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

net/publication/358267186

Air Demand of a Hydraulic Jump in a Closed Conduit

Article in Journal of Hydraulic Engineering · February 2022

DOI: 10.1061/(ASCE)HY.1943-7900.0001963

CITATIONS READS
7 427

4 authors:

Pengcheng Li Zhu David

Ecofish Research; University of Alberta Lenovo
11 PUBLICATIONS 28 CITATIONS 254 PUBLICATIONS 3,984 CITATIONS

SEE PROFILE SEE PROFILE

Tingyu Xu Jian Zhang

Hohai University Hohai University
9 PUBLICATIONS 17 CITATIONS 77 PUBLICATIONS 420 CITATIONS

SEE PROFILE SEE PROFILE

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

Study of thermally stratified flows View project

Basin ecological regulation View project

All content following this page was uploaded by Tingyu Xu on 22 April 2022.

The user has requested enhancement of the downloaded file.

 Air Demand of a Hydraulic Jump in a Closed Conduit
Pengcheng Li 1; David Z. Zhu, M.ASCE 2; Tingyu Xu 3; and Jian Zhang 4

Abstract: Understanding the air demand of a hydraulic jump in a closed conduit is important in hydropower operations and urban sewer
designs. In this study, physical experiments are described to study flow regimes and the air demand of a hydraulic jump in a closed conduit
with various submerged outlet depths. Flow regimes with a submerged outlet were defined following previous studies based on outlet depth.
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.

Free-surface supercritical flow with a hydraulic jump can induce a relative air demand (air flow rate to water flow rate) of approximately 3%–
14%. If the hydraulic jump is followed by pressurized pipe flow, the air demand decreases with increasing outlet depth until the return roller is
fully developed in the pipe. For the partially submerged hydraulic jump, the relative air demand is significantly reduced to less than 1%. Field
measurements of the air demand at the Hugh Keenleyside Dam, British Columbia, were consistent with the experimental measurements
for the partially submerged hydraulic jump. The dynamics of the air pocket upstream of the hydraulic jump was studied. If the air supply
was constrained by nozzles of various sizes placed on the top of the air vent, the air pressure in the closed conduit decreased and the
hydraulic jump was pushed upstream. The energy loss coefficient in the air vent was also studied. DOI: 10.1061/(ASCE)HY.1943-
7900.0001963. © 2021 American Society of Civil Engineers.
Author keywords: Air demand; Air vent; Closed conduit; Hydraulic jump; Limited air supply.

Introduction
Air can be entrained into flowing water by a hydraulic jump.
Understanding the issues of air demand and proper ventilation
of a hydraulic jump inside a closed conduit are important in hydro-
power operations and urban sewer designs. For instance, negative
pressure, blowback, and other problems can be caused by entrained
air in pipes or closed conduits in hydropower facilities (Bosman
et al. 2016). Air entrainment and transport are also important in
municipal engineering applications as in ventilation and sewer
odor–related issues (Edwini-Bonsu and Steffler 2004; Qian et al.
2017). Additionally, the entrained air bubbles can be dissolved into
water in the tailrace of hydropower plants and cause gas bubble
disease to fish due to supersaturated total dissolved gases or dis-
solved oxygen (Weitkamp and Katz 1980; Chapman 1986; Stefan
and Fang 1994). Thus, it is crucial to predict the air demand in a
closed conduit under different outlet conditions to ensure the proper
design and operation of these hydraulic structures.
Flow in closed conduits can have different flow patterns and
flow regimes, some of which involve hydraulic jumps in different
forms. Sharma (1976) classified six types in the study of air entrain-
ment downstream of a sluice gate in a low-level outlet for various
1 conduit with a vertical gate to investigate the air demand for
Ph.D. Candidate, Dept. of Civil and Environmental Engineering, Univ.
of Alberta, Edmonton, AB, Canada T6G 1H9. Email: pli5@ualberta.ca
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of
Alberta, Edmonton, AB, Canada T6G 1H9; Professor, College of Civil
and Environmental Engineering, Ningbo Univ., Zhejiang 315211, China
(corresponding author). Email: david.zhu@ualberta.ca
3
Visiting Student, Dept. of Civil and Environmental Engineering, Univ.
of Alberta, Edmonton, AB, Canada T6G 1H9; Ph.D. Student, College of
Water Conservancy and Hydropower Engineering, Hohai Univ., Nanjing
210098, China. Email: tingyu_hhu@163.com
4
Professor, College of Water Conservancy and Hydropower Engineer-
ing, Hohai Univ., Nanjing 210098, China. Email: jzhang@hhu.edu.cn
Note. This manuscript was submitted on March 30, 2021; approved on
October 15, 2021; published online on December 7, 2021. Discussion per-
iod open until May 7, 2022; separate discussions must be submitted for
individual papers. This paper is part of the Journal of Hydraulic Engineer-
ing, © ASCE, ISSN 0733-9429.
gate openings and the downstream submergence. Stahl and Hager
(1999) identified four different types of hydraulic jumps, from un-
dular hydraulic jump to hydraulic jump with a near full pipe flow,
based on the approach flow Froude number and the gate opening.
The characteristics of a hydraulic jump and its return roller have
been systematically investigated by Rajaratnam (1967) and Long
et al. (2011); air entrainment and transport are affected by the hy-
draulic jump (Chanson 1996; Mortensen et al. 2012; Takahashi and
Ohtsu 2017). Additionally, air flow induced by flowing water in
pipes has received significant attention in recent decades (e.g., Qian
et al. 2020). Air entrainment in a closed conduit with supercritical
flow with or without a hydraulic jump has also been widely inves-
tigated. Kalinske and Robertson (1943) conducted the very first
experiments on air entrainment of hydraulic jumps within circular
pipes. The measured air entrainment rate was correlated to the
Froude number upstream of the jump and found to be independent
of pipe length and slope. Rajaratnam (1967) proposed that the air
entrainment of a hydraulic jump originates mainly from the drag of
the surface of the supercritical flow and the turbulent mixing gen-
erated by the hydraulic jump. Wisner (1967) correlated the air de-
mand induced by free-surface flow as well as hydraulic jumps to
the inflow Froude number based on experimental measurements.
Sharma (1976) conducted a series of experiments in a rectangular
free-surface flow with a hydraulic jump. Wisner et al. (1975) and
Rabben et al. (1983) also developed air-demand relations for hy-
draulic jumps in different cross sections, suggesting that the air
demand in rectangular pipes was larger than that in circular pipes.
Escarameia (2007) compared these existing data and indicated
that the downstream outlet conditions could substantially affect
air demand, but no detailed analysis was conducted, i.e., the ef-
fects of the outlet conditions on air demand were not systemati-
cally assessed. Even though flow conditions and hydraulic jump
types have been widely discussed in closed conduit flow, knowl-
edge on air flow demand under various submerged outlet depths
remains limited.
Air vents allow air to enter a structure to ensure proper air sup-
ply and prevent structural damages or cavitation. Thus, the size of
air vents must be sufficient to meet the required air demand. The air

© ASCE 04021058-1 J. Hydraul. Eng.

