Accepted for publication at ICOLD European Symposium, Geiranger, Norway, 25-27 June 2001

A new approach for better assessment of rock scouring due to high

velocity jets at dam spillways
E. Bollaert & A. Schleiss
Laboratory of Hydraulic Constructions, Swiss Federal Institute of Technology, EPFL, Lausanne, Switzerland

ABSTRACT: High velocity plunging jets, emerging from spillways of large dams, often create erosion of the
downstream rock bed. Traditionally, scour is estimated with (semi-) empirical formulae that neglect part of
the physics involved. Above all, the relationship between hydrodynamic pressures in rock joints and pressures
at the plunge pool bottom is unknown. Experimental tests in simulated closed-end joints, at prototype jet ve-
locities, outlined that scouring is highly dynamic, governed by the interaction of three phases (air-water-rock)
and characterized by transient pressure phenomena, such as oscillations and resonance. Based on tensile stress
and dynamic uplift, a rock mass failure criterion is proposed for the assessment of scouring. The application
of measured dynamic pressures to tensile stress failure criteria, such as the linear elastic (LE) or the fracture
mechanics (LEFM) approach, is discussed more in detail. This physically based approach provides a better
understanding of the formation of ultimate scour depth.

1 INTRODUCTION the application of realistic instantaneous pressure

differentials over and under rock blocks or concrete
Results of a theoretical and experimental analysis of slabs of protection linings. The existing approaches
a research study concerning the pressure fluctua- can be classified with the help of the three phases
tions inside closed-end rock joints, due to the impact involved: the liquid phase (water), the gas phase
of high velocity jets emerging from dam spillways, (air) as well as the solid phase (rock). The three-
are presented. A better understanding of these dy- dimensional cube in Figure 1 summarizes the most
namic pressures is necessary to assess the basic important existing methods of ultimate scour
physical processes of rock mass destruction by hy- evaluation and compares them with the objective of
drodynamic fracturing and jacking of the existing the present research and possible future develop-
joints and hydrodynamic uplift of the formed ments. As it can be seen, the challenge is the devel-
blocks. opment of an improved scour evaluation method
Pressure measurements inside artificially created that accounts for the mentioned physical processes
1D and 2D rock joints were performed which high- in a 3-phase transient manner.
lighted the formation of violent transient two-phase In the following, special attention will be drawn
flow phenomena in the form of standing waves, on the global methodology that is proposed for a
shock waves and resonance conditions (Bollaert more appropriate prediction of ultimate scouring.
2001). In that way, extreme pressures inside the Above all, the hydrodynamic action of the air-water
rock joints can attain values of up to several times mixture inside the rock joints has to be defined and
the kinetic energy of the impacting jet. The presence applied to appropriate rock failure criteria for the
of air bubbles in the pressurized flow on this wave physical-mechanical processes of hydrodynamic
propagation process appears to be of crucial signifi- fracturing and uplift. The fracturing and jacking
cance, because it highly influences the governing process is related to the tensile stress resistance and
wave celerity and thus makes the problem non- initial stresses of the rock mass and can be described
linear (Bollaert & Schleiss 2001). by a linear elastic stress field or by a fracture me-
The actual state-of-the-art on ultimate scour chanics approach (Haimson & Zhao 1991). Uplift is
depth evaluation methods comprises empirical and defined as a dynamic equilibrium of forces on a rock
semi-empirical formulas, methods of rock block up- block as a function of time, thus procuring the im-
lift by applying a maximum possible underpressure pulsive action that ejects the block.
combined with a minimum overpressure, and finally
 non-linear
Fully coupled interactive
3-phase model dynamics
Aim of LCH-EPFL oscillations,
Research (1998-2001) resonance
pressure

WATER
differential

turbulence

Fiorotto & Rinaldo 1992, Liu & al. 1998,

May & Willoughby 1991 Fiorotto & Salandin 2000 jet diffusion
Franzetti & Tanda, Puertas
McKeogh & Elsawy 1980, Bin 1984, Ervine 1998, Bohrer et al. 1998
& Dolz 1994, Ervine et al. 1997
Xu Duo Ming 1983, Cui Guang Annandale 1995, Annandale et al. 1998
hydrostatic
Tao 1985, Franzetti & Tanda 1987 R rock aeration
Lewis 1996 AI Spurr 1985
Albertson et al. 1948, Cola 1965, Beltaos & Poreh & Hefez 1967, Stein et al. 1993, pool aeration
Rajaratnam 1973, Hartung & Häusler 1973,… Chee & Yuen 1985, Bormann & Julien 1991
Fahlbusch 1994, Hoffmans 1998 Mason 1989 Montgomery 1984 jet aeration
Whittaker & Schleiss 1984,
Mason & Arumugam 1985
Kirschke 1974 empiricism
empir. initiat. rock 1D 1D 2D ROCK
formulas motion index joint net net

Figure 1. 3D visualization of the actual state-of-the-art on ultimate scour depth evaluation methods and of the LCH-EPFL project’s
objective (Bollaert & Schleiss 2001).

