Experimental Study of Flow in a Vortex Drop Shaft

Can-Hua Zhao1; David Z. Zhu, M.ASCE2; Shuang-Ke Sun3; and Zhi-Ping Liu4

Abstract: Model experiments were conducted to investigate the performance of a vortex drop structure with a relatively small height to
diameter ratio. Detailed measurements of wall pressure and water thickness of annular jet flow were obtained along the vertical drop shaft,
and the rate of air entrainment was measured. The results confirmed the high efficiency of energy dissipation in the vortex drop structure
even for a relatively small drop height. The air entrainment rate was found to be significant, and good correlation was observed between
the rate of air entrainment and the water jet velocity. The one-dimensional frictional free-vortex flow model was extended to include the
effects of pressure forces. While the energy loss in the drop shaft can be simulated by correcting the friction factor, both the frictional
model and the extended model significantly underpredict the wall pressure.
DOI: 10.1061/共ASCE兲0733-9429共2006兲132:1共61兲
CE Database subject headings: Vortices; Drop structures; Energy dissipation; Air entrainment; Friction; Shafts.

Introduction tical shafts: the flow clings to the wall and spirals down while an
air core forms in the center, and under some circumstances an
Drop structures are commonly used to convey water from a annular hydraulic jump occurs close to the bottom. Flows in a
higher to a lower elevation in urban sewers and hydropower sys- vortex drop shaft can be classified into three regions 共Fig. 1兲:
tems. Two common types are plunge-flow drop structures annular jet flow, transition flow, and water cushion region 共Jain
共Rajaratnam et al. 1997兲 and vortex drop structures 共Jain 1987; 1988兲. An outlet structure aims to direct the nearly vertical annu-
Vischer and Hager 1995兲. In a plunge-flow drop structure, water lar jet to a horizontal conduit, to deaerate the air–water mixture,
falls uncontrolled in a vertical shaft. In a vortex drop structure, and to safely dissipate the remaining energy 共Vischer and Hager
angular momentum 共i.e., vortex兲 is imparted into the flow through 1995兲.
a special inlet design and the water spirals down clinging to the Vortex drops are widely used nowadays. Jain and Kennedy
wall of the drop shaft. Vortex drop structures are generally con- 共1984兲 reported the application of vortex drop shafts in the Mil-
sidered superior to plunge-flow ones 共Jain 1984兲 given that: 共1兲 waukee Metropolitan Sewerage District. Some of the drop struc-
they maintain a stable flow pattern for various discharges; and 共2兲 tures carry a discharge of up to 90 m3 / s over a drop height of
energy dissipation is enhanced in the vortex flow in the drop 80 m. Vischer and Hager 共1995兲 reported the vortex drop of Cur-
shaft. bans, Italy 170 m high and 7.30 m in diameter for a design dis-
Vortex drop structures consist of three main elements: an inlet charge of 140 m3 / s. Recently a vortex drop was built in China for
structure, a vertical shaft, and an outlet structure 共Fig. 1兲. The Shapai Power Station with a drop height of about 100 m and a
inlet structure produces a swirling flow as discussed by Jain and design discharge over 200 m3 / s 共Dong and Gao 1995兲. A vortex
Ettema 共1987兲 and Vischer and Hager 共1995兲. A typical inlet con- drop shaft was also proposed to convey water into an existing
figuration includes a spiral inlet 共Hager 1990兲, a tangential inlet diversion tunnel so as to form a tunnel spillway in another power
共Jain 1984兲, and a slot vortex inlet 共Quick 1990兲. Regardless of station of China 共Zhao et al. 2001兲. The drop shaft will have a
the inlet configuration, flow patterns are generally similar in ver- design discharge of 1,400 m3 / s and a drop height of 190 m.
Given high drops, the design of vortex drop shafts requires a
1 proper consideration of energy dissipation and air entrainment.
PhD Candidate, Dept. of Civil and Environmental Engineering, Univ.
Substantial energy dissipation in vortex drop shafts is ex-
of Alberta, Edmonton, AB, Canada, T6G 2W2. E-mail:
pected. The helical track of the vortex flow promotes energy dis-
[email protected]
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of sipation as a result of prolonged wall friction. Estimates of the
Alberta, Edmonton, AB, Canada T6G 2W2 共corresponding author兲. energy dissipation in earlier studies were based on the assumption
E-mail: [email protected] that the drop shaft is so long that the flow reaches its terminal
3
Senior Engineer, Dept. of Hydraulics, China Institute of Water velocity. For L = 50D 共where L is the length and D is the diameter
Resources and Hydropower Research, A-1 Fuxing Rd., Beijing 100038, of drop shafts兲 and a Manning’s n of 0.012, Vischer and Hager
China. E-mail: [email protected] 共1995兲 estimated 85% of energy dissipation. Jain and Kennedy
4
Professor, Dept. of Hydraulics, China Institute of Water Resources 共1984兲 predicted a 90% energy loss in a drop shaft of 100D with
and Hydropower Research, A-1 Fuxing Rd., Beijing 100038, China. a friction factor of 0.03. For drop shafts with a relatively small
E-mail: [email protected] ratio of L / D, the terminal velocity may not be reached and less
Note. Discussion open until June 1, 2006. Separate discussions must
energy loss is expected. Jeanpierre and Lachal 共1966兲 reported
be submitted for individual papers. To extend the closing date by one
month, a written request must be filed with the ASCE Managing Editor. 62% of energy dissipation for a drop of about 9D. For drop shaft
The manuscript for this paper was submitted for review and possible with small L / D the efficiency of energy dissipation needs to be
publication on September 12, 2002; approved on March 25, 2005. This investigated.
paper is part of the Journal of Hydraulic Engineering, Vol. 132, No. 1, Air entrainment is of another concern for a large drop. Air is
January 1, 2006. ©ASCE, ISSN 0733-9429/2006/1-61–68/$25.00. entrained by the high speed falling jet and its impinging onto the

JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006 / 61

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
 共1990兲 suggested ␴ = 1.25 as an economical and hydraulically ef-
ficient design. In the present study ␴ = 1.16 was used, which gives
D = 0.223 m for a maximum design discharge of 0.051 m3 / s. The
total drop height of the structure is L = 3.15 m and the length of
the shaft is 2.70 m. An aerator was set up on the top of the volute
chamber to replenish the air core, while five vent pipes with 4 cm
of diameter were mounted on the top of the outlet tunnel for
deaeration 共Fig. 1兲.
This model study was erected to provide design guidance for a
Chinese power station where a large vortex-flow drop shaft was
proposed to convey water to an existing diversion tunnel 共Zhao et
al. 2001兲. Some aspects of the model design were site specific.
Given the large discharge, a volute chamber with a diameter of
1.25D was installed above the drop shaft 共Fig. 1兲. To reduce con-
struction cost, the diameter was gradually reduced to D through a
transition section at a slope of 10:1. The reduction of shaft diam-
eter was also discussed by Jain 共1987兲, and a similar design has
been successfully applied at the Shapai Power Station in China
共Dong and Gao 1995兲. A deflector was installed at the entrance of
the volute chamber to mitigate the collision of the spiraling flow
with the incoming flow to improve the intake efficiency. A sump
Fig. 1. 共a兲 Sideview of experimental setup; and 共b兲 inlet was added at the base of the drop shaft to increase the energy
configuration. Locations of piezocrystal sensors 共쎲兲 共not to scale; all dissipation and to improve the flow pattern at the entrance of the
dimensions are in meters兲 outlet tunnel. However, Zhao et al. 共2001兲 found that the depth of
the sump had little effect on energy dissipation in this case.
Experiments were carried out for a design discharge of
bottom pool. While air demand in a vortex drop shaft is believed 0.051 m3 / s. Additional experiments with slightly smaller dis-
to be less than in a plunge-flow drop shaft 共Jain 1988兲, it never- charges of 0.047 and 0.049 m3 / s were also conducted to verify
theless could still be significant. Air supply in the structure is the consistency of the measurements. Discharges were measured
important in maintaining a stable air core. In the downstream with a rectangular weir. Wall pressure was measured by a total of
pipe, on the other hand, proper deaeration is needed to prevent the 50 piezocrystal pressure sensors arranged at 13 sections along the
reduction in the pipe discharge capacity due to the entrained air. volute chamber and the drop shaft 共Fig. 1兲. These sensors were
Some work on air entrainment in vortex drop structures has been calibrated to within 0.5% of the full scale 共20 kPa兲. The sampling
published 共Jain and Kennedy 1983; Jain 1988; Vischer and Hager frequency was 40 Hz and the sampling length was 30,000 data
1995兲. However, given the complexity of air-entrainment mecha- points. The thickness of the annular jet flow was measured using
nisms, further study is needed. a specially designed small L-shaped probe. The probe consisted
In this study, experiments were conducted to test the perfor- of a horizontal pipe of 3 mm in diameter and a small inlet tube
mance of a vortex drop structure with a small L / D ratio. Detailed 共1.5 mm in height and 1.2 mm in diameter兲 standing upright at
flow measurements were obtained on the variation of the water the head. When inserted slowly and horizontally into the water
depth and the wall pressure along the drop shaft. The energy through the shaft wall, the conduit guides water out until its inlet
dissipation and air entrainment were measured and the results reaches the water surface. It was tested in a free circular jet and
were compared with earlier studies. The theoretical vortex-flow was found sensitive in locating the air–water interface. Up to
model of Jain 共1987兲 was also extended and the predictions were about 2 mm of measurement error in the jet thickness is expected
compared with the measurements. given that the diameter of the probe head is 1.2 mm, the mini-
mum reading scale of the probe is 1 mm, and some uncertainty in
detecting the air–water interface. To reduce the measurement
Experiments error, five readings were taken and the data were averaged. The
flow thickness was measured at seven sections.
A model vortex drop structure was built at the China Institute of At Q = 0.051 m3 / s, water depth in the approach channel 共sec-
Water Resources and Hydropower Research 共IWHR兲, as shown in tion 1兲 was h1 = 0.130 m. The approaching flow of velocity
Fig. 1. It consisted of a rectangular approach channel, a volute V1 = 3.92 m / s was supercritical with a Froude number of 3.5. A
chamber, a drop shaft, and a rectangular outlet tunnel with an hydraulic jump occurred in front of the entrance of the volute
arched ceiling. The outer guide wall of the approach channel was chamber. The deflector deviated the spiraling jet in the volute
connected to the volute chamber with a 1 / 4 ellipse curve. The chamber from impinging on the incoming flow to avoid choking
channel was nearly horizontal and the vortex inlet may be classi- the entrance. In the volute chamber and the drop shaft, a spiral
fied as a tangential slot type. The model was made of Plexiglas flow with a stable air core was formed. An annular hydraulic
for visual observation. jump occurred at an elevation of about 0.50 m 共Fig. 1兲. Below the
The diameter of the vertical drop shaft was selected according jump a water cushion was formed and the lower end of the shaft
to Jain 共1984兲 as was submerged. This flow pattern was defined as Regime II by
Jain and Kennedy 共1983兲. As flow turbulence in the pool can
D = ␴共Q2/g兲1/5 共1兲
dissipate a large amount of energy, it is therefore an advantage in
where ␴ = safety factor; Q = discharge; and g = gravitational accel- this study.
eration. Jain 共1984兲 used ␴ = 1 for a tangential inlet, but Hager The measured instantaneous pressure showed its root mean

