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

net/publication/262962249

Stability of granite drystone masonry retaining walls: I. Analytical design

Article in Geotechnique · November 2012

DOI: 10.1680/geot.10.P.112

CITATIONS READS

25 4,156

4 authors, including:

Leandro Alejano
University of Vigo
228 PUBLICATIONS 4,323 CITATIONS

SEE PROFILE

All content following this page was uploaded by Leandro Alejano on 04 January 2016.

The user has requested enhancement of the downloaded file.

Alejano, L. R. et al. (2012). Géotechnique 62, No. 11, 1013–1025 [http://dx.doi.org/10.1680/geot.10.P.112]

Stability of granite drystone masonry retaining walls: I. Analytical design

L . R . A L E JA N O , M . V E I G A † , J. TA B OA DA  a n d M . D Í E Z - FA RTO ‡

A limit-equilibrium-based technique is proposed for the La présente communication présente une technique basée
design and analysis of drystone masonry retaining walls, l’équilibre limite pour l’étude et l’analyse de murs de
focusing particularly on granite walls as traditionally soutènement en pierre sèche, en se concentrant en parti-
constructed in Galicia in northwest Spain. To date, walls culier sur des murs en granite, du type utilisé tradition-
have typically been designed on the basis of past experi- nellement dans la construction en Galicie, dans le nord
ence and rigid wall approaches. The method presented de l’Espagne. Jusqu’à présent, les murs étaient générale-
analyses wall stability, at different heights associated with ment conçus en tenant compte de l’expérience acquises et
block rows, against sliding and against two overturning des méthodes de murs rigides. La méthode proposée
mechanisms. As a result, design widths are obtained for analyse la stabilité des murs, à différentes hauteurs
different wall depths. The method can also be applied to relatives à des rangées de parpaings, en fonction du
estimates of the stability of existing walls with known glissement ainsi que de deux mécanismes de renverse-
dimensions and properties. The analytical method was ment. En conséquence, on obtient des largeurs nominales
applied experimentally to the design of a 20 m high wall. pour différentes profondeurs de murs. Cette méthode
A sensitivity analysis of the design was also performed, peut être appliquée également à des estimations de la
with results presented in the form of a spider diagram. stabilité de murs existants aux dimensions et propriétés
Estimates of the most relevant parameters for this type connues. On applique cette méthode d’analyse de façon
of design and some back-analyses to assess the technique expérimentale à l’étude d’un mur de 20 m de haut. On
are addressed in an accompanying paper. également procède à une analyse de la sensibilité du
modèle, et présenté les résultats sous forme de schéma en
« toile d’araignée ». Enfin, un article joint à cette com-
munication se penche sur les paramètres les plus perti-
nents pour ce type de modèle ainsi que sur certaines
KEYWORDS: design; limit equilibrium methods; retaining walls rétro-analyses pour évaluer la technique.

INTRODUCTION A problem with using drystone walls is the lack of

Drystone walls date from early Neolithic times, when they scientific knowledge of drystone wall behaviour. A stan-
were used to build dwellings and protective walls for settle- dard reinforced concrete wall is correctly assumed to
ments. They eventually came to be used to retain soil and so behave as a homogeneous rigid structure that yields
increase the quantity of flat land for agricultural purposes. according to solid mechanics, and its stability is assessed
Other early civilisations, mainly mountain communities such in terms of its limit equilibrium (LE) and factor of safety
as the Inca, improved this technique by constructing anti- (FoS). The stability of a drystone retaining wall is likely
seismic walls that still stand 600 years after construction to depend on factors not usually taken into account in
(Fig. 1(a)). In the Mediterranean region, and in north-west conventional analyses. Given that a drystone wall is built
Spain and Portugal, retaining walls were used to terrace hilly as rows of masonry, the LE approach is reasonable, but
areas for agriculture (Fig. 1(b)). The industrial revolution in must be applied to each and every row of blocks, from
Europe and the development of rail and road networks led the upper part of the wall down to its base. This is the
to a significant increase in the number and size of these approach adopted by the authors for the case of traditional
walls. According to the UK Department of Transport (1987) retaining drystone granite walls as constructed in Galicia
there are still more than 2700 km of this type of wall in the (Fig. 2).
British infrastructure. Figure 2 shows how the granite walls being dealt with
The development of concrete and new materials in the here differ from traditional retaining walls as reflected in the
20th century led to diminished use of drystone masonry literature (Zimbabwe’s walls, studied by Dickens & Walker
walls. However, there has been a growing interest in using (1996), or British walls as studied by O’Reilly et al. (1999),
drystone walls in recent decades, to preserve old structures Harkness et al. (2000), Claxton et al. (2005) and so on) in
and to build new ones, given their lower environmental terms of the large size of the blocks (sometimes up to some
impact and their better landscape integration. Walls of this tonnes) and the wall height (up to 20 m). According to
kind are also used in modern construction projects as retain- Cooper (1986) and Wong & Ho (1997), drystone retaining
ing walls (Fig. 1(c)), and are frequently incorporated in walls are generally between 2 m and 10 m high.
modern architectural buildings (Fig. 1(d)). A different approach to modelling drystone walls follows
standard contact laws, and assumes that they are made of
very rigid pieces of rock in contact. Such walls can be
Manuscript received 14 October 2010; revised manuscript accepted modelled by means of distinct-element codes such as UDEC
21 March 2012. Published online ahead of print 23 August 2012. (Itasca, 2004). Dickens & Walker (1996) used an early
Discussion on this paper closes on 1 April 2013, for further details see version of this code for the first time for masonry walls,
p. ii.
 Department Ing. de Recursos Naturales y Medio Ambiente, following applications (Harkness et al., 2000; Powrie et al.,
Universidad de Vigo, Spain. 2002) that analysed full-scale tests of dry masonry walls
† Itasca Consultants AB, Stockholm, Sweden. Formerly of Universi- (Burgoyne, 1853).
dad de Vigo. Another possible approach to the problem – developed by
‡ S. A. de Obras y Servicios COPASA, Spain. Pande et al. (1989), and applied to masonry structures by de

1013
1014 ALEJANO, VEIGA, TABOADA and DÍEZ-FARTO

(a) (b)

(c) (d)

Fig. 1. (a) Retaining walls in the Inca city of Machu Picchu (Peru), built to increase urban space; (b) drystone retaining terraces for
agricultural purposes in the Gêres region of northern Portugal; (c) granite drystone masonry wall built to retain factory facilities in Vigo
(Spain); (d) the Museum of Man in A Coruña (Spain) with a drystone base built of pink granite

& Walker, 1996; Harkness et al., 2000; Powrie et al., 2002;

Claxton et al., 2005) focus on analysing existing or pre-
defined structures, whereas the present authors emphasise
design.