2 EXPERIMENTAL FACILITY
The jet outlet has diameters of either 57 or 72 mm
2.1 General description and is of cylindrical or convergent shape. The water
The experimental set-up can be divided in two main in the basin is restituted over two rectangular sharp
elements: an upper 3m-diameter cylindrical element weirs that are located radially symmetrically at the
in PVC reinforced by a steel structure, and a lower border of the basin. 10 pressure sensor taps are in-
element simulating the joint by a 1 mm thin inox stalled at the plunge pool bottom and 6 inside the ar-
strip that is prestressed with the help of 10 steel bars tificial rock joint.
between two 100 mm thick, 1 ton heavy, galvanized
steel plates (Figure 2 and Photos 5b, c).
2.2 Jet outlet structure
Two different types of jet outlet configurations were
1 used during the tests (Figure 3). The jet of cylindri-
cal shape has diameters of 57 or 72 mm and a total
2 length of 450 mm. One third of this length is situ-
ated inside the upstream water supply conduit, the
3 other two thirds are visible from the outside.

Conduit Conduit
4
6
5 Cylindrical Convergent
nozzle nozzle

Figure 2. Side view of experimental facility: 1) cylindrical jet Jet Jet

outlet, 2) cylindrical basin, 3) pre-stressed two-plate steel
structure, 4) PC-DAQ and pressure sensors, 5) restitution sys-
tem, 6) thin steel strips (1D and 2D fissures).
Figure 3. Cylindrical (57 or 72 mm) and convergent (only 72 characteristics inside. The influence of wall friction
mm) jet outlet configurations used for the tests. along the simulated rock joint, as well as the effect
of the modulus of elasticity of the steel (Est = 2.1 x
The second configuration consists of a conver- 1011 Pa) on pressure wave celerity or structural vi-
gently shaped nozzle of 400 mm long (300 mm out- bration, can be neglected in the present analysis.
side the water supply conduit) and a diameter that
gradually changes from 300 mm to 72 mm.
2.4 Electronic data acquisition equipment
2.3 Cylindrical basin and lower steel structure The data acquisition equipment comprises an auto-
matic data measurement system that was specially
The plastic basin reinforced with steel girders repre- designed for simultaneous and dynamic signal ac-
sents the plunge pool and has a water depth that can quisition and analysis.
vary between 0 and 0.7 m, for a total height of 1m The signal conditioning hardware has been de-
(Figure 4). veloped at the Hydraulic Machines Laboratory
The lower galvanized steel structure prestresses a (LMH) at the Swiss Federal Institute of Technology
1mm inox strip between two thick plates as shown (EPFL, Lausanne, Switzerland) and uses an 8-
in Figure 4. This is obtained by the use of a series of channel platform with the following performances
10 prestressable steel bars of 36 mm of diameter (per channel): programmable sensor excitation and
(Photo 1a). The inner boundary of the inox strip de- amplification, electrical isolation, jumper-selectable
fines the geometry of the simulated rock joint, while lowpass filtering with different settings, simultane-
the force induced into the bars automatically pro- ous 14 bit A-D conversion, and finally routing to-
vides a watertight sealing. wards the PC by use of a high speed ARCNET-
interface. Each channel can stock up to 65,536 val-
ues at maximum 20 kHz acquisition rate.
cylindrical nozzle outlet

a) b)

reinforced plastic basin

galvanized steel plates

micro pressure sensor
prestressed steel bar
inox strip

Figure 4. Perspective view of experimental facility presenting

the alimentation conduit, the upper plastic basin and the lower
steel structure.

