709-724, 1995
Copyright ~5:1995 Elsevier Science Ltd
0148-9062(95)110018-6
Printed in Great Britain. All rights reserved
0148-9062/95 $9.50 + 0.00
Pergamon
Th& paper deals with the study of rockfalls us&g a mathematical model,
codified for computer use. Called CADMA, it allows predictions to be made
of fall trajectories and of the relevant parameters (energy, height of bounce,
run out distance of the falling blocks)for the design of remedial works.
Designed with the experience gained from several in situ tests, this model is
based on rigid body mechanics, and statistically analyses a fall in a two-dimensional space. The main features of the program are presented in this paper, as
well as the criteria for choosing the trajectory to be studied, and the techniques
for the assessment of the most relevant parameters required for the execution
of the rockfall analysis (particularly the dynamic parameters: restitution and
rolling friction coefficients). Some practical aspects of the rockfall mathematical analysis are also discussed. These include the effect of topographical detail
on the results and the optimal number of simulations to be carried out. The
characteristics and potentials of the program were evaluated by comparing the
results of in situ tests: in all cases, the program supplied generally accurate
predictions in terms of fall velocity, energy, height of bounce and stopping
distance.
INTRODUCTION
In the context of slope instability phenomena, the
detachment of blocks from steep walls and their subsequent falls along slopes are particularly significant
[1]. This phenomenon involves high risk in densely
populated mountain areas, such as the Alps, where
slopes are usually long and steep, and where housing
estates and most man-made constructions are generally located at the bottom of valleys. It is particularly
important in these areas to have the best possible
knowledge of rockfall trajectories and energies in order
to determine accurate risk zoning and construct
adequate defence systems near the threatened areas.
Until recently, rockfall problems, and specifically,
remedial activities were mostly managed on an empirical basis, since understanding of the subject was somewhat limited. Today, computers represent an invaluable
instrument in dealing with highly variable phenomena
(such as rockfalls). Their development, together with
valuable experience gathered through a more rational
observation of the phenomenon (in particular with
flSMES SpA, Via Pastrengo 9, 24068 Seriate, Bergamo,Italy.
~ENEL CRIS, Via Ornato 90/14, 20121 Milan, Italy.
special/n situ and laboratory tests), has increased rockfall knowledge considerably. Such knowledge now
allows us to perform more rational and repeatable
analyses and gain more accurate predictions and thus
more effective protective structures.
MAIN APPROACHES TO THE PROBLEM
710
L
I
Real ~
physical
system
Experimental tests
Mathematicalmodelof ~ _ ~
the real physicalsystem
Mathematical
model
analysis
No I
Stop
]~
Mathematical .~
synthesisof the
phenomenon "
Yes
AZZONI et
ANALYSISOF ROCKFALLS
al.:
711
IDEALIZEDTOPOGRAPHIC
PROFILE
A
TRUE TOPOGRAPHIC
PROFILE
1-I' 2
TRUE
TRAJECTORY
~ ~ ~ -
I PROJECTED
TRAJECTORY
," ' . ,
C'
",,.
Fig. 2. The kinematics of the motion is studied in a vertical plane obtained by rotation into a single plane of all different
vertical planes.
Y
J,
X
Z
Fig. 3. Model of the block at the impact.
The natural variability of some important parameters (such as the shape of the block, the mechanical characteristics of the slope, local slope angle
at impact, detachment area and inclination on the
slope of the trajectory after detachment) requires
that both description and analysis of the phenomena be statistical rather than deterministic. For this
purpose, the model takes into consideration a large
number of falls and adopts random values (chosen
within a previously determined range) for each of
the above-mentioned parameters.
Rolling
Free fall
Impact
'.Vm
\
\
",Freefall
\ IVy
"mpact
a)
~llV
b)
712
AZZONI
ef al.:
ANALYSIS
OF ROCKFALLS
X
Fig. 6. Assumptions
to+ At
F(t) dt
s fo
represents the impulse of the active forces
(Fig. 6),
-after
impact a rotation
point.
r,,+Ar
s '0
fO>dt
F(t)
f(t)
t,
At
t
and active F(r) forces in the
to the impact.
this
= Gax Ko
AZZONI et al.:
ANALYSIS OF ROCKFALLS
o,F
a)
-
713
a)
d)
b)
~03
P3
02
P ~
i
X,
04
Pl = Ps
i X,
C2
o (X)
Fig. 7. Interpretation of the rolling friction coefficient for a wheel.
