You are on page 1of 25

DE BEER, E. E. (1970).

Gotechnique

20, No. 4, 387-411.

EXPERIMENTAL DETERMINATION OF THE SHAPE FACTORS AND THE BEARING CAPACITY FACTORS OF SAND
E. E. DE BEER*
SYNOPSIS In order to determine the values of the shape factors to be introduced in the formula of ultimate bearing capacity of shallow foundations, an extensive series of tests on small footings resting on fine sand was performed at Ghent. To obtain a homogeneous and given density the sand was placed in a fully automatic way, under control of the output and the height of fall of the sand. To eliminate the depth effect the overburden pressure was realized by air pressure in inflated rubber bags covering the surface of the sand. From the tests formulae for the shape factors sq, s, and sy are deduced. It appears that the shape factor sy in the weight term is independent of the angle of friction, while the shape factors sq and s, are not. In the formulae given for sq and s, the influence of the state of strain is implicitly included; however, different formulae are given depending on whether the effect of the curvature of the intrinsic curve on the introduced shear strength parameters has been considered or not. Afm de determiner les coefficients de forme a introduire dans la formule de la capaciti: portante limite sous des fondations directes, un t&s grand nombre dessais sur petites semelles reposant sur un sable fin ont CtC effect& a Gand. Afin dobtenir une densite donnee, homogbne, le sable a Cte place dune facon completement automatique, permettant de regler la hauteur de chute et le debit. Afin deliminer linfluence du coefficient de profondeur, la surcharge laterale a Cte rCalisCe au moyen dune pression dair dans des sacs en caoutchouc recouvrant la surface du sable. A partir des essais des formules ont et.6 Btablies pour les coefficients de forme sq, s, et sy. On constate que le coefficient de forme sy dans le terme du poids-propre du sol est independant de langle de frottement interne, tandis que par centre les coefficients sq et s, varient avec cet angle. Dans les formules donnees pour sp et s, linfluence de letat de deformation est imphcitement inclue. Toutefois les formules sont differentes, selon que le parametre de cisaillement introduit tient compte ou non de la courbure de la couche intrinseque du materiau.

INTRODUCTION

As proposed by Brinch Hansen (1961) the ultimate loaded footings can approximately be expressed by
~5, = N,d,s,i,q+

bearing capacity syiyyk, ib


. .

under centrally
. . .

N,d,s,i,c +

-2-

InIthe tests carried out only vertical loadings were considered. simplified to 9, = N,d,s,q+
NC&c+ 2 syyk, ib

Therefore equation (1) can be

(2)

Applying the theorem of the corresponding states of Caquot, it can be shown that N, = (N,-l)cot+ . . . . . . . . (3)

Therefore even if the tests are performed on cohesionless materials, the values of s, and d, can be obtained from expressions (4) and (5).
* Professor, Universities of Ghent and Louvain; Director of the Belgian Institute of Soil Mechanics. 387

E.

E.

DE

BEER

NOTATION

width of the footing, m cohesion, t/m2 depth factor for the cohesion term depth factor for the overburden term diameter of the particles corresponding to the ordinate of 10% of the grain size distribution diagram coefficient of uniformity, d,,/d,, coefficient of uniformity, d,,/d,, inclination factor for the cohesion term inclination factor for the overburden term inclination factor for the weight term length of the footing, m percentage of voids maximum percentage of voids minimum percentage of voids unit load on the footing, t/m2 unit rupture load on the footing, t/m2 initial lateral overburden pressure, t/m2 lateral overburden pressure at the moment of rupture, t/m2 shape factor for the cohesion term shape factor for the overburden term shape factor for the weight term settlement of the footing, m settlement of the footing at the moment of rupture, m depth of the foundation underneath the soil surface, m dimensionless quantity defined by the expression (10)

points in the load settlement graphs corresponding to the rupture criterion of Christiaens (1966) relative density, (nmsx-4/@max-niZmin) modulus of elasticity of the solid particles points in the load settlement graphs corresponding to the rupture criterion of Bent Hansen (1961) bearing capacity factor for the cohesion term bearing capacity factor for the overburden term bearing capacity factor for the weight term hydraulic radius of the footing, 4x1 m dry weight, t/m3 effective volume weight of the soil underneath the foundation level, t/m3 shape ratio, b/l specific weight of the grain material, t/m3 normal stress component, t/m2 mean normal stress, t/m2 normal stress on the sliding plane, t/m2 mean value of the normal stress on the sliding surface, t/m2 angle of internal friction secant angle of internal friction angle of internal friction, varying with the percentage of voids, and the mean normal stress along the rupture surface angle of internal friction, varying only with the percentage of voids perimeter of the footing, m surface of the footing, m2

D= Eq H

N Ncl N* R
Yd Yk.1

6 6, (T
=rn

ug

0g.M

VW x w

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF

SAND

389

Table

1.

Characteristics

of the Mol sand

Mean grain size Degree of uniformity

d,, = 0.19 mm 0.135 d g=cO=m=

0.66 0.56

d g=z=w= Effective grain size Sphericity Roundness Specific weight &linimum density ILZaximum density d,, = 0.79 0.40 8, = ya = yd =

0.15

5.135 mm 2.65 t/m3 1.391 t/m3, nmax = 47.5oj, 1.696 t/m3, nmrn = 36%

I.45 is 0.1 is 0.2 44 0.3

I.50 43 Oh 42

I.55
yk :

I.60

I.65 is 1 0.8 37 d.9

I.70 36 I!0 8 I.1

t/d

41

4b n:7. d.6 Dr 0.7

0:5

Fig. 1

The aim of the tests was to determine the expression of the shape factors sq and s,. From the tests it was also possible to deduce the variation of the bearing capacity factors N, and N, in relation to the relative density.
SAND USED

Its principal characteristics are given in All tests were performed with dry Mol sand. Table 1. It is a uniform fine sand, composed almost exclusively of quartz. The angles of shearing resistance as obtained in normal triaxial tests are given plotted against relative density in Fig. 1 (de Beer and VesiC, 1958). They vary between 29 and 45. Under normal triaxial tests, tests are run with three different cell pressures, e.g. 5000, 10000 and 15000 kg/m2, and a common tangent is drawn as nearly as possible through the origin. In such tests the discrepancies between the circles and the origin are levelled out although they are often wrongly considered as testing errors. The intrinsic law of a cohesionless material for a given density is not a straight line, but a Thus for a given density the shearing strength of a curve turning its concavity to the (5 axis. cohesionless material cannot be expressed by a unique value of the angle of shearing resistance. In order to account for the influence of the normal stress on the shearing strength parameter the value of the secant angle & is introduced, obtained by ,drawing the tangent to the Mohr circle defining the limit state of equilibrium through the orlgm (Fig. 2).

