Вы находитесь на странице: 1из 10

ie anaysiso'resu:s 'rom

rou:ine oi e oac:es:s
by BENGT H. FELLENIUS', PEng, DI. Tech, MEIC, MASCE

THIS PAPER deals with the analysis of zero settlement is reached, defined as ite improvement of the "standard method"
results from axial testing of vertical single 0.01in/h ( = 0.002in/10 min.) The final and it has been incorporated as an op-
piles, i.e. the most common field test per- load, the 200% load, is maintained for 24 tional method in the ASTM Designation
formed. Despite the numerous tests which hours The "standard method" is very D-1143. However, it is the author's opin-
have been carried out and the many time consuming requiring from 30 to 70 ion that a test consisting of 16 equal in-
Papers which have reported on such tests hours to complete. It should be realised crements of say 30 tons applied every
and the analysis thereof, the understand- that the words "zero settlement" are very 30 minutes would provide a better test
ing of pile test loading in current engi- misleading, as the settlement rate of 0.01in than 8 increments of 60 tons applied every
neering practice leaves much to be de- (0.25mm)/hr is equal to a settlement of 1.0 hour, because it would provide a bet-
sired. The reason for this is that the en- 7ft (2.1m) /yr. ter defined load-movement curve. Also,
gineers have concerned themselves with The "standard method" can be speeded a similar yield value, and one not much
mainly only one question: "Does the pile up by using the method of equilibrium different, can be evaluated from the move-
have a certain least capacity?", finding proposed by Mohan et al (1967), where ment during the last 15 minutes, provided
little of practical value in analysing the the load (jack pressure) is allowed to that readings are taken often enough and
actual capacity and the pile-soil interac- drop rather than being maintained by that they are accurate. But why stop at
tion. This Paper aims to show that en- pumping. The equilibrium load value is 16 increments, when 32 every 15 min-
gineering value can be gained from ela- taken as the load applied on the pile. utes determine the load deformation curve
borating on a pile test —
during the ac- Housel (1966) proposes that each of the even better? The load is still applied at
tual testing in the field, as well as in the eight increments be maintained exactly a constant rate in terms of tons per hour
analysis of the results. one hour regardless of having reached and no principal change is made.
The first portion of the notes make use "zero" settlement or not. The Housel Actually, the duration of each load is
of an earlier Paper by the author (Fellen- method of applying the load at equal time less important, be it 1.0 hour or 15 min-
ius, 1975) However, additional views and intervals allows an analysis of movement utes; it is the fact that the duration of
recent literature have been incorporated. with time, which is not possible with each load is the same which is important.
the "standard method". By plotting the From this realisation, we can progress
Testing methods magnitude of movement obtained during to the one that even shorter time inter-
The most common test procedure is the last 30 minutes of each one-hour load vals, and an increase of the rate of load-
the slow maintained load method refer- duration versus the applied load, two ing in tons/hour, are possible without
red to as the "standard loading procedure" approximately straight lines are obtained. impairing the test. Actually, by using as
in the ASTM Designation D-1143 (ASTM Provided the test has approached failure, short time intervals as practically pos-
1974) in which the pile is loaded in eight that is. The intersection of the two lines sible, the time-dependent influences are
equal increments up to a maximum load, is termed yield value. reduced and a more truly undrained test
usually twice the predetermined allowable A maintained-load test according to is obtained. In those cases where a study
load. Each increment is maintained until Housel's method takes a full day to per- of the time-dependent, or drained condi-
form The points on the curve are still tions, creep, etc. is desirable, the test
*University of Ottawa, Canada very few, but Housel's method is a defin- duration should be measured in weeks,

300 500 I I
EXAMPLE I

-0
EXAMPLE N' CLAY +gy
DAV I S SON'S METHOD
C. R. P. —TEST - 60

DRIVEN JULY 9, 1975 GLACIAL 'a o


TILL o 6
TESTED AUGUST 22, 1975
DEPTH 76FT I

PREBORED 30 FT

200— 200— 1
I
I
fn f I

0I- 0I- I
I
I
I

0 CI I
I

0
I 00—

0 I I
0 I
I
0 I.O 2.0 0 I.O 2.0
MOVEMENT ( I NCHES) MOVEMENT (INCHES)
Fig. 1. Load-movement diagram from CRP test Fig. 2. Construction of Davisson's limit
September, 1980 19
IO IO 3 300-
EXAMPLE I EXAMPLE I

