Local Scouring in Lowand High Gradient Streams at Bed Sills
Local Scouring in Lowand High Gradient Streams at Bed Sills
Local Scouring in Lowand High Gradient Streams at Bed Sills
Affouillement local dans des courants à faible et forte pente sur des seuils de fond
MARIO A. LENZI, Assoc. Prof., Dept. of Land and Agroforest Environment, Univ. of Padua, via Romea, Agripolis, 35020 Legnaro
(Padova), Italy
ANDREA MARION, Assist. Prof., Dept. of Hydraulic, Maritime and Geotechnical Engineering, Univ. of Padua, via Loredan 20, 35100
Padova, Italy
FRANCESCO COMITI, PhD Stud., Dept. of Land and Agroforest Environment, Univ. of Padua, via Romea, Agripolis, 35020 Legnaro
(Padova), Italy
ROBERTO GAUDIO, Research Eng., I.R..P.I , National Research Council (C.N.R.), Via Cavour, 87030 Rende (Cosenza), Italy
ABSTRACT
The main characteristics of local scouring downstream of bed sills, forming a staircase-like system in high-gradient streams with non-uniform alluvium,
have been investigated through 13 clear-water laboratory runs. Three initial longitudinal slopes and different flow rates were considered, keeping the
same distance between the baffles. The grain size distribution of the sediment is that of a real alpine torrent scaled to the model dimensions. The
measured scour depth, length and shape are compared to previous results concerning low gradient and uniform sediment gradings. A dimensional
analysis approach appears to remain valid; nevertheless some simplifications cannot be made, since the jet regime plays an important role both for the
depth and the length of the scour, and consequently affects the scour shape. Two equations are proposed for the estimation of the maximum scour depth
and length. The equations are from previous data sets on low-gradient tests and a new one of experimental results on high-gradient cases.
RÉSUMÉ
Les principales caractéristiques d’affouillement à l’aval de seuils de fond, formant un système en marches d’escalier dans des courants à forts gradients
avec des alluvions non uniformes, ont été étudiées à travers 13 essais de laboratoire en eau claire. Trois pentes longitudinales initiales et différents débits
furent étudiés en gardant la même distance entre les déflecteurs. La distribution des tailles de grains de sédiments est celle d’un réel torrent alpin à
l’échelle des dimensions du modèle. Les mesures de profondeur, longueur et forme de l’affouillement sont comparées aux résultats antérieurs concernant
de faibles pentes et des granulométries uniformes. Une approche par analyse dimensionnelle semble rester valable : cependant quelques simplifications
ne peuvent plus être faites car le régime de jet joue un rôle important à la fois pour la profondeur et la longueur de l’affouillement, et par suite en affecte
la forme. Deux équations sont proposées pour l’estimation du maximum de la profondeur et de la longueur de l’affouillement. Les formules proviennent
des ensembles de données précédentes sur les essais à faible pente, et une nouvelle formule des résultats expérimentaux dans les cas de forte pente.
Keywords: bed sills, channel erosion, local scouring, high-gradient streams, laboratory flume.
prevent the stream bed from being excessively degraded and in-
1. Introduction
cised ([5], [32], [33], [24], [16], [15]).
Mountain streams are often subject to channel incision. One Referring to mountain steep streams (torrents), peculiar hydraulic
method to stabilise them is to use a sequence of transverse grade- and sediment features are observed: highly heterogenous grain
control structures, or bed sills. Local scour downstream of the size distribution of the bed material and very low relative submer-
sills can endanger their stability and create the risk of failure if gence. Armouring processes and high roughness coefficients are
the foundations are not designed taking into account the maxi- the principal results of such conditions ([20],[21]).
