You are on page 1of 9

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

net/publication/279251379

Vortex-Induced Air Entrainment Rates at Intakes

Article  in  Journal of Hydraulic Engineering · June 2015


DOI: 10.1061/(ASCE)HY.1943-7900.0001036

CITATIONS READS
20 696

3 authors:

Georg Moeller Martin Detert


IUB Engineering AG ETH Zurich
2 PUBLICATIONS   26 CITATIONS    23 PUBLICATIONS   272 CITATIONS   

SEE PROFILE SEE PROFILE

Robert Michael Boes


ETH Zurich
153 PUBLICATIONS   985 CITATIONS   

SEE PROFILE

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

2nd International Workshop on Sediment Bypass Tunnels View project

Probabilistic Dam Breach Modeling View project

All content following this page was uploaded by Martin Detert on 09 February 2016.

The user has requested enhancement of the downloaded file.


Vortex-Induced Air Entrainment Rates at Intakes
Georg Möller 1; Martin Detert 2; and Robert M. Boes 3

Abstract: This paper presents an intrinsic approach to estimate air entrainment rates due to intake vortices based on large-scale laboratory
measurements. Quasi-continuous measurements of the amount of entrained air were conducted using a sophisticated de-aeration system.
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

The data analysis of 34 experimental runs resulted in a parameter fit describing the air entrainment rates at horizontal intakes. Additionally, a
prediction band is given that provides a handle to inspect probability aspects. Furthermore, an approach to determine the critical
intake submergence is presented. These guidelines enable hydraulic design engineers to estimate both the amount of entrained air and
the critical intake submergence at horizontal intakes, so that the efficiency and safety of pressurized waterway systems is further increased.
DOI: 10.1061/(ASCE)HY.1943-7900.0001036. © 2015 American Society of Civil Engineers.
Author keywords: Air entrainment rate; De-aeration; Hydraulic structure; Intake vortices; Intake submergence; Zero-air-criterion.