390

E. E. DE BEER

Fig. 2

/Top

silo

Fig. 3.

View on the sand container with rubber bags

600

Propulrion of
the trolley

Propulrion of the vertk.I translation of the bottom place

1.44

I .48
yJ :

I .52 dry unit weight:

I .56 f/m3

I.60

Fig. 4.

Set-up

of the test apparatus

Fig. 5

2 T

Fig. 6 (right)

IO

After

Verit

(1963)

footing

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF SAND

391

From triaxial tests performed with different relative densities and different cell pressures, Ladanyi (1960) determined for the Mol sand the variation of the secant angle. He published a diagram giving the variation of the secant angle against the relative density D, with the ratio CT,..JE~ as a parameter, where (T, is the mean normal pressure existing on the sample at rupture and E, is the elasticity modulus of the quartz constituting the grains.
THE TESTS

All the tests are small-scale. The width or diameter of the small footings varies between 36 and 150 mm. The shape of the footings is characterized by the ratio 6 = b/l, where b is the width and 1 is the length. Two kinds of footings were used, one with S=l (circular and square footings) and another with 6 = 4. The base of all footings was covered with a rough material to give the case of a perfectly rough foundation. A series of tests was performed with footings at the surface of the sand, in order to determine the factors N, and sY. A second series of tests was performed with various overburden pressures q. Most of these tests were performed with overburden pressures of 1 and 3 t/m2. In order to separate the variables the overburden pressure was obtained by covering the surface of the sand with air inflated rubber bags, covered by a steel plate (Fig. 3). In that way the overburden pressure is exerted by a material without shearing strength, and the influence of the approximate value of the depth factor d, is therefore greatly reduced. As both the bearing capacity and the angle of shearing strength are extremely sensitive functions of the density, utmost care is needed to obtain a homogeneous density and to determine as accurately as possible the actual density. In order to obtain a homogeneous density the set up of the apparatus was as shown in Fig. 4. The sand ran from a silo into two lateral containers, which automatically filled a spreader box with a split of 3.26 mm running with a constant speed back and forth over the test box. The density obtained depends on the height of the fall and on the sand output. In order to maintain a constant height of fall while filling the test box, the bottom of the test box is lowered automatically in proportion to the increase of the height of sand in the test box. The relationship between the density and the height of fall for a speed of translation of the spreader box of 130 mm/s is given in Fig. 5. By increasing or decreasing the speed of the spreader box, for a given height of fall, higher or lower densities can be obtained. In all tests the water content was below 0.1%. A total of 662 tests was performed, 350 at the surface, and 312 with an overburden pressure.
CRITERION OF RUPTURE

As is also stated by VesiC (1963), the rupture underneath a footing may be obtained by a general shear failure, by local shear or by punching. The appearance of one of these phenomena depends on the relative density and on the overburden pressure. If the ratio q/yk, iR or z/R is introduced, where z is the depth of the foundation under the soil surface, yk,i is the effective volume weight of the material under the footing plane and R is the ratio of the surface w of the footing to its perimeter x, W/X;the zones of general shear, local shear and punching shear are located as shown in Fig. 6 (VesiC, 1963). As the overburden pressure increases, the zone of general shear has a tendency to disappear. For very heavy relative overburden pressures only the phenomenon of punching remains. In the case of general shear there is no difficulty in determining the rupture load, but in the case of local shear and punching it becomes difficult to define clearly the value of the ultimate bearing capacity. When the footingissinkinggradually into the soil, the overburden pressure increases and, in the case of very low densities, the density also increases. Both phenomena

392
0 4 8 I2

E.

E.

DE

BEER n:t/ml -20

16

24

28

32

36

E P E 16 i ; f 14

220

\
242

,
230 244 248 239 246

32 OC . B

qo=o

Fig.

7 (above right)

and

_. T
Footing %JzJ$Y$

150 mm

dia.

Fig.

8 (left).

Rupture

criterion

of Brinch

Hansen

(1963)

produce a gradual increase of the bearing capacity, and in a load-settlement diagram a continuous increase of the load against settlement is found (Fig. 7). For loading tests at the surface of the sand Bent Hansen (1961) has given a method which eliminates the influence of the increasing overburden pressure. However, for loading tests with an initial overburden pressure this method does not work quite satisfactorily. Brinch Hansen (1963) has defined the ultimate bearing capacity as the stress, for which the strain is twice the strain at a 10% smaller stress (Fig. 8). If the settlements are represented by ZJ Aw -=-

wr

1 2

for

AP

pr=TO

AP -= &

1Aw -_ 5w,

(6;)

SHAPE

FACTORS

AND

THE

BEARING

CAPACITY

FACTORS

OF

SAND

393

This can be expressed

more generally

as dp -_=ap dlnp dw w . . . . . . . . as . . . . . . (9) . . . . . . . . . . (7) (8)

= adlnw

where a is a characteristic value depending on the relative density. In order to obtain quantities without dimension, this is expressed dlnL where, according to Bent Hansen, = adlnW

Ay,, &

the factor A for the present

tests is in a first approximation . . . . . . (10) and (11) (12) (13) (14)

A =(-&+;)~,(1+035;)+~
The expression of the factor Brinch Hansen. A is based on the following N,=NY s,=l-0.4; s,=1+0.2;
d, = 1+0.35;

assumptions . . . . . . . . . . . . . . . .

of Bent Hansen . . . . . . . . . . . . . * .