DE BEER S METHOD
CHIN S METHOD
200- Iee
P
( I N. /TONS)
150-

V>

O
I 00 ~

CI

5.10 +

40-

30
0.05 0 IO O.I5 0.20 0.30 040 050 I 00 1.50 2.00

MOVEMENT (INCHES)
Fg. 4. Construction of De Beer's yield limit

C2

pULT = —
4.25 Io = 235roas

I.O 2.0
MOVEMENT ( INCHES)
Fig. 3. Ultimate failure according to Chin

months or even years. A 48 or 72 hour A quick test which has gained much the Quick ML method is preferable for
test is then vastly inadequate, and re- use in Europe is the Constant Rate of instrumented piles.
sults only in confusion Penetration test (CRP test), first pro- A fourth test method is cyclic testing.
Tests which consist of load increments posed internationally for piles by Whitaker However, cyclic methods will not be des-
applied at constant time intervals of 5, (1957 and 1963) and Whitaker & Cooke cribed here; for details see Fellenius
10 or 15 minutes are called Quick Main- (1961). Manuals on the CRP test have (1975), and references contained therein.
tained-Load Tests (ML tests) and are been published by the Swedish Pile Com- In routine tests, cyclic loading or even
from both technical, practical and econo- mission (1970) and New York Depart- single unloading and loading phases must
mical points of view superior to the slow ment of Transportation (1974). In the CRP be avoided. It is a common misconception
ML tests, They have been relatively re- test, the pile head is forced to settle at that unloading a pile every now and then
cently introduced into North America, but a predetermined rate, normally 0.02in/ according to some more or less "logical"
are steadily gaining acceptance. The latest min (0,5mm/min), and the load to achieve scheme will provide information on the
version of the ASTM Designation has one the movement is recorded, Readings are tip movement. It will only result in a
quick ML method as an optional method. taken every two minutes and the test is destruction of the chances to analyse the
For instance, recently the Federal High- carried out to a total penetration (i.e. test results and the pile load-deformation
way Administration published an exten- movement of the pile head) of 2-3in behaviour, In non-routine tests and for a
sive users'anual for a Quick ML method (50-75mm) or to the maximum capacity specific purpose, cyclic testing can be
(Butler & Hoy, 1977), of the reaction arrangement, which means used, but then after completion of an
The Quick ML method should aim for that the test is completed within two to initfal test and when the pile is instru-
30 to 40 increments with the maximum three hours. mented with at least a tell-tale to the pile
load determined by the amount of reac- The CRP test has the advantage over tip.
tion load available or the ultimate capa- the Quick Ml test in that it enables an There is absolutely no logic in believing
city of the pile. For routine cases, it may even better determination of the load- that anything of value can be obtained
be diplomatically preferable to stay at a deformation curve, This is of particular from cyclic testing, or occasional un-
maximum load of 200% of the intended value in testing friction piles, when loadings, or one or a few resting periods
allowable load, For ordinary test arran- sometimes the force needed to achieve at certain load levels, when it is realised
gements, where only the load and the the penetration gets smaller after a peak that we are testing a unit which is sub-
pile head movement are monitored, time value has been reached. It also agrees jected to the influence of several soil
intervals of 5 minutes are suitable and with the testing in most other engineering types, is already under stress of unknown
allow for the taking of 2 to 4 readings fields, which regularly use CRP methods magnitude, exhibits progressive failure,
for each increment (for instance, when to determine strength and stress-strain etc., and that all we know is what we
reaching the load, and at 2.5, 4.0 and 5.0 relations. apply and measure at the pile head, while
minutes after starting to load) . When To perform a CRP test, access is re- we are really interested in what h'appens
testing instrumented piles, where the in- quired to a mechanical pump that can at the pile end.
struments take a while to read (scan), the provide a constant and non-pulsing flovv
time interval may have to be increased. of oil. Ordinary pumps with a pressure- Interpretation of failure load
To go beyond 15 minutes, however, holding device, manual or mechanical, are For a pile which is stronger than the
should not be necessary Nor is it advis- not suitable because of unavoidable load soil, the ultimate failure load is reached
able, because of the potential risk of in- ing variations. Also, the absolute require- when rapid settlements occur under sus-
ffuence of time-dependent movements
which may impair the test results. Us-
ment of simultaneous reading of all load
and deformation gauges (changing contin-
tained or slightly increased load —
the pile
plunges. However, this definition is in-
ually, a Quick ML test is completed within uously) could be difficult to achieve adequate, because plunging requires very
two to three hours without a trained staff, For these reasons, large movements and it is often less a
20 Ground Engineering
7.0
BRI NOH HANSEN'S METHOD
90 96 - CRlTERlONI
(
BRINCH HANSEN'S METHOD
205
200— lao% cRITERIDN)
P
909 205 = 185
Vl
Z
0
O
ct
0
IOO-
6.0— =C,B+C2
= 1.414. 10 l66 +3.956.10

CII-'l + C2

/0.25&v BOTH ON

IO 20 ES P„ = 211 TONS
5096 I 67:0 66 I 69
MOVEMENT (INCHES) 5.0+ I I I
Fig. 5. Ultimate failure according to the 0.500 1.000 I. 500 2.000
90% criterion by Brinch Hansen
MOVEMENT (INCHES)
Fig. 6. Ultimate failure according to the 80% criterion by Brinch Hansen

