See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/233564083

Analytical active earth thrust on cantilever walls

with short heel

Article in Canadian Geotechnical Journal · December 2008

DOI: 10.1139/T08-078

CITATION READS

1 72

1 author:

Venanzio Greco
Università della Calabria
27 PUBLICATIONS 269 CITATIONS

SEE PROFILE

All content following this page was uploaded by Venanzio Greco on 06 February 2016.

The user has requested enhancement of the downloaded file.

 Pagination not final/Pagination non finale
1

Analytical active earth thrust on cantilever walls

with short heel
Venanzio R. Greco

Abstract: The use of Rankine’s method is inappropriate for calculating active thrusts in cantilever retaining walls with a
short heel because the thrust wedge is interrupted by the wall backface. The use of Coulomb’s approach is preferable, but
at present only numerical solutions have been proposed to solve the problem. This paper presents an analytical solution,
based on Coulomb’s approach, for evaluating the active thrust on cantilever walls with a short heel subjected to homoge-
neous backfill with a regular topographic profile and without pore pressure. The solution is given by an algorithm where
two equations, one quadratic and the other cubic, are solved in turn, in an iterative procedure that converges rapidly. The
distribution of lateral pressure and the position of the point of application of the thrust are also given in analytical terms.
Key words: active earth force, cantilever walls, lateral pressure, active condition.
Résumé : Dans le cas des murs de soutènement cantilever avec un court pied arrière, l’utilisation de la méthode de
Rankine pour le calcul des poussées est inadéquate parce que le coin de pression est coupé par la face arrière du mur.
L’utilisation de la méthode de Coulomb est préférable, mais jusqu’à maintenant, on a proposé seulement des solutions nu-
mériques pour résoudre le problème. Cet article présente une solution analytique basée sur l’approche de Coulomb pour
évaluer la poussée sur les murs cantilever avec un pied court, retenant un remblai homogène avec un profil topographique
régulier et sans pression interstitielle. La solution est donnée par un algorithme où on résout à tour de rôle deux équations,
une quadratique et l’autre cubique, dans une procédure itérative qui converge rapidement. La distribution de la pression la-
térale et la position du point d’application de la pression sont également données en termes analytiques.
Mots-clés : force de poussée des terres, murs cantilever, pression latérale, condition de poussée.
[Traduit par la Rédaction]

Introduction thrust with respect to the inclination angles of the planes

limiting the thrust wedge. The latter condition is equivalent
Thrusts on cantilever retaining walls are generally calcu- to that of minimizing the safety factor against wall sliding
lated using Rankine’s method (Rankine 1857). This method along the base. In comparison to Rankine’s method, that of
is based on the assumption that, due to the wall movement Coulomb, apart from taking into account the friction be-
and the consequent reduction in lateral pressure on the back- tween the soil and wall, has a larger range of applicability
fill, a part of the backfill behind the wall becomes plastic. when considering load and geometrical conditions. Although
Rankine’s analysis proves that the soil inside this plastic the two methods differ markedly in their formulation (one is
zone fails along two sets of planes inclined at angles  and a limit equilibrium method and the other is a stress method),
 with respect to the horizontal (Fig. 1). if they are used with the same input data then they give ex-
However, if the heel is not sufficiently long, the slip plane actly the same results (Greco 1999). The differences that
inclined at  starting from the heel intersects the wall sometimes arise are derived from differences in input data:
(Fig. 2a), and the use of Rankine’s method is no longer ap- in Rankine’s method, the thrust inclination is a result of the
propriate. This is because the stress on the plane DB is af- analysis, while in Coulomb’s method it is a datum of the
fected by the friction between the wall and the soil, which problem.
is not taken into consideration in Rankine’s method. In walls with a long heel, Rankine’s method is preferable
In light of this shortcoming, Barghouthi (1990) searched because it requires fewer computations. In contrast, the
for a solution by applying Coulomb’s method (Coulomb thrusts on cantilever retaining walls with a short heel should
1773) to the BDCA wedge (Fig. 2a) by maximizing the be correctly analyzed with Coulomb’s procedure, as formu-
thrust Sa with respect to angles  and  (Fig. 2c). Cou- lated by Barghouthi (1990).
lomb’s method is, in fact, an alternative procedure to that However, re-examining the method of Barghouthi, Greco
of Rankine to calculate thrusts on retaining walls. It is very (1992) pointed out that, because the inclination of Sa is not a
easy in its formulation, being based on the equilibrium of constant (as in the original method of Coulomb) but varies
forces at limit equilibrium and on the maximization of the with , the maximization of the thrust Sa is not sufficient to
give the critical condition for wall stability. In fact, a lesser
Received 24 November 2006. Accepted 7 July 2008. Published thrust could have a higher horizontal component and thus
on the NRC Research Press Web site at cgj.nrc.ca on . prove to be more critical for the wall stability. From an ex-
V.R. Greco. University of Calabria, Department of Structural amination of the numerical results, it emerged that the hori-
Engineering, 87036 Rende (Cs), Italy (e-mail: zontal component (Sah) of the thrust is maximized by the
[email protected]). same value of  that is minimizing the vertical component

