2 CHARACTERISTICS OF AIR-REGULATED

SIPHON SPILLWAYS

2.1 Introduction
This chapter consists of two parts. The first (Sn 2.2) gives background information
about siphon spillways with definitions and a description of simple and air-regulated
siphons. General criteria for siphon design are then considered. The second part (Sn 2.3)
is devoted to the results of experiments on the 1:10 scale Pergau siphon model. The aim
of this investigation was to gain insights into the air supply and air demand for siphon
spillways. The characteristics of this model for various air inlet areas and positions are
presented. The influence on the air supply of a free vortex which forms around the
siphon’s hood is considered in Sn 2.9.3.

2.2 A review of siphon spillways

2.2.1 Basic definitions

A simple siphon is a conduit bent into an inverted U-shape with two limbs of unequal
length so that a liquid may be transferred from a higher to a lower level at a rate
proportional to the difference in head. The exit from the siphon may have a water seal
or it may be freely discharging to the atmosphere, but in either case the air in the
conduit must be removed before the siphon runs full. During this priming process, a
partial vacuum is created within the siphon as air is replaced by water.

When air is removed by the natural flow of water through the siphon it is said to be self-
priming. When all air is removed from the siphon barrel, priming is complete and
blackwater flow commences. Alternatively, priming may be achieved with an air pump.

26
Chapter2 Characteristics of Air-Regulated Siphon Spillways

If a drawing down of the reservoir allows air to enter the siphon, the flow is interrupted
and depriming will occur. The priming and depriming of a siphon can occur rapidly
and lead to violent structural loading, but when air is allowed to enter a siphon in
controlled volumes, the siphon will prime at a moderate rate and the loadings are
diminished. Such an air-regulated siphon is designed for stable operation at all
discharges up to blackwater flow. The difference in level between the reservoir and the
siphon exit or tailwater level is known as the working head. This vertical distance
determines the maximum discharge available from the siphon when it carries
blackwater flow. At lower discharges moderated by a flow of air through the siphon, the
working head is represented by the sum of the head over the crest of the siphon, as for a
conventional weir, and the subatmospheric pressure head in the barrel.

2.2.2 Development of siphon spillways

The principal of a siphon was first applied by Hirsch at Mittarsheim in 1866 but did not
come into common use despite the advantages of a higher discharge per unit spillway
width, the ability to pass a flood with an almost constant water level and the absence of
moving parts to control operation. For many years, however, siphons were not
considered for spillways owing to uncertainty about the priming process, the threat of
debris blocking the throat and the danger of sudden surges downstream of the dam
when the siphon primes.

The introduction of air to regulate the discharge of water through a siphon was an
important innovation to ensure a smooth, gradual and controlled priming action. This
led to the use of such siphons on dams and other flood control projects such as
reservoirs, flood control works and hydroelectric power systems. The earliest air-
regulated siphons were designed by Crump and constructed at Renala, India in 1922. In
Britain one of the earliest air-regulated siphons was built at Eyebrook in 1959 (Oliver,
1959).

More recently, siphon spillways have been constructed for Spelga dam, Craigavon
(Ervine, 1976), Shek Pik (Young and Tucker, 1958), Plover Cove (HRSW, 1971) and
High Island (HRSW, 1973), but owing to their complex behaviour and the sparsity of
design guidelines, almost all siphon spillways have had to be developed on the basis of

27
Chapter2 Characteristics of Air-Regulated Siphon Spillways

physical model studies. A recent example is an adjustable air-regulated siphon spillway

for the Pergau scheme (Hardwick and Grant, 1997). The laboratory model used for this
design formed an important vehicle for the fundamental studies of the present
investigation.

2.2.3 Classification of siphon spillways

2.2.3.1 Working head

One of the common ways for classifying siphon spillways is based on working head. It
is often an arbitrary distinction, but it is generally accepted that a high head siphon has a
working head greater than 6 m; medium head siphons operate in the range 3 m-6 m and
low head siphons function under heads of less than 3 m (Ervine, 1974).

2.2.3.2 Priming method

Perhaps a more precise classification of siphons is on the basis of their priming
mechanism, i.e. whether conventional or air-regulated. The conventional method
involves the entrainment and evacuation of air by a nappe of water, either plunging into
the sealing pool at the siphon’s exit or passing over a deflector in the invert and
impinging on the soffit of the siphon’s barrel. Early siphon designs involving auxiliary
priming devices such as baby siphons and weirs are not in common use because they are
difficult to construct and are somewhat impractical.

In a conventional siphon the entry of air tending to deprime the siphon is avoided by
locating the entrance well below the pond waterlevel. The downstream exit may not
necessarily be sealed except during the initial priming. Thereafter, entrainment by the
moving water is normally sufficient to evacuate any remaining air from the siphon. The
siphon is designed to discharge water free of air, hence the name blackwater to describe
the flow.