The artificially created rock joints are thus one-or

two-dimensional, of any particular form, but with a c)
constant thickness. In this paper, only the results for
the first type of rock joints, i.e. non-persistent or
closed-end rock joints, are presented. This configu-
ration is particulary interesting in view of hydrody-
namic fracturing. The second type of joints, i.e. the
open-end joints, is of interest when evaluating hy-
drodynamic uplift of rock blocks. Tests on these
joints are actually ongoing. A 1D joint (0.80 m long,
0.01 m wide, 0.001 m thick) and an analogous 2D
joint (0.80 m long, 0.60 m wide, 0.001 m thick)
were tested. Although they constitute a very simpli-
fied representation of realistic rock joints, neglect-
ing parameters as joint apertures and macro-
roughness, points of contact, joint walls, filler mate-
rial, a.s.o. they give a good idea of the pressure
Photo 5. a) View of prestressable steel bars at opened steel Because secondary currents in the upstream con-
structure; b) View of the cylindrical jet operating at an outlet duit system could not be totally avoided, the out-
velocity of 15 m/s; c) general view of the experimental facility
with PC and DAQ system in the foreground.
coming jets showed some low frequency (< 1 Hz)
The typical acquisition rate is 1 kHz, with a low- instabilities, particularly visible at jet outlet veloci-
pass filtering at 500 Hz, according to the Nyquist ties of less than 15 m/s.
Table 1. Plunging jet characteristics for the 72 mm diameter
theorem. This generates 65 seconds of values for cylindrical nozzle outlet system
every run and reproduces representative and ergodic _________________________________________________
V Qj Fr Re We Tu L L/Lb L/Dj Y/Dj
_________________________________________________
j
statistical values. Regular control runs were per- 3
formed at acquisition rates of up to 20 kHz in order m/s m /s - - % % cm -
_________________________________________________- -
5
to check the transient character of the measured 7.4 30 8.8 4E10 232 4.45 3-50 0.13 0.4-7 2.1-9.7
pressure peaks. 9.8 40 11.7 5E105 308 5.45 3-50 0.12 0.4-7 2.1-9.7
12.3 50 14.6 7E105 386 4.89 3-50 0.11 0.4-7 2.1-9.7
The used software is written in the LabVIEW en- 14.7 60 17.5 8E105 462 4.31 3-50 0.10 0.4-7 2.1-9.7
vironment and has been developed at the Laboratory 17.2 70 20.5 9E105 540 4.25 3-50 0.09 0.4-7 2.1-9.7
of Hydraulic Constructions of the EPFL. 19.7 80 23.4 1E106 619 4.49 3-50 0.09 0.4-7 2.1-9.7
The surface mounted micro pressure sensors (3 22.1 90 26.3 1E106 694 4.35 3-50 0.09 0.4-7 2.1-9.7
mm in diameter) are of the piezoelectric type 24.6 100 29.3 1E106 773 4.37 3-50 0.08 0.4-7 2.1-9.7
27 110 32.1 2E106 848 4.26 3-50 0.08 0.4-7 2.1-9.7
(KULITE XTC-190C), with an absolute pressure 29.5 120 35.1 2E106 926 4.39 3-50 0.08 0.4-7 2.1-9.7
range between 0 and 17 bar and a total precision of _________________________________________________
+/- 0.1% of the full scale output. They can easily be
screwed into the steel structure. 3.2 Plunge pool bottom pressure influence on rock
With 6 sensor dowels inside the artificially cre- joint pressures
ated rock joints and 10 sensor dowels at the plunge
pool bottom, an easily changeable system of pres- The impact of a high velocity jet into a plunge pool
sure measurement locations can be obtained. is governed by the concept of jet diffusion through a
At the same time, the pressure sensors can be medium at rest. Momentum exchange with the pool
used for pressure measurements directly at the jet generates a progressively growing shear layer, ex-
outlet to determine initial jet turbulence intensity pressed by an increase of the jet’s total cross section
values. and a convergence of the jet core region with con-
stant velocity profile (Rajaratnam 1976). Therefore
dynamic pressures on the rock-water interface can
3 HYDRODYNAMIC CHARACTERISTICS be generated by direct jet core impact, appearing for
small plunge pool depths, or indirectly by the
The first step of a physically based approach of ul- macroturbulent shear layer flow, appearing for ra-
timate scour depth evaluation consists of the hydro- tios of pool depth to jet thickness (Y/Dj) higher than
dynamic action of plunging jet impacts inside pre- 4 to 6 in the case of plunging jets (Ervine et al.
existing or created rock joints. Three physical 1998). Terminology used in this paper thus distin-
phases are involved and have to be studied: plung- guishes between jet core impact and developed jet
ing jet characteristics, resulting plunge pool bottom impact (Figure 6).
pressure fluctuations and finally the directly de-
a) b)
pending rock joint pressure fluctuations. Dj

