Plunge pool scour in prototype and laboratory

Stefano Pagliara
Department of Civil Engineering, University of Pisa, I-56 100 Pisa, Italy
Willi H. Hager
VAW, ETH-Zentrum, CH-8092 Zurich, Switzerland
Hans-Erwin Minor
VAW, ETH-Zentrum, CH-8092 Zurich, Switzerland

ABSTRACT: Scour is a major concern with flip buckets in hydraulic engineering, given the
several serious incidents over the past 50 years. The present research is based on previous
laboratory observations conducted at VAW, ETH-Zurich, Switzerland, to explore the effect of
cross-sectional jet shape, the effect of tailwater jet submergence and the jet impact angle on the
plunge pool scour features. It was observed that the maximum scour depth may be significantly
larger for so-called dynamic flow conditions than for static conditions when the jet action has
seized. This was attributed to large dynamic forces with a twofold effect: (1) Suspension of
sediment by the highly turbulent flow, and (2) Steeper scour hole slopes as compared with the
natural angle of repose. The following thus intends to relate the static end scour depth to the
maximum dynamic scour depth in order to allow prediction of the effective scour maximum for
prototype conditions.

1 INTRODUCTION

Plunge pools are economic means to dissipate hydraulic energy for high-speed flows, provided
the geological site conditions are favorable and the upstream jet velocity is larger than typically
20 m/s. For smaller velocities, stilling basins are often used, but these are highly unreliable for
larger velocities because of concerns with cavitation damage, large spray production, unstable
flow and tailwater wave generation (Vischer and Hager 1998). The limitations of plunge pools
are dictated by large scour depth, its sufficient distance from other hydraulic structures to inhibit
structural damages and control of spray action.
Plunge pools are usually at the end of either surface spillways or bottom outlets as applied in
dam engineering. The flood discharge for surface spillways is first conveyed over a gated or
ungated spillway crest that is followed by a chute. Once the discharge has reached an elevation
close to the dam foot, it is directed onto a flip bucket to be ejected into the air. Accordingly, an
air-water jet travels through the atmosphere and eventually strikes the plunge pool, in which the
hydraulic energy is dissipated, and a stilled tailwater may be conveyed to the downstream
receiving water course.
Plunge pools are currently a standard design in hydraulic engineering, although its scour
patterns are not yet fully understood. Therefore, a fundamental research project was initiated at
Versuchsanstalt für Wasserbau, Hydrologie und Glaziologie VAW, at ETH Zurich, Switzerland.
Its purposes was: (1) Set-up of a laboratory installation to experimentally investigate such
flows, (2) Define the main parameters of plunge pool scour and (3) Present design relations that
may be applied in dam engineering. An exploratory study was provided by Canepa and Hager
(2003), in which the complicated laboratory set-up was simplified in an air-water high-speed
flow generated with a circular pipe, thereby profiting from well-defined upstream boundary
conditions. Also, a three-phase Froude number was introduce that accounts for the effects of
water, air and sediment involved in the flow. Minor, et al. (2002) presented implications of this
research and applied results to hydraulic practice. Recently, Pagliara, et al (2004) presented a
large experimental study in which various points were further clarified. The main findings may
be summarized as follows: (1) The effect of jet shape on plunge pool scour is small provided the
cross-sectional average mixture velocity is considered, (2) the impinging jet angle has a
relatively small effect provided it is included between 30° and 90° measured from the
horizontal, and (3) there are distinct effects of both submergence of jet by the tailwater level and
pre-aerated flow characteristics. The research is accompanied with generalized scour hole
profiles and the effect of temporal scour depth development was also investigated. The present
paper would like to add to previous research by relating maximum scour depths measured
during scour progress, and as observed after jet flow was stopped. The motivation is simple:
The prototype scour hole may exclusively be observed when floods have ended, and the true
extent of prototype scour thus normally remains unknown. Using laboratory observations, this
unknown may be determined. The present laboratory data are then compared with prototype
observations to check the fit. A typical computational example is also added to explain the
design procedure.