Air-regulated siphons, in contrast, draw in a continuous stream of air which controls the
discharge through mixing with the water to form a white-water stream. The priming of
such devices is gradual and is generally controlled by the water level of the upstream
pond.

28
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.2.4 Principles of air-regulated siphons

A siphon spillway takes advantage of the level difference between its upstream and
downstream pools; this is in contrast with a conventional spillway with atmospheric
pressure at the water surface where the working head is initially the upstream water
level relative to the weir crest. Once a simple siphon is primed, its full working head
acting on the stream causes the discharge to increase to the maximum, blackwater flow;
the drawback of such behaviour is that the siphon is essentially an on-off device where
the discharge hunts between the maximum and minimum flows with levels at the intake
fluctuating correspondingly. This difficulty has been overcome in certain instances
through the use of multiple siphons with different crest elevations but this approach is
neither elegant nor economical.

The admission of air to a siphon hood moderates the build up of flow and with the
upstream water level controlling the amount of air admitted, the spillway is able to
regulate the rise of reservoir level within a narrow range through a process described
below. Owing to such fine control, almost the full storage capacity of the reservoir can
be realised.

2.2.5 Characteristics of siphon spillways

The relationship between the reservoir level and the discharge through a siphon is the
siphon’s characteristic. Fig. 2.1 shows a typical characteristic of a simple siphon where
OA represents the range of weir flow before priming is completed. Thereafter the
discharge increases rapidly from A to B for virtually no increase in reservoir level. The
section BC represents blackwater conditions. A lowering of reservoir level induces a
rapid depriming when air enters the hood at B  . This behaviour with its hysteresis is
not generally acceptable.

29
Chapter2 Characteristics of Air-Regulated Siphon Spillways

Fig 2.1 Typical characteristic of an on-off siphon spillway

An air-regulated siphon (Fig. 2.2) shows a similar weir flow section OA and then an air-
regulated section A-B, where there is a gradual increase of discharge for a relatively
small rise of reservoir level. When the air inlet becomes submerged, no air is admitted
and blackwater conditions prevail at section BC as before. With a falling reservoir level,
the flow process is reversed with no hysteresis and a virtually steady flow can be
achieved for any upstream condition.

Fig 2.2 Typical characteristic of an air-regulated siphon spillway.

Fig. 2.3 shows a typical characteristic for an air-regulated siphon which incorporates the
benefits of both conventional weir behaviour and closed conduit flow. Achieving an
appropriate air-controlled, linking section demands an understanding of the behaviour
of the air supply system and air entrainment under subatmospheric pressure. Recent
researches have added to this understanding but there are still many uncertainties for
designers. The present study aims to extend the knowledge of air-water mixtures with
special reference to air-regulated siphons. While it may always be necessary to make
physical models of novel devices, it should be possible to improve the guidance for
designers at a preliminary stage, at least to define the general form of a siphon’s
geometry with heightened confidence.

30
Chapter2 Characteristics of Air-Regulated Siphon Spillways

Fig 2.3 An air-regulated siphon spillway’s characteristic in relation to weir and closed conduit
flows (after Charlton 1971).

2.2.6 Operation of an air-regulated siphon spillway with a deflector

With a progressively rising water level in a pond, four stages of operation can be
identified as described below (Charlton, 1971) and are common to the behaviour of the
Pergau siphon.

2.2.6.1 Weir flow

Before priming commences the siphon operates as a conventional spillway. The
descending water in the barrel approaching the deflector has insufficient momentum to
be diverted to the soffit and so the pressure in the barrel remains atmospheric.

2.2.6.2 Subatmospheric weir flow

With increasing discharge, the flow in the barrel has sufficient momentum to impact on
the soffit of the barrel beyond the deflector and thus seals the barrel against the ingress
of air from downstream. Thereafter, air is entrained and evacuated from the spillway,
the pressure falls in the barrel and there is an associated increase of water flow through
the siphon which in turn evacuates more air and lowers the pressure still further. At a
certain stage the pressure in the hood is so low that air is drawn through the holes in the
hood and the growth of discharge is halted. Eventually a stable condition is established
where the inflowing air is just equal to the outflowing air and the water discharge is

31
Chapter2 Characteristics of Air-Regulated Siphon Spillways

held at a moderate level. There is little mixing of air and water in the descending limb of
the siphon until the flow encounters the deflector.

2.2.6.3 Air-partialized flow

At still higher water discharges, culvert flow in the barrel approaches full flow with
well-mixed air bubbles distributed throughout the body of the water stream.

2.2.6.4 Blackwater flow

When a further rise of reservoir level eventually seals the air holes through the hood,
air-regulation ceases and blackwater flow commences. Beyond this point, an increase of
discharge can be accommodated only by an appreciable rise in reservoir level. Owing to
the threat of flow overtopping the dam, blackwater flow should be avoided in practice.

