You are on page 1of 13

Mining Science and Technology, 13 (1991) 3 0 1 - 3 1 3

301

Elsevier Science Publishers B.V., A m s t e r d a m

Sill mat evaluation using centrifuge models


R.J. Mitchell
Queen's Universtty, Kingston, Ont. K7L 3N6, Canada
(Recmved August 16, 1990; revised version accepted February 26, 1991)

ABSTRACT
Mitchell. R.J., 1991. Sill mat evaluation using centrifuge models. Min. Sci. Technol., 13: 301-313.
Sill mats are structural elements used in cut-and-fill mining of low to moderate width, steeply dipping ore zones in order
that ore sill pillars may be mined out. Traditional sill mat design appears to have developed from experience and timber mats
are often used to ensure stability. Centrifuge model studies are combined with equilibrium analyses in this paper to provide
insight into sill mat behaviour. Physical model testing provides particular insight into potential modes of failure under
complex boundary stress or displacement conditions. The results show that sill mat reinforcing might be considered as an
alternate to timber mats in cases where plain cemented sill mats would not be expected to be suitable.

Introduction
Sill pillars are ore blocks left between
working levels in an underground mine to
support the overlying mine backfill during
removal of the underlying ore (upper ore levels
are generally mined first for economic reasons). Sill pillars are c o m m o n l y used in steeply
dipping ore zones of limited width and a
typical situation is shown in Fig. 1. Such sill
pillars m a y be candidates for rock bursts and
the values in the ore generally make it economical to mine out the sill pillar. It is then
necessary to create an artificial sill which will
support the overlying fill after the pillar is
being mined out. Such sill mats are often cast
from cemented sand backfill materials, usually without reinforcements but often underlain by a timber mat. A typical timber mat is
shown in Fig. 2. The loading conditions on a
cemented sand sill mat and the potential
failure modes are examined in this paper.
Analytic results are combined with centrifuge
model studies to provide insight into the be-

,ALLY < lOrn

UPPER LEVEL
MINED FIRST

BACKFILLED
FOR
GROUND CONTROL

/
MAT
~BROKEN
~CUSHION

ROCK

SILL PILLAR

LOWER LEVEL
MINED UP TO
EXPOSE
MAT

LOWER LEVEL
MINED LATER

Fig. 1. A typical sill pillar a n d silt mat.

haviour of sill mats. The use of embedded


reinforcements in place of timber mats is
discussed.

0 1 6 7 - 9 0 3 1 / 9 1 / $ 0 3 . 5 0 1991 - Elsevier Science Publishers B.V. All rights reserved

302

R.J.MITCHELL

Loading conditions and failure models


Figure 3 shows a cemented sill mat at the
critical operational stage, just after the underlying sill pillar is removed. It will be supporting a non-uniform vertical stress, Ov, of
unknown magnitude and a lateral closure
stress, % which will increase as rock deformations develop and which may be monitored as ore removal approaches the sill. Sill
mats are normally of such a length that the
problem can be idealized as a two-dimensional plane strain stability problem. The external stresses are resisted by bending and
shear stresses in the sill mat.
Possible modes of failure are crushing, caving, wall shear, torsional shear a n d / o r flexural failure. Crushing would occur if the closure
stress exceeds the plane strain unconfined
compressive strength (UCS) of the cemented
sand. If it is not possible to ascertain that
crushing will not occur, a timber mat would
be considered necessary for pillar removal.
In order to develop some analytical predictions it is necessary to estimate the vertical
stress acting on the sill due to the overlying
uncemented fill. Theoretical considerations
derived from Terzaghi [1], model studies and
field measurements presented by Mitchell [2],

FABRENE~

WIRE SCREEN~
SL,.G

//

J?
y

ULLHORN

Fig. 2. A typical fill mat.

UNCEMENTED

="EAV'
.-+~-~-.