STATEMENT OF THE PROBLEM AND ANALYTICAL

DESIGN METHOD
A drystone granite masonry wall is constructed of piled-
up dry stones that rely on weight for stabilisation purposes.
In the current case, the dry stones are more or less cuboid-
shaped granite blocks. Small rock fragments or cuttings are
used to immobilise the blocks, and to fill the gaps between
blocks during construction (Fig. 2). Backfill usually consists
of completely decomposed granite, as it is abundant in
Galicia, is easy to obtain, and has good draining features.
Although the fill can provide a degree of cohesion, for the
sake of simplicity it is not considered in the analytical
Fig. 2. Pink granite dry masonry retaining wall near Vigo (Spain)
measuring 14 m high design procedure adopted here.
To optimise construction of these retaining walls and
achieve a predefined FoS against sliding and overturning, a
Buhan & de Felice (1997) and to retaining walls by Colas et method was developed that calculates the necessary dimen-
al. (2008) – is to homogenise the parameters. However, sions of the wall for each and every row, given that the
although the interest of this approach for particular cases is contacts between blocks have no tensile strength. The same
recognised by the present authors, it has been excluded from method can also be used to assess the stability of existing
this study. walls if their dimensions are known. Stability against rota-
Note that most of the approaches described (e.g. Dickens tional sliding under the wall base was also considered as
 STABILITY OF GRANITE DRYSTONE MASONRY RETAINING WALLS: I 1015
well as the bearing capacity of the ground below the wall water table, a force U 1 was considered to be acting in the
foundation. An accompanying paper (Alejano et al., 2012) base. The water pressure at the base of block 1 was thus
describes a test of the method based on back-analysis of estimated as
fallen walls and numerical models to compare results.
To date, drystone walls have been calculated as rigid, but u ¼ ªw h1w (1)
this does not properly represent their actual behaviour.
Mundell et al. (2009) recently described an interesting LE Considering a uniform distribution of water pressure acting
approach to analysing the stability of drystone retaining on the horizontal joints between blocks, the value of force
walls. In their approach they determined the horizontal and U 1 can be estimated. The total force integrating this pressure
vertical stresses acting at each level at the back of the wall, distribution depends on the width of the block, which is the
and the direction of the overall thrust. They then considered design parameter to be calculated.
the standard stable construction criterion that the eccentricity The factor of safety (FoS) against the sliding of a block
of the thrust line must remain within the middle third of the was estimated as the ratio of available and mobilised shear
base of the structure. This is a convenient approach, not only forces, as
from a practical perspective; it can also be used for analys- Tavailable
ing bulging behaviour. However, it does not account expli- FoS ¼ (2)
citly for overturning failure, and does not calculate an FoS. Tmobilised
The earth pressures on the active and passive sides of the Considering the force equilibrium for block 1 (Fig. 3) and
wall, together with the active and passive forces on the wall given that, for LE, the sum of forces in the horizontal and
and the accompanying shear stresses and their lines of vertical directions should equal zero, the following calcula-
action, were estimated following classic soil mechanics tions can be made.
approaches (Coulomb, 1776; Caquot & Kerisel, 1948; Berry
& Reid, 1987; Das, 2002) (see the Appendix). ÓFvert: ¼ 0 ) N 1 ¼ W 1 þ T 1a  U 1
(3)
ÓFhoriz: ¼ 0 ) T 1b ¼ Tmobilised ¼ P1a
STABILITY AGAINST SLIDING
In the present study’s walls, sliding may take place at Therefore, it is possible to write
different heights of the wall through contacts between  1 
blocks, so the calculation approach focused on stabilisation N 1 tan b W þ T 1a  U 1 tan b
FoS ¼ ¼ (4)
against sliding for each row of blocks. Rows of blocks were Tmobilised P1a
considered to have a constant height (typically 0.5 m in
Galicia), and the calculation process started with the upper where b is the friction angle between blocks. The value of
block (Fig. 3). The forces acting on this block are its weight P1a was estimated according to equation (32), which was
W 1 , the normal reaction N1 , the active shear force T 1a , the updated to take into account the conditions of the case. The
shear force between blocks T 1b , and finally the active force values for W 1 and U 1 , which depend on block width w1s , are
P1a , estimated according to equation (32) and accounting for
any possible water pressure. The passive normal and shear W 1 ¼ ªb h1 w1s
forces act only below ground level. (5)
Even though the fill is supposed to be highly permeable, U 1 ¼ ªw h1w w1s
in particular designs pore pressure may arise under heavy
rainfall conditions. For the sake of simplicity, hydrostatic Finally, introducing these values in equation (4), the design
conditions were assumed, as an approximation. Accordingly, parameter w1s can be calculated as
it was assumed that the water surface is horizontal in the fill,
and also within the wall from its back to the wall face. Even FoS: P1a = tan b  T 1a
w1s ¼ (6)
if this assumption is not rigorous, it is mathematically ªb h1  ªw h1w
simple, and reasonably representative of conditions in the
lower layers of the wall during heavy rainfall. The width of the first row of blocks ensuring stability
So, in the case of the first row located under a ground- against sliding was thus calculated. Calculation proceeded
for the next rows thus: once the ith row was stabilised, row
i + 1 was designed. The blocks over row i + 1 were already
w1s stable, and their widths and weights were known. The forces
acting on the blocks in row i + 1 are depicted in Fig. 4.
The total weight of blocks and fill acting on the block in
row i + 1 was calculated by adding the weight of the
i
T1a materials, called W over , to the contact for rows i and i + 1.
P1a W iþ1
is the weight of the block needed to stabilise row
h1 W1
i + 1, and W fi is the weight of the fill located over the step
between rows i and i + 1, that is, over the part of the blocks
h1w in row i + 1 protruding from row i. These unknowns, W over i

and W fi , can be written in terms of wsiþ1 , which is the

T1b parameter to be calculated in the next calculation step.
U1
N1 The equilibrium of forces in the horizontal and vertical
directions now reads
ÓFhoriz: ¼ 0 ) T mobilised
iþ1
¼ Paiþ1  Ppiþ1

ÓFvert: ¼ 0 ) N iþ1 ¼ W over

i
þ W fiþ1 þ W iþ1 (7)
Fig. 3. Stability against sliding analysis: forces acting on the first
row of blocks þ T aiþ1  T piþ1  U iþ1
1016 ALEJANO, VEIGA, TABOADA and DÍEZ-FARTO

w is

w if
w iover

T i⫹1
a
P i⫹1
a

T i⫹1
p

hi⫹1
Row i ⫹ 1 w

P i⫹1
p W i⫹1

T i⫹1
b
i⫹1
N
U i⫹1

Fig. 4. Stability against sliding analysis: forces acting on block row i + 1

where Paiþ1 and Ppiþ1 are the normal active and passive  X
i
iþ1
W over i
¼ W over þ ªb wsiþ1 h iþ1 þ ª wsiþ1  wsi hi (11)
forces, according to equations (32) and (35), needed to
i¼1
analyse the stability of the wall at the depth of row i + 1.
These forces vary for every depth of analysis, but they refer The calculation process continues down to the wall base.
to the integration of pressures acting from the block of
analysis to the surface. T aiþ1 and T piþ1 are the corresponding
active and passive shear forces, according to equations (34) STABILITY AGAINST OVERTURNING
and (36), for row i + 1. Overturning may occur in the full width of a row. A mode
The FoS is calculated, again as in equation (2), to yield I toppling mechanism tends to occur where a single block
forms the full width of the wall at a particular depth (Fig.
N iþ1 tan b
FoS ¼ iþ1
5(a)). However, more common is the mode II toppling
T mobilised mechanism, where the failure occurs as a result of the
  overturning of part of the wall whose base forms an angle 
i
W over þ W fiþ1 þ W iþ1 þ T aiþ1  T piþ1  U iþ1 tan b with the horizontal (Fig. 5(b)). Although the second situation
¼ is more common in standard high retaining walls, both
Paiþ1  Ppiþ1
mechanisms are accounted for in this study.
(8)