Head (1975), in a model study of a low-head, air-regulated siphon, identified five stages
for siphon operation as: weir flow, deflected nappe, depressed/drowned nappe, air
partialized flow and blackwater flow. In Head’s example the depressed/drowned
condition was associated with subatmospheric weir flow conditions. Owing to the small
working head, the distinction between these two phases appeared when the tailwater
drowned the deflected nappe.

2.2.7 Siphon design

This section outlines some of the main considerations when developing the geometry of
a siphon spillway. A well-designed structure has the following advantages over a
normal overflow weir:

1. The reservoir retention level may be raised and so the storage capacity of the
reservoir will be increased.
2. The stable operation of an air-regulated siphon spillway will maintain the reservoir
water level within narrowly defined limits.
3. The higher working head of a siphon permits greater discharge, improved flood
control and a relatively compact spillway structure.
A typical siphon characteristic (Fig. 2.3) can be used as a basis for designing a new
spillway to achieve the design discharge, a certain reservoir retention level and reservoir

32
Chapter2 Characteristics of Air-Regulated Siphon Spillways

surface level at the design discharge. Thereafter the cross-section, flow geometry,
priming system and number of barrels must be established. Stages in this process are set
out below.

2.2.7.1 The design discharge

The maximum discharge is determined from hydrological predictions and a routing of
the design flood through the reservoir. If this flow were to be discharged in the
blackwater regime, its relationship with the working head would be

Qwmax  Cd A 2 gH , (2.1)

where, Cd is the siphon’s coefficient of discharge for blackwater flow;

A is the siphon’s cross sectional area;
H is the working head across the siphon between reservoir and tailwater levels.

2.2.7.2 Working head

The working head is determined from the maximum permissible flood reservoir level
relative either to the centre of the siphon outlet or to the downstream water level. Its
value depends on the general siphon geometry and the maximum discharge for which
the siphon is to be designed. Theoretically and practically, there is no limit to the length
of the siphon’s downstream leg (Zanker et al, 1968), but for high-head siphons the
designer must ensure that low pressures approaching the cavitation threshold in the
region of the crest are avoided.

2.2.7.3 Siphon cross-sectional area

The necessary cross-sectional area of the siphon is determined from the maximum
discharge, the available working head and the siphon’s discharge coefficient. The
resistance to flow is considered to be greatest in the blackwater regime when the siphon
then functions as a closed conduit under the working head. The siphon discharge
coefficient depends on the geometry of the siphon and varies between 0.65 and 0.85
(Charlton, 1971).

33
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.2.7.4 Priming head

Just prior to priming, the air holes in the hood are sealed but the downstream end of the
siphon is vented to atmosphere, or in the case of a exit-sealed siphon, the water jet in the
barrel has not achieved the threshold conditions to entrain air and evacuate the barrel
from air. The pressure in the barrel remains atmospheric and the flow is thus still
governed by a conventional weir relationship (OA in Fig. 2.2).

Qw  Kbhw3 / 2 , (2.2)

where, K is the coefficient of discharge for weir flow;

b is the width of siphon’s hood at the crest;
hw is the height of water above the crest at point A in Fig. 2.3.

2.2.7.5 Exit geometry

There are two types of exit for siphons: exit-sealed and exit-free. The former is the
preferred type as the priming process is more certain and positive than for the latter. In
cases where it is not practicable to seal the siphon’s exit, a deflector is located in the
barrel to divert the flow to the soffit of the conduit and so achieve a seal. Fig. 2.4 shows
an exit-free installation where a battery of 16 siphons located on two sides of an eight-
sided bellmouth spillway increases the reservoir retention level (Oliver, 1959).

In comparison with a sealed exit, the sealing of an exit-free design tends to be weaker;
as a consequence the priming may be slower and the required head to complete the
process is often higher.

2.2.7.6 Priming system

Priming systems designed in the past (Fig. 2.5) have varied from an auxiliary weir
(Leliavsky, 1957) and baby siphons (Mc Birney, 1958) to the simple deflected nappe
which is now considered to be most efficient for air extraction. In certain circumstances
a hydraulic jump has been used successfully to prime siphons (Charlton, 1962).

34
Chapter2 Characteristics of Air-Regulated Siphon Spillways

Fig 2.4 Views of Eyebrook Reservoir siphon spillway (after Ervine and Oliver, 1980).

(a) Baby siphon (after Mc Birney, 1958) (b) Auxiliary weir (after Leliavsky, 1957)

(c) Siphon primed by a hydraulic jump (after Charlton, 1962)

Fig 2.5 Some priming systems used in siphon spillways.

35
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.2.7.7 Deflector
To ensure effective priming in an exit-free siphon, the deflector should continue to
divert the stream as the water flow increases during the priming process; if the device is
too small the increasing flow may cease to be diverted effectively and the seal will be
broken. A rectangular cross-section is preferred for the barrel in the region of the
deflector to ensure that the ends of the nappe retain contact with the solid boundary
(Charlton, 1971). Conversely, if the deflector is too big the overall siphon resistance
will be unnecessarily increased. On the strength of experiments in the present research,
recommendations for the appropriate position of the deflector are given in Chapter 5.