2 DESCRIPTION OF LABORATORY OBSERVATIONS

The experiments were conducted in a laboratory channel of 0.500 m width, 0.70 m height and 7
m length. The jet was generated with pipes of different diameters ranging up to 0.100 m, in
which also nozzle elements were placed to vary the cross-sectional jet geometry. Sands and
gravels of almost uniform grain size of up to d=0.007 m were used to replace a completely
disintegrated rock surface, of which the height was up to almost 0.35 m. Air was added to the
pipe flow to simulate an air-water high-speed jet. Standard observations resulted in discharges
for water QW and air QA. The water surface was recorded with a point gage, whereas the
sediment surface was read with a special gage having a 0.04 m circular plate at its end to
measure the sediment surface to approximately d/2 (Canepa and Hager 2003).

a) b)

c) d)

Figure 1. Photographs of plunge pool scour (45° jet angle) development (a) close to scour initiation, (b)
close to scour end; view of a test with jet angle of 60°: (c) dynamic condition; (d) ‘dry’ conditions.
Prior to experimentation, the sediment was horizontally inserted into the test channel, and air
discharge was added by the in-house pressurized air supply system. Then, water discharge was
started, with the temporal start of a run set at the initiation of scour. It should be noted that scour
initiated immediately once the air-water mixture flow impinged on the plunge pool. Readings of
the sediment surface were made after 1, 5 and 20 minutes, and sometimes after longer time. For
the present observations, the end scour was normally reached after 20, or even after 5 minutes.
At the end of an experiment, the water flow was stopped, the jet at the same time deflected to
inhibit further scour action and the resulting sediment surface measured after water had been
drained from the test reach. This condition was referred to as the ‘dry’ scour measurement.
Figure 1 shows typical photographs close to scour initiation, and scour end.
It was at once obvious that the scour profile close to end scour conditions, thus under a dynamic
load, was significantly different from the ‘dry’ scour profile. Evidently, this difference results
due to two reasons: (1) During jet flow, a considerable amount of sediment is in suspension that
deposits as soon as the jet flow stops, and (2) much steeper slopes of sediment surface may
result due to dynamic jet forces, in addition to gravity. Figure 2 shows typical scour hole
profiles for four jet impact angles α and three subsequent observational times, in addition to the
‘dry’ condition. It may be observed that both the maximum scour depth increases with time, as
the maximum aggradation height increases to reach the final scour profile. Temporal changes
for α=30° appear to be larger than for α=90°. Also, the ‘dry’ scour profile reaches never the
scour depth under dynamic conditions, although the maximum aggradation is normally larger,
due to the deposition of suspended sediment after the jet being stopped. Note that differences
between the ‘dry’ or static condition and the dynamic condition are relatively small for small jet
angle, and increase as the jet angle α increases.
Figure 3 shows the cross-sectional shape of the scour hole under maximum scour depth. The
tests involved a jet angle of 45° and 90°. As for the longitudinal scour hole profiles, this plot
demonstrates the significant differences between the final dynamic and the dry scour hole
geometries. In certain case, the dry scour surface was practically horizontal, and one might have
thought that there was absolutely no scour action. Such a presumption may be dangerous
because the real extent of scour may be largely underestimated. Means to detect the maximum
scour depth during dynamic scour action are thus an important basis for adequate hydraulic
design.

