SECONDARY CURRENTS AND SEDIMENTS CONTROL BY DELTA VANES IN OPEN CHANNELS

Terunori Ohmoto, Hiroyuki Sugimoto and Norimitsu Nariai

Department of Civil and Environmental Engineering, Kumamoto University, 2-39-1, Kurokami, Kumamoto city, Japan

Fax/telephone:+81-96-342-3507 /+81-96-342-3507, E-mail:ohmoto@gpo.kumamoto-u.ac.jp

Abstract：The streamwise development of longitudinal vortices and their effects on the sediments were investigated experimentally for different vortex pair configurations. Three-components velocity measurements have been made in an open channel flow with a vortex pair artificially generated by two delta vanes protruding from the flat bed. The experimental results showed that delta vanes produced sand ribbons and controlled the longitudinal vortices, the mean motion of the vortex centers qualitatively followed a vortex filament model and the vortices with the flow depth spacing indicated a significant interaction between them.

Keywords: delta vanes, longitudinal vortex, sand ribbons, sediments

1  INTRODUCTION

At bends of a river, secondary currents of Prandtl's first kind generated by centrifugal force cause scour and sedimentation near outer and inner banks, respectively. In recent years, blade-like plates called vanes are installed on the bed near outer banks to settle the local scouring problems [1-3]. Unlike groynes, conventional bank erosion control works, which move the scouring position to the front face of the groyne by distancing the axis of stream from the riverbank, vane works control secondary currents at the bend to mitigate scouring. The vane works studied in this study are aimed at actively extracting secondary currents of Prandtl's second kind that generally exist in straight river courses due to anisotropy of turbulence, and at controlling main flow and sediments. They are similar to vane works used at bends because both are installed to control secondary currents but different because they control different types of secondary currents than the vanes at bends. An attempt was made to mitigate local scouring near the vanes by changing their shape to a triangle. The authors tried to identify the effect of a boundary on the stabilization mechanism for a pair of longitudinal vortices. The authors then made a theoretical analysis of stability of pairs of longitudinal vortices by a vortex filament model for cases where a pair of vortices existed in the flow field in an infinite, semi-finite or finite area. As a result, it was illustrated that a pair of longitudinal vortices with the flow depth spacing that had an axis in the main flow direction and were aligned in alternate directions were stabilized by the influence of bottom surface and of free surface. In this study, focusing on the above points, the authors artificially generated secondary currents with delta vanes and made an experimental analysis about the possibility of control of secondary currents and sediments.

2  TEST APPARATUS AND PROCEDURE

For testing, a variable-gradient model channel 10 m long and 40 cm wide was made of acrylic resin. In a movable-bed test, almost uniform silica sand of a medium diameter d50=0.94 mm was used as bed material. The material was laid throughout the channel for a thickness of 6cm. Delta vanes made of 5-mm-thick acrylic boards in the form of a rectangular equilateral triangle were placed at a height ∆of 3 cm from the initial bed, at an attack angleθof 10 degrees and with the spacing obtained by multiplying the initial flow depth by a certain integer as shown in Figure 1. In the test, water was supplied under the above boundary conditions and hydraulic conditions listed in Table 1. Flow depth was measured with an ultrasonic depth sounder to identify bed shape after 120 minutes passed since the start of water supply. The amount of sediments was measured with a sand trap consisting of compartments with the spacing of 1 cm. Changes in flow velocity were measured in two cross sections 8 and 20 cm downstream of the vanes under the hydraulic conditions listed in Table 2. A two-component electromagnetic current meter was used as a sensor. Velocity components in the direction of main flow and in the vertical direction, and components in the main flow and transverse directions were measured at the same time at one and the same point. The results of measurement were converted from analog to digital data at a sampling frequency of 20 Hz. Then 2,048 data per measuring point were processed statistically. O (h) in the Vane alignment column in the table indicates a case where vanes were aligned with the spacing of flow depth h, vanes opened toward downstream and the spacing was gradually increased in the flow direction. C (h) is a case where the spacing between the vanes was identical to flow depth h and was gradually decreased in the flow direction.