2.2.7.8 Upper bend design

The upper bend must sustain a mixed flow of air and water around the crest of the
siphon under a radial pressure gradient. It is reasonable to assume that the flow in the
hood approximates to a free vortex and on this basis the radial pressure gradient for
various bend radii and flow conditions may be determined. To assist the mixing of air
and water, the centripetal pressure gradient should virtually neutralise the hydrostatic
pressure gradient. It has been shown that a change of sign of the combined pressure
gradients during the air-regulated phase can upset the air-water flow pattern and cause
instability (Charlton, 1971). It is suggested, therefore, that the bend radii be selected so
that over the flow range, the radial pressure gradient approaches the neutral condition
but that its sign does not change. For a free vortex flow

c c
u1  and u2  , (2.3)
R1 R2

where u is the velocity, R is the radius, c is a constant and 1 and 2 refer to parameters at
crest and crown respectively. If the gradient of pressure, p, in the bend is to be nearly
neutral, then

p1  p2 . (2.4)

Equating the energies between crest and crown gives

36
Chapter2 Characteristics of Air-Regulated Siphon Spillways

p1 u12 p u2
  2  2  ( R1  R2 ) . (2.5)
g 2 g g 2 g

Substituting of (2.3) and (2.4) in equation (2.5) gives an expression

2g
c  R1 R2 . (2.6)
R1  R2

Eliminating c from (2.3) yields

R1  R2 R1  R2
R1  u 2 and R2  u1 . (2.7).
2g 2g

To a first approximation the velocity gradient between crest and crown may be taken as
u1  u 2
linear and then U mean  . In this way the curvature of crest and crown, appropriate
2
to blackwater flow can be established.

2.2.7.9 Entrance and air inlet

To have stable operation and a smooth characteristic for the siphon spillway, a sufficient
area of air inlet must be provided to permit the maximum demand for air to enter
without restriction.

37
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.3 Experiments with the 1:10 scale Pergau siphon model

This section describes a series of laboratory investigations carried out with a 1:10 scale
siphon model to discover the influence on air and water flows of differing areas and
locations of air holes through the hood. The aims of this series of experiments were
threefold:

1. to gain an increased understanding of the physical processes of siphon operation;

2. to investigate the influence on siphon behaviour of the upstream water level and

3. to investigate parameters influencing air supply and air demand in the model.

2.3.1 Apparatus
The main features of the 1:10 scale model were reported previously by Hardwick,
(1992) and will be summarised here. The model (Fig. 2.6) consisted of a timber header
tank 1.23 m high and 1.0 m by 1.5 m in cross section, a perspex approach channel, a
perspex siphon spillway of cross-section 292 mm wide by 145 mm high with a stilling
basin at the downstream end. The air inlet consisted of a line of 22 cylindrical air holes,
6.35 mm in diameter and 65 mm long located on a horizontal axis in a plane 19 mm
above the siphon’s crest.

The number of air holes and their elevation relative to the siphon’s crest were
determined through a series of trials to provide a satisfactory head-discharge
characteristic for the model (Hardwick, 1992). A skimming weir drew water from the
header tank to minimise undesirable level fluctuations in the approach channel arising
from turbulent processes in the siphon.

38
Chapter2 Characteristics of Air-Regulated Siphon Spillways

Fig 2.6 vxm

39
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.3.2 Normal operation of the siphon

Plate 2.1 shows the main features of the flow pattern with the model operating in the
air-regulated phase. The flow in the approach channel was decelerated at the protective
beam adjacent to the air holes. The beam provided a quiescent pool of water between its
downstream face and the air holes and it is the level of this pool which, in rising with a
rising pond level, restricts the ingress of air into the hood and thus permits an increase
of water discharge through the siphon.

Within the hood and above the level of the air holes, a turbulent mixture of air and water
can be seen. Further downstream, this air-water mixture broke away from the soffit of
the hood and air and water separated, the water forming a supercritical wall jet with a
subatmospheric pressure at its surface. Near the deflector, the jet was directed toward
the soffit of the barrel. The air released at the top of the barrel was drawn downstream
by friction with the water jet and was entrained by the water along the length of the
deflected nappe. It will be shown (Chapter 6) that the curvature of the deflected
streamline increased the jet’s turbulence level and so enhanced the air/water mixing.

An aerated roller may form between the nappe and the soffit of the barrel depending on
the value of subatmospheric pressure in the barrel, deflection angle and the influence of
higher tailwater levels with increasing discharge. In a siphon spillway on the other hand,
as the air-water mixture impacts on the soffit, some of the entrained air is detrained and
drawn back upstream under the impress of the barrel’s subatmospheric condition. The
remainder of the originally entrained air lying below a stagnating streamline is carried
downstream beyond the trailing edge of the barrel where it is diffused and eventually
expelled across the free surface of the turbulent stilling pool (plate 2.1).