714
N oo
oo
'
".00
.
e'oo
'.
.
,6oo'
. 2~oo'
3~oo
.4 0 . 0. 0
4. 8 . 0 0
56.00
64.00
72.00
I m)
!
.OO
8.00
16.00
24.00
32.00
VELOCITY
~6.34"
8.63
40.00
48.00
FREQUENCY
2~01"
V (m/s)
OBSERVATION SECTION X--35.00m
66.00
64 -00
?2.00
Ira)
++
4~07
--.0!
f(Hz)
HEIGHT OF BOUNCE
.67
1.34
2.00
2.66
3.33
H (m)
Experimental method
Experimentation on physical models allows visualization of all the aspects of a phenomenon that, because of
their specific and aleatory character, could be difficult to
715
To predict the rockfall characteristics through computer analysis, the mechanical characteristics and
geometry of the blocks and slope (topography) are
needed. The former are usually represented not directly,
but through some coefficients which allow the modelling of the amount of energy dissipated during the
various phases of the fall. In particular, the restitution
coefficient expresses the amount of energy dissipated
during the ground impact. This is generally considered to have an elastoplastic behaviour. The rolling
friction coefficient expresses the frictional effect of the
ground on the rolling block.
The block's geometry is usually expressed by its
volume and the ratio between its main axes, while the
topography, once the section for study has been defined,
is described in a profile which should be as detailed as
possible. Besides these elements, the mathematical model
utilized for this research requires the assessment of other
parameters (namely, the modulus and direction of the
starting velocity, and the velocity at which the block
changes its movement from rolling to bouncing and vice
versa). The following sections describe the more relevant
dynamic parameters used in computer analysis and their
assessment [25, 26].
Restitution and rolling friction coefficients
716
AZZONI et al.:
ANALYSIS OF ROCKFALLS
Table 1. Values of the restitution and rolling friction coefficients adopted for the calibration of the mathematical model
Maximum
restitution coefficient
Block size
Rock (limestone)
Fine angular debris and earth, compacted
(gravel and cobbles, dia < 20 cm)
Fine angular debris and earth, soft
Medium angular debris
with angular rock fragments
(20-40 cm dia)
Medium angular debris with scattered trees
Coarse angular debris
with angular rock fragments
(40-120 cm dia)
Earth with grass and
some vegetation
Ditch with mud
Yard (fiat surface of
artificially compacted ground)
Road
different falls. The values represented in Table 1 correspond to the maximum values of the restitution
coefficient assessed for different geological materials.
The values of rolling friction coefficients were also
evaluated through back-analysis of two different experimental tests, carried out with blocks of different shape
and volume (a prismatic block, of about 1.2m 3 in
volume, and a spherical one of 0.3 m3).
Since the rolling-friction coefficient in this case
depends on the roughness of the slope in relation to the
size of the falling block, two different values (depending on the volume of the blocks) were determined.
Following some tests, it was noted that values obtained
through back analysis alone, yet unconfirmed by other
methods, do not always provide correct results when
used on other slope types (particularly if topography
is not very detailed). For this reason, another method
for the assessment of the restitution coefficient was
established.
Assessment by elaboration of in situ test. The ground's
restitution coefficient was evaluated according to its
more rigorous definition, namely the ratio between the
total energy of the falling block before and after its
impact on the ground. Energies were assessed by
0.754).90
0.554).60
0.3 m 3
0.40-0.45
0.504).60
1.2 m 3
0.40
0.40
0.35-0.45
0.454).50
0.70-0.80
0.60-0.70
0.60-0.70
0.50-0.60
0.40~.50
0.554).70
0.65-1.20
0.60-0.80
0.504).60
0.554).65
0.454).50
0.70-1.00
<0.20
0.504).65
0.85
0.50-0.65
0.75
0.404).45
measuring both rotational and translational block velocities, slope inclination and sizes of the block at impact.
In this way it has been possible to assess that the
maximum restitution coefficient (Emax)ranges on normal
slopes from 0.35 to 0.95, depending on geological conditions (Fig. 10). In this figure, values larger than 1
should be disregarded, since they indicate an increase
of energy (these values are related to some limitations
in the analysis of video records).
Like the previous parameter, the rolling friction
coefficient was assessed both through back analysis
(Table 1) and the elaboration of in situ tests. This
experimental method was set up partially in line with
the concepts proposed by Statham [7]: coefficient
depends on the ratio between the size of the rolling
block (D) and the debris (d), and basically corresponds to the slope angle at which the block moves
with a steady velocity (neither accelerating nor
decelerating) (Fig. 11).