62 / JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Fig. 2. Measurements of annular jet flow: 共a兲 average relative water thickness; 共b兲 wall pressure for each direction; and 共c兲 fluctuation coefficient
of wall pressure 共C p兲

square 共RMS兲 of up to 10% of the mean pressure. The measured all tests, the outlet tunnel was pressurized, water velocity was
jet flow thickness also fluctuated within about ±8% of its mean small 共⬍0.6 m / s兲, and no air bubbles were observed in the tail
value. Both measurements showed significant variations over a tank. Accordingly, it was assumed that the entrained air was com-
cross section: 20% for wall pressure and 15% for water thickness. pletely discharged through the vent pipes. Centerline velocities of
Identical patterns of the wall pressure and the thickness were airflow in the vent pipes were measured with a QDF-3 hot-wire
observed in the experiments with Q = 0.047 and 0.049 m3 / s. Thus anemometer 共Yuanda Company, China兲. Adopting the one-
the actual flow was not axisymmetric 共or annularly uniform兲 but seventh power law velocity profile, air discharges were calculated
spiraled downward. In the following, the flow will be assumed by Qa = ⌺共0.8VcAa兲, where Vc is the measured centerline velocity
axisymmetric and the measurement will be averaged over each and Aa is the area of the vent pipe. A possible error of 15% in air
cross section so as to apply a one-dimensional 共1D兲 analysis. Note discharge was estimated, given the uncertainty in both the veloc-
that the cross-sectional averaged values can have a combined ity measurement and the assumption for the velocity profile.
error of about 20% in both wall pressure and water thickness as a
result of measurement errors, data fluctuations and the variations
over a cross section. Performance of Drop Structure
The time-averaged wall pressure P and the relative water
thickness t = b / R are shown in Fig. 2, where b⫽water thickness
Energy Dissipation
and R = D / 2. Water thickness decreases quickly beyond the intake
and then gradually towards a constant. The wall pressure is Four sections were identified to examine the energy losses in the
mainly caused by centrifugal force. In the volute chamber the drop structure. The approach section 1 has an elevation of
direct impact of incoming flow on the wall also increases the Z1 = 3.150 m 共Fig. 1兲. The elevation of section 2 below the intake
pressure. The wall pressure decreases fast down the drop shaft. It, to the drop shaft is Z2 = 2.385 m. Section 3 is located at
however, remains positive along the drop shaft and thus excludes Z3 = 0.785 m, just above the annular hydraulic jump. Section 4 is
the possibility of cavitation. Below the annular hydraulic jump, at the outlet tunnel, where Z4 = 0.
the wall pressure gradually becomes hydrostatic. Between sections 2 and 3, the flow can be assumed as gradu-
The level of the wall pressure fluctuation can be represented ally varied annular jet flow for which a free vortex is typical
by a fluctuation coefficient defined as C p = Prms / P, where P and 共Quick 1961; Jain 1987兲. Accordingly, ␯tr = ⍀, where ␯t is tangen-
Prms⫽time-averaged and the RMS of the wall pressure at each tial velocity; r is the radial coordinate, and ⍀ is the circulation.
section, respectively. As shown in Fig. 2共c兲, C p increased from Assuming that the pressure in the vortex flow is caused by cen-
0.02 to 0.1 when the water dropped down in the shaft. Below the trifugal force and that atmospheric pressure prevails in the air
annular jump, C p reached 0.27. The fluctuation was small com- core, the pressure distribution over each section is then
pared with C p = 0.4 in a plunge pool with a water depth of 8–10
times the circular jet diameter, as reported by Ervine et al. 共1997兲.
The fluctuation was also found to be comparable with the mea-
surement by Toso and Bowers 共1988兲 in a hydraulic jump. In a
p共r兲 = 冕
r

R−b
1
␳␯2t /r · dr = ␳⍀2 2
2 R 共1
1
冋
− t兲
1
2 − 2
r
册 共2兲

power spectrum analysis of the instantaneous wall pressure, low where ␳ = density of water and t = relative flow thickness, t = b / R.
dominant frequency in the range of 0 – 2 Hz was observed. Simi- Following the definition of specific energy head E in open chan-
lar phenomena were observed by Gardiner and Hay 共1982兲 in a nel flows and based on Eq. 共2兲
stilling basin and Liu et al. 共2004兲 in hydraulic jumps.
In studying air entrainment, the water discharge was varied ␯z2 ␯2t p共r兲 ␯z2 2⍀2
from 0.026 to 0.051 m3 / s and the tunnel outlet was submerged in E= + + = + 共3兲
2g 2g ␳g 2g gD2共1 − t兲2
a tail tank with controlled water levels. The entrance of the tunnel
was also submerged in the water cushion region of Regime II. For where ␯z = vertical velocity. Thus the total head of the flow is

JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006 / 63

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
 H=E+Z 共4兲

Assuming a constant vertical velocity over a cross section and ␯z

in Eq. 共3兲 can be replaced by the average vertical velocity

Q 4Q
Vz = = 共5兲
A ␲D2t共2 − t兲

where A = cross-sectional flow area. With measured wall pressure

head and flow thickness, one can solve Eqs. 共2兲–共5兲 for the energy
heads H at sections 2 and 3. The flow at sections 1 and 4 was Fig. 3. Variation of relative air discharge 共␤兲 with Q*: present study
open channel flow and H = h + V2 / 共2g兲 + Z, where h is water depth. 共䊏兲; Jain and Kennedy 共1983兲: L / D = 25 共䉱兲; L / D = 11 共䉭兲; Ogihara
Accordingly, energy losses in the drop structure can be evaluated and Kudou 共1997兲: L / D = 25 共䊊兲; L / D = 37.5 共쎲兲; Eq. 共6兲 共—兲; Eq.
for each segment. 共7兲 共----兲
The efficiency of energy dissipation in each part of the drop
structure can be quantified by ␩, and ␩ = 共1 − Hi+1 / Hi兲 ⫻ 100%,
where Hi is the total head at section i. With the measured water
depth at section 4 in the outlet tunnel h4 = 0.08 m, the efficiency air was discharged through the last vent pipe, while the remaining
of the energy dissipation in each part can be estimated as: half of the air was discharged by the rest three with a similar
␩12 = 24%; ␩23 = 34%; and ␩34 = 80%. The overall energy dissi- efficiency.
pation rate in the drop structure ␩14 = 90%. Note that Jain 共1988兲 correlated the air entrainment in a vortex drop
␩14 = 1 − 共1 − ␩12兲共1 − ␩23兲共1 − ␩34兲. shaft to the local circulation of the annular jet flow. However, the
Between sections 1 and 2, the flow pattern was complicated determination of the local circulation is difficult in practice. In
and the energy loss included that due to the hydraulic jump at the addition, the vertical velocity of the jet flow could also be impor-
inlet, the residual collision between the incoming flow and the tant as discussed below. Vischer and Hager 共1995兲 related the
spiraling water, and wall friction. A significant energy loss re- entrained air discharge in a vortex drop shaft 共Qa兲 to the water
sulted in this region even though the drop height between the two discharge 共Q兲 and the pipe-full discharge 共Q p兲 as
sections was relatively small 共0.765 m兲. It is interesting to note
that a free hydraulic jump in a rectangular channel with an incipi- ␤ = Qa/Q = 共Q p/Q兲1/2 − 1 共6兲
ent Froude number of 3.5 will only contribute about 30% of the
This relationship, however, is for Regime I flow where the annu-
loss in ␩12 while the rest of 70% is due to the other factors
lar flow discharges freely into the outlet tunnel and the lower end
mentioned.
of the drop shaft is open to the atmosphere. According to Vischer
From section 2 and 3, the drop height was 1.6 m 共7.2D兲 and
and Hager, the velocity head of the approach flow can be ne-
the energy dissipation efficiency was ␩23 = 34%. The wall friction
glected in deep drops and Q p is then a function of wall friction,
and the turbulence production in the annular jet flow were the
the diameter 共D兲, and the length 共L兲 of the drop shaft. Thus Eq.
major sources of the dissipation. As will be discussed in detail
共6兲 can be rewritten in the form of ␤ as a function of L / D and Q*,
below, a typical friction factor of 0.02 will result in a head loss of
where the nondimensional water discharge is Q* = nQ / 共␲D8/3兲.
only around 20%. Thus the vortex flow in this region has high
For relatively small water flow, Vischer and Hager 共1995兲 sug-
efficiency in energy dissipation.
gested that
The most significant energy loss occurred between sections 3
and 4. Thus the plunge and the mixing in the water cushion were
␤ = 0.01共L/D兲2/3共Q*−3/5 − 1兲 共7兲
very effective in energy dissipation. Quick 共1961兲 related energy
loss in an annular hydraulic jump to the upstream velocity. For a which results in a smaller ␤ than Eq. 共6兲 when Q is small. *

flow of 5 m / s as measured in the present experiment the energy Five sets of experimental data are plotted in Fig. 3, together
loss is expected to be about 65%. The energy loss can also be with the predictions of Eqs. 共6兲 and 共7兲 for corresponding L / D
estimated from the measurement. Assuming a uniform flow with values: 共1兲 the present study with D = 0.223 m and L / D = 14; 共2兲
hydrostatic pressure distribution under the annular jump, the flow Jain and Kennedy 共1983兲 with D = 0.127 m and L / D = 25; 共3兲 Jain
velocity was estimated to be 1.34 m / s, such that the efficiency of and Kennedy 共1983兲 with D = 0.292 m and L / D = 11; 共4兲 and 共5兲
the annular hydraulic jump reached as high as about 70%, coin- Ogihara and Kudou 共1997兲 with L / D = 25 and L / D = 37.5 and
cident with Quick’s prediction. Thus, in the water cushion region, D = 0.08 m. Here n = 0.009 was used for Plexiglas. Fig. 3 shows
the annular hydraulic jump played an important role in energy that for Regime I flows, Vischer and Hager’s prediction 关Eqs. 共6兲
dissipation. and 共7兲兴 compare well with the experimental results of Ogihara
and Kudou for Q* ⬎ 0.01. On the other hand, the relative air dis-
charge for Regime II flows appears to be significantly smaller
Air Entrainment
than that for Regime I flows. The present study has ␤ ranging
Air entrainment was measured through five vent pipes in the from 0.47 to 0.68 for Q* of 0.004–0.008 in Regime II flow, which
present study. These vent pipes were mounted along the 5.5 m is fairly comparable with the measurement of Jain and Kennedy
long outlet tunnel 共Fig. 1兲 at 1.25De, 3De, 7De, 11.25De, and 共1983兲 for L / D = 11 and D = 0.292 m. Notice that the uncertainty
15De downstream from the axis of the drop shaft. Here, in the measured ␤ is about 15% in the present experiment. The
De = 0.3 m is the equivalent diameter of the tunnel cross-sectional other data set of Jain and Kennedy 共1983兲 for L / D = 25 gives
area for our setup. It was observed that the third vent pipe at 7De quite small ␤, ranging from 0.1 and 0.4. Even smaller ␤ values
discharges about half of the entrained air. A negligible amount of 共less than 0.1兲 were reported by Jain 共1988兲. The smaller ␤ for