Jet core
3.1 Characteristics of plunging jet 4-6 x Dj
Dj Y
Table 1 summarizes the most relevant plunging jet Developed
characteristics for the 0.072 m diameter cylindrical Y Jet core Plunge pool
Jet region
Plunge pool
nozzle outlet used for the experiments. Similar tests
Rock joint Rock joint
were conducted on convergent nozzle outlets but are
not presented here because they generated pressure
results of the same kind as the cylindrical nozzle. Figure 6. a) jet core impact appearing for Y/Dj < 4-6; b) de-
Attention has to be paid to the small jet fall veloped jet impact appearing for Y/Dj > 4-6
heights L (max. 0.50 m), and the small degree of
break-up L/Lb (max. 0.13) of the jets. Therefore, the The influence of plunge pool bottom pressures on
generated jets are of a rather compact nature. Fur- rock joint pressures is governed by a flow modifica-
thermore, measured initial turbulence intensities are tion from macroturbulent conditions into pressur-
of around 4 to 5 %. High Reynolds and Weber num- ized flow through a bounded medium. The impact of
bers were obtained, avoiding the influence of vis- a jet on a joint contains principally all the elements
cous and surface tension effects on the results. of a resonator system. The problem lies in the exci-
tation capability of the jet. Transient pressures act-
 ing on a joint with a length of maximum 10 m and In Figure 7 a comparison is made between the
wave celerities of about 1000 m/s can create oscilla- measured pressure at the plunge pool bottom, next
tory conditions for a frequency range that is slightly to the rock joint entrance, and the pressure inside a
beyond 35-70 Hz (fundamental resonance mode fr = closed-end 1D rock joint. It can be seen that the sur-
cj/(4Lf) or = cj/(2Lf) for closed or open-end resona- face pressure signal gets strongly modulated inside
tor, Wylie & Streeter 1978). Resonance is not possi- the rock joint. The pressure inside the joint is char-
ble for the macro-turbulent flow in a plunge pool, acterized by the appearance of important oscillatory
which has its highest energy at low frequencies (< conditions, giving rise to considerable peak values.
25 Hz, Toso & Bowers 1988). The tests performed These peaks indicate the capability of high velocity
by the authors indicate however that a high velocity jets to create oscillatory and resonance conditions
jet has sufficient energy beyond this range to create inside underlying rock joints. Their importance
a resonant excitation inside open or closed-end rock highly depends on the macroturbulent flow pattern
joints (Bollaert 2001). The high air content in at the plunge pool bottom and on the available air
plunge pools (Bin 1984, Ervine 1998) and the trans- content that can be released inside the joint. Jet core
fer of flow at the rock-pool interface from macro- impact is characterized by low air contents (because
turbulent flow into pressurized flow are responsible not influenced by the surrounding highly aerated
for considerable air bubble presence inside the rock shear layer) and relatively moderate pressure
joints. changes. Therefore, the obtained peak values are
This can be explained by alternating air bubble less than those for developed jet impact, which cre-
release and re-solution effects of the flow mixture ate a combination of highly turbulent flow condi-
propagating in the joint. In fact, if a liquid with a tions in the plunge pool, close to the joint entry,
certain gas content in solution undergoes a pressure with an important available air content. For such
drop, supersaturation and thus gas release occurs conditions, well pronounced resonant phenomena
(Bhallamudi & Chaudry 1990). The amount of re- inside the joints, and thus high peak pressures, re-
leased gas directly depends on the pressure drop be- sult.
low the governing saturation pressure and on the This is illustrated with Figure 8, in the case of a
degree of agitation of the mixture. A very slight 1D closed-end joint. The maximum obtained peak
change in free gas drastically changes the mixed pressure values are described by means of a Cmax
fluid compressibility and thus wave celerity. coefficient, defined as the ratio of extreme pressure
The air contents inside 1D and 2D simulated rock head to the incoming jet’s kinetic energy V2/2g.
joints have been quantified by means of the corre- Peak values of up to 5 to 6 times the jet energy in-
sponding wave celerity. For jet core impact, the jected in the plunge pool were measured inside the
mean free air content stays more or less constant be- joints.
tween 0.5 and 2%. This indicates low plunge pool Furthermore, Figure 8 shows a limit of estimated
air contents and ineffective pressure drops inside. maximum values according to the Y/Dj ratio. At the
Developed jet impact allows considerable air bubble beginning, the curve grows with increasing Y/Dj,
release and thus very low wave celerities, even for due to increasing pool turbulence and air release,
high mean pressures. Celerities less than 100 m/s followed by a decrease at Y/Dj > 10 due to consid-
and air contents higher than 10 % have been ob- erable diffusion of the plunge pool turbulence.
served.
6
8
Pressure measured at pool bottom
7 5
Pressure measured inside 1D joint
Absolute pressure [10+1m abs]

6 4
5
Cmax

3 ESTIMATE OF MAXIMUM
4

3 2

2 1

1 core impact developed jet impact

atmospheric pressure 0
0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
0 10 20 30 40 50 60 70 80 90 100 110 Y/Dj [-]
Time units (5 msec/unit)

