See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337274000

LOW-LEVEL DAM OUTLETS HYDRAULIC DESIGN OPTIMISATION FOR THE

FLUSHING OF SEDIMENT

Conference Paper · November 2019

CITATION READS

1 333

3 authors, including:

Gerrit Basson Eddie Bosman

Stellenbosch University Stellenbosch University
88 PUBLICATIONS   141 CITATIONS    19 PUBLICATIONS   37 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Real time modelling of irrigation releases in river systems View project

Lake St Lucia South Africa View project

All content following this page was uploaded by Gerrit Basson on 15 November 2019.

The user has requested enhancement of the downloaded file.

LOW-LEVEL DAM OUTLETS HYDRAULIC DESIGN OPTIMISATION
FOR THE FLUSHING OF SEDIMENT
LR van der Spuy1, GR Basson1 and DE Bosman1
1. Department of Civil Engineering, Stellenbosch University, Stellenbosch, South Africa.

PRESENTER: L.R. VAN DER SPUY

ABSTRACT

This paper aims to provide design guidelines for optimising a postulated configuration for a sediment
flushing system of a low-level outlet at a dam for control of reservoir sedimentation to protect the
hydropower intakes. The postulated design, which comprises of a hybrid low-level outlet conduit and
low weir, is optimized by means of numerical and physical modelling for effectively flushing sediments
during different flood and water level scenarios. The low weir (a semi-circular, ogee spillway-type
structure with wing-walls) is designed and positioned in the vicinity of the low-level intake. The main aim
of the structure is to produce desirable supercritical flow conditions upstream of the conduit intake, as
well as through it, in order to help optimise the flushing of sediments during free-surface flow conditions.
The submerged weir is however also designed for pressure flushing where the aim is: (1) local sediment
removal; and (2) preventing main reservoir sediments from depositing near the outlet gate and hindering
gate closure.

1. INTRODUCTION

Reservoir sedimentation is an ongoing critical concern worldwide with a recorded global average of 33%
of reservoir storage capacity already been lost therefrom, which is expected to exceed 50% by 2050
(i.e. not considering new dam projects) (ICOLD, 2009). Drawdown or pressure flushing through low-
level dam outlets below hydropower intakes are common mitigation measures for removing locally
deposited sediment from reservoirs. This is periodically necessary to especially keep the hydropower
intakes free of non-cohesive sediment, as sand fractions will typically damage turbines in hydropower
plants. The low-level intake area also needs to be kept clear of blockage caused by debris and sediment
deposition as this could threaten sluice gate operation, prevent drawdown flushing and ultimately result
in the uncontrolled emptying of the reservoir. There are various international cases of existing dams
where little provision has been made for the flushing of coarse, large-diameter sediments (i.e. gravel
and boulders) that occur at the respective dam sites. This lack of provision implies that certain low-level
outlet designs may be deficient for such cases, especially during flood events where high sediment
concentrations are expected. Boulders can quickly become a notable concern in small reservoirs that
fill up quickly with sediment. At larger reservoirs, large-diameter sediments can also accumulate near
hydropower intakes once the reservoir has filled with sediment. It is therefore proposed that new
optimised, versatile and robust low-level outlet designs need to be implemented in future dams to ensure
suitable dam operation efficiency and sustainability. In order to address these issues, a literature study
of the hydraulic designs of existing low-level outlet conduits and intakes that have resulted in effective
pressure and free-flow flushing (with drawdown) of coarse non-cohesive sediments during flood events
was performed as a background to this investigation. Based on the information obtained from the
literature study, a postulated design was created for further refinement/optimisation by means of
numerical and physical model studies. The postulated design comprised of a semi-circular, ogee-
shaped spillway structure (with wing-walls) located around the intake area of the low-level outlet and
inside of the typical sediment scour hole that forms during pressure flushing (Dreyer, 2018). The main
aim of the structure was to produce desirable upstream supercritical flow conditions to help optimise the
flushing of sediments during free-surface flow/spilling conditions to prevent sediment deposition and
blockage at/near the outlet gates. The submerged weir was however also designed for pressure flushing
where the aim was: (1) local sediment removal/scouring; and (2) preventing main reservoir sediments
from depositing near the outlet gates and hindering gate closure. Based on the results of the numerical
and physical models of the postulated design, design guidelines for such an outlet will be developed in
a follow-up project. The guidelines will provide for different and varying conditions, such as site location,

1
upstream water levels, downstream tailwater levels, preferred flushing techniques, as well as sediment
size and type to be flushed.