. . . .

Christiaens (1966), found that by drawing the values of w/b against plAy,,ib scale, a diagram is obtained which in many cases consists of an upper curved part which is a straight line (Fig. 9). As the value of a is a measure of the diagram, the intersection of the curved part and the straight line can be rupture point. The criterion defined by Christiaens is therefore in close

on a double log part and a lower slope angle of the considered as the relation with the

Footing: 38 mm dia. qo= I t/m2

Fig. 9.

Rupture

criterion of Christiaens (1966)

394

E.

E.

DE

BEER

criterion defined by Brinch Hansen, the only difference being that a constant value for a is not introduced, but a value varying with the relative density, For some tests it was impossible to deduce a well-defined intersection point. In such cases an upper and lower limit for the ultimate bearing capacity is defined (points C, and C, on Fig. 9). With the criterion of Christiaens it is finally possible to deduce from the tests the corresponding ultimate bearing capacity 9,.
DEDUCTIOK FRO35 LOADINGTESTS AT SURFACE OF FIRST VALUE OF N,s,/Z VERSUS $(a CT~,~) APPROXIhfATE

The shearing strength of a cohesionless material for a given density cannot be defined by a unique value of the angle of shearing strength, as the intrinsic law is curved, and furthermore another intrinsic law is found for each state of strain (plane or triaxial). From a great number of triaxial tests performed with different densities and different mean normal pressures, the secant angle of the Mol sand in relation to the relative density D, and the mean normal stress (T, is the variation at disposal (Ladanyi, 1960). Fig. 10 shows the variation of the secant angle 4 plotted against the relative density, and considers the value of the normal stress (TV in the shearing plane. Meyerhof (1950) has shown that the mean normal stress (T=,~ along the shearing plane underneath a footing is about one tenth of the ultimate bearing capacity $J,. The value of (J=, M can also be expressed by ~~,M=p*(l-sin+)

Following Meyerhofs work the assumption is now made that the curved intrinsic law AMN may be replaced by the straight law OMN defined by the secant angle +(n, Us, M) corresponding to the mean normal stress along the shearing surface (Fig. 11). This is a very rough approximation, but it eliminates intricate calculations based on the curved shape of the intrinsic curve, and finds justification in the good correlation which Meyerhof (1950) obtained between the values calculated with this approximation and his test results. Further, the triaxial angle d(rt, a,, M) is used irrespective of the shape of the footing. This means that the influence of the state of strain on the shearing characteristics is not directly taken into consideration. Therefore this influence will be hidden in the experimental values of NPs, and N,s,.

44

Fig.10

Fig.11

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF SAND

395

The way in which the value of N,s, can be deduced from the tests without overburden pressure is now described. Equation (2) shows that at the start the problem is indeterminate. For loading tests without overburden equation (2) can be written as NY J, = Nsdsss~lr, $@a/,+~ %yk, which includes the unknowns N,s,
s 8 . . . . . . (16)

and N,s,. y2
-Y=-_

Solving for s,,N, gives

Pr
Yk, ib

sNd qqqb

(17)

Solving

for N, gives

N, =
yk, ib s,dq~

P=
+ S, &f

1 9

......

(18)

In order to find a more approximate value of N, expressions (1 l)-(14) are introduced into the second term of equation (18). Introducing this value of N,, together with expressions (13) and (14) for s, and d, into (17) one obtains a first approximate value N$%$l) of Ngsy. The value of ug, M can be determined from equation (15) and, from this value and the value of D,, one can determine the value of +(rt, a,,,). For each tested footing the values of Nkl)~$~)/2can be plotted against the secant angle +(% ug,M). On Fig. 12 the experimental points corresponding to rectangular footings (S=&) are shown by small rectangles, those corresponding to square footings by squares and those corresponding to circular footings (6= 1) by circles. The curve AB gives the mean curve obtained for all circular footings irrespective their size, and the curve CD the mean curve for all rectangular footings (6 = 6) irrespective their size. From the ratio of the ordinates of the curves AB and CD it can be deduced that the shape factor sp) can be expressed by $,l) = l-04b/l . . . . . . . . (19)

if

IO.8 12.3
IS.9 13.8 IS.8 225 27.6 34.5 44-4 60.6 85.2

19.1 16.8 ;::: 29.2 4P9 34.9 53.6 69.0 94.2 132.5 6=&

20.5 l&O 265 23.0 31.3 46.0 37.5 57.5 7+0 101.0 I420 6=0

:t=0 40 ::a 0 43 44 45 46 -___

s=

Fig.

12

396

E.

E.

DE

BEER

Using this expression for 6 = 0 the values Nil)/2 of the curve EF as a function of #lz, Q, J are obtained. The ordinates corresponding to the same values of +(n, Up,M)are compared; thus the derived shape factor does not directly give the ratio of the bearing capacity of two footings of different shapes and with the same small dimension placed on a sand with a given relative density. The defined shape factor is related to the case of two footings with the same small dimension placed on sand with the same straight substitution line of the intrinsic curve.
DEDUCTION OF VARIATION OF FIRST APPROXIMATE TESTS WITH OVERBURDEN VALUE OF Nbsb PRESSURE FROM LOADING

From equation (2) $J~may be given as p, = N~s~d,(q, + yr&?$) +y s$lylc,JJ . . . . * (20)

where q,, is the initial overburden pressure and d, = 1 +O-35w,/b. By introducing into equation (20) the approximate values N$%il), an approximate value Ny)s!$) for N,s, is obtained. As q,, is large compared with yk,,b the error introduced in the calculation of N!$)shl_) by introducing an approximate value N$!)s$~)of N,s, is rather small. From equation (15) Us,Mis calculated and with this value and the value of D, from Fig. 10 the value of +(n, CT,,~). For each footing and for each value of qo, Nhl_kh]) can be plotted against #(n, up,M).

I
I 44.

