Vous êtes sur la page 1sur 10

Journal of Hydraulic Research Vol. 41, No. 4 (2003), pp.

417426 2003 International Association of Hydraulic Engineering and Research

Scour development downstream of a spillway Dveloppement dun affouillement laval dun dversoir
BIJAN DARGAHI, Assoc. Prof. Div. Hydraulic Engrg., The Royal Institute of Technology, Stockholm, Sweden. E-mail: bijan@kth.se

ABSTRACT The scouring process downstream of spillways is an important research topic of value in engineering practice. The objectives of the present experimental study were to examine the similarity development of scour proles, the controlling scour mechanism and predictions of the scour geometry. No experimental evidence was found in support of the similarity assumption for the scouring process. Scouring downstream of a spillway is controlled by large secondary ows and a vortex system induced by a hydraulic jump. The main dimensionless parameters controlling the scouring process are the relative operating head, the relative sediment size, and the relative roughness of the scour protective plate. The maximum scour depth and the rate of sediment transport can be estimated by using simple, power-type equations that relate the scour geometry to the controlling scour parameters. RSUM Le processus daffouillement laval des dversoirs est un sujet de recherche important au service de lingnierie pratique. Les objectifs de ltude exprimentale prsente ici taient dexaminer la similitude des prols, le mcanisme contrlant laffouillement et les prvisions de sa gomtrie. On na trouv aucune preuve exprimentale permettant de valider lhypothse dune similitude pour le processus daffouillement. En aval dun dversoir, laffouillement est contrl par des coulements secondaires importants et un systme de vortex induit par un ressaut hydraulique. Les principaux paramtres sans dimensions contrlant le processus daffouillement sont la hauteur de chute de fonctionnement relative, la taille relative de sdiment, et la rugosit relative de la semelle de protection. La profondeur maximum daffouillement et le taux de transport de sdiment peuvent tre estims au moyen dquations simples de type puissance qui relient la gomtrie de laffouillement aux paramtres de contrle de celui-ci.

Keywords: Scouring; spillways; scour similarity; scour equations; sediment transport; temporal scouring; hydraulic jumps; ow visualization; secondary ows. Introduction Background Local scour downstream of hydraulic structures such as low head and high head structures, spillways and culverts is an important research eld due to its signicant practical value. Scour induced by low head structures such as gates and low dams has been investigated by many researchers. These works can be classied into two groups. One group concerns scour depth equations and the other group deals with the ow and scour characteristics and the temporal behaviour of scouring. There are many scour depth equations that have been developed for different ranges of experimental parameters. Most of these equations relate the scour depth to the ow depth, discharge and sediment size. One of the earliest equations was given by Schoklitsch (1932). He used model tests to propose two scour depth equations for overow and underow gates with short horizontal sills. The equation for the overow case reads: Z + Zo = 4.75H0.2 q0.57 D0.32 90 (1) In which Z = scour depth, Zo = downstream ow depth, H = difference between upstream and downstream water surfaces, q = ow discharge per unit width and D90 = sediment size in mm (90% passing by weight). Equation (1) is in metric units and is intended for coarse sediment in the range D90 = 9 36 mm. Jaeger (1939) reanalysed previous experimental data to arrive at a similar form of scour depth equation (2) for plunging jets. Z + Zo = 6H0.25 q0.5 H D90
0.33

(2)

A similar form of scour depth equation was given by Eggenberger (1944) for a gate with underow and overow Z + Zo = CH0.5 q0.6 D0.4 90 (3)

In which C is a constant that depends on the ratio of the underow to the overow discharges, being 22.8 when the underow is zero. In Eq. (3) all the variables are in metres with exception of D90 which is in millimetres. Shalash (1959) extended the previous works by including the effect of an apron with length L for an

Revision received February 17, 2003. Open for discussion till December 31, 2003.