Can. Geotech. J. 45: 1–10 (2008) doi:10.1139/T08-078 # 2008 NRC Canada

PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
2 Can. Geotech. J. Vol. 45, 2008

Fig. 1. Thrust wedge acting on a cantilever wall with a long heel ECA, whose weights are W’ and W@, respectively (Fig. 2a),
and sets of failure planes. with
1
½1 W 0 ¼ bðh þ H  yÞ
2

1
½2 W 00 ¼ Hx
2
where (Fig. 2a)
cos " cos  cos " cos 
½3 x ¼ ðh þ b tan "Þ ¼H
sinð  "Þ sinð  "Þ
The active thrust S1 can be obtained through the equili-
brium conditions of forces acting on the wedge BDF
(Fig. 2b),
1 cos  cos " sinð  0 Þ
½4 S1 ¼ ðh  yÞ2
2 sinð  "Þ cosð  0  Þ
where
½5 y ¼ b tan 

Analogously, the active thrust S2 is calculated through the

equilibrium conditions of forces acting on the wedge BDCA.
(Sav), therefore the critical condition for the wall stability
was stated by maximizing Sah (or by minimizing Sav) with ðW 0 þ W 00 Þsinð  0 Þ S1 cosð  0  Þ
½6 S2 ¼ 
respect to the unknown angle . This also corresponds to sinð þ 20  Þ sinð þ 20  Þ
minimizing the wall safety factor against sliding along the
base. Finally the thrust components Sah and Sav are obtained
through the equilibrium conditions, in horizontal and verti-
Successively, this was also proven by a theoretical analy-
cal directions, respectively, of forces acting on the wedge
sis (Greco 2001) in a more general study that included the
BDCE
presence of water tables, lines or strips of surcharge, and ir-
regular topographic profiles. The solution of the problem ½7 Sah ¼ S1 cos þ S2 sinð þ 0 Þ
was, however, obtained by means of a numerical procedure.
This paper presents the analytical solution of the active
thrust on retaining walls with a short heel and backfill of ½8 Sav ¼ S1 sin  S2 cosð þ 0 Þ  W 0
regular topographic profile not subject to surcharges and
pore pressure. The position of the point of application of The thrust components Sah and Sav are thus functions of
the thrust is also given by an analytical solution. the angles , , and , and they should be differentiated
with respect to these angles to obtain the critical condition
for the wall. However, it was shown that (Greco 2001):
Coulomb-type analysis
(1) The maximum of S1 is attained for the same value of 
A cantilever wall with a short heel is represented in that maximizes Sah and simultaneously minimizes Sav.
Fig. 2a. The thrust wedge BDCA behind the wall is bounded Moreover, maximizing S1 with respect to  also corre-
by the plane CA inclined at  with respect to the horizontal, sponds to minimizing the safety factor against sliding Fs
by the plane CD inclined at , by the topographic profile with respect to the same angle.
and the wall back face, which is assumed to be vertical.
The soil is assumed to be homogeneous, not subject to pore (2) The components Sah and Sav vary proportionally as the
pressure and, moreover, cohesionless with friction angle ’ angle  varies. Therefore, maximizing S2 corresponds to
and unit weight . maximizing Sah and Sav with respect to . Maximizing S2
produces, moreover, a minimization of Fs.
The thrust wedge BDCA is subjected to the following
(3) The critical condition for a wall against sliding along the
forces: W, the self weight of the wedge BDCA; S1 the thrust
base is obtained for the value of  that maximizes Sah
acting on BD; S2 the thrust on DC; and R, the thrust on CA.
and simultaneously minimizes Sav.
The force W is vertical and downwards; the thrusts S2, R,
and S1 are directed towards the inside of the wedge; the first Point (1) permits the thrust S1 to be put in the form
two are inclined at ’ with respect to the normals to the 1
planes on which they act; the third is inclined at , this ½9 S1 ¼ ðh  yÞ2 Ka1
being the friction angle between soil and wall along BD. 2
Drawing the vertical plane CE crossing the heel C, the where Ka1 is the coefficient of active thrust (Müller-Breslau
wedge BDCA is divided into two subwedges, BDCE and 1906)
# 2008 NRC Canada
PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
Greco 3

Fig. 2. Walls with short heel. (a) Forces acting on the wedge ABDC. (b) Forces acting on the wedge BDF. (c) Forces acting on the wedge
BDCE.