36

Fig. 13

27.0 35.0 50.0 75.0 120.0

ii3

Il.7 12.8 14.2 16.0 18.3

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF SAND

397

All experimental points for circular footings are shown on Fig. 13. The same mean curve AR is drawn, based on the points for all circular footings (6 = 1) and for all values of q,,. On Fig. 14 are shown all experimental points for the rectangular footing 38 x 228 mm. Similar diagrams have been drawn for the other rectangular footings which were used. Based on the points for all rectangular footings (6 = $J and for all values of q0 the same mean curve CD has been drawn. The curves AB and CD are shown on Fig. 15. From the ratio of the ordinates of these two curves it may be deduced that b sql) = l+dtan+(rz,u,,,) . . . . . . . (21)

Based on this value, the curve EF for the values of NY) against +(Pz,(TV, M)is obtained. The shape factor, expressed by equation (21) is not related to the case of a sand with a given relative density D,, but to the case of sands characterized by the same value of #(n, ug,M), and thus by the same straight substitution line of the intrinsic curve.
DEDUCTION VARIATION FROM LOADING TESTS WITHOUT OVERBURDEN OF MORE EXACT VALUE OF N,s, IN FUNCTION PRESSURE OF OF $(n, (I%,~)

With equation (18) deduced from the general law (2) a more exact value of N, can be obtained by calculating the ratio syN,/2N, of the second member from the approximate

1:o
0

33.0 273 40.0 50.0 63.5 85.0

Fig.

16. Regression analysis : circular footings, q0 = 0

Fig. 15

___-

2,, 45' 46 47"

I IS.0 M

398

E.

E.

DE

BEER 2

Table

Data regression analysis syN,/2 = F[$(n, Q, M)] Footing Limits 4(n, c~, M)

! Number
of tests

log syN,/2 =a+(%, og, M) -b

Co;~daton

Fig. 16

37 55 42 40

41 45 46 40 41 45 46 55

33 39 37 33

0.064 0.114

1.210 3.308 0.874 2.435

0.737 0.736 0.789 0.863

Fig. 17

1 Rectangular I footings

36 50 42

0.060 0.098

Data regression analysis ssiV, = F ;+(n, (tg,M)] Footing Limits +(ti, ug, M Number of tests log ssiVs =

a+,

og,

M) -b

Correlation factor

Fig. 19

Circular

32 50

footings 6=1

/ 41
- 33 45 39 43

t::

40

;: 71 114 34

0.069 0.249

Fig. 20

Rectangular footings S=&

39 42 50 46 55

-I

0.068 0.091 0.292

1.318 2.220 10.758

0.767 0.767 0.913

values s$~)N$~)and NY) already (21) giving

obtained.

Further,

sq can be calculated

with the expression

(22)

A better approximation for syN, is obtained by introducing this value of N, and that of s, from equation (21) into equation (17). The experimental points obtained for all circular footings (S = 1) are shown on Fig. 16. The linear regression analysis could be applied on the obtained points. However, on a semilog diagram the variation of s,N, is not exactly linear but is represented by a slight curve with its concavity turned to the large values of jyN,. Therefore the experimental points were not considered as a whole, but were divided into groups on which the linear regression analysis was applied separately. From the linear relationship obtained a smooth curve a/3 has been deduced. The data of the linear regression analysis are given in Table 2. The experimental points obtained for all rectangular footings (6 = Q) are shown on Fig. 17. As for the circular footings the mean curve yS has been obtained. The data of the linear regression analysis are given in Table 2. The two curves a/3 and yS are shown on Fig. 18. From these the value of (syNy)d=1,6/(sYNy)6=1 can be deduced. The ratio appears to be independent of the value of +(n, (TV, M). Its value is to have

[(sY),=~,GI/II(s~)~=~I = 1.476
In order to be able to determine correctly the variation of s, in B = b/l, it is necessary experimental data for some other values of the shape ratio 6.

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF

SAND

I
20.9 23.0 25.5 29.2 34.7 40.3 48.7 59.0 73.8

Fig.

93.7

125.5 169.7

17. Regression analysis: rectangular footings, qo=O,

S=b/f=1/6

YS

Curves drawn AB CD

ar sight EF 18.0 20.2 23.0 26.5 31.3 37-5 46.0 57.5 74.0 101.0 142.0 208.0 6=0

I
IO.7 12.0 13.8 15.9 IS.8 22.5 27.6 34.5 44.4 60 6 05.2 122.0 s= I

lb-7 18.9 21.4 24.7 29.2 34.9 42.9 53.6 69-O 94.2 132.5 192.0 -__ s=_6

Fig.

18

With lack of such data, an approximate relationship between sy and S may be obtained in one of the following two ways. First, based on previous test data (de Beer and Ladanyi, 1961) one could in a first approximation assume that for S = 1 the value of sy is 0.6. If this assumption is made, the experimental value of (s~)~=~,~should be given by (s&=~,~ = 1.476(s,),=, Now two different expressions of sy 1+0*2S SY=m s> = l-O.46 = 1.476x 0.6 = 0~3857

* .

. .

. .

. .

. .

. .

. .

. -

(23)

(19)

400 both give sy= O-6 for S = 1.

