You are on page 1of 9

Geotech Geol Eng (2007) 25:499–507

DOI 10.1007/s10706-007-9124-9

ORIGINAL PAPER

Upper bound limit analysis for finding interference effect of


two nearby strip footings on sand
Jyant Kumar Æ Priyanka Ghosh

Received: 3 February 2006 / Accepted: 7 April 2007 / Published online: 11 May 2007
Ó Springer Science+Business Media B.V. 2007

Abstract The ultimate bearing capacity of two


closely spaced strip footings, placed on a cohesion- Keywords Bearing capacity  Failure  Flow rule 
less medium and loaded simultaneously to failure at Foundations  Plasticity
the same magnitude of failure load, was determined
by using an upper bound limit analysis. A logarithmic
spiral radial shear zone, comprising of a number of 1 Introduction
triangular rigid blocks, was assumed to exist around
each footing edge. The equations of the logarithmic In many cases foundations encountered in practice
spiral arcs were based on angles /L and /R rather are not isolated and these foundations interfere with
than soil friction angle /; the values of /L and /R each other on account of their close spacing. The
were gradually varied in between 0 and /. The interference effect of two strip footings on their
ultimate bearing capacity was found to become ultimate bearing capacity was studied theoretically by
maximum corresponding to a certain critical spacing Stuart (1962) on the basis of the limit equilibrium
between the footings. For spacing greater than the method. For the problem of three closely spaced strip
critical, the bearing capacity was found to decrease footings on sand, Graham et al. (1984) have
continuously with increase in the spacing. The extent employed the method of characteristics so as to
of the spacing corresponding to which the ultimate obtain the solution for determining the interference
bearing capacity becomes either maximum or equal effect on the central footing (not the two outer
to that of a single isolated footing increases with footings). This theory was meant for obtaining the
increase in /. The results compare reasonably well interference effect due to a number of strip footings,
with the available theoretical and experimental data. for instance the foundations of railway sleepers. In
order to study the effect of the interference of the two
footings a number of experiments have also been
conducted by various researchers (Stuart 1962; West
J. Kumar (&)
and Stuart 1965; Saran and Agarwal 1974; Das and
Civil Engineering Department, Indian Institute of Science,
Bangalore 560012 Karnataka, India Larbi-Cherif 1983; Kumar and Saran 2003). From
e-mail: jkumar@civil.iisc.ernet.in available experimental and theoretical studies it is
understood that the ultimate bearing capacity, due to
P. Ghosh
the unit weight component, increases on account of
Department of Civil Engineering, Indian Institute of
Technology, Kharagpur 721 302, India the interference of the other footing. The aim of the
e-mail: priyog@iitkgp.ac.in present study was to examine further, with the help of