417

418

Bijan Dargahi

underow structure. The proposed equation reads: Z + Zo = 9.65H0.5 q0.6 D0.4 90 1.5H L
0.6

(4)

Equation 4 is based on D50 = 0.7 mm, 1.9 mm and 2.65 mm and q = 0.01 0.07 m2 /s. The studies on the ow eld and temporal variation of the scour hole are more limited in comparison to the rst group. Experiments show (e.g. Breusers (1966) and Dietz (1969)) that for a given geometric arrangement, the shape of the scour hole is almost independent of the ow velocity and bed material at the equivalent values of Zo /Zmx (Zmx = maximum scour depth). The ow eld depends on the type of the ow. In a two-dimensional scour hole the ow is similar to that of a turbulent mixing layer. The ow character is dominated by vortices in case of a threedimensional situation. The temporal variation of the scour depth was investigated by Breusers (1966) and Kotoulas (1967). Breusers (1966) suggested a power type relationship for computing the temporal variation of scour depth. Kotoulas (1967) found that in case of coarse sand about 64% of the nal scour occurred in the rst 20 s, and about 97% of scour depth was attained in 2 h. Novak (1955) studied the inuence of a stilling basin on local scouring. He found that the scour depth could be reduced to 45 to 65% if the basin is long enough to contain the hydraulic jump. In a recent work Farhoudi and Smith (1985) studied the scour proles downstream of a spillway. They found a similarity relationship for the temporal variation of the scour proles. A detailed review of the foregoing works and further studies are given by Simons and Sentrk (1977). Further information can be found in the recent design manual for hydraulic structures issued by IAHR (Breusers and Raudkivi (1991)). Generally, scouring downstream of hydraulic structures is associated with a jet type ow characterised by high velocity and velocity decay laws. The physics of jetow induced scouring is less complex than the scouring developed around hydraulic obstacles such as bridge piers and abutments. However, an important common feature in all local scouring problems is the existence of strong secondary ows. Secondary ows are generated during ow separation or interaction with ow obstacles such as bridge piers. Secondary ows induce complex vortex structures that are the main scouring agents in local scour problems. Experiments have served to improve understanding of the physics and to yield useful relationships to predict the maximum scour depths. Regarding bridge pier and abutment scouring, the threedimensional nature of the scouring process has been thoroughly investigated (Dargahi (1990)). In the case of scouring downstream of hydraulic structures, most investigations have been limited to two-dimensional planes.

similarity development of the scour proles; second, the nondimensional scour vs. time function; and third, the controlling scour mechanism. Similarity of the scour proles is an important consideration, because if it exists it could provide a means to predict the time factor. Many local scouring results are based on small scale model tests, which are excellent tools to study the physics of the problem and to nd useful relationships. However, scaling problems occur when one needs to know the temporal scour development in a prototype situation. The difculty would be overcome if a temporal similarity relationship could be found. Previous research on local scouring has not yet resulted in such a relationship, although in a few cases where the ow can be simplied a solution is possible. The present study investigates the temporal variation of the scour proles downstream of a spillway. As will be shown no evidence was found of the existence of a similarity relationship. In this respect, the results do not conrm the ndings of Farhoudi and Smith (1985). Dimensionless parameters The ow eld and local scouring downstream of a spillway with an apron (bed protection plate) depends on a large number of ow and sediment variables. These variables can be grouped into the following dimensionless parameters (5) by use of dimensional analysis: h o ks ho Ls u , , Fr, Re, , , , hd ho D m h o u c (5)

in which h0 = operating head, hd = design head, ks = spillway roughness, Fr = Froude number, Re = Reynolds number, Dm = characteristic sediment mean size, Ls = length of the protection plate (apron), u = shear velocity, uc = critical sediment particle velocity and = geometrical standard deviation of sediment sizes. The rst four parameters are related to the ow over the spillway and the remaining parameters are related to the scouring process. The present study showed that the operating spillway head (h0 ) and the sediment size (D50 ) were the dominating variables for local scouring. Experiments The experiments were conducted in a 22-m-long, 1.5-m-wide (B = 1.5 m) and 0.65-m-deep ume. A xed layer of uniformly graded ne sand covered the channel bed. The dimensionless wall roughness height dened by Y+ = u ks / ( = kinematic viscosity) was approximately 1.8 (ks = 0.36 mm), so that the surface was effectively smooth. Scouring tests were carried out in a 0.26-m-deep, 4-m-long and 1.5-m-wide sediment recess section situated 16.5 m from the inlet. The ow entering the test section was uniform within 1% of the free stream velocity and the turbulence intensity was 5%. An overow spillway of 0.205 m crest height was placed at a distance of 16.5 m from the inlet. The spillway was designed according to the Waterways Experiment Station (WES) standard (USArmy Corps of Engineers (1952)) for a design head of hd = 0.1 m. A Perspex plate was placed downstream of the spillways toe having dimensions 1.5 0.615 m

Objectives The present study is an attempt to extend the previous experimental works by providing detailed 3-D measurements downstream of a spillway. Three issues were of major interest: rst, the

Scour development downstream of a spillway

419

0.205m

Flow

hO
Protection plate
Z

ZO X -Z Z
mx

Scoured surface Control gate 0.26m


Lower basin

LS= 0.615 m 16.5 m

sediment recess section 4m

Figure 1 The details of the test section in a vertical plane i.e. xz-plane (not to scale).