Careful observation of the blocks' velocities at different positions in the fall trajectories and the measurement of the slope angles corresponding to the different
positions of blocks, enabled us to assess rolling friction
coefficients which provide a good match with those
oo:o
0.4
Unreliable value ]
lain
BARE ROCK
0.2
0.6
O~
0.8
1.2
.4
1.6
1.8
Restitution coefficient
Fig. 10. Restitution coefficient values assessed for impacts on different types of ground (after [26]).
717
50
._. 45
40
. . . . . . . . . .
~v 35
~
...........
. . . . .
',
__
. . . . . . .
O r
o;
i
i
. . . . . . . . . . . .
i . . . . . . . . . . .
=,
'
i
A'--
30
< 25
20
15
0.00
..........,
~3-o . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . ~. . . . . . . . . . . ~. . . . . . . . . . . . ,. . . . . . . . . . . . ~O. 10
0.20
0.30
0.40
0.50
diD
Fig. 11. Assessment of rolling friction coefficientand comparison with Statham's values (after [26]). discoidal block,
accelerating; [] discoidal block, decelerating; spheroidal block, accelerating; 0 spheroidal block, decelerating; tabular
block, accelerating; A tabular block, decelerating; columnar block, accelerating; O columnar block, decelerating;
A' = Statham's upper boundary; A" = Statham's lower boundary; B = rolling friction coefficientby back analysis.
Starting velocity
Even if the real starting velocity generally equals
zero or is slightly greater, in computer analysis, so as to
" m o v e " the block, it is sometimes necessary to provide
a certain initial velocity (usually 1-3 m/sec), particularly
on rough or low inclination slopes. This parameter is not
so relevant to the general trajectory when the fall starts
from a steep wall, since here the velocity is much more
influenced by the effect of the force of gravity. In some
real cases, it is useful to use a certain starting velocity,
such as when only the profile of the lower part of the
slope closest to the threatened area is available.
CALIBRATION
AND COMPUTER
MODEL
RESULTS
718
A Z Z O N I et al.:
ANALYSIS OF ROCKFALLS
Slope top
Slope top
Rock f o i l / / , \ I
oreo / ,
\
/
I
Limits of r o c k / (
,fell/..
/..
"
='k
t
"
t'
I-
,~/
10000
i
"~
V
. . . . . .
1000
,--l - - - - - -
.........
-*-
"1
f7
i
1
'
r
"I
['1
1
I
,- . . . . . . . . .
:::::c::~:::
,-_
i '1"
I
.
. . . . . . . . . . . . . . . .
: : = :'' ~ : : " . . . . . . .
J-J-L~JJ
i
. . . . .
-- T
-- ~
-- ~
-- r
. . . . .
T--i-
' I
r-
7-
d . . . . . .
LLI
I. . . . . .
I
25-30*
~ - - . 0
-F_l_
i i
T~_I_Iq
r - - 7 - - r -i-
- : : 3 : 2 _ - i : : [ -- [ 3 = r
J__II__L_LJ_LL
i
i
. . . . .
~ - - - i - - ~ - r
i i
'l--
i i i
i i i
i ~
I i /
100
6 0
~ _ ~ r r l
>
10
"t - 1 - 1 - 1 -
50-60*
[E
i
t
i
i i
, - - T - r - i - i - r r
i
i
I
i i
I I
Slone annie
J _ _1_ _1_ ~ J _ l _ l _
T. - - .I - . . 7 .- r . . " I 7 7
--
1l
. . . . . . . . . . . . .
--=--:-=aZ_---,;;~-_-c~_:ccc
-~ . . . . .
I
[ - [ 7 - - F F
.t
--'1
I
. . . . .
'
I
& .....
:I:=;::"
- - - - ]
I'l
I
" ...................
I. . . . . . . . . . . . .
100
a.....
c~!:::::r::~::c:,:z~-:,
F ...................
. . . . . .
1 7 7 " - -
T777
T--I-q-F~777
r-r
7-rrr
1000
VELOCITY (m/s)
ABCDE
24
ABCDE
ABCDE
ABCDE
ABCDE
I I l l l
~16
.......
--:.:
_ _
--i--
~ &
16
24
29
DISTANCE OF OBSERVATION POINT (m)
ABCDE
ABCDE
35
ABCDE
ABCDE
ABCDE
I
I
I
I
E'~-6
I"1O
AAAAA
~3
......................