function of the capacity of the pile-soil compression of the pile by 0 value of Brinch Hansen (1963), who defines fail-
system and more a function of the capac- 0,15in (4mm) plus a factor equal to the ure as the load that gives twice the
ity of the man-pump system. diameter of the pile divided by 120. For the movement of the pile head as obtained
In the past, a common definition of 12in dia. example pile, the value is 0.25in for 90% of that load. This method, also
failure load has been the load for which (6mm). The Davisson limit was developed called the 90% criterion, has gained wide-
the pile head movement exceeds a cer- in conjunction with the wave equation spread use in Scandinavia (Swedish Pile
tain value, usually 10% of the diameter analysis of driven piles and has gained Commission, 1970), Brinch Hansen (1963)
of the pile end. This definition does not widespread use in phase with the in- also proposes an 80% criterion defining
consider the elastic deformations of the creasing popularity of this method of the ultimate load as the load that gives
pile, which can be substantial for long analysis. It is primarily intended for test four times the movement of the pile head
piles, while it is negligible for short piles. results from driven piles tested in ac- as obtained for 80% of that load. The
In reality, a limit movement relates only cordance with quick methods. 80% criterion failure load can be estima-
to the allowable deformation limits of the Fig. 3 gives the method proposed by ted by extrapolation from the curve to
superstructure to be supported by the Chin (1970 and 1971) for piles in applying be about 210 tons. (Some references have
pile, and not to the load test results. the general work by Kondner (1963). The confused the 80% and 90% criteria, and
Sometimes, the failure value is defined method assumes that the load-movement use, erroneously, for the 80% criterion the
as the load value at the intersection of curve when the load approaches the failure movement of the 90% criterion),
load is of hyperbolic shape. By the Chin In Fig. 6, Brinch Hansen's 80% criterion
two lines, approximating an initial pseudo-
elastic portion of the load-movement curve method, each load value is divided with is shown in a plot — which is very simi-
and a final pseudo-plastic portion, This
interpreted failure
its corresponding movement value and
the resulting value is plotted against the
ilar to that of Chin — V p/ II plotted against
definition results in ~). The ultimate failure value is deter-
loads, which depend greatly on judgement movement. As shown in Fig 3, after some mined from the criterion that a point, co-
initial variation, the plotted values fall on
and, above all, on the scales of the
a straight line. The inverse slope of this
ordinates P6 3„, on the curve is the point
graph. Change the scales and the failure of ultimate failure when the point, co-
value changes also. A load test is influ- line is the Chin failure load, ordinates 0.80P6 0.253„, also lies on the
enced by many occurrences, but the Generally speaking, two points will de- load-movement curve. The criterion gives
draughting manner should not be one of termine a line and a third point on the the following simple relationships to use
these. same line confirms the line. However, be- in calculating the ultimate failure, P„:
To be useful, a failure definition must ware of this statement when using Chin's
1
be based on some mathematical rule and method. It is very easy to arrive at a
generate a repeatable value that is in- false Chin value if applied too early in P
dependent of scale relations and the opin- the test. Normally, the correct straight 2VC, C„
ions of the individual interpreter. In some line does not start to materialise until
C2
way, it has to consider the shape of the the test load has passed the Davisson
load-movement curve or, if not, it must limit. As a rule, the Chin failure load is C,
consider the length of the pile (which about 20% to 40% greater than the Davis-
son limit. When this is not the case, it is where C, is the slope of the straight line
the shape of the curve indirectly does) .
Without such proper definition, every in- advisable to take a closer look at all the and C2 is the y-intercept in the V p/A„plot,
terpretation becomes meaningless. test data. Fig. 6.
The test results given as a load-move- The Chin method is applicable to both When using the Brinch Hansen 80% cri-
ment curve in Fig. 1 will be used to pre- quick and slow tests, provided constant terion, it is important to check that the
sent nine different definitions of failure. time increments are used. The ASTM point 0.80P6/0.253„ indeed lies on the
The example pile is a 12in (305mm) con- "standard method" is therefore usually not measured load-movement curve.
crete pile installed through 60ft (18.3m) of applicable. Also, the number of monitored In the example case, P„ is 211 tons,
sensitive clay, 10ft (3.0m) of clayey silt values are too few in the "standard test"; which agrees well with the value extra-
and 6ft (1.8m) of silt. The pile was the interesting development could well polated from the load-movement curve,
tested six weeks after driving. Method appear between load increment number directly.
of testing was the CRP method. The pile seven and eight and be lost. Brinch Hansen's 80% criterion postulates
started to plunge when the test load Fig. 4 presents a method proposed by that the load movement curve is approxi-
reached 200 tons, but at the maximum De Beer (1967) and De Beer fk Wallays mately parabolic. Chin postulates that it
load of 206 tons the load necessary to (1972), where the load movement values is approximately hyperbolic. The shape
achieve the movement was still incroac- cre plotted in a double logarithmic dia- of the actual curve is obviously close
ing, gram. When the values fall on two ap- enough to both mathematical curves to
Fig, 2 is applied a method proposed
In proximately straight lines, the intersec- allow both approximations. Brinch
by Davisson (1972), also referenced by tion of these defines the failure value. Hansen's 80% criterion results generally
Peck et al (1974). Davisson's limit value De Beer's method was originally proposed in a failure value about 10% lower than
is defined as the load corresponding to for slow tests. Chin's value, No'te that both methods al-
the movement which exceeds the elastic Fig. 5 illustrates a method proposed by low the latter part of the curve to be
September, 1980 21
EXAMPLE i
EXAMPLEI
250
MAZURKIEWICZ'S METHOD 250- FULLER AND HOY S METHOD
208 BUTLER
/
AND HOY S METHOD
200 ZD3
/X
200—
I/3
150 En
I-
0 150- ON