Some of these variables can be further decomposed, to give

W iþ1 ¼ ªb h iþ1 wsiþ1

U iþ1 ¼ ªw hwiþ1 wsiþ1

(9)
 X
i
W fiþ1 ¼ ª wsiþ1  wsi hi
i¼1

Finally, introducing these values in equation (8), the design

parameter wsiþ1 can be calculated as 
h   i
FoS: Paiþ1  Ppiþ1 = tan b  W over i
Xi
ªwsi h i þ T piþ1  T aiþ1
wsiþ1 ¼ Xi¼1
i (10) (a) (b)
ªb h iþ1 þ ª i¼1 h i  ªw hwiþ1
Fig. 5. Failure modes due to overturning, depending on wall type:
iþ1
The value of W over needed for further calculations can now (a) mode I, full-wall-width blocks; (b) mode II, standard-width
be estimated as blocks
 STABILITY OF GRANITE DRYSTONE MASONRY RETAINING WALLS: I 1017
Stability against overturning for full-wall-width blocks (mode I) w1tl
The procedure for calculating wall stability against mode w1tl/2
I toppling follows the same sequence as for stability against
sliding. The width of the first block is calculated to ensure
that toppling will not occur and, on this basis, the width for
the second block, and so on to the wall base.
The lines of action of the different forces need to be T1a
P1a
1 W1
known in order to be able to analyse overturning. Taking h
into account the rotation line located in the external part of h1a
the base in each row, the lever arms for the different forces h1w
can be estimated. Accordingly, the FoS can be calculated in
U1
terms of the ratio between stabilising or resisting moments w1tl/2
and overturning moments as Rotation line

X
M resist:
FoS ¼ X (12)
M overt:
Fig. 6. Stability against overturning analysis: forces acting on first
row of blocks
For the first row of blocks (Fig. 6), the sum of resisting and
overturning moments can be estimated as
w1tI ¼
X
M resist: ¼ T 1a w1tI þ 12 W 1 w1tI rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h i ffi
 1 2  
X (13) T 1a þ T a  4 ªb h =2  2 FoS  ªw h w FoS  Pa ha
1 1 1 1 1

M overt: ¼ P1a h1a þ 12 U 1 w1tI

ªb h1  FoS  ªw h1w
(15)
where the weight of the block and the water force (if any) in
the base of the block – according to the simplified ground- Since the active shear force T 1a is always positive or null,
water approach described above – depend on the design there is a negative solution that is ruled out. If w1tI , 0, then
width of the block, w1tI , according to this means that the vertical backfill slope is self-stable, and
calculation of the first row ends.
W 1 ¼ ªb h1 w1tI Once the width of the wall up to the ith row of blocks is
(14) designed, the weight of the materials over the contact of
U 1 ¼ ªw h1w w1tI rows i and i + 1, called W overi
, and its lever arm to the face
of the wall, called x9w i, can be obtained (see Fig. 7).
W iþ1 is the weight of the block needed to stabilise row
Introducing equation (14) in equation (13) and equation (13) i + 1. The centre of gravity of this block is located at
in equation (12), for a particular value of the FoS, it is distance wtIiþ1 =2 from the wall face. These unknowns can be
possible to calculate the width of the first block to obtain written in terms of wtIiþ1 , which is the parameter to be
the design as calculated in the next step.

w itl⫹1
x if
w i⫹1
tl /2

x⬘wi

W iover Wif

T i⫹1
a P i⫹1
a

P i⫹1
p
h ia⫹1 h iw⫹1
i⫹1 Row i ⫹ 1
W
h i⫹1
p

U i⫹1
Rotation line w i⫹1
tl /2

Fig. 7. Stability against overturning analysis: forces acting on block row i + 1

1018 ALEJANO, VEIGA, TABOADA and DÍEZ-FARTO
W fi is the weight of fill located over the step between in the base (Fig. 8) that alters force equilibrium. The angle
rows i and i + 1, that is, over the part of the blocks in row  plays a non-negligible role.
i + 1 protruding from row i. The line of action for this The width of row i + 1 needed to stabilise the wall against
weight goes vertically through the centre of this zone; its mode II overturning at the corresponding depth is now
iþ1
horizontal distance to the line of rotation is called xfi : called wtII : In this case, the point of rotation for row i + 1
Paiþ1 and Ppiþ1 (which can be derived from the active and descends vertically for a value d iþ1 : Since most of the forces
passive earth pressure coefficients) represent the active and acting on the overturning wall are originally vertical, the
passive forces at this wall depth, and haiþ1 and hpiþ1 are their lever arms do not vary, and so are kept as in the mode I
respective lever arms, obtained according to equations (33) calculation. However, for the horizontal forces, and particu-
and (35). T aiþ1 represents the active shear force between the larly for the active force at the wall back, the lever arm is
fill and the wall on the active side, and wtIiþ1 is the distance increased in a distance equal to the descent of the rotating
between its line of action and the rotation line. point, which depends on the width of the wall at that depth
Therefore, for the equilibrium calculation W over i
, Paiþ1 , and on the angle , according to
iþ1 iþ1 i iþ1 iþ1
Pp , T a , x9w i , wtI , ha and hp are known. The values for
W fi , W tiþ1 , xfi and U iþ1 , meanwhile, are dependent on wtIiþ1 , d iþ1 ¼ wtII
iþ1
tan  (20)
according to
 X
i
W fi ¼ ªf wtIiþ1  wtIi hi
i¼1
Accordingly, a new triangular prism or wedge located under
the base of row i + 1 must overturn, together with the upper
W iþ1 ¼ ªb h iþ1 wtIiþ1 (16) part of the wall. Its weight can be calculated starting from
iþ1
  the width of row i + 1 (wtII ) and its lever arm is one third
1 iþ1
xfi ¼ 12 wtIiþ1 þ wtIi of this distance, which is 3 wtII : The weight of the wedge in
the lower part and its lever arm are estimated as
U iþ1 ¼ ªw hwiþ1 wtIiþ1
 iþ1 2
iþ1
W wdg ¼ 12 ªb tan  wtII
The corresponding stabilising and overturning moments (Fig.
7) can now be calculated as iþ1 iþ1
xwdg ¼ 13 wtII (21)
X
i
M resist: ¼ W over x9w i þ 12 W iþ1 wtIiþ1  
iþ1 3
xwdg ¼ 16 ªb tan  wtII
iþ1 : iþ1
W wdg
þ W fi xfi þ T aiþ1 wtIiþ1 þ Ppiþ1 hpiþ1 (17)
X
M overt: ¼ Paiþ1 haiþ1 þ 12 U iþ1 wtIiþ1 Therefore, for the equilibrium calculation W over i
, Paiþ1 , Ppiþ1 ,
iþ1 iþ1 iþ1 iþ1
T a , x9w i , wtII , ha and hp are known. The values for
iþ1 : iþ1
Recalling that the ratio of these two expressions is the FoS, W fi , W tiþ1 , xfi , Uiþ1 , diþ1 and W wdg xwdg are dependent on
iþ1
a quadratic equation for wtIiþ1 is solved to obtain the positive wtII , according to equations (22).
value indicating the width of the blocks in row i + 1 that
fulfil the stability criterion. This value is calculated as
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 12 0 1 3
u
u X i X i
u Bªf h i
C6 B ª hCi
7
u @ i¼1 A 4 2 @ f i¼1 A 5
t iþ1 2 ªb h i 1 
1
T a þ Ta 4 þ  FoS  ªw hw iþ1 :
wt i
 i iþ1 iþ1
þ W over x9wi þ Pp hp  FoS  Pa ha iþ1 iþ1
2 2 2 2
wtIiþ1 ¼
X i
ªf h i þ ªb h i  FoS  ªw hwiþ1
i¼1
(18)

