Sunteți pe pagina 1din 9

CONSOLIDATION THEORY APPLIED TO FOUNDATION PILE

TIME EFFECTS
by
LARS 0. SODERBERG, M.ASCE

SYNOPSIS
The time increase in the bearing capacity of a L’accroissement horaire de la capacite portante
friction pile in clays and silts is explained in terms d’un pieu flottant dans l’argile ou le limon est cx-
of the consolidation theory. The case of a single plique selon les termes de la theorie de consolidation.
pile is examined in detail and with respect to the Le cas d’un seul pieu est 6tudie en detail et par
available experimental data. The problems con- rapport aux donnees experimentales disponibles.
netted with large clusters are outlined and limits Les problemes se rapport-ant a des groupes impor-
are proposed for the time effect on a cluster. tants sont esquisses et on propose des limites pour
l’effet horaire sur un groupe.

NOTATION
area cm2
horizontal coefficient of compressibility cmz/kg
horizontal coefficient of consolidation cm2/sec.
vertical coefficient of consolidation cm2/sec.
voids ratio -
space increment cm
time increment sec. (second)
coefficient of permeability cm/set.
constant -
whole number in approximate solution -
flow of water ccjsec.
radius cm
pile radius cm
time set
hydrostatic excess pressure kg/cm2
hydrostatic excess pressure at pile skin kg/cm2
at t = 0
excess pressure as zt but time different kg/cm2
volume of water
unit weight of water :g,cc
angle radian
angle of internal friction radian
initial stress kg/cm2
radial stress kg/cm2
tangential stress kg/cm2

INTRODUCTION
For many years people who have been intimately associated with pile driving have recog-
nized the fact that the driving resistance of a pile may increase with time. This phenomenon
is described as “set up” and has been observed in most parts of the world. Quantitatively
this phenomenon can vary from an imperceptible one to a situation where a few minutes delay
makes any further movement of the pile impossible.
The first mention in literature of this time effect was at the turn of the last century when
Wended reported some test loading results he had made in Sweden. The bearing capacity of
217
218 LARS 0. SODERBERG

his piles increased rapidly with time and reached their maximum in less than a month (Bjerrum
and Flodin, 196O).r The Scandinavian countries were among the first to recognize this effect
and in Norway it has become a common practice to load test piles a month after driving in
order to take advantage of this time effect (Bjerrum, et cd., 1958).
The time effect is mentioned quite generally in the literature, although it is often incidental
to the main points the particular Authors were making. Fig. 1 shows a plot of pile bearing
capacity versus the log of time for the available cases from which numerical evaluations of the
time effect could be made. The bearing capacity is maximum at the horizontal axis and
decreases in the vertical direction. These cases will be examined in more detail later, but for

Fig. 1.

the present they serve to indicate how large this increase in bearing capacity can be. An
increase of the order of 10 is apparent in these cases.
Any factor that contributes to the pile bearing capacity by this large amount will be very
interesting commercially because, if properly understood, it will permit a significant increase in
allowable pile load. It is tempting to generalize from tests such as these and consider “set
up” as simply a soil characteristic. This generalization would permit a very large increase in
allowable pile load in certain soils. The problem, however, is not so simple, and it turns out
that other factors are as important as soil characteristics.
A fundamental understanding of this problem is necessary before this increase in bearing
capacity can be safely used to support an engineering structure, and it is the purpose of this
Paper to analyse the fundamentals of this phenomenon for a few simple cases.