Introduction effects caused by unsteady flow behavior of entrapped air. Studies


by Zhou et al. (2002, 2013a, b) indicate that entrapped air pockets
Air entraining vortices are a typical phenomenon at intakes of hy- are able to induce an abnormal pressure surge, which can easily
droelectric power plants (HPP) and other pressurized systems. The burst the pipe. The third group refers to negative, quasi-stationary
entrained air changes the flow properties from a single-phase to a effects. An accumulation of air reduces the effective cross-sectional
two-phase flow, characterized by the relative air entrainment rate area and, consequently, the flow discharge decreases. Furthermore,
local corrosion damages of steel linings may occur.
Qa
β¼ ð1Þ A literature review concerning intake vortices indicates that only
Qw a few publications deal with air entrainment as to its quantification,
see Table 1. The first approaches to quantify the relative air entrain-
giving the relationship between the volumetric air discharge Qa and ment rate β were presented by Iversen (1953) and Denny and
the volumetric water discharge Qw . Young (1957). According to their research, the air entrainment rate
Air in pressurized systems may have strong consequences on at pump sumps is typically β ¼ 5%, but may reach β ¼ 10% in
the operation and safety of pressure systems and hydraulic machi-
extreme cases. Investigations of Hattersley (1965) indicate air en-
nery. These effects include (1) efficiency reduction of turbines and
trainment rates between 0.06 ≤ β ≤ 0.73%, thus one to two mag-
pumps, (2) unsteady flow behavior, i.e., pulsations and pressure
nitudes smaller than the former value. However, up to the 1970s no
surges in the system with corresponding mechanical loads on
valuable information appears to be provided to quantify β as a func-
the involved components, and (3) stationary effects, e.g., flow re-
tion of common flow features such as the intake (pipe) or combined
duction from the presence of air bubbles and local corrosion dam-
(submergence) Froude numbers, i.e., FD ¼ vD =ðgDÞ0.5 and
age, respectively. The first group refers to direct effects of air in
Fco ¼ vD =ðghÞ0.5 , respectively, where vD = average velocity at
hydraulic machinery. Even minute β values may cause a reduction
the characteristic intake cross section, D = inner intake diameter,
in efficiency of around 1%. An air entrainment rate of β ¼ 1.5%
g = gravity acceleration, and h = intake submergence relative to the
leads to an efficiency reduction of up to 16% (Denny and Young
pipe axis. Padmanabhan (1984) presents maximum values of air
1957; Papillon et al. 2000). For air entrainment rates up to β ¼ 4%
concentration rates (void fraction) Ca ¼ β=ðβ þ 1Þ versus Fco
there is a further exponential decrease in efficiency. Depending on
(Fig. 1). With Ca;max ¼ 15%, the maximum relative air entrainment
the type of pump (axial pumps are more sensitive than centrifugal),
rate amounts to β ¼ Ca =ð1 − Ca Þ ¼ 18%. The majority of all
at air entrainment rates between 7 ≤ β ≤ 20%, a sudden drop of
efficiency down to a total flow interruption may occur (Chang data are below Ca ¼ 1%. The data scatter is considerable, and
1977; Poullikkas 2000). The second group includes the destructive no specific trend becomes obvious. No correlation of air entrain-
ment and flow parameters was performed, but the author provided
1
Project Manager, IUB Engineering AG, Technoparkstrasse 1, CH-
an envelope line of maximum air entrainment. Although the studies
8005 Zürich, Switzerland; formerly, Laboratory of Hydraulics, Hydrology of Denny and Young (1957) and Padmanabhan (1984) are compa-
and Glaciology (VAW), ETH Zurich, CH-8093 Zürich, Switzerland rable, the measured entrainment rates again vary widely over sev-
(corresponding author). E-mail: georg.moeller@iub-ag.ch eral orders of magnitude.
2
Research Engineer, Laboratory of Hydraulics, Hydrology and Furthermore, a large number of publications using the term air
Glaciology (VAW), ETH Zurich, CH-8093 Zürich, Switzerland. E-mail: entraining vortices provide no detailed information on entrainment
detert@vaw.baug.ethz.ch rates or void fractions (e.g., Haindl 1959; Jain et al. 1978; Chang
3
Professor and Director, Laboratory of Hydraulics, Hydrology and and Lee 1995), see also Möller et al. (2012). Recently, Suerich-
Glaciology (VAW), ETH Zurich, CH-8093 Zürich, Switzerland. E-mail:
Gulick et al. (2014a, c) presented a model to estimate the key char-
boes@vaw.baug.ethz.ch
Note. This manuscript was submitted on March 23, 2014; approved on
acteristics of free surface vortices at intakes. However, a linkage to
March 12, 2015; published online on June 8, 2015. Discussion period open the amount of entrained air is missing here as well.
until November 8, 2015; separate discussions must be submitted for indi- To close the knowledge gap concerning vortex-induced air en-
vidual papers. This paper is part of the Journal of Hydraulic Engineering, trainment rates at intakes, the goal of the present research is to
© ASCE, ISSN 0733-9429/04015026(8)/$25.00. quantify β in terms of the hydraulic characteristics of the local flow.

© ASCE 04015026-1 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


Table 1. Summary of Investigations Concerning Air Entrainment Rate
Graphical data
Author(s) Air entrainment rate β presentation Regression Experimental setup
Iversen (1953) Normal 1–5%, extreme > 10% h=D No Intake orientated vertically upwards; scale model, prototype
Denny and Young (1957) Normal 1–5%, extreme > 10% n/a n/a Intake orientated vertically upwards and horizontal; scale model
Hattersley (1965) 0.06–0.73% n/a n/a Intake orientated vertically upwards; prototype
Padmanabhan (1984) ≤18% Fco [0.0–2.2] No Two intakes orientated horizontally; scale model
≤23% Fco [0.0–2.0] No Intake orientated vertically downwards; scale model