J. Hydraul. Eng., 2022, 148(2): 04021058

 Fig. 1. Schematic of experimental setup.
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
vent design, such as its size and the loss coefficient, can influence
the air demand in the closed conduit (Hohermuth et al. 2020). Cur-
rently, literature studies are limited regarding the relation between
air vent size, air demand, and hydraulic jump behaviors (Falvey
1980; Tullis and Larchar 2011; Hohermuth et al. 2020). Addition-
ally, while the characteristics of a hydraulic jump and the air en-
trainment prediction within closed conduits or sewer systems have
been studied (Chanson 1996; Unsal et al. 2008; Vos 2011; Ozkan
et al. 2014), no attention has been paid specifically to the effects of
a submerged outlet. This study examines the influences of submer-
gence on hydraulic jump characteristics and air demand. Addition-
ally, the characteristics of the hydraulic jump with limited air
supply under a submerged outlet are also investigated. This study
will aid to predict the air demand and ensure the proper operation of
hydraulic structures with a hydraulic jump in a closed conduit
under various outlet conditions.
Experimental Program
Experiments using a hydraulic jump in a closed conduit with
various outlet submergence levels were conducted in the T. Blench
Hydraulics Lab at the University of Alberta. The experimental
setup consisted of a 2-m-long (L) horizontal perspex pipe with a
diameter (D) of 0.20 m, a control gate with an adjustable opening
(a), and a water tank downstream of the pipe with an adjustable gate
to control the outlet depth (H), as shown in Fig. 1. To change the
Froude number, two different relative openings of the control gate
(a=D) were selected. A gradually varied flow profile of increasing
flow depth downstream of the control gate was formed from the
vena contracta (h0 ) to the toe ahead of the hydraulic jump (h1 ).
The hydraulic jump was recorded by a video camera at a speed
of 30 frames per second and a resolution of 1,920 × 1,080 pixels
for 30 s to determine the average distance from the jump toe to the
pipe outlet (Lp ) using an image processing tool as described in
what follows. A visual measurement using a tapeline from outside
the pipe was also used to validate Lp .
A vertical air vent of diameter d0 ¼ 50 mm was placed down-
stream of the control gate. Different sizes of nozzles were inserted
on top of the air vent (d ¼ 7–25 mm) to study the effect of con-
strained air supply on the characteristics of the hydraulic jump.
Experiments were first conducted without a nozzle under 30%
or 40% gate openings (Experiments A and B, respectively). Then
various nozzle sizes (d) were placed on the top of the air vent with
30% gate opening (Experiment C). Details of these experiments are
shown in Table 1.
Water was supplied by a pump, the flow rate Qw was controlled
by a gate valve and measured by a magnetic flow transmitter (Fox-
boro, I/A series, Schneider Electric Systems, Rueil-Malmaison,
France) with an accuracy of 0.5% of the full scale. The air
© ASCE 04021058-2
J. Hydraul. Eng., 2022, 148(2): 04021058
Table 1. List of experiments
Experiments D (m) d (mm) a=D Qw (L=s) F0 H=D
A 0.2 50 30% 15–30 5.1–10.2 0.6–3.0
B 0.2 50 40% 15–30 2.9–5.8 0.7–2.25
C 0.2 0–50 30% 25 8.5 1.5
velocity was measured on the top of the air vent using a hotwire
air velocity transmitter (FMA 1000 series, Omega, Laval, Quebec).
The air velocity transmitter had an accuracy of 1.5% of the full
scale. Two points were selected when measuring the air velocity,
one at the center, the other at the half radius from the center. Air
velocities were recorded at a frequency of 2 Hz for 180–300 s. The
influence of measurement duration on the mean air velocity can be
neglected when the measurement duration is not less than 180 s.
The air pressure inside the air vent was measured using a pressure
transducer (Model 264 Differential Pressure Transducer, Setra Sys-
tems, Boxborough, Massachusetts) at the top, center, and bottom of
the air vent (Fig. 1) at a frequency of 4 Hz for 300 s. The pressure
transducer had an accuracy of 0.6 Pa. The air pressure was then
used to calculate the mean air velocity U avg and the air flow rate
using the energy equation. The point air velocity measurements
were used to compare with U avg .
If a hydraulic jump was partially submerged and the air entrain-
ment rate was low, a high-speed camera (Phantom v211, Vision
Research, Wayne, New Jersey) was placed near the downstream
end of the pipe to record the bubble movement and bubble size
distribution for calculating the air demand. High-speed videos
were taken at a speed of 200 frames=s for approximately 5 s with
a resolution of 1,280 × 800 pixels. Because it was difficult to
directly capture each single bubble due to its overlap using an
automated image-processing tool (MATLAB version R2019b),
some bubbles were traced manually in Fiji (Fiji Is Just ImageJ
2017), which is a Java-based open-source image-processing
program. The bubble shape was considered to be ellipsoidal, of
which the major axis in the horizontal direction and minor axis
in the vertical direction were measured directly from the image.
The third axis perpendicular to the image plane (also in the hori-
zontal direction) was assumed to equal the major axis because the
difference between the two horizontal axes was less than 10% (Liu
et al. 2015).
Results and Discussions
Hydraulic Jumps in Circular Conduits
Five flow regimes can be defined following the classification pro-
posed by Sharma (1976), as illustrated in Fig. 2. In Regime 1, the
flow downstream of the control gate is free-surface flow and no
J. Hydraul. Eng.
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Different flow regimes.

hydraulic jump is formed when the outlet depth H (water depth
above the invert of the pipe outlet) is sufficiently small. In Regime
2, a hydraulic jump is formed and followed by free-surface flow
with H < D. Under this flow regime, the jump is similar to a
classical hydraulic jump with a surface roller and a bottom forward
flow axially concentrated as a surface jet (Hager 1992). As H in-
creases, the hydraulic jump with a flow recirculation is followed by
pressurized full-pipe flow, under which the downstream end of the
roller is not confined within a closed conduit (Regime 3a) or the
roller is completely confined within a closed conduit (Regime 3b).
In Regime 4, the hydraulic jump is partially submerged when the
conjugate depth is larger than the pipe diameter. The flow charac-
teristics and air demand of Regimes 3 and 4 need further investi-
gation since they have not been fully investigated. If the hydraulic
jump is fully submerged, the water level rises into the air vent pipe
without any air entrainment in Regime 5.
The effects of the outlet depth (H) on the flow regime for
Experiment A with a=D ¼ 30% are shown in Fig. 3(a). The Froude
number of the flow is calculated at the vena contracta with
F0 ¼ 8.5, where F0 ¼ V 0 =ðgh0 Þ0.5 , V 0 = average water velocity,
and h0 = hydraulic depth at the vena contracta, which is the division
of the cross-sectional flow area by the water surface width. If
H=D ¼ 0.8, a hydraulic jump was observed but the flow remained
free surface within the pipe, indicating Regime 2 flow. If H=D in-
creased to between 1.0 and 1.75, a hydraulic jump was followed by
pressurized flow in the pipe (Regime 3). In Regime 3a, under which
H=D ¼ 1.0 and 1.25, the downstream end of the return roller was
not observed within the conduit as shown in Fig. 3(b). If the dis-
tance from the toe of the hydraulic jump to the pipe outlet (Lp )
increased and was longer than the roller length [Lr in Fig. 3(b)],
a full turbulent roller was developed inside the pipe (Regime 3b),
with H=D ¼ 1.5 and 1.75. The measurements of Lr were in the
range of 4 to 4.5D, which were close to those of Mortensen et al.
(2012) with a similar Froude number. Additionally, Lr could also
be reasonably predicted by the method of Stahl and Hager (1999)
using the approach flow Froude number. If H=D was 2.0 and
2.25, hydraulic jumps were partially submerged (Regime 4). The
hydraulic jump is fully submerged without air entrainment for

