11/33

Efficiency of Horizontal Drains on Slope Stability

L'Efficacite des Drains Horizontaux dans la Stabilite des Talus

E. NONVEILLER Professor of Foundation Engineering, Faculty of Civil Engineering, Zagreb, Yugoslavia

S Y N O P S I S . T h e h y d r o d y n a m i c pressure field of spatial flow toward horizontal tubular drains installed in unstable slopes
has been c o m p u t e d by m e a n s of a numerical finite difference formulation of the Laplace equation. T h e pore pressure
field induced by the change f r o m planar h y d r o d y n a m i c pressure without drains to spatial pressures obtained after intro­
ducing the drains is c omputed. Pore pressure dissipation with time depends on compressibility, permeability of slope
material, slope geom e t r y , length and spacing of drains. It is c o m p u t e d by finite difference approximation of the three
dimensional Fourrier differential equation. T h e solution is used for parametric presentation of influence of c o m p r e s ­
sibility and permeability of slope material, slope geom e t r y , drain length and spacing on the delay of improving slope
stability. Results presented m a y be used to evaluate the appropriate drain g e o m e t r y in order to assure a satisfactory
safety factor of the slope in desired time after installation of drains.

INTRODUCTION
Drilled horizontal drains have been used, for m o r e than
40 years to stabilise sliding earth slopes (Stanton,1948).
In Yugoslavia they w e r e used for the first time in 1966 on
a slide in the valley of the river Rjecina. Later a slide on
the outskrits of Z a g r e b w a s stabilised with drilled hori­
zontal drains (Nonveiller, 1970). The length and the s p a ­
cing of the drains in these applications w e r e inferred ini-
tuitively f r o m consideration of the geologic and hydraulic
conditions of the slope. The influence of the drains on the
m e a n h y d r o d y n a m i c pressure w a s conjectured in stability
analyses f r o m planar flow nets, since there w e r e no p r a c ­
tical solutions of the three dimensional h y d r o d y n a m i c
ground water flow in the drained slope. Solu;ions of spati­
al flow have b e c o m e practical with numerical analyses.
S o m e typical cases w e r e solved with a c o m p u t e r p r o g r a m
(Nonveiller, Tadic, 1978). K e n n e y et al. (1977) have publi­
shed an experimental solution for two typical boundary
conditions of 3:1 slopes.

Th e time needed to activate tubular drains is of practical

concern for their application to slope stabilization in c o m ­
pressible soil of low permeability. T h e difference b e twe­
en the h y d r o d y n a m i c pressures of the planar in the u n d r a ­
ined slope and the spatial groundwater flow in the d o m a i n Q
(FIG. l) around a drain produces a transient pore p r e s ­
Fig 1 Idealised slope section with drain and boundaries
sure field, w hich gradually dissipates. H e n c e the full ef­
ficiency of the drains is delayed depending on the length
T h e solution is established with the scalar field H(x,y,z)
of drainage paths and the consolidation coefficient c of
in S2 vhich satisfies the Laplace differential equation
the soil.
A H (x,y,z) = 0 in n, ( l)

SPATIAL S E E P A G E F L O W T O W A R D A D R A I N IN T H E SLOPE and the boundary conditions.

3
Th e h y d r o d y n a m i c pressure field in the d o m a i n ficR T h e boundaries of fi are r = r u r ^,
(Fig l) for the undrained and the drained condition is
c o m p u t e d by m e a n s of a numerical approximation of with r = S u S o S o L, and r = B u B uS
1 1 Z J ' Z 1 “
the Laplace equation in the nodal points of a discretised
field (x, y, z ) .

495
11/33

as s h o w n on Fig. 1. O n the boundaries S^, S^, and L T h e h o m o g e n e o u s Laplace equation (1) is approximeted
in any node of the region a by the difference equation:
the potential H (x,y,z) is given in the f orm:

H / r j = ♦, (2 )
h~2 (H - 2H . . . + H . , .,) + h ~ 2 (H . . , -
x 1 i+ l,jk ljk l- l,jk y i,j+ l,k
w h e r e ♦is a k n o w n function, given in e very point (x,y,z)
on r .
- 2H H . . , . ) + h (K . . .
ljk i,j- l,k z ij,k + l
In our climatic conditions the surface S 2 of a slope of m o ­
derate or low permeability can be regarded as the m a x i ­
- 2 H . .. + H . . . .) = 0 (5)
m u m ground water surface, and stationary percolation in ljk ij,k - l
the slope can be a s s u m e d .
S4 is a n impervious base and and are boundary pla­ 1 < i <1-1, 1 < j <J - 1, 1 <k<K-l.
in w i c h H. . is the approximation of the function H(x,y,
nes with s y m m e t r i c reflection. O n these planes the b o u n ­ 'iJ**
z) in the node (ih , ih , kh ).
d ary condition x y’ z

T h e boundary conditions are:

(3 )
- o n Sj and the potential is constant at the ground w a ­
shall b e satisfied.
ter leved,
T h e s y s t e m of equations (1) and the boundary conditions
- on the potential is equal to the elevation of the node,
(2) and (3) present a m i x e d boundary p r o b l e m of the h o ­
m o g e n o u s Laplace equation. Generally it can not be sol­ - on the drain L the potential is equal to the elevation of
ved analytically, thus a convenient numerical m e t h o d m u s t its exit o n the slope and w e write:
be adopted. In our case the finite difference m e t h o d with
iterative solution of the s y s t e m of equations w a s used H = *(Ph ) f o r P h £ r / h> (6 )
(Krylov, 1972).
T h e region 0 is devided by m e a n s of a net of planes paral­ O n the boundaries u “2 u 4 discre­
lel to the coordinate planes at equidistant intervals h^, tised boundary conditions are expressed by m e a n s of the
h ,h (Fig. 2). T h e intersections of the planes define a function H in two adjacent s y m m e t r i c reflected layers.
y ^ On: ljk
net of nodes, the set
^ h) * (ihx , jhy , khz ) , lci<I-l, l<j<J-l, l<k*K-l, (4)

B (. h> : H - H . - 0
I ,J,K is the n u m b e r of intervals in the direction of x,y,z 1 ilk 1, -1 , k

f r o m the origin to any node. T h e discretised m o d e l of

( 7)
B 2 h> Hi,J + l ,k - Hi,J - l,k = °
has N internal nodes. Beside these there are sets of nodes
,
on the boun d a r y planes c (h) c (h) c n
as well „(h) : H ... - H . . . = 0.
s'4 ijl 1,J,~1

Inserting Eq. (7) in (5) the fictitious layers of the net are
ij,k+l (2)
eliminated.
T h e s y s t e m of equations (5), (6), (7) contains only the
u n k n o w n values of the function H . i n the N internal nodes.
ijk
T h e resulting s y s t e m of linear algebraic equations is sol­
ve d iteratively. F r o m the k n o w n values of the function in
the p ^ 1 iteration step the values in the (pfl)^1 step are
obtained f rom:

H„ P +J
ijk C{ h ~ x (HPi +l,jk + * 1 - 1 ,3 k> + \ 2 (lf,i,j+ l,k +

Fig 2 Designation of nodes *i,j-

■l1 , k* > +hl2<*ij,k+l +HPij,k-l>}' (8>
as on B B ^ . T h e set of boundary nodes S ^ is
1 2 ^
defined a p p r o x i m a t i v e ^ with nodes at m i n i m u m distance -2
with C = [2 { h + h * + h f
L x y
f r o m the surface S^.

496
 11/33

T h e iterative computation is continued until the criterion: pressure dissipates. T h e process of consolidation is g o ­
verned by the diffusion differential equation:

z
2
**■
i j k1 -
(9) 3 U 32U 32U 3£7
ABC
ijk , + C — - + C ( 12 )
3x 2 y 3 y 2 3z 3t
e a specified small n u m b e r , is satisfied in two subse­

quent iterations in selected nodes A , B , C . T h e n u m b e r in w hich C ( x , y , z ) are consolidation coefficients

of n e ce ssary iteration steps is reduced if initial values
k M
a a
of the function are input, established a p p r o x i m a t i v e ^ a = (x ,y,z) (13)

b y interpolation f r o m the boun d a r y values given by eq.(6).