123
500 Geotech Geol Eng (2007) 25:499–507

an upper bound limit analysis, the effect of the Michalowski 1997; Soubra 1999; Zhu 2000; Kumar
variation of the spacing between two rough surface 2004). On account of the symmetry about the centre
footings on the bearing capacity. It is known that an line OEQ between the two footings, only one footing
upper bound limit analysis ensures the kinematic on the right side of OEQ was chosen for carrying out
admissibility of the chosen collapse mechanism. An the analysis. The boundary of each radial shear zone,
upper bound analysis based upon finite elements, in comprising of a number of adjoining rigid triangular
conjunction with optimization, provides a rigorous blocks, was assumed to be bounded by an arc of the
numerical solution and this approach does not require logarithmic spiral. The equations for the logarithmic
any assumption regarding the geometry of the spiral arcs were based upon on angles /L and /R; the
collapse mechanism (Anderheggen and Knöpfel values of /L and /R were varied from 0 to /. It
1972; Sloan 1989). However, such an exercise is should be mentioned that in Fig. 1, FL E ¼
quite cumbersome and it often requires considerable FL CehL tan /L , and FR R ¼ FR CehR tan /R ; where hL and
computational time and memory especially when the hR are the values of angles \EFLC and \RFRC,
size of the domain is large and the soil friction angle respectively. The positions of the focus of the
becomes high. The upper bound analysis in this study logarithmic spiral arcs, CE and CR, were varied
was carried out by assuming the geometry of the along the lines FLBE and FRAR, respectively. The arc
collapse mechanism. The work is an extension of the CE of the logarithmic spiral was joined tangentially
recent analysis of the first author (Kumar 2004) for an to the boundary of the triangular symmetrical rigid
isolated footing. The boundary of the radial shear block in between the two footings; therefore, the
zone emerging from each edge of the footing was value of the angle \OEB becomes equal to p/4 + /L/
assumed to be bounded by an arc of the logarithmic 2. Likewise, the logarithmic spiral arc CR was joined
spiral; the equations of the logarithmic spiral arcs tangentially to the line RG; consequently, the angle
were, however, based on angles /L and /R rather \ARG becomes equal to p/4 + /R/2. The chosen
than / and the values of /L and /R were varied in collapse mechanism can be completely defined with
between 0 and /. In addition, the focus of the the help of six independent input variables, namely,
logarithmic spiral arc was also varied rather than hL, lL, /L, lR, /R and gR; where gR = \RAG, lL and
keeping fixed at the edge of the footing. The results lR are the distances FLB and FRA, respectively. When
obtained from the analysis were compared with the the spacing between the footings becomes very small,
available experimental and theoretical data. there will be a possibility that the logarithmic arc CR
of the rupture surface may intersect the vertical centre
line OEQ; a number of collapse mechanisms in such
2 Definition of the problem cases will become inadmissible. As a result, the
magnitude of the failure load will become higher. If
Two rough strip footings, each of width B, are placed the magnitude of the total collapse load due to the
on a cohesionless medium at a clear distance s as two interfering footings becomes greater than that of
shown in Fig. 1. The magnitude of the ultimate a single isolated footing of width (2B + s), the
collapse load Pu (per unit length) for each footing has collapse mechanism associated with the single iso-
to be computed. Both the footings are assumed to fail lated footing having width (2B + s) was chosen to be
simultaneously at the same magnitude of the failure applicable.
load.

4 Velocity hodographs
3 Collapse mechanism
At collapse, it was assumed that the footing and the
A non-symmetrical triangular trapped wedge ABC, underlying block ABC move as a single rigid unit in
as shown in Fig. 1, was considered to be a part of the the vertical direction with a velocity V0. The zones
footing itself; it is known that for a perfectly rough BCE and ACR were divided into n and m number of
footing such a non-plastic trapped wedge forms triangular blocks, respectively; the value of both n
below the footing base (Terzaghi 1943; Chen 1975; and m was kept equal to 700. For the purpose of

123
Geotech Geol Eng (2007) 25:499–507 501

s/ 2 B
Pu
FL FR
lL θL θR lR
O B A G
ηR
U4 U43 V0 φ
φ U10 φ φ V43
(π/4+φL/2) U32 V10 φ V4 φ
U21 V21
φ φ
φ
V32 V4 3
E U1 C (π/2+φR) V4
U3
φ
U2 V1
φ V3
E1 φ
E2 φ V3 R
R1 V2 φ
φ R2
V3
Q V3 2
Footing is defined by line AB
V2
V2
U4 3 V1 V2 1
U3 2 U1 V1
U3 U2
U2 1 V0 V0
U4 U1 U1 0 V1 0
U2
Pu Pu