Fig. 3. (a) Hydraulic jumps with various outlet depths with F0 ¼ 8.5 and a=D ¼ 30%; and (b) photos of Regimes 2 and 3.

© ASCE 04021058-3 J. Hydraul. Eng.

J. Hydraul. Eng., 2022, 148(2): 04021058

 H=D ≥ 2.5. Note that Lp and Lr are influenced not only by the pipe
length but also by the outlet depth and the Froude number.
Downstream of the control gate, the water depth increases to-
ward the hydraulic jump as a gradually varied flow. With increasing
outlet depth H, the hydraulic jump was pushed toward the control
gate, so the depth at the toe of the hydraulic jump h1 (Fig. 1) would
increase. If h1 equaled the depth at the vena contracta h0 , the jump
was no longer a free jump (i.e., the flow regime was transferred
from Regime 3b to Regime 4). In Regimes 1–4, the mean air pres-
sure P1 at the bottom of the air vent could drop below atmospheric
pressure.
The depth at the toe of the hydraulic jump (h1 ) in a circular
horizontal closed conduit can be calculated by applying the mo-
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
mentum equation across the hydraulic jump when the downstream
outlet is submerged for Regimes 3 and 4, and the friction is
assumed to be negligible compared to other terms in the equa-
tion (Chow 1959):
Q Q
ρgA1 h̄1 þ ρ w Qw ¼ ρgA2 h̄2 þ ρ w Qw − P1 A2 ð1Þ
A1 A2
where A1 = cross-sectional flow area at toe of jump; A2 ¼ πD2 =4 =
flow area at downstream end of pipe; and h̄1 and h̄2 = distances
from water surface to centroid of the up- and downstream sections
of hydraulic jump, respectively, with h̄2 ¼ H − D=2.
If h1 ¼ h0 , the critical Froude number at the vena contracta (F0c )
can be used to determine whether the hydraulic jump is partially
submerged using Eq. (1) by dividing by gh0 :
F20c ¼ ½A0 h̄0 − A2 ðh̄2 − P1 =ρgÞ=½ðA20 =A2 − A0 Þh0  ð2Þ
Under a specific outlet depth, the hydraulic jump will be par-
tially submerged (Regime 4) if F0 is less than the critical value F0c
as shown in Figs. 4(a and c). Otherwise, the hydraulic jump is not
submerged. The calculated F0c along with various outlet depths has
Fig. 4. Froude number under specific submerged water depth in circular closed conduit with (a and b) a=D ¼ 30%; and (c and d) a=D ¼ 40%
(assuming P1 is negligible).
© ASCE 04021058-4
J. Hydraul. Eng., 2022, 148(2): 04021058
good agreement with the experimental data, as shown in Figs. 4(a
and c). Note that F0c increases with H, indicating that a hydraulic
jump with a higher outlet depth needs a larger flow rate to remain
unsubmerged.
Furthermore, a fully submerged hydraulic jump in a circular
pipe can also be determined by applying the momentum equation
[Eq. (1)] between the control gate and the downstream end of the
pipe. For a fully submerged hydraulic jump in a circular pipe, the
hydrostatic pressure at the control gate can be calculated using
the full-pipe flow depth D. The critical Froude number at the con-
trol gate Fac , and the corresponding critical outlet depth, above
which the hydraulic jump would be fully submerged, is in good
agreement with the measurements, as shown in Figs. 4(b and d).
Therefore, the flow regimes can be determined using the momen-
tum equation along with Fig. 4.
Air Demand of Different Flow Regimes
The air demand was determined by the vent pipe area and the mean
air velocity, U avg , which was calculated from the air pressure mea-
sured inside the air vent (d0 ¼ 50 mm) using the energy equation.
The instantaneous air velocity (U a ) was calculated using the energy
equation between the atmosphere and the point inside the air vent:
−Pi ¼ ½ð1 þ KÞρa U 2a , where Pi is the instantaneous air pressure
at the top of the air vent, ρa is the air density, and K ¼ 1 is the local
loss coefficient of the inward-projecting (reentrant) entrance
(Mott 1994). The mean air velocity was calculated from U avg ¼
ð∫ T0 U a dtÞ=T, where T ¼ 300 s is the total measuring duration,
and dt is the time interval. A number of measurement durations
T were examined and the influence of T on U avg can be neglected
if T > 240 s. Based on the measurement of hotwire air velocity
transmitter, the air velocity at the center of the air vent was approx-
imately 3%–22% higher than that at the half radius from the center,
and the average velocity at these two points was approximately
J. Hydraul. Eng.
 7%–14% less than U avg . Overall, the mean air velocity U avg calcu-
lated from the air pressure is considered reasonable, so it was sub-
sequently used to calculate the air flow rate (Qa ) and the relative air
demand (β ¼ Qa =Qw ) in this study.
The air pressure behaviors were different at various locations in
the air vent, as shown in Fig. 5. When the Froude number at the
vena contracta F0 ¼ 8.5 with H=D ¼ 1.5, the mean air pressure
was between −4.2 and −3.8 Pa at different locations. However,
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
Fig. 5. Measured air pressure at different locations of air vent with
F0 ¼ 8.5 for H=D ¼ 1.5.
Fig. 6. (a) Comparison of measured air demand β for Regimes 2, 3, and 4 as a function of Froude number with previous prediction equations; and
(b) effects of outlet depth on β with 30% gate opening (circle) and 40% gate opening (square) for Regimes 2–4. (Regime 5 has no air demand.)
© ASCE 04021058-5
J. Hydraul. Eng., 2022, 148(2): 04021058
the pressure transducer had an accuracy of 0.6 Pa. The air pres-
sure fluctuation at the bottom of the air vent was damped toward the
top of the air vent, and the influence of air pressure fluctuation
caused by the hydraulic jump can be neglected at the top of the
air vent through the development of a computational fluid dynam-
ics (CFD) model (see Supplemental Materials for a discussion of
the air pressure fluctuation and CFD model setup). The measured
air pressure was sometimes positive, as shown in Fig. 5, indicating
flow out to the atmosphere. A few tufts were placed on the top of
the air vent during the experiment to indicate that the flow direction
of the air was upward or downward. For instance, the tuft direction
became upward about every 2.2 s and stayed approximately 0.2 s
before it changed downward for H ¼ 1.5D with F0 ¼ 8.5, which
was close to the cycle time (approximately 1.9 s) of air pressure
changed from negative to positive (Fig. 5). Note that the value
of the upward air velocity was approximately 45% of the down-
ward velocity, and the duration of the outward air flow was short
(Fig. 5). Overall, the measurements of pressure fluctuation com-
pared reasonably with the tuft movement.
From the measurements with an air vent size of d0 ¼ 50 mm,
the relative air demand β was found to be influenced by the outlet
depth as well as the flow regime in the pipe, as shown in Fig. 6.
Note that for Regime 4, the air flow rate is small, and only indi-
vidual air bubbles in the submerged pipe can be measured. Calcu-
lated from the bubble sizes and movements, the relative air demand
was between 0.03% and 0.63%, which was in a range (β ¼
0.03%–0.72%) similar to that obtained from the air velocity
(U avg ) and the vent tube area. Thus, the measurement of air demand
is reliable. Air demands in the experiments at various Froude num-
bers and outlet depths were compared with the previous air demand
prediction models in the literature, as shown in Fig. 6(a). Previous
models could only predict the air demand for free-surface flow with
or without hydraulic jump (e.g., Kalinske and Robertson 1943;
Sharma 1976; Hohermuth et al. 2020), but these equations did
not reasonably predict the air demand at different outlet depths
under the same F0 (e.g., Regimes 3 and 4). Fig. 6(b) indicates that
for a low-level outlet, the air demand in a free-surface flow is higher
than that of a submerged outlet at the same F0 . Clearly, F0 is not the
only governing parameter influencing the air demand in a closed
conduit with a submerged outlet [Fig. 6(b)]. Therefore, it is neces-
sary to improve the existing air demand prediction under various
flow regimes.
J. Hydraul. Eng.
 In Regime 1 (supercritical free-surface flow without a hydraulic
jump), the free surface produces a drag force on the air mass above
the water surface (Valero and Bung 2016). Because of the drag
force on the air, the water velocity has a significant influence on
air velocity (Campbell and Guyton 1953). Although the air demand
in Regime 1 was not measured in this study, Sharma (1976) pro-
posed a linear curve to predict its relative air demand in Regime 1
(β ¼ 0.09F0 ) based on the Froude number at the vena contracta.
Hohermuth et al. (2020) also proposed an equation for estimating
the air demand in Regime 1 (β 1 ) incorporating the effects of tunnel
length and air vent loss coefficient:
 0.8  0.25
Ad L K 1 ðH
β 1 ¼ 0.037F1.3 ð3Þ
0
AD ðζ þ 1Þ 0.5
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
higher than that of a submerged outlet (Regimes 3 and 4), as illus-
trated in Fig. 6(b).
If the outlet depth H is greater than 1D (Regime 3), the hy-
draulic jump is followed by pressurized pipe flow (H > D and
F0 > F0c ). It is found that the relative air demand in Regime 3a
β 3 starts decreasing from H ¼ D until the flow becomes that in
Regime 3b, under which the air demand is independent of the outlet
depth. Therefore, the relative air demand in Regime 3a can be cal-
culated from Eq. (4) by adding a new term determined by F0 and
H=D as
D − 1Þ
β3 ¼ β2 − ð5Þ
F0 − 1