/
\, ~ .

c7'" "2ORE REMOVED

UNCEMENTED FILL

Fig. 3. Stresses and failure modes in a sill mat.

indicate that arching between the H W / F W


rocks will reduce the vertical stresses to:
"yL
v = 2 K tan q>

(1)

where:
7 = the unit weight of the fill;
K = a soil constant, often assumed to be unity.
In the absence of knowledge on the vertical
stress distribution it is prudent to assume that
this stress acts uniformly on the sill.
A wide thin sill mat would, quite obviously, be susceptible to flexural failure due to
the relatively low tensile strength of cemented
tailings. Using standard flexural formulae for
a fixed end uniformly loaded beam, failure is
predicted when:
(L)2
-~ > 2 ( o t + o c ) / w

(2)

where:
a t = the tensile strength of the cemented sill;
w = the uniform loading which should include
the self-weight of the sill mat;

SILL MAT EVALUATION

303

d = the sill depth.


A thick narrow sill mat might be more
prone to caving or undergo side wall shear
failure. If it is assumed that caving would
extend to a stable arch of height L/2 (semicircular arch), then all unreinforced sills would
be formed to a depth, d > L/2 and caving
would develop when:

the sill mat combined with a complete separation in the H W contact (Fig. 3). Making
some allowance for the potential for an increased value of o v on the hanging wall side,
rotational failure would develop when:

Lr

The above analyses are derived for unreinforced, cemented tailings sill mats. Reinforcements (steel wire, mesh or geogrid systems) can be used to improve the tensile or
flexural performance of the mat. Reinforcing
elements could also be fastened to the walls,
particularly the hanging wall, to prevent wall
shear or tortional failures. A timber mat is
generally used as a separate support b u t can
b e designed to be compatible with the cemented sill mat.

> 8ot/q/" =

2.5ot

(3)

F r o m equilibrium, block sliding of the sill due


to side shear failure occurs when:
( Ov + d r ) > 2( ~-f/sin2fl ) ( d / L )

(4)

where:
~-f= the shear stength in the fill-wall rock
contact;
/3 = the H W / F W (hanging w a l l / f o o t wall)
dip angle.
Tortional failure is most likely to develop
when the shearing resistance at the hanging
wall (HW) contact is low due to poor quality
H W rocks a n d / o r low H W dip angles, which
allow separation in this contact. F o r low dip
angles, a simple approximate prediction m a y
be arrived at b y assuming tensile failure in

d 20t

(Ov+rd)> 2L(L_dcotfl) sin2fl

Centrifuge modelling of sill mats


Acceleration of a model in the geotechnical
centrifuge causes p r o t o t y p e gravitational

TABLE 1
Properties of model materials
Test method

7:1T:C

cast

Strengths measured (kPa)


Compressive
Shear
1000
2050

20:1 T:C

control

250

30:1 T:C
Wood dowel

samples
2.1ram diameter

200
N/A

Wood dowel

2.1mm diameter

N/A

Steel rebar
Wood stulls

2.3ram diameter
4.7ram diameter

N/A
N/A

150 MPa
8.7 MPa

390 MPa
78 MPa

Wood lagging

2.5ram square

N/A

4.3 MPa

38 MPa

Steel wire

0.4ram diameter

N/A

Material

Size

N / A = not appficable.
T : C = tailings : cement sill mats.

(5)

120
90
12.5 MPa
Y/i

N/A

Tensile
400
50
40
125 MPa
190 MPa

550 MPa

unconfined
compression,
shearbox
and direct
tension
direct
tension (dry)
3 point
bending (dry)
direct tension
direct tests
(wet condition)
direct tests
(wet condition)
direct tension