The obtained value of wtIiþ1 is the width for row i + 1 needed

to stabilise the wall against mode I overturning at the
corresponding depth.
iþ1
The value for W over can now be calculated as in equation
(11), and x9w i for the next step can be estimated as
i
W over x9w i þ W iþ1 wtIiþ1 =2 þ W fi xfi
x9w iþ1 ¼ iþ1
(19)
W over
P i⫹1
a
The calculation proceeds in this way down to the wall base.
Rotation point for mode I h i⫹1
pa

Stability against overturning for standard wall-width blocks

(mode II)  d i⫹1
The previous approach is a particular case of this more
complex case. The calculation for the first block can be
performed as described above. At a certain depth where Rotation point for mode II
there are many blocks, however, the most common failure w i⫹1
tll

mechanism is overturning (depicted in Fig. 5(b)); the falling

part of the wall does not have a plane base, but has a wedge Fig. 8. Features of mode II overturning
 STABILITY OF GRANITE DRYSTONE MASONRY RETAINING WALLS: I 1019
 iþ1 X
i deduced, and the solution to the equation can thus be easily
W fi ¼ ªf wtII i
 wtII hi entered in a calculation sheet.
i¼1 As for the value of , from a theoretical perspective, the
FoS can be expressed as a function of  to obtain
W iþ1 ¼ ªb h iþ1 wtII
iþ1

 iþ1  c1 þ c2 tan 
xfi ¼ 12 wtII i
þ wtII (22) FoS ¼ (25)
c3 þ c4 tan 
U iþ1 ¼ ªw hwiþ1 wtII
iþ1

where c1 to c4 are constants. The derivative of this function

d iþ1 ¼ wtII
iþ1
tan  indicates that the FoS is a monotonic growing function for
 iþ1 3 , between 08 and 908 in all the studied cases. This means
xwdg ¼ 16 ªb tan  wtII
iþ1 : iþ1
W wdg that  tends to be as large as the structure allows from an
equilibrium perspective. In practice, this value will be the
The corresponding stabilising and overturning moments can highest one compatible with wall geometry. This issue is
now be calculated according to studied in more detail from the practical perspective in the
X accompanying paper (Alejano et al., 2012). Although Colas
i iþ1
M resist: ¼ W over x9w i þ 12 W iþ1 wtII þ W fi xfi et al. (2008) suggest optimising the value of this angle, they
do not explain clearly how to do so.
þ T aiþ1 wtII
iþ1
þ Ppiþ1 hpiþ1 þ W wdg
iþ1 iþ1
xwdg (23) iþ1
The obtained value wtII is the width of row i + 1 needed
X   to stabilise the wall against mode II overturning at the
M overt: ¼ Paiþ1 haiþ1 þ d iþ1 þ 12 U iþ1 wtII
iþ1
corresponding depth.
The angle , which tends to be constant along the height
Recalling that the ratio of these two expressions is the FoS, of the wall, does not hold a constant value in the lower part
iþ1 of the wall, owing to geometrical constraints. Therefore
a third-order equation in wtII is obtained
analysing the wall in the lower rows of blocks, when a row j
1  iþ1 3 is obtained in such a way that
6 ªb tan  wtII
!   Xn
X
i  2 j
d j ¼ wtII tan  . hi (26)
2 ªf h þ 12 ªb h ªw hwiþ1
1 i iþ1 iþ1
þ  12 FoS  wtII
i¼ j
i¼1
  iþ1 
þ  FoS  Paiþ1 tan  wtII
T aiþ1 the constant angle  will have a rotation point under the
" ! wall base – but this is not mechanically possible (Fig. 9).
Xi  i 2 Therefore, starting from this row and working downwards,
þ  2 ªf
1
h i i
wtII þ W over x9w i the angle  – now called now  j 9 – should be selected in
i¼1
such a way that the rotation point coincides with the external
#
base of the wall. Starting from this row, the angle  j 9 is
þ Ppiþ1 hpiþ1  FoS  Paiþ1 hpiþ1 ¼0 accordingly defined as
0X n 1
i
(24) h
 j9 ¼ tan1 @
i¼ j A
j1
(27)
wtII
This equation can be solved (Olver et al., 2010) to obtain
the positive value indicating the value for the width of
j1 j
blocks in row i + 1 that fulfils the stability criterion. Even if where wtII is chosen instead of wtII for the sake of
this equation has three roots, the realistic root can always be simplicity (otherwise a fourth-order equation would result),