U3
B s B

Fig. 1 Velocity hodographs and the chosen collapse mechanism

illustrations, however, only three such blocks and the U10, U21, U32 and U43 are also inclined at an angle /
associated velocity hodograph triangles are illustrated with the lines AC, AR1, AR2, AR, BC, BE1, BE2 and
in Fig. 1. In this figure, V1, V2, V3 and V4 are the BE, respectively. It should be mentioned that the
respective absolute velocities of the blocks ACR1, central block (BOE) in between the two footings, due
AR1R2, AR2R and ARG. Similarly, U1, U2, U3 and to symmetry, was permitted to move only in the
U4 are the respective absolute velocities of the blocks vertically upward direction with a velocity U4. The
BCE1, BE1E2, BE2E and BOE. V10 is the velocity of directions of the velocities V0, V1, V2, V3, V4, U1, U2,
the block ACR1 relative to the block ABC; V21 is the U3, U4, V10, V21, V32, V43, U10, U21, U32 and U43 are
velocity of the block AR1R2 relative to the block all fixed. Therefore, by using the velocity hodograph
ACR1; V32 is the velocity of the block AR2R relative triangles, the values of the velocities V1, V2, V3, V4,
to the block AR1R2; V43 is the velocity of the block U1, U2, U3, U4, V10, V21, V32, V43, U10, U21, U32 and
ARG relative to the block AR2R. Likewise, U10 is the U43 can all be computed in terms of V0.
relative velocity of the block BCE1 with respect to It should be mentioned that the velocity hodog-
the block ABC; U21 is the relative velocity of the raph triangles were based on the consideration of
block BE1E2 with respect to the block BCE1; U32 is velocity discontinuities along the interfaces of
the relative velocity of the block BE2E with respect to various rigid blocks not along the radial lines
the block BE1E2; and U43 is the relative velocity of joining the focus (FL and FR) to any point on the
the block BOE with respect to the block BE2E. The logarithmic spiral arc.
interfaces of all the triangular blocks were treated as
velocity discontinuity lines. The soil mass was
assumed to obey Mohr–Coulomb failure criterion 5 Analysis
and an associated flow rule, the directions of veloc-
ities V1, V2, V3, V4, U1, U2, and U3 make an angle / Since the soil mass was considered to be cohesion-
with the corresponding rupture lines. Likewise, the less, the rate of the total work done by all the body
directions of the relative velocities V10, V21, V32, V43, and external forces should become equal to zero. The

123
502 Geotech Geol Eng (2007) 25:499–507

collapse load Puc per unit length of the footing due to factor Nc was defined by using the following standard
the component of soil unit weight for a single footing expression:
can be expressed by using the following expression.

  m 
P
n P
WiL Ui;v þ L
Wnþ1 Unþ1  W 0 V0 þ WiR Vi;v þWmþ1
R
Vmþ1;v
i¼1 i¼1
Puc ¼ ð1Þ
V0

where W0 is the weight of the block ABC; WiL is the Puc =B


Nc ¼ ð2Þ
weight of the ith block in the radial shear zone BCE; 0:5cB
WiR is the weight of the ith block in the radial shear
L
zone ACR; Wnþ1 is the weight of the block BOE; It may be noted that Puc in the above equation is
R
Wmþ1 is the weight of the block ARG. Also, Ui,v is the magnitude of the failure load per unit length (due
the vertical component of the velocity of the ith block to soil unit weight) for a single isolated footing of
in the zone BCE (upward positive); Vi,v is the vertical width B.
component of the velocity of the i-th block in the
zone ACR (upward positive); Un+1 is the velocity of 6.2 Efficiency factor nc
the block BOE (upward positive); Vm+1,v is the
vertical component of the velocity of the block ARG The efficiency factor nc is the ratio of the failure load
(upward positive). Therefore, for a chosen collapse for a single footing of width B in the presence of the
mechanism by using Eq. 1 the magnitude of Puc can other footing to the corresponding magnitude of the
be determined. Subsequently, the magnitude of Puc failure load for a single isolated footing of the same
can be minimized with respect to independent width B.
variation of six variables namely, hL, lL, /L, lR, /R
and gR. It was always assured that the collapse
mechanism remains always kinematically admissible. 7 Results and comparisons
A mechanism is kinematically admissible if the
magnitudes of all the velocities V1, V2,..., Vm+1, U1, For doing the optimization (i) hL and gR were varied
U2,..., Un+1, V10, V21,..., Vm+1,m, U10, U21,..., Un+1,n in from 0 to 908 with an interval of 18, (ii) /L and /R
terms of V0 remain always positive; the directions of were varied from 0 to / with an interval of 18, (iii)
the velocities V1, V2,..., Vm+1, U1, U2,..., Un+1, V10, distances lL and lR were varied with an interval of
V21,..., Vm+1,m, U10, U21,..., Un+1,n are indicated in 0.05B. The results were obtained for different spacing
Fig. 1 for n = m = 3. between the two footings by changing the varying /
in between 0 and 508. The results are presented
below.