3  CONTROL OF SEDIMENTS

A coordinate system is shown in Figure 1. The position of the channel bed at the end at the midpoint between vanes was defined as the origin. The x-axis was in the main flow direction. The y-axis was transverse to the flow direction. The z-axis was vertical. Photos 1 through 4 show plan views of sand wave after 240 minutes of water supply for the vane alignments shown. While Photo 1 for vane alignment C (h) shows a sand ridge at the center of the channel downstream of the vanes, Photo 2 for vane alignment O (h) shows a sand trough at the same position. Thus the bed heights in the two cases were of opposite phase compared to each other. While with vane alignment C (h), the width of sand ridge decreased as water flowed down, the sand trough became wider with alignment O (h). This suggests, based on an analysis with the vortex filament model reported earlier [4] and hydrogen bubble tests by Smith & Schwartz [5], that longitudinal vortices with the flow depth spacing came closer to each other for vane alignment C (h) and separated from each other for vane alignment O (h) in the flow direction. The local scour around the vanes was larger with O (h) than with C (h) shown in Figure 2(7 photos). Photos 3 and 4 show bed configurations corresponding to V8C (h) and V8O (h), respectively. As seen from the ridges in Photo 3 and the troughs in Photo 4 at the center of the channel, these bed configurations were of opposite phase compared to each other. Figure 2 shows bed configurations at x=40 cm when the spacing between two vanes was gradually reduced in the flow direction from h (initial flow depth) to 4h. With the vane spacing of h or 2h, an outstanding sand ridge was formed parallel to the main flow at the center of the channel or at y=0. The wave height was approximately 5 mm. With the vane spacing of 3h or 4h, the figure shows, a sand trough was formed at the center of the channel and two sand ridges were formed on both sides of the trough. Figure 3 shows bed configurations at x=40 cm when the spacing between two vanes was gradually increased in the flow direction from h (initial flow depth) to 4h. The figure shows that a sand trough was formed between the vanes placed with the spacing of h or 2h. With the vane spacing of 3h or 4h, a sand ridge with a wave height of 2 mm was found at the center of the sand trough near y=0. Figures 2 and 3 show that when the vane spacing exceeded 3h, the distribution of bed heights started changing and interaction of longitudinal vortices with the flow depth spacing caused by the vanes weakened.

4  FLOW CONTROL

Flow around vanes

Photos 5 through 7 visualize flow pattern around vanes by a dye-injection method using condensed milk. Condensed milk was applied thinly on the channel bed near the point approximately 10 cm immediately upstream of the vanes. The condensed milk near the upstream wall did not enter the bed, which turned black, in areas before and after the vanes. In front of the vanes, horseshoe-shaped vortices as if in front of a column were expected to exist. With vane alignment C (h) in Photo 5, the width of the dye that passed the vanes narrowed in the flow direction, and the color was clean white. With vane alignment O (h) in Photo 6, the dye that passed the vanes widened, became paler and was spread in the flow direction. This corresponds to the bed configurations in Photos 1 and 2. With vane alignment C (h), a pair of longitudinal vortices rotating in opposite directions produced upward flow at the midpoint between the vanes. With vane alignment O (h), a pair of longitudinal vortices rotating in opposite directions produced downward flow at the midpoint between the vanes. It is conceivable that interaction between a pair of vortices and the channel bed reduced the distance between the longitudinal vortices for vane alignment C (h) and that the interaction also increased the distance between the longitudinal vortices for vane alignment O (h). With vane alignment V8O (h), four clear lines ran parallel to the main flow downstream of the vanes, which indicate the existence of stable longitudinal vortices with the flow depth spacing transverse to the channel.

Main flow velocities and secondary currents

Figures 4 through 7 show variations in main flow velocity and secondary current. Main flow velocities and secondary currents were investigated using two delta vanes under the hydraulic conditions in case 11 shown in Table 2 while varying the distance between the vanes and the direction of the vanes. In Figure 4, vanes were located at y=±2 cm. The centers of a pair of longitudinal vortices with the flow depth spacing were located at y=±2 cm and z=±2 cm, almost at the same position either at x=8 cm or 20 cm. The areas that turned black in Photos 5 and 6 as no condensed milk entered are found to be in the place where downward flow occurred. Secondary currents were more outstanding with vane alignment C (h) than with O (h). The upward flow experienced no reduction in main flow velocity at x=8 cm under the influence of the flow behind the vanes. At x=20 cm, however, main flow velocity, influenced by secondary currents, was the lowest in the upward flow section and highest in the downward flow section. The flow velocity distribution was similar to that at the field of flow on beds with a longitudinal ridge [6]. Figures 5 through 7 suggest that the centers of longitudinal vortices with the flow depth spacing corresponded to the locations of vanes and that reduction in interaction between the vortices led to attenuation. While either of the two longitudinal vortices took a form of an explicit cell, the secondary currents became weaker as the spacing between the vanes increased.

5  CONCLUDING REMARKS

Longitudinal vortices with the flow depth spacing, although weak in shear flow, generate   three-dimensional flow pattern. They persist for a long length in the main flow direction with little attenuation. They exist as mean flow. They, therefore, play an important role in the transport of momentum or materials and in the ecosystem. In this study, triangular vanes were installed on the bed at a certain attack angle to the main flow to extract cellular secondary currents embedded in an open channel turbulent flow and to control flow and sediment transport. The results are summarized below.

(1) When the spacing between vanes was gradually reduced in the flow direction, an outstanding sand ridge parallel to the main flow was formed between vanes placed with the spacing of h or 2h. When the spacing was 3h or 4h, a sand trough was formed between vanes and two sand ridges appeared on both sides of the trough. When the spacing between vanes was gradually increased in the flow direction, a sand trough parallel to the main flow was formed between vanes placed with the spacing of h or 2h. When the spacing was 3h or 4h, a sand ridge was formed between vanes. When the spacing between vanes exceeded 3h, the distribution of bed heights started changing, suggesting the weakening of the interaction between longitudinal vortices with the flow depth spacing induced by the vanes.

(2) Local scouring around vanes was small. Scouring was somewhat large for vane alignment O (h), which could be mitigated by combining vanes.

(3) With vane alignment C (h), a pair of longitudinal vortices rotating in opposite directions produced upward flow at the midpoint between the vanes. With vane alignment O (h), a pair of longitudinal vortices rotating in opposite directions produced downward flow at the midpoint between the vanes. In the flow immediately downstream of vanes, interaction between a pair of vortices and the channel bed reduced the distance between longitudinal vortices for vane alignment C (h) and that the interaction also increased the distance between the longitudinal vortices for vane alignment O (h).

(4) It was suggested that the centers of longitudinal vortices with the flow depth spacing corresponded to the locations of vanes and that reduction in interaction between the vortices led to attenuation. While either of the two longitudinal vortices took a form of an explicit cell, the secondary currents became weaker as the spacing between the vanes increased.

(5) Flow visualization tests and point measurements showed that a pair of longitudinal vortices with the flow depth spacing artificially generated by delta vanes existed in a stable state when the spacing between the vanes was equal to the flow depth.

References

[1]  A.J.Odgaard and J.F.Kennedy: River-Bend Protection by Submerged Vanes, J.Hyd.Div, ASCE, Vol.109, HY8, pp.1164-1173, 1983.

[2]  A.J.Odgaard and J.F.Kennedy: Prevention of Local Scouring at River-Bend by Iowa Vanes, JHHE, JSCE, Vol.2, No.1, pp.1-13, 1984.

[3]  Shoji Fukuoka and Akihide Watanabe: Theoretical Study of Flow and Bed Profiles in Vane Installed Curved Channels, Proceedings of the Japan Society of Civil Engineering, JSCE, Vol.447/II-19, pp.45-54, 1992 (in Japanese).

[4]  Terunori Ohmoto and Muneo Hirano: Stability Mechanism and Control of Longitudinal Vortex Streets, Proceedings of Hydraulic Engineering, JSCE Vol.37, pp.495-502, 1993(in Japanese).

[5]  C.R. Smith＆S.P. Schwartz：Observation of streamwise rotation in the near-wall region of a turbulent boundary layer, Phys. Fluids, 26,No.3, pp.641-652, 1983.

[6]  Terunori Ohmoto, Muneo Hirano and M.S.Pallu: Three-dimensional Turbulent Shear Flow Structure over Sand Ribbons, Proceedings of Hydraulic Engineering, JSCE Vol.33, pp.529-534, 1989(in Japanese).

Table 1  Experimental condition in case of movable bed

                       (h:flow depth, U:mean velocity, I₀:channel slope, θ:angle attack)

Table 2   Experimental condition in case of fixed bed

Fig. 2    7 photos

Fig. 5  Contours of main flow velocity and secondary current vectors
 when the spacing of two vanes was h

Fig. 6  Contours of main flow velocity and secondary current vectors
when the spacing of two vanes was 2h

Fig. 7  Contours of main flow velocity and secondary current vectors
when the spacing of two vanes was 3h

Fig. 8  Contours of main flow velocity and secondary current vectors
when the spacing of two vanes was 4h