BASIC CONDITIONS

Cummings, Kerkhoff and Peck (1950) published the results of some tests on a large cluster
of piles with relatively large centre-to-centre pile spacing in soft clay. They did not measure
bearing capacity but they did measure the water content of the clay at various radii from the
pile over a period of a year after driving. They established quite clearly that there is a hori-
zontal migration of the pore-water which was initiated by driving. They showed further that
the change in water content was approximately constant with depth and varied essentially
with time.
1 The references are given on p. 225
PILE TIME EFFECTS 219
A driven pile will be surrounded by soil grains, at a certain pressure, and pore-water with
an hydrostatic excess pressure. The skin friction of the pile will be developed from the inter-
granular pressure according to the laws of friction. A decrease in the hydrostatic excess
pressure will result in an increase in the intergranular pressure and therefore increase the pile
bearing capacity. This is essentially a partial pressure effect of the soil-water system with the
water diffusing outward from the pile and is the phenomenon reported by Cummings, et al.
(1950).
The literature provides much indirect support for this thesis. For example, Seed and
Reese (1957) showed, on a test pile in California, that the water content in the clay adjacent
to the pile decreased with time after driving, while the bearing capacity increased. In 1958
Bjerrum, in Norway, noticed that a permeable wood pile “set up” much faster than an
impervious one. He also measured increased excess pore pressures that were constant with
depth (Bjerrum, et al., 1958, pp. 10 and 13).
It is quite clear that “set up” is a consolidation phenomenon in clays and silts and that
any analytical treatment of it will require the basic assumptions of Terzaghi’s consolidation
theory. The empirical validity of these assumptions are beyond the scope of this Paper and
are fully discussed elsewhere (Biot, 1941; Taylor, 1948; Terzaghi and Peck, 1948). The
assumption that the soil is isotropic is not necessary when only flow in the horizontal plane is
considered, and the methods proposed herein can be adapted to an anistropic material.
A real pile is installed by a series of approximately instantaneous increases in depth.
The soil at the tip will be remoulded with each increase and a complicated three-dimensional
diffusion system will be initiated with each blow. The pile above the tip will slide against
the soil and there will be continuing remoulding effect with each blow. When the pile is very
long compared with its diameter, the diffusion will be essentially horizontal, except at the
ground surface and at the pile tip; there will be a vertical flow of water at these two extremes.
The problem is then that of a long cylinder surrounded by a radially symmetric water-pressure
distribution which varies somewhat with depth. This variation arises from the finite driving
time and the vertical flow at the top and bottom.
A great deal can be learned by idealizing this situation somewhat. This is done by neglect-
ing the vertical pressure differences and by taking a horizontal plane through the central
portion of the pile as representative of the true pile behaviour. With the assumption of an
impervious pile the problem becomes one of predicting the hydrostatic excess pressures at the
pile skin as a function of time. The pile bearing capacity will be maximum when these
pressures are zero and a linear relationship between water pressure and bearing capacity will
be valid.

SINGLE PILE DERIVATION

The partial differential equation of this phenomenon can be derived by standard methods
as presented by Terzaghi, Biot, Taylor, and others. The element to be considered is illustrated
in Fig. 2. There is no pressure variation in the tangential direction and lines of constant
hydrostatic excess pressure, U, will be concentric circles. These considerations constrain the
flow of water to the radial direction only.
The flow of water from the shaded element less the in flow is the change in the volume of
water contained by the element, by Darcy’s law this volume per unit time is:

WV,) = (hmr/y,)
- x ($ + 2) . . . . . .
at
where dVw is the volume of water contained by the element, R is the coefficient of permeability,
Y is the radius, 0 is an angle, ym is the unit weight of water, and u is the hydrostatic excess
pressure.
220 LARS O.SODERBERG

The time derivative of the volume of water in the element is:


WVW) rd0 dr de
-=(l+.z . . . . . . . . (2)
St
by substituting ah x du for de and by defining:

- . . . . . . . . .
Equations 1 and 2 can be conveniently equated. ah is the horizontal coefficient of compres-
sibility and e is the voids ratio of the soil. The partial differential equation for the case of a
single pile in an infinite medium is then:

$c$$+.$) . . . . . . . (4)

The assumption that ch is constant is admittedly open to discussion. It has however, been
made in all the consolidation work to date
with good results and it will be accepted
here. ch will vary over the radial distance
that is disturbed by driving. An average
value for chwill be used and it will be assumed
that ch does not vary with the radius.
Equation (4) introduces the phenomenon
as one of diffusion and the mathematical
background of diffusion and heat flow may
now be used (Carslaw and Jaeger, 1948).
One consequence of this is that the solutions
Fig.2
become functions of the parameter &/4cht.
Hence, any enlargement of scale lengthens the time by the square of the physical dimension
while an increase in the coefficient of consolidation can only reduce the time by the first power.
This leads immediately to the important conclusion that the horizontal dimension of the
cylinder can be more important than the characteristics of the soil surrounding it.