O
CI Ci
IOO
0
100-
50 I NE

50-
I

0.50 1.00 1.50 2.00


MOVEMENT (INCHES)
Fig. 7. Ultimate failure load according to Mazurkiewicz
0 I I I
0.50 1.00 I.50 2.00
plotted according to a matnernatical rela-
tionship, and —
which is cften very tempt-
ing — they make an "exact" extrapolation MOVEMENT (INCHES)
of the curve possible. That is, it is easy
Fig. 8. Ultimate failure according to Fuller & Hoy and Butler & Hoy
to fool oneself into believing that the
extrapolated part of the curve is as true
as the measured. the intersections according to Mazurkie- The Fuller 82 Hoy method penalises the
In Fig. 7, the method put forward by wicz, some disturbing freedom of choice long pile, because the larger elastic move-
Mazurkiewicz (1972) is illustrated, A ser- is usually found. ments occurring for a long pile, as op-
ies of equal pile head movement lines In Fig. 8 a simple definition proposed posed to a short pile, cause the slope
are arbitrarily chosen and the correspond- by Fuller 82 Hoy (1970) is shown. The of 0.05in/ton to occur sooner, The But-
ing load lines are constructed from the failure load is equal to the test load for ler 82 Hoy development takes the elastic
intersections of the movement lines with where the load movement curve is sloping deformations into account, substantially
the load-movement curve, From the in- 0.05in/ton (0.14mm/kN), offsetting the length effect.
tersection of each load line with the load Fig. 8 also shows a development of the Fig 9 shows the construction of the
axis, a 45" line is drawn to intersect with above definition proposed by Butler & Hoy failure load as proposed by Vander Veen
the next load line These intersections (1977) defining the failure load as the load (1953), A value of the failure load, PH„,
fall, approvimately, on a straight line, the at the intersection of the tangent sloping is chosen and values calculated from
intersection of which with the load axis 0.05in/ton, and the tangent to the initial 1n (1 —P/PDI,) are plotted against the
defines the failure load. Also, this method straight portion of the curve, or to a line movement. When the plot becomes a
is based on the assumption that the load- that is parallel to the rebound portion of straight line, the correct PH„has been
movement curve is approximately para- the curve. As the latter portion is more chosen, The Vander Veen method was pro-
bolic. Consequently, the interpreted fail- or less parallel to the elastic line (see posed long before programmable pocket
ure load of Mazurkiewicz's method is close Fig, 2), the author suggests that the in- calculators were available. Without those,
to that of Brinch Hansen's 80% criterion. tersection be that of a tangent parallel however, the application is very time-
However, when drawing the line through to the elastic line, instead. consuming.

190
500
i 195 EXAMPLE I

VANDER VEEN'S METHOD


200 COMPARISON OF FAILURE CRITERIA
-40-

CHIN 235

8RINCH HAN5EN 80 / ZH MAZURHIEWICZ 208


FULLER 203
200— AND HOY
90 /o 205'~VANDER VEEN 205
-30- HOY 185 DE REER I86
I
60
X
2IO O
I-

C3
220
O
-2.0-
~ 230
100—

-I 0 I
06 07 08 09 IO 12 13 f4
MOVEMENT DNCHES)

Fig. 9. Ultimate failure according to Vender Veen 0 I


I
0 I.O 2.0
MOVEMENT (INCHES)
F'ig >0 Comparison of nine failure criteria
22 Ground Engineering
50 200

/
/
/
Is INCHES DIAMETER FLAT JACK LOAD CELE
O)O
e
40-
I) No

/''Ir

30—
1
<u f—
'XAMPLE N~l
150

jV
20— ~J
l
/
EXAMPLE N'I A
O
o IOO
ih
O

WL''''i'' 1'!/
'' i'''' i'
IOO 200 300
LOAD CELL (TONS)

—I 0-

—20-

—30-
50 100 150
READING
—40—
Fig. 12. Variationsin load cell calibration due to eccentric load
application, temperature change, and reduced loading area

—50- chosen
for the latter can be wrongly
Fig. 1 f. Errorin jack-load determined with manometer vs. correct values of pile elastic modulus or pile
load determined with load cell length, or imprecise or erroneous load or
movement values, Also the Butler & Hoy
method permits an acceptance criterion for
In Fig, 10, the nine values determined elastic column compression of the pile, proof-tested piles to be formulated and
above are plotted together, As shown, the plus 0.15in, plus a value equal to the included in the specifications However,
Davisson limit of 181 tons is lower than diameter divided by 120. The factor F the Butler & Hoy method requires the
all the others and the Chin value of 235 is a safety factor and should be chosen pile head movement to be large enough to
tons is the highest, The other seven to a value of 1.5 to 1.8 depending on cir- reach the Fuller & Hoy point, which
values are grouped more or less together cumstances, restricts the use of the definition in this
around an average of 200 tons, The Chin method is chosen because it context,
It is difficult to make a rational choice allows a continuous check on the test,
of the best criterion to use, because the if a plot is made as the test proceeds, and Influence of errors
one preferred is heavily dependent on a prediction of the maximum load that The test results shown in Fig. 1 and
one's past experience. One of the main will be applied during the test. Sudden used in the preceding discussions are
reasons for having a strict criterion is, kinks or slope changes in the Chin line from a test where an electrical strain
after all, to enable a set of compatible indicate that something is amiss with gauge load cell was used to determine
reference cases to be established. The either the pile or with the test arrange- the load applied on the pile. In the test,
author prefers to use not one but three or ment, The Chin value has the additional the pressure in the jack was monitored
four of the criteria given in Fig. 10, The advantage of being less sensitive to im- by means of a manometer, which had been
preferred criteria are the Davisson limit precisions of the load and movement calibrated together with the jack. Yet the
load, the Chin failure load, the Brinch values, load determined from the manometer read-
Hansen 80% criterion, and the Butler The Brinch Hansen 80% criterion is ings was inaccurate. Fig. 11 shows the
& Hoy failure load. In the case of an en- chosen because it usually gives a P„value difference between the load determined
gineering report, the preference and ex- which is close to what one subjectively from the jack pressure and the load deter-
perience of the receiver of the report may accepts as the true ultimate failure value. mined by the load cell, as plotted against
result in the use of one of the other The value is smaller than the Chin value. the load cell load.
criteria, in addition. However, the criterion is more sensitive The error (overestimation) in the jack
The Davisson limit is chosen because to inaccuracies of the test data than is pressure load is substantial and varies
it has the tremendous merit of allowing the Chin criterion between 10 and 25%, being mostly 15-
the engineer, when proof testing a pile The Butler & Hoy method is chosen 20%. In unloading the pile, the error was
for a certain allowable load, to determine primarily because of its resemblance to much smaller. This is not the worst, nor the
in advance the maximum allowable move- the Davisson method In some cases a best, example the author has met, but is
ment for this load with consideration of Davisson limit load can be obtained with- a typical case for the equipment used
the length and size of the pile. Thus, as out the interpreter being willing to ac- in the industry of today,
proposed by Fellenius (1975), contract cept intuitively that the pile has reached Fig. 11 also shows similar results from
specifications can be drawn up including failure. (In such cases, the Chin value another test, called Example IA, when
an acceptance criterion for piles proof- will be much higher than the Davisson in loading the error was less than 5%.
tested according to quick testing methods. limit), However, the Butler & Hoy slope On the other hand, the error in unloading
The specifications can simply call for a of 0.05in/ton is not approached unless was large, This seems to have involved
test to at least twice the design load, as failure is imminent, and absence of a
usual, and declare that at a test load Butler & Hoy failure indicates in addi- — jacking equipment of a much better qual-