R.J.MITCHELL

304

stresses to be simulated by the inertial stresses


created by centrifugal force. Since 7 = Pa and
the centrifuge scale factor is, h = a/g (where
a is the centrifugal acceleration induced), the
induced unit weight is y = p hg. The model
then experiences a self-weight stress increase
proportional to the scale factor, and the induced stresses are identical to those in the
prototype. Actual areas remain at the model
scale and forces (stress times area) are reduced, in the model, according to the square
of the linear scale. In flexure the loading scale
is unity, the length scale is 2~, the m o m e n t
scale is h 2, and the fibre stress scale is unity
(as required for similitude). Similitude can be
achieved for centrifuge models by scaling
structural elements on an area basis. This
allows both the spacing and the size of
elements to be varied in order to produce
similitude with a prototype. In order to have
similitude of the vertical stress distribution it
is considered necessary to have the unce-

mented overburden fill extend to a height of


at least 2L above the sill mat.
The timber mat design shown in Fig. 2 is
typically used for stopes up to 5 m in width.
The stuns are considered to act as independent beams with free end conditions. Bending
compatability between the mat and the cemented sill should be considered if these two
structural components are to be complementary in supporting the overburden uncemented fill. Thus:

(W -- g ) t 4
384(EI)sILL

5qZ4S
-

384(E1)sTULL

where:
w = (Ov + d7) is the total unit loading;
q = the unit loading supported by the stulls;
S = stull spacings;
EI = section modulus.
Then:
W
q = 1 + [5S(EI)sILL/(EI)sTULL ]
(7)

TABLE 2
Scaled Equivalents
Item
Linear scale
Wall roughness
W o o d dowel
reinforcement
Steel rebar
reinforcement
Steel wire
reinforcement

Reinforcement cover

Timber support mat

C - C = centre to centre.

(6)

Model

Prototype (typical)

1/50
Milled 10 m m
rounds at 5 m m
2.1 m m diameter
at 20mm c - c
free end condition
2.3ram diameter
at 20mm c - c
free end condition
0.4ram diameter
at 20mm c - c covered
by light plastic mesh
anchored to walls
Placed 5 m m to
10ram from base of sill
of sill
4.7ram dowels on 40mm
c - c (free ends) with 2.5mm
lagging at 20mm c - c
covered by geotextile

4 - 5 m sill width
+ 0.13 m breakage on
0.5 m (moderately rough wall)
0.10 m round timbers
on l m centres to 0.14m
round timbers on 2m centres
80mm diameter steel
on 0.5m centres
or equivalent hollow sections
0.1m 0.1m mesh of
c o m m o n steel reinforcing
wire anchored to rock walls
0.25-0.5m cover
(from sill base)
0.3m round stulls
on 2m centres with
0.1m lagging on 0.8m
centres; screen
(fabrene covered)

SILL MAT EVALUATION

305

To accomodate the possibility of longitudinal


shear, the standard shear formula gives at
failure:

% L S > 1.3(tA)s

(8)

where (tA)s is the shearing resistance (sectional area times the allowable shear strength)
of the stull. For flexural failure

SL 2
w - -t3

> 0.785o s

(9)

where:
o s = the tensile (flexural) s t r e n g t h o f the stull
material, t = the d i a m e t e r o f the stull.

Centrifuge model tests


The materials listed in Table 1 were used in
a variety of combinations to form models

representing plain, reinforced and timber mat


supported prototype sill mats of 4-5 m width
at a model scale of about 50. The model sill
mats were cast nominally 30-60 m m in depth, representing prototypes of 1.5-3 m in
thickness. Other scale eqivalents are noted in
Table 2. Models were cast at the rate of two
per day and cured for 28 days in a moist
room. On the 28th day, each model was
m o u n t e d in a centrifuge strongbox and conditioned for several minutes at centrifuge speed
of about 60 r p m (producing accelerations of
a = 10 g). The centrifuge speed was then increased at a rate of about 2 r p m / m i n until a
sill mat failure condition was achieved.
Closure stresses were not applied to these
models in order to represent the weakest case
for transverse shear or flexural failures. Model
tests were carried out in two series. The first

TABLE 3
Centrifuge models with varying dip, sill depth and reinforcements
Model

HW/FW
dip (/3 )

Reinforcement
type and
cover

Depth of
sill (mm)