3 EFFECT OF JET SHAPE

The effect of jet shape was further tested with a pipe nozzle 29 mm wide and 99 mm long in
both the horizontal and vertical positions, for a jet angle of α=30°. Figure 4 compares the
maximum scour depths for dynamic Zm and static Zs conditions, where Z=z/D and D=equivalent
pipe diameter. Note that the full symbols represent unsubmerged (U), and the open symbols
submerged (S) impact conditions into the tailwater. These tests were made for blackwater (BW)
flow conditions, i.e. there was no air addition to the water flow.
Figure 4 shows absolutely no shape effect because all respective data for submerged and
unsubmerged conditions lie on well defined, yet different curves. Also included is the line
Zs=Zm, from where it follows that all static scour depths are smaller than those under dynamic
condition. In the following these differences are determined, based on a large data series that
involved four jet impact angles, submerged and unsubmerged tailwater conditions and white
water (WW) and black water (BW) approach flow conditions. The ‘dry’ scour data are related to
the dynamic scour data here, because the effects of the previous parameters on the dynamic
scour depth have been determined previously by Pagliara et al. (2004).
 60 60

z (cm)
z (cm)

30 30

a) b)
y (cm) y (cm)
0 0
240 300 360 420 200 250 300 350

60 80

z (cm)
z (cm)

30 40

c) d)
y (cm) y (cm)
0 0
170 230 290 100 150 200

Figure 2. Scour profiles for dynamic and static conditions for (a) α=30°, Run BW30UB183, (b) α=45°,
Run WW45UB21, (c) α=60° Run WW60SC67, and (d) α=90° Run BW90UC89 at times (⋅⋅⋅⋅) 1, (-⋅-) 5,
(−−−) 20 minutes and (−−−) ‘dry’ condition. Circles in (a) to (c) show the impact region of the falling
suspended material when flow is stopped.

40 60
z (cm)
z (cm)

20 30

(a) (b)
BW45SA14(t=20') WW90SC77 (t=20')
BW45SA14(dry) WW90SC77 (dry)

0 0
0 25 x (cm) 50 0 25 x (cm) 50

Figure 3. Cross-sectional scour profiles for (−−−) dynamic and (−−−)static conditions, (a) α=45° (test
BW45SA14) and (b) α=90° (test WW90SC77).
 8
30°BWS
Zs
30°BWU
R1-30°BWU
R2-30°BWU
R1-30°BWS
R2-30°BWS

Zm
0
0 4 8

Figure 4. Shape effect for maximum static and dynamic scour depths, full symbols represent
unsubmerged, open submerged impact conditions.

4 STATIC VERSUS DYNAMIC SCOUR DEPTHS

The data relative to the static and dynamic conditions were further considered to propose a
generalized correlation. Figure 5 shows all the data available for the four jet angles α,
blackwater and whitewater conditions, and submerged and unsubmerged impact conditions. It
may be noted that the angle α=30° involves the smallest deviation, whereas α=90° results in
significant deviations between Zs and Zm, under otherwise identical conditions.
The data shown in figure 5 may be approximated with α [in deg.] as

Zs = 0.75Zmε, ε=Eα -0.75, 30°≤α≤90° (1)

where the coefficient E=14 for unsubmerged, and E=10 for submerged jet flow. Figure 6 is a
summary plot in which the previous blackwater tests are shown again, together with the
respective curves from (1). Accordingly, the impact angle has a significant effect on the
reduction of the static, as compared with the dynamic scour depth. The effect of tailwater
submergence is relatively small for angles α≥45°, but considerable for small impact angles α of
the order of 30°. Once the maximum dynamic scour depth Zd is known from previous research
(Pagliara et al. 2004), equation (1) may be applied to find the reduction of the static scour depth,
in terms of all other hydraulic parameters involved in the plunge pool scour features. Also, a
measured static scour depth recorded after a flood may be used to predict the effective dynamic
scour depth that had occurred.
Figure 7 shows the data in a more concise arrangement, by using the abscissa proposed in (1). It
is seen that the present data cover a wide range of the combined parameter Zmε, namely from 0.5
up to about 4. Both submerged and unsubmerged data lie evenly spaced along the straight of
perfect agreement, indicating a small bias of (1). Figure 7 may be considered a condensed plot
of a large number of data to determine the ratio of static to dynamic plunge pool scour depths.
The application of (1) to cases of practice is straightforward, once the basic scour parameters
such as impact angle α, impact jet velocity V, determining median sediment size d90, jet air
content β and equivalent jet diameter D are known. The following intends to compare prototype
data with (1).
 8 8
30°BWS 45°BWS
Zs Zs
30°WWS 45°WWS
30°BWU (a) 45°BWU (b)
30°WWU 45°WWU