64 / JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Fig. 4. Relation between entrained air rate and jet velocity: present
measurement 共䊏兲; Jain and Kennedy 共1983兲: L / D = 25 共䉱兲;
L / D = 11 共䉭兲; Ogihara and Kudou 共1997兲: L / D = 25 共䊊兲; L / D = 37.5
共쎲兲; Rajaratnam et al. 共1997兲 共⫻兲; Eq. 共8兲 共—兲

Regime II flows is expected since a portion of entrained air es-

capes back to the air core due to the pressure rise in the water Fig. 5. Control volume in annular jet flow
cushion at the base of the drop shaft.
Fig. 3 also shows that ␤ decreases when Q* increases for rela-
Ervine 共1998兲 stated that Eq. 共8兲 is applicable for a jet velocity
tively larger flow rate. Results from Jain and Kennedy 共1983兲 and
from 1.5 to 15 m / s, and has an uncertainty of around 30% in air
Jain 共1988兲, however, showed that ␤ initially increased with Q*
flow rate. Apparently Eq. 共8兲 involves some scale effect of air
before the decreasing trend appears. Vischer and Hager 共1995兲’s
entrainment. It suggests that the scale factor for entrained air flow
prediction 关Eqs. 共6兲 and 共7兲兴 gives larger air entrainment for a
rate is smaller than Lr2.5 for the water flow rate in Froudian models
larger L / D. This is confirmed by experimental data in Fig. 3 and
共Lr is the length scale兲. The prototype jet velocity needs to be
the measurements of Jain 共1988兲 with the exception of Jain and
obtained in order to estimate air entrainment using Eq. 共8兲. Our
Kennedy’s data for L / D = 25.
model has a relatively short drop where the air entrainment is
For models with a limited drop height, Rajaratnam et al.
dominated by the falling jet impinging into a pool, thus our re-
共1997兲 suggested that air entrainment is predominated by the fall-
sults can be well predicted using Eq. 共8兲. For prototypes with a
ing jet impinging into the water cushion. Ervine 共1998兲 reviewed
large drop height, other air entrainment mechanisms, such as self
previous measurements on air entrainment in various structures
aeration and air dragged down by falling water, are expected to be
including plunge-flow drop shafts and inclined conduits, and re-
important. Scaling effects are also different for these mechanisms.
lated the air entrainment rate to the jet velocity 共V j兲 at the plunge
Discussions on these types of air entrainment can be found in
point as
Wood 共1991兲.
Qa = ␭W关0.00002共V j − 1兲3 + 0.0003共V j − 1兲2
+ 0.0074共V j − 1兲 − 0.0058兴 共8兲 Annular Jet Flow in Drop Shaft
where W = jet width; and ␭ = coefficient introduced to account for
the loss of air returned back to the upstream conduit. Eq. 共8兲 is Analysis
considered applicable for conduits at different slopes with a jet
In studying vortex drops, Jain 共1987兲 and Vischer and Hager
thickness larger than 20– 30 mm and a jet velocity of up to
共1995兲 proposed that the energy in the vortex flow is dissipated by
15 m / s. Note that V j in Eq. 共8兲 needs to be larger than 1 m / s to
wall friction. Jain 共1987兲 presented a frictional model based on
initiate air entrainment.
conservation equations and the assumptions that: 共1兲 the flow is
In Fig. 4, Eq. 共8兲 is compared with the five data sets from Fig.
axisymmetric; 共2兲 the vertical velocity component is a constant
3 together with the measurements of Rajaratnam et al. 共1997兲 in a
over a cross section; 共3兲 the radial velocity is negligible; and 共4兲
plunge-flow drop shaft. For the vortex drop shaft W = ␲D. Based
the tangential velocity follows the free vortex distribution. The
on Ervine’s discussion 共1998兲, ␭ = 0.85 was adopted for the Re-
mass conservation was given as Eq. 共5兲, while the momentum
gime II flows studied here to account for air escaping back to the
equations can be deduced based on the control volume in Fig. 5:
shaft. For Regime I flow of Ogihara and Kudou 共1997兲 ␭ = 1 was
vertical momentum
assumed. Here the terminal velocity for a vortex drop shaft 共V⬁
discussed in the next section兲 was used as the jet velocity. For d共␳QVz + F p兲/dz = ␳gA − ␲D␶z 共9兲
Rajaratnam et al.’s 共1997兲 plunge flow, W = 0.75␲D and ␭ = 1
were assumed according to their observation of W in the range of angular momentum
共0.5– 1兲␲D. Fig. 4 shows a good agreement between Eq. 共8兲 and d共␳Q⍀兲/dz = − 2 ␲D2␶t
1
共10兲
the four data sets. Thus the air entrainment is primarily controlled
by the jet velocity, irrespective of whether the drop shaft is of where Vz = V sin ␪ with V being the mean flow velocity and
vortex or plunge-flow type. This finding is interesting, as it is ␪ = angle of the mean flow velocity from the horizontal;
typically believed that the vortex drop shaft entrains less air. The ␶t = ␶ cos ␪ and ␶␯ = ␶ sin ␪, ␶ = wall shear stress; F p = vertical pres-
reason for the discrepancy between the data of Jain and Kennedy sure force acting on the control volume; and z = vertical coordi-
共1983兲 and Eq. 共8兲 is not clear. nate with a downward direction. The wall shear stress is given by

JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006 / 65

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
 This leads to an even larger T. It is not clear why in this case the
thickness t will continue increasing. One possible reason is that
the assumption of free vortex flow is no longer valid in this case.
For known initial conditions, Vz and t at a downstream section
can be found from Eqs. 共5兲 and 共14兲. ⍀ and Vt can then be solved
from Eqs. 共11兲 and 共12兲. These equations were solved using the
fourth order Runge–Kutta method and the results were compared
with the experimental measurements. Note that the terminal ve-
locity V⬁ = 共8Qg / f␲D兲1/3 results from Eq. 共14兲 by letting dVz / dz
approach 0 and tan ␪ approach ⬁.

Comparison with Measurements

To facilitate the comparison with the theoretical predictions, the
Fig. 6. Variation of M and T with water thickness and mean flow
origin of the vertical axis z is now set to coincide with the start of
direction ␪
the drop shaft at elevation 2.700 m, and its direction is measured
positively downward. The initial flow thickness and wall pressure
␶ = 1 / 8 f␳V2, where f = friction factor. Using Vt = V cos ␪, Eq. 共10兲 were obtained from the experiments. Unfortunately, it is difficult
may be expressed as to determine these values with an acceptable accuracy due to the
complicated flow pattern right at the beginning of the vertical
d⍀ f␲D2 shaft. The predictions were therefore started from a point slightly
=− VVt 共11兲
dz 16Q below at z = 0.315 m 共i.e., section 2兲, where the flow is gradually
varied.
The averaged tangential velocity is
The friction factor, f, was estimated from the Moody’s dia-

冕
R
1 4⍀ gram to be 0.013–0.015 for a smooth pipe with a Reynolds num-
Vt = ␯t␲r dr = 共12兲 ber R = 4RhV / ␯ = 3 ⫻ 105, where Rh = t共2 − t兲D / 4 is hydraulic ra-
A R−b 共2 − t兲D
dius; and ␯ is the kinematic viscosity of water. Jain and Kennedy
By neglecting the pressure force and assuming a small t, Jain 共1983兲 reported that f in an annular jet flow would be signifi-
共1987兲 obtained an analytical solution. However, the pressure cantly higher than that from Moody’s diagram due to the curva-
force could be important when the water thickness varies quickly ture effect of the drop shaft, and f = 0.02 was used by Jain 共1987兲.
such as at the beginning of the drop shaft 共Fig. 2兲. Besides, for an The friction factor was also estimated from Manning’s n, i.e.,
economic design of vortex drop shafts, the area of air core is f = 8g / C2 = 8gn2 / R1/3
h , where C is Chézy coefficient. Here f will
recommended to be 25% of the drop shaft area at the throat sec- vary with Rh: f increases from 0.021 to 0.027 along the drop
tion 共Jain and Ettema 1987兲. This results in a relatively large t shaft. These numbers are consistent with Jain’s calculations of
value of close to 0.5. Here effects of t and pressure force are 1.6–1.7 times of Moody’s f. In this study, friction factors of 0.02
studied. and 0.03 will be used.
From Eq. 共2兲 the pressure force at any section The predictions are compared with the measurements in Fig. 7.

冕
R The comparison is reasonably good for water thickness but poor
Fp =
1
p共r兲 · 2␲rdr = 2 ␲␳⍀2关1/共1 − t兲2 + 2 ln共1 − t兲 − 1兴 for wall pressure. The increase of the friction factor f from 0.02 to
R−b 0.03 increases the water thickness because the vertical velocity Vz
is reduced and decreases the wall pressure as the tangential ve-
共13兲
locity Vt is also reduced. However, the effect of the friction factor
By using ␶ = 1 / 8 f␳V , substituting Eqs. 共5兲 and 共13兲 into Eq. 共9兲
2 on the wall pressure appears to be small, yet the wall pressure is
results in significantly under-predicted. Compared to a friction factor of

冋 册冒
0.02, f = 0.03 gives a better prediction of Vz but a worse prediction
dVz g ␲DVVz
= −f 共1 − M兲 共1 − T兲 共14兲 of Vt.
dz Vz 8Q The poor prediction of the wall pressure is attributed to vari-
in which the two terms, M and T, are functions of ␪ and t ous factors. However, the uncertainty in the determination of the