60

30

0.3

55

60

60

0.6

60

Wood dowel
5 m cover
Wood dowel
5 mm cover
Steel rebar
5 mm cover

30

0.3

80
92
77
108

60

Steel rebar
5 mm cover

60

0.6

77
132

90

30

0.3

52

90

Wood dowel
5 mm cover
Wood dowel
5 mm cover

60

0.6

84

90

Steel rebar
5 mm cover

30

0.3

68
137

90

Steel rebar
5 mm cover

60

0.6

110

Scale factor
at failure,

Failure
mode
Sill rotation,
dowel shear
Caving,
sill shear
Caved to rebar,
sill rotation
and rebar
bending
Caving to rebar,
sill rotation
and rebar
bending
Flexural
failure
Dowel shear
and caving,
stable arch
Caved to rebar,
sill shear and
rebar bending
Caved to rebar,
sill shear and
rebar bending

306

series used 20:1 tailings: cement sill mats


(5% cement by mass of dry tailings) to study
the effects of H W / F W dip, sill depth and
embedded reinforcements. The second series
used a typical 70 H W / F W dip and investigated the use of anchored reinforcements and
timber mats with three different tailings cement contents. All model sill mats were overlain by about 250 m m of uncemented fill.
Since the reinforcing elements and timber
mat are correctly modelled for the selected
scale of 50, if the model is stable at a centrifuge speed of 130 r p m (giving)~ = a/g = 50),
the prototype is expected to be stable. Research models were accelerated beyond this
stress level to produce failures and the value
of 2~ at failure expressed as a multiple of 50
might be considered as a reasonable measure
of the prototype safety factor. In reality, however, the scaled equivalents listed in Table 2
are only correct at )~ = 50. Should a model
achieve equilibrium at 2~= 150 (225 rpm), for
example, it would represent a sill mat of 12 m
width and any structural reinforcing elements
would be 3 times larger in area than given in
Table 2. In some cases the resulting prototype
would not be practical.
Table 3 contains data on the centrifuge sill
mat model series designed to study H W / F W
dip, sill depth and reinforcements. The failure
of model I is shown in Fig. 4 where, as in all
of these models, the cemented sill mat is
painted white for ease of in-fright visual observations. This failure developed suddenly,
without prior cracking, by dowel shear and
sill mat rotation. Even with a high degree of
wall roughness, the HW contact provides little resistance to slippage. Rotation about the
FW contact appears to be a primary failure
mode. Having a deeper sill mat, model 2
exhibited a caving failure just prior to collapse, as indicated by the post-failure photograph in Fig. 5. The dashed line in Fig. 5
indicates the main uncemented fill flow zone
as failure developed. Models 3 and 4, which
contained steel rebar reinforcements, suffered

R.J. MITCHELL

.;.~

,.-~.'

i""

.."

. .

I~

-,If,

!/""
~...

, ....

,~'~. !

, ~)..

.,:

,_.f?

.;..-

~. ,,lit ...e 'r


"k, ,,.r .........
~.
,I~,:- -~.: ~,,~.'~':.
.... Z , " " ~ .~"."
.- ....e~,.;". .:~ ;,,-We.:
.:!~.;" :.-.,
. . . . ,.~.
" "~

~'v~

~..;

e.,...

:,L,,.,

. ~r

."

""

" " " "~" "T

"

(b)

;~

, _.'.;"
: ...

.,f,"

..' ,.~
~-"

t '
..,-"

":"

,..

,.

~'

,. "

:
~"

. .....

Fig. 4. Failure of wooden dowels. (a) General view of


failure. (b) Sheared dowel in sill mat.

caving of material below the steel reinforcements at a scale factor of about 80,
followed by rebar bending as the sill mat
collapsed. The failure of model 4 is shown in
Fig. 6 and it can be noted that rebar is still
supporting the overlying materials although
plastic deformation has occurred.
Model 5 exhibited a classic flexural failure
with rupture of the wooden dowels but model
6, being a deeper sill mat, suffered caving
with shear of the dowels. The cemented sill
mat maintained a stable arch at a scale factor
of X = 77. Photographs of these model sills
mats are shown in Fig. 7. Both of the 90 o
H W / F W sill mat models with steel rebar

307

SILL MAT EVALUATION

~ q l m _ .