P i⫹1
a

P aj
h i⫹1
pa

h jpa
 d i⫹1

dj

⬘

w i⫹1
tll

w jtll

Fig. 9. Features of mode II overturning, with â9 in lower part of wall

1020 ALEJANO, VEIGA, TABOADA and DÍEZ-FARTO
and because it does not vary the results significantly. This the wall; the height of the individual rows; the in situ and
ensures a rotation point quite close to the external base of saturated unit weights of the fill and the blocks; the friction
the wall. angle of the fill, between the fill and the wall, and between
blocks; a correct estimate of the angle ; and, finally, the
design FoS. Using this information, the sheet estimates safe
FINAL DRYSTONE WALL DESIGN widths against sliding and mode I and II overturning for
Guidelines on how to estimate the parameters needed for each row, and produces the final wall geometry. A calcula-
drystone wall design are provided in an accompanying paper tion sheet has also been designed to analyse the stability of
(Alejano et al., 2012). With wsiþ1 , wtIiþ1 , wtII
iþ1
and the width constructed walls. Entering the geometry of the wall yields
of the overlying row w i, the width of row i + 1 can now be the FoS against sliding and overturning for each row. In the
obtained as case of wall design, the SLIDE code (Rocscience, 2005) is
  used to analyse stability against circular failure under the
w iþ1 ¼ Max w i , wsiþ1 , wtIiþ1 , wtII
iþ1
,0 (28) wall. Foundation stability analysis is also accounted for in
the calculation sheet.
The largest of these values is selected for the final design
and then rounded up to a multiple of 0.5 m, since these are
the dimensions usually considered in practice. EXAMPLE OF WALL DESIGN
Once the drystone retaining wall geometry is defined in a For illustrative purposes, a 20 m high drystone masonry
way that ensures stability, it is useful to analyse the stability granite wall was designed. Its base was 2 m under the
of the wall against possible circular failure under its base, ground surface, where rock was moderately weathered. Row
and also to check the stability of the foundation. Even height was 0.5 m, and the estimated upper height of the
though these failure mechanisms are not very common, they groundwater table over the foundation was 6 m. The dry and
must still be contemplated in a design project. saturated unit weights of the fill were estimated as 1650 kg/
Circular failure under retaining walls has sometimes been m3 and 1900 kg/m3 respectively, and the unit weight of the
observed, especially in riverbanks. The most popular meth- rock was 2600 kg/m3 : The friction angle of the fill was
ods for analysing circular failure are those based on dividing measured as an average 33.68, and cohesion (not accounted
the sliding mass into slices (Bishop, 1955; Morgenstern & for in this analysis) was estimated as 20 kPa. The contact
Price, 1965; Spencer, 1967). These methods are implemen- friction angle between blocks and fill was estimated as 338
ted in codes such as SLIDE (Rocscience, 2005), which and the friction between rock blocks as 408. The minimum
facilitates their application to estimates of wall stability design FoS was 1.2, and  for this type of construction was
against sliding under the wall base. If a low FoS is observed estimated as 458. These data entered in the calculation sheet
(under a predefined design value, e.g. FoS ¼ 1.2), the wall produced the wall design reflected in Table 1.
should be increased in depth or width to ensure stability in The profile of the resulting wall is depicted in Fig. 10 for
the prescribed terms. 458 and for various values of , including 638, 37.58 and
All engineered structures apply a load to the underlying 22.58. As can be observed, an increase in the value of 
soil or rock that produces a deformation in the ground – dictates a greater volume of stone, even if the width of the
and therefore ground settlement – and, occasionally, foun- wall in the lower part is similar for all  values. The role of
dation failure. Foundations are not usually necessary under this and other parameters is studied further in the accom-
the kind of walls studied in the present authors’ research, panying paper (Alejano et al., 2012).
as they are usually built over more or less fresh rock. Since the width of each row is upgraded to a multiple of
Nonetheless, wall foundation stability has to be examined 0.5 m, the actual FoS values at each depth are increased in
in the light of ground carrying capacity, so as to avoid the final design. The calculation sheet can also compute
general and local failure, and to control settlement follow- these values. Table 2 shows the mode II FoS against over-
ing traditional approaches to foundation engineering (Codu- turning for the case where  ¼ 458.
to, 2001). The implementation of the analytical method in a calcula-
The entire analytical approach described above was pro- tion sheet enables sensitivity analyses to be easily performed,
grammed in a calculation sheet with the intention of making which is very useful in a discipline such as geotechnics,
drystone masonry wall design as easy and simple as possible where a certain degree of uncertainty can always be ex-
once the material parameters are known. Since results are pected. These analyses are not possible, or are extremely
obtained in virtually no time, the calculation sheet is very time-consuming, in numerical approaches (see, for instance,
useful for rapid analyses of the outcome of changes in input the sensitivity study of dry masonry retaining walls by
parameters. The input data are: wall height and underground Powrie et al., 2002), whereas they may take as little as 1 h
height; the position of the groundwater table at the back of in the present authors’ calculation sheet. This approach also

Table 1. 20 m high drystone masonry retaining wall design

Depth: m Row width: m Depth: m Row width: m Depth: m Row width: m Depth: m Row width: m

0.5 0.5 6.0 2 11.5 3.5 17.0 5.5

1.0 0.5 6.5 2 12.0 4 17.0 5.5
1.5 0.5 7.0 2. 5 12.5 4 18.0 5.5
2.0 1 7.5 2. 5 13.0 4 18.5 5.5
2.5 1 8.0 2. 5 13.5 4.5 19.0 5.5
3.0 1 8.5 3 14.0 4.5 19.5 5.5
3.5 1.5 9.0 3 14.5 4.5 20.0 5.5
4.0 1.5 9.5 3 15.0 5 20.5 5.5
4.5 1.5 10.0 3. 5 15.5 5 21.0 5.5
5.0 2 10.5 3. 5 16.0 5 21.5 5.5
5.5 2 11.0 3. 5 16.5 5.5 22.0 5.5
 STABILITY OF GRANITE DRYSTONE MASONRY RETAINING WALLS: I 1021
 ⫽ 63°  ⫽ 45°  ⫽ 33·75°  ⫽ 22·5° Depth: m
0

10

15

20

22

Volume ⫽ 84·5 m3 Volume ⫽ 77 m3 Volume ⫽ 72·25 m3 Volume ⫽ 68·25 m3

0 2 4 6m 0 2 4 6m 0 2 4 6m 0 2 4 6m

Fig. 10. Dimensions of 20 m high drystone masonry retaining wall for various values of parameter â and including rock volume
necessary per metre of wall

Table 2. Final factors of safety against mode II overturning for individual rows of designed wall

Depth: m FoS overturning, Depth: FoS overturning, Depth: FoS overturning, Depth: FoS overturning,
mode II m mode II m mode II m mode II