4 4

Zm
Zm
0 0
0 4 8 0 4 8

8 8
60°BWS 90°BWS
Zs 60°WWS Zs 90°WWS
60°BWU 90°BWU
60°WWU (c) 90°WWU
(d)

4 4

Zm Zm
0 0
0 4 8 0 4 8

Figure 5. Maximum scour depths Zs versus Zm for α= (a) 30°, (b) 45°, (c) 60°, and (d) 90° and both
blackwater BW and whitewater WW conditions, and submerged, and unsubmerged jets.

9
45°BWS
Zs 45°BWU
60°BWS
60°BWU
90°BWS 30° U
6 90°BWU
30°BWS
30°BWU
Zs=Zm
45° U
30° S
3 60° U
45° S
90° U 60° S
90° S

Zm
0
0 3 6 9

Figure 6. Regression of data for unsubmerged U and submerged S conditions for different angles α
(eq.1).
 5
45°BWS
Zs 45°BWU
60°BWS
60°BWU
90°BWS
90°BWU
30°BWS
30°BWU

0.75Z m ε
0
0 5
ε
Figure 7. Condensed data plot Zs(0.75Zm ) according to (1) with (−−−) line of perfect agreement.

5 COMPARISON WITH PROTOTYPE DATA

The previous results were applied to typical cases reported in the literature, to check whether
agreement with the analysis may be obtained. Figure 8 compares data relative to Picote dam
(Portugal), Ukai dam (India), Kilickkaya dam (Turkey) and Cabora Bassa dam (Mozambique).
In all evaluations, an unsubmerged air-water jet was assumed to occur, as is typical for
hydraulic structures. The various parameters needed for the analysis according to Pagliara et al
(2004) were estimated based on Yildiz and Üzücek (1994) and Whittaker and Schleiss (1984),
with the jet air content β calculated according to Ervine and Elsawy (1987) and a determining
sediment size d90 of 0.30 to 2 m. It should be noted that this effect is relatively small and does
not lead to a significant modification of the results. The zm values were calculated by means of
Eq. (4) (Pagliara et al 2004), while the corresponding zs values originate from (1).
Figure 8 indicates reasonable agreement between prototype observations and prediction
according to laboratory measurements. The zs values calculated are close to the measurements
(black squares), while the zm values calculated (white squares) indicate a deeper scour during
flow as compared with the static condition.
It should be noted that the effects of jet impact velocity V and effective jet diameter D on both
the static and the dynamic scour depths are significant, because both are linearly related to the
maximum scour depth. Both of these parameters may be difficult to estimate, given that the
exact development of cross-sectional average blackwater velocity and the air entrainment
characteristics along the chute and over the flip bucket involve some degree of uncertainty,
typically of the order of some 20%. Also, the jet development across the atmosphere, from the
bucket to the tailwater impact is actually not simple to predict, given a serious lack of research
in this field. The present project may be considered a step forward in this direction.

6 CONCLUSIONS

Prototype scour may only be determined under static conditions, i.e. after a flood wave has
passed and the scour hole is accessible again. This research was concerned with the maximum
scour depth under dynamic conditions, based on an extended laboratory program. It was found
that the ratio between static and dynamic scour depths depends particularly on the jet impact
angle, whereas the effect of jet submersion is relatively small except for small impact angles.
The present results were verified with literature data of prototype static scour holes, of which
reasonable agreement resulted. The present results may thus be a basis of future scour hole
assessments, and can also be a design basis for plunge pools.