40
Chapter2 Characteristics of Air-Regulated Siphon Spillways

Plate 2.1 The 1:10 scale Pergau siphon model operating in the air-regulated phase.

2.3.3 Measurements of air and water flows

The water discharge through the siphon model was calculated from the pressure
difference across an orifice plate located in the supply pipe to the header tank. This
difference was measured using a differential mercury manometer with the accuracy of
0.5 mm. Any flow over the skimming weir was measured in a similar fashion and
subtracted from the inlet flow. The airflow entering the hood through bellmouthed
intakes of known cross-sectional area was deduced from the fall in pressure across these
orifices. An alcohol manometer with a 1:16 amplification was used to measure this
difference. Detailed descriptions of airflow measurements are given in Chapter 4.

2.3.4 Operating ranges

Typical characteristic curves relating pond elevation to water discharge are shown in
Fig. 2.7 for two different air hole areas. The curves demonstrate four distinct ranges of
operation: weir flow, air-regulated flow, air partialised flow and blackwater flow; these
are considered in turn below.

41
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.3.4.1 Weir flow

The flows from 0 to about 4 l/s generated by pond elevations, yp, up to 20 mm above the
crest level had insufficient momentum to seal the barrel when negotiating the deflector.
The pressure in the barrel was thus atmospheric and so the siphon operated as a
conventional overflow spillway.

For slightly higher pond elevations and discharges between 4 l/s and 15 l/s, the siphon
hunted between weir flow and a partially-primed condition when a temporary seal of the
deflector was effected; this sealing led to a fall of pressure in the barrel and a
consequent increase of discharge. This increased flow in turn caused a drawdown of the
upstream pond and an exposure of the air holes such that atmospheric conditions were
restored in the barrel and weir flow recommenced. Such hunting would be unlikely at
full-scale because the large prototype reservoir could not be so readily drawn down.
Accurate measurements of pond elevations, air flow and pressure in the barrel in this
unstable range were not possible and what is plotted in Fig. 2.7 is the best estimate of
average values.

2.3.4.2 Air-regulated and air-partialised flow

For pond elevations yielding discharges above 15 l/s, stable, air-regulated behaviour
was initiated when a seal was effected over the deflector. As a consequence, air was
evacuated from the barrel causing both a reduction of pressure and an increase of water
discharge until the pressure in the hood was sufficiently low to draw in air through the
air holes. This penetrating air halted both the fall of pressure in the barrel and any
further increase of water discharge. A stable operating point was thus established by an
equality between the air supply and the air demand. Such a flow pattern with a distinct
interface between air and water in the barrel persisted for discharges up to about 60 l/s.

In the discharge range 60 l/s to 80 l/s, air-partialized flow was observed when the flow
throughout the siphon became an air-water mixture.

2.3.4.3 Blackwater flow

The pond levels generating flows over 80 l/s submerged the air holes and blackwater
flow commenced. In Fig. 2.7a, the characteristic curves show an abrupt increase of

42
Chapter2 Characteristics of Air-Regulated Siphon Spillways

slope at this point which could lead to a rapid loss of freeboard below the top of a
reservoir embankment and thus render operation in this range unsuitable at full-scale.

2.3.5 Influence of air hole area

The air hole area is seen in Fig. 2.7a to influence the slope of the characteristic curve in
the air-regulated range. A reduction of air hole area led to a slight flattening of this
slope which, ultimately with the provision of no air holes, becomes horizontal and the
operating modes of the siphon are limited to either weir flow or blackwater flow.

2.3.6 Air flow characteristics

Airflow, Qa, and corresponding values of subatmospheric pressure, Hsub, in the barrel
for two air hole areas are shown in Figs 2.7b,c. It can be seen that a reduction in hole
area at the beginning of the air-regulated phase tended to throttle Qa but when the siphon
progresed to air-partialized flow, the influence of hole area weakened. Three ranges of
flow are identified in Fig. 2.7b. Maximum Qa is observed at the commencement of the
first range (i) and thereafter, as Qw increased, Qa decreased sharply as the water level in
the hood progressively rose toward the crown of the siphon in a process which appeared
to impede the inflow of air.

In the second range (ii), Qa was virtually constant and it will be seen later that this is
probably the result of two opposing processes: the rising pond level on the one hand
tending to restrict the air supply while on the other, the corresponding increase of Q w
tending to increase air entrainment near the deflector. In the third range (iii), Qa
increased, owing to the formation of what was understood to be a free vortex in the
hood of the siphon (Sn 2.3.9). Greater measured levels of subatmospheric pressure at
the crown than those further downstream in the barrel are consistent with this idea and
the generally lowered pressure in the hood led to increased Qa through the air holes.
Beyond range (iii), the air supply was increasingly throttled until blackwater flow
prevailed, the scattered results indicating the difficulty in measuring both Qa and Hsub in
unsteady flow conditions.