Experimental Setup
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

Hydraulic Model
Experiments were conducted at a large-scale experimental tank-
pipe system at VAW, ETH Zurich (Fig. 2), consisting of three main
parts: (1) 50 m3 experimental steel tank, (2) 14 m long pressurized-
pipe system, and (3) 130 m3 underground reservoir. Water dis-
charge was supplied by two frequency-controlled pumps providing
a maximum discharge of Qw ¼ 0.510 m3 =s in a closed loop.
The stilling basin was designed to promote a uniform velocity
distribution in the main tank. In the stilling basin the end fittings
of the feeding pipes were perforated to diffuse the incoming jet
homogeneously.
The flow was guided through a vertically arranged filter fleece
to the main tank. An additional wave absorber raft (not drawn in
Fig. 2) damped surface waves. In a top view the tank was symmet-
rically arranged to the main flow direction containing no other in-
Fig. 1. Air concentration Ca ¼ β=ðβ þ 1Þ (‘void fraction’) versus
combined Froude number Fco ¼ vD =ðghÞ0.5 at horizontal intakes, with
stallations. The unsteady nature of intake vortex generation was
envelope line giving maximum air entrainment rate Ca ðFco Þ; perturbed judged to be undisturbed by the boundary conditions of the tank.
flow tests were conducted while using partially blocked approach flow The water flow was from the main tank into the pressurized-pipe
[reprinted from Padmanabhan (1984) with permission] system through a horizontal intake pipe, whose sharp-crested open-
ing had an inner diameter of D ¼ 0.389 m (DN400). Visual access

Fig. 2. Cross-sectional view of experimental facility: (a) 50 m3 experimental tank with stilling basin and filter fleece for flow uniformization and
intake pipe; (b) 14 m long pressurized-pipe system with de-aeration device consisting of three riser pipes and two air storages; (c) 130 m3 reservoir
and pumps

© ASCE 04015026-2 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


to the tank was secured by three glass windows at the bottom wall, respect to the data transferability to prototype scale. To check
the right wall, and the head wall. The Froude numbers FD and Fco the validity of our Froude model hypothesis mentioned above, scale
varied up to 2 and 2.8, respectively, whereas the relative submer- family tests were performed. The goal was also to define RD;cr
gence h=D varied up to 4. and WD;cr .
The water discharges were measured by magnetic inductive dis- This study’s scale family tests were performed by Meyer
charge (MID) meter, the water levels by ultrasonic sensors, the (unpublished data, 2012; see also Möller 2013) at the same
pressures in the rising pipe system by piezoresistive pressure sen- tank facility as used for the main experiments. Parameter were
sors sensors, the approach flow velocities by acoustic Doppler ve- varied in the range 0.2 ≤ FD ≤ 1.5, 0.75 ≤ h=D ≤ 3.0, and
locimetry (ADV), and the flow field around the intake vortices by for nominal diameters D ¼ ½DN200; DN300; DN400; DN500 ¼
large-scale particle image velocimetry (PIV) (Keller et al. 2014). ½190; 291; 389; 486 mm, where h=D ¼ 0 refers to a water level
A vortex type box with seven buttons, in relation to the vortex at the pipe axis. Measurements of β (see next section) were only
types VT ¼ 0 − 6 (see below), allowed for a continuous and performed at constant vortex phases. To compare the results of dif-
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