" pffiffiffiffiffiffiffiffiffiffi #2
cos 0 cos " ½15 a2 ¼ Ka1 ½bcosð0  Þ  2 hsinð0  Þ  b cos0
½10 Ka1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cos cos " þ sinð0 þ Þsinð0  "Þ
½16 a1 ¼ Ka1 h½hsinð0  Þ  2 bcosð0  Þ
while the angle  maximizing S1 is þ b2 sin0 þ bðh þ HÞcos0
2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
ffi 3
cos " sin ð þÞ 0
1 4 cos  sin ð0 "Þ  sin ½17 a0 ¼ Ka1 bh2 cosð0  Þ  b2 ðh þ HÞsin0
½11 0
 ¼  þ tan 0
tanð  "Þ 5
sin0 þ coscos "
ð0 "Þ tan
½18 b1 ¼ Hcos 0
Points (2) and (3) state that the minimization of Fs is
achieved by differentiation of Sah (or Sav) with respect to ½19 b0 ¼ bHsin 0
the angles  and . However, following the approach of Sil-
vestri (2006), the distances x and y are used instead of the
angles  and , because this leads to less complicated equa- ½20 d1 ¼ tan "cos 0  sin 0
tions.
By substituting eqs. [1]–[6] and [9] and [10] in eqs. [7] ½21 d0 ¼ Hcos 0
and [8], we have
½22 e1 ¼ tan "sin 0 þ cos 0
1 a3 y3 þ a2 y2 þ a1 y þ a0 þ ðb1 y þ b0 Þx
½12 Sah ¼ 
2 ðmy þ nÞx þ py þ q
½23 e0 ¼ Hsin 0
 ðd1 x þ d0 Þ
½24 m ¼ cos 20 þ sin 20 tan "
1
½13 Sav ¼ Hx
2
1 a3 y3 þ a2 y2 þ a1 y þ a0 þ ðb1 y þ b0 Þx ½25 n ¼ bðtan "cos 20  sin 20 Þ
 
2 ðmy þ nÞx þ py þ q
½26 p ¼ Hsin 20
 ðe1 x þ e0 Þ
where ½27 q ¼ bHcos 20
½14 a3 ¼ Ka1 sinð0  Þ The critical condition is then given by

# 2008 NRC Canada

PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
4 Can. Geotech. J. Vol. 45, 2008

@Sah Note that coefficients L1, M1, and N1 only formally differ
½28 ¼0
@y from L2, M2, and N2, respectively. Therefore eqs. [37] and
[41] provide the same solution
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
@Sah M  M 2  LN1
½29 ¼0 ½45 x¼
@x L
or, alternatively, by where L = L1 = L2, M = M1 = M2, and N = N1 = N2.
Equation [32] admits three real roots
@Sav qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½30
@y
¼0 B 3
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi q 3
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½46 y0 ¼ þ Q þ Q2 þ P 3 þ Q  Q2 þ P 3
3A

@Sav qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½31 ¼0 B  3 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
@x ½47 00
y ¼ þ c Q þ Q2 þ P3
3A qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Both eqs. [28] and [30] give  3 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
þ c2 Q  Q2 þ P3
½32 Ay3 þ By2 þ Cy þ D ¼ 0
with qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B  2 3 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
000
½33 A ¼ 2a3 ðmx þ pÞ ½48 y ¼ þ c Q þ Q2 þ P 3
3A
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 3
þ c Q  Q2 þ P 3
½34 B ¼ a2 ðmx þ pÞ þ 3a3 ðnx þ qÞ
where c is a complex number
pffiffiffiffiffiffiffi
1 þ 3 2 2
½35 C ¼ 2a2 ðnx þ qÞ ½49 c¼ ¼ cos  þ isin 
2 3 3
and P and Q are expressed by real numbers.
½36 D ¼ ða1 þ b1 xÞðnx þ qÞ  ða0 þ b0 xÞðmx þ pÞ  3
B BC D
½50 Q¼ þ 2
Equation [29] gives 3A 6A 2A
½37 L1 x2 þ 2M1 x þ N1 ¼ 0
 2
with C B
½51 P¼ 
½38 L1 ¼ d1 ðmy þ nÞðb1 y þ b0 Þ 3A 3A