0.5 7.67 6.0 1.52 11.5 1.29 17.0 1.32

1.0 2.90 6.5 1.37 12.0 1.42 17.0 1.32
1.5 1.65 7.0 1.61 12.5 1.35 18.0 1.31
2.0 2.86 7.5 1.48 13.0 1.29 18.5 1.30
2.5 2.11 8.0 1.36 13.5 1.41 19.0 1.30
3.0 1.64 8.5 1.56 14.0 1.35 19.5 1.30
3.5 2.17 9.0 1.45 14.5 1.30 20.0 1.30
4.0 1.82 9.5 1.36 15.0 1.41 20.5 1.31
4.5 1.56 10.0 1.53 15.5 1.35 21.0 1.32
5.0 1.91 10.5 1.44 16.0 1.30 21.5 1.34
5.5 1.69 11.0 1.36 16.5 1.38 22.0 1.37

opens the door to statistical applications such as Monte between fill and rock blocks, 9. For mode I overturning 
Carlo simulation techniques and point-estimate methods. does not affect results (it is fixed and equal to zero), whereas
A sensitivity analysis was performed for the 20 m high 9 and 9 are both significant. Neither failure mode is affected
wall (22 m high, with 2 m underground) and the results by the friction angle of contacts between blocks. As for
depicted in the form of spider diagrams. A significant output stability against sliding, the most significant input are b and
parameter is represented along the vertical axis – in this 9. Note that when b is highly reduced, sliding may be the
case, the wall base width that ensures stability (Fig. 11), and dominant failure mechanism (Powrie et al., 2002).
the volume of stone needed to build a single metre of wall The importance of this analysis is that it tells us which
(Fig. 12). Percentages along the horizontal axis varied in parameters are most important for wall stability, as it high-
terms of the different significant input variables: the friction lights the properties ensuring the safety of the structure that
angle of the fill, 9; the friction angle between blocks, b ; need to be very carefully estimated. It is also useful in
the friction angle of the contact wall fill, 9; the dip of the guiding the selection of materials so as to create walls that
base plane  for mode II overturning; the unit weights of are more stable or less costly.
the fill, ª, and of the masonry, ªb ; and the position of the
groundwater table at the wall back, hw .
It can be seen that the most significant inputs for stability CONCLUSIONS
against mode II overturning are the friction angle of the fill, Inadequate dimensioning of the width of a drystone
9, the angle , and also the friction angle of the contact granite retaining wall can result in its failure. Conversely, a
1022 ALEJANO, VEIGA, TABOADA and DÍEZ-FARTO
Friction angle of fill, φ⬘ Friction angle of fill top, mode II, φ⬘
Friction angle of block, φb Friction angle of block top, mode II, φb
Friction angle of contact wall fill, δ⬘ Friction angle of contact wall top, mode II, δ⬘
Dip of base plane,  Dip of base plane top, mode II, 
Unit weight of fill, γ Unit weight of fill top, mode II, γ
Water head, hw Water head top, mode II, hw
Unit weight of block, γb Unit weight of block top, mode II, γb
7

6 Toppling (mode II)

5
Width of wall base: m

3
Sliding

0
50 60 70 80 90 100 110 120 130 140 150
% of change of variable

Fig. 11. Sensitivity analysis of dimensions of base of 20 m high drystone masonry retaining wall against sliding and
against mode II overturning in relation to the most significant input parameters (values corresponding to mode I
overturning are quite similar to those for mode II overturning and are not presented, in the interest of brevity)

Friction angle of fill, φ⬘ Friction angle of fill top, mode I, φ⬘ Friction angle of fill top, mode II, φ⬘
Friction angle of block, φb Friction angle of block top, mode I, φb Friction angle of block top, mode II, φb
Friction angle of contact wall fill, δ⬘ Friction angle of contact wall fill top, mode I, δ⬘ Friction angle of contact wall fill top, mode II, δ⬘
Dip of base plane,  Dip of base plane top, mode I,  Dip of base plane top, mode II, 
Unit weight of fill, γ Unit weight of fill top, mode I, γ Unit weight of fill top, mode II, γ
Water head, hw Water head top, mode I, hw Water head top, mode II, hw
Unit weight of block, γb Unit weight of block top, mode I, γb Unit weight of block top, mode II, γb
120

100
Stone volume per 1 m of wall: m3

Toppling (mode II)

80

60

Toppling (mode I)

40
Sliding

20

0
50 60 70 80 90 100 110 120 130 140 150
% of change of variable

Fig. 12. Sensitivity analysis of volume of stone needed to build a single metre of a 20 m high retaining drystone masonry wall
proof against sliding and against mode I and II overturning in relation to the most significant input parameters
 STABILITY OF GRANITE DRYSTONE MASONRY RETAINING WALLS: I 1023
suitable design procedure not only guarantees stability but where 9 is the effective friction angle of the fill, Ł is the battered
will avoid oversizing, thereby optimising the time and slope of the wall, 9 is the effective friction angle of the interface
money spent on construction and maintenance. Stability between fill and blocks, and i is the slope of the ground surface
analysis of drystone walls is marked by the occurrence of behind the wall. The obtained values of K9a do not significantly differ
contacts between blocks that act as weakness planes. from those proposed by Caquot & Kerisel (1948), a suitable
alternative for performing a more rigorous calculation.
An analytical LE-based technique was described for the The Coulomb theory is not usually accurate enough for the
design of drystone granite retaining walls as traditionally calculation of passive earth pressure, given the curved shape of the
built in north-west Spain. The focus was on the actual failure surface. So, in line with the work by Caquot & Kerisel
mechanisms that produce instability, namely sliding and two (1948), we adopt the approach described by Berry & Reid (1987).
types of overturning. Even though mode II overturning Since the derived equation to obtain the passive earth pressure
seems to be the major cause of drystone retaining wall coefficient is quite complex, results are presented in the form of
instability, existing design and analysis techniques do not Table 3 for the case of a logarithm spiral surface, considered to be
explicitly account for this mechanism. The LE-based ap- the critical shape.
proach balances simplicity and accuracy. It is also very The stresses can be analysed at a particular depth on the active or
internal side of the wall. It is necessary to know the depth of the
useful in providing a better understanding of the behaviour
analysis. Total depth at a row will be Óhi for hi as the height of a
of drystone granite retaining walls, and so improves design stone row. The height of the water head at the point of analysis is
techniques. called hwi , and the difference between both – called the dry depth –
Since the LE-based technique implies a large number of is hdi ¼ Óh i  hwi : Unit weights for the dry soil (ª) and the saturated
calculations, it was implemented in a calculation sheet. It is soil (ªsat ) are also necessary parameters. Remember that the
therefore particularly suitable for industry. submerged unit weight of the soil can be computed as
Data inputs in the user-friendly sheet in the form of a ª9 ¼ ªsat  ªw , where ªw is the unit weight of water. For the sake
series of parameters referring to the wall materials and of simplicity, we assume hydrostatic pore pressure conditions.
geometry produce suitable dimensions for the design of a With this information, the pore pressure and the total and effective
stable wall. The calculation sheet is also useful for analysing vertical stresses can be computed at the depth of row i as
changed circumstances affecting the wall (water, weathering,  v ¼ ªhdi þ ªsat hwi
etc.). Furthermore, better knowledge of the design input
parameters is ultimately paramount in ensuring wall stability. u ¼ ªw hwi
(30)
 v9 ¼  v  u ¼ ªhdi þ ªsat hwi  ªw hwi