where Ad = cross-sectional area of air vent; AD = cross section of

pipe; and ζ = air vent loss coefficient. Fig. 7(a) compares the air
demand for Regimes 1 and 2 with the existing equation proposed
by Sharma (1976), Hohermuth et al. (2020), and Kalinske and
Robertson (1943). Note that the significantly larger air demand
in Regime 1 calculated from Sharma (1976) is likely due to its
larger air vent and longer tunnel.
In Regime 2, there is a hydraulic jump followed by free-surface
flow, and the relative air demand β 2 can be well predicted by the
equation proposed by Kalinske and Robertson (1943), as shown in
Fig. 7(a):
β 2 ¼ 0.0066ðF0 − 1Þ1.4 ð4Þ
However, Eq. (4) only applies in predicting the relative air de-
mand in a closed conduit with a free-surface outlet (H < D). The
relative air demand in a free-surface low-level outlet (Regime 2) is
where K 1 is the dimensionless coefficient for Regime 3a. If
Lr < Lp , the flow is in Regime 3b, and the relative air demand
is not changed by H. At K 1 ¼ 0.294, the calculated relative air de-
mands are in good agreement with the experimental data of Regime
3, as shown in Fig. 7(b). The physics of the air demand will be
discussed in what follows.
For Regime 4 where the hydraulic jump is partially submerged
(H > D and Fac < F0 < F0c ), air bubbles are mainly entrained by
the breaking of the surface roller instead of by the advective dif-
fusion region (Rajaratnam 1967; Takahashi and Ohtsu 2017).
Therefore, the relative air demand is significantly decreased com-
pared with the other two regimes, as shown in Fig. 6(b). In this
case, a correction factor K 2 =ðF0 − 1Þ2 was added to develop the
prediction equation. The relative air demand also decreases with
(H=D − 1) and (F0 − 1). Therefore, the prediction equation for
the relative air demand in Regime 4 β 4 can be written

Fig. 7. Relative air demand under (a) Regimes 1 and 2; (b) Regime 3; and (c) Regime 4.

© ASCE 04021058-6 J. Hydraul. Eng.