INITIAL PORE PRESSURE DISTRIBUTION

The initial pore-pressure distribution and the coefficient of consolidation must be known
before a solution to equation (4) can be attained. Presumably the coefficient of consolidation
can be determined in the laboratory, at least within limits, but the initial pore-pressure distri-
bution is a much more difficult problem.
Several assumptions must be made in order to determine this distribution analytically.
The most important of which is that the excess pore pressure is proportional to the radial stress
which is imposed by pile driving. This assumption is quite reasonable if the soil surrounding
the pile is thought of as a series of concentric cylinders, and it is remembered that the stresses
in the outer cylinders will remain unchanged if the inside cylinder is replaced by one of an
hydrostatic pressure equal to the radial stress (Timoshenko and Goodier, 1951). The physical
assumption implied here is that the elastic-plastic stresses surrounding the pile will bereplaced
at the instant after driving by an equivalent hydrostatic pressure. With radial symmetry
the radial stress will be the major principal stress.
By further assuming that the soil acts as a perfectly plastic-elastic system with a constant
compressive strength (4 = 0) f or an instant after driving these stresses can be calculated.
This problem has been solved for the case of plane stress in the plastic zone. This case is
applicable to pile driving as it requires a large vertical strain (heave). The maximum radial
pressure distribution from this case is shown in Fig. 3 (Nadia, 1950).
PILE TIME EFFECTS 221
An alternate assumption can be made in place of the elastic-plastic system and this is that
for the last instant of driving the soil behaves as a viscous substance that will not support
tension.
The equilibrium equation in the horizontal plane for any system with radial symmetry
(Timoshenko and Goodier, 1951) is:

$?+,- UT- (Ta =o . . . . . . . . (5)


r
In this case 0,. will be a compressive stress and Q will be tension, but the assumption is that
the material cannot support tension. Therefore, the tangential stress must be 0. Equation 5
becomes
. . . . . . . . . (6)
a, = K/r is a solution that fits the end con-
ditions a, = (TOat rr, a, = 0 at I = infinity, 1.00
and by remembering the first assumption the
solution becomes: * *75
% Yl ,_\ .50

and this distribution is also shown in Fig. 3.


Bjerrum has observed a pore-pressure
distribution surrounding a pile that approxi-
mately fits this latter case and an interpre- r
‘1
tation of his points are shown in Fig. 3
(Bjerrum, Correspondence, 1959). His data Fig. 3
was taken 1 m from the pile outwards, and
these points were obtained by extrapolating his data to the pile skin. This one case is of
course, not conclusive and the observed distribution could fit either case by allowing for experi-
mental errors.
The approach used here will be to solve the diffusion equation for both distributions and
compare the results. The result will then be compared to the available experimental data to
check the whole theoretical structure.

SOLUTION OF THE SINGLE PILE CASE


With an assumption for the initial pore pressure distribution a solution, in terms of the
pile radius and the coefficient of consolidation, of Equation (4) is possible.
A convenient method for this solution is offered by the technique of finite differences
(Hildebrand, 1952; Richart, 1957; and Timoshenko and Goodier, 1951). The notation is
illustrated in Fig. 2. Point 0 is that under consideration, point I is one radial space incre-
ment (h,) inside and point 2 is one radial space increment outside point 0 . U denotes the
hydrostatic excess pressure at time t + hztwhere ht is the chosen time interval, while u denotes
the same pressure at time t. With this notation Equation (4) can be approximated by the
difference equation:
uo - UIJ ug - 2260+ 261 % - %
--&-- = Ch + r (3)
( h2, 1 ...
Next chht/h2, is arbitrarily set at 0.25 and this parameter is factored out. The radius Y is
then replaced by nh, where n is the number of space intervals from the point in question to
the pile centre. The approximation now reduces to:
U. = ~~(0.25 - l/&z) + ~~(0.25 + l/&z) + .50uo . . . . (9)
222 LARS 0. SODERBERG

This now makes the pressure of any point a function of the pressure of that point and the two
adjacent points, considered one time increment earlier.
One further condition is necessary for a solution. That is that there is no flow of water

across the boundary of an impervious pile. Darcy’s Law requires $ to be 0 at this boundary.