- k(a) D a r c y koefficient in direction (a)
For the stability computation of a planar slice of the slope
the average of the potential H(x, y , z ) along the y axis is - M ( a ) m o d u l u s of compressibility in (a)
c o m p u t e d f r o m the approximation: A numerical finite difference solution is developed for the
solution of eq. (12) for the discretised region 12 . T h e fi­
nite difference equivalent of eq. (12) is (with the notation
H .. = ( 1 / J l l „ H. for (ik} on B (1 0 ) of the nodes in parenteses o n Fig 2):
ik j=0 ijk

T h e algorithm is p r o g r a m m e d in Fortran and n a m e d T R O - 2 <U1 2U0> + J (U 5 + U6 - 2V 0 ) +

FIL. T h e values of the potential head in the nodes are c o m ­ <hx> % >
puted f r o m eq. (8), the m e a n values f r o m eq. (10). A
plotter subroutine is included for presentation of the equ- (14 )
1 (U 2 2V - ~ ' V a t " V
ipotentials f r o m eq. (10). ' V
C At
a
P r o b l e m s with one or several drains at different elevati­ With M =
2 '
(xyz),
ons can be solved. T h e case of a drainage trench in the 'V
plane is approximated by tubular drains of appropriate
the pore pressure in the central node at t i m e t+ At is
lenght at every z elevation in the B^ plane.
f r o m eq. ( 14):

C O N S O L I D A T I O N O F I N D U C E D PORE PRESSURE
U0 , t + L t \ (V 1 + U3> V U5 + V + V U2 + V

T h e initial pore pressure field due to the introduction of 2 (M + M + M - 0,5)U„ (1 5)

x y z 0
a drain is c o m p u t e d f r o m

For stability of the solution it is required that m a x M <

< 1/6.
U( xyz) y
w
{ H( x y z)
o
- H (x y z )
a
} a n
The solution m u s t satisfy the boundary conditions:
in w hich index (o) refers to undrained and (d) to drained - for nodes on Sj, S ^ , S ^ , L : „ = 0, U
t=0 ’
condition (Fig. 3).
- for nodes on B , B^ around which the pore pressures
are s y m m e t r i c respective to the y axis: U = U ,
T h e respective effective stress change can develop only 5 o
after the pore v o l u m e has changed accordingly by expel- - for nodes on S^ the pore pressures are s y m m e t r i c to
the z axis: = U .

Th e algorithm is p r o g r a m m e d in Fortran and n a m e d

S P A C O N . A Subroutine c o m p u t e s the pore pressure f r o m
the TROF I L output and links the two p r o g r a m s for a u t o m a ­
tic processing.
Th e output of S P A C O N is linked by a Subroutine to the
Stability p r o g r a m S S T A B - 1 (Wright, 1974) and the sta­
bility for selected slip surfaces at prescribed times are
c o m p u t e d and plotted in a graph. T h e ordinate s h o w s the
normalised stability gain

Fig 3 H y d r o d y n a m i c head in node before and after ( 16 )

draining, initial pore pressure head

ling water toward the boundaries. W h e n the drain is in­ in which subscripts 0,t and d denote the stability factor
at time 0,t and infinite resp. (Fig. 7).
stalled pore pressure of magnitude U will develop and
the effective stress shall increase gradually as the pore

32-017129 497
11/33

T h e abscissa s h o w s n o r malised time: s = S/H. - drain spacing over input hydraulic head
difference
9,= t-C -h 7 2 (17)
' v 1 1^ = L/S - drain length over spacing
in which t is the time, C the consolidation coefficient, 1^ = L/H. - drain length over input hydraulic head dif­
v ’
ference.
H. the input head difference.
T h e influence of increasing drain length and reduced s p a ­
O n Fig. 8 the results are s h o w n with the parameters: cing is clear f r o m the d i a g r a m s on Fig. 6. T h e specific
drain length 1 needed to achieve a required safety fac­
F=F :F , 0 = t-c .L-/H.-4”2 (18) tor increase 1^ decreases with longer drains at larger spa­
t o * v ' 1 ^
cing. Longer drains are needed to achieve the s a m e d e ­
gree of stability increase in the flatter slope.
INFLUENCE OF LENGTH A N D SPACING OF TUBULAR "------ ^2?—f
D R A I N S O N S L O P E STABILITY