(Fig. 1). To measure the temporal variation of the bed, a specially designed instrument called here a prole-metre was used. The prole-metre was mounted on a motorized carriage systems. The XY-coordinates were measured by two potentiometers attached to the carriage. The origin of the coordinates was at the end of the plate touching the right ume wall (looking in the ow direction). The experimental variables were the sediment size, the plate roughness and the ow discharge. Two different bed materials were used, one ne sand with D50 = 0.36 mm (D90 = 0.50 mm) and the other medium size gravel with D50 = 4.9 mm (D90 = 7 mm). Both materials were effectively uniformly graded, the values of geometric standard deviations begin less than 1.4. Two series of experiments were carried out, one with a smooth plate and the other with additional roughness on the plate, in the form of hexagon nuts glued to the plate. Each nut was 1.1 cm high and 2.0 cm wide. They were placed in three staggered rows with a spacing of 7.5 cm between each row and between each nut. The rst row was 10 cm downstream of the spillway. The ow discharge was varied from 20 to 100 l/s in 20 l/s intervals. Table 1 shows the key ow parameters. Each test is characterized by the material type, ow discharge, plate roughness and the section location. For example S20R means, the bed material is sand (S), the ow discharge is 20 l/s, the protection plate has added roughness (R). S20R-250 (in case of Fig. 8) refers to S20R results at a section located at y = 250 mm or Yn = y/B = 0.167.

Each test was run for 8 h, which was sufcient for most of the tests to reach a quasi equilibrium state of scouring. However, in tests S40, S60, and G100, 8 h might not have been sufcient to reach equilibrium. This was because the bed sediment thickness (0.26 m) was not sufcient to retain the maximum scour depths in these tests. The measurements were computer based and consisted of temporal bed proles, water surfaces and ow visualizations. The bed proles were measured continuously during 8 h both in longitudinal (X) and spanwise (Y) directions. Each measurement set consisted of seven longitudinal sections at y = 0.033B, 0.167B, 0.333B, 0.5B, 0.66B, 0.833B and 0.967B and ten spanwise sections in the range y = 0.033B to y = 0.967B. The spanwise sections covered the entire scoured area. Water surfaces were measured upstream of the spillway, at the toe of the spillway, at the end of the scour protection plate, and along the longitudinal sections. To visualize the ow eld and the scouring process a potassium permanganate solution was added to the ow. A video camera was used to record the visualizations in both XY (horizontal) and XZ (vertical) planes. The video records were analysed by a digital image processing system developed by the author (for details see Dargahi (1997)). The rst two hours of each test were recorded continuously, as the scour process was intensive. Later, 10-minute video records were taken at 30 min intervals. At the end

Table 1 Experimental data.

Test S20 S40 S60 G20 G40 G60 G80 G100 S20R S60R G20R G60R G100R

Additional D50 roughness Scour material [mm] No No No No No No No No Yes Yes Yes Yes Yes Medium Sand Medium Sand Medium Sand Fine Gravel Fine Gravel Fine Gravel Fine Gravel Fine Gravel Medium Sand Medium Sand Fine Gravel Fine Gravel Fine Gravel 0.36 0.36 0.36 4.90 4.90 4.90 4.90 4.90 0.36 0.36 4.90 4.90 4.90

Q [l/s] 20 40 60 20 40 60 80 100 20 60 20 60 100

ho [mm] 38.8 56.4 73 38.5 57 71 83.2 95.8 38 72.8 37.0 70.6 96.2

hn = ho /hd (hd = 0.1 m) 0.388 0.564 0.73 0.385 0.57 0.71 0.832 0.958 0.38 0.73 0.37 0.706 0.962