Figure 8. Coefficient Cmax of extreme peak pressure value in

Figure 7. Pressure result in the time domain: comparison of function of the Y/Dj ratio. Curve of estimated extremes for ap-
the plunge pool bottom pressure with the corresponding 1D plication of failure criteria based on tensile stress (see § 4).
rock joint pressure.
 This limiting curve can be used for the estimation
of maximum peak pressure values, as required when
applying to a failure criterion based on tensile resis-
tance of the rock mass, as explained hereafter in § 4.
The input parameters are the impacting jet velocity
Vj, the jet diameter Dj and the plunge pool depth Y.
All these parameters can be rather easily deter-
mined.
4 GEOMECHANICAL CHARACTERISTICS
4.1 Description of the rock mass Sj1 Sj2 Sj1 Sj2

The analysis of the hydrodynamic action of a high

velocity air-water jet on the underlying, jointed rock
Lj1 Lj2
mass needs an accurate description of the main geo- Lj1
mechanical characteristics of the rock itself and of ej1
its discontinuities. In the following, only two- αj1 αj2
αj1 αj2
dimensional rock mass representations will be con- ej2 φj1
sidered. Table 2 summarizes the most important pa- ej1 ej2
rameters necessary to describe how the rock mass
will resist against the hydrodynamic forces. Most of a) b)
the parameters can be obtained by simple field ob-
Figure 9. Two often encountered rock mass layer situations: a)
servations and borehole tests. The parameter TYPE non-persistent joint set pattern (Pj < 100 %); b) persistent joint
is related to the crystallographic composition of the set pattern (Pj = 100 %)
rock mass whereas STRUC refers to the evolution of
this composition according to depth (Table 2). When assuming that physical-mechanical proc-
esses are responsible for rock mass destruction and
thus for scour hole development, it is obvious that
Table 2. Main geomechanical parameters of the rock mass
________________________________________________ the behavior of the non-persistent rock mass is
Parameter Symbol Dim
________________________________________________ highly governed by the hydrodynamic fracturing
Rock mass type TYPE [-] principle. In order to express the resistance of such a
Rock mass structure STRUC [-] rock mass, it is necessary to relate these forces with
Depth of layer Hl [m] a failure criterion expressing whether the joint will
Rock Quality Designate RQD [%]
Uniaxial compressive strength σc [MPa] propagate or not. This aspect has been largely inves-
Uniaxial tensile strength σt [MPa] tigated by linear elastic (LE) tensile stress theory,
Young’s modulus of elasticity Er [MPa] often used to determine in-situ horizontal stresses,
Material density ρ [kg/m3] and by (static) hydraulic fracturing techniques,
Number of joint sets Nj [-] mainly used in the petroleum industry, and based on
Joint set dip angle αj [°]
Joint set persistency Pj [%] a linear elastic fracture mechanics (LEFM) ap-
Joint set typical length Lj [m] proach. However, the particularity of the present ap-
Joint set spacing Sj [m] plication lies in the highly dynamic character of the
Joint set width ej [m] impacting hydrodynamic forces: the rate of applica-
Joint set friction angle ϕ
________________________________________________
j [°] tion generally modifies the static fracture propaga-
tion resistance of a rock mass and should be taken
into account in a fully dynamic analysis.
The rock mass structure STRUC allows a subdivi- In the case of a completely fissured rock mass, its
sion of the rock mass into different layers of a cer- resistance against hydrodynamic failure can be ex-
tain type of mineralogy and of a certain depth Hl. pressed by some typical rock block characteristics,
For each of the layers, it is essential to know how such as size, shape, weight and shear and/or cohe-
the internal layer parameters are interrelated. sive forces along the joints. This resistance can thus
Figure 9 illustrates in a two-dimensional way the be introduced in a dynamic equilibrium of the forces
two mostly encountered layer situations: in Figure acting on the rock block. This equilibrium will
9a, a non-persistent rock mass pattern is shown, highly depend on the instantaneous difference in
characterized by two (Nj = 2) joint sets that only in- pressure distribution above and underneath the block
tersect at some of their joints. In other words, the po- and therefore has to be formulated in function of
tential 2D-joint pattern of the rock mass is not fully time (impulse). The calculated net impulse of a pres-
established and certain joints are of the so-called sure wave under a block will determine whether the
“closed-end” type. These rocks have joint set persis- considered typical rock block will be uplifted out of
tencies Pj less than 100 %. its matrix or not.
The second layer (Figure 9b) represents a much For each of the aforementioned failure types, the
more completed stage of rock mass break-up, i.e. the ultimate scour hole development can be estimated as
2D-joint pattern is completely established and the follows. Firstly, maximum scour will be reached
fully persistent (Pj = 100 %) rock mass can be sub- when the pressure peaks in the closed-end joints are
divided into a large number of similarly shaped, not capable to propagate the joint further anymore.
regularly distributed rock cubes. Despite the persis- Secondly, for the uplift criterion, scour will be fin-
tent joint pattern, local contact surfaces between the ished when the equilibrium of forces in function of
blocks exist. time doesn’t allow to push a rock block out of the
mass anymore. In the following, each of these fail- the elastic stress field induced by the hydrodynamic
ure criteria will be discussed more in detail. force distribution along the joint. This constant is the
“stress intensity factor” KI and is linearly related to
the hydrodynamic action (stress σ) and directly re-
4.2 Non-persistent rock mass failure criteria lated to the square root of a characteristic length, of-
The presented failure criteria arte limited to the as- ten chosen as the joint length Lj. With reference to
sumption of pure tensile modes of loading (mode I), Figure 9b for the used parameters, this is mathemati-
without any shear force effects. The criteria are es- cally expressed as (Ewalds & Wanhill 1986):
sentially based on the parameter Pb0, named as
“breakdown pressure under zero initial pore pressure
and zero far-field stresses”, also called the “zero  L j1 
K I = σ ⋅ πL j1 ⋅ f   (3)
breakdown pressure” (Haimson & Zhao 1991).  S j2 
 