E. E. DE BEER

For S = 6 the first expression gives

The second expression gives s,, 6=1,6 = 1 -y = 1 - 0.0667 = 0.9333

Therefore, if s,, d= r = 0.6 expression (23) should correspond better with the experimental results than expression (19). Second, however, it is possible to make no assumptions concerning the value of sy for S = 1, and to assume arbitrarily a linear relationship between s,, and 6. The tests give s,=I-ms
sy,6=1/6 =

(24)

1.476s,,,=,

l-;

= l-476(1 -m) 0.476 m = = O-363 1.318 s, = l-0.3636 . . . . . . . . (25)

For S= 1 s, = 0.637 The experimental results are therefore also covered by expression (25), in which case for foottings S = 1, the shape factor s,, related to #(n, (TV, M) is no longer 0.6 but 0.637. However, as long as experimental data for other shape ratios are not available, it is impossible to find out which of expressions (23) and (25) is the more exact. It can be seen that the difference between the results is not very great, and that for practical purposes one of the expressions (19), (23) or (25) can be used. The following deductions use expression (23) giving s, = 0.6 for S = 1. Given the expression 5 shown in Fig. 18 gives the variation of s, the values of NJ2 can be calculated. The curve l ). For comparison also shown are the curves AB, of NJ2 against the secant angle +(n, (T~,~ It can be seen CD and EF corresponding to the first approximate values N$l)~$,~)/2 of N,s,/2. that the difference between the value of the first approximation and the more exact ones is not negligible.
DEDUCTION FROM LOADING TESTS WITH OVERBURDEN VARIATION OF MORE EXACT N,s, AS A FUNCTION PRESSURE OF +(n, Q,~) OF

Knowing the more exact values of syN,/2 (curves c+Iand yS in Fig. 18), these values can be introduced into equation (18) to give a more exact value of N,s, . . . .

(26)

A more exact value of ssN, can now also be deduced from the tests without initial overburden pressure (q,,=O). Equation (18) can be written as ssN, =

2%

......

(27)

In equation (27) are introduced the values of syN,/2 from the curves c@ or yS in Fig. 18, and

SHAPE

FACTORS

AND

THE

BEARING

CAPACITY

FACTORS

OF SAND

401

the values of shl)Nhl)from the curves AB or CD of Fig. 15, and thus a more exact value of s,N, is obtained. For all circular footings the points representing the values of ssN, in function of +(rt, ug, M) are shown in Fig. 19. All points are shown, irrespective of the diameter of the footing and the overburden pressure. The open circles represent the points corresponding to the tests with an initial overburden pressure, and the solid circles those corresponding to the tests without an initial overburden pressure. The circles are quite well related. The experimental points are divided into two groups, and for each group the linear regression analysis is applied. From the two straight lines shown in Fig. 19 the curve cls is deduced. The data of the linear regression analysis are given in Table 2. As all points are located around the same curve, irrespective of the value of the overburden pressure q,,, it is obvious that if the values of ssN, are drawn against the secant angle +(n, ug, M) the values of sPN, are independent of the overburden pressure. For all rectangular footings 6 = 8, irrespective of their width or the initial overburden pressure, the points giving ssN, against +(+z,Ok,J are shown in Fig. 20. The open circles represent

l-

36'

40'

17.0 15.2 21-4 19.0 24.6 28.5 42.0 34.4 72.0 53.0 173.0 107.0 308.0 660.0 da' Fig. Regression 19. circular analysis: footings Fig. Regression 20. analysis:rectanguh&oomgs: S= b/l

IO.2 II-2 12.5 13.9 IS.8 18.1 21.5 26.0 32.5 43.5 64.0 102.0 180.0 380.0 Y6

:z* 0 Zk0 :L0 41"

E.

E.

DE

BEER

I
t >
D 0

d(n. ~&Y.M)

SqN,
analysis

Regression

: I
Curves AB

drawn CD

at sight EF

16.2 17-6 IV-5 21.9 25.0 28.8 34.2 43.6 57.2 82.8 25.5 !lO.O 6=l

Fig. 21

the results of the test with initial overburden, and the solid circles those of the tests without The experimental points are divided into three groups, giving initial overburden pressure. three straight lines, from which the curve yS is deduced. The data of the regression analysis are given in Table 2. From the ratio of the The two curves c$=c(/? and y6 are shown together on Fig. 21. ordinates of these two curves, the following expression for sq is found

s, = l+ptan

[+(n,a,,,)]

The variation of N, against Given the values of s,, the values of N, can be calculated. IS iven by the curve 65 in Fig. 21. d(V %.M) . g Also in Fig. 21 are given the curves AB, CD and EF corresponding to the first approximation shl)N&l) of s N Again the de&r&l shape factor does not give the ratio between the bearing capacities of two footings of different shape but of the same small dimension, placed on a sand with a given density, but the defined shape factor is related to the same value of $(n, u=, M).
COMPARISOS OF THEORETICAL AND EXPERIMENTAL VALUES OF N,

In Fig. 22 the angle of shearing strength 4 is plotted against the theoretical values of NJ2 as given by Brinch Hansen (1961), Buisman (1940), Kerisel and Caquot (1956), Lundgren and Mortensen (1953) and Meyerhof (1955). The curve l 5 represents the experimental values of N, against the secant angle +(n, cr=,M). The experimental values are seen to be lower than the values of Lundgren and Mortensen (1953). That the experimental values are lower than the theoretical ones can be explained easily by the fact that the theoretical values are based on the assumption of a general shear failure of a material with constant volume. In fact at medium and low densities the failure is

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF

SAND

403

.
I6 Fig. 22 &,, Fig. 23

32.

40

induced by local shear failures or by punching, phenomena which are not considered in the theoretical deductions. It can also be seen that at high relative densities the experimental curve has the tendency to give higher values than the theoretical ones. Also at low densities the experimental values of NY/2 tend to become larger than the theoretical values. An explanation for this anomaly can be sought in the fact that in very loose soils as soon as the footing starts to penetrate into the soil an increase in density occurs. Therefore to be correct the values of NY/2 should not be related to the angle $(n, ug, J corresponding to the initial relative density, but to a higher value of that density. Further, no absolute meaning must be attached to the relative situation of the curve l 5 of the experimental values of NJ2 with respect to the theoretical curves. The experimental values have been obtained using a crude approximation based on the secant angle +(n, (T~,,J and on a crude approximation of the mean normal stress along the shearing plane. Therefore the values of NY/2 are to be considered rather as a tool which will enable the calculation of the ultimate bearing from the classical formula than as a purely physical parameter.
COMPARISON OF THEORETICAL AND EXPERIMENTAL VALUES OF N,

A curve has been drawn in Fig. 23 which gives the variation of the theoretical values of N, against the angle of shearing strength. The theoretical values are given by the equation
l&T, =

eTtsn@ tan2

The curve l 5 in Fig. 23 gives the experimental values of N, against the values of the secant angle +(n, 0 =, M). The experimental values are lower than the theoretical ones, except for very high densities where they tend to become larger. In order to take the influence of local shear and punching into account Terzaghi (1943) suggested for low and medium dense sand calculating the ultimate bearing capacity by introducing in the classical formula an arbitrary angle 4 given by tan 4 = $ tan C#J
N, =

(;+g ) .......