6 Definitions of bearing capacity factor Nc and 7.1 Nc for an isolated rough strip footing
efficiency factor nc
The obtained variation of Nc for a single isolated
6.1 Bearing capacity factor Nc for an isolated rough strip footing is given in Table 1. The values of
footing Nc were compared with the upper bound limit
analysis solutions of Chen (1975), Michalowski
For an isolated footing placed on ground surface (1997), Soubra (1999), Zhu (2000) and Kumar
without any surcharge pressure, the bearing capacity (2004). The comparison of all these results is

123
Geotech Geol Eng (2007) 25:499–507 503

Table 1 A comparison of Nc values for an isolated rough strip footing using the upper bound limit analysis
/ Present analysis Chen (1975) Michalowski (1997) multi- Soubra (1999) multi- Zhu (2000) multi-blocks
and Kumar logarithmic spiral blocks symmetrical blocks symmetrical symmetrical mechanism
(2004) mechanism mechanism mechanism

108 0.74 1.16 0.71 – 0.71


208 4.47 5.87 4.47 4.49 4.47
308 21.42 26.7 21.39 21.51 21.38
408 119.14 147.0 118.83 119.84 118.75
508 1030.12 – 1025.98 1042.48 1025.06

presented in Table 1. It should be mentioned that the 6

analysis of Chen (1975) was based on the logarithmic


5 ξγ = ( 2 + s / B ) / 2
2

spiral rupture surface with its focus at the footing

Efficiency factor, ξγ
edge; and the equation of the spiral arc was based on φ= 5 0o
4 45 o
soil friction angle /. On the other hand, the analyses 40 o
35 o
of Michalowski (1997), Soubra (1999) and Zhu 30 o
3 25o
(2000) were all based on multi-blocks collapse 20 o
15 o
mechanism, whereas the collapse mechanism of 10o
2
Kumar (2004) considers the employment of a loga-
rithmic spiral rupture surface. The equation of the 1
spiral arc in this analysis was based on an angle /L 0 2 4 6 8 10 12
s/B
which varies in between 0 and /, and the focus of the
logarithmic spiral was varied rather than keeping Fig. 2 The variation of nc with s/B and /
fixed at the edge of the footing. It should be noted
that the present Nc values are exactly the same as
given by Kumar (2004) and compare very favourably especially for values of / greater than 208. The
with the existing results of Michalowski (1997), variation of smax/B and scr/B with / is shown in
Soubra (1999), and Zhu (2000). The Nc values of Fig. 3. The values of both smax/B and scr/B increase
Chen (1975) have been generally found to be the with increase in /.
highest.

7.2 The variation of efficiency factor (nc) 7.3 Comparison of the efficiency factor

The variation of nc with respect to changes in the For / = 308 and 408, the nc values of Stuart (1962)
values of s/B and / is shown in Fig. 2. The magnitude were compared with the present solution. The
of nc becomes equal to 2.0 and 1.0 for s/B = 0 and s/ comparison of the two theories is shown in Fig. 4a.
B = smax/B, respectively; where smax is the value of s For s/B > scr/B, the values of nc obtained in
beyond which the two footings can be assumed the present analysis were found to be smaller
isolated. It can also be noted that starting from s/ than those provided by Stuart (1962). For a given
B = smax/B, the value of nc increases continuously value of /, the values of scr/B of Stuart (1962)
with decrease in s/B up to a value of s/B = scr/B. were seen to be a little higher. The values of smax/
Corresponding to s/B = scr/B, the value of nc becomes B from both the analyses compare well with each
maximum. For values of s/B < scr/B, the magnitude of other.
nc decreases with decrease in s/B; the expression of The values of nc were also compared with the
the efficiency factor between s/B = 0 and s/B = scr/B experimental data of Saran and Agarwal (1974),
is defined by the expression, nc = (2 + s/B)2/2. The Das and Larbi-Cherif (1983) and Kumar and Saran
maximum value of nc was found to be greater than 2 (2003). The experimental results are for values of /