manual documentation of the visually identified vortex types ferent scales, a measured air entrainment factor
(Hecker 1987). Furthermore, both the fluid temperature and the
conductivity were measured. A complete list of the instrumentation β i ðjÞ
Rf ðjÞ ¼ ð3Þ
and measurement equipment used in the experimental investiga- β i ðDN500Þ
tions is presented by Möller (2013). The data acquisition of all
measurement subsystems was continuously performed with a was defined to analyze possible scale effects, where j denotes the
1 Hz frequency. four different pipe diameters D, and i stands for the flow conditions
at the same FD and h=D, respectively. Rf values are only deter-
mined for measuring points where air entrainment was measurable
Scale Similarity in all four scales. Fig. 3 summarizes the findings. The median of
A dimensional analysis conducted by Möller (2013) showed that, Rf ðDN200Þ is 0.10, i.e., β is about 10 times smaller than for experi-
besides the FD =Fco , the Reynolds number RD ¼ vD D=ν, with ν = ments with a DN500. The median of Rf ðDN300Þ is equal to 0.33,
kinematic viscosity, and the Weber number WD ¼ ρv2D D=σ, with i.e., scale effects are still present, but to a smaller extent. The
σ = surface tension between air and water, are decisive parameters median of Rf ðDN400Þ is 1.05 ≈ 1.0, thus the median of β-values
for scaling the laboratory model. These findings are in line with the is almost the same for DN400 and DN500. In other words, the air
literature, e.g., Tastan and Yildirim (2014) as well as Suerich- entrainment rates match well for all the experiments with DN400
Gulick et al. (2014b). and DN500, and, consequently, scale effects are seen to be insig-
The authors’ initial hypothesis was that inertial forces are dom- nificantly small for diameters ≥ DN400. However, the scatter of
inant for air-entraining vortex flow, so that Froude similarity was 20% for the 25th and 75th percentiles is quite large and has
chosen for the scale model. To keep the effects of viscosity and to be kept in mind for the discussion of the scatter of the measure-
surface tension small, the experimental conditions must be main- ment results.
tained above corresponding threshold (critical, subscript cr) values The authors conclude that for the current experimental setup
of RD;cr and WD;cr . Through the compliance of empirical limits, vortex-induced air entrainment based on a Froude model applies
their influence should be negligible to avoid scale effects. Classical to intake diameters of D ≥ 389 ≈ 400 mm with regard to scale
limits for similitude criteria of intake-vortex investigations are effects. A further data analysis showed, that viscosity and surface
RD;cr ¼ 3.2 × 104 according to Daggett and Keulegan (1974), and tension effects turned out to be negligible if RD ≥ 6 × 105
WD;cr ¼ 121 according to Ranga Raju and Garde (1987). For air- and WD > 3.2 × 103 , respectively, what is in reasonable agree-
water flow Pfister and Chanson (2014) proposed RD;cr ¼ 2 − 3 × ment to the Reynolds limit suggested by Pfister and Chanson
105 to avoid relevant scale effects in terms of air concentrations (2014).
within 5 ≤ F ¼ v=ðghÞ0.5 ≤ 15, where v and h are characteristic
air-water flow velocity, and flow depth, respectively. For F < 5,
these limits have to be selected more conservatively. If one limita-
tion is considered, the other is implicitly respected. The difficulty,
however, is to define a meaningful reference velocity and length for
the present investigation. The critical velocity vcr for air entrain-
ment to occur, also termed inception limit, is a function of relative
turbulence intensity. According to Kobus (1984), vcr depends on
the magnitude of the fluid parameter Z ¼ ðgη4w Þ=ðρw σ3 Þ, with
ρw as density, and ηw as dynamic viscosity of water. Z depends
on liquid properties only independent of the boundary scale and
the flow velocity. In experimental two-phase flow investigations
the liquid parameter applies alternatively to W, with the advantage
neither to contain a reference length nor a reference velocity. For
equal water properties in model and prototype, the air inception
limit is given with ν w ¼ ηw =ρw as the kinematic viscosity of water
by Kobus (1984)

v3cr
¼ const: ¼ ð0.5 ÷ 1Þ · 105 ð2Þ
gν w
Fig. 3. Boxplot of analyzed data Rf of the scale family tests with the
four tested pipe diameters; bold line = median, box = 25th and 75th
The order of magnitude indicates air inception at velocities
percentiles, whiskers = extreme values
of 0.8 − 1 m=s. Obviously, some scale uncertainties exist with

© ASCE 04015026-3 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Air discharge Qa at h=D ¼ 1.5, FD ¼ 0.8 (images by Dr. Georg Möller): (a) air-entraining vortex; (b) de-aeration at rising pipe #2

Air Entrainment Rate for 0 ≤ ti ≤ 2 h) with the median VT6 phase related to Qa;j marked
by a circle. Fig. 5(c) visualizes the progress of the running
Experimental Determination average Q̂a;i .