420

Bijan Dargahi

of each experiment, the total amount of eroded material caught in a downstream basin was measured. The measurements were used to compute the geometrical properties of the scour cavity, the temporal rate of transport and the total load. Results and discussions The ow eld The main ow feature was the rapid development of a hydraulic jump from the initial super-critical condition. The character of this hydraulic jump differed from the classical stationary case where the bottom (bed) boundary is rigid. The mobile bed caused the jump to undergo variation both with respect to its kinematics and dynamic characteristics. The hydraulic jump resembled more a shock wave than a travelling wave that is typical of stationary jumps. The changes in the hydraulic jump were imposed as the sequent depths and the local Froude numbers changed during the scouring process. The hydraulic jump induced large secondary ows and clockwise vortices. Secondary ow regions resembling large ow circulating zones dominated the ow patterns, both in horizontal and vertical planes. The development of the scour cavity caused the ow to undergo separation at its edge. A clockwise vortex was created during the separation process. In addition, a small roller ow zone formed within the hydraulic jump. In contrast to the classic jump, the roller zone occupied a small portion of the jump, whereas the clockwise vortex has no counterpart in a stationary jump. To illustrate some of these features, reference is made to the video clips shown in Fig. 2(a)(j). The gures show the temporal development of the vortex system close to the ume side wall in a vertical plane (XZ) for test G100. The schematic sketches of the corresponding idealised ow lines are also included. A cyclic behaviour was observed that lasted for about 20 s. Figure 2(a), and (j) mark the start and the end of the cycle, respectively. During the initial stage (bc) a breaking tail water wave grew in height and reached the highest point of the hydraulic jump. The wave moved rapidly upstream and the approaching supercritical ow was not sufciently strong to push the rising tailwater downstream. Instead, the approaching ow was pressed down towards the bed, and vortices were created (d) through ow separation. During the rst six- second period (ad), the scoured prole had no inexion point. A double-slope scour hole was created by the vortex as it interacted with the bed material. Following this period, the vortex attained its maximum strength and was detached from the ow lines (e). The vortex formation was followed by a vortex breakdown stage (f) that created small vortices (f and g). No regular shedding frequency could be found for the vortices. Following the vortex breakdown, the roller ow grew in strength and nally the cycle was complete (hj). The secondary ow system induced a complex ow pattern in the horizontal plane (XY) that was composed of several ow circulation regions. The number of regions depended primarily upon the ow discharge and the bed material. A two-region circulating ow pattern developed in all tests except G20. The two ow regions were separated by a narrow section referred to here as the main channel where the ow direction was streamwise. An

Figure 2 Temporal development of the scour proles and the vortex system (test G100), xz-plane (video shot and schematic h : m : s).

Scour cavity

Despositation

Flow

Main scoured channel

Scour cavity

Despositation

Protection plate

Figure 3 Schematic layout of the two scoured regions and the secondary ows, xy-plane.

example is given in Fig. 3. The number of these regions increased to four in the gravel test G20. The increase could be due to a lower ow discharge and a larger particle size. An interesting aspect of the ow is that each scour hole can be considered as a separate ow region having a separate vortex system and secondary ows. The results also emphasize the need

Scour development downstream of a spillway

421

to consider the 3-D character of the ow eld. The author believes the ow division into several regions is related to ow instabilities that are inherent in the hydraulic jump. Once the instabilities interact with the mobile bed, scouring cavities develop. The ow then becomes conned to the cavities. No signicant change in the general ow patterns was observed when additional roughness was added to the scour protection plate. However, the vortex system became weaker and the hydraulic jump shifted upstream. Transport mechanism The main transport agents were the hydraulic jump and the induced secondary ows. The temporal variation of the bed caused signicant changes in the boundary conditions. These changes in turn led to complex transport patterns that included all the modes of transport. During the initial stage, the transport took place randomly. That is to say the intensity and duration of the transport varied irregularly. The irregularity was related to the strong unsteady nature of the hydraulic jump that maintained an oscillating position for about 30 min. Thereafter, the rate of transport decreased and the hydraulic jump resembled more the classical case. The scour cavities illustrated in Fig. 3 developed simultaneously in all the tests. Each cavity caused a local ow separation that created clockwise vortices. An example of the nal scoured surface for test S20 is shown in Fig. 4. Within a scour cavity two sediment transport regimes existed, one backward due to the reversed ow and the other streamwise. The mean position of the line separating the two regimes for sand tests is shown in Fig. 5. In the case of the gravel tests, the line moved upstream and the length ratios of the two regimes were changed from 7/8 to 4/5 and 1/8 to 1/5. The results imply that the larger the sediment size, the greater the portion of the reversed ow in the scour cavity. The general sediment transport pattern followed closely the two ow regimes. Some of the material carried by the reversed ow was deposited near the edge of the plate and some was transported out of the scour hole. The streamwise transport led to the formation of a mound downstream of the scour cavity. In addition to the foregoing transport patterns, a vortex-type transport was observed. The sediment particles were trapped in the vortex circular motion and signicant local erosion took place. The motion was maintained for several minutes until the vortex was convected downstream by the main ow. The vortex did not a xed position, and could be convected upstream by the reversed ow or downstream by the main ow. One issue was the transport mode and its dependence on the ow and sediment variables. The transport mode depended on the ow discharge and the grain size of the bed material. Generally, higher discharges led to a higher proportion of suspension or saltation. Different modes existed within the scour cavity. In the tests with sand, the sand particles were suspended in the reversed ow region, whereas in tests with gravel, saltation dominated. In both cases, vortex type transport was present. In the lower part of the scour cavity (Fig. 5: 1/8 and 1/5), rolling/gliding and saltation dominated the transport process. With increasing scour