2. CONCEPT DESIGN OF POSTULATED SEDIMENT FLUSHING SYSTEM

2.1 Layout
The intake of a low-level outlet conduit is normally situated inside of the original main river channel at
the dam and below the hydropower intakes as recommended by Basson & Rooseboom (1999). These
intakes should also preferably be positioned at a location that limits the amount of sedimentation in front
of them (e.g. on the outer, concave bank at a river bend). The conduit inlet invert level is typically
designed at or below the original river bed level in order to allow for: (1) desirable drawdown flushing
(emptying) conditions and retrogressive erosion of locally deposited sediments (Basson & Rooseboom,
1999); (2) faster flow velocities into the conduit to improve free-flow flushing of sediments (discussed
later); and (3) the formation of local sediment scour zones during pressure flushing that effectively limit
sediment entrainment into the hydropower intakes. For the postulated design of this study (Figures 2 &
3), the low-weir, spillway structure and wing-walls were positioned upstream of the conduit inlet and
inside of the sediment scour hole which forms around the low-level intake area for sediment sizes in the
sand regime. This location ideally allows for easier construction (i.e. minimal need for sediment removal)
and optimal access to further transported sediment during flushing, as well as limits the risk of altering
the local sediment deposition patterns (e.g. underwater angle of repose). Dreyer (2018) used physical
modelling to derive equations to predict the geometrical dimensions of the sediment scour hole that will
typically form during pressure flushing based on different low-level dam outlet conduit shapes, upstream
reservoir heads (𝐻𝑢 ) and upstream sediment levels (𝐻𝑠𝑒𝑑−𝑢𝑝 ). The sediment considered was fine, non-
cohesive silica sand with an effective diameter of 0.09 mm. The most critical scour cone dimensions are
illustrated in Figure 1.

Figure 1. Side view of sediment scour cone geometry for sediment in the sand regime
(adapted from Dreyer, 2018)

2.2 Design of the structural configuration of the postulated sediment flushing system
The cross-sectional side view and top view of the flushing system design for this study is illustrated in

Figure 2 & 3, respectively. The numbered components illustrated in both figures indicate the following:
(1) upstream sediment bed; (2) wing-wall; (3) low-weir; (4) ogee spillway; (5) low-level outlet conduit; (6)
inlet structure; (7) gate chamber; (8) air vent; and (9) dam. The parameters 𝑝1 , 𝑝2 , 𝑝3 and 𝑝𝑝 indicate
the chosen hydraulic control points at which flow velocities and water depths were determined from

2
theory for analysis. The remaining hydraulic parameters are described and discussed in the following
sections.

Figure 2. Cross-sectional side view of postulated flushing system

(water profile indicated for free-flow scenario)

Figure 3. Cross-sectional top view of postulated flushing system

2.2.1 General conduit structure

The low-level outlet conduit incorporates two sluice gates in series (i.e. upstream emergency gate and
downstream service gate) with the gate chamber being located as far upstream as possible to minimize
the length of the pressurised portion of the conduit. The conduit section downstream of the emergency
gate was designed as straight as possible to minimise the change in flow direction and flow velocity
during flushing. The conduit section downstream of the gate chamber incorporates a raised and
horizontal soffit level to ensure free-surface flow under all possible flow conditions to prevent possible
air entrapment and air blowback issues inside of the conduit during pressure flushing. The soffit along
the pressurised portion of the conduit should be designed parallel to the bed slope and tangent to the

3
inlet roof curve in order to guide streamlined flow into the conduit during full-flow pressure flushing (i.e.
sluice gates are 100% open).