--, " i ,

,Id~lllll

U ail i
i

lm--

e' ,"~{'q~',-~;.".'"

li

||~

L
i

.......

~r~..::.~,:.......'.:.~ ... "t.':t

.. !-f:.'~-..,:~'~ .::.~,]

Ik~|:
.

:.~. :i~'~;?.: .,~.~


.!~ ,

... ~;2., ,~,,,,~.,

,~.:,:~ .~ :" .?~

~-.. ;,,,..,...;~.~

..

Fig. 5. Sill rotation failure (dashed line shows boundary

of fill flow).

exhibited caving to the rebar, as noted in


Table 3, and considerable flexural cracking in
the sills as deformation developed. The badly
cracked remains of model 8 are shown in Fig.
8. Models with vertical walls generally exhibit
flexural, rather than torsional, failure as a
result of the better shearing resistance in the
H W contact, but this improvement appears to
be offset by larger surcharge stress, o v, such
that the H W / F W angle does not appear to
be a major factor in sill mat design. This
conclusion is certainly not intuitive.
Fig. 7. Sill flexure and caving failures. (a) Model 5,
showing rupture of the wooden dowels. (b) Model 6
showing caving and shear of the dowels.

=~R,,.-~..~'.'~~f~
'


....,

,",

,
"~.~

~ ".;' "~~..~ .~ .~" .'1 ;'."


~...~..,"~,.~,,.
"~ ~..-, .....",~.

:,-

'

g.
!

Fig. 6. Bending of steel rebar due to sill rotation.

The wooden reinforcements marginally increased the sill mat stability but the steel
reinforcements provided m u c h more substantial support. The wooden members failed in
shear while the steel members eventually succumbed to large flexural deformations. Hollow or flanged steel sections could be considered for prototype use but it is doubtful that
these would be economical when stability can
be achieved by increasing the sill mat depth.

308

R.J MITCHELL

Fig. 8. Steel reinforced sill after flexural failure (Model


8).

D o u b l i n g the sill m a t f r o m 1.5 m to 3 m


p r o t o t y p e d e p t h increased the stability b y over
40%.

H i g h c e m e n t c o n t e n t u n r e i n f o r c e d sill m a t s
have b e e n successfully used in practice where
there is g o o d wall rock a n c h o r a g e a n d the
closure strains are sufficiently low t h a t lateral
crushing of these relatively stiff m a t s is n o t a
problem. T h e m o d e l tests listed in Table 4
were designed to c o m p a r e high strength a n d
low strength sill m a t s w i t h a n d w i t h o u t supports. R o u g h H W / F W contacts at a dip of
70 a n d a m o d e l sill m a t d e p t h of 30 m m
were s d e c t e d for the tests. This m a t d e p t h
p r o d u c e s c o m p a t i b i l i t y in b e n d i n g b e t w e e n
the sill a n d the t i m b e r mat, giving q = 0.4w
in eqn. (8) for a typical ratio of (EI)sTULL = 8
(El) sILL.
T h e effect of the t i m b e r m a t o n stability is
d e m o n s t r a t e d b y the p e r f o r m a n c e of models
C1 ( a l t h o u g h the sill m a t is b a d l y fractured, it
is stable at ~ - - 1 6 0 ) a n d C2 ( u n s u p p o r t e d
failure at ~ = 60) s h o w n in Fig. 9. M o d e l s C3
a n d C4 are s h o w n after failure in Fig. 10. T h e
7 : 1 tailings:cement sill m a t (C3) was o n l y 2.5

TABLE 4
Centrifuge models with anchored supports, mats and high cement content
Model
(T: C)

HW/FW
dip (fl o )

Reinforcement
type and
cover

Depth of
sill (ram)

Scale factor
at failure

Failure
mode

C1
(20 : 1)
C2
(20 : 1)