This in turn requires that the imaginary zbl, at the boundary, be replaced by a reflexion of
~0, the pressure on the boundary. In Equation (9) ~0 will replace Z.Qat the pile skin, r = rl.
With this approximate method it is possible to attain a solution within 5% of the true
solution without an undue amount of labour. The accuracy of any solution can be verified
by reducing the net spacing.
With the elastic-plastic assumption the dotted curve A Fig. 4 shows the behaviour of the
excess pore pressure at the pile skin in terms of the initial pore pressure at the skin, ~0.
With the viscous assumption the pore pressure at the skin behaves as the dotted curve B.
It will be noticed that only the behaviour of the pressure at the skin relative to its initial
value is considered. The viscous assumption requires a very much larger absolute value of
pressure initially at the skin, but it is the relative behaviours that are compared here. On this
basis the curves are very much alike.
The radial distributions are also quite different between these two assumptions, but again
it is the behaviour at the pile surface which is significant and from this point of view the curves
are quite similar.
The real solution lies probably somewhere between these two curves and this uncertainty
will be reduced by taking a solution between the two. The real solution for a single pile in
an infinite medium will be close to curve I in Fig. 4.
The accuracy of this curve I will be quite good during the early stages of consolidation.
Z&O will be predictable within 10% of its value for any given time up to a.+0 = O-50. At
larger values of consolidation the solution will be progressively worse until it will be unreliable
after 80% consolidation.

COMPARISON T;Z’ITH EXPERIMENTAL DATA

Curve I, Fig. 4, can be used to calculate the horizontal coefficient of consolidation. The
relationship between c, and ch is not very clearly established, but it is known that ch is at least
equal to c, and is likely to be greater. The best check that can be made from the available
experimental data is to evaluate c, from this data by the use of Terzaghi’s empirical relation-
ship (Terzaghi and Peck, 1948). This will provide a lower limit for ch, and ch as calculated
from the diffsion theory can be compared with this empirical limit. This will not be a positive
proof but if the ranges of ch check it is likely that the theoretical structure is correct.
This comparison will be made between a theoretical 50% consolidation and a 50% increase
in bearing capacity as measured either by load tests or hammer blows. The bearing capacity
increase is measured as the percentage between maximum and minimum capacity, not in
absolute values. From curve I it can be seen that the 50% increase will occur at a time of
r21/Ch. This figure will be equated to the time at which the experimental pile reaches its
50% increase.
Seed and Reese (1957) reported a 6-in.-clia.pipe pile, driven into soft clay, on which
they made extensive measurements. At various times they measured the failure
loads after driving and the pertinent results are shown in Fig. 1. With the proper scale
adjustment this curve looks very much like curve I. This adjustment can be made by equa-
ting +I/& to 37 hours as can be seen by inspection of Fig. 1. Here r1 is 7.6 cm. With these
numbers the theoretical ch becomes 4.35 X lo-4 cm2/sec. This soil had a liquid limit of
41 and from Terzaghi’s relationship the range of c, will be from 10-3 to 10-4 cmz/sec.
As ch will be equal to or greater than c, this is a good check.
PILE TIME EFFECTS 223
1.00 t
.90

.80

.70

.60

.40

.30

.ro

Fig. 4.

In a discussion of reference No. 6, House1 gave data for a 16-in. O.D. pipe pile based on test
loadings of two full-scale piles in soft clay. This data is again shown in Fig. 1. In this case
the 50% time is 65 hours and the radius is 20.3 cm. This gives a theoretical ch of 1.76 x
10-s cms/sec. The liquid limit was roughly 30 which gives an empirical c, of the order of
10-a cme/sec. Again the check is as good as can be expected.
Yang reported the results of some redriving tests on a 117-lb. H pile in 1956. Two results
are shown in Fig. 1: one in 190 ft of silt and the other in 220 ft of the same silt underlain by
clay (Yang, 1956). These readings were taken on different piles within a cluster. While
data on the liquid limit is unavailable an interesting comparison can be made between silt
and the silt-clay combination.
By assuming that a cylinder can be substituted for the H section the equivalent radius
becomes 8.5 cm. For the silt the 50% time is about 15 hours and ch is 1.34 times 10-3
cms/sec. For the silt-clay combination the 50% time is 30 hours and ch becomes 6.70 times
10-4 cmz/sec. The difference between these two coefficients is just as would be expected.
It should be mentioned in passing that Yang also reported a reverse time effect in a sand.
That is the bearing capacity decreased with time in this sand. This is a much less common
phenomenon than the increase so frequently noticed in silts and clays and this paper will be
restricted to the latter case.
Bjerrum, Hansen and Sevaldsen (1958), reported the curve shown in Fig. 1 for the Horton
Quay. The piles were 35 cm square in clay. If the comparison is made on the basis of a
single pile, ch will be unreasonably small. The most likely explanation is that this pile is
part of a cluster that is diffusing as a group (thus rsr is very large) and this is, in fact, the
case as reported by Bjerrum.