Figure 4 The nal scoured surface (tf = 8 h), test S20 (video shot).

Flow

7 8

1 8

Figure 5 Schematic of the ow division in the scour cavity ( sand tests), xz-plane.

time, the ow velocities were reduced and the vortex weakened. The transport mode then changed from suspension/saltation to rolling/ gliding. Erosion under the protection plate was initiated by the moving vortex system, that caused scouring near the edge of the plate. Once the scouring started, the secondary side ows in the XY plane (Fig. 3) continued to erode the material under the plate. This process was counteracted by the reversed ow (XZ-plane), that transported eroded material from the scour holes towards the plate. The net effect was to limit scour under the plate. The extent of erosion under the plate depended primary on sediment size. In gravel tests, little erosion was observed. The additional plate roughness had a signicant inuence on the transport intensity. In all the tests the intensity reduced sharply and the scour cavities were 1857% smaller than the case with

422

Bijan Dargahi

no plate roughness (see scour depths in Table 3). The lower range corresponded to higher the discharge values.

Temporal scour variation The temporal variation of scour is an important issue that involves three aspects. One is the inuence of ow and sediment variables upon the development of the scour geometry and the maximum scour depths. The second issue is whether the scour proles follow a similarity law. The third issue is whether a general law could be found to estimate the maximum scour depths at a given time. To address these issues, large numbers of scour proles were analysed. Two representative longitudinal sections were selected, one centre line section (Yn = 0.5) and one side section (Yn = 0.167). The choice of the sections was related to the scour patterns shown

in Fig. 3. In addition, there were signicant differences between the two sections regarding the development of the proles. The scour proles at Yn = 0.167 had one minimum point, in contrast to section Yn = 0.5 which had several minima. The latter resembled large anti-dunes. Some examples are given in Fig. 6, which shows a set of scour prole measurements at the side section Yn = 0.167. The XZ-coordinates in these gures are normalized using the corresponding operating heads. The plots show the inuence of the operating head and the bed material upon the temporal development of the proles. With increasing operating head (ho ) or discharge, the scour cavity increased in size and its shape changed (decreasing values of Zn and Xn ). In the case of the sand tests, the shape of the cavity changed from an inverted cone into an inverted frustum for ho > 0.4hd . A similar change was observed for the gravel tests at section Yn = 0.5.

0 1 0 1 Zn 2 3 4 5
0 1 0 1 Zn 2 3 4 5 0 1 0 1 Zn 2 3 4 5 0 1 0 1 Zn 2 3 4 5

20

40

Xn

60

0 1 0 1 Zn 2 3

20

40

Xn

60

S20250 hn=0.38

4 5

G20250 hn=0.38

20

40

Xn 60

0 1 0 1 Zn 2 3

20

40

Xn 60

S40250 hn=0.56
Xn

4 5

G40250 hn=0.57

20

40

60

0 1 0 1 Zn 2 3 4 5

20

40

Xn 60

S60250 hn=0.73
20 40 Xn 60

G60250 hn=0.73
Xn

0 1 0 1 Zn 2 3

20

40

60

G80250 hn=0.83

4 5

G100250 hn=0.958

Figure 6 Temporal scour proles normalized with the spillway operating head at section Yn = 0.167, tf = 8 h.

