Clays 65

These values might also be used to assess the end bearing capacity of bored piles.
However, it is very important to recognize that these failure values correspond to
movements of 10% of the pile diameter and not 30% as assumed in Table I.

Correlations of Failure Stresses with Other Field Test Results

Failure stresses may also be correlated to other in-situ tests such as the CPT and the
PMT. Lee and Salgado (2005) proposed to correlate (qu)c with the cone resistance, qc,
of the CPT, suggesting that this correlation would depend on the sand relative density
(Dr) and the coefficient of earth pressure at rest, K0. Disregarding extreme values, like
relative densities, Dr of 30% and 90% and K0 values lower than 0.45 and higher than
0.7, (qu)c would be in the range 0.25/0.30 qc, the smaller coefficients corresponding to
higher values of both Dr and K0:

(qu)c = 0.25 / 0.30 qc (11)

Combarieu (1997) suggested a correlation between (qu)c and the limit pressure of
the pressuremeter, PL.

(qu)c  PL (12)

Stress-Movement of Shallow Foundations and Bases of Bored Piles

The stress corresponding to the conventional failure is established on basis of


penetration test results N (SPT/SPT-T), qc (CPT) or PL (PMT)). The other unknown
parameter is the Coefficient of Intrinsic Compressibility, C, Décourt (1999).

C is a function of mineralogical and some other characteristics of soil grains and,


therefore, it does not need to be determined for each specific case. Once load test
results exist in a given area, and therefore the results of C are known, they may be
used in other cases dealing with similar soils.

For most of the soils of São Paulo city, with quartz particles, for clays as for sands,
sedimentary or residual, C varies between 0.35 and 0.45. A value of 0.42 has been
successfully used by the author whenever no specific load test results are available.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
468 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

For the dune sands from Labenne, France, C is higher, varying between 0.38 and
0.63, C = 0.55 fitting well most of the test results. For the much more compressible
grains of calcareous sands from Kuwait and Australia, the values of C are much
higher, varying between 0.6 and 0.83, C = 0.75 fitting well most of the load test
results.

Once (qu)c and C are determined, load-movement relationships for any stress level,
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

may be easily determined:

log (q/(qu)c) = C + C log (s/Beq) (13)

The higher the adopted value of C, the more conservative the load-movement
prediction will be.

Load-Movement of Piles

The available methods for assessing the bearing capacities of piles aim to calculate
the pile toe capacity and the unit shaft friction. Considering that in almost all the
cases it is not clearly specified which failure values have been considered by the
authors, it is assumed that conventional bearing capacities have been considered
(values of loads for settlements of 10% of the pile diameter).

Décourt (1993) proposed the already mentioned Modified Brierley Method, MBM,
for determining the actual load-settlement curve in a schematic and conservative way.

However, the determination of a more-accurate curve is highly desirable. The


author is presently working in this direction. The idea is to take into account the
computed values of (Qs)c and (Qu)c and log x log correlations.

One of these curves would represent the branch of low settlements, that is mostly a
shaft friction behavior. The other would represent the branch of high settlements, that
is mostly an end bearing behavior. The values to be considered in the regressions are
shown in Table V.

TABLE V. Predicted load-settlement curve


Loads Settlements (mm or %d)
I – Shaft Friction 0.45 (Qs)c 2.0
0.9 (Qs)c 10.0
II – End Bearing 0.9 (Qs)c 10.0
(Qu)c 0.1d

In Figure 16 a comparison is made between the predicted curve and the loadtest
results in Fig. 12.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING 469

Q (kN)
0 200 400 600 800 1000 1200 1400
0

5 d = 300.0mm

10

15 logQ = 2.59 + 0.43logs


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

s (mm)
2
20 R =1
Q = 0.45Qs s = 2.0mm
25 Q = 0.9Qs s = 10.0mm logQ = 2.84 + 0.18logs
2
R =1
30
Q = 0.9 (Qs)c s = 10.0mm
35 Q = (Qu)c s = 0.1d = 30.0mm

40

FIG. 16. Comparison between the predicted load-settlement curve and the actual
data.

CONCLUSIONS

Methods are proposed for a more-correct interpretation of load test results. These
methods allow conventional and ultimate failure values to be properly determined.
Besides, in case of non-instrumented piles, procedures are suggested for obtaining the
shaft friction and the toe loads. Considering that for routine jobs load tests are usually
not available, methods have been proposed to predict what these curves would have
been like. The basic parameters (qu)c and (qs)c are determined on basis of
conventional field tests like the SPT (especially the SPT-T), the CPT and also the
PMT (the limit pressure PL).The movements of the foundations, for any load (stress)
level, may be easily determined using the correlations here proposed.