Pervious slope

With the T R O F I L p r o g r a m a parametric study of the influ­

ence of drain spacing and length on the stability of regu­
lar slopes of fairly pervions soil w a s carried out. T h e
average of the h y d r o d y n a m i c pressures in the d o m a i n
between two adjacent drains (Eq. 12) w a s c o m p u t e d for
several circular slip surfaces on slopes 3:1 and 2:1
with drains 50, 75, 100, 150 m long, spaced at 20, 30,
40, 60, 100 m . T h e safety factor F for planar seepage
without drain and F for spatial seepage with a tubular
drain, w e r e c o m p u t e d using Bishops m e t h o d without
interslice forces. The soil strength w a s a s s u m e d for
1:2 slope c'= 2,6 kPa, 0'= 29°, slope 1:3 c'= 2,2 kPa,
0 = 22°, bulk density Y w= 20 k N / m 3 .

1 0 _____________ 120_______________ , 20
Fig 5 A v e r a g e pressure head, drained slope 1:2,
drains on two and on 10 levels
T h e degree of stability increase achived with two and
with 10 shorter drains in a plane is s h o w n on Fig. (6a).
It is seen that with one long drain at the toe the safety
factor is increased m u c h m o r e than with several drains
of the s a m e total length placed at higher levels.

H 03 S (m )— - ^ S(m) —

Fig 4 A v e r a g e pressure head, drained slope 1:2,

L = 100 m , S = 20 and 100 m

T h e resulting equipotentials for s o m e of the cases are

s h o w n on Fig 4. T h e results s h o w n are similar to those
obtained by K e n n e y et al. (1977) on physical models. O n
Fig. 5 the equipotentials of the seepage with two 50 m
long drains spaced at 30 m , and with 10 drains each 15 m
long spaced at 20 m are s h own. T h e results of this inve­ Fig 6 Parametric presentation of drain efficiency,
stigation are s h o w n o n Fig. 6 with normalised parameters: slope 1:2 and 1:3

F = F^/F q - specific safety factor increase,

498
 11/33

T e n 15 m long drains m o d e l a planar drain trench 7,5 m Table 1.

deep, the efficiency of which is obviously low and the cost Range of values k, M ^ , c^
high c o m p a r e d to tubular drains in the slope toe.

2 2
Slope of low permability S o i l type k m/s M kPa/m C m /s
V
T h e time needed to achieve the full efficiency of tubular Sand 10-1- io ~ 4 io 4 - i o 5 io ~ 2 - 10~3
drains in a slope of low permeability w a s studied for the
3 4
1:2 slope with 50,75 and 100 m long tubular drains on 10, Clay CL 10~8- 1o ' 9 5 .1 0 - 5.10 5. 1 0 ~ 6 - 5 . 1 0 ~ ?
20, 60 and 100 m spacing. T h e soil p a r a m e t e r s are as
Clay CH i o - 9- i o - 10 i o 4 - io 5 10-6 - 1o ~7
-3 2
before, and C = 1 0 m /s.
v

12,U=50m,Hj=50m, cv-IO~3mter
■F* 1,20 o
1,15 ^

a s.»*t
,
20
SJ

Fig 7 Stability gain vs normalised time 9_ for 50,75 and 100 m long drains

T h r e e d i a g r a m s on Fig 7 s h o w the results of the c o m p u t a ­ Experience s h o w s that a safety factor increase of 1,2 is
tion in dimensionless p a r a m e t e r s according to Eqs. (16) needed for lasting stabilization of large slides, s o m e ini­
and (17). tial effect is achieved with F = 1,1. Regarding the time
needed for completion of drainage w o r k s between one and
six months, it can be a s s u m e d that within that period of
Fig 8 s h o w s the results in p a r a m e t e r s according to Eq.
time a safety factor increase to 1,2 F( should be satisfa-
(18). In this presentation the time p a r a m e t e r 0 of all
ctory.
c o m p u t e d cases lays in a n a r r o w range. For informative F r o m Fig g w e have 9 = 16.10 (m ) an for 1,2 F
checks of the consolidation time, with consolidation coef­ -3 °
and 0^ = 3,7-10 for 1,1 F • F r o m :
ficient, drain length and spacing other than c o mputed,
the approximated time p a r a m e t e r 0 can be used.
cv = (H.- S ) 2 - (L • t)"1 . 0 2 ,
6
w e have with t = 2 ,63*10 s and t = 15,8-10 s for one and
for six m o n t h s resp. the following results for a 1:2 slope
with H. = 50 m , L = 100 m , S = 10 m :
l ’ ’
Table 2
2
C m /s for
-■* * pi ^7 ;1.6 F = 1 ,2 0 F = 1 ,1 0
iff1 Iff4 Iff1 to1 Iff' s-iff'd -5 -6
One 1 ,5 .1 0 3 ,5 .1 0
Fig 8 Safety factor increase F vs time p a r a m e t e r 0^ Six
„ „
2 ,5 .1 0
. -6
6 ,0 .1 0
- 7