J. Hydraul. Eng., 2022, 148(2): 04021058

 K2 β3
β4 ¼ ð6Þ
ðF0 − 1Þ2
where K 2 ¼ 4 is the model coefficient fitted from the experimental
data. Fig. 7(c) compares the predicted and measured air flow rates
in Regime 4. Note that the relative air demand in Regime 4 was
very low, which can be neglected in many predictions and appli-
cations. However, even a small amount of relative air demand can
lead to a severe air supersaturation and fish kills, as discussed in Li
et al. (2020), where about 0.4% relative air demand generates harm-
fully high air supersaturation in the water downstream of a low-
level outlet. Therefore, the relative air demand prediction in Regime
4 is still relevant even at a small value of less than 1%.
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
The physics of the air entrainment and bubble transport in vari-
ous flow regimes are discussed here. From experimental measure-
ments, Lp and Lr are strongly affected by the outlet depth (H).
When air is entrained into the jump, it is subsequently either trans-
ported downstream or detrained in the roller, as shown in Fig. 8. As
suggested by Takahashi and Ohtsu (2017), the air concentration
distribution in the return roller is nearly independent of the hy-
draulic jump location if the jump is unsubmerged. When the outlet
depth is relatively low (e.g., H ¼ 1.25D and F0 ¼ 8.5, Regime 3a),
the downstream roller end is not within the conduit as illustrated in
[Fig. 8(a)], indicating that the roller is not fully developed in the
pipe. Entrained air bubbles that typically recirculate within the
roller are released immediately at the pipe outlet to the water tank
rather than being detrained to the upstream air pocket. As the outlet
depth increases, the jump moves upstream. The length of the roller
in the pipe also increases, causing more entrained air bubbles to
recirculate within the roller rather than being released downstream.
Bubbles recirculated in the roller would eventually be released up-
stream through the roller front (Skartlien et al. 2012). Therefore,
more entrained air is detrained as the outlet depth increases.
If the outlet depth increases, e.g., to H ¼ 1.75D, Lp is suffi-
ciently long for a fully developed roller to form, as shown in
Fig. 8(b). For flow conditions in Regime 3b, the whole return roller
is formed within the pipe and Lr barely changes. The same phe-
nomenon was also observed by Mortensen et al. (2012). Thus, there
Fig. 8. Entrained air bubble transport with (a) H ¼ 1.25D; and (b) H ¼ 1.75D. Data for F0 ¼ 8.5 and a=D ¼ 30%.
© ASCE 04021058-7
J. Hydraul. Eng., 2022, 148(2): 04021058
is no major difference in the amount of entrained air transported
upstream and terminated with detrainment under Regime 3b with
various H [Fig. 6(b)]. Additionally, flow with a high Froude num-
ber entrains more air and creates smaller bubbles due to more en-
ergetic rollers, resulting in reduced air detrainment (Schulz et al.
2020). Overall, increasing the outlet depth leads to an increase in
the length of both Lp and Lr , which would enhance the air detrain-
ment and lead to a smaller β.
Hydraulic Jump with Limited Air Supply
Sufficient air supply through air vents is needed to maintain the air
pressure and to reduce potential structural damage or operational
issues. To investigate the effects of constrained air inflow on the
characteristic of a hydraulic jump in the submerged conduit, a noz-
zle of various sizes was placed on the top of the air vent (d ¼ 7, 10,
15, 20, and 25 mm). The submerged outlet depth was H ¼ 1.5D
and F0 ¼ 8.5 with a gate opening a=D of 30% (Experiment C in
Table 1). Fig. 9 illustrates the hydraulic jump under various
amounts of air supply. The hydraulic jump characteristics appeared
to be identical when d=d0 was larger than 0.3, below which the
jump moved upstream. Owing to the reduced subatmospheric air
pressure P1 as shown in Fig. 10(a), the jump was sucked upstream
by the subatmospheric pressure. From Eq. (1), a negative P1 can be
regarded as an increase in the outlet depth H. The effect of P1 can
be accommodated by adjusting h̄2 [defined in Eq. (1)] by an
amount of −P1 =ðρgÞ in the tailrace: h̄20 ¼ h̄2 − P1 =ðρgÞ. Note that
−P1 =ðρgÞ increases with decreasing d0 , as shown in Fig. 10(a). The
roller length with limited air supply (d=d0 ¼ 0.2, Lr ¼ 4.9D) was
longer than that with unconstrained air supply (Lr ¼ 4.0D). As
d=d0 was reduced to 0.14, the bubbly flow became more transpar-
ent compared to that with unconstrained air supply as shown in
Fig. 9(d), indicating that the quantity of bubbles in the water
was reduced.
Similarly, the relative air demand decreased dramatically if d=d0
was less than 0.3, as illustrated in Fig. 10(b). This is because the air
supply was restricted by the high air vent loss and a small air vent
opening. If the air supply was limited, the magnitude of the
J. Hydraul. Eng.
 (a) (b)
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.

(c) (d)

Fig. 9. Hydraulic jumps with different air supply: (a) β ¼ 8.7% (d=d0 ¼ 0.5); (b) β ¼ 8.2% (d=d0 ¼ 0.3); (c) β ¼ 5.9% (d=d0 ¼ 0.2); and
(d) β ¼ 2.1% (d=d0 ¼ 0.14). Data for F0 ¼ 8.5, a=D ¼ 30%, and H ¼ 1.5D.

Fig. 10. (a) Air pressure; (b) relative air demand β; and (c) air vent loss coefficient ζ for various nozzle sizes. Data for F0 ¼ 8.5, a=D ¼ 30%, and
H ¼ 1.5D.

negative air pressure P1 further increased and the hydraulic jump modified outlet depth (h̄20 ). Fig. 10(b) illustrates that the measured
was pushed upstream or became partially submerged, which was and predicted β values are in good agreement at different d=d0 ,
similar to what happened after increasing the outlet depth. β can which indicates that the modified outlet depth h̄20 can be used to
be predicted using the developed equation in this study using a predict β for constrained air supply.

© ASCE 04021058-8 J. Hydraul. Eng.