123
504 Geotech Geol Eng (2007) 25:499–507

in between 378 and 418. The theoretical results were 7.4 Failure pattern and critical collapse
obtained for / = 408. The comparison of all these mechanisms
results is shown in Fig. 4b. It can be seen that the
results from the present approach compare reasonably By varying the spacing between the two footings, the
well with the experimental data. However, the peak failure patterns were drawn for / = 308 and 408. The
values of nc, associated with scr/B, predicted from the value of s/B was varied from 0.5 to 1.5 for / = 308. On
present theory were found to be generally greater the other hand, for / = 408, the value of s/B was varied
than the experimental data. in between 1.5 and 3.5. The failure patterns have been
illustrated in Figs. 5 and 6; it should be mentioned that
in drawing the failure patterns, the horizontal and the
10 vertical scales were kept always the same. It can be
noted that the size of the radial shear zone BCE
8
around the interior edge of the footing, as compared to
the size of the radial shear zone ACR around the
smax/B, scr/B

6
exterior footing edge, decreases continuously with
sma x/B
4
decrease in the value of s/B.
For / varying from 108 to 508, the values of the
2 input parameters hL, lL, /L, lR, /R and gR associated
scr/B with the critical collapse mechanisms are also
0 presented in Table 2. The value of s/B lies in
10 15 20 25 30 35 40 45 50
between (i) 0.10 and 0.25 for / = 108, and (ii) 3.0 and
φ
7.0 for / = 508. It may be noted that the values of
Fig. 3 The variation of smax/B and scr/B with / /L and /R remain generally different from /.
The critical values of /L and /R lie in the range of

Fig. 4 A comparison of (a) 4


obtained nc with (a) Stuart
(1962)’s theory and (b) φ = 40o
Stuart (1962)
Efficiency factor, ξγ

experimental data 3 Present Analysis

2
φ = 30o

0
0 1 2 3 4 5 6
s/B
(b) 4
Present analysis, = 40o
Das and Larbi-Cherif (1983), 38
Efficiency factor, ξγ

3 Saran and Agarwal (1974), 41


Kumar and Saran (2003), 37
Present study,
2

0
0 1 2 3 4 5 6 7 8
s/B

123
Geotech Geol Eng (2007) 25:499–507 505

Fig. 5 The variation of the φ = 30o


failure pattern with spacing
for / = 308 O
B A G

C
E

s/B = 0.5
φR = 23o, φL = 0o, lR/B = 0.155, lL/B = 0.005, θL = 89o, ηR = 37o

O
B A G

C
E
R

s/B = 1.0
φR = 24o, φL = 10o, lR/B = 0.155, lL/B = 0.055, θL = 89o, ηR = 36o

O
B A G

C
E
R
Q

s/B = 1.5
φR = 25o, φL = 20o, lR/B = 0.155, lL/B = 0.155, θL = 89o, ηR = 35o

0.0–0.9/ and 0.8/–/, respectively. The values of the 9 Conclusions


lL/B and lR/B were found to lie in between 0.005 and
0.355. The critical values of hL were found to be By using the upper bound limit analysis, the variation
close to 908 for most of the cases. However, no clear of the efficiency factor nc with change in the spacing
trend was observed for the variation of gR with between two closely spaced rough strip footings was
respect to the change in s/B. obtained. It was noticed that the magnitude of nc
becomes maximum corresponding to s/B = scr/B. For
s/B > scr/B, the value of nc decreases with increase in
8 Remarks the spacing between the two footings. When the
spacing between the two footings becomes equal to
The extent of dilation experienced during the shear zero, the value of nc becomes exactly equal to 2.0.
failure for most of the cohesionless soils is generally The size of the radial shear zone at the interior
greater than that predicted on the assumption of the footing edge, in relation to the size of the radial shear
associated flow rule. Since non-associated flow rule at the exterior edge of the footing, reduces continu-
generally results in smaller values of failure load ously with the decrease in the spacing. The values of
(Frydman and Burd 1997), the collapse load pre- nc were found to be generally lower as compared to
dicted on the basis of the present upper bound the analysis of Stuart and match reasonably well with
analysis will be a little greater than the true failure the available experimental data. The peak values of
load. nc associated with scr/B for a given value of / were