INTERPRETATION SINGLE PILE SOLUTION

Curve I has some interesting ramifications if numerical values are assigned to y1 and ~1~.
If it is assumed that the single pile has a radius of 15 cm and that ch can vary between 10-i
and 10-4 cms/sec. +i/Ch will vary from 0.63 hours to 630 hours. This implies that 50%
of the “set up” of a single isolated pile will take place between these two time limits. Roughly
10% consolidation will have taken place in l/l00 of this time or between 23 seconds and
6.3 hours.
224 LARS 0. SODERBERG

These numbers in turn imply that with high consolidation coefficients a significant part of
the consolidation will take place between strokes of the hammer, while with a low coefficient
the driving time will be negligible.
In the worst case the driving time would have the same order of magnitude as the 50%
consolidation time and in this case the initial pressure assumptions that were made would
be questionable. However, ordinarily a pile will not have more than a lo-minute driving time
in silts and clays, and in these cases the driving time will be less than 20% of the 50% con-
solidation time. Therefore, in general, the assumed original pressure distribution is reasonable
in this respect.
In many cases it has become a common practice to interrupt driving for a few minutes in
order that the pile will “set up” and redrive at a specified blow count (Cummings, et al.,
1950). This practice is very nicely explained by these numerical values. An interruption of
10 minutes could develop as high as a 30% consolidation effect.

CLUSTERS CLOSELY SPACED PILES

It is of considerable interest to apply this reasoning to clusters of piles. It has been shown
that one pile affects its surroundings for a large distance outside the pile. It is therefore clear
that if there is a multi-pile cluster with spacings of the order of 2 pile diameters, the inter-
action will be serious with respect to this diffusion phenomenon because the single pile no
longer bleeds to infinity. This inquiry will be restricted to cases where this interaction is
complete or nearly so.
This problem is best understood if it is considered from two points of view. The first is
the individual pile with respect to the mass of piles that surround it, and the second is the
entire cluster with respect to its surroundings.
If one pile is driven it will diffuse as has been developed. If this pile is then surrounded
by a ring of piles the diffusion of the first will be affected. In addition, the hydrostatic excess
from the outer ring will bleed inwards and add to the normal excess pressures of the first pile.
If an additional ring is added, this phenomenon will compound itself. If enough rings are
added the tendency is for the hydrostatic excess pressures to build up to a fairly high value
throughout the cluster and for diffusion to take place from the perimeter of the cluster out-
wards.
The above reasoning is supported by the fact that it has frequently been observed in the
field that a large mass of piles will often show much less driving resistance than a single pile
or that of the first few driven in the mass. Real cluster will then behave as the ideal circular
one to a greater or lesser extent depending on the pile spacing and the driving time.
The second point of view is to consider the problem as a large mass, with the dimensions of
the pile cluster, and associated with a certain hydrostatic excess pressure distribution. This
pressure will decrease with time by the large scale diffusion phenomenon outwardly from the
cluster.
It should be emphasized that each case of a pile cluster should be considered on its own
merits. The assumption for horizontal consolidation is not likely to be valid unless it is a very
deep small diameter cluster. In addition when the horizontal dimension becomes large there
are liable to be other boundary conditions that limit the excess pressures. There will usually
be significant vertical flow and differences between ch and c, will have to be considered. The
assumption that ch is unaffected by driving must be made and is something that should be
investigated experimentally.
Nevertheless, these methods can be applied to clusters although they will usually require
the use of a computer for a three-dimensional case.
A number of general conclusions about clusters are possible. The time required for a
given degree of consolidation will depend primarily on r2i/Ch and the initial pressure distribu-
tion. This means that as an upper limit, the time difference between two clusters in the same
PILE TIME EFFECTS 225

location will depend on the square of their horizontal dimensions. This time will be reduced
somewhat by any vertical flow that is present. Going one step further, an upper limit for a
cluster compared to a single isolated pile will also depend on the square of the respective
horizontal dimensions. Here the initial pressure distribution as well as vertical flow will
reduce the time.