Concerning joint tensile failure, two major ap-
proaches can be distinguished. The first is the linear
elastic (LE) theory which considers the rock mass to with f(x) a function of the actual length of the
be linear elastic, homogeneous, isotropic, initially joint (Lj1) and its maximum possible length (Sj2).
continuous and impermeable to the fluid. It neglects Joint propagation will only occur when the product
plastic yielding of the rock and is based on a of stress times the square root of the joint length at-
straightforward comparison of the stresses induced tains a critical value. LEFM analysis thus allows to
close to the joint end (by the hydrodynamic action) take into account the joint geometry based on a clas-
with the in-situ stresses σh and σH (minor and major sical linear elastic stress analysis. Furthermore, it
principal stresses in a plane perpendicular to the can characterize to some extent the processes of
joint) and the tensile strength σt of the rock mass. subcritical joint propagation, such as fatigue.
This statement determines the so-called “breakdown Regardless of the adopted failure criterion, the
pressure Pb” and is formulated in the following best way to obtain realistic results would be to per-
manner (Hubbert & Willis 1957, cited in Haimson & form classical hydraulic fracturing or jacking bore-
Zhao 1991): hole tests, in order to determine at which pressure
the pre-existing joints will propagate. This could be
done at different depths, depending on the rock
Pb = σt + 3σh – σH – P0 (1) structure STRUC and the layer depths Hl, and results
in static fracture toughness values KIc,stat for every
with P0 the local initial pore pressure. It has to be layer.
underlined that the zero breakdown pressure Pb0 cor- However, depending on the rate of application of
responds to the rock mass tensile strength σt and thus the hydrodynamic forces, physically correct joint
represents a constant rock mechanical property. propagation has to take into account dynamic ef-
This method is often used in order to estimate the fects, modifiying both the rock’s modulus of elastic-
in-situ horizontal stress field by means of vertical ity Er and its tensile strength σt. A rate of pressure
borehole hydraulic fracturing or jacking tests. For an raise Rp [MPa/s] can be taken into account by per-
uncracked rock, such tests are useful to determine in forming laboratory dynamic fracturing tests on rock
a direct way the hydraulic fracturing breakdown specimens (Haimson & Zhao 1991, Zhao & Li
pressure Pb, without particular knowledge on the in- 2000).
situ stress field (near the surface often neglectable). It may be concluded that the use of rock joint
The second approach of rock joint tensile failure failure criteria in practice is particularly interesting
is based on linear elastic fracture mechanics when hydraulic fracturing or jacking tests can be
(LEFM). The major difference with LE theory is the carried out in-situ. They should be performed at dif-
assumption that the zero breakdown pressure Pb0 ferent depths, following the layered structure of the
gets joint size-dependent and thus is not a constant rock mass. The most reliable test results could be
material property anymore. The expression for the obtained by performing dynamic fracturing borehole
breakdown pressure becomes (Rummel 1987, cited tests, taking into account the influence of the rate of
in Haimson & Zhao 1991): pressure raise on the fracture toughness. But those
are difficult to realize. If no tests can be made, the
analysis has to be based on available values of simi-
Pb = KIc + k1σh + k2σH – P0 (2) lar rock formations regarding the tensile strength of
the rock mass, the local initial pore pressure and the
with KIc the “fracture toughness” (for plane strain in-situ minimum stresses. A first order estimate can
conditions and mode I fracturing) and k1 and k2 both nevertheless be obtained by neglecting the in-situ
parameters depending on the joint geometry. The stress, thus by only taking into account the tensile
breakdown pressure Pb can now be replaced by a strength and an initial pore pressure.
constant that gives an idea about the magnitude of
4.3 Persistent rock mass failure criteria In this equation, the pressure distributions above
The second type of failure criteria refers to totally and underneath the block have been spatially inte-
broken-up rock masses. The approach is based on grated in the calculation direction of the force.
the definition of a representative rock block geome- Destabilization and thus uplift of the rock block
is clearly governed by the pressure differences that
try, called the “characteristic block” (Figure 9b).
This block will be subject to a dynamic equilibrium can appear during the considered time interval. This
of all the forces acting on as a function of time. The pressure difference depends on two aspects that are
most relevant forces are: difficult to determine: first of all the relationship be-
1) The stabilizing force Gb, defined as the immerged tween the overpressure distribution and the plunge
weight of the rock block. pool macroturbulence, and secondly the influence of
the change of the joint width during block uplift on
2) Fo(x,t), which is defined as the force resulting
from the time and space dependent pressure dis- the pressure under the block at subsequent time
tribution acting over the block. This force results steps. The former depends on the ratio of plunge
from the macroturbulent pressure pattern in the pool eddy size compared to rock block size and can
plunge pool and can under certain conditions be be obtained by statistical analysis of model tests.
destabilizing when reaching negative values. This ratio continuously changes during the uplift
3) The stabilizing force which is expressed by the process, in function of the time evolution of the im-
shear force Fsh(t,ej). This force depends on sev- pulsion. As a first approximation, a conservative re-
eral parameters, such as joint roughness, aperture, sult is obtained by neglecting these pressure drops,
filler material, etc. It can be approximately as- therefore assuming a pressure under the block that is
sumed only depending on joint width and time. independent of the block movement, and by apply-
4) The most important destabilizing force results ing a maximum negative pressure at the upperside of
from the time, space and joint width dependent the rock block. This will result in an upper limit of
pressure distribution Fu (x, t, ej1(t), ej2(t)), acting ultimate scour hole depths.
along the joints under the block. This distribution A more realistic equilibrium of forces will be
is highly influenced by 2-phase transient phe- worked out by the authors, based on ongoing model
nomena such as oscillations, resonance, a.s.o. tests with open-end rock joints. This will allow
evaluation of the maximum possible impulsion on a
block in function of jet, pool and joint parameters.
Fo(x, t)