123
506 Geotech Geol Eng (2007) 25:499–507

Fig. 6 The variation of the φ = 40o


failure pattern with spacing
for / = 408 O B A G

C
E

s/B = 1.5
φR = 34o, φL = 17o, lR/B = 0.105, lL/B = 0.005, θL = 89o, ηR = 31o

O B A G

C
E
R

s/B = 2.5
φR = 34o, φL = 26o, lR/B = 0.105, lL/B = 0.055, θL = 89o, ηR = 31o

O B A G

C
E R

s/B = 3.5
φR = 34o, φL = 32o, lR/B = 0.105, lL/B = 0.105, θL = 89o, ηR = 31o

Table 2 The values of /L, /R, lL/B, lR/B, hL, gR and nc associated with the critical collapse mechanism
/ s/B /L /R lL/B lR/B hL gR nc

108 0.10 0.0 10.0 0.205 0.255 89.0 9.0 1.473


0.25 3.0 10.0 0.255 0.205 89.0 14.0 1.189
208 0.25 0.0 20.0 0.105 0.305 89.0 26.0 1.358
0.50 5.0 20.0 0.155 0.255 89.0 28.0 1.114
308 0.25 4.0 25.0 0.055 0.205 81.0 35.0 2.042
1.00 10.0 24.0 0.055 0.155 89.0 36.0 1.152
1.50 20.0 25.0 0.155 0.155 89.0 35.0 1.035
408 1.50 17.0 34.0 0.005 0.105 89.0 31.0 1.409
2.50 26.0 34.0 0.055 0.105 89.0 31.0 1.098
3.50 32.0 34.0 0.105 0.105 89.0 31.0 1.014
508 3.00 31.0 44.0 0.005 0.055 86.0 26.0 1.690
5.00 36.0 44.0 0.005 0.055 89.0 26.0 1.172
7.00 41.0 44.0 0.055 0.055 89.0 26.0 1.034

123
Geotech Geol Eng (2007) 25:499–507 507

seen to be generally greater than the experimental Michalowski RL (1997) An estimate of the influence of soil
data. weight on bearing capacity using limit analysis. Soils
Found 37(4):57–64
Saran S, Agarwal VC (1974) Interference of surface footings
on sand. Indian Geotechnical J 4(2):129–139
References Sloan SW (1989) Upper bound limit analysis using finite ele-
ments and linear programming. Int J Numer Anal Meth-
Anderheggen E, Knöpfel H (1972) Finite element limit anal- ods Geomechanics 13:263–282
ysis using linear programming. Int J Solid Struct 8: Soubra AH (1999) Upper-bound solutions for bearing capacity
1413–1431 of foundations. J Geotechnical Geoenviron Eng ASCE
Chen WF (1975) Limit analysis and soil plasticity. Elsevier, 125(1):59–68
Amsterdam Stuart JG (1962) Interference between foundations, with spe-
Das BM, Larbi-Cherif S (1983) Bearing capacity of two clo- cial reference to surface footings in sand. Géotechnique
sely-spaced shallow foundations on sand. Soils Found 12(1):15–22
23(1):1–7 Terzaghi K (1943) Theoretical soil mechanics. Wiley, New
Frydman S, Burd HJ (1997) Numerical studies of bearing York
capacity factor Nc. J Geotechnical Geoenviron Eng ASCE West JM, Stuart JG (1965) Oblique loading resulting from
123(1):20–29 interference between surface footings on sand. Proceed-
Graham J, Raymond GP, Suppiah A (1984) Bearing capacity of ings of the 6th international conference on soil mechanics
three closely-spaced footings on sand. Géotechnique and foundation engineering, Montreal, vol. 2, pp. 214–217
34(2):173–182 Zhu D (2000) The least upper-bound solutions for bearing
Kumar J (2004) Effect of footing-soil interface friction on capacity factor Nc. Soils Found 40(1):123–129
bearing capacity factor Nc. Géotechnique 54(10):677–680
Kumar A, Saran S (2003) Closely spaced footings on geogrid-
reinforced sand. J Geotechnical Geoenviron Eng ASCE
129(7):660–664

123

You might also like