2.2.2 Conduit shape & dimensions

A wide, flat-rectangular shaped conduit with a height-to-width ratio of 1: 2 was considered for this study.
According to various authors, this specific shape appears to result in the largest volume of sediment
being scoured locally upstream during singular operation for most scenarios. It has also been found that
rectangular-shaped conduits are typically easier to construct (i.e. in terms of entrance, openings, tunnel,
etc.) as well as make it easier to install and operate gates and other hydraulic equipment. The choice of
conduit width (𝐵) and inlet height (𝐷) should depend on: (1) the structural and economic limitations with
regard to installing gates (i.e. service and emergency) inside of the conduit; and (2) the size of sediment
and floating debris (e.g. trees, etc.) designed to be flushed through the conduit. Based on these
conditions and the specifications of the study, a width of 6 𝑚 and a height of 3 𝑚 was chosen.

2.2.3 Conduit bed slope

A conduit bed slope (ɸ) of 1: 12 was chosen for this study, with the choice being influenced by the
following factors: (1) SANRAL (2013) recommends ɸ > 1% to avoid typical sediment deposition inside
of the conduit; (2) the bed slope should be steep enough to facilitate the flushing of the largest design
sediment (i.e. gravel & boulders) without deposition; and (3) as the horizontal conduit ceiling
downstream of the gate chamber will result in an increase in conduit cross-sectional area as one moves
downstream, a smaller slope choice is desired. Based on the physical properties of the in-situ sediment
that is designed to be flushed (i.e. according to the upstream river bed-load transport capacity), incipient
motion calculations (i.e. Modified Liu Diagram or Shield’s parameter) must be used to verify that the
chosen conduit dimensions and bed slope are suitable for the provisional flushing of the design sediment
inside of the conduit during different critical flushing conditions.

2.2.4 Conduit inlet structure

A bell-mouth shaped roof and side-wall entrance was designed and implemented to maximise flow
acceleration (i.e. as in the case of a flow jet forming from a sharp-edged orifice) into the conduit and to
minimise inlet energy losses. The inlet was also designed to streamline the flow into the conduit to
prevent flow separation. Flow separation can cause cavitation to occur as well as result in the formation
of dead zones inside of the conduit in which sediment could deposit. The design of the roof and side-
wall surface profiles are based on that recommended by the Bureau of Indian Standards (IS 9761, 1995).

2.2.5 Conduit outlet structure

An energy dissipation structure (e.g. stilling basin) immediately downstream of the conduit outlet would
naturally be required in the case of erosive flows with high velocities that exit the conduit and influence
the downstream river conditions. This study does not explicitly focus on the design and implementation
of an energy dissipation structure, but in the case of a general stilling basin design, the USBR (1987)
guidelines should suffice. The conduit outlet in this study comprises an upright rectangular shape and
avoids any mechanism (e.g. a radial gate) that could constrict the air and water flow at the downstream
end of the conduit (especially at the soffit of the conduit). The outlet is also designed to release the
exiting flow in the original direction of the river flow in order to prevent the alteration of the downstream
erosion pattern.

2.2.6 Air vent

An air vent was installed in the region immediately downstream of the gate chamber as this is where air
demand is most critical and can reach a maximum when the gates are being operated under the highest
reservoir head at some partial opening. According to Sharma (1976), a gate opening of 80% can be
assumed to result in the maximum air demand immediately downstream of the gates. For this study, the
air vent outlet is located within 2 𝑚 away (i.e. downstream) from the service gate, as recommended by
Erbisti (2004), in order to be most effective. The air vent inlet is located above the maximum reservoir
water level in order to avoid interference with air flow. The air vent shaft is circular in cross-section (as
typically used in practise) as well as straight and uniform in order to minimise the number of bends and
sharp corners and to avoid sudden changes in cross-section. The required sizing of the air vent was
provisionally designed by first calculating the maximum airflow rate (𝑄𝑎 ) in the vent using the equation
defined by Kalinske & Robertson (1943). Once the upstream reservoir head (𝐻) and gate opening
percentage were known/chosen, the discharge under the gates (𝑄𝑤 ) under free-surface flow conditions

4
was calculated using the gate control equation. The equation defined by USACE (1980) was then used
to calculate the air demand ratio (𝛽). Finally, the required radius (𝑟𝑎 ) of the air vent was calculated
considering the calculated 𝑄𝑎 and limiting the airflow velocity (𝑣𝑎 ) in the vent to less than 45 m/s
(limitation by USACE, 1980). For this study, the following values were conservatively chosen/calculated:
𝐻 = 34 𝑚 (maximum available water level in physical model); gate opening = 80 %; 𝑄𝑤 = 223.2 𝑚3 /𝑠;
𝛽 = 0.196; 𝑄𝑎 = 43.7 𝑚3 /𝑠; 𝑟 = 0.6 𝑚; 𝑣𝑎 = 38.7 𝑚/𝑠. According to Bosman & Basson (2012), the
empirical equations that have been defined in literature for calculating air demand in low-level outlets
have however been found to be generally inaccurate. The same authors therefore recommend that
physical modelling of the gated conduits be done in order to accurately determine the air demand ratio
(𝛽), the cavitation index and pressures along the conduit, the gate rating curve, as well as the flow
conditions downstream of the gate.

2.2.7 Wing-walls
Wing-walls were constructed tangent to the curved faces of the bell-mouth side-walls to effectively aid
in accelerating and streamlining flow into the conduit during free-flow flushing. Wing-walls were also
needed to ensure that supercritical flow did not approach the conduit inlet at an angle, otherwise a
hydraulic jump could form at the entrance during free-flow conditions. The wing-wall angle (𝛼) to the
conduit longitudinal centreline was chosen through trial and error in order to meet various structural and
hydraulic requirements. A range of angles between 30˚ and 45˚ initially appeared suitable to meet the
requirements. CFD modelling was subsequently used to compare designs incorporating 𝛼 = 30˚ and
𝛼 = 45˚ (Section 3). It was found that 𝛼 = 30˚ yielded the best results. The curved intake profile of each
wing-wall was designed similar to that of the bell-mouth side-wall recommended by the Bureau of Indian
Standards (IS 9761, 1995). The wall thickness was designed as 0.40 𝑚 to minimise the volume of
sediment that could deposit and silt up along the wall crest. In order to limit structural costs, the final
wall crest level was designed to match the safety design head (𝐻𝑒 ) above the weir crest at which no flow
spills over during the flushing of 𝑄𝑒 (Section 2.3.1). From CFD and physical modelling, a wall crest
height of 2.00 𝑚 above the weir crest was found to suffice for this study.

2.2.8 Low-weir & ogee-spillway structure

The initially calculated values of the upstream design parameters (

Figure 2 & 3) during the specified free-flow flushing conditions are indicated in Table 1. These calculated
values were used to determine the radii of the ogee spillway profile according to the USBR (1987) design
for general uncontrolled ogee spillways having a vertical upstream face. The reverse bottom curve at
the downstream end of the spillway profile was connected to a linear surface with a bed slope (𝑚) of
1: 10 that extended until it intersected with the conduit bed slope (ɸ) of 1: 12. Steeper approach bed
slopes have shown to typically increase the inlet capacity of long conduits with inlet control (SANRAL,
2013).

Table 1. Initial calculated design parameter values for free flow cases

Parameter Equation Value Unit Comments

Upstream initial 𝐻𝑠𝑖 = 𝐷𝑐𝑜𝑠(ɸ) 2.990 m Measured relative to the conduit
conduit submerged invert level.
water depth (𝐻𝑠𝑖 )
Upstream design 𝐻𝑠𝑖 3.737 m Calculated assuming initial
𝐻𝑤 =
water depth (𝐻𝑤 ) 0.8 conduit inlet submergence being
80 %.
Upstream design 1.600 m Chosen to result in large design
approach flow depth head over weir crest for flushing.
(𝑦𝑢 )
Upstream design 1.595 m/s Corresponds to 𝑦𝑢 on
approach flow velocity permissible velocities graph
(𝑉𝑢 ) (SANRAL, 2013) for assumed
upstream fine gravel bed.
Local design sediment 𝐻𝑠𝑒𝑑 = 𝐻𝑤 − 𝑦𝑢 2.137 m Assumed levelled along
depth (𝐻𝑠𝑒𝑑 ) upstream weir face.

5
 Upstream channel 𝑄𝑒 20.828 m Width upstream of weir crest at
𝑏𝑢 =
width (𝑏𝑢 ) 𝑦𝑢 ∗ 𝑉𝑢 which approach flow conditions
occur; 𝑄𝑒 = 53.153 𝑚³/𝑠.
Chosen weir crest 12.00 m Meets structural requirements.
radius (𝑅𝑐 )
Weir crest arc length 19.400 m Influenced by 𝑅𝑐 and wing-wall
(𝑏𝑐 ) angle (𝛼 = 30˚), with 𝑏𝑐 < 𝑏𝑢 .
Safety design head 1.156 m Solved so 𝑄 = 𝑄𝑒
(𝐻𝑒 )
Design head (𝐻𝑜 ) 𝐻𝑜 = 𝐻𝑒 /1.33 0.869 m Recommended design ratio
from USBR (1987)
Weir crest height (𝑃) 𝑉𝑢 2 0.574 m Measured above local upstream
𝑃 = 𝐻𝑤 + − 𝐻𝑠𝑒𝑑 design sediment level.
2𝑔
− 𝐻𝑒
Total weir crest height 𝐻𝑡 = 𝑃 + 𝐻𝑠𝑒𝑑 − 𝑑 2.499 m 𝑑 = height between local NGL
(𝐻𝑡 ) and invert level (𝐼𝐿1 ) = 0.212 m
Safety design 𝐶𝑒 53.153 m³/s Uncontrolled ogee spillway
𝑄= 𝐶 𝑏 𝐻 1.5
discharge (𝑄) 𝐶𝑜 𝑜 𝑐 𝑒 equation for straight weir in plan
(USBR, 1987)

2.3 Hydraulic design

2.3.1 Design discharge for free-flow flushing
The maximum design discharge (𝑄𝑒 ) for free-flow flushing was initially calculated as 53.15 𝑚3 /𝑠 using
upstream (inlet) control. This initial value was used to design the weir and ogee spillway structure for
subsequent CFD simulations and physical modelling. It was then later refined in CFD simulations and
physical modelling as the maximum design discharge that could be flushed without causing
submergence of the conduit inlet (i.e. not touching the conduit soffit), with no flow spilling over the wing-
walls. The final 𝑄𝑒 value was determined to be 119 𝑚3 /𝑠.

2.3.2 Design discharge for pressure flushing

The maximum design discharge (𝑄𝑝 ) during pressure flushing that is relevant to this study was
determined in the physical model as the discharge that was required to maintain the maximum available
upstream head (𝐻𝑢 ) in the model setup, i.e. 34 m (prototype), during pressure flushing. The value was
determined to be 369 𝑚3 /𝑠 for inlet control conditions.

2.3.3 Minimum submergence of conduit inlet for pressure flushing

The equation proposed by Knauss (1987) was used to provisionally estimate the required submergence
(ℎ𝑚𝑖𝑛 ) of the conduit inlet to avoid air-entraining vortex formation during pressure flushing. Knauss’
equation is valid for median and small size installations (i.e. 𝐹𝑟 ≥ 0.33) and for intakes with symmetric
approach flow conditions, but without the use of special devices for vortex suppression. The value of
ℎ𝑚𝑖𝑛 was calculated as 27.65 𝑚 above the conduit inlet centreline for this study. Due to the fact this
empirical equation has been defined according to general structural and hydraulic conditions for design
purposes, it was necessary to accurately determine the specific submergence requirement through
physical modelling. Physical modelling yielded a value of 22.67 𝑚 (i.e. 24.16 𝑚 above the inlet invert
level) before vortices visibly began to form.

3. CFD MODELLING & RESULTS

Computational Fluid Dynamic (CFD) simulations in 3D ANSYS Fluent 19.1 were used to ultimately
compare two alternative wing-wall orientations (i.e. 𝛼 = 30˚ and 𝛼 = 45˚) within the sediment flushing
design illustrated in

Figure 2 &

6
Figure 3. The purpose of the CFD simulations was to analyse and compare the influence of the
wing-walls on the critical flow vectors (i.e. looking at the degree of streamlined flow) as well as
on the critical flow depths & velocities (i.e. looking at the degree of supercritical flow), especially
around the conduit inlet area, during free-flow flushing considering 𝑄𝑒 (Section 2.3.1).The
finalised conceptual design geometry that was modelled in ANSYS SCDM (i.e. SpaceClaim) 19.1,
and later simulated in ANSYS Fluent 19.1, is illustrated in

Figure 4.
Figure 5 &

Figure 7 illustrate the water volume fraction contour results generated by Fluent 19.1 along a
created longitudinal vertical mid-plane for the two designs.

Figure 6 & Figure 8 illustrate the flow velocity magnitude vector results generated by Fluent 19.1 along
the geometry floor surface for the two designs.

Figure 4. Final conceptual design (free-flow flushing)

Figure 5. Water flow profile result for 𝜶 = 𝟒𝟓˚

7
 Figure 6. Water velocity magnitude vector results for 𝜶 = 𝟒𝟓˚

Figure 7. Water flow profile result for 𝜶 = 𝟑𝟎˚

Figure 8. Water velocity magnitude vector results for 𝜶 = 𝟑𝟎˚

The CFD simulation results illustrate that the 𝛼 = 30˚ wing-wall result in faster flowing supercritical flow
along the spillway, with significant flow acceleration into the conduit around the inlet side-walls. This is
essential for reducing the risk of sediment deposition at potential dead zones near the gates. This design
also appears to result in significantly less flow convergence and damming at the inlet in comparison to
the 𝛼 = 45˚ wing-wall. The wide 45˚ angle, together with the fast approach velocities, result in very high
damming at the centre, poor streamlining of flow vectors into the conduit and clear dead zone formations
near the gates. The 𝛼 = 30˚ wing-wall design was therefore accepted as the best design option.

4. PHYSICAL MODEL

A 1:40 scale physical model of the design evolved from the findings of the CFD model simulations
(Figure 4Error! Reference source not found.) was built for further testing and refinement. The model
was used to test the robustness, reliability and actual flushing capability of the design, considering
different sediment sizes, sediment depths and flushing (i.e. pressure & free-flow with drawdown)
conditions. Relatively coarse sediment was tested due to their presence in run-of-river schemes as well
as at larger dams where the reservoir contains a high percentage of sediment. The physical model was
placed in a 1 𝑥 40 𝑥 1.15 𝑚 deep laboratory glass flume with a flat bed for hydraulic testing. The four
model sediment sizes (𝑑50 ) considered for testing were: (1) 0.095 𝑚𝑚 (3.8 𝑚𝑚 𝑝𝑟𝑜𝑡𝑜𝑡𝑦𝑝𝑒); (2)
13.2 𝑚𝑚 (0.528 𝑚 𝑝𝑟𝑜𝑡𝑜𝑡𝑦𝑝𝑒); (3) 19 𝑚𝑚 (0.760 𝑚 𝑝𝑟𝑜𝑡𝑜𝑡𝑦𝑝𝑒); and (4) 37.5 𝑚𝑚 (1.5 𝑚 𝑝𝑟𝑜𝑡𝑜𝑡𝑦𝑝𝑒).
(Note: All magnitudes with reference to physical parameters and otherwise including 𝑑50 sediment
sizes, water depths, flow rates etc. are hereafter referred to as prototype magnitudes). For free-flow
flushing, two different upstream sediment depths were tested, namely: (1) sediment depth 1 – intake
area filled with respective sediment group until level with the conduit inlet soffit level & upstream area
filled with 𝑑50 = 0.528 𝑚𝑚 sediment group until level with the weir crest level; and (2) sediment depth 2

8
– entire upstream area filled with respective sediment group (besides 𝑑50 = 1.5 𝑚) until level with the
proposed wing-wall crest level. The critical results obtained from free-flow flushing are indicated in Table
2, with measuring point, 𝑃6 , occurring at the conduit outlet invert, and 𝑃7 occurring 180 𝑚 downstream
of 𝑃6 (for tailwater levels controlled by downstream sluice gate). The centreline flow profile of each
discharge (𝑄𝑓𝑐 ) for sediment depth 1, as well as of the maximum discharge (𝑄𝑒 ) for free-flow flushing,
along the model is illustrated in Figure 9. It must be noted that no singular free-flow discharge was able
to fully flush the intake area of the 𝑑50 = 1.5 𝑚 sediment group. The maximum volume of scour that
could be obtained from steadily increasing the discharge until the conduit inlet became submerged is
illustrated in Figure 10 (Left). A reverse bottom curve with a radius of 4 𝑚 was then implemented to
attempt to minimise the risk of sediment deposition near this area. The test was repeated with the flow
being incremented to 99 𝑚3 /𝑠 until scouring stabilised, where after it was lowered to 72 𝑚3 /𝑠. This
resulted in scouring as illustrated in Figure 10 (Right). In order to fully flush the intake area of the 𝑑50 =
0.760 𝑚 sediment group for the sediment depth 2 setup, the flow was incremented to 99 𝑚3 /𝑠 until
scouring stabilised, where after it was lowered to 66 𝑚3 /𝑠. A discharge of 158.4 𝑚3 /𝑠 was, however,
required to fully flush the downstream river channel.

Table 2. Critical free-flow flushing results for different sediment groups

Sediment group size (𝑑50 ) (𝑚) 0.004 0.528 0.760 1.500

Sediment depth 1
Minimum discharge to fully flush model (𝑄𝑓𝑐 ) (𝑚3 /𝑠) 8.3 39.8 71.9 -
Time taken to fully flush intake area (𝑚𝑖𝑛) 28.5 9.5 10.0 -
Flow depth at 𝑃6 (𝑚) 0.180 0.740 0.900 -
Flow depth of 𝑇𝑊𝐿𝑚𝑎𝑥 at 𝑃6 (𝑚) 0.180 0.820 1.100 -
Flow depth at 𝑃7 (𝑚) 0.740 1.180 1.380 -
Flow depth of 𝑇𝑊𝐿𝑚𝑎𝑥 at 𝑃7 (𝑚) 0.740 1.700 2.180 -
Sediment depth 2
Minimum discharge to fully flush model (𝑄𝑓𝑐 ) (𝑚3 /𝑠) 28.0 78.0 158.4 N/A
Time taken to fully flush intake area (𝑚𝑖𝑛) 6.9 20 14.7 N/A
Flow depth at 𝑃6 (𝑚) 0.620 1.100 2.180 N/A
Flow depth of 𝑇𝑊𝐿𝑚𝑎𝑥 at 𝑃6 (𝑚) 0.620 1.100 2.180 N/A
Flow depth at 𝑃7 (𝑚) 1.060 1.500 2.260 N/A
Flow depth of 𝑇𝑊𝐿𝑚𝑎𝑥 at 𝑃7 (𝑚) 1.980 2.340 2.260 N/A
*where: “-“ = did not flush; “N/A” = did not test

9
 Figure 9. Centreline flow profiles of discharges (𝑸𝒇𝒄 ) for sediment depth 1

Figure 10. Min & max scour of 𝒅𝟓𝟎 = 𝟏. 𝟓 𝒎 sediment depth 1 setup (free-flow flushing)

For pressure flushing, both sediment depth 1 and sediment depth 2 scenarios were tested. The three
prototype upstream design water levels (𝐻𝑢 ), measured relative to the conduit inlet invert level, that were
considered for pressure flushing were: 𝐻𝑢1 = 24.16 𝑚 (equals ℎ𝑚𝑖𝑛 ); 𝐻𝑢2 = 29.08 𝑚 (intermediate value);
𝐻𝑢3 = 34.00 𝑚 (maximum available head in physical model). The critical results obtained from the
pressure flushing tests of the 𝑑50 = 0.760 𝑚 sediment depth 2 setup are indicated in Table 3. It must be
noted that during flushing of the intake area under the different upstream water levels (𝐻𝑢 ), the sluice
gate took about 2.3 𝑚𝑖𝑛 (prototype) on average to open fully. It was seen that the intake area did not
scour fully under any pressure flushing or sediment setup scenario. The longitudinal scour profiles of
the 𝑑50 = 0.760 𝑚 sediment depth 2 setup is illustrated in

Figure 11. The ogee-shaped spillway structure and wing-walls were finally removed in order to perform
critical comparison tests between the resulting sediment depth 2 scour cones during pressure flushing
at water level, 𝐻𝑢1 . The corresponding longitudinal scour profile of the 𝑑50 = 0.760 𝑚 setup, without the
structure, is illustrated in

Figure 11. Top view images of post free-flow flushing of the 𝑑50 = 0.528 𝑚 and 𝑑50 = 0.760 𝑚 sediment
depth 2 setups without the intake structure are indicated in Figure 12.

Table 3. Critical pressure flushing results of 𝒅𝟓𝟎 = 𝟎. 𝟕𝟔𝟎 𝐦 sediment depth 2 setup

Upstream water levels (𝐻𝑢 ) (𝑚) 24.16 29.08 34.00

Measured discharge to maintain 𝐻𝑢 (𝑄𝑚 ) (𝑚3 /𝑠) 296.4 329.3 368.8
Time to flush intake area (𝑚𝑖𝑛) 1.4 2.7 1.7
Flow depth at 𝑃6 (𝑚) 2.220 2.380 2.500
Flow depth of 𝑇𝑊𝐿𝑚𝑎𝑥 at 𝑃6 (𝑚) 7.900 8.140 8.260
Flow depth at 𝑃7 (𝑚) 2.180 2.060 1.900

10
 Flow depth of 𝑇𝑊𝐿𝑚𝑎𝑥 at 𝑃7 (𝑚) 10.140 10.780 11.060

Figure 11. Observed pressure flushing scour profiles of 𝒅𝟓𝟎 = 𝟎. 𝟕𝟔𝟎 𝐦 sediment depth 2 setup

Figure 12. Free-flow flushing of 𝒅𝟓𝟎 = 𝟎. 𝟓𝟐𝟖 & 𝟎. 𝟕𝟔𝟎 𝒎 sediment depth 2 setups (no structure)

5. CONCLUSION

The postulated sediment flushing design appears to be working successfully in general considering the
physical model test results acquired for sand, gravel and boulders up to about 1.5 𝑚 in diameter. The
difference in scour cone profile appears minimal during pressure flushing with and without the intake
structure. The significant impact of the structure, however, appears to be during free-flow flushing where
the weir is required to hold back main reservoir sediments from moving towards/depositing near the

11
 outlet gate operation area. Even though this region is always free of sediment deposition following
flushing, the only present concern is the inability to scour all the sediment in the intake structure for all
sizes during pressure flushing. In order to improve the design, the following is proposed: (1) increase
the intake area bed slope to larger than 1: 10; (2) further increase the reverse bottom curve radius
downstream of the ogee-spillway profile; and (3) place the weir closer to the low level outlet.

6. ACKNOWLEDGEMENTS

The authors wish to thank the South African Water Research Commission (WRC) for their permission
to publish this paper. The opinions and views presented in this paper are, however, those of the authors
and do not necessarily reflect those of the WRC.

7. REFERENCES

1. Basson, G. R. and Rooseboom, A. (1999) Dealing with reservoir sedimentation: Guidelines and case studies. ICOLD
bulletin 115.
2. Bosman, E. and Basson, G. R. (2012) Investigation of unsteady flow conditions at dam bottom outlet works due to air
entrainment during gate closure. Volume I: Physical modelling. Pretoria: SA Water Research Commission (WRC).
3. Dreyer, J. S. (2018) Investigating the Shape and Layout of Low-Level Outlets for Pressure Flushing of Sediments in HPP
Reservoirs. University of Stellenbosch.
4. Erbisti, P. (2004) Design of Hydraulic Gates. 1st edn. Lisse, Netherlands: Swets & Zeitlinger B. V.
5. ICOLD Bulletin 147 (2009) Sedimentation and Sustainable Use of Reservoirs and River Systems.
6. IS 9761 (1995) Hydropower Intakes - Criteria for Hydraulic Design. 1st edn. Bureau of Indian Standards.
7. Kalinske, A. A. and Robertson, J. M. (1943) Entrainment of Air in Flowing Water: A Symposium: Closed Conduit Flow,
Transactions of the American Society of Civil Engineers, 108(1), pp. 1435–1447.
8. Knauss, J. (1987) ‘Prediction of Critical Submergence’, in Knauss, J. (ed.) Swirling Flow Problems at Intakes. 1st edn.
Rotterdam: IAHR, Balkema.
9. SANRAL (2013) Drainage Manual. 6th edn. Pretoria: The South African National Road Agency.
10. Sharma, H. (1976) Air-Entrainment in High Head Gated Condiuts, Journal of the Hydraulics Division, 102(11), pp. 1629–
1646.
11. USACE (1980) Hydraulic Design of Reservoir Outlet Structues (EM 1110-2-1602). Washington, DC: United States Army
Corps of Engineers.
12. USBR (1987) Design of Small Dams. 3rd edn. Water Resources Technical Publication.

12

View publication stats