mum scour depth. The issue concerning the choice of a representative diameter for
Local scour by free jets is difficult to treat theoretically due to the local scouring with graded sediment mixtures has not been com-
complexity of its dynamics. Consequently experimental studies pletely solved yet. Most authors suggest the choice of D90 as the
play a major role in relating the scour features (depth, length, effective diameter ([33], [6], [9]), some others indicate Dm (me-
shape, time-development) to the hydraulic and sediment vari- dian diameter) or D85 [27] instead. A very interesting evaluation
ables. Field measurements present logistic problems along with in the case of scouring by horizontal submerged jets was pre-
difficulties in evaluating hydraulic and sediment parameters, sented by Aderibigbe and Rajaratnam [1]. They found that the
therefore most of the research concerning this particular form of best correlation between the non-dimensional scour depth and
scour has been carried out using laboratory tests. The majority of length and the densimetric Froude number (the only parameter
the studies have addressed the problem of single, isolated drop seemingly affecting their results) was using D95. Such diameter
structures ([13], [25], [30], [27], [26], [6], [9], [17], [30], [10], was also found to be roughly the median size of the bed material
[28]); much less is known about the case of staircase-like systems inside the final scour hole.
of low check-dams or bed sills, the most popular technique to Another concern in predicting the scour depth and length for slop-
Revision received June 27, 2002. Open for discussion till April 30, 2003.
ys a
= 0.180 1 + 0.369 (12)
Hs ∆D50
Fig. 2. Grain size distribution curve.
which covers the range 1.3 a1/(∆D50) 9.1, with a correlation co-
efficient R=0.94. adopting a geomorphologic criterion ([11], [22]).
Similarly, adopting the same dimensional analysis for the scour In all the tests the initial bed was set at a constant slope, achieved
length ls, the authors obtained the following empirical relation by tilting the flume. The slope was determined from the differ-
(valid only for the gravel sediments): ence in elevation from a horizontal water surface. Three slopes
were tested: 0.0785, 0.1145 (which is the actual reach gradient)
ls a and 0.1480. The range was chosen in order to cover the most
= 1.87 1 + 4.02 (13) common slopes occurring in natural step-pool systems. Three
Hs ∆D50
wooden baffles (1 cm-thick, and the width of the flume) were
inserted into the granular fill of the flume. The distance between
which covers the same range, with R=0.94. the sills was 1.050 m. In order to guarantee long-term equilibrium
As in the other studies, the non-dimensional scour holes were conditions the test duration was set at 18 hours for all tests. The
found to be self-affine, with the maximum scour depth occurring duration was selected on the basis of direct observation and by
at a distance of 0.3·ls to 0.4·ls from the sill. using a video-camera to record the local scouring zone. An elec-
It is important to point out that in all these tests the scour length trical point gauge with an acoustic device was used to measure
was never long enough to be constrained by the following sill. the bed profile and the water surface elevation: for the former the
Thus the distance L did not produce effects of interference on the longitudinal interval was 2.5 cm in the scour hole and 5 cm else-
development of the scour. where, for the latter it was 10 cm. The profiles were taken along
The goal of the present research was to assess whether equations the centre-line only when the scour holes were two-dimensional,
similar to (12) and (13) were applicable in mountain high-gradi- otherwise two other longitudinal bed profiles were measured. The
ent streams with heterogenous grain size distribution, or whether maximum depth in the cross-section was also measured. The
two parameters are needed as expressed by Eq.(10). scour length was determined visually as the point where the equi-
librium slope was obtained downstream of the hydraulic jump
zone.
3. Experimental set-up and data processing
Although the initial bed was set-up with extreme care, the scour-
Thirteen clear-water tests were carried out using a mobile-bed ing process exhibited some asimmetry because the development
physical model set up in the Sloping Sediment Duct in the labora- of lateral bars produced a concentration of flow in a narrower
tory of HR Wallingford. This facility allows the easy modifica- section. This phenomenon was more significant with low to me-
tion of channel slope for a wide range of values (±65%). The dium flow rates. This is in agreement with the well-known result
flume is 5.57 m long, 0.6 m wide, 0.245 m deep, has a metal bot- that planform instabilities arise with low water depth/channel
tom and alternate glass and metal side walls [12]. width ratios.