-60

Picote dam, 1962

z m, z s
(calc)[m]
-40
Kilickaya, 1990

-20
z s (calc)

Cabora Bassa
Ukai dam, 1974
z m (calc)

0 -20 -40
z s (meas)[m] -60
0

Figure 8. Comparison between prototype and calculated values of maximum scour depth zm and zs.

REFERENCES

Ahmad, I., Chishti, M.I., Rasheed, M.A. 1979. Tarbela left bank irrigation tunnel: Some special features
of design, construction and operation. 13th ICOLD Congress New Delhi Q50(R63): 13-32.
Canepa, S., Hager, W.H. 2003. Effect of jet air content on plunge pool scour. J. Hydraulic Engng.
129(5): 358-365.
Coleman, H.W. 1982. Prediction of scour depth from free falling jets. ASCE Conference Applying
Research to Hydraulic Practice Jackson MI: 298-304.
Ervine, D.A., Falvey, H.T. 1987. Behaviour of turbulent water jets in the atmosphere and in plunge
pools. Proc. Inst. Civ. Engrs. 83: 295-314; 85: 359-363.
Fahlbusch, F.E. 1994. Scour in rock riverbed downstream of large dams. Hydropower & Dams 1(4): 30-
32.
Häusler, E. 1972. Der Kolk unterhalb der Kariba-Staumauer. Tiefbau 12(10): 953-962 (in German).
Häusler, E. 1980. Zur Kolkproblematik bei Hochwasser-Entlastungsanlagen an Talsperren mit freiem
Überfall. Wasserwirtschaft 70(3): 97-99 (in German).
Kulkarni, V.N., Patel, I.C. 1981. Ski-jump spillway for India’s Ukai dam. Water Power & Dam
Construction 33(9): 44-48.
Lowe, J. III, Chao, P.C., Luecker, A.R. 1979. Tarbela service spillway plunge pool development. Water
Power & Dam Construction 31(11): 85-90.
Mason, P.J., Arumugam, K. 1985a. Free jet scour below dams and flip buckets. J. Hydraulic Engineering
111(2): 220-235; 113(9): 1192-1205.
Mason, P.J., Arumugam, K. 1985b. A review of 20 years of scour development at Kariba dam. The
Hydraulics of Floods & Flood Control Cambridge A5: 63-71.
Minor, H.-E., Hager, W.H., Canepa, S. 2002. Does an aerated water jet reduce plunge pool scour? Rock
scour due to falling high-velocity jets: 117-124, A.J. Schleiss & E. Bollaert, eds. A.A. Balkema: Lisse,
the Netherlands.
Pagliara, S. Hager, W.H., Minor, H.-E. 2004. Hydraulics of plunge pool scour. J. Hydraulic Engng.
Subm.
Quintela, A. de Carvalho, Fernandes, J. de Salvador, Cruz, A.A. da. 1979. Barrage de Cahora-Bassa:
Problèmes posés par le passage des crues pendant et après la construction. 13th ICOLD Congress
New Delhi Q50(R41): 713-730 ; 5: 536-540.
Sen, P. 1984. Spillway scour and design of plunge pool. J. Irrigation and Power India 41(1): 51-66.
Vischer, D.L., Hager, W.H. 1998. Dam hydraulics. John Wiley & Sons: Chichester.
Whittaker, J.G., Schleiss, A. 1984. Scour related to energy dissipators for high head structures. Mitteilung
73, Versuchsanstalt für Wasserbau, Hydrologie und Glaziologie, ETH-Zurich: Zurich.
Yildiz, D., Üzücek, E. 1994. Prediction of scour depth from free falling flip bucket jets. Water Power &
Dam Construction 46(11): 50-56.
Zvorykin, K.A., Kouznetsov, N.V., Akhmedov, T.K. 1975. Scour of rock bed by a jet spilling from a
deflecting bucket of an overflow dam. 16th IAHR Congress Sao Paulo 2: 418-422.