¼ ªhdi þ ª9hwi
ACKNOWLEDGEMENTS
The authors thank the Autonomous Government of Galicia From this the corresponding horizontal stresses can be derived, as
for funding under contract reference number 08TMT010E  h9a ¼ K9a ªhdi þ K9a ª9hwi
for the research project entitled ‘Calculation method for (31)
analysing drystone masonry walls for retaining purposes’.  ha ¼  h9a þ u ¼ K9a ªhdi þ K9a ª9hwi þ ªw hwi
Ailish M. J. Maher provided assistance with English usage
in a version of the manuscript. Finally, the total and the effective active force on the wall can be
obtained by integrating the total and effective horizontal pressure
along the wall depth, Ha , following the approach of Das (2002) (see
Fig. 13). In this way we obtain
APPENDIX: EARTH PRESSURES AND ACTIVE AND ð Ha
PASSIVE FORCES ON WALL Pa ¼  h dh
0
The Coulomb (1776) earth pressure theory was selected to
 2  2
calculate the active earth pressures (Berry & Reid, 1987). Coulomb ¼ K9a ª hdi þ K9a ªhdi hwi þ 12 ð K9a ª9 þ ªw Þ hwi
1
2
assumed the failure surface to be a plane. Observations of fallen (32)
ð Ha
walls and numerical models with UDEC suggest that this assumption
is sufficiently close to reality (Harkness et al., 2000). For the sake of P9a ¼  h9 dh
0
simplicity, and with the idea of proposing a reasonably simple design
 2  2
method, the Coulomb approach for frictional soils (cohesion is ¼ K9a ª hdi þ K9a ªhdi hwi þ 12 K9a ª9 hwi
1
2
disregarded) is taken to estimate active forces for the different
heights of the walls to be analysed. The line of action of Pa is not relevant for an analysis of sliding, but
The active pressure coefficient is taken as certainly is relevant for an analysis of stability against overturning.
cos2 ð9  ŁÞ To estimate the position of this line, the addition of the moments
K9a ¼ 0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi12 exerted by every acting force (the sum of products of every acting
force multiplied by the lever arm) has to equal the moment exerted
@cos Ł þ sin j9 ð1 þ tan 9= tan 9Þð1  tan i= tan 9ÞA by the resulting force (sum of forces). We can thus obtain the
ð1  tan Ł tan 9Þð1 þ tan Ł tan iÞ
position of the lever arm or height of the line of action of the active
(29) force at the point of calculation as

Table 3. Critical values for passive earth pressure coefficient derived from considering logarithm spiral sliding surface (following Berry
& Reid, 1987)

9/9 K9p

9 ¼ 58 9 ¼ 108 9 ¼ 158 9 ¼ 208 9 ¼ 258 9 ¼ 308 9 ¼ 358 9 ¼ 408

0 1.19 1.42 1.70 2.04 2.46 3.00 3.69 4.60

0.25 1.22 1.48 1.82 2.26 2.84 3.62 4.70 6.26
0.5 1.23 1.53 1.92 2.43 3.13 4.13 5.58 7.80
0.75 1.24 1.56 1.98 2.55 3.36 4.52 6.31 9.18
1024 ALEJANO, VEIGA, TABOADA and DÍEZ-FARTO

Unit weight of soil ⫽ γ

Σhi ⫺ hiw Σhi ⫺ hiw

hi
i
Σh

i i
K⬘a γ(Σh ⫺ hw)

γsat
hiw hiw
σ⬘h u hiw σh

hi

K⬘aγ(Σh i ⫺ hiw) γ whiw

i i i i
K⬘a[γ(Σh ⫺ hw) ⫹ γ⬘hw)] ⫹ γ whw

K⬘a[γ(Σhi ⫺ hiw) ⫹ γ⬘hiw)]

Fig. 13. Active pressure distribution on retaining wall with partially submerged soil (adapted from Das, 2002)