70

30

0.38

160

70

Timber
support mat
None

35

0.43

70

None

25

0.40

35
50
80
125

Sill rotation,
stull bending
Caving,
cracking,
shear failure
Sill rotation

70

None

30

0.38

53

70

Timber
support mat

30

0.38

68
115

70

Light wire
10 mm cover
Timber
support mat
Light wire
10 mm cover

30

0.38

68

30

0.38

240

Caving, sill
rotation
Sill cracking,
flexural
failure
Sill rotation,
wire rupture
No failure

30

0.38

88
215

Sill cracking,
wire rupture

C3
(7: 1)
C4
(30: 1)
C5
(30 : 1)
C6
(30 : 1)
C7
(7 : 1)
C8
(7 : 1)

70
70

T: C = tailings:cement

SILL

MAT

309

EVALUATION

times as stable as the 30:1 taifings:cement


sill mat (C4) but the unconfined compressive
strength of these mixes are different by an
order of magnitude. Model C5, shown in Fig.

EEl

-.

,d

,:
~d

11, demonstrates that it is possible to fail a


timber mat below a low cement content sill
mat but it was not within the normal working
limits of the centrifuge to fail the combina-

,?..
-

v-"

.j.~.,-.

,........~., . . . .

.'"

~
,

.i

~
r.

.:

.,

..,

~IK"

"7. ,

'"

;i

,T

Fig. 9. The effects of a timber mat. (a) and (b) Model C1 with a timber support mat showing only rotation and
bending. (c) Model C2, without reinforcement, showing caving and cracking.

310

RJ. MITCHELL
cemented tailings sill mat having a d/L ratio
as low as 0.5 and a strength of 300 kPa or
better could be adequate for spans of up to 5
m, providing that the hanging wall contact is
stable.

tion of a timber mat and 7 : 1 tailings : cement


sill mat (model C7). Models C6 and C8 were
wire reinforced to represent p r o t o t y p e
anchored wire mesh. Failure, in b o t h cases,
developed b y sill mat cracking followed b y
rotation and rupture of the reinforcing wires.
This type of reinforcement was more effective
with the 7 : 1 tailings : cement sill mat than
with the 30 : 1 tailings : cement sill mat, giving
improvement factors (over the plain sill mats)
of 1.7 and 1.3, respectively, for the two cement contents.
Perhaps the main conclusion from the
combined model test series should be that the
tradition of using timber mats beneath cemented sill mats is a good solution to a
complex design problem. This is particularly
true in cases where high closure displacements m a y cause crushing of the cemented
sill mat. In cases where crushing is not a
problem, the use of anchored wire mesh reinforcements in the cemented mat would be an
effective alternative to the timber mat. A plain

Comparisons of predictions with model results


Assuming that arching is maintained in the
overlying uncemented fill, the surcharge stress
Calculated from eqn. (1), with 3' = 20 k N / m a,
L=5m,
K = l , and f f - - 3 3 , i s 7 7 k P a . T h e
value of d~, is 30 k P a for a 1.5 m deep sill
mat and 60 kPa for a 3 m deep sill mat. An
average value of (o v + d3') = 120 kPa can be
used for an approximate prediction of the
unit sill mat loading at a scale factor of 50.
To predict the failure scale factor, w = (o v +
d3') = 2.4X is used to calculate the values
given in Table 5. Predictions are not made for
stull shear because this is not a critical condition for the timber mat design. Consideration

i i

. . . .

i-

'

"-E',,','

~!

--~l,~

, =.~=

....... L

n!!

:-"

,~,,i

I;-=.[

'~US__-~ " ~ 2 7 ~
L~ ....
---~

~=~
i

./1

t.:,

j,

-- ,i,,.

~==-~
- '.

~-~=-_
'" ~-C
I

,it'

,-i-',i

l.,I-i

Fig. 10. Failures in plain cemented sills.