. .

. .

. .

. .

* .

(39)

ezntanN3

tan2 [

i+karctan(j!jtan+)]

E.

E.

DE

BEER

%
0

4
0 D

156 17-O 18-5 20.3 22.5 25.2 28.5 :6:; 42.0 48.7 57.0 66.5 78 0 92 0 108.0 GH 41

Fig.

Regression 24. analysis : circular footings, q0 = 0

Fig.

25. Regression analysis: rectangular footings, q,=o,


tj=b/l=1/6

0 ::a 0
::,J

24.1 26.2 28.8 31.9 35.4 39.2 44.5 50.5 57.5 66.5 77.0 90.0 104-o 121.0 143.0 169.0

30 31

15.6 17.0 18.5 :;:: t;.: 32.1 36.8 42.0 40.7 57.0 66.5 78.0 92.0 108.0

24.1 26 2 28 8 31.9 35-4 39.2 44.5 66.5 77.0 90.0 104.0 121.0 143-O 169-O

I-545 l-541 I.557 I.571 I.573 I.556 I.561 I.568 563 I.583 l-581 I-579 I.564 .55 I I.554 I .565

26.0 28.3 30.8 33.8 37.5 42 0 47.5 53-7 61.3 70.0 81.2 95.0 I IO.8 130.0 153.3

180.0

Fig. 26

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF

SAND

405

The values of the bearing capacity factor N, calculated in this way are also given against the angle of shearing strength 4 in Fig. 23. The experimental curve ~5 gradually tends to become the Terzaghi curve.
VARIATION OF s,N,/2 AGAINST IN ANGLES NORMAL OF SHEARING TRIAXIAL STRENGTH 4(n)

DETERMINED

TESTS

Instead of drawing the experimental values of syN,12 against the secant angles $(n, ug, M)r it is possible to draw them against the angles C(S) as determined in conventional triaxial tests. These values are given as a function of the percentage of voids n in Fig. 1. For all circular footings, irrespective of their diameter, the points giving sVN,/2 against 4(n) are shown in Fig. 24. They are divided into two groups, giving the straight lines, from which the curve GH is deduced. The data of the linear regression analysis are given in Table 3. For all rectangular footings (a=&), irrespective of their width, the points giving s,N,/2 against q%(n),are shown in Fig. 25. They are considered as one group, and by the method of regression a straight line is obtained, which is replaced by the curve IJ. The data of the regression analysis are given in Table 3.
Table 3
Data regression analysis s&,/2 Footing = F[4(n)] Limits C(n) Number of tests log s~N,/~ =a+(%) -b Correlation

a Fig. 24 Circular footings 6=1 Rectangular footings s=g 30 25 38 10 38 45 05 z: 0.0445 0.0682

b 0.151 1.038 0.784 0.875

Fig. 25

29 15

45 05

70

0.057

0.373

0.949

Data regression analysis s,N,

= F[$(n)]

Footing

Limits 4(n)

Number of tests

log s,N,

= a+(n) -b

Correlation

a Fig. 27 Circular footings a=1 0 ___~___ 30 24 38 09 37 39 45 06 45 48 37 39 48 37 55 41 05 36 42 45 54 46 ____34 38 :: 0.094 0.151 0.080 0.045 0.039 0.085 0.144 0.048 0.062 0.045

b
1,547 3.721 1.416 0.025 0.029 1.381 3,531 0.331 0.904 0.355 0.942 0.972 0.903 o+Q39 0.945 0.952 0,964 0.854 0.895 0.951

1
-~ 3 ~____ 0 -~~-_-1

39 24
27 27 27 27 29 15 38 IO 28 42 37 39 ~___ 29 06

19 z:

Fig. 28

Rectangular footings s=g

----

---

42 31 67

406

E.

E.

DE

BEER

Curves GH and IJ are shown together curves gives

in Fig. 26.

The ratio of the ordinates

of both

sy = 1-0.4; Here the defined shape factor corresponds to the usual definition, giving the ratio between bearing capacities of two footings with different shape and same width, placed on a sand with

I
0

40=0

go=0

.
0

qo=ltlrn2

q0=3tlm2
Fig. 27

__4fi__
_4(n)

Regression

analysis

MO
16-3 18.4 21.0 25.0 30.2 37.7 47.4 60.7 78.2 IO-l.0 140.0 190.0 270.0 390.0 576 0 13.8 14-9 16.2 17.9 19.8 22.0 24-5 27.5 31.0 35.0 39.9 45.5 52.0 60.2 71.0 85 0 qo= I r/m: !

I I.2 12.1 12.8 14.0 15.0 16-5 18.0 19.9 21.8 24. 26.5 29.8 33.2 37.5 42.6 48-O

I
J-

40=0

lo= 3r/m*

I I -I

go =3rlm2 I 44.

Fig.