Over the discharges associated with ranges (i) and (ii), Fig. 2.7c shows a steady increase
of subatmospheric pressure with increasing Qw which reflects the growing entrainment

43
Chapter2 Characteristics of Air-Regulated Siphon Spillways

of air by a more vigorous mixing process in the region of the deflector. In range (iii),
this growth of Hsub appeared to be moderated by the greater supply of air noted above in
connection with free vortex action. The slight trend toward higher values of Hsub with
reducing hole area will be seen later to be more marked when the air hole area is further
diminished. Beyond range (iii), where blackwater conditions develop and pressure
measurements were less reliable, the trend of Hsub toward lower values is nevertheless
clear.

The parameters shown in Figs 2.7a,b,c are linked by the siphon’s geometry and flow
pattern; in an attempt to eliminate the influence of pond level on Qa, the 1:10 scale
model was modified in the following way (Sn 2.3.7):

44
Chapter2 Characteristics of Air-Regulated Siphon Spillways

150
(a) B lackwater flo w
A ir-partialized flo w
120 Weir flo w
Unstable flo w
A ir-regulated flo w

yp [mm ] 90

60

22 air ho les
30
16 air ho les

0
0 15 30 45 60 75 90
Qw [l/s]

6
(b) 22 air holes
5
16 air holes

4
Qa [l/s]

2
i ii iii
1

0
0 15 30 45 60 75 90
Qw [l/s]

210
(c)
22 air holes
180
16 air holes
150
Hsub [mm ]

120

90

60

30

0
0 15 30 45 60 75 90
Qw [l/s]

Fig 2.7 Characteristics of Pergau siphon model for normal operation, (a) Head-discharge curves,
(b) Air flow through siphon, (c) Subatmospheric pressure head in barrel.

45
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.3.7 An attempt to deduce air demand

As a means of breaking the link between air supply and air demand, the entrance to the
inner pool between the beam and the air holes was sealed such that conditions upstream
had no influence on the air hole inlets. Measurements of head, discharge, airflow and
pressure were repeated and the results presented in Fig. 2.8.

An unobstructed supply of air prevented blackwater flow (c.f. Figs 2.8a and 2.7a) and
reduced the gradient of Hsub (c.f. Figs 2.8c and 2.7c). Higher pond elevations relative to
the weir crest, yp, were generally necessary to yield a given discharge because of these
reduced levels of Hsub. In range (i) of Fig. 2.8b, the rising level of water in the hood
with increasing Qw again progressively reduced Qa. It will be shown in Chapter 3 that a
rising waterlevel above the air hole elevation reduced the driving head for Qa and as a
result the rate of air entering the siphon decreased.

In range (ii) with no further change of flow geometry in the hood and no restriction on
the entry of air, Qa increased in response to the formation the free vortex flow in the
hood. Another important agent for increasing airflow was the change in the flow pattern
over the deflector with increasing Qw. The thicker water jet with higher momentum was
deflected at a lower angle to the horizontal, impinging on the siphon’s soffit with a
milder angle so that the roller above the nappe diminished in size and almost all the
entrained air was transported by the water out of the siphon. This entrainment was offset
to some extent by both the influence of an increasing Hsub tending to increase the size of
the roller and a rising pressure downstream of the deflector imposed by a higher
tailwater. Toward the end of range (ii), Qa tended to become constant and the gradient
of Hsub diminished with the onset of air partialisation when the entraining processes near
the deflector were suppressed; to maintain an increase in Qw, the response of the siphon
to such a reduced gradient for Hsub was seen in a steeper gradient for yp in Fig. 2.8a.

As before, a reduction of air hole area generally reduced both yp and Qa and increased
Hsub, but this attempt to break the link between air supply and air demand failed owing
to the effects of a varying flow pattern in the hood; the rise of water level towards the
crown in range (i) and the suggested free vortex effect at higher flows tended
respectively to reduce and increase the gradient of Qa. In an effort to eliminate these
influences, an alternative air inlet was devised as described below.

46
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.3.8 Air inlet downstream of the hood

To deduce the air demand arising solely from the entrainment / detrainment process
near the deflector, the air holes through the hood were sealed and new holes were
opened in the soffit of the barrel, well downstream of the siphon’s crest where they
would be unaffected by changes of flow pattern in the hood. It was also considered that
such an arrangement would show the influence of Hsub in detraining air from the mixing
zone near the deflector. The results of tests for this new air inlet location with a
combination of hole areas identical to those used previously are shown in Fig. 2.9

The head-discharge curves of Fig. 2.9a are in marked contrast with those previously
considered and show regions where stable operating conditions could not be achieved.
For discharges up to about 4 l/s, weir flow was seen over the crest of the siphon until a
seal developed over the deflector and siphonic flow commenced. This transition was
most obvious for the smallest hole area where the pond elevation was rapidly drawn
down and there was a step-increment in discharge. At this transition, Hsub increased
sharply from near-atmospheric conditions but the unobstructed access of air limited its
rise to levels which were generally below those shown in Figs 2.7 and 2.8; increases of
water discharge were thus mainly achieved by an increase of pond elevation.