Only one of the roots in eqs. [46]–[48] is admissible,

½39 M1 ¼ d1 ðpy þ qÞðb1 y þ b0 Þ being real, positive, and not higher than h. From the analysis
of a large number of cases, it emerged that:
½40 N1 ¼ d0 ðpy þ qÞðb1 y þ b0 Þ þ ða3 y3 þ a2 y2  P is always negative
þa1 y þ a0 Þ½d1 ðpy þ qÞ  d0 ðmy þ nÞ  Q is generally positive, but for high values of  and ", it
can be negative
Equation [31] gives
 Q2 + P3 is always negative
½41 L2 x2 þ 2M2 x þ N2 ¼ 0  The admissible root is always y000 , which can be rewritten
in the following form (and avoids the use of complex
with numbers):
½42 L2 ¼ e1 ðmy þ nÞ½ðb1 y þ b0 Þ  Hðmy þ nÞ  
B pffiffiffiffiffiffiffi pffiffiffi # #
½52 y¼ þ P 3sin  cos
3A 3 3
½43 M2 ¼ e1 ðpy þ qÞ½ðb1 y þ b0 Þ  Hðmy þ nÞ
with
2 3
½44 N2 ¼ e0 ðb1 y þ b0 Þðpy þ qÞ  Hðpy þ qÞ þ ða3 y
þ a2 y2 þ a1 y þ a0 Þ½e1 ðpy þ qÞ  e0 ðmy þ nÞ
The solution of the problem can be obtained using a trial
and error procedure, by solving eqs. [32] and [37] (or
eq. [41]) iteratively.
# 2008 NRC Canada
PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
Greco 5

Fig. 3. (a) Subwedge BDPA’ high z and forces acting on it. (b) Stress disribution on planes BD and DC.

8 0pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
> 2 3 thrust plane BDC is broken (Fig. 3a); hence we assume that
>
> @ Q  P A
>
> tan1 if Q > 0 in general the inclination of these planes can vary with
>
> Q
>
> depth z.
>
<  However, the slip planes starting from plane BD are ac-
½53 #¼ if Q ¼ 0
> 2 tually parallel to one another and the stresses on BD act
>
> 0pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 with inclination  and modulus given by (Das 1987)
>
>
>
> 2 3
>
> tan1
> @ Q  P A þ  if Q < 0 @S1 ðzÞ
: Q ½54 sa ðzÞ ¼ ¼ Ka1 z
@z
where z is the depth of the point considered with respect to
The iterative procedure for solving eqs. [32] and [37] con- point B, and S1(z) is the value of S1 relative to a thrust wedge
verges very quickly. of height z. It is given by eq. [9] by substituting (h – y) with
If, however, eq. [52] gives a value of y higher than h, then z. Consequently, the active thrust S1 is applied to (2/3)(h – y)
the wall has a long heel and the thrust must be calculated below point B (Fig. 3b).
with Rankine’s method. On the DC plane, the stresses are inclined at an angle  –
Note that eq. [32] is a cubic equation only if A = 0. This /2 + ’ with respect to the normal, while the modulus can
occurs if a3 = 0, i.e., if ’ > . Because for concrete walls we be obtained from the relationship:
generally assume the  value to be between 1/2’ and 2/3’,  
the previous mathematical analysis is the useful one in most dS2 ðzÞ @S2 ðzÞ @S2 ðzÞ @ðzÞ
½55 sa ðzÞ ¼ ¼ sin þ
cases. If, on the contrary, one assumes  = ’, then eq. [32] dz=sin  @z @ðzÞ @z
becomes a simpler quadratic equation.
where S2(z) is the value of the thrust S2 relative to the
wedge BDPA’ (Fig. 3a) and is obtained, as in the previous
Stress distribution section, by substituting h with z for wall height.
The determination of stress distribution sa(z) on the thrust ½W 0 ðzÞ þ W 00 ðzÞsinð  0 Þ
planes BD and DC permits the calculation of the position of ½56 S2 ðzÞ ¼
sinð þ 20  Þ
the thrust and the analysis of the wall stability and moreover S1 cosð  0  Þ
the distribution of shear forces and bending moments acting 
on the wall to be obtained. This determination is based on sinð þ 20  Þ
the basic hypothesis, formulated by Huntington (1957), that
the wedge BDCA fails internally along planes starting from
both BD and DC. Originally, Huntington assumed that these
planes were parallel. In the case under consideration the

# 2008 NRC Canada

PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
6 Can. Geotech. J. Vol. 45, 2008

Fig. 4. Inclination angle of the plane PA’ versus depth z for the case examined in the section entitled ‘‘A numerical application’’ (’ = 368,
 = 248, " = 08,  = 16 kN/m3, h = 6 m, b = 0.6 m).