28

528

32 36 $() 40e

SHAPE

FACTORS

ASD

BEARING

CAPACITY

FACTORS

OF SAND

407

given relative density. From this expression of s, the values of NY/2 can be deduced. They are given as a function of d(s) by the curve KL. The curve KL is shown in Fig. 22 with the value 4(s) on the abscissa. On the same graph are also given the theoretical curves from other workers, and the curve ECgiving the observed values against the secant angle $(+z,I+, M). The curve KL which gives the experimental values of NY/2 against the conventional triaxial angles d(n) is well above the theoretical values. The observed values are two to four times larger than the theoretical values and this corresponds with the data found by other workers. The explanation of the difference between the observed and experimental values is in the curvature of the intrinsic curve, which plays an important role in case of very small model footings tested without overburden pressure, and also in the influence of the state of strain.
VARIATION OF sqN, AGAINST CONVENTIONAL TRIAXIAL ANGLES @z)

Instead of drawing the values of ssNs against the secant angles +(Pz,Q, M) they can also be drawn against the angles C(N) deduced from conventional triaxial tests. The values s,N, of the tests made by a given initial overburden pressure qO, and with different footings but characterized by the same shape factor 6, are considered as a separate group. For each group a graph is made of the points representing s,N, as a function of $(s). For circular footings (6 = 1) the experimental points are shown for q0= 0, q0= 1 t/m2 and q,,= 3 t/m2 in Fig. 27. For rectangular footings (S=&) the experimental points are shown for q,,=O, q,,= 1 t/m2 and q,,=3 t/m2 in Fig. 28. For each of the graphs corresponding to q. =0 and q,,= 1 t/m2 the experimental points are subdivided into two groups and for each of these groups the linear regression analysis is applied. For each of the graphs corresponding to q,, =3 t/m2 the experimental points are considered as a whole, on which the linear regression analysis is applied. The straight lines obtained are shown in Figs 27 and 28. From these straight lines curves GH and I J are deduced. The data of the linear regression analysis are given in Table 3. All the curves in Figs 27 and 28 are shown together in Fig. 29. Each curve corresponds to a given value of the initial overburden pressure q,, and to a given value of the shape ratio 6.

4a=

I.

4(n)

41 42 43 3 ::o

_ 1
F;;(6= &)(theor.)

:q(6=

I]

__ i

t/m

40= 3, t/m? (s,N& (%&),I, I.384 I.388 I.390 I.414 I.440 I.442 I.444 I.452 I.466 I -477 I.483 I.493 I.506 I.509 I.519 I.527

I.385 I.395 I.406 I-416 I ,426 I.436 I.446 I.456 I ?65 I.475 I ,484 I.493 I 502 I.510 I.519 I.527

i --

I.387 I.397 I-409 I.408 I 424 I.432 I.436 I.466 I.470 I.471 I 500 I.500 I.51 I I.51 I 1.528 -_

I.391 I.396 I.395 I.413 I -424 I.436 I.445 I.455 I.465 I.475 I.486 I.497 I.502 I.51 I I.518 I 527

Fig.

29

408 1000

E.

E.

DE

BEER

__ qa=o
*(n)
40= t/m

I,

40=3,
t/m2

__

N, __K&o
15-l 170 19.3 22.8 27 6 34.3 42.9 55.6 71-2 93.9 127.8 172.1 244 4 350.8 519.3 800-O

N,
K,L
12.8 13.7 14-8 16.4 18.1 20 I 22 0 25.0 28. I 31.7 36. I 40.6 46 8 54. I 63 6 76 0

-- N,
LLI
103 I I.1 I I.7 12.8 139 15.1 16.4 18.0 19-E 21-a 23-9 26 9 30.0 33-7 38.2 42.9

N 1.

theor.

18 4 20 6 23-2 26.1 ::.: 37-a 42.9 48 9 55.9 64.2 73-9 85.4 99 0 II53 134.9

I100

I
IO 5
I 32 I 36 $ Cn) I 40
I

Fig.

30

26-

44

All the curves give the variation of sqNq against the angle 4(n). The solid lines GH correspond to circular footings (6 = 1) and the dashed lines (I J) to rectangular footings (6 = 4). Clearly, for a given shape of footing and a given relative density and thus a given value-of d(n), the value of s,N, depends on the initial overburden pressure q,,. If the ratio of the ordinates of two curves 8 = 1 and 6 = -k corresponding to the same value of q. is calculated, irrespective of the value of qO,the expression sq is given by sq = l+Psin #z) . . . . .

The value sq gives the ratio between the bearing capacities of two footings with same width but different shape placed with the same overburden on sand with the same relative density. However, N, is now a function of qo. Using expression (32), the values of N, given by the curves KL in Fig. 30 are found. The solid lines were obtained directly and the dashed lines were obtained by interpolation. It is seen that for a given relative density or a given angle 4(n) the values of N, decrease when the overburden pressure q. increases. Fig. 30 also shows the theoretical values of N,. For the extreme case of q. = 0 the experimental values of N, are lower than the theoretical ones in the case of low densities, and become much larger than the theoretical values for high densities. On the other hand for very large overburden pressures the experimental values become smaller than the theoretical values, even for very high densities. The general trend of the curves agrees well with the fact that the zone of general shear failure gradually disappears when the relative density decreases and the overburden pressure increases. The curves therefore tend to flatten when q. increases. Further, this also shows that the dependence of N, on q. is to be explained by the fact that for a given density or C(n), when gradually increasing values of q,, are considered, the rupture phenomenon gradually changes from a general failure to a local failure, and finally to a punching failure. It is evident that such different states of failure cannot be covered by a unique value of N,.

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF SAND

409

It must be stressed that the graph of Fig. 30 is not in accordance with the dimensional analysis, as 4(n) and N, are dimensionless quantities, where q. has the dimensions of a stress. Thus q. has to be replaced by a dimensionless ratio. Trials have shown that q. cannot be replaced by qO/ylc, ib as the experimental points are then scattered over the entire area.
WAYS OF CALCULATING THE ULTIMATE BEARING CAPACITY