The grain size distribution of the sediment used in tests is shown The dimensions of the scour holes were never large enough to
in Fig.2, where the characteristic diameters are also reported. It occupy most of the distance between the sills. Therefore no inter-
was a very heterogeneous mixture, ranging from coarse sand to ference between the sills occurred during the scouring process
small cobbles. The geometric standard deviation σ g = D84 / D16 = ([8], [23]).
5.38 was much greater than the threshold proposed by Breusers The maximum scour depth, ys (Fig. 1), was calculated down-
and Raudkivi [7] for the definition of nonuniform gradings stream of the central sill, identifying the cross-section where the
σg=1.35. The relative submerged density was ∆=1.63. maximum depth occurred and computing the cross-section aver-
The size grading reproduces at a 1:40 scale the sediment distribu- age depth (on 13 values). The equilibrium slope Seq was evaluated
tion of a reach of the Maso di Spinelle Torrent, a major tributary from the bed profiles. The values of the sill step a2 were also
of the Brenta River, in Valsugana (Trentino region, Italy). Along evaluated from the bed profiles. Once the equilibrium slope was
this stream several boulder check-dams have recently been built, obtained, the morphological jump a1 was calculated using Eq. (2).
Tab. 1. Initial condition and measured quantities, (hsill, flow depth at the downstream edge of the sill; for the
sediment characteristics, see Fig. 2)
Test Q q S Seq L ys ls hsill a2
(m3/s) (m2/s) (m) (m) (m) (m) (m)
H1 0.0111 0.0185 0.0785 0.068 1.050 0.050 0.21 0.024 0.007
H2 0.0125 0.0208 0.0785 0.065 1.050 0.066 0.30 0.027 0.010
H3 0.0143 0.0238 0.0785 0.053 1.050 0.082 0.38 0.028 0.010
H4 0.0165 0.0275 0.0785 0.046 1.050 0.095 0.42 0.032 0.015
H5 0.0175 0.0292 0.0785 0.044 1.050 0.106 0.42 0.033 0.020
H6 0.0044 0.0073 0.1145 0.096 1.050 0.035 0.15 0.013 0
H7 0.0075 0.0125 0.1145 0.088 1.050 0.064 0.19 0.018 0
H8 0.0098 0.0163 0.1145 0.076 1.050 0.075 0.25 0.021 0
H9 0.0125 0.0208 0.1145 0.062 1.050 0.106 0.30 0.026 0.004
H10 0.0143 0.0238 0.1145 0.053 1.050 0.122 0.35 0.029 0.004
H11 0.0040 0.0067 0.1480 0.104 1.050 0.071 0.18 0.013 0
H12 0.0060 0.0100 0.1480 0.090 1.050 0.095 0.20 0.017 0
H13 0.0090 0.0150 0.1480 0.073 1.050 0.133 0.30 0.022 0.004
a1 a1 ys ls
Test Hs (m) a1 (m)
Hs ∆D95 Hs Hs
H1 0.049 0.011 0.225 0.161 1.020 4.28
H2 0.053 0.014 0.267 0.207 1.244 5.65
H3 0.058 0.027 0.462 0.391 1.416 6.56
H4 0.064 0.034 0.535 0.498 1.488 6.58
H5 0.066 0.036 0.546 0.529 1.597 6.33
H6 0.026 0.019 0.734 0.284 1.323 5.67
H7 0.038 0.028 0.737 0.406 1.696 4.90
H8 0.045 0.040 0.899 0.590 1.668 5.56
H9 0.053 0.055 1.039 0.805 1.998 5.65
H10 0.058 0.065 1.115 0.943 2.107 6.04
H11 0.025 0.046 1.861 0.675 2.861 7.05
H12 0.033 0.061 1.872 0.890 2.921 6.15
H13 0.043 0.079 1.848 1.150 3.121 7.04
The use of the D95 produced the best correlation, nevertheless this
turned out to be almost insensitive to the choice of the representa-
tive grain size.