EXPERIMENTAL DATA
The present need is to have the results of a large number of re-driving tests in conjunction
with pore-pressure measurements and loading tests. To obtain this type of data does not
necessarily require a large expenditure of money. Much can be accomplished by simply
re-scheduling the standard load tests which are a common practice today.
These redriving tests can be used to make predictions of bearing capacity that could be
checked against the load-test results. These comparisons would, in turn, give much indirect
information about the consolidation coefficients and the pore-pressure distribution. These
comparisons would be even more informative if field data on the pore pressure distribution
and laboratory data on the consolidation coefficients were known.
It is hoped that this Paper will stimulate interest in gathering this important data.

CONCLUSIONS
1. The bearing capacity of a friction pile is dependent on the diffusion time of the hydro-
static excess water pressure that is created by pile driving. The time required for this
phenomenon to reach a specified state is proportional to the square of the horizontal dimension
of the pile, and is inversely proportional to the horizontal coefficient of consolidation.
2. A static load test on an isolated pile should never be relied upon to give the bearing
capacity of piles in a cluster. Consideration must be given to the diffusion time of the group.
3. In order to predict this diffusion phenomenon without extensive field tests a large back-
log of information on the coefficients of consolidation and pressure distributions will be neces-
sary. This information will necessarily have to be collected from field construction projects.

REFERENCES
BIOT, M. A., 1941. “General theory of three-dimensional consolidation.” J. AppZ. Physics, 12 : 4266430.
BJERRUM, L., HAi%EN and SEVALDSON, 1958. “Geotechnical Investigations for a Quay Structure in
Horton.” Norwegian Geotech. Publ. No. 28, Oslo. (See p. 12.)
BJERRUM, L., 1959, letter from Dr Bjerrum to author dated May 8, 1959, describing pore pressure measurc-
ments for a bridge in Southern Norway.
BJERRUM, L., and NILS FLODIN, 1960. “The development of soil mechanics in Sweden, 1900-1925”.
GCotechpzique, 10 : 1 : 1-18. (See p. 4.)
CARSLAW, H. S., and J. C. JAEGER, 1947. “Conduction of heat in solids.” Oxford Univ. Press.
CUMMINGS, A. E., G. 0. KERKHOFF and K. B. PECK, 1950. “Effect of driving piles into soft clay.” Trans.
A nzer. Sot. ciz~. E~zgrs, 115 : 275-286. (See p. 50.)
HILDEBRAND, F. B., ” Methods of applied mathematics.” Prentice-Hall.
NAD~I, A., 1950. “ Theory of flow and fracture of solids.” McGraw-Hill, N.Y. (See p. 479.)
RICHART, F. E., 1957. “ A review of the theories for sand drains.” Amer. Sot. civ. Engrs, SM3, July 1957.
SEED, H. B., and L. C. REESE, 1957. “The action of soft clay along friction piles.” Trans. Amer. Sot.
cio. Engrs, 122 :739, 83(SM3) :1301/l-38.
TAYLOR, D. W., 1948. “ Fundamentals of soil mechanics.” Wiley, New York.
TERZAGHI, K., and R. 13. PECK, 1948. “Soil mechanics in engineering practice.” Wiley, New York.
(See pp. 76, 77 and 228.)
Trn%osnENKo, S., and J. N. GOODIER, 1951. “Theory of elasticity.” McGraw-Hill, N.Y. (See p. 58.
equation 43-exraluated for case of zero outside pressure on a cylinder with an infinite outside diameter;
also p. 58 : equation 10.)
YANG, NAI-CHEN, 1956. “ Redriving characteristics of piles.” J. Soil Mech. Div., Amer. Sot. civ. Engrs,
x-01. 82, SM3. (See p. 17.)

S-ar putea să vă placă și