equal to a factor, F, times the design tion to a high Chin value that the — ity than that used in Example 1. However,
Example 1A is from an identical pile lo-
load, the movement shall be less than the Davisson value is imprecise. The reasons cated about 20ft (6m) away at the
September, 1980 23
44f ~l R

I I I I I

,'.-;-~
(Left), A 650 ton kentledge arrangement
e
(lor testing 200lt long 16.5in concrete
~ I K'N
piles). Note that the measuring beam is
J ~m ~ ~
shielded from sunshine and wind
ltg
(Above). Load cell and swI vel plate
(spherical bearing) on a hydraulic jack

same site and tested two days later us- ing only a centre area of the load cell. inserted down to the pile tip to monitor
ing the same equipment and method. Of course, the load cell of Fig. 12 is the compression of the pile and the pile
Based on the above and many similar unsuitable for use in the field, where tip movement, As will be shown, this
measurement results, the author con- temperature variations and eccentric load- relatively simple and cheap addition to
cludes that if one wants to ensure an ing cannot be avoided. In a load test, the test arrangement greatly enhanced the
imprecision smaller than about 20%, a the geometric centre does not necessarily value of the test results.
load cell must be used, The jack and jack coincide with the load centre. Therefore, The graph in Fig. 13 also shows the
pressure are too erratic to be reliable A it is necessary to check the calibration of movement of the pile tip and the mea-
calibration of the jack and manometer for the cell and its sensitivity to eccentric load sured compression of the pile. After a
one pile is not relevant to even a neigh- application. load of 70-90 tons, the measured compres-
bouring test pile. The reason for the un- The foregoing discussion has dealt with sion plots in a straight line, indicating
reliability is that the system is being the imprecision of the load value. But the that the part of the added load used for
required to do two things at the same precision of the movement values can overcoming shaft resistance is constant.
time; both provide the load and measure also be critical, If the "failure" criterion It would be highly improbable that the
it, and load cells with moving parts are is a maximum settlement of 1.75in constant value is other than zero. There-
considerably less reliable than those (44mm), an error of 0.25in (6.5mm) is fore, the applied additional load remains
without, Calibrating testing equipment in of no consequence when the maximum unreduced by shaft friction straight down
the laboratory ensures that no eccentric movement recorded is 1.5in (38mm) or to the pile tip, and the slope of the
loadings, bending moments, or tempera- less, which it is on most proof testing compression line is equal to the slope
ture variations inffuence the calibration. occasions. However, errors of this de- of the elastic line. The combined elastic
However, in the field, all these factors are gree of magnitude greatly influence the modulus of the pile determined from this
present to influence the test results to an shape of the curve and the various met- slope is 5,1 x 10" psi (35000 Mpa).
unknown extent, unless a load cell is used. hods of interpretation of failure loads. In According to a method of analysis pro-
Naturally, many structures are safely particular, Davisson's limit is sensitive to posed by Trow (1967), the pile tip starts
supported on piles which have been tested these errors, to move when the elastic line becomes
with erroneous loads, and as long as we It must be remembered that the mini- tangential to the load movement curve
are content to stay with the old rules, mum distances from the supports of mea- of the pile head, and the load applied
loads and piling systems, we do not need suring beam to the pile and the platform thereafter goes straight to the pile
to improve the precision. The error is etc., as recommended in the ASTM Desig- tip. The analysis by Trow is valid for a
included in the safety factor. That is why nation, are really minimum values; gen- linear, i.e. triangular or rectangular, dis-
factors as large as 2.0 and 2.5 are applied erally they give rise to errors of little con- tribution of shaft resistance. The test
and such numbers are really more ignor- cern for ordinary testing, but they are too results presented in Fig. 13 show that at
ance factors than safety factors. However, close for research or investigative testing a load of about 70-90 tons, the elastic
if we want to economise and continue to purposes. line, established from the measured com-
increase allowable loads as geotechnical One of the greatest villains for spoiling pression, becomes parallel to the load-
knowledge increases, we cannot accept a load test is the sun The measuring movement curve. Consequently, according
potential errors as large as 20 to 25%. In beam must be shielded from sunshine at to Trow's method of analysis, the shaft
the author's opinion, we cannot accept all times, friction must be approximately linearly
errors exceeding 10%, and this require- distributed, and the shaft friction value
ment necessitates the use of load cells. Analysis of results using tell-tale cannot be greater than about 70-90 tons.
However, the fact that a load cell is data When assuming constant unit shaft fric-
used is no guarantee for precise loads. Fig. 13 shows the results from a Quick tion, i.e. rectangular shaft resistance,
Fig. 12 shows calibrations performed on ML test on a 130ft long (40m) 12in the distribution of load in the pile be-
a flat-jack load cell under varying condi- (300mm) precast concrete pile. The pile comes linear and from knowledge of the
tions. The heavy centre line is a regular had a total cross-section of compression of the pile, Fellenius (1969)
calibration curve obtained when using 3in the area of steel reinforcement
124in'800cm'),