5 POSSIBLE APPROACH FOR A NEW

Gb
ULTIMATE SCOUR METHODOLOGY
F sh

Fsh(t,ej2)
(t ,
e j1

Based on the hydrodynamic action inside rock joints

)

(chapter 3) and on rock mass failure criteria (chapter

Fu(x, t, ej1(t), ej2(t)) 4), a possible new approach for better assessment of
ultimate scour hole development is outlined. At the
x time of writing, measurements of impulsion on rock
blocks were not yet completed. In result, the here
proposed methodology is restricted to non-persistent
Figure 10. Force balance on a characteristic block
rock masses.
The significant time dependence of all of the
above forces requires determination of the dynamic,
time dependent impulse on the rock block. The total
impulse I∆tpulse on the block during the time interval
∆tpulse of a certain pressure pulse is obtained by inte-
grating the net force equilibrium at every time step
dt, in a defined direction. This defines the final ve-
locity V∆tpulse and thus the total uplift height of the
block (m = mass of the block):

∆tpulse

I ∆tpulse = ∫ (F
0
u − Fo − G b − Fsh ) ⋅ dt = m ⋅ V∆tpulse (4)
 Vj, Dj 0 1 2 3 4 5 σadm, σmax [Mpa]

core
tailwater depth Plunge pool
Sj1 Sj2 2
Y
4
layer 1: persistent layer 1
shear layer
6
layer 2
layer 2: non-persistent 8
RQD,Nj,σh,σH,σt,P0