The main experiments were conducted at a constant conduit diam-


eter of D ¼ 389 mm within the parameter matrix 0.4 ≤ FD ≤ 1.2 Characteristic Results
and 1.25 ≤ h=D ≤ 2.5, corresponding to 0.36 ≤ Fco ≤ 0.80 for a A comprehensive and systematic analysis of all experimental runs
total of 34 experimental runs. By choosing this range, the experi- has been conducted (Möller 2013). Two average values were used
ments were covering the typical range of hydropower intake con- to characterize the measured air entrainment, namely, Q̄a giving
ditions in terms of submergence and intake velocity (Knauss 1987). a characteristic mean value of air entrainment over the full
However, at ½FD ; Fco  ¼ ½0.36; 0.50 the air entrainment was far be- v
measurement duration, whereas Qa denotes a characteristic maxi-
low the measurable limit of β ¼ 10−5 , although an almost stable
mum value as it is related to VT6 phases. Note that the latter does
vortex was visible during 2 h of measurement, so that finally
not equal the absolute maximum of an experimental run, but it is a
the results of 33 runs were analyzed concerning their air entrain-
two times averaged value, first within a single VT6 phase and sec-
ment rate. ond over all VT6 phases in a run. This value is related to the vortex
Fig. 4(a) shows a typical air-entraining intake-vortex at h=D ¼ phases in which high air discharge was measured and is therefore
1.5 and FD ¼ 0.8. The resulting bubbles in the de-aeration (rising) referred to as the characteristic maximum. Fig. 6 shows that β
pipe #2 (Fig. 2) are shown in Fig. 4(b). The water level in the rising strongly correlates with Fco . vA regression analysis with β̄ as the
pipes equals the piezometric head at atmospheric ambient pressure.
mean air entrainment rate, β as mean air entrainment rate of
If the pipes are closed on top, the pressure increases above atmos-
VT6 phases, and Fco ¼ vD =ðghÞ0.5 as combined Froude number
pheric pressure with each accumulated bubble rising in the pipe,
results in the exponential functions
resulting in a water level drop measured by an ultrasonic sensor
and a pressure sensor, respectively. At each further time step the β̄ ¼ 1.04 × 10−8 expð16Fco Þ and ð4Þ
mass of collected air is determined using the ideal gas law. Sub-
sequently, the difference of the actual pressure and the initial atmos-
pheric pressure allows for the computation of the entrained volume v
of air, and thus indirectly the identification of the air discharge Qa . β ¼ 2.00 × 10−8 expð16Fco Þ ð5Þ
The air entrainment data were tested for their accuracy within a
broad range of Qa ¼ 1.9 × 10−5 − 3.6 × 10−4 m3 =s at different The coefficient of determination R2 ≥ 0.9 is high for both
h=D and FD values using a forced aeration. Moreover, the system mean air entrainment rates. The boundaries are given by Fco ¼
was optimized by Bühlmann (unpublished data, 2012) toward a ½0.36 − 0.8, because the air-entrainment rate at Fco ¼ 0.36 was be-
low thev measurement limit of β ¼ 1 × 10−5 . As a rule of
data acquisition that allowed for quasi-continuous Qa recordings
with a frequency of up to 1 Hz, so that the measurement duration thumb β ≈ 2β̄.
was extended to ensure that the full Qa range was caught. There-
fore, two external air storages, each with a volume of 0.4 m3 , were Comparison with Padmanabhan’s Data
added to the rising pipes to enlarge the air volume (Fig. 2).
Fig. 5 shows a typical time series of measured Qa for a test with To the authors’ knowledge, Padmanabhan (1984) gives the most
h=D ¼ 1.5 and FD ¼ 0.8, highlighting the strong fluctuation of air comprehensive data set available so far concerning measured
entrainment. The plot contains three subplots. Fig. 5(a) illustrates air entrainment rates (Fig. 1). Although all recorded data were
the occurrence of an air-entraining vortex by bars, color-coded with plotted, no regression analysis was conducted. Fig. 7 shows the
white, corresponding to vortex types < VT6, gray denotes VT6, lower section of Padmanabhan’s plot, digitized to an accuracy of
v
and black short phases of VT6. Fig. 5(b) gives the actual Qa;i Ca ≈β ⪞ 10−3 due to the restrictions of the original data plot.