Scour development downstream of a spillway

423

The scour geometry varied considerably as the bed material was changed from sand to gravel. In gravel tests, the scour cavity became smaller and the slope angles were reduced by 10 to 20%. A common feature in all tests was that 40% of the nal scour depth was reached after about 20 min or 4% of the test duration. One important issue was where in the scoured area the maximum erosion depth occurred. The maximum scour depths were found at the side section within a distance range 10Xn 13Xn (Xn = x/ho ). The range remained more or less independent of the ow and sediment variables. Regarding the sand test, slightly higher scour depths (5%) were recorded at the centre line section compared to the side section. The bed proles were characterized by upstream and downstream slopes given by angles and , respectively. An example for the gravel tests is given in Fig. 7. The live-bed transport conditions caused a strong periodic variation of the angles. The angles reached at the end of the test durations are summarized in Table 2. For the sand tests, the values of and decreased as the operating head increased, although the upstream slope was steeper than the downstream slope. This result agrees with the positions of the secondary ow eld in the upper region of the scour cavity, where secondary ow could maintain a steeper scour prole. For the gravel tests, the angles were reduced but the downstream slopes were steeper than the upstream slopes. Similarity among temporal developments of the scour proles implies that the proles could be collapsed into a single curve, if normalized using appropriate variables. A common procedure is to use the temporal maximum scour depths to normalize both the lengths and depths. Using this approach, the temporal scour

proles were normalized to investigate the validity of a similarity relationship. Some examples are given in Fig. 8(a)(e) (section Yn = 0.167). There is a considerable variation between the shapes of the proles with respect to time, operating head and bed material size. The use of other variables such as the equilibrium scour depths also did not produce the desired results. In conclusion, the author could not nd any experimental evidence in support of the similarity assumption. This conclusion contradicts the ndings of Farhoudi and Smith. According to their study, the scour proles downstream of a spillway are geometrically similar if normalized with the temporal maximum scour depths. Their results were based only on scour prole measurements close to the ume wall, whereas the present study considered the entire scoured area. In addition, the use of special instrumentation made it possible to collect data with high spatial and time resolutions. It can be speculated that the Farhoudi and Smith results are related to mean behaviour. In the opinion of the author it is unlikely that a similarity law can be found for problems involving more than one set of time and length scales. Despite the lack of similarity, useful results can be found that describes some mean behaviour. The temporal variation of the maximum scour depth (Zmx ) and the volume rate of transport are two examples. Figure 9 shows the variation of the maximum scour, where the coordinates were normalized with the maximum scour depth (Zf ) and time (tf ) reached at the end of the experiments. The data can be described with a power type equation as follows: Zmx tmx =A Zf tf
n

(6)

30

25

Slope angle , in degrees

20

15

10

0 0.0 0.1 0. 2 0.3 0.4 0.5 0.6 0. 7 0.8 0.9 1.0

t / tf G20, G60, G20, G80, G40, G80, G40, G100, G60, G100,

Figure 7 Temporal development of the upper and lower angles of the scour cavity-gravel tests.

Table 2 Scour cavity slope angles in degrees.

Test

S20 27 20

S40 24 23

S60 19 17

G20 11 19

G40 10 18

G60 13 22

G80 13 21

G100 16 22

S20R 21 13

S60R 23 17

G20R 8 7

G60R 14 18

G100R 17 14

424

Bijan Dargahi
b: S40-250
5 10 15
0 0.4 0.2 0 0.2 5 10 15

a: S20-250
0 0.4 0.2 0

z/Z mx, t

z/Z mx, t

0.2 0.4 0.6 0.8 1 1.2 x/Z mx, t

0.4 0.6 0.8 1 1.2

x/Zmx,t

c: S60250
0.4 0.2 0 0.2 0 5 10 15

d: S60R250
0 0.4 0.2 0 0.2 5 10 15

z/Z mx, t

z/Zmx, tt

0.4 0.6 0.8 1 1.2

0.4 0.6 0.8 1 1.2

x/Z mx, t

x/Z mx, t
10 15

e: S60750
0 0.4 0.2 0 0.2 5

f: G20250
0 0.4 0.2 0 5 10 15

z/Z mx, t

0.2
0.4 0.6 0.8 1 1.2

z/Zmx, t

0.4 0.6 0.8 1 1.2

x/Zmx, t

x/Z mx, t

Figure 8 Temporal development of the scour proles at section Yn = 0.167, normalized with maximum scour depth at time t (tf = 8 h).