"1-
AAA~A
l #
Ill
il 1
16
24
29
DISTANCE OF OBSERVATION POINT (m)
35
FREQUENCY (Hz)
ABCDE
ABCDE
ABCDE
ABCDE
ABCDE
._.3
i
.
~
. . . . . . . . . . . . . . . . . . . . . . . . .
z
i
IJ.
I I I
tit
#####
iII
-._
_. .................
AAAAA
__llll
II
I[
z~;~Z
tl
....
t*l
II
I I
A AAA~
Ii-0
16
24
29
DISTANCE OF OBSERVATION POINT (m)
- Computer max.
value
- Computer min.
value
35
Computer mean
value
;':::.~......
SLOPE A
SLOPE B
...."
R~ck
40
40
'~':,~
30
' r,ne
'"
,'~,.J Compacted
-~,;~%
.~/k
'";~:'~'~'~.~. ":~"
Rock
20
20
; i}k~Fine-medium angular
:-'%'~ debris, loose
'; " i " : ~ / d i t c h
~-i;~ (
yard
10
10
20
30
40
- ";'k~
loose
10
,'o
2'o
debt, s, c o m p o c , e d
angu,ar
719
go
l'~: ~
;o
5'~
720
AZZONI et al.:
ANALYSIS OF ROCKFALLS
compared amongst themselves and with the experimental data. The tests showed that analyses with 100 and 200
simulations were positive and totally similar, while
analyses with 20 falls were inaccurate. The results of the
tests performed with 50 falls were basically similar to
those with 100 simulations, but 50% faster. Thus, this
number of simulations is considered advisable when
dealing with detailed slope profiles or parametric analyses. In fact, for a slope about 60 m long, described with
approx. 170 points, an analysis carried out with 50
simulations took about 8 min on a 386 PC and produced
an output file of about 1.2 Mbytes. The same analysis,
run with 100 simulations, required about 17min for
calculations with a resulting output file of about 2.3
CONCLUSION
VELOCITY (m/s)
24.0
E 16.0
0
o
8.0
.._1
m
0.0
7.0
16.0
24.0
29.0
35.0
?
E 6.0
T
(3
~'~ 3.0
"1"
0.0
7.0
16.0
24.0
29.0
35.0
FREQUENCY (Hz)
9.0
T
v
>" 6.0
O
Z
UJ
O 3.0
UJ
rY
LL
0.0
7.0
16.0
24.0
29.0
35.0
Computer av.
mean value
4. Computer av.
modal value
O Av. experimental
value
O Max. experimental
value
Computer max.
value
Fig. 15. Comparison between the computer analysis results and the experimental data, for slope A.
AZZONI et al.:
ANALYSIS OF ROCKFALLS
V E L O C I T Y (m/s)
20.0
15.0
E
---- 10.0
D
._1
LU
>
5.0
0.0
40.0
54.0
-~
0.8
I.U
"I"
0.4
c
i
0.0
40.0
L
54.0
O Av. experimental
value
Computer max.
value
721
Fig. 16. Comparison between the computer analysis results and the experimental data, for slope B.
722
B O U N C E H E I G H T (m)
V E L O C I T Y (m/s)
25
20
v
15
o,
lo
>
16
24
29
35
DISTANCE OF
OBSERVATION POINT (m)
- Computer
max & min.
Computer
mean value
16
24
29
35
DISTANCE OF
OBSERVATION POINT (m)
Computer
modal value
o Experimental
value
value
Fig. 17. Comparison between the results of the computer back-analysis and the experimental data for roekfall No. 26.
to further improve output data and render
u s e r - f r i e n d l y f o r all r o c k fall s p e c i a l i s t s .
it m o r e
REFERENCES
1. Spang R. M. Protection against rockfalls---stepchild in the
design of rock slopes. Proceedings of 6th Int. Congress on Rock
Mechanics, Montreal, Canada, pp. 551-557 (1987).
2. Richards L. R. Rockfall protection: a review of current analytical
and design methods. Secondo Ciclo di Conferenze di Meccanica e
Ingegneria delle Rocce, MIR, Politecnico di Torino, pp. 11.1-11.13
(1986).
3. Ritchie R. M. Evaluation of rockfalls and its control. Highways
Res. Record. 17, 14-28 (1963).
4. Camponuovo G. F. ISMES experience on the model of St.
Martino. Proc. Meet. Rockfall Dynamics Protective Works
Effectiveness 90, 25-39 (1977).