Joint set 1 Joint set 2 Hl 10

Pj1 Pj2
αj1 αj2 12 ULTIMATE
ej1 ej2 SCOUR DEPTH
Lj1 Lj2 14
φj1 φj2
σadm
16
σmax
Figure 11. Example of 2-layered rock mass showing the initial Y/Dj
plunge pool-rock mass interface (continuous line) and the cal-
culated ultimate scour elevation (dotted line).
An example of a two-layered rock mass is shown Figure 12. The maximum hydrodynamically induced stress
in Figure 11. The first rock layer is of the persistent σmax and the admissible rock mass stress σadm (in function of
type and has already been scoured out by the im- Y/Dj) graphically determines the ultimate scour hole depth.
pacting jet. It corresponds to the initial condition for
the analysis. The second layer is still non-persistent This analysis can be repeated several times,
and is formed by two main joint sets (Nj = 2). Each sublayer per sublayer, until the maximum hydrody-
joint set has its own parameters and will be loaded namic stress in the joint is less than the admissible
by a hydrodynamic action. This action is determined stress, thus not able to propagate the joint anymore.
step-by-step, firstly based on the ratio of plunge pool This comparison is illustrated with Figure 12. The
depth to jet impact diameter Y/Dj. Every step of the location where the two curves intersect indicates the
analysis considers a sublayer with a height corre- approximate elevation of the ultimate scour hole
sponding to the characteristic rock block size that bottom as a function of Y/Dj. If the initial stress in
will develop. For a given Y/Dj ratio, the extreme the rock is influenced only by overburden pressure,
positive pressure coefficient Cmax, as defined in Fig- progressing scouring development of course will
ure 8, is obtained. Multiplying this coefficient with change the initial stress conditions.
the incoming kinetic energy of the jet (depending on This approach is purely static and doesn’t take
the jet impact velocity Vj) automatically results in a into account the fatigue effects in the fracture grow-
(dimensional) extreme pressure and thus maximum ing zone in front of the fracture tip, nor for dynamic
stress σmax inside the joint. This extreme pressure effects induced by the rate of pressure raise Rp.
value will then be compared with the maximum ad-
missible stress in the joint σadm, obtained by in-situ
fracturing or jacking tests. 6 CONCLUSIONS

A possible new approach for better assessment of ul-

timate scour depth was outlined, based on a detailed
analysis of the physical-mechanical processes of hy-
drodynamic fracturing and jacking, i.e. rock joint
propagation, and hydrodynamic uplift of rock
blocks. Extreme pressure values inside 1D closed-
end rock joints were determined, based on experi-
ments carried out with prototype jet velocities.
These values were then introduced in a tensile fail-
ure criterion of the rock mass.
For the time being, the proposed general ap-
proach of ultimate scour depth evaluation only deals
with non-persistent rock masses and is based on a
graphical comparison of the maximum hydrody-
namical induced stress σmax with the admissible
stress σadm inside the joints, as a function of the ratio
of plunge pool depth to jet diameter Y/Dj.

REFERENCES

Bhallamudi, M.S. & Chaudry, M.H. (1990). Analysis of Tran-

sients in Homogenous Gas-Liquid Mixtures. Proceedings
of the 6th Int. Conference on Pressure Surges. BHRA. UK.
Bin, A.K. 1984. Air Entrainment by Plunging Liquid Jets, Pro-
ceedings of the IAHR Symposium on scale effects in model-
ing hydraulic structures: paper 5.5-1. Esslingen.
Bollaert, E. 2001. Spectral Density Modulation of Plunge Pool
Bottom Pressures inside Rock Fissures. Paper submitted to
the XXIXth IAHR Congress. Beijing. China.
Bollaert, E. & Schleiss, A. 2001. Air Bubble Effects on Tran-
sient Water Pressures in Rock Fissures due to High Veloc-
ity Jet Impact. Paper submitted to the XXIXth IAHR Con-
gress. Beijing. China.
Ervine, A. 1998. Air Entrainment in Hydraulic Structures: A
Review. Proceedings Institution of Civil Engineers, Wat.,
Marit. & Energy 130: 142-153.
Ervine, D.A. et al. 1998. Pressure Fluctuations on Plunge Pool
Floors. Journal of Hydraulic Research 35 (2): 257-279.
Ewalds, H.L. & Wanhill, R.J.H. (1986). Fracture Mechanics.
Netherlands: Delftse Uitgevers Maatschappij.
Haimson, B.C. & Zhao, Z. (1991). Effect of Borehole Size and
Pressurization Rate on Hydraulic Fracturing Breakdown
Pressure. Proceedings of the 32nd US Symposium on Rock
Mechanics. Rotterdam: Balkema.
Rajaratnam, N. 1976. Turbulent Jets. Netherlands : Elsevier.
Toso, J.W.; Bowers, C.E. (1988). Extreme Pressures in Hy-
draulic Jump Stilling Basins. ASCE. Journal of Hydraulic
Engineering, Vol. 114, N° 8: 829-843.
Wylie, E.B. & Streeter, V.L. 1978. Fluid Transients. USA:
McGraw-Hill Inc.
Zhao, Z. & Li H.B. (2000). Experimental Determination of
Dynamic Tensile Properties of a Granite. International
Journal of Rock Mechanics and Mining Sciences, Vol. 37:
861-866.