h   2 i 1 i   1 i 
K9a ª hdi dj descent of the rotating point for row j due to the
3 hd þ hw þ K9a ªhd hw
1 i i i
2 2 hw
h  i 2 i1 i  occurrence of toppling mode II at the lower rows
þ 2 ð K9a ª9 þ ªw Þ hd
1
3 hw
ÓF sum of forces
ha ¼  i 2  i 2 (33) Ha wall depth at active side
2 K9a ª hd þ K9a ªhd h w þ 2 ð K9a ª9 þ ªw Þ hw
1 i i 1
H p wall depth at passive side
h1 height of the stone row 1
On the active side, and due to pushing by the fill, a slight outward
h i height of the stone row i
displacement of the wall may be expected, in such a way that a iþ1
h height of the stone row i + 1
downward shear force, Ta, occurs in the fill in contact with the wall.
ha lever arm or height of the line of action of the active
This contributes to wall stabilisation, and can be estimated as
force
T a ¼ P9a tan 9 (34) h1a position of the line of action of the active force at
row 1
where 9 is the friction angle between the wall and the fill, which haiþ1 position of the line of action of the active force at
tends to be a value close to the friction angle of the fill. row i + 1
In similar terms, the passive forces in the wall face and the hdi dry depth at row i
distance between the rotation line and the line of action are
estimated as hpiþ1 position of the line of action of the passive force at
ð Hp row i + 1
 2 hsi location under the ground surface in front of the
Pp ¼  h dh ¼ 12 K9p ªsat hsi
0
wall
ð Hp hw position of the groundwater table at the wall back
 2 (35) h1w height of the water head at row 1
P9p ¼  h9 dh ¼ 12 K9p ª9 hsi
0 hwi height of the water head at row i
iþ1
hw height of the water head at row i + 1
hp ¼ 13 hsi i slope of ground surface behind the wall
where Hp is the depth of the analysis, and hsi is the location under K9a active earth pressure coefficient
the ground surface in front of the wall, assumed to be fully saturated K9p passive earth pressure coefficient
for the sake of simplicity and conservativeness. Mresist: resisting moments
On the passive or external side, owing to pushing by the wall, a Movert: overturning or toppling moments
slight inward displacement of the existing soil or ground takes place, N normal reaction force
in such a way that an upward shear force, Tp, occurs in the ground in N1 normal reaction at row 1
contact with the wall. This can be calculated as N iþ1 normal reaction at row i + 1
Pa total active force on the wall
T p ¼ P9p tan 9 (36) P9a effective active force
The shear force between blocks, Tb, can be estimated as P1a active force acting in row 1
iþ1
Pa active force acting in row i + 1
T b ¼ N tanðb Þ (37) P p total passive force
where b is the friction angle between the blocks, and N is the P9p effective passive force
normal force applied to the contact surface. This shear strength is Ppiþ1 passive force acting in row i + 1
also that which operates against sliding through the block contacts. Ta active shear force
T 1a active shear force at row 1
T aiþ1 active shear force at row i + 1
NOTATION Tb shear force between blocks
c1 , c2 , c3 , c4 constants for the formulation of the FoS as a T b1 shear force between blocks at row 1
function of  for toppling mode II T bi shear force between blocks at row i
iþ1
d iþ1 descent of the rotating point for row i + 1 due to the Tp passive shear force at row i + 1
occurrence of toppling mode II Tavailable shear force available at a surface contact
 STABILITY OF GRANITE DRYSTONE MASONRY RETAINING WALLS: I 1025
Tmobilised shear force required to mobilise a contact Berry, P. L. & Reid, D. (1987). An introduction to soil mechanics.
iþ1
T mobilised shear force required to mobilise a contact at row London, UK: McGraw-Hill.
i+1 Bishop, A. W. (1955). The use of the slip circle in the stability
Tp passive shear force analysis of slopes. Géotechnique 5, No. 1, 7–17, http://dx.doi.
U 1 water force acting on the base of the row 1 org/10.1680/geot.1955.5.1.7.
U iþ1 water force acting on the base of the row i + 1 Burgoyne, J. (1853). Revetments of retaining walls. Corps R. Engng
u water pressure Papers 3, 154–159.
W 1 weight of the block needed to stabilise row 1 Caquot, A. & Kerisel, J. (1948). Tables for the calculation of
W i weight of the block needed to stabilise row i passive pressure, active pressure, and bearing capacity of
W iþ1 weight of the block needed to stabilise row i + 1 foundations. Paris, France: Gauthier-Villars.
W fiþ1 weight of fill located over the step between row i and Claxton, M., Hart, R. A., McCombie, P. F. & Walker, P. J. (2005).
i+1 Rigid block distinct-element modeling of dry-stone retaining
W iover weight of blocks and fill acting over the block of row walls in plane strain. J. Geotech. Geoenviron. Engng ASCE 131,
i No. 3, 381–389.
iþ1
W over weight of blocks and fill acting over the block of row Coduto, D. P. (2001). Foundation design: Principles and practices,
i+1 2nd edn. Upper Saddle River, NJ, USA: Prentice Hall.
iþ1
W wdg weight of the wedge in toppling mode II acting on Colas, A. S., Morel, J. C. & Garnier, D. (2008). Yield design of
row i + 1 dry-stone masonry retaining structures: comparisons with analy-
w 1 width of row 1 tical, numerical, and experimental data. Int. J. Numer. Anal.
w iþ1 width of row i + 1 Methods Geomech. 32, No. 14, 1817–1832.
w1s width of the row 1ensuring stability against sliding Cooper, M. R. (1986). Deflections and failure modes in dry-stone
wsiþ1 width of the row i + 1 ensuring stability against retaining walls. Ground Engng 19, No. 8, 28–33.
sliding Coulomb, C. A. (1776). Essai sur une application des régles de
w1tI width of the block needed to stabilise row 1 against maximis et minimis a quelques problèmes de statique, relatifs à
toppling (mode I) l’architecture. Mem. l’Acad. R. Sciences, Paris 3, 38 (in French).
wtIiþ1 width of the block needed to stabilise row i + 1 Das, B. M. (2002). Principles of geotechnical engineering. Pacific
against toppling (mode I) Grove, CA, USA: Brooks Cole/Thomson Learning.
iþ1
wtII width of the block needed to stabilise row i + 1 de Buhan, P. & de Felice, G. (1997). A homogenization approach to
against toppling (mode II) the ultimate strength of brick masonry. J. Mech. Phys. Solids 45,
wtIIjþ1
width of the block needed to stabilize row j against No. 7, 1085–1104.
toppling (mode II) in the lower part of the wall Department of Transport (1987). The assessment of highway bridges
x9w i perpendicular distance between action line of Woveri and structures: bridge census and sample survey. London, UK:
and the rotation line on row i DoT.
x9w iþ1 perpendicular distance between action line of Wover iþ1 Dickens, J. G. & Walker, P. J. (1996). Use of distinct element
and the rotation line on row i + 1 model to simulate behaviour of dry-stone walls. Struct. Engng
xfi perpendicular distance between action line of Wfi and Rev. 8, No. 2–3, 187–199.
the rotation line Harkness, R. M., Powrie, W., Zhang, X., Brady, K. C. & O’Reilly,
i
xwdg perpendicular distance between action line of Wwdg i M. P. (2000). Numerical modelling of full-scale tests on
and the rotation line drystone masonry retaining walls. Géotechnique 50, No. 2,
 angle with the horizontal line formed by the base of 165–179, http://dx.doi.org/10.1680/geot.2000.50.2.165.
overturning piece of wall in toppling mode II Itasca (2004). UDEC code: Vers. 4.1. Minneapolis, MN, USA:
 j 9 angle with the horizontal line formed by the base of ITASCA Consulting Group.
overturning piece of wall in toppling mode II at the Morgenstern, N. R. & Price, V. E. (1965). The analysis of the
lower rows when a standard value of  cannot be stability of general slip surfaces. Géotechnique 15, No. 1,
used 79–93, http://dx.doi.org/10.1680/geot.1965.15.1.79.
ª unit weight of the dry soil Mundell, C., McCombie, P., Bailey, C., Heath, A. & Walker, P. J.
ª9 submerged unit weight of soil (2009). Limit-equilibrium assessment of drystone retaining
ª b block unit weight structures. Proc. Inst. Civ. Engrs Geotech. Engng 162, No. 4,
ª sat unit weight of saturated soil 203–212.
ª w unit weight of water Olver, F. W. J., Lozier, D. W., Boisvert, R. F. & Clark, C. W. (eds).
9 effective friction angle of the interface between fill (2010). NIST handbook of mathematical functions. Cambridge,
and blocks UK: Cambridge University Press.
Ł battered slope of the wall O’Reilly, M. P., Bush, D. I., Brady, K. C. & Powrie, T. R. L.
 h total horizontal stress (1999). The stability of drystone retaining walls on highways.
9h effective horizontal stress Proc. Inst. Civ. Engrs Municip. Engr 133, No. 2, 101–107.
 ha total active horizontal stress Pande, G. N., Liang, J. X. & Middleton, J. (1989). Equivalent
9ha effective active horizontal stress elastic moduli for brick masonry. Comput. Geotech. 8, No. 3,
v total vertical stress 243–365.
9 v effective vertical stress Powrie, W., Harkness, R. M., Zhang, X. & Bush, D. I. (2002).
9 effective friction angle of the fill Deformation and failure modes of drystone retaining walls.
 b friction angle between blocks Géotechnique 52, No. 6, 435–446, http://dx.doi.org/10.1680/
geot.2002.52.6.435.
Rocscience (2005). SLIDE. Toronto, Canada: Rocscience Inc.
Spencer, E. (1967). A method of analysis of the stability of
REFERENCES embankments assuming parallel interslice forces. Géotechnique
Alejano, L. R., Veiga, M., Gómez-Márquez, I. & Taboada, J. 17, No. 1, 11–26, http://dx.doi.org/10.1680/geot.1967.17.1.11.
(2012). Stability of granite drystone masonry retaining walls: II. Wong, H. N. & Ho, K. K. S. (1997). The 23 July 1994 landslide at
Relevant parameters and analytical and numerical studies of real Kwun Lung Lau, Hong Kong. Can. Geotech. J. 34, No. 6, 825–
walls. Géotechnique, http://dx.doi.org/10.1680/geot.10.P.113. 840.

View publication stats