5. Broili L. Relations between scree slope morphometry and dynamics of accumulation processes. Proc. Meet. Roekfall Dynamics
Protective Works Effectiveness 90, 11-24 (1977).
6. Habib P. Notes sur le robondissement des blocs rocheux. Proc.
Meet. Rockfall Dynamics Protective Works Effectiveness 90,
123-125 (1977).
7. Statham I. A simple dynamic model of rockfall: some theoretical
principles and field experiments. International Colloquium on
Physical and Geomechanical Models, pp. 237-258 (1979).
8. Falcetta J. L. Etude cynematique et dynamique de chute de blocs
rocheux. Th~se, INSA, Lyon (1985).
9. Chan Y. C., Chan C. F. and Au W. C. Design of a boulder fence
in Hong Kong. Conference on Rock Engineering in an Ubran
Environment, Inst. Min. Metall., Hong Kong, pp. 87-96 (1986).
10. Mak N. and Blomfield D. Rock trap design for pre-splitting slopes.
Conference on Rock Engineering in an Urban Evironment, Inst.
Min. Metall., Hong Kong, pp. 263-270 (1986).
11. Piteau D. R. Computer rockfall model. Commun. Proc. Meet.
Rockfall Dynamics Protective Works Effectiveness 90, 123-125
(1977).
12. Hacar B., Bollo F. and Hacar R. Bodies falling down on different
slopes. Dynamic studies. Proc. 9th Int. Conf. Soil Mech. Found
Engng 2, 91-95 (1977).
13. Azimi C., Desvarreux P., Giraud A. and Martin Cocher J.
Methode de calcul de la dynamique des chutes de blocks.
Application ~ l'Otude du versant de la montagne de la Pale
(Vercors), Bull. liaison Labo P. et Ch. 122, 93-102 (1982).
14. Hock E. A program in Basic for the analysis of rockfalls from
slopes. Unpublished notes (1987).
15. Hungr O. and Evans S. G. Engineering evaluation of fragmental
rockfall hazard. Proceedings of the 5th International Symposium on
Landslides, Lausanne, pp. 685~590 (1988).
(A1)
(A2)
and the co-ordinates of the initial position of the center of mass at time
to by:
X(to) = XA
Y(to) = YA + ho"
(A3)
(A4)
723
X
Fig. AI. Definition of the free falling problem in an assumed O X Y reference frame.
Then:
V x ~ (d)z d r
V0,
V,, = -
Applying the principle of the conservation of the angular momentum over the infinitesimal time interval, before and after the impact
(Fig. B1), the following relation can be written:
1.~oo+Vo~'d,-Vo~..d~=l.co+V~.d,.-L.d~
(al)
(B4)
coz . de
assuming co = coz:
(B5)
1 coo+Vo~'dy-Vo,.'d~
"
"
I + d2~+ d~
(86)
(B2)
and P G = ( X G - X p ) . i + ( Y o - Y p ) .
Since t o = O - i + O . j - o ~ . k
j+O.k
~
V = to
x PG =
0/"
(xo - xp)
f0
( r o - Y~)
-co~
CASEa):Xa'>X"
(83)
~',~o
( ~ ) ~
(~)~
7,
CASEb):x,~:x~
CASE ): X~ < X e
~0 ( ~
x
Fig. BI. Configuration of the block before and after the impact.
e_
II I
724
A Z Z O N I et al.:
ANALYSIS OF ROCKFALLS
(B7)
Therefore, it is possible to evaluate a coefficient of restitution of energy
with the following relation:
K
Q~
co2
co "Q0
E*
( / + r 2) =
(B8)
K o 2 ' K o - ( l + r 2) 2 . K o
2-K o
within (0~<E* < l) where:
Qo = I " coo + Vo.,=" ~, - Voy " d,.
0=N-m'g'cos~t
m XG = m - g . sina - T
d28
I.~-~=
(C1)
X~
Fig. C1. Definition of the rolling problem in the assumed O X ' Y '
reference frame.
T. R-N.a.
1
(C2c)
I.-~=T.R-N.~
-~.i'~+N.~.
T _I
m+~
(C3)
Defining
A = -
XG(t) -- ,~'o(t0)
t = A ' g ' cos at' (tan ~t -- tan ~bd)'
(C2)
I
)~o(t) = x / 2 ' A ' g 'cos =- (tan ~ - tan ~bd). [XG(t) -- X~(t0)] + ~ ( t 0 ) .
(C2a)
(C4)
From equation (C4), the rolling friction coefficient can be determined
as:
(C2b)
(C5)