In Tab. 3 are reported the experimental ranges of the two parame-
Fig. 3. Non-dimensional maximum scour depths as a function of ters a1/Hs and a1/(∆D95) for the previous low-gradient tests and
a1/(∆D50). the new steep runs.
3.0
2.6
2.2
Observed Values
1.8
1.4
1.0
0.6
0.2
0.2 0.6 1.0 1.4 1.8 2.2 2.6 3.0 3.4
18
Observed Values
14
10
2
2 6 10 14 18 22
Numerical example:
A mountain stream subject to incision is to be protected with bed
sills and the distance between the structures has been chosen to be
L=20 m: determine the clear water long-term maximum scour
depth and the scour length given the following torrent characteris-
Fig. 10. Similarity of the scour holes. tics:
– design discharge with 100-yr return period: Q100=30 m3/s;
– channel average slope: S=0.11;
cal point. – channel width: B=10 m;
Marion et al. [24], whose experiments were characterised by high – sediment characteristic diameters: D65=0.40 m; D90=0.75 m;
relative submergence ratio, Sr= hu/D84>10, found good agreement D95=0.90 m.
between the measured equilibrium slope and the one calculated
through Eq.(7) with θc=0.040 and D=D50, assuming that the Man- The following parameters are assumed:
ning’s coefficient could be determined using the expression: – Shields incipient motion parameter: θc=0.076;
– relative submerged density of the sediments: ∆=1.63.
D901 / 6
n= (16)
26 Calculation:
– unit width discharge: q =Q/B=3 m2/s;
On the contrary, the present tests feature very low submergence – Manning roughness coefficient (Eq. 17): n=0.063 s/m1/3;
ratios Sr, roughly ranging from 0.7 to 1.7, as found in natural – equilibrium slope (Eq. 7, with D65): Seq=0.057;
mountain torrents. Therefore Manning’s coefficient has been – morphological jump (Eq. 2): a1=1.06 m;
evaluated through the Keulegan’s equation developed for macro- – critical specific energy above the sill (Eq. 8): Hs=1.44 m;
roughness conditions: – non-dimensional parameters: a1/Hs=0.74; a1/(∆D95)=0.72;
– non-dimensional scour depth (Eq. 14): ys/Hs=1.59;
D901 / 6 – maximum scour depth: ys=1.59x1.44 m=2.29 m;
n= (17) – non-dimensional scour length (Eq. 15): ls/Hs=6.26;
15.1
– scour length: ls=6.26x1.44 m= 9.01 m.
which gives n=0.038. The critical Shields’ mobility parameter, θc, 6. Conclusions
was set equal to 0.076, following the indications found in the sci- Bed sill sequences are useful to prevent erosion in incised chan-
entific literature for coarse grain size distributions [19]. Finally, nels, but the local scour hole downstream of the structures needs
the most appropriate diameter to be used in Eq. (7) was assessed an adequate estimation, particularly in the case of high-gradient
by searching for different diameter the best fit between the exper- streams, where the scour depth is enhanced by the effect of larger
imental points and the theoretical curve assuming the above val-
ues for n and θc. This calibration process led to the identification 0.20
of the D65 (equal to 0.016 m) as the ‘‘characteristic’’ grain size
for the equilibrium slope. 0.15
Fig. 11 shows the measured equilibrium slope and the curve 5
expressed by Eq. (7) versus the unit discharge q: it can be ob-
S eq
0.10
served that the agreement is fairly good except the three runs with 0
the smallest discharges, which have submergence ratios Sr less
than 1.
The evidence that for these tests the equilibrium slope is less than
the predicted one might be due to the larger flow resistance in-
duced by surface effects (i.e. small hydraulic jumps in correspon-
dence with protruding cobbles) when the submergence drop be-
low the unity, as found by Bathurst ([3], [4]). Fig. 11. Measured and calculated equilibrium slope values as a func-
Better results in predicting the equilibrium slope might presum- tion of the unit discharge.