SILL MAT EVALUATION

311

of the flexural and shearing resistance in the


free end wood dowel reinforcements, however, would add scale factors of X = 108 and
X = 20, respectively, as indicated in brackets
on Table 5, to all cases where the failure
m o d e required rupture of these reinforcements. Failures in sill mats containing
the free end steel rebar reinforcements occurred due to caving of the cover and bending
of the rebar. These bars can theoretically perform in flexure to an ultimate stress level
given by X = 340, but incompatibly large deformations occur at levels of about 50% of
this value. Detailed predictions of failure for
the anchored steel reinforced cemented sill
mats are not considered in this paper because
the bond strength is an unknown factor.
From the data in Table 5 it is concluded
that sill mat flexure and sill rotation (or tar-

sional shear) are critical failure modes for


typical unreinforced sill mats. Sill shear and
caving are potential modes of failure for shallower mats. The question of how these predictions compare with centrifuge model results is
addressed in Table 6. The best predictions are
associated with timber mat flexural support
(models C1 and C7) and the poorest predictions with plain sill mat flexural support
(models C2 and C4). Wood reinforced sill
mat torsional shear predictions are conservative by a factor of about 2 (models 1 and 2)
due, possibly, to the assumption of negligible
HW contact shearing resistance. Wood reinforced sill mat shear predictions (models 5
and 6) are fairly accurate and the steel rebar
reinforced sill mat ultimate load predictions
are quite good (models 3, 4, 7 and 8).
The model test results appear to be con-

TABLE 5
Predictions for unreinforced and wood supported sills
Failure
mode

Sill
flexur
SiH
caving
Sill
shear
fl = 90
Sill
shear
fl = 60
Sill
rotation
fl = 60

Derived formula

Predicted value of h at failure

(d/L)

X = (ot/1.1)(d/L) 2
eqn. (3) with
o=O
X = l . 4 at
eqn. (4)
X = 0.91 at(d/L
eqn. (5)
)t = 3.64
eqn. (5)

20:1

30:1

7:1

20:1

30:1

T:C

T:C

T:C

T:C

T:C

T:C

33
(140)

4
(110)

3
(110)

131
(240)

16
(124)

13
(120) *

560

70
(90)
33
(53)

56
(76)
25
(45)

560

70
(90)
66
(86)

56
(76)
50
(70)

130
(150)

100
(120)

N/A

260
(280)

200
(220)

39
(59)

5
(25)

4
(24)

192
(212)

24
(44)

19
(39)

156

156

156

156

156

156

273

at(d/L )

N/A

1.1[L]2

0.58(L

= 0.6

7:1

)t

(d/L)

= 0.3

546

O-(-t k "~ ] -- Ot k d)

(not applicable to 13 = 90 o )
Timber mat
h = 2(o~MPa)
flexural
eqn. (10)
failure

T : C = tailings : cement
* Numbers in brackets indicate expected improvements due to wooden dowel reinforcements.
N / A = not applicable.

312

R..J. MITCHELL

sistent:

even the ratio

performance
and

o f 3.2 i n t h e f l e x u r a l

of the anchored

C 6 ) is o f t h e s a m e

2.4 for the

plain

wire models

order

models

of

content

(models

C3 and C4). Models

C6 and

(C8

C8 were

as the ratio of

formance

similar

sill m a t s a n d t h e r e s u l t s i n d i c a t e t h a t t h i s t y p e

cement

designed

to evaluate

of light mesh

the likely per-

reinforced,

prototype

TABLE 6
Comparison of predictions and observations
Model
no.

Description

d
~

Dip
(/3 o )

Predicted
failure

Observed
failure

Comments

Wood dowel
in 20 : 1 T : C
W o o d dowel
in 20 : 1 T : C
Steel rebar
in 20 : 1 T : C

0.3

60

0.6

60

0.3

60

sill rotation
~ = 25
sill rotation
~, = 44
caving, ~ = 70
rebar bending
at 2~ = 170

sill rotation
~ = 55
caving, sill
shear ~ = 92
caving, X = 77
rebar bending
at )~ = 108