冋 册
initial wall pressure cannot explain all the difference. Even using
1 1 1 the upper limitation of the measured P at the initial section, the
M= · + 2 ln共1 − t兲 − 1
tan ␪ 2t 共1 − t兲2
2 prediction is still significantly smaller than the measurements, as
共15兲 the effect of the larger initial P value dies off very quickly 关see
1 t共2 − t兲3 dotted line in Fig. 7共b兲兴. In addition, given the large size of the air
T= · core 关see Fig. 7共a兲兴, the effect of the air movement on the pressure
tan ␪ 8共1 − t兲4
2
head is expected to be on the order of 1 mm based on a mean air
Note that neglecting both M and T will reproduce Jain’s solutions. velocity obtained from the air flow rate. It is possible that the
The variations of M and T as functions of ␪ and t are shown in free-vortex frictional model is unable to predict the wall pressure
Fig. 6. It is clear that M and T can be close to or even larger than in a vortex drop shaft since the assumptions of free vortex and
1 when t is large and ␪ small, i.e., at the upper portion of a drop circularly uniform flow are no longer valid for the spiral flow.
shaft, and thus cannot be neglected. If T ⬍ 1 initially, Vz will Variations of total energy head, energy gradient and energy
increase, and consequently t will decrease while ␪ will increase, dissipation rate along the drop shaft are shown in Fig. 8. Here the
which results in smaller T as shown in Fig. 6. However, if T ⬎ 1 total energy head is H = E − z as z axis is downward positive.
initially, Vz will decrease and t will increase while ␪ will decrease. The rate of local energy dissipation was defined by

66 / JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Fig. 7. Comparison of predictions with measurements of annular jet flow: 共a兲 flow thickness; 共b兲 wall pressure; and 共c兲 tangential velocity Vt and
vertical velocity Vz. Measurements 共䊏兲; predictions with pressure force and f = 0.02 共—兲; without pressure force and f = 0.02 共—兲; without
pressure force and f = 0.03 共----兲; without pressure force and f = 0.03 with large initial P as initial condition 共…兲.

␩ = 共E2 + ⌬z − E兲 / 共E2 + ⌬z兲 ⫻ 100%, in which E2 is the specific en- cally decreasing along the whole reach of the prediction 关Fig.
ergy at section 2 and ⌬z is the vertical distance from section 2 to 8共b兲兴. This is more reasonable as the pressure head is part of the
the section of interest. It is clear from Fig. 8共a兲 that a larger flow energy.
friction factor of f = 0.03 gives a good prediction of the energy
loss. This implies that, when attributing the energy loss to wall
friction, the frictional model misses the part of the loss due to the
Summary and Conclusions
vortex motion, which can be compensated by using a large fric-
tion factor.
The energy gradient, dH / dz tends to −1 when z / D is large An experimental study on the performance of a vortex drop shaft
关Fig. 8共b兲兴. It indicates that the potential energy 共or elevation was conducted using a drop structure of 3.15 m high and 0.223 m
head兲 is mostly consumed by friction at large z / D and t tends to in diameter. The drop shaft had a tangential slot inlet with a
constant while V tends to the terminal velocity. From Fig. 8共c兲, nearly horizontal approach channel. In all experiments, Regime II
the local energy dissipation rate increases quickly with the ratio flow was formed with an annular hydraulic jump at the base of
z / D. The rate of energy dissipation is about 65% at z / D = 20, and the drop shaft partly due to some site-specific features of down-
is about 40% at about 10D. Therefore the energy dissipation in stream conditions. Detailed measurements of wall pressure and
vortex drop shafts is significant. flow thickness were made along the depth. The wall pressure
The effects of the pressure force in the predictions are shown remained positive along the drop shaft, which is favorable for
in Figs. 7 and 8. With the inclusion of T and M 共at the starting preventing the occurrence of cavitation. An overall energy dissi-
section T = 0.49 and M = 0.36兲, the vertical velocity Vz increases pation of up to 90% was measured in the vortex drop. This con-
faster 关Eq. 共14兲兴, thus both t and P decrease faster. However, firmed the performance of vortex drops as energy dissipators. In
these effects become negligibly small after z / D ⬎ 5. Similarly the the annular flow, the efficiency of wall friction and vortex motion
pressure term generate some difference in Fig. 8 at the starting in energy dissipation was about 34% over a drop height of
part but tend to be negligible at downstream part. With the pres- 1.60 m. The annular hydraulic jump at the bottom of the drop
sure force included, the energy gradient dH / dz was monotoni- shaft was found to be very efficient with a high efficiency of

Fig. 8. Predictions of 共a兲 energy head, 共b兲 energy gradient with depth, and 共c兲 rate of energy dissipation with elevations. Measurements 共䊏兲;
predictions with pressure force and f = 0.02 共—兲; without pressure force and f = 0.02 共—兲; without pressure force and f = 0.03 共----兲.

JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006 / 67

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
about 70%, consistent with the prediction of Quick 共1961兲. There- Gardiner, S. R. M., and Hay, D. 共1982兲. “Dynamic force measurement on
fore, the water cushion of the drop structure plays a very impor- stilling basin floor and sidewalls.” Proc., Int. Conf. on Hydraulic
tant role in energy dissipation. It should be mentioned, however, Modelling of Civil Engineering Structures, BHRA Fluid Engineering,
that care should be taken as a hydraulic jump could cause flow 123–130.
instability problem. Hager, W. H. 共1990兲. “Vortex drop inlet for supercritical approaching
The 1D free-vortex frictional model of Jain 共1987兲 was evalu- flow.” J. Hydraul. Eng., 116共8兲, 1048–1054.
ated for flow in the drop shaft. A large friction factor of f = 0.03 is Jain, S. C. 共1984兲. “Tangential vortex-inlet.” J. Hydraul. Eng., 110共12兲,
needed to properly simulate the energy loss. This is larger than 1693–1699.
Jain, S. C. 共1987兲. “Free-surface swirling flows in vertical dropshaft.” J.
the value of f = 0.014 estimated from Moody diagram but close to
Hydraul. Eng., 113共10兲, 1277–1289.
f = 0.024 from Manning’s n. While the water thickness along the
Jain, S. C. 共1988兲. “Air transport in vortex-flow drop-shafts.” J. Hydraul.
depth was reasonably well predicted, the wall pressure was sig- Eng., 114共12兲, 1485–1497.
nificantly underpredicted. Jain’s model was extended to incorpo- Jain, S. C., and Ettema, R. 共1987兲. “Vortex-flow intakes.” IAHR Hydrau-
rate the pressure force. However, its effect was overall small on lic Structures Design Manual, Vol. 1, A. A. Balkema, Rotterdam, The
the wall pressure and the flow thickness, but significant on under- Netherlands.
standing energy dissipation. Jain, S. C., and Kennedy, J. F. 共1983兲. “Vortex-flow dropstructures for the
The rate of air entrainment was measured in the outlet tunnel Milwaukee Metropolitan Sewerage District inline storage system.”
for a range of water discharges. A variation of 0.47–0.68 in the IIHR Rep. No. 264, Univ. of Iowa, Iowa City, Iowa.
relative air discharge 共␤兲 was observed and it decreased with Jain, S. C., and Kennedy, J. F. 共1984兲. “Vortex-flow drop.” Proc., 1984
increasing water discharge. The air entrainment was found to be Int. Symp. on Urban Hydrology, Hydraulics and Sediment Control,
primarily controlled by the jet velocity at the plunge point. The Univ. of Kentucky, Lexington, Ky., 115–120.
measured air entrainment in vortex drop shafts was also found to Jeanpierre, D., and Lachal, A. 共1966兲. “Dissipation d’énergie dans un
be comparable to that in plunge-flow drop shaft. This could be puits a vortex.” Houille Blanche, 21共7兲, 825–831.
important and further study is needed as it is typically believed Liu, M., Rajaratnam, N., and Zhu, D. Z. 共2004兲. “Turbulence structure of
that a vortex drop shaft entrains less air than a plunge-flow drop hydraulic jumps of low Froude numbers.” J. Hydraul. Eng., 130共6兲,
shaft. The above result is obtained from small scale models with 511–520.
relatively small drops where the air entrainment is dominated by Ogihara, K., and Kudou, T. 共1997兲. “Theoretical analysis of air-entrained
flow in vertical drop shafts of the channel in urban drainage system.”
jet impinging into a water pool. It is also noted that the rate of air
Proc., 27th Congress of IAHR, Water Resources Engineering Division,
entrainment does not follow the Froude similitude.
Theme A, 69–74.
Quick, M. C. 共1961兲. “The annular hydraulic jump.” Civ. Eng. Public
Works Rev., 56共662兲, 1176–1179.
Acknowledgment Quick, M. C. 共1990兲. “Analysis of spiral vortex and vertical slot vortex
drop shafts.” J. Hydraul. Eng., 116共3兲, 309–325.
The writers would like to thank Professor N. Rajaratnam for his Rajaratnam, N., Mainali, A., and Hsung, C. Y. 共1997兲. “Observations on
helpful advice during the preparation and revision of this paper. flow in vertical dropshafts in urban drainage systems.” J. Environ.
Eng., 123共5兲, 486–491.
Toso, J. W., and Bowers, C. E. 共1988兲. “Extreme pressures in hydraulic-
References jump stilling basins.” J. Hydraul. Eng., 114共8兲, 829–843.
Vischer, D. L., and Hager, W. H. 共1995兲. “Vortex drops.” Energy dissi-
Dong, X., and Gao, J. 共1995兲. “Report on model study of retrofitting a pators: Hydraulic structures design manual, No. 9, Chap. 9, A. A.
diversion tunnel into a vortex dropshaft spillway in Shapai PowerSta- Balkema, Rotterdam, The Netherlands, 167–181.
tion.” IWHR Research Rep., China Institute of Water Resources and Wood, I. R. 共1991兲. “Air entertainment in free-surface flows.” IAHR hy-
Hydropower Research, China 共in Chinese兲. draulic structures design manual, No. 4, Hydraulic design consider-
Ervine, D. A. 共1998兲. “Air entrainment in hydraulic structures: A review.” ations, Balkema, Rotterdam, The Netherlands.
Proc. Inst. Civ. Eng., Waters. Maritime Energ., 130, 142–153. Zhao, C. H., Sun, S. K., and Liu, Z. P. 共2001兲. “Optimal study on the
Ervine, D. A., Falvey, H. T., and Withers, W. 共1997兲. “Pressure fluctua- depth of stilling well for rotation-flow shaft flood-releasing tunnel.”
tions on plunge floors.” J. Hydraul. Res., 35共2兲, 257–279. Water Power, 2001共5兲, 30–33 共in Chinese兲.

68 / JOURNAL OF HYDRAULIC ENGINEERING © ASCE / JANUARY 2006

Downloaded 04 May 2011 to 95.131.110.102. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org