in which the exponent n is a variable. Analysis showed that A 1 and that n can be given as a logarithmic function of the relative operating head ho /hd n = 0.17 ln ho hd + 0.04 (7)

scour time. The coefcient A and exponent n were found to be expressible as logarithmic functions of the relative sediment size ho /D50 ho D50 ho D50

A = 0.029 ln n = 0.115 ln

0.028 0.178

(9)

A similar power type equation was found for the volume of the eroded bed material t Vt =A Vf tf
n

(10)

(8)

in which, Vt = volume of the transported material at time t, Vf = volume of the initial bed material and tf = total

The correlation coefcients (r) for Eqs. (6)(10) were in the range 0.750.9.

Scour development downstream of a spillway


1

425

0.8

0.6

0.4

found, then a simple set of dimensionless equations were t to the data. The equations are expressed in terms of two main ow and scouring parameters; i.e. h0 /hd , ho /D50 . The maximum scour depths (Zmx ) and its location (Xmx ) are given in Eqs. (11)(14). Equations (11) and (12) are valid for the side section Yn = 0.167 and Eqs. (13) and (14) are valid for the centre line section Yn = 0.5 Zmx = h0
0 0.2 0.4

Zmx /Zf

0.2

h0 D50

1/3

(11)
1/12

t mx /tf

0.6

0.8

Xmx ho = 10 D50 h0 Zmx h0 = 1.7 ho D50 Xmx ho =5 D50 h0

(12)
1/4.5

Figure 9 Temporal development of the maximum scour depth measured, section Yn = 0.167. S20 S4 S60 G20 G40 G60 G80 + G100 S20R S60R.

(13)
1/3

(14)

Scour depth equations The maximum scour depth and its location downstream of a spillway are important design factors. To the knowledge of the author, there are no equations that deal with scouring depths downstream of a WES-type spillway. However, scour depth equations for various types of weirs, plunging jets, and stilling basins are available in the literature (see Breusers and Raudkivi, 1991). For comparison purposes some of these Eqs. (1)(4) were used to calculate the maximum erosion depths in the present experiments. Of these equations, the Jaeger (1939) equation gave a good estimate of the measured scour depths (Table 3). To develop scour depth equations, one customary approach is to use non-linear multi-regression analysis to t equations to experimental data. The method has the disadvantage that the resulting equation normally assumes a complex form and is difcult to use. Furthermore, the use of such an equation may require knowledge of a large number of variables. In the present work, rst the main parameters that control the scouring process were

Table 3 also gives the values of the maximum scour depths predicted by Eq. (13) for tests S40, S60 and G100. For these test, the maximum scour depths could not be recorded. One other issue is to predict the time needed to reach a certain scour depth. In this study the scour depth equal to the operating head is chosen. The time Ts required to reach a scour depth equal to the operating head, could be estimated by use of Eq. (15). Ts uc = 8 103 ho Conclusion The experimental study of scouring downstream of a spillway with an apron (a protective plate) led to the following conclusions. 1. The main ow feature was the rapid development of a hydraulic jump from the initial super-critical conditions. The ho D50
1/1.25

(15)

Table 3 Comparison of maximum scour depths at section Yn = 0.5.

Predicted using literature equations [m] Present study Test identity [m] S20 S40 S60 G20 G40 G60 G80 G100 S20R S60R G20R G60R G100R 0.2 >0.26 (0.3: Eq. (13)) >0.26 (0.4: Eq. (13)) 0.07 0.11 0.16 0.2 >0.26 (0.3: Eq. (13)) 0.1 >0.26 0.03 0.13 0.19 Stocholitsch (1932) Jaeger (1939) Eggenberg (1944) Shalash (1959) Eq. (1) Eq. (2) Eq. (3) Eq. (4) 0.33 0.49 0.61 0.12 0.19 0.23 0.26 0.29 0.33 0.61 0.12 0.23 0.29 0.16 0.26 0.36 0.04 0.08 0.11 0.13 0.15 0.16 0.36 0.04 0.11 0.15 0.99 1.48 1.87 0.31 0.51 0.63 0.73 0.82 0.99 1.87 0.31 0.63 0.82 0.25 0.37 0.45 0.06 0.12 0.13 0.14 0.15 0.25 0.45 0.06 0.13 0.15

426

Bijan Dargahi

2.