Steel rebar
in 20 : 1 T : C

0.6

60

Wood dowel
in 20 : 1 T : C
W o o d dowel
in 20 : 1 T : C

0.3

90

0.6

90

caving, h = 70
rebar bending
at 2~--- 170
sill shear
)~ = 53
sill shear
)~ = 86

model stronger
by factor of 2
model stronger
by factor of 2
rebar bending
is a progressive failure
mode
sill supported
by bending in
rebar
dowels
sheared
sill caving
predicted at
)~ = 90
caving
observed at
)~ = 68
caving
observed at
)~ = 110
very good
prediction
some caving at
2~= 50
sill rotation
predicted,
)~ 135
poor
prediction

2
3

Steel rebar
in 20 : 1 T : C

0.3

90

rebar bending
2~ = 170

caving, 2~ = 77
sill rotation
at 2~ = 132
sill flexure
h = 52
sill caving,
dowel shear
at )~ = 84
rebar bending
)~ = 137

Steel rebar
in 20 : 1 T : C

0.6

90

rebar bending
h = 170

rebar bending
)~ = 188

C1

Timber mat
20 : 1 T : C
Plain
20:1 T: C
Plain
7 :1 T :C

0.38

70

0.43

70

0.40

70

mat flexure
)~ = 156
sill flexure
)~ = 1 0
sill flexure
~ = 82

mat flexure
)~ = 160
sill shear
)~ = 80
sill rotation
~ = 125

sill flexure
or rotation,
~<10
mat flexure
X =156
no formal
prediction

caving, sill
rotation,
~=53
mat flexure
X =115
wire rupture
at ~ = 68

sill and mat


flexure at
)~ = 238
no formal
prediction

no failure
at h = 240

stable sill to
25 m width

wire rupture
at )~ = 215

some cracking
at X = 88

C2
C3

C4

Plain
30 : 1 T : C

0.38

70

C5

Timber mat
30:1 T : C
Anchored
wire in
30:IT:C
Timber mat
in 7 : 1 T : C

0.38

70

0.38

70

0.38

70

0.38

70

C6

C7

C8

Anchored
wire in
7:IT:C

sill cracking
at 2~ = 6 8
reinforcement
carrying load

313

SILL M A T E V A L U A T I O N

Fig. 11. Failure of a timber mat.

of reinforcing has potential. Further evaluation of this alternative to timber mats is certainly warranted.

Conclusions
Data from sixteen centrifuge model tests
on sill mats with various types of reinforcements a n d / o r support systems have been
compared to equilibrium analytical performance predictions with the following conclusions:
(1) Analytical predictions of the flexural
performance of plain cemented tailings sills
are not very accurate. Centrifuge models indicate that plain sills may be stable for spans
up to 5 m as long as an effective hanging wall
contact can be provided.

(2) Sill rotation (torsional shear) appears


to be a prevalent failure mode for H W / F W
dips of 70 or less. Rotation may be prevented by the use of reinforcements or other
support systems.
(3) The traditional use of timber sill mats
is supported by the centrifuge model study. A
significant advantage of timber mats is their
ability to undergo large deformation before
rupture. To ensure flexural compatibility,
timber mats should be designed using site
specific information.
(4) Reinforced sill mats, particularly mats
with light mesh reinforcements anchored to
the walls, appear to be an effective alternative
to timber sill mats.
Model sill mat failures were produced at
scale factors ranging from X = 5 to X = 215,
representing prototype sill mats of 5-22 m in
width and having H W / F W dips from 60 to
90 o. This covers a wide range of prototype
geometries for which cemented sill mats might
be considered. The possibility of sill mat
crushing due to closure displacements is noted
but not taken into account in the model work.
Cemented sill mat design is a complex and
often site-specific problem. Further research
is needed in order to provide more confidence
in the applications of the above conclusions
to prototype design.

Acknowledgements
Financial support from the Natural Sciences and Engineering Research Council
(NSERC) for centrifuge modelling of mine
backfill structures is greatly appreciated. The
interest of the Canadian mining community
in promoting research to improve their understanding of sill behaviour and design is most
encouraging.

References
1 Terzaghi, K., Theoretical Soil Mechanics. Wiley, New
York (1943).
2 Mitchell, R.J., Earth Structures Engineering. Allen
and Unwin, Boston, Mass. (1983), 260 pp.

You might also like