J. Hydraul. Eng., 2022, 148(2): 04021058

 Air removed by the hydraulic jump results in a subatmospheric
pressure in the air pocket upstream of the jump and the air demand
through the air vent. The air pressure inside the pipe can be as-
sessed based on the air demand, air vent diameter, and loss coef-
ficient. The air vent loss coefficient ζ (based on the air velocity
at the nozzle) for various relative nozzle sizes d=d0 is shown in
Fig. 10(c). Note that the loss coefficient for an orifice plate is de-
termined by the orifice opening size, which can reach about 20
when the orifice opening is 30% of the pipe size (Potter and
Wiggert 2016; Hohermuth et al. 2020). In this study, a nozzle of
different sizes was placed on the top of the air vent. The energy loss
was mainly caused by the reentrant loss and the sudden expansion
loss following the nozzle. If there was no nozzle placed on the top
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
of the air vent (d=d0 ¼ 1), the air vent loss coefficient is 1.0, re-
flecting only reentrant loss, as reported in Mott (1994). With a noz-
zle, the flow at the nozzle expands into the air vent pipe (d0 ),
leading to an additional sudden expansion loss. The sudden pipe
expansion loss can be estimated analytically as ð1 − d2 =d20 Þ2 . Thus,
there is no expansion loss if d=d0 ¼ 1, and it increases to 1.0 when
d=d0 ¼ 0. The air vent loss coefficient ζ (based on the air velocity
at the nozzle), which can be considered the summation of the re-
entrant loss and the sudden expansion loss, increases from 1.0 to
about 2.0 if d=d0 decreases from 1 to 0.14 [Fig. 10(c)]. The fore-
going estimate of the loss coefficient is used to calculate the air
velocity at the nozzle. The calculated air velocity is in good agree-
ment with the anemometer measurements [Fig. 10(c)], indicating
the reliable prediction of the loss coefficient.
The air velocity at the nozzle also changes with decreasing
nozzle size and eventually becomes stable around 25 m=s for
d=d0 ¼ 0.14, as shown in Fig. 10(c). In this study, whistling started
if the air velocity exceeded 25 m=s (d ≤ 1 cm). Note that the noz-
zle and its expansion are likely among the factors enhancing the
whistling. Falvey (1980) proposed that the whistling sound oc-
curred for air velocity above 30 m=s. Thus, it is recommended that
special attention be paid to whistling noise when the air velocity in
an air vent exceeds 25 m=s. Note that an air velocity higher than the
recommended limit without whistling noise has been reported in
prototypes and larger models. Associated with a high air velocity,
one should also pay attention to possible subatmospheric pressure,
significant compressibility effects, or damage to hydraulic struc-
tures (Hohermuth et al. 2020).
The characteristics of the hydraulic jump and air pressure inside
the pipe was also monitored once the air vent was suddenly sealed
(d=d0 ¼ 0). Then the air pressure oscillated and dropped signifi-
cantly from approximately zero to approximately −1,400 Pa on
Fig. 11. (a) Cross section of low-level outlet in Hugh Keenleyside Dam; and (b) relative air demand β versus gate opening under fieldwork
conditions.
© ASCE 04021058-9
J. Hydraul. Eng., 2022, 148(2): 04021058
this experimental scale. The volume of the air pocket upstream
of the hydraulic jump decreased, and the amount of the air demand
by the hydraulic jump decreased from the normal value to zero in
about 20 s. More details are provided in the Appendix.
Prototype Application and Scale Effects
It was reported that the low-level outlets at the Hugh Keenleyside
(HLK) Dam, British Columbia, Canada, can cause supersaturation
of total dissolved gases (TDG). To investigate why and evaluate the
generation of TDG from the operation of the rectangular low-level
outlet gates, fieldwork was carried out from July 13 to 16, 2020.
The length of the closed conduit is 17.1 m with a width of 6.1 m and
a height of 8.0 m. The slope of the closed conduit is 14.8°. The air
entrainment into the low-level outlet mainly originates from the
gate hoist room, which is connected with the atmosphere through
windows, as shown in Fig. 11(a). Spot measurements of air velocity
were taken on the top of the gate well using a hotwire velocity me-
ter (TSI Alnor 9535 A, TSI, Shoreview, Minnesota) in the low-
level machinery room during the fieldwork. The size of the gate
well is 5.9 m × 2.7 m, and a platform equipped with ropes for lift-
ing the gate was placed on the edge of the top of the gate well. Spot
measurements of air velocity were taken at four different points.
The air velocity was recorded for 1 min with a 1-s interval at each
measurement point. Additionally, five SmartReaders (Plus 4, two-
channel pressure and temperature data logger, ACR Systems, Van-
couver, British Columbia, Canada) were used to continuously mea-
sure the differential pressure between the top of the gate well and
inside at different depths (2, 4, 6, 8, and 10 m) in the low-level
machinery room. However, the pressure difference was too small
to be accurately measured.
In the low-level outlet of the HLK Dam, the flow is in Regime 4,
and β can be calculated using the developed Eq. (6) based on lab
measurements. Overall, the predicted β shows a trend similar to that
of the field measurements, and both give relatively low air demands
(β < 1%), as illustrated in Fig. 11(b). The difference between the
prediction and the measurement is noticeable at small gate openings.
This is likely caused by the measurement uncertainties due to the
low velocity in Regime 4 in both the lab and field measurements.
Additionally, there could be scale effects, but likely not significant.
In general, the air demand prediction using Eq. (6) in Regime 4 can
provide a reasonable estimate for the air demand in the prototype.
Sakhuja et al. (1984) found that the scale effect factor for the
relative air demand in free-surface flows (Regime 1) increased with
model scale and could reach approximately 1.8 if the model scale
J. Hydraul. Eng.
 was 50. In Regime 2 (free-surface flow with a hydraulic jump),
Chanson and Gualtieri (2008) reported that the relative air demand
in a small-scale model (1:2) was underestimated by approximately
45%–60%. If a hydraulic jump is completely confined within a
closed conduit (Regime 3b), Mortensen et al. (2011) found that
the air demand was independent of the pipe size. Note that the pipe
diameter used in their study was between 76 and 591 mm, whereas
the pipe diameter in this study is 200 mm. Additionally, our air
demand in Regime 3a-b (β ¼ 2%–14%) is similar to that measured
in Mortensen et al. (2012) with β ¼ 0% to 20%. Thus, the air
demand is not sensitive to the pipe size in Regime 3 for the tested
size range. From the comparison between the field measurement at
the HLK Dam and model prediction of the air demand in Regime 4
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
hydraulic jump at the air–water interface. If the outlet of the conduit
is submerged and the hydraulic jump is fully developed, the relative
air demand is insensitive to the water surface turbulence. The tur-
bulence will be damped if the hydraulic jump is submerged, leading
to relatively insignificant scale effects on the relative air demand.
Therefore, scale effects appear to exist in Regimes 1 and 2, likely
not sensitive to the pipe sizes in Regime 3 in the test range. From
field measurement, the scale effect appears to be unimportant in
Regime 4.
Conclusion

[Fig. 12(b)], the air demand model prediction appears to provide a
good estimate for the prototype.
In Regime 1, the air demand is directly affected by the surface
turbulence of the water flow, which likely has scale effects. The
relative air demand in Regime 2 is influenced by both the water
surface turbulence and the turbulent shear layer generated by the
Regimes of supercritical flow with and without a hydraulic jump in
a closed conduit were investigated to predict the air demand for
different outlet depths. Regime 1 is the free-surface flow without
a hydraulic jump, in which the relative air demand (β ¼ Qa =Qw )
was estimated to be larger than other flow regimes. Regime 2 rep-
resents a hydraulic jump followed by free-surface flow, for which

Fig. 12. (a) Air pressure; (b) air demand; and (c) hydraulic jump patterns after sealing air vent. Data for F0 ¼ 8.5, a=D ¼ 30%, and H ¼ 1.5D.

© ASCE 04021058-10 J. Hydraul. Eng.

J. Hydraul. Eng., 2022, 148(2): 04021058

 the relative air demand depends only on the Froude number at the Data Availability Statement
vena contracta F0 . Regime 3 describes the hydraulic jump followed
by pressurized pipe flow with an undeveloped or fully developed All data, models, or code that support the findings of this study are
surface roller. The relative air demand is between 3% and 14%, and available from the corresponding author upon reasonable request.
decreases with increasing roller length until the roller is fully
developed in the pipe. Regime 4 gives an air demand significantly
smaller than 1% due to the partially submerged hydraulic jump. Acknowledgments
The air demand was also measured in a prototype low-level outlet
The writers gratefully acknowledge the financial support from the
of a dam with a partially submerged hydraulic jump, which was
Natural Sciences and Engineering Research Council (NSERC) of
consistent with the developed prediction equation based on exper- Canada and the British Columbia Hydro and Power Authority (BC
imental measurements. Hydro). The authors would also like to thank Perry Fedun for his
It is found that the pressure fluctuation at the bottom of the air technical assistance.
vent is much larger than that at the top. Additionally, various sizes
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.

of nozzles on the top of the air vent were tested to investigate the
influence of the air supply on the hydraulic jump. Limited air sup- Notation
ply caused a subatmospheric pressure, and the magnitude of the
subatmospheric pressure significantly increased after sealing the The following symbols are used in this paper:
air vent. The hydraulic jump was pushed upstream with decreasing A1 = cross-sectional flow area at jump toe;
air demand due to the reduced air pressure in the closed conduit. A2 = cross-sectional flow area at downstream pipe end;
Furthermore, the air vent loss coefficient increases with decreasing a = gate opening;
nozzle size and can be simplified as the addition of the reentrant D = pipe diameter;
loss and the sudden expansion loss if the relative nozzle size d=d0 is d = nozzle diameter;
less than 1. This study improves the general understanding of the
d0 = air vent diameter;
air demand of a hydraulic jump with a submerged outlet to ensure
F0 = Froude number at vena contracta;
the proper design and operations of urban sewer systems and
hydropower facilities. F0c = critical Froude number at vena contracta;
Fac = critical Froude number at control gate;
g = gravity acceleration;
Appendix. Hydraulic Jump with Sealed Air Vent H = water depth above invert of pipe outlet;
h0 = water depth at vena contracta;
The size and pressure of the air pocket in front of the hydraulic h1 = water depth upstream of hydraulic jump;
jump was also monitored when the air vent was sealed. When the h0 = hydraulic depth at vena contracta;
air vent was fully sealed at t ¼ 10 s, the air pressure oscillated and K = local loss coefficient of reentrant entrance;
decreased from approximately zero to approximately −1,400 Pa K 1;2 = dimensionless coefficient;
[Fig. 12(a)], which caused the hydraulic jump to move upstream. L = pipe length;
The mass of the air pocket at different times was estimated from the Lp = distance from jump toe to pipe outlet;
images of the air pocket size and the measured pressure assuming Lr = roller length;
the ideal gas law. After the air vent was sealed, β was calculated
Pi = instantaneous air pressure;
from the estimated decrease of the mass of the air pocket, as shown
P1 = mean air pressure at bottom of air vent;
in Fig. 12(b). β remained at between 5% and 8% in the first 10 s
Qa = air discharge;
after sealing the air vent (t ¼ 10–20 s) and then decreased signifi-
cantly to approximately 0.1%. After the air vent was sealed, the Qw = water discharge;
reduction of β from normal to around zero would take approximately T = total measuring duration;
20 s [Fig. 12(b)]. The hydraulic jump flow regimes at different times U a = instantaneous air velocity;
after sealing the air vent are illustrated in Fig. 12(c). Owing to U avg = mean air velocity;
the closed air vent, air mass in the air pocket was entrained into the V 0 = average water velocity;
hydraulic jump and transported downstream, contributing to the β = relative air demand, Qa =Qw ;
decreased air mass in the air pocket. The air pocket volume and ζ = air vent loss coefficient; and
pressure kept falling until t ¼ 31 s, as shown in Fig. 12(a). Later, ρa = air density.
the white water in the pipe became relatively clear (t ¼ 150 s),
suggesting that less air was transported downstream.
Although the air vent was sealed, some air bubbles were ob- Supplemental Materials
served to be continuously transported downstream, and the flow
did not transit to a completely drowned hydraulic jump even when Appendix S1 (air pressure fluctuation and CFD model setup) and
the experiment ran for 20 min, as shown in Fig. 12(c). One possible Fig. S1 are available online in the ASCE Library (www.ascelibrary
reason for the source of this small amount of air is that the low air .org).
pressure results in a decreased dissolved gas concentration from
100% saturation to approximately 98% with some air bubbles re-
leased into water. Using the air–water mass transfer model (Li et al. References
2020) and assuming the mass of dissolved air could be fully trans- Bosman, A., G. R. Basson, and D. E. Bosman. 2016. “Hydraulic model
ferred to the gas phase due to the decreased solubility, the calcu- study of the blowback behaviour of the bottom outlet of the Berg River
lated air void ratio (0.042%) was close to the measurement Dam, South Africa.” J. S. Afr. Inst. Civ. Eng. 58 (1): 43–52. https://doi
data (0.063%). .org/10.17159/2309-8775/2016/v58n1a5.

© ASCE 04021058-11 J. Hydraul. Eng.

J. Hydraul. Eng., 2022, 148(2): 04021058

 Campbell, F. B., and B. Guyton. 1953. “Air demand in gated outlet works.”
In Proc., 5th IAHR and ASCE Joint Conf., 529–533. Reston, VA:
ASCE.
Chanson, H. 1996. Air bubble entrainment in free-surface turbulent shear
flows. San Diego, CA: Academic Press.
Chanson, H., and C. Gualtieri. 2008. “Similitude and scale effects of air
entrainment in hydraulic jumps.” J. Hydraul. Res. 46 (1): 35–44.
https://doi.org/10.1080/00221686.2008.9521841.
Chapman, G. 1986. Ambient water quality criteria for dissolved oxygen.
Washington, DC: USEPA.
Chow, V. T. 1959. Open-channel hydraulics. New York: McGraw Hill.
Edwini-Bonsu, S., and P. M. Steffler. 2004. “Air flow in sanitary sewer
conduits due to wastewater drag: A computational fluid dynamics
approach.” J. Environ. Eng. Sci. 3 (5): 331–342. https://doi.org/10.1139
/s03-072.
Downloaded from ascelibrary.org by Hohai University Library on 03/23/22. Copyright ASCE. For personal use only; all rights reserved.
Escarameia, M. 2007. “Investigating hydraulic removal of air from water
pipelines.” Proc. Inst. Civ. Eng. Water Manage. 160 (1): 25–34. https://
doi.org/10.1680/wama.2007.160.1.25.
Falvey, H. 1980. “Air-water flow in hydraulic structures.” In USBR
engineering monograph. Denver: US Bureau of Reclamation.
Fiji Is Just ImageJ. 2017. “Fiji/ImageJ.” Accessed November 4, 2020.
https://imagej.net/Fiji.
Hager, W. H. 1992. Energy dissipators and hydraulic jump. Dordrecht,
Netherlands: Kluwer.
Hohermuth, B., L. Schmocker, and R. M. Boes. 2020. “Air demand of
low-level outlets for large dams.” J. Hydraul. Eng. 146 (8): 04020055.
https://doi.org/10.1061/(ASCE)HY.1943-7900.0001775.
Kalinske, A. A., and J. M. Robertson. 1943. “Closed conduit flow.” Trans.
ASCE 108 (1): 1435–1447.
Li, P., Y. Ma, and D. Z. Zhu. 2020. “Mass transfer of gas bubbles rising in
stagnant water.” J. Environ. Eng. 146 (8): 04020084. https://doi.org/10
.1061/(ASCE)EE.1943-7870.0001760.
Liu, L., H. Yan, and G. Zhao. 2015. “Experimental studies on the shape and
motion of air bubbles in viscous liquids.” Exp. Therm. Fluid Sci. 62 (Apr):
109–121. https://doi.org/10.1016/j.expthermflusci.2014.11.018.
Long, D., N. Rajaratnam, P. M. Steffler, and P. R. Smy. 2011. “Structure of
flow in hydraulic jumps.” J. Hydraul. Res. 29 (2): 207–218. https://doi
.org/10.1080/00221689109499004.
Mortensen, J. D., S. L. Barfuss, and M. C. Johnson. 2011. “Scale effects of
air entrained by hydraulic jumps within closed conduits.” J. Hydraul.
Res. 49 (1): 90–95. https://doi.org/10.1080/00221686.2010.536695.
Mortensen, J. D., S. L. Barfuss, and B. P. Tullis. 2012. “Effects of hydraulic
jump location on air entrainment in closed conduits.” J. Hydraul. Res.
50 (3): 298–303. https://doi.org/10.1080/00221686.2012.670008.
Mott, R. L. 1994. Applied fluid mechanics. 4th ed. Upper Saddle River, NJ:
Prentice-Hall.
Ozkan, F., M. C. Tuna, A. Baylar, and M. Ozturk. 2014. “Optimum
air-demand ratio for maximum aeration efficiency in high-head gated
circular conduits.” Water Sci. Technol. 70 (5): 871–877. https://doi.org
/10.2166/wst.2014.305.
Potter, M. C., and D. C. Wiggert. 2016. Mechanics of fluids. 4th ed. Boston:
Prentice Hall.
Qian, Y., W. Shao, D. Z. Zhu, K. A. A. Mohamad, P. M. Steffler, S. Edwini-
Bonsu, D. Yue, and D. Krywiak. 2020. “Modeling air flow in sanitary
sewer systems: A review.” J. Hydro-environ. Res. 38 (Sep): 84–95.
https://doi.org/10.1016/j.jher.2020.10.003.
Qian, Y., D. Z. Zhu, W. Zhang, N. Rajaratnam, S. Edwini-Bonsu, and P.
Steffler. 2017. “Air movement induced by water flow with a hydraulic
© ASCE 04021058-12
View publication stats J. Hydraul. Eng., 2022, 148(2): 04021058
jump in changing slope pipes.” J. Hydraul. Eng. 143 (4): 04016092.
https://doi.org/10.1061/(ASCE)HY.1943-7900.0001252.
Rabben, S. L., H. Els, and G. Rouve. 1983. “Investigations on flow aeration
at offsets downstream of high-head control structures.” In Proc., 20th
IAHR Congress Moscow USSR, 354–360. Esslingen, Germany:
International Association for Hydro-Environment Engineering and
Research.
Rajaratnam, N. 1967. “Hydraulic jumps.” In Advances in hydroscience.
New York: Academic Press.
Sakhuja, V. S., T. C. Paul, and S. Singh. 1984. “Air entrainment distortion
in free surface flows.” In Proc., Int. Symp. on Scale Effects in Modelling
Hydraulic Structures, 4.8.1–4.8.4. Esslingen, Germany: International
Association for Hydro-Environment Engineering and Research.
Schulz, H. E., J. G. Vasconcelos, and A. C. Patrick. 2020. “Air entrainment
in pipe-filling bores and pressurization interfaces.” J. Hydraul. Eng.
146 (2): 04019053. https://doi.org/10.1061/(ASCE)HY.1943-7900
.0001672.
Sharma, H. R. 1976. “Air-entrainment in high head gated conduits.”
J. Hydraul. Div. 102 (11): 1629–1646. https://doi.org/10.1061/JYCEAJ
.0004650.
Skartlien, R., J. A. Julshamn, C. J. Lawrence, and L. Liu. 2012. “A gas
entrainment model for hydraulic jumps in near horizontal pipes.” Int.
J. Multiphase Flow 43: 39–55. https://doi.org/10.1016/j.ijmultiphase
flow.2012.02.013.
Stahl, H., and W. H. Hager. 1999. “Hydraulic jump in circular pipes.” Can.
J. Civ. Eng. 26: 368–373. https://doi.org/10.1139/l98-068.
Stefan, H. G., and X. Fang. 1994. “Dissolved oxygen model for regional
lake analysis.” Ecol. Modell. 71 (1–3): 37–68. https://doi.org/10.1016
/0304-3800(94)90075-2.
Takahashi, M., and I. Ohtsu. 2017. “Effects of inflows on air entrainment in
hydraulic jumps below a gate.” J. Hydraul. Res. 55 (2): 259–268.
https://doi.org/10.1080/00221686.2016.1238016.
Tullis, B. P., and J. A. Larchar. 2011. “Determining air demand for small to
medium-sized embankment dam low-level outlet works.” J. Irrig.
Drain. Eng. 137 (12): 793–800. https://doi.org/10.1061/(ASCE)IR
.1943-4774.0000345.
Unsal, M., A. Baylar, M. Tugal, and F. Ozkan. 2008. “Increased aeration
efficiency of high-head conduit flow systems.” J. Hydraul. Res. 46 (5):
711–714. https://doi.org/10.3826/jhr.2008.3360.
Valero, D., and D. B. Bung. 2016. “Development of the interfacial air layer
in the non-aerated region of high-velocity spillway flows. Instabilities
growth, entrapped air and influence on the self-aeration onset.” Int. J.
Multiphase Flow 84: 66–74. https://doi.org/10.1016/j.ijmultiphaseflow
.2016.04.012.
Vos, A. 2011. Unsteady flow conditions at dam bottom outlet works due
to air entrainment during gate closure: Berg River Dam model.
Stellenbosch, South Africa: Univ. of Stellenbosch.
Weitkamp, D. E., and M. Katz. 1980. “A review of dissolved gas supersatu-
ration literature.” Trans. Am. Fish. Soc. 109 (6): 659–702. https://doi.org
/10.1577/1548-8659(1980)109<659:ARODGS>2.0.CO;2.
Wisner, P. 1967. “Hydraulic design for flood control by high head gated
outlets.” In Proc., 9th ICOLD Congress, 495–507. Paris: International
Commission on Large Dams.
Wisner, P. E., F. N. Mohsen, and N. Kouwen. 1975. “Removal of air from
water lines by hydraulic means.” J. Hydraul. Div. 101 (2): 243–257.
https://doi.org/10.1061/JYCEAJ.0004201.
J. Hydraul. Eng.