3.

4.

5. 6.

7.

jump induced large secondary ows and clockwise vortices. The vortices were created during the ow separation process. The secondary ow system caused several scour cavities to develop. The numbers of cavities depended upon the bed material and the ow discharge. The main transport agents were the hydraulic jump and the induced secondary ows. The temporal variation of the bed caused signicant changes in the boundary conditions. These changes in turn led to complex transport patterns that included all the modes of transport. No signicant changes in the general ow patterns were observed when additional roughness was added to the scour protection plate. However, the transport intensities were reduced sharply and the scour cavities became 1857% smaller than in the case with no plate roughness. No experimental evidence was found in support of a similarity assumption for temporal development of the scouring process. The main dimensionless parameters controlling the scouring process are the relative operating head, the relative sediment size and the relative roughness of the protective plate. Power type equations to predict the scour geometry were found in terms of the controlling scour parameters.

Xn = x/ho Xmx = x-position of Zmx Yn = y/B Z = scour depth Zn = Z/ho Zmx = maximum scour depth Zmx,t = maximum scour depth at time t Zf = scour depth at time tf = geometrical standard deviation of sediment sizes = kinematic viscosity References 1. Breusers, H.N.C. (1966). Conformity and Time Scale in Two-dimensional Local Scour, Proc. Symp. on Model and prototype conformity, Hydr. Res. Lab., Poona, India, 18. 2. Breusers, H.N.C. and Raudkivi, A.J. (1991). Scouring, Hydraulic Structures Design Manual, A. A. Balkema, Rotterdam. 3. Dargahi, B. (1990). Controlling Mechanism of Local Scour, J. Hyd. Engrg., 116(10), 11971214. 4. Dargahi, B. (1997). Generation of Coherent Structures in Turbulent Boundary Layers, J. Engrg. Mech., 123(7), 686695. 5. Dietz, J.W. (1969). Kolkbildung im feinem oder leichter Sohlmaterialen bei strmenden Abuss. Mitt. Theoder Rehbock Flussbaulab., Karlsruhe, Heft 155, 1119. 6. Eggenberger, W. (1944). Die Kolkbildung bein einen berstrmen und beider Kombination berstrmenUnterstrmen. Disseration, ETH Zrich. 7. Farhoudi and Smith. (1985). Local Scour Proles Downstream of Hydraulic Jump, J. Hyd. Res., 23(4), 343 358. 8. Jaeger, Ch. (1939). ber die hnlichkeit bei ussbaulichen Modellversuchen. Wasserwirtschaft und Wassertecknik 34, No 32/27, 269. 9. Kotoulas, D. (1967). Das Kolkproblem unter Berchsichtigung der Faktoren Zeit und Geschiebemischung im Rahmen der Wildbachverbauung. Diss. T.U. Braunschweig. 10. Novak, P. (1955). Study of Stilling Basins with Special Regard to their End Sill, Proc. 6th IAHR Conference, The Hague, paper C15. 11. Simons, D.B. and Sentrk. (1977). Sediment Transport Technology, Water Resources Publications, Fort Collins, Colorado. 12. Schoklitsch, A. (1932). Kolkbildung unter berfallstrahlen. Die Wasserwirtschaft, p. 341. 13. Shalash, M.S.E. (1959). Die Kolkbilung beim Aususs unter Schtzen. Diss. T.H. Mchen. 14. U.S. Army Corps of Engineers (1952). Corps of Engineers Hydraulic Design Criteria. prepared for Ofce of the Chief of Engineers, Waterways Experiment Station, Vicksburg, Miss.

Acknowledgments The rst series of the experiments were carried out by Matthias Bartzsch and Insa Will and the second series by Michael Bach. The reviewers provided useful advice and suggestions to improve the manuscript.

Notations B = ume width Dm = sediment characteristic size D50 = sediment size (50% ner) D90 = sediment size (90% ner) Fr = Froude number h0 = operation head hd = spillway design head hn = ho /hd ks = spillway surface roughness Ls = length of the scour protection plate Re = Reynolds number t = time tf = duration of the experiments tmx = scour time corresponded to maximum erosion depth Ts = time to reach a scour depth = ho u = shear velocity uc = critical shear velocity according to Shields Vf = volume of the initial bed material Vt = volume of transported material in time t XYZ = coordinate system in streamwise, spanwise and vertical directions