REFERENCES

de Beer, E. E. (1988). “Different Behavior of Bored and Driven Piles”. Proceeding


of the 1st International Geotechnical Seminar on Deep Foundation on Bored and
Auger Piles, pp. 47-78, Ghent, Belgium.
Brierley, G. S., Thompson, D. E. and Eller, C. W. (1979). “Interpreting end-bearing
Pile Load Test Results”. ASTM Special Publication 670 – Behavior of Deep
Foundations, Boston, pp. 181-198.
Combarieu, O. (1997). “Capacité Portante des Fondations Superficielles.
Pressiomètre et Essais de Laboratoire”. Bulletin des Laboratoires des Ponts et
Chaussées, 211, pp. 53-72.
Décourt, L. (1982). “Prediction of the Bearing Capacity of Piles Based Exclusively
on N Values of the SPT”. Proceedings of ESOPT II, pp. 29-34, Amsterdam.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering
470 FROM RESEARCH TO PRACTICE IN GEOTECHNICAL ENGINEERING

Décourt, L. (1993). “Predicted and Measured Behavior of Non Displacement Piles in


Residual Soils”. Proceeding of the 2nd International Geotechnical Seminar on
Deep Foundations on Bored and Auger Piles, pp. 369-376. Ghent, Belgium.
Décourt, L. (1993/1995). “On the Load-Settlement Behavior of Piles”. III Odair
Grillo Lecture, given in 1993 and published in 1995. Soils and Rocks, vol.
XVIII, no 2, pp. 93-112.
Décourt, L. (1995). “Prediction of Load Settlement Relationships for Foundations on
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 02/19/15. Copyright ASCE. For personal use only; all rights reserved.

the Basis of the SPT-T”. “Ciclo de Conferencias Internacionales Leonardo


Zeevaert”, pp. 87-104, Mexico.
Décourt, L. (1996a). “Foundation Failures Assessed on Basis of the Concept of
Stiffness” (in Portuguese). SEFE III, Vol. I, pp. 215-224, São Paulo.
Décourt, L. (1996b). “Foundations, Theory and Practice”. Chapter 8.1 - Piles (in
Portuguese), pp. 265- 301, edited by Waldemar Hachic, Frederico . Falconi, José
Luiz Saes, Régis G. Q. Frota, Celso S. Carvalho and Sussumu Niyama,
ABMS/ABEF.
Décourt, L. (1999). “Behavior of Foundations Under Working Load Conditions”.
Proc. of XI Panamerican Conference on Soil Mechanics and Geotechnical
Engineering. Vol. 4, pp. 453-488, Foz do Iguaçu (Brazil).
Décourt, L. (2003). “Behaviour of a CFA Pile in a Lateric Clay”. 4th International
Geotechnical Seminar on Deep Foundations on Bored and Auger Piles, pp. 301-
308. Ghent, Belgium.
Décourt, L. (2006a). “Development of the Lateral Friction Between Pile and Soil on
Basis of Load Tests Using Hidrodynamic Expansive Cells” (in Portuguese).
COBRAMSEG 2006, Curitiba, PR. (Brazil).
Décourt, L. (2006b). “Discussion of ”Estimation of Bearing Capacity of Circular
Footings on Sands Based on Cone Penetration Test” by Junhwan Lee and
Rodrigo Salgado. ASCE, Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 132, no. 11, pp. 1511-1513.
Décourt, L. and Quaresma, A. R. (1978). “Bearing Capacity of Piles on Basis of
SPT” (in Portuguese). Proceedings of VI COBRAMSEF, Vol. I, pp. 45-53, Rio
de Janeiro, RJ. (Brazil).
Fellenius, B.H. (1980). “The Analysis of Results from Routine Pile Loading Tests”.
Ground Engineering, pp. 19-31.
Fellenius, B. H. (2001). “What Capacity Value to Choose from the Results of a Static
Loading Test”. Fulcrum. The Newsletter of the Deep Foundations Institute.
Lee, J. H. and Salgado, R. (2005). “Estimation of Bearing Capacity of Circular
Footings on Sands Based on Cone Penetration Test”, ASCE, Journal of
Geotechnical and Geoenvironmental Engineering, Vol. 131, no. 4, pp. 442-452
Osterberg, J. O. (1984). “A New Simplified Method for Load Testing Drilled Shafts”.
Foundation Drilling. Vol. XXIII, no. 6, ADSC, pp. 9, July/August, 1984.
da Silva, P.E.C.A.F. (1983). “Hidrodynamic Expansive Cell. A New Way to Perform
Loading Tests”. Belo Horizonte, MG. (Brazil): 106 p.
Terzaghi, K. (1942). “Discussion of Pile Driving Formulas”. Proceedings of ASCE,
68 (2), pp. 311-323.

Copyright ASCE 2008 From Research to Practice in Geotechnical Engineering Congress 2008
From Research to Practice in Geotechnical Engineering

You might also like