for 1:2 slopes

F r o m tables 1 and 2 it is clear that within o n e m o n t h satis­

The efficiency of the stabilization of h o m o g e n e o u s slopes
factory stabilisation is achieved in sandy or silty material
with tubular drains depends on:
and within six m o n t h s in clay.
- the size of the slope expressed b y H . ,
Conslusions
- the disposition of drains, expressed b y L and S,
With the algorithm presented before the effectivenes of tu­
- the consolidation coefficient c .
v bular drains in h o m o g e n e o n s slopes can be studied and
the convenient length and spacing of the drains can be
In order to estimate the efficiency of draining clay slo­
established. It is not difficult to adapt the algorithm to
pes the range of c^ values for typical overconsolidated
include stratified slopes and non isotropic permeability.
soils s h o w n on Table 1 shall be considered.
Th e results s h o w that the t i m e needed to achieve an a d e ­
quate safety gain increases with decreasing consolidation

499
11/33

coefficient. Thus longer drains and a closer spacing

needed in clay slides of low permeability and hi.gh c o m ­
pressibility, in order to reduce the time w h e n the safety
gain is sufficient to stabilise the slide.
T h e experience with tubular drains in clay slopes gained
b y the author s h o w s encouraging results (Nonveiller,
1970). Subhorizontal stratification with m o r e pervious
interbeds greatly increase the drain efficiency.

M o s t slides occur in highly overconsolidated clay with a

ligher consolidation coefficient. Fissuring contributes to
increase the overall permeability. Such favourable g e o ­
logic features should be dully considered in the design
of any tubular drain installation.

ACKNOWLEGEMENTS
T h e author is indebted to J. Tadic, Eng. M a t h , w h o p r e ­
pared the Trofil p r o g r a m , V.Ceric, M r . Sc. Phys. w h o
prepared the Spacon p r o g r a m , to Z.Balija and V . S u n a r a
w h o prep a r e d the authomatic link of the p r o g r a m s and the
plotter subroutines, and to N.Petrovic, Civ.Eng. w h o
w o r k e d out the e x a m p l e s in the computing Centre of the
Civil Eng.Fac. Univ. of Zagreb. T h e w o r k w a s partly
financed b y the Scientific fund of Croatia and by funds
of the Civ.Eng.Facuity, Univ. of Zagr e b , w hich is
gratefully a c k n o w l e d g e d .

REFERENCES

K e n n e y T . C . , P a z i n M . , Choi W . C . , 1977: D e sign of

horizontal drains for soil slopes, Journal
of the Geotechnical Engineering Division,
A S C E , Vol. 103, N o G T 11, pp. 1311 - 1323

Krylov C.I., Bobkov, V . V . , Monastirnyj, P.I., 1972: V y -

cislitelnye m e t o d y vysej matematiki,
Vysesjsaja skola, M i n s k
Nonveiller E. 1970: Sanierung v o n Hangrutschungen
mittels Horizontalbohrungen, European
Civil Engineering, Bratislava - Praha -
Wien, N o 5, p. 221-228
Nonveiller E . , Tadic J . , 1978: Izbor duljine i r a z m a k a
cijevnih drenova za stabiliziranje kosina,
Proc. XlVth Cong r . Yugoslav S S M F E ,
Sarajevo, Vol. II, p. 313-321

Stanton, T.E., 1948: California experience in stabilizing

earth slopes through the installation of hori­
zontal drains by the hydrauger method,
Proc. Ilnd ICSMf'E, Rotterdam, Vol.III.,
p. 256-260
Wright S . G., 1974: S S T A B - 1 Slope stability p r o g r a m ,
(installed at the C o m p u t i n g Centre of the
Fac. of Civil Eng. Sciences, Zagreb, by
Prof. R.Schiffman in 1977).

500