© ASCE 04015026-4 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


(a)

(b)
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

(c)

Fig. 5. (a) Time series of vortex types VT; (b) actual air discharge Qa;i and VT6 phase median Qa;j ; (c) running average Q̂a;i for experimental run with
h=D ¼ 1.5, FD ¼ 0.8

(a) (b)

Fig. 6. Air entrainment rates β versus Fco with 95% prediction band. Subdivision at Fco ¼ 0.66 accounts for varying prediction widths: (a) Mean air
v
entrainment rate β̄ðFco Þ; (b) Mean air entrainment rate of VT6 phases β ðFco Þ

(a) (b)
v
Fig. 7. Mean air entrainment rate according to single vortex phases β compared with Padmanabhan’s data (1984) digitized from Fig. 1 and recast to
β ¼ Ca =ð1 − Ca Þ: (a) linear plot; (b) semi-logarithmic plot to facilitate interpretation

© ASCE 04015026-5 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


v
Superimposed is the air entrainment rate β as determined within the vortex formation. Thus, up to now both hcr and the critical
this study as characteristic maximum value, as well as Eq. (5). situation concerning air entrainment were determined without
v any air entrainment measurements. However, the present quantita-
The measured values of β within the parameter range of Fco tive air entrainment measurements enable a completely new ap-
lie within the data cloud as determined by Padmanabhan. The proach, as hcr is determined in direct correlation with the air
inception point of air entrainment matches well at Fco ≈ 0.3 for entrainment rate β measured to a minimum air entrainment rate
Padmanabhan (1984) and was determined to 0.4 in the present of β ¼ 1 × 10−5 . The critical submergence corresponds to the in-
study. Due to the sophisticated present measurement technique ception point for which air is entrained. Therefore, the relative criti-
the measured air entrainment rates are judged to be determined cal intake submergence was reanalyzed at the point when the
much more precisely herein. Note that the measurement accuracy v
of Padmanabhan’s data amounts to only Ca ¼ 1%. This wide smallest air entrainment rate distance between β cr and β compared
range of scatter may lead to the conclusion that almost all data with the two classical methods of Knauss (1987) and Gordon
are within this range, rendering both data sets difficult to be com- (1970). The two design values Fco;cr ¼ vD =ðghcr Þ0.5 and FD ¼
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

pared. Note that the maximum air entrainment line (Padmanabhan v


vD =ðgDÞ0.5 are recast to ðh=DÞcr ¼ F2D =F2co;cr , with Fco;cr ðβ ¼
1984, see also Fig. 1) allows for a distinction between the two
data sets. With β ¼ Ca =ð1 − Ca Þ, the line is approximated by 10−5 Þ ¼ 0.39 from Eq. (5) and, as a conservative approach,
v Fco;cr ðβ̄ ¼ 10−5 Þ ¼ 0.43 from Eq. (4), resulting in the
β ¼ 0.117Fco − 0.044. Padmanabhan’s envelope curve is clearly critical submergence for the test range of 1.25 ≤ ðh=DÞcr ≤ 2.5.
v
Statements outside of this range remain uncertain, e.g., for h=D >
above the corresponding values of β as determined within the
2.5 the curve increases over-proportionally as compared with
present research. In particular, air entrainment within the current
Knauss (1987). To include this shortcoming, a further extension
test runs never reached Padmanabhan’s entrainment level. Conse-
to 0.75 ≤ ðh=DÞcr ≤ 3.0 was considered, based on additional data
quently, Padmanabhan (1984) line of maximum air entrainment
from scale family tests of Meyer (unpublished data, 2012; see
gives a conservative overestimation to the incipient point of air
Möller 2013) conducted at the same test setup. These data and
entrainment. As the comparison with Padmanabhan (1984) data
the findings of the present study resulted in an approximation
showed no general disagreement, the current air entrainment rates
for the critical submergence by curve fitting as
are seen to be validated.