has shown that simple relations can be


(76mm) thick full-width steel plates on was 1.9in'12cm'), and the pile circum- established for the load at the pile end
both sides of the load cell and applying ference was 41in (107cm). The pile was and the total shaft resistance, as shown
the load through a spherical bearing loaded in steps of 22.4 tons, and a load in Fig. 14. At pile head loads of 224, 246
(swivel plate). This curve is readily re- cell was used to determine the test load. and 280 tons, the measured compressions
peatable. However, by moving the load The failure loads evaluated in accordance were 096, 1.07 and 1.24in, respectively.
only 2in (51mm) off-centre, a different with the nine methods are given in the The values result in calculated pile tip
calibration was obtained. By letting the graph. Scatter of the values is similar to loads of 160, 182 and 216 tons, respec-
temperature drop, a third line was ob- that shown in Fig. 10. tively. The corresponding calculated pile
tained. The greatest influence was obtain- In the test, a centre pipe had been shaft resistance was 64 tons for all three
ed by removing the steel plates and load- cast in the pile allowing a tell-tale to be pile loads.
24 Ground Engineering
Pskin
EXAMPLE N'2
QUICK M.L. TEST Rip Pave Phead
255
PILE HEAD— LOAD
300—
JJ 250
VANDER VEEN

MA 2 U R K EW C 2
I I

P 270
IFULLER HOY

DE BEER
L250 ASSUMED LOAD
O BUTLER-HOY PILE DISTRIBUTION IN THE
240 LENGTH=L
UJ DAVISSON PILE (CON "TANT UNIT
x 200— AREA -"A L/p SHAFT FRICTION ALONG
UJ THE PILE I

a.

Pave =A E
BL
CJ L
KE
0 tIP ave head
100— skin head tID

P I LE T IP—
tip

DEPTH
Fig. 14. Shaft and tip load calculation from measurement of
'
I pile compression (tip movement) when assuming constant unit
I
I
100 2 00 3.00 I 4 00 500 shaft friction

0 25
increments of 22 tons are too large to
MOVEMENT ( INCHES)
justify the refinement. An increment of
Fig. 13. Load-movement diagram from Quick M.L. test with 10 tons, instead, would have shown much
measurement of pile tip movement. Example 2 more precisely the load-movement de-
velopment during the first 100 tons of
The value of 64 tons is less than the C is 0.5. The previously mentioned three applied load.
previously established maximum possible loads and average loads give values of Lacking adequate soil data, the true
of about 70-90 tons. For many reasons, it x of 0.714, 0.740 and 0,772, respectively. tip and shaft loads cannot be determined.
is prob'able that the unit shaft resistance Insertion in the Leonards 8I Lovell rela- However, they lie somewhere in between
is not constant. Recently, Leonards 82 tion gives values of the tip loads, which the mentioned figures. The results of the
Lovell (1978) proposed a method of ana- are equal to the ones calculated pre- complete analysis have been plotted in
pile compression, viously, i.e. 160, 182 and 216 tons. Fig. 18 showing the load-movement cur-
lysis using measured
which allows a variety of distributions The nomogram of Fig. 16 is applicable ves for the tip and the shaft (vs head
of shaft resistance to be tried in the when assuming a triangular distribution movement) for the two extreme distri-
analysing of the test data. of shaft resistance. In this case, the ratio butions of the shaft resistance, Detailed
Leonards 8I Lovell established the fol- C becomes 0.667 and the calculated knowledge of the soil profile could nar-
lowing refations: values of x are 0.571, 0.610 and 0.658 row the ranges. However, for most prac-
C' C respectively, resulting in the correspond- tical purposes, determining the shaft re-
ing pile tip loads of 128, 150 and 184 sistance to be somewhere in between 64
1 —C tons, and a shaft resistance of 96 tons and 96 tons, as in the subject case, is
where for all three loads. good enough
= between the pile tip In these days of the pocket calculator, The C'alue has additional analytical
x ratio
load and the load applied it is easier to work with the equations significance. The ratio C's
plotted in
tO the pile head (P„P x X P),= directly, as opposed to using nomograms. Fig 19, both as a function of the load
In Fig. 17, the equations behind the no- at the pile head, and as a function of the
C' ratio of measured compres- mograms in Figs. 2 and 3 are presented. inverse of the load. According to the
sion to column compression, pattern of shaft resistance derivation by Leonards fk Lovell (1978),
the latter being the com-
A third
added, which is useful for piles in homo-
is
the plot of C's the inverse of P is a
pression of a free column geneous clay. The reduction of shaft resis- straight line, if the change of compression,
subjected to the same load tance at depth ~ is intended for use when dII, for a change of load, dp, is a constant
as the pile, analysing a progressive mobilisation of value, This is the case when the maximum
shaft resistance is reached and surpassed
C = ratio of elastic compression shaft resistance.
To discuss the results of the analysis by the applied load.
of the pile at a load P sup- The equation for the line is:
ported totally by shaft fric- of the load test presented above, the as-
tion to the column compres- sumption of constant unit shaft friction
sion for tho same load. along the entire length of the pile re-
sulting in a shaft load of 64 tons is pro-
C' n — K- 1