1 zsin ð  "Þ þ ðh  yÞsin ð þ "Þ

½57 W 0 ðzÞ ¼ ðh  y  zÞcot
2 sin cos "
 
00 1 cos cos " zsin ð  "Þ þ ðh  yÞsin "cos  2
½58 W ðzÞ ¼ 
2 sin ð  "Þ sin cos "
Moreover, the angle  is a function of z (Greco 2007), with
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
0 1 C1 þ C12  C0 C2
½59 ðzÞ ¼  þ tan
C2

½60 C2 ¼ A1 ½cosð  0 Þcosð"  0 Þsinð þ 0 Þ  cosð þ 0 Þsinð  "Þ þ ½A0 sinð þ 0 Þ

þ S1 cosð þ 0  Þcos 2 ð"  0 Þ

½61 C1 ¼ A1 cosð  0 Þsinð"  0 Þsinð þ 0 Þ þ ½A0 sinð þ 0 Þ þ S1 cosð þ 0  Þsinð"  0 Þcosð"  0 Þ

½62 C0 ¼ A1 sinð  0 Þsinð"  0 Þsinð þ 0 Þ þ ½A0 sinð þ 0 Þ þ S1 cosð þ 0  1 Þsin 2 ð"  0 Þ

where can be seen in Fig. 5 for the case examined in the sec-
  tion entitled ‘‘A numerical application’’.
1 cos sin " 2 sinð  "Þ (2) In proximity of the maximum the function S2(z) varies
½63 A1 ¼  z þ ðh  yÞ
2 sinð  "Þ sin2  very little in terms of , (as can be seen in Fig. 6 rela-
tive to the case examined in the section entitled ‘‘A nu-
1 cos cos " merical application’’). Therefore if instead of the value
½64 A0 ¼  ðh  yÞ2 of  maximizing S2(z) we introduce a value differing by
2 sinð  "Þ 18 or 28 from it, we have a negligible error, which does
The use of eq. [55] clearly appears prohibitive. However, not exceed 1% of the value of S2(z).
note that: Therefore, we assume for (z) an angle  constant (as in
the procedure proposed by Huntington 1957) and equal to
(1) The angle (z) varies very little (generally within 28–38) the average between the values of  and  obtained from
along DC, especially for low values of the ratio b/h. This eqs. [3] and [11], respectively. The average value should re-

# 2008 NRC Canada

PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
Greco 7

Fig. 5. Thust S2 versus the inclination angle  of the plane CA for the case examined in the section entitled ‘‘A numerical application’’ (’ =
368,  = 248, " = 08,  = 16 kN/m3, h = 6 m, b = 0.6 m).

duce the total error linked to the assumption of (z) as a ½71 SRah ¼ SRa cos "
constant and better distribute the error along DC.
With this simplification, q/qz = 0 and eq. [55] gives a ½72 SRav ¼ SRa sin "
linear distribution of sa(z) on DC
Rankine’s method has been widely used to calculate thrusts
z hþy on retaining walls with a short heel because, being a numer-
½65 sa ðzÞ ¼ sD
a þ ðsCa  sD
aÞ
y ical procedure, Barghouthi’s (1990) method requires a large
number of computations.
where sD C
a and sa are the stresses in points D and C on the However, note that when  = " Rankine’s method gives
plane DC, with
exactly the same results as the present method independent
sinð  Þ sinð  0 Þ of the value of the ratio b/h. In this case, Rankine’s method
½66 sD
a ¼ ðh  yÞcos" requires a much smaller number of computations and is
sinð  "Þ sinð þ 20  Þ
therefore preferable. However, when  = ", the thrust com-
ponents Sah and Sav obtained with Rankine’s method differ
from those obtained with the present method.
sinð  Þ sinð  0 Þ
½67 sCa ¼ Hcos" To compare the present method with that of Rankine, a
sinð  "Þ sinð þ 20  Þ
sensitivity analysis was undertaken by varying the geometri-
Consequently, the depth z2 of the point of application of S2 cal and geotechnical parameters, as shown in Table 1. The
is given by comparison was developed by reviewing 711 cases concern-
 C  ing walls with a short heel. From this comparison, a first ob-
2sa þ sD a y servation regards the angle  of inclination of the plane CA.
½68 z2 ¼ h 
sCa þ sD
a 3 Barghouthi (1990) asserted that this angle is always equal to
that obtained from Rankine’s theory. From the analysis of
the results it emerges that this is true only when  = ". In
Comparison with Rankine’s method the general case (i.e.,  = ") the angle  differs from that
given by Rankine’s theory; in fact it is also dependent on
Using Rankine’s method the active thrust on CE is given the angle  and the ratio b/h, which Rankine’s theory does
by not take into account.
1 The errors in the components Sah and Sav caused by an in-
½69 SRa ¼ KaR H 2 appropriate use of Rankine’s method are represented in
2
Fig. 6, where abscissae report the ratio b/h, which is the
where (Bowles 1982; Das 1987) most important parameter, while ordinates report the relative
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi errors (Sah) and (Sav), defined as
R cos "  cos 2 "  cos 2 0
½70 Ka ¼ cos" pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cos " þ cos 2 "  cos 2 0 SRah  SCah
½73 ðSah Þ ¼
SCah
The thrust components are given by

# 2008 NRC Canada

PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
8 Can. Geotech. J. Vol. 45, 2008

Fig. 6. The errors (Sah) and (Sav) versus the ratio b/h. Table 1. Values used for the comparison.