−0.45
ðh=DÞcr ¼ −2.5FD þ 5.3; for 0.26 ≤ FD ≤ 1.2 ð8Þ
New Intake Design Approach Based on
Air-Allowance-Criterion
Eq. (8) is also plotted in Fig. 8. It is not in line with the linear
Eqs. (4) and (5) enable to estimate the air entrainment rates due to relation as suggested by Gordon (1970) and Knauss (1987). How-
intake vortices. As to the possible variability of these estimations, ever, both the latter approaches and all other published criteria for
Stahel (2008) suggests a prediction band to determine limits for intake submergence are mainly based on observations of presumed
future observations. His methodology allows for a statement of air entrainment due to occurring vortices. In contrast to this, the
a stochastic variable. In general, a prediction band has to be dis- present approach is based on real air entrainment data, where
tinguished from a confidence band. The latter does not contain the occurrence, frequency, and type of vortices are of minor impor-
the random deviation and is smaller, therefore. Limits for prospec- tance. As the current approach is based on factual measured data it
tive observations using a probability of 95% are computed. Fig. 6 is considered superior to the classical linear approaches of Knauss
shows the prediction bands graphically, whereas Table 2 lists the (1987) and Gordon (1970) for estimating the critical submergence.
associated limits as the exponential functions

β̄ ¼ limitðβ̄Þ × 10−8 expð16Fco Þ; and ð6Þ


v v
β ¼ limitðβ Þ × 10−8 expð16Fco Þ ð7Þ

Therefore, the common flow parameters h; D and vD determine


the air entrainment rates.

Critical Intake Submergence

The critical submergence hcr is widely used as parameter to deter-


mine an incipient state for which no air is entrained due to intake
vortices. Typically, this parameter is determined by observation of

Table 2. 95% Prediction Band Limits Based on Eqs. (4) and (5) Fig. 8. Critical relative intake submergence defined at air entrainment
Concerning Air Entrainment Rates β rate of β ¼ 1 × 10−5 and approximation of critical submergence in
v comparison with Knauss (1987) and Gordon (1970) (symmetrical
Fco Limit β̄ β v
<0.66 Upper 2.08 4.09 approach flow). Parameter matrix of measured β , is presented by
Lower 0.52 0.97 the shaded circles, their size being linearly correlated to measured
≥0.66 Upper 4.36 6.80 air entrainment rates. The open squares give additional data measured
Lower 0.25 0.60 by Meyer (unpublished data, 2012)

© ASCE 04015026-6 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


An increased submergence is thus necessary to avoid air entrain- The approach is intrinsic, as it accounts for the basic hydraulic
ment in the range of 0.5 ≤ FD ≤ 1.2 (Fig. 8). parameters of the mean pipe flow velocity vD and the submergence
height h. Additionally, an approach to determine the critical intake
submergence hcr based on the air entrainment rate measurements
Discussion has been developed [Eq. (8)].
Due to the overall negative consequences of air in pressurized
Based on systematic scale family tests the experimental setup with systems, a zero-air-criterion has evolved as the classical safety
a DN400 pipe is scale-invariant, thus findings related to β can be design. However, if the consequences are determined and can be
applied to prototype scale. This conclusion is further supported handled, vortex-induced air entrainment at intakes could be toler-
by the critical inception velocities of about 0.8–1.0 m=s given ated to a certain extent. The new approach to estimate β [Eqs. (4)
by Kobus (1984), which were reached or exceeded for all 33 test and (5)] is a prerequisite to apply this air allowance criterion. The
runs with DN400 considered. This statement is based on the design of alternative deaeration devices (e.g., Wickenhäuser 2008)
assumption that the scatter between the entrainment rates measured is also improved due to the knowledge of β.
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

for DN400 and DN500 (Fig. 3) are attributed to uncertainties in air


entrainment measurements, but not in limitations of the scaling
technique itself. The results showed a clear asymptotic trend for Acknowledgments
β to be constant for ≥DN400. However, the extensive character
of these experiments and the boundary conditions of the tank The authors would like to thank swisselectric research and
limited the number of experimental series to only four. Thus, the Swiss Federal Office of Energy (SFOE) for their financial
the appropriateness to scale up the current results to prototype scale support.
is, strictly speaking, only based on four (median) data points.
Consequently, an unbiased error estimation of entrainment rates
for prototype scales is illusory. There is, after all, