You are on page 1of 7

Jw,1955

181

The Bearing-Capacity of Eccentrically Loaded


Foundations on Sandy Soils
By W. Eastwood, B.Eng., Ph.D., A.M.I.C.E., A.M.1.Struct.E.

Introduction

N the past it has been usual when designing eccentrically loaded footings to assume that the pressure
distribution across the base of the footing wouldbe
linear and that the maximum value
of the pressure must
be limited to that permissible under a centrally loaded
footing.
Recently, however, Meyerhofl has suggested that a
more valid approachis to treat the eccentrically
loaded
foundationas if it were centrally loaded buthada
width equal to the actual width less twice the eccentricity (see fig. 1 ) . The results of experiments on small
scale model footings are reported which appear to confirm this hypothesis. For footings on sand theMeyerhof

theory gives somewhatsmaller ultimateloadsthan


the older theory except a t small eccentricities.
At the time that paperwas published the author was
investigating theeffects of eccentricity on stripfootings
on sand. His experiments on larger
scale models than
Meyerhofs gave results which are not in agreement with
that authors findings, and the object of this paper is
to report the resultsof these experiments and to postulate a possible reason for the disagreement.
The authors tests havealso indicated that there may
not be an ultimate load in the accepted sense if the
footing is completely restrained from slipping sideways.
The load increases indefinitely with increasing penetration of the footing.

FOOTINC

FOOTINC A

Fig. 1. Eccentrically loaded Footing A assumed to be equivalent

to centrally loaded Footing B

DETAILSOFAPPARATUSAND
INVESTIGATION
All the
tests
were
carried
out
on
sand
in
a
timber box braced with steel. The size of the box was
10 ft. x 5 ft. x 3 ft. deep.
The footings testedwere cut fromrolled steel channel
and were either 6, 8 or 10 in. wide. Grooves in which
the knife-edge loading device was located were cut at
various eccentricities (see Fig. 2). The thickness of the
channel under the knife edgewas so small that the
eccentricity was not appreciably affected by the tilt of
the channel under the eccentric test load. The
eccentricitiesusedwere up to b/6 (i.e. loading within the
middle-third) except for the 6 in. wide footing. I n this
case themaximumeccentricity wasalmostb/4,
i.e.
loading well outside the middle-third.
Fig. 2 also shows the method of applying the load.
To ensure that the footing could tilt endways or sideways without any restraint from the loading arrangement,thevariouspinnedjoints
wereincorporated.
The length AB was dso made fairly large (12 in.) so
that, if the footingmovedlaterallyrelative
tothe

loading beam during a test the inclinationof the thrust


through the loading arrangement
would changeonly
a negGgible amounr from ;he vertical. It was found
in the tests that
at the time the ultimate load
was
reachedthischange of inclinationwasnevergreater
than about io,and generally was much less.
It had been noticed in an earlier investigation that
even with a centrally applied load there was often a
measurable lateral movement before the ultimate load
was reached.Witheccentricloading
it was thought
that this lateral movement might
be considerably increasedunlesssome
restraint were applied. Inan
actualstructuretheremay
be little or norestraint
against lateral movement as in the case
of a footing
at the bottom of a relatively slender column, or the
structure may be so stiffened that lateral movement
wouldbe insignificant. Accordingly, it wasdecided
to repeatallthetests
usingtwo
separateloading
arrangements, one with lateral restraint of the footing
and one without (see Fig. 2).

182

The

i
LOADINC BEAM

I '

ERLL RESTRAIN

PINNED JOINT 0
FOOTINC

Fig. 2. Loadingarrangement
The sand used in the tests had a grading curve as
shown in Fig. 4. It was placed in 9 in. layersand
compactedto refusal bya ' Kango ' hammer.The
average density obtained was 108 lb. per cu. ft. After
each test the sand was dug over to a depth of 18 in.
(i.e. greater than the measured depth
of disturbance
of the sand in a test) and recompacted. The compaction was continued until a straight
screed run along
the top edges of the box produced negligible scrapings.
As the weight of sand in the box was unchanged the
mean density of the sand was also constant from one
test to another.
The lateral slip of the footing was measured continuously during the tests by means
of micrometer dial
gauges.A 30 in.longpointer,
part of which can be
seen attached to the far end of the footing in Fig. 3,
moving over a stationary scale gave the angle of tilt
of the footing, a correction being applied to the scale
reading to take account of the vertical settlement.
During a test the footings were driven into the sand
at approximately 1 in. perminute,measured
atthe
point of application of the load.

RESULTS OF TESTS
(a) Mode of Failure
( i ) Footings not restrained against lateral movement
In the tests in
which the footings were not restrained
against lateral movement the load increased steadily
withsettlementuntil
slip surfaces were suddenly
developed. This development of slipsurfaces
was
quite audible and the load instantly dropped to about
half its ultimate value. When the footing was driven
still further into the sand the
loadslowlyincreased
again, but in general the ultimate load had not
been
redeveloped at several times the settlements at which
it was first attained.

Fig. 3. Photograph of test on footing restrained


against lateral slip

June, 1955

B.S. SIEVE No.

100

90
80
70
U

vl

60

v)

2, SO
W

< 0

l-

z
30

20
10

O*/
C R A M SIZE

001

Fig. 4.

Gradingcurveof

I.0

10

mms.

sandused in tests

II. Zones of plastic shear.


Ill. Passive Rankine zones.

soil in region II moves outwards to


replace it.

Fig. 5.

Comparison of two-way failure assumed in some theories and one-way failure which

occurs in practice

The StructuralEngineer

184
Several of the more notable theories of foundations
are based on the assumption that for centrally loaded
foundationsfailure
occurs by slidingwedgesbeing
formedonboth
sides of the footing (see Fig. sa).
Most investigators have found,
however, that inpractice
failure occurs by sliding to one side only (see Fig. 5b).
Thiswassoineverytestexceptone
of the present
series, the dimension A in Fig. 5b being about 4b on
theaverageforcentralloading.
Althoughonly oneslipsurfacewasformed
at the
ultimate load, a further surface diddevelop later if the
footing waspushed far enoughintothesand.With
central loading this second sliding surface was usually
on the same sideof the footing as the first, but in about
25 per cent. of the tests it was on the opposite side.
Witheccentricloadingthefirstslidingsurface
was
invariablyonthesame
side astheeccentricity
(see
Fig. 6), and if the footing was then pushed further into
the sand a second surface was usually formed further
out on the same side. The ' outcrops ' of these first
andsecondsurfaces
were roughlyatdistances
of b
and up to 4b from the footing respectively, this latter
dimensiontending to be less for tests with large eccentricity.
I n a tleast one of the well-known foundation theories
it is assumed that the footing fails by rotation about

some centreas shown in Fig. 7a. Thetilt measurements showed that the rotation was always away from
the slip surface as in Fig. 7b.
Generally the angle of
tilt when theultimateload
wasreachedwasabout
l " for centrallyloaded footings, andasmuchas
8"
for eccentrically loaded footing, increasing with the
eccentricity.

(ii)Footings restrained against lateralmovement


There were some important differences in behaviour
when the footings were restrained laterally.
With central loading on the 8 in. and 10 in. wide
footings the first sliding surfacetoform
sometimes
outcropped at approximately one footing width away
on the side towards which the footing tilted, andsometimes at about three footing widths outon the opposite
side. There was not, however,
a
suddendrop
in
bearing power when this first surface formed as is invariablythe
case with unrestrained footings. The
load continued to rise as the footing was driven further
into the sand, although a t a decreased rate in general.
Even when further sliding surfaces were formed there
it appears that there is
was no drop in the load, and
no clearly defined ultimate load in the usual sense.

Fig. 6. Position of slipsurfacesintests

(ai) FOOTINC

ROTATION

ACCOAOINC TO SOME

THEOPIES.

Fig. 7. Direction of rotation of footings according to


some theories and actual direction as observed in tests

June, 1955

185

(b) Ultimateloads
The average ultimate load for three tests with each
set of conditions is given in table I .
Load 18in x 6in footing 18in x 8in footing 18in x loin footing
_____
eccenlateral Laterally No lateral Laterally No lateral Laterally
t i c i t y No
restraint
restraint restrained
restrained restraint
~

zero
b/18
b/9
b/6
2b/9

1.63
1.36

~
Q)

2.73 2.523.53
2.12 2.062.72
Q) E
2.20 1.77a 2
1.78 1.43
1.16 1.08

2.46
2.01

a2

~~

Table 1.
Average ultimate loads in ton per sq. ft. for tests on sand with
various eccentricities of loading.
f O O l l N C WIDTH

- IN.

Fig. 8. Variation of ultimate load with footing width


and eccentricity of application (no lateral restraint)

m
so
.l

The results for the tests without restraint are plotted


in Fig. 8. I t will be seen that the results for all eccentricities are in accordance with previous investigations
using centralloading,theloadperunitarea
for a
giveneccentricityratiobeingnotquiteproportional
to thefooting width.
Theresults for thetestswithandwithoutlateral
in
restraint for the 6 in. wide footings are compared
Fig. 9. From this figure it will be seen that the partial
restraintcausedanaverage
increase intheultimate
load of about 6 per cent.
Figs. 10 a, b and c compare the variation of bearing
valuewitheccentricitywhichwasobtained
in the
testswiththat
whichwouldbe
expected according
to the Meyerhof theory, and also with the older theory
in whicha
straight line pressuredistribution,the
maximum value of which is constant, is assumed. I t
will be seen that for the 8 in. and 10 in. wide footings
the older theory gives bearing values in closer agreement with the experiments than the Meyerhof theory
throughout the range of eccentricities. With the 6 in.
footing each of the theories agrees more nearly with
theexperimentsovercertainranges,butthe
older
theoryhasasmallermaximumdeviationand
also
has the advantage thatdoes
it not give an over-estimate
for any eccentricity.

0
LOAD ECCENTRICITY

Fig. 9. Effect of partially restraining lateral movement


on ultimate load for 6 in. wide footing

Similar behaviour was recordedfor eccentric loading


on the 8 in. and 10 in. wide footings.
Unfortunately, the restraining device was incapable
of preventinglateralmovementcompletely
because
of the slight play whichwasnecessary
to allowunrestrictedverticalmovement.Themagnitude
of this
lateralmovementvariedbetween
0.01 and 0.025 in.
This is quite small compared with lateral movement
in unrestrained tests(generally between 0.1 and 0.2 in.)
but was sufficiently large nevertheless for an ultimate
load to be obtained in the normal way for some of the
6 in. footingtests.Thevalue
of theultimateload
obtained in these tests
was somewhat larger than in
the completely unrestrained tests as will be seen later.

DISCUSSION OF RESULTS
The present tests are obviously insufficient in scope
to draw general conclusions for all types of footing.
Buttheydoindicatethatundercertainconditions
a t least the old-established assumptions may be better
than Meyerhofs suggested alternatives. Since the
tests were very similar to some of those carried out
by Meyerhof except that they were to a much larger
scale (Meyerhofs footings wereonly
1 in. wide in
general) a reason was sought for the apparent differences in the two sets of results.
A probablereason for the divergence is apparent
fromanexamination
of Meyerhofs apparatusand
method of test. Photographs in his paper show that
theloads wereapplied byanarrangementapproximately as in Fig. 11. It will be seen that as thefooting
tilted under the actionof eccentric load the eccentricity
would also increase. As no measurements of the angle
of tilt appear to have been made it is probable that
nocorrectionwasmade
tothe eccentricity value.
Thus the ultimate loads obtained
will correspond to
greatereccentricitiesthanthereported
values. The

The Structzcral Engineer

186

+-+

TEST RESULTS

0 . 5 1 * - 4 USUAL THEORY
O- --d MEYERHOf

ECCENTRICITY

ECCENTRICITY

Figs. loa, b, and c. Comparison of test results with usual theory and Meyerhof hypothesis
for 6in., 8in., and loin. wide footings

r
Fig. 11. Changeof eccentricity as footing tilted in Meyerhofs tests

June, 1955

187

0 6

D- ---Q

MLYLRHOf THEORY

olb

OOJb

Oysb

ECCENTRICITY

Fig. 12. ComparisonofRamelotandVandeperre'stest


results on square footings with the usualtheory
and the Meyerhof hypothesis

error in the
eccentricity was probably quite considerable
if the angle of tilt was as great in Meyerhof's tests as
it was intheauthor's.
At ultimate load the value
variedbetween
about 0.5" with central loading to
about 6" when the eccentricity was b/6.Thus if the
ratio h/b in Fig. l l were say 0.5, the actual eccentricity
atultimate load in a test in which it was initially
b/6 would be b/6 4.b/20 approximately.
Ramelot and Vandeperre have also reportedthe
results of some tests on square and circular footings on
sand, usingeccentric loading. Theauthorhas
compared their results for square footings with the usual
theoryandthe Meyerhof hypothesis. Although footings of variouswidths were tested, only in the case
of the 30cm. square footing were sufficient repeat tests
madetodrawanaccurategraph
of ultimate load
against eccentricity for eccentricities up to b/6.This
graph is shown in Fig. 12. Thereis some scatter of
the points partly because some of them represent only
a single test, but again it willbeseen that the usual
theory is upheld rather than the Meyerhof hypothesis.
CONCLUSIONS
( 1 ) Thetests carried out by theauthor on strip
footings indicate that the usualpractice of assuming
thatthere isa straight line distribution of pressure
under an eccentrically loaded foundation,andthat

the ultimate value of that pressure is the same as that


under a centrally loaded foundation, is sound for
footings on sandandeccentricities
u p tob/6.
An
alternative hypothesis put forward by Meyerhof, in
which theultimate load for an eccentricallyloaded
footing isassumed to be equal to that for a footing
of width equal to theactual widthminustwice
the
eccentricity, does notagree so well withsome of the
experiments.
(2) Footings on sand which arerestrained
from
slipping
sideways
have no definite ultimate load
whether centrally or eccentrically loaded, there being
no sudden drop of bearing power when slipsurfaces
are formed. When only partial
restraint
against
lateral movementis
allowed, theremay be a welldefined ultimate load somewhat
higher
thanthat
obtainedwith no lateralrestraint of the footings.
Acknowledgments
Theexperimental workdescribed inthispaper
was carried
out in the University of Aberdeen,withtheactivehelpand
encouragement of Professor Jack .411en, D.Sc.,M.I.C.E.

References
(1) Mcyerhof, G. C;. (1953).

The Bearing Capacity of FoundationsUnderEccentricandInclinedLoads.


Proc.Third
Int. Conf. Soil Mech. and Found. Engineering, Vol. 1 . p. 440.
( 2 ) Ramelot, C. andVandeperre, L. (1950). T r a v a u x de la
Cornrnissiosz d'Etude
des
Fondations
de Pylones. Compt.
Rend. Rech., I.R.S.1.-4., Brussels, No. 2 .

You might also like