Variable Values
’ (8) 20, 25, 30, 35, 40, 45
/’ 0, 1/3, 1/2, 2/3
"/’ 0.00 to 0.80 with step 0.10
d/b 0.10 to 0.50 with step 0.05

Rankine’s method undestimates Sah and overestimates Sav.

The difference between the two methods increases as the ra-
tio b/h decreases and the differences  – " and ’ – " in-
crease. Because we assume generally  = 2/3’, the method
of Rankine is conservative, except for the steepest backfill.
In the range of the examined parameter values, the errors
attributable to an inappropriate use of Rankine’s method in
Sah do not exceed 9% in overestimation and 3% in underes-
timation for b/h = 0.10. For b/h > 0.2 the errors are less than
5% for the overestimate and 1.5% for the underestimate.
As regards Sav, Rankine’s method gives higher errors,
which can amount to 15% of SCah for the overestimate and
20% for the underestimate.
Finally, concerning the location of the point of application
of Sa, Fig. 7 shows the ratio yS/H obtained with the present
method versus the ratio b/h. The dashed line refers to the
position given by Rankine’s method (H/3). The present
method gives higher values of yS (up to 0.340H) if  > "
and lesser (up to 0.327H) for  < ".
In summary, if  > ", then Rankine’s method is conserva-
tive with respect to the thrust components Sah and Sav and
slightly unconservative with respect to the arm of Sah. The
contrary occurs when  > ".

A numerical application
The previously described method is used to solve a prob-
lem of active earth pressure with the following data: ’ =
368,  = 248, " = 08,  = 16 kN/m3, h = 6 m, b = 0.60 m.
From these input data we have:
 = 59.38, H = 6.00 m, Ka1 = 0.235, a3 = 0.049, a2 = –0.934,
a1 = 6.140, a0 = 2.424, b1 = 4.854, b0 = –2.116, d1 = –0.588,
d0 = 4.854, e1 = 0.809, e0 = 3.527, m = 0.309, n = –0.571, p =
SRav  SCav 5.706, q = 1.112.
½74 ðSav Þ ¼ Assuming tentatively y = h, we have:
SCah
L = –20.38, M = –561.20, N = 4196.9, x = 3.515 m, A =
where the index R refers to thrusts calculated with Ran- 0.663, B = –6.47, C = 1.67, D = 13.33, P = –9.74, Q =
kine’s method, while the index C refers to thrusts calculated 20.25, w = 0.280, y = 1.749 m.
with the present method based on Coulomb’s approach. Assuming now y = 1.749 m, we obtain:
Positive values (Sah) and (Sav) correspond to overesti- L = 0.113, M = –41.56, N = 275.9, x = 3.334 m, A =
mates of the related thrust components using Rankine’s 0.658, B = –6.40, C = 1.47, D = 13.56, P = –9.78, Q =
method. Overestimates of Sah and underestimates of Sav are 20.22, w = 0.283, y = 1.754 m.
conservative for the wall stability against sliding and foun- Generally, two iterations are sufficient to obtain a preci-
dation bearing capacity, while underestimates of Sah and sion of 1 cm in y and with three iterations the precision
overestimates of Sav are unconservative for the same col- reaches 1 mm. In fact, beginning a new iteration starting
lapse mechanisms. The inconsistencies of the lumped safety with y = 1.781 m, we obtain:
factor approach for this collapse mechanism have been dem-
L = 0.108, M = –41.81, N = 277.6, x = 3.334 m, A =
onstrated (Greco 1997) with respect to the traditional over- 0.658, B = –6.40, C = 1.47, D = 13.56, P = –9.78, Q =
turning around the wall toe. 20.22, w = 0.283, y = 1.754.
From this study it emerges that when  > ", Rankine’s With these values of x and y, we obtain finally:
method overestimates Sah and underestimates Sah, with re-  = 60.98,  = 108.98, S1 = 33.9 kN/m, S2 = 62.72 kN/m,
spect to the present method. On the contrary, if  < ", then Sah = 67.0 kN/m, Sav = 15.9 kN/m.
# 2008 NRC Canada
PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
Greco 9

Fig. 7. Ratio ys/H versus the ratio b/h.

In comparison, Rankine’s method gives: as the ratio b/h decreases and the differences  – " and ’ –
Sah = 74.8 kN/m, Sav = 0.0 kN/m,  = 63.08. " increase. More specifically, for  > " Rankine’s method
Because  = (59.58 + 60.98)/2 = 60.28, the stress distribu- gives higher values of Sah and lower values of Sav, with re-
tion is characterized by: spect to the method proposed here. The latter gives a
slightly higher position of the point of application of the
sD C
a = 28.04 kPa, sa = 39.61 kPa. thrust Sa, but for walls of usual height this difference is
With these stresses, we have: only a few centimetres. The use of Rankine’s method,
z2 = 5.17 m and S2 = 62.69 kN/m. although inappropriate from a theoretical point of view, is,
This value of S2 differs very little from the value previ- in this case, conservative. The contrary occurs if  < ".
ously obtained (62.72 kN/m). The small error (0.03%) is
due to the simplified assumption of constancy of the angle References
(z) along DC.
Barghouthi, A.F. 1990. Active earth pressure on walls with base
The equilibrium condition of force moments acting on the projection. Journal of Geotechnical Engineering, 116(10): 1570–
wedge BDCE with respect to point C, gives the position of 1575. doi:10.1061/(ASCE)0733-9410(1990)116:10(1570).
the point of application of Sa: yS = 2.04 m. Bowles, J.E. 1982. Foundation analysis and design. McGraw-Hill.
Coulomb C.A. 1773. Essai sur une application des règles de maxi-
Conclusions mis et minimis à quelques problèmes de statique, relatifs à l’ar-
chitecture. Mémoires de Mathématique et de Physique présentés
The active thrust on cantilever walls with a short heel is à l’Academie Royale des Sciences, 7: 343–382. [In French.]
often calculated using Rankine’s method, although this pro- Das, B.M. 1987. Theoretical foundation engineering. Elsevier, Am-
cedure has shortcomings from a theoretical point of view. sterdam, The Netherlands.
Coulomb’s method is more appropriate. However, only nu- Greco, V.R. 1992. Discussion of ‘‘Active earth pressure on walls
merical procedures, requiring a large number of computa- with base projection’’ by Amjad F. Barghouthi (October, 1990,
tions are currently available. Vol. 116, No. 10). Journal of Geotechnical Engineering, 118(5):
This paper has presented an analytical solution for the ac- 825–827. doi:10.1061/(ASCE)0733-9410(1992)118:5(825).
tive thrust on retaining cantilever walls with a short heel, Greco, V.R. 1997. Stability of retaining walls against overturn-
formulated as an algorithm articulated on the alternate solu- ing. Journal of Geotechnical and Geoenvironmental Engineer-
tions of a system of two equations, one quadratic and the ing, 123(8): 778–780. doi:10.1061/(ASCE)1090-0241(1997)
other cubic. Despite the apparent complexity, the proposed 123:8(778).
procedure can be easily implemented in a spreadsheet or a Greco, V.R. 1999. Active earth thrust on cantilever walls in general
computer code. Moreover, it is rapidly convergent. conditions. Soils and Foundations, 39(6): 65–78.
Greco, V.R. 2001. Active earth thrust on cantilever walls with short
The results obtained with the proposed method have been heel. Canadian Geotechnical Journal, 38(2): 401–409. doi:10.
compared with those obtained using Rankine’s method. 1139/cgj-38-2-401.
From this comparison, it emerged that, when the friction an- Greco, V.R. 2007. Analytical earth thrust on walls with bilinear
gle between soil and wall () is equal to the inclination an- backface. Proceedings of the Institution of Civil Engineering -
gle of the topographic profile of the backfill ("), Rankine’s Geotechnical Engineering, 160(1): 23–29. doi:10.1680/geng.
method gives exactly the same results as the present method. 2007.160.1.23.
In the case of  = ", the present method gives results that Huntington, H. 1957. Earth pressure and retaining walls. John Wi-
differ from those of Rankine, and these differences increase ley & Sons, New York.
# 2008 NRC Canada
PROOF/ÉPREUVE
 Pagination not final/Pagination non finale
10 Can. Geotech. J. Vol. 45, 2008

Müller-Breslau, H. 1906. Erddruck auf Stützmauern. Alfred Kro- R1 force acting on the failure plane DF (Fig. 2b)
ener. Stuttgart [In German]. sa(z) stress on a point at depth z below point B along the
Rankine, W.J.M. 1857. On the stability of loose earth. Philosophi- thrust planes BD and DC (Fig. 4)
cal Transactions of the Royal Society of London, 147: 9–27. sCa stress on point C of the plane DC (Fig. 3)
doi:10.1098/rstl.1857.0003. sDa stress on point D of the plane DC (Fig. 3)
Silvestri, V. 2006. Discussion of ‘‘Active earth thrust by backfills S1 active thrust acting on the plane BD (Fig. 2a)
subject to a line surcharge’’. Canadian Geotechnical Journal, S2 active thrust acting on the plane DC (Fig. 2a)
43(11): 1220–1223. doi:10.1139/T06-103. S2(z) active thrust acting on the plane DP (Fig. 3)
Sa active earth thrust acting on the vertical plane CE
List of symbols (Fig. 2c) calculated with the present method
SRa active earth thrust acting on the vertical plane CE
a0 coefficient given by eq. [17] (Fig. 2c) calculated with Rankine’s method
a1 coefficient given by eq. [16] Sah horizontal component of the active earth thrust Sa
a2 coefficient given by eq. [15] (Fig. 2c)
a3 coefficient given by eq. [14] SCah horizontal component of the active thrust on CE
A coefficient given in eq. [33] (Fig. 2c) calculated using the present method
A0 coefficient given by eq. [64] SRah horizontal component of the active thrust on CE
A1 coefficient given by eq. [63] (Fig. 2c) calculated using Rankine’s method
b length of wall heel (Fig. 2) Sav vertical component of the active earth thrust Sa
b0 coefficient given by eq. [19] (Fig. 2c)
b1 coefficient given by eq. [18] SCav vertical component of the active thrust on CE
B coefficient given in eq. [34] (Fig. 2c) calculated using the present method
c complex number given by eq. [49] SRav vertical component of the active thrust on CE
C coefficient given in eq. [35] (Fig. 2c) calculated using Rankine’s method
C0 coefficient given by eq. [58] W weight of the wedge BDCA (Fig. 2a)
C1 coefficient given by eq. [61] W(z) weight of the wedge BDPA’ (Fig. 3)
C2 coefficient given by eq. [60] W’ weight of the wedge BDCE (Fig. 2a)
d0 coefficient given by eq. [21] W’(z) weight of the wedge BDPE’ (Fig. 3)
d1 coefficient given by eq. [20] W@ weight of the wedge ECA (Fig. 2a)
D coefficient given in eq. [36] W@(z) weight of the wedge E’PA’ (Fig. 3)
e0 coefficient given by eq. [23] W1 weight of the wedge BDF (Fig. 2b)
e1 coefficient given by eq. [22] y height of the thrust plane DC (Fig. 2a)
Fs safety factor of wall against sliding along the base y’, y@, y’’’ roots of eq. [32]
h wall height (Figs. 1 and 2a) y2 height of point of application of S2 above the wall
H thickness of backfill on a vertical plane CE through base (Fig. 4)
wall heel (Fig. 2a) yS height of point of application of the active thrust Sa
Ka1 coefficient of active earth thrust on BD (Fig. 2a) gi- above the wall base (Fig. 2c)
ven by eq. [10] z depth of point P below point B (Fig. 3)
KaR coefficient of active earth thrust on CE (Fig. 2a) z2 depth of the point of application of S2 below point
calculated with Rankine’s method (eq. [70]) B (Fig. 3)
L coefficient in eq. [45], L = L1 = L2  inclination angle of failure plane CA (Fig. 2a)
L1 coefficient given in eq. [38] (z) inclination angle of failure plane PA’ (Fig. 3) given
L2 coefficient given in eq. [42] by eq. [55]
m coefficient given by eq. [24]  constant approximate value assumed for (z)
M coefficient in eq. [45], M = M1 = M2  inclination angle of failure plane DC (Fig. 2a)
M1 coefficient given in eq. [39]  unit weight of backfill soil
M2 coefficient given in eq. [43]  friction angle between backfill soil and wall back-
n coefficient given by eq. [25] face (Fig. 2a)
N coefficient in eq. [45], N = N1 = N2 " inclination angle of topographic profile (Fig. 2a)
N1 coefficient given in eq. [40]  inclination angle of failure plane DF (Fig. 2b)
N2 coefficient given in eq. [44] w parameter given by eq. [53]
p coefficient given by eq. [26] (Sah) relative error on Sah linked to inappropriate use of
P coefficient given in eq. [51] Rankine’s method (eq. [65])
q coefficient given by eq. [27] (Sav) relative error on Sav linked to inappropriate use of
Q coefficient given in eq. [50] Rankine’s method (eq. [66])
R force acting on failure plane CA (Fig. 2a) ’ internal friction angle of backfill soil
R(z) force acting on failure plane PA’ (Fig. 3)

# 2008 NRC Canada

PROOF/ÉPREUVE
View publication stats