P
C's known from the mea- bably incorrect, However, the shaft load

—x—
The ratio where
sured data, The purpose of the analysis of 96 tons calculated on the assumption AE da
is, either from knowledge of x, i.e. the of triangular distribution of shaft resis- n =
tip load, to determine C, i.e. the relative tance is greater than the maximum pos- L dp
distribution of shaft resistance, or, in- sible shaft load. To arrive at a shaft load
versely, from knowledge of relative dis- in between 64 and 96 tons, the analysis The factor n is equal to 1 only if all shaft
tribution of shaft resistance determine could be repeated with a C ratio between friction has been mobilised, as in this
the tip load. 0.500 and 0.667, chosen either from Fig. case, which, as pointed out by Leonaras
It is not necessary to know the factual 15 with two rectangular shaft friction Ik Lovell, is not necessarily always the
shaft resistance in order to establish the patterns or from Fig. 16 with an upper case.
ratio C. Leonards & Lovell (1978) have triangular and a lower rectangular pattern. In Fig. 20, additional example is
an
determined C for two principal patterns For instance, C = 0.58 determines the given. Resultsare shown from a Quick
of shaft friction, and these are presented shaft load to be 76 tons. However, no ML test on an 84ft long, 12in precast
in the nomograms in Figs 15 and 16. The justification is available for further re- concrete pile driven through clay and into
case of constant unit shaft resistance is finement. Such justification would have a very competent glacial till, The ditgram
a special case of the nomogram of Fig. been, for instance, a definite change of shows the measured pile head movement,
15; the two friction values are equal and soil profile at some depth, Also, the load pile compression, and the head load versus
September, 1980 27
I.O l ~ I.O I I I K I

SHAFT FRICTION PAT SHAFT FRICTION PATTERN


GROUND fS fs fs fs
GROUND
SURFACE SURFACE

0.9 0.9 I
I I
I i
I I

P LE
I

T P I
PILE I

0 8 'll

I—
C= 2
I

I-
),'(I ((I
K
I
II K) )
O.&e-s-RATIOS OF —
SI
fs,
= k
0.8

UJ )—
O Z
UJ
U O
UJ
O U
o C7 UJ
O
0.7

0.6
0.6
0.55-
)- Xk
)- )(k .
0.50
l I 05 I

0 05 I.O
0 05
I
I.O

RATIO 7/L = RATIO Jf/I


Fig, 15. Coefficient C for various distributions of unit shaft friction Fig. 16. Coefficient C for various distributions of unit shaft friction

measured tip movement, plus the calcu-


lated shaft resistance and tip load. The
20 tons. This load acts on the pile be-
fore the start of the test loading and, as a
r k

calculations are performed assuming that result, the compression of the pile does
the distribution of shaft resistance fol- not start from zero, but has an initial
lows the third pattern in Fig. 17 with value of about 0.08in. Adjusting the
l = L, and k equal to the same ratio accordingly, and recalculating the C'alues

of shear resistance as found by vane shear results show that the adjusted maximum 2

testing. (For details on the soil profile, shaft resistance is 210 tons and the maxi-
and an older, much more time-consuming mum tip load is 235 tons. The difference
and arbitrary, method of analysis of the in calculated tip load is 50 tons, 2.5 times
test results, see Fellenius & Samson, the estimated residual load.
1976). The above shows how sensitive the
The pile test is not carried to ultimate method of analysis is to residual loads,
failure, However, the Leonards-I ovell and, therefore, also to inaccuracies of the
method of analysis of the simple tell- measurements. However, the old subjec-
tale measurements of the tip movement tive methods are actually even more sen- 2
()-kl- ~
2
k
makes it possible to establish that the sitive, but because of their arbitrary nature, 2-2 A i)-k)- x k
maximum shaft resistance acting along this is not always evident. In contrast,
the pile is 240 tons and the maximum tip the Leonards & Lovell method allows a
load mobilised is 185 tons. Indeed, this determination of the extent of uncertain-
is a result well worth the expenditure of a ties influencing a test It provides, there-
bit of time and money, fore, the engineer interpreting the data
As in the previous example, additional with a zone of reliability and a confi-
supportinq information is gained from a dence he would otherwise have felt only )

plot of C'ersus 1/P for the results, and because of his ignorance of the uncer-
also C'ersus P. As seen in Fig 21, when tainties involved and of their effects,
)+k ~ z)I- 8
2

L s k +
qoing from riqht to left in the diagram, To take full advantage of the Leonards
the C's 1/P is at first a curved line & Lovell method of analysis, the testing
later becominq a straight line pointing to method should not be the "standard Fig. 17. Mathematical expressions for co-
C' 1.00, 1/P = 0. The P value for when loading procedure", which for all techni- efficient C for various distributions of unit
the curve becomes a straight line deter- cal and economical purposes is the worst shaft friction
mines the point (load) where all shaft method to use, but the "Quick load test
friction is mobilised, in this case for P = method". The load increments must be reaching the maximum test load, Naturally,
300 tons. applied at constant intervals (for practical a reliable load cell should be used to
The results of the second test are al- reasons usually 5 minutes, to allow for at supplement the jack manometer, and every
most certainly affected by a residual load least three readings per increment). The effort must be made to ensure reliable
caused by the reconsolidation of the clay increments should be small enough to allow movement values.
after driving and estimated to be about for at least 30 or 40 increments before Fig 22 shows another method of pre-
28 Ground Engineering
300 INVERSE OF LOAD AT PILE HEAD ( /P)
~(
'XAMPLE
I