At a discharge of about 30 l/s in Fig. 2.9a, the flow pattern again changed with a
migration of the water surface towards the crown of the siphon and the development of
free vortex action in the hood. At this point there was a step-fall in pond elevation and a
step-rise in Hsub.

The trends for Qa in Fig. 2.9b show a common increase of air demand until point A in
the figure when, for the smallest air hole area, an equilibrium was achieved between a
tendency to greater air entrainment with increasing Qw and a tendency to greater
detrainment owing to higher values of Hsub. With larger air hole areas, this equilibrium
was established at points B and C at higher values of Qw.

47
Chapter2 Characteristics of Air-Regulated Siphon Spillways

140
A ir partialized flo w
120 Weir & unstable flo w
A ir-regulated flo w
100
80

yp [mm ] 60
22 air ho les
40 16 air ho les
11 air ho les
20
6 air ho les
0

-20
Qw [l/s]
-40
0 15 30 45 60 75 90

(a) Head-discharge curves

8
22 air ho les
7 16 air ho les
11 air ho les
6 6 air ho les

5
Qa [l/s]

1 i ii iii
Qw [l/s]
0
0 15 30 45 60 75 90

(b) Air demand curves

160
22 air ho les
150 16 air ho les
11 air ho les
140 6 air ho les
Hsub [mm ]

130

120
110

100

90

80
0 15 30 45 60 75 90
Qw [l/s]

(c) Subatmospheric pressure head in barrel

Fig 2.8 Characteristics with inlets to air holes unrestricted.

48
Chapter2 Characteristics of Air-Regulated Siphon Spillways

180

150

120

yp [mm ]
90

60 22 air ho les
11 air ho les
30 6 air ho les

0
Qw [l/s]
-30
0 15 30 45 60 75

(a) Head-discharge curves

18
22 air ho les
16 11 air ho les

14 6 air ho les C

12
Qa [l/s]

10

8
B
6
4 A

2
Qw [l/s]
0
0 15 30 45 60 75

(b) Air demand curves

160

140

120
Hsub [mm ]

100
80

60
40 22 air ho les
11 air ho les
20
6 air ho les
0
0 15 30 45 60 75
Qw [l/s]

(c) Subatmospheric pressure head in barrel

Fig 2.9 Characteristics for an air inlet downstream of the siphon’s crest.

49
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.3.9 Influence of a free vortex in the siphon’s hood

The rate of airflow in siphon spillways generally depends on the characteristics of the
siphon’s air supply and the air demand of the water jet over the deflector.

It was recognised from the experiments performed on the Pergau model described in
this chapter that the major parameters influencing air supply were as follows (Fig.
2.10): the waterlevel upstream of the air hole, ya, the cross-sectional area of the air
holes, Aa; the height of water above the air holes in the hood, Z; and the subatmospheric
pressure in the siphon’s hood. The air demand was found to be influenced by the flow
pattern and water jet velocity at the deflector, the surface roller and the angle of the jet
impingement on the siphon’s soffit and the downstream water level. The
subatmospheric pressure in the hood acted as the driving head for airflow entering the
siphon and also served as an agent for controlling the angle of jet impingement and the
associated roller above the nappe which detrained part of the originally entrained air
back to the barrel.

It is believed that the formation of a free vortex in the siphon’s hood affected both the
supply of air to the siphon and the demand for air over the deflector. To examine this
idea, precise measurements of all the above parameters were performed when the
siphon was operating in the normal condition with air entering 22 holes through the
hood (Fig. 2.6). Fig. 2.11 shows the variation with water flow, Q w, of the following: the
airflow, Qa; the subatmospheric pressures in the hood at the level of air holes, Hsub-hole
(Fig. 2.10); the subatmospheric pressure at the siphon’s crown, Hsub-crown, and above the
water jet in the barrel, Hsub-barrel. The variation of waterlevel upstream of the air holes is
plotted in Fig. 2.12.

In this section, attention is mainly concentrated on the variation of subatmospheric

pressure in the hood. Fig. 2.11, shows for Qw  30 l/s that Hsub-hole, the driving head for
air entering the siphon, experienced a slight fall; the corresponding fall of Qa is believed
to be partly due to the obstruction of air holes by the upstream water level, ya, (Fig.
2.12) but more importantly, the decreasing subatmospheric pressure in the hood as the
water level above the air holes, Z, rose. In the range 30  Qw  50 l/s, Qa remained
virtually constant where Z reached its upper limit at the siphon’s crown. In this range

50
Chapter2 Characteristics of Air-Regulated Siphon Spillways

there appeared to be a balance where the increasing Hsub-barrel (tending to increase Qa) is
opposed by the gradual obstruction of the air holes together with the increased
detrainment by the roller.