It is now well known that the intrinsic curve of a sand with a given density is not a straight line but a curve. Further, it is known that the shearing strength characteristics depend on the state of strain. When taking these two influences into account correctly the classical trinome formula has to be abandoned and special tests have to be made to define the shearing strength characteristics in plane strain. This postulates that the results of rather elaborate calculations based on a curved intrinsic law should become available and that the plane strain tests should become routine tests. When this is not the case, approximate calculation methods based on an assumed straight intrinsic law, obtained by conventional test equipment, have to be used. The triaxial apparatus is such a piece of conventional test equipment. There are now two possibilities. The first is to run a large series of tests with the triaxial equipment in order to define the variation of the shearing strength as a function of the relative density, and the mean normal pressure om or the normal stress up in the shearing plane. Thus the curved intrinsic law for triaxial strain for a given relative density is obtained. According to the simplification introduced by Meyerhof (1950) this curved law is replaced by a straight law giving the same values as the curve law for Q M=p,/lO. The second possibility is to run only a series of conventional triaxial tests giving the variation of the conventional angle of shearing strength against the relative density. Both these possibilities give only data about the shearing strength in triaxial strain. In order not to overlook the influence of the state of strain, it therefore becomes necessary to base calculations not on the theoretical values of the bearing capacity factors but on the experimental values, in which the influence of the state of strain is implicit. Against the use of the secant angle $(n, (T=, M ) the objection can be made that it necessitates a large number of triaxial tests. However, compensating for this disadvantage is the fact that the values of N, to be used are independent of the overburden pressure. The second method has the advantage of being based on a smaller number of conventional tests. However, when the angles +(n) are used, it becomes essential to introduce values of N, which depend on the overburden pressure q,,. Further, in certain cases the method can give rise to misleading results. If properly used with the corresponding values of the bearing capacity factors and shape factors, both methods will in normal cases lead to acceptable results.
SHAPE FACTOR s, FOR COHESION TERM

According to the law of the corresponding states of Caquot, the shape factor s, for the cohesion term is given by equation (4). In this expression if the secant angle +(lz, ug,,J is used the experimental values of N, and the values sq as given in equation (23) have to be introduced. This gives % = Ltan$(lz,a,,,) +rN,_1
b

The values of s, against +(n, +,J obtained from the experimental values of N, are given in Fig. 31. The values of s, with the theoretical values of N, can also be determined, giving the dashed lines in Fig. 31. By using expression (33) and the method of 1Hospital it is found that for

410

E.

E.

DE

BEER

SHAPE

FACTORS

AND

BEARING

CAPACITY

FACTORS

OF SAND

411 1951). (commany used,

$=O one obtains sC= 1.2; this is the experimental value found on stiff clays (Skempton, Using the values N, corresponding to the reduced shearing angles after Terzaghi pressible material) gives the full lines and s,= 1.3 for $=O. This is the value found years ago in Delft on soft clays (Polder clays). If instead of the secant angles 4(n, u,,J the conventional triaxial angles 4(n) are equation (32) for s, has to be introduced into equation (4) which gives 0 = &sin +zN,_1 b

C(n)

However, a value for N, which depends on qo has to be introduced into equation (34). This is shown Thus theoretically for each value of qo, another value of s, should be obtained. However, the calculations show that the influence of by the experimental curves in Fig. 32. q. on s, is small and can be neglected. Introducing the theoretical values of N, in equation (34) gives the dashed curves in Fig. 32, Again with the equation (34), and introducing the Terzaghi values gives the solid curves. using the rule of 1Hospital gives s, = 1.2 for 4 = 0, with the theoretical values of N,, and s, = I-3 with the Terzaghi values. It can be concluded that with the expressions given for sq, the experimental values s, = 1.2 for $=O can theoretically be proven; this shows that the proposed formulae do not contradict experimental evidence.
CLOSING REMARK

The Author found it worthwhile to give a description of the small-scale tests performed in Ghent to show that as soon as knowledge concerning the bearing capacity of the cohesionless materials is refined, the secondary parameters (curvature of the intrinsic curve, influence of the state of strain, incomplete development of the shearing surface) appear to complicate the problem. However, these parameters have to be taken into account in interpreting correctly the results of loading tests, especially those of a small scale.

REFERENCES

The bearing capacity of sand, tested by loading circular plates. Froc. 5th Int. Conf. BENT HANSEN (1961). Soil Mech., Paris 1, 659-664. BRINCH, HANSEN J. (1961). A general formula for bearing capacity. Bull. geotek. Inst. No. 11. BRINCH HANSEN, J. (1963). Discussion on hyperbolic stress-strain response : cohesive soils. J. Soil Me&. Fdns Div. Am. Sm. civ. Engrs 89, SM 4, 242. BUISMAN, K. (1940). Grondmechanica. Delft. CHRISTIAENS(1966), Private communication. Etude experimentale de la capacite portante du sable sous des fondaDE BEER, E. E. & LADANYI, B. (1961). tions circulaires Btablies en surface. C. Y. 5 Congr. Int. Mec. Sols Trav. Fond., Paris 1. DE BEER, E. E. & VESIC, A. B. (1958). Etude experimentale de la capacite portante du sable sous des fondations directes etablies en surface. An& Trav. Publ., No. 3. KERISEL, J. & CAQUOTA. (1956). Traits de mecanique des s&s. Paris: Gauthier-Villars. LADANYI, B. (1960). Etude des relations entre les contraintes et les deformations lors du cisaillement des sols pulverulents. Annls Trav. publ. Be&., No. 3. LEUSSINK, H., BLINDE, A. & ABEL, P. G. (1966). Versuche tiber die Sohldruckverteilung unter starren Vertifl. Inst. Boden-Mech. Felsmechanik Technischen Grtindungskorpern auf kohasionslosem Sand. Hochsch. Fredericiana, Karlsruhe 22. LUNDGREN, H. & MORTENSEN,K. (1953). Determination by the theory of plasticity of the bearing capacity of continuous footings on sand. Proc. 3rd Int. Conf. Soil Mech., Ziirich 1. Ph.D. thesis, University of London. MEYERHOF, G. G. (1950). The bearing capacity of sand. MEYERHOF, G. G, (1955). Influence of roughness of base and ground-water conditions on the ultimate bearing capacity of foundations. Gdotechnique 5, No. 3, 227-242. SKEMPTON,A. W. (1951). The bearing capacity of clays. Proc. BZdgs Res. Congr., London. TERZAGHI, K. (1943). Theoretical soil mechanics. New York: Wiley. Soil Mechanics Laboratory Report. Bearing capacity of deep foundations in sand. VESIc, A. B. (1963). Georgia Institute of Technology.

You might also like