3 -3 .3
~-TIP VS BUTT LOAD 0 I)0 510 IDIO
1.00
2
MEASURED COMPRESSION
COLUMN COMPRESSION

0 90-

f.-TIP
200-
0 80-

VI
X /P
O
0.70-
C3
ct
O
0.60.
IOO- ———p ———--p--p-
3
-p .p

HEAD(TONS)
PILE

— SHAFT

----0-----O--O--O-— 0.50-

0.4 0
;7 0 IOO 200 300
0-, LOAD AT
0 1.00 2.00 3.00 Fig. 19 Example 2. Ratio C'lotted vs load at the pile head
MOVEMENT (INCHES) and vs the inverse load
Fig, 18. Example 2. Load-movement diagram for shaft and tip loads
500

HEAD LOAD
YS TIP MOVEMENT
COMPRESSION INVERSE LOAD (TONS )
HEAD -3
-3 -3
0 5 IO 10.10 15 10
400— 1.00—

~ r
V3
Z VS P
O
SHAFT LOAD-
.r~
CI
HEAD
VS
MOVE MENT C .r wJ
I
O
200— TIP LOAD
0. 50—
VS
HEAD MOVEMENT
~

100—

0.10—

0 0 1 I

0 I 00 200 300 400 500 600


0 2.00
MOVEMENT ( INCHES) LOAD ( TONS 'I

Fig. 20. Example 3. Load-movement diagram from Quick M.L. test Fig. 21. Example 3. Ratio C'lotted vs load at the pile head
with measurement of pile tip movement and vs the inverse load

senting the results of an analysis of a pile stant unit shaft resistance (rectangular tributions give the same average load in
test (Example 2). The load distribution distribution) and the curved line that for the pile, the two areas A'nd A" must
in the pile is shown for three different a triangular distribution of shaft resistance. be equal.
loads at the pile head. The straight line The interesting point in this graph is that The condition indicated in Fig, 12 can
represents the load distribution for con- to fulfill the condition that both load dis- be developed to determine non-linear load
September, 1980 29
LDAD (Toifs) PILE HEAD
B(I TT LOAD
0 100 200 300
0

50—

TELLTALE
SE

zi-
CL
iu
to

TiP TELLTALE

Fig. 23. Determination is ri


of loa d distribution from tell-tale measure-
100— ments

'
E. Ik Wa//ays, M (1972): "Franki
piles
pi with overexpande d bases",
as La Technique
des Travaux, No 333, 48 pp.
"Static measurements o f
ur". Design and I ns tli a at f P'I
Foundations Ce! u ar Structures, Envo Publ.
and Cellular
Co., Edited by H-Y Fang, pp,
0 -~ — ' Davisson, M. T. 1972: "High capacity piles",
Lecture Series,
Foundation Construction, ASC i oi S
1.0- 52 pp,
H. (1969): "Bearing capacity of
Z ) fnction piles Results s I t f!I Ro t",
X 2.0- 5 edish)
Iu Research, Report No. 24 pp (I
3.0- 1975): "Test loading of piles.
Methods, interpretation an d new oof testing
procedure", Proc. ASCE, Vo
i-
Samson, L. (1976): "Testing
22. Example 2.. V eric
tical load distributions in the i le during of dnveability of concrete piles an
the test
p'ig.

to sensitive clay", Canadian o,


pp.. 139-160.
~

ick-load test method conven-


, as roposed by va lue that measurements t oof tii movemen
recom-,...
tional methods and interpre a i
wo- can provide the anay
in . 78-86,
'd'

that is strongly e
m~~d~d aalso for routine tests. For clos -
san,

tip and a second one


'

ocate
ave been In
23, the diagram en d t It b il
precast prestress COnCrete
ires a little bit of advance planning
pi eS respo,
sM4 pp

response cohesive soils", ASCE,

f the

g ph
SO that a Centre tube

"
ile. H-piles could
can bee cast in the
be Can
necessitate
din g and a few hours of pre
som,d, L . 388-
ts on
osium on
hig
e awo

d paration. Cast-in-place pi es a
2 from the pi e e u h eluded.
t e w ds, lotted at In view o of the simplicity and low re a-
mi ell-tale locations tive
iv cost coupled with t e arg "
an te II of extra information gai,
ined, by no means n w tea~a, eo e

'fill
POSSI considering ex hausted
a in this Paper, ale Is itt 6

distribution
i line must fu i
, A'nd A" and
e true ' loa
the conditions

h
B'n
athema-
ex
m n

8 8rBnCE)S
in ti meaSure- N y

ec...,,
DOT
c dure SCP4/74, 35 pp,
sSo

"Foundation
.
Engineering",
E. Ei Tho
Secon
nbu, T.

p we I oa d" A I 8 k 1 weis
the "true" load distributionOn has irs been
aS firat ASTM Standards, Part
assumed to cons' I
88
Ic nsan J. (1963); Discussion Hyper- o m
1
SV SM4 pp 241-242 r to he 48th nnua
load. It also mean Ho, H. E. (1977): "Users manual o a
tion losad testin
'ck-load
"
method
F dere I Highway
ig
for founda-
Adminis-
De op
sistance resu s i
"'gh
Chin, F. K. (1970): "Estima on
oad of Piles not earned, .
Whr
'
k
ou s, eo ec n

In the ana
use was made
movement pi tr
ys'rkansas
o
Southe t A Co f, o Soi
i

River
ngn
Discussion, "Pile tests.
Project", AS
0-932.
Whiraker,....:
PP
h. an oun
new

has been to demonstrate t e trem


qen van zend onder
het grensdraag
funde ingen
Tijdshrift der openbar Werken van
op
van
',
vermo-
staal"
.
i a
Whiraker,
tration tes
'a
T. (1963): "The constan
o
acity o a
r

Seep tember, 1980 31

Вам также может понравиться