For Qw  50 l/s, it was suggested that a free vortex flow formed in the hood around the
siphon’s crest. Fig. 2.11 shows an increase in the subatmospheric pressures in the hood,
Hsub-hole and Hsub-crown, and despite the increase of ya, Qa increased. The additional air
supplied to the siphon in this range diminished the gradient of the subatmospheric
pressure in the barrel; this led in turn to a milder angle of impingement of the deflected
nappe on the siphon’s soffit and less detrainment of air from the roller.

In an attempt to confirm the formation of free vortex flow in the hood for Qw  50 l/s, it
was decided to compare the measured water discharge with that predicted by an
equation developed for an ideal fluid in a rectangular conduit bend (Webber, 1993) as:

p 2  p1
2
Qw
 [1  ( R1 / R2 ) 2 ] , (2.8)
 2 g[bR1 Ln( R2 / R1 )] 2

in which 1 and 2 refer to the parameters at the inner and outer radii of the bend in the
horizontal plane respectively. For a siphon’s hood in the vertical plane, the difference in
pressure head at the crown and crest would be

2
Qw
H 2  H1  [1  ( R1 / R2 ) 2 ]  ( R2  R1 ) , (2.9)
2 g[bR1 Ln( R2 / R1 )]2

where, b is the throat width and H2, R2 and H1, R1 are pressure head and radius at the
crown and crest respectively.

The ratio of predicted discharge, Qw-p, (Eqn 2.9) to the measured discharge, Qw-m,
together with the variation of subatmospheric pressure in the hood and barrel, are shown
in Fig. 2.13. It is seen that for Qw  50 l/s, the ratio of Qw-p/Qw-m reached a fairly
constant value of about 1.3 and appeared to confirm the development of free vortex
flow in the hood. The 30% difference between the measured and predicted discharges is
believed to be due to general turbulence in the mainstream and boundary layer friction.

51
Chapter2 Characteristics of Air-Regulated Siphon Spillways

Fig. 2.11 shows that for Qw  70 l/s, despite the increasing Hsub in the hood, the airflow
decreased to zero as the air holes were deeply submerged by the upstream water level,
ya, (Fig. 2.12). The fall of Hsub values evident in Fig. 2.11 reflects the complete filling of
the barrel as blackwater conditions developed.

Hsub-crown
Beam
Hsub-barrel

Hsub-hole
Z

ya
19 mm
22 holes
 mm
l = 65 mm
49.35o Hsub-crest

Fig 2.10 Parameters influencing free vortex flow in the siphon’s hood (Not to scale).

300 7
Hsub-cro wn
Hsub-barrel
6
Hsub-ho le
Qa
5
200
Hsub [mm ]

Qa [l/s]

100
2

0 0
0 25 50 75 100

Qw [l/s]

Fig 2.11 Variation of Qa and Hsub in the siphon’s hood and barrel.

52
 Chapter2 Characteristics of Air-Regulated Siphon Spillways

10 7
ya

Qa 6
8

5
6
ya [mm ]

Qa [l/s]
4
3

2
2
mm
0
1

-2 0
0 20 40 60 80 100

Qw [l/s]

Fig 2.12 Variation with Qw of Qa and the upstream water level relative to the invert of the air
holes.

500 5
Hsub-crest
Hsub-cro wn
Hsub-barrel

Qw-predictedted / Qw-measured
400 4
Qw-p/Qw-m
Hsub [mm ]

300 3

200 2

100 1

Free vo rtex flo w

0 0
0 25 50 75 100

Qw-measured [l/s]

Fig 2.13 The ratio of predicted water discharge for free vortex flow in the hood and the measured
discharge in the siphon model.

53
Chapter2 Characteristics of Air-Regulated Siphon Spillways

2.4 Summary
The results of experiments on the Pergau siphon model can be summarised as follows:

1. For the siphon geometry of Fig. 2.6, the water discharge, Q w, was found to depend
on

i a combination of pond elevation, yp, and subatmospheric pressure, Hsub;

ii air hole area which was further modified by the rising waterlevel upstream of
the air holes, ya, as Qw increased and
iii varying flow patterns in the hood and near the deflector which altered the supply
and demand curves for air.
2. The air entering the siphon depended on the air supply and demand characteristics.
It was found that the height of water in the hood above the level of the air holes, the
subatmospheric pressure in the hood and the formation of a free vortex flow around
the siphon’s crest all had an influence on the air supply.

3. In addition to the turbulence level in the deflected stream, the angle between the
impinging nappe and the siphon’s soffit was found to influence the air demand, the
maximum occurring when a mild angle generated no roller above the nappe.

4. For the geometry represented in Fig. 2.8 where the hole area remained constant, the
air demand characteristics were no longer influenced by (ii) above and blackwater
conditions were eliminated, but the influence of (iii) above remained.

5. The geometry relating to Fig. 2.9 with air holes beyond the crown provided the
desired air demand curve because it was believed to be uninfluenced by the air
supply characteristics.

In an attempt to gain a deeper understanding of the air supply process itself, a simplified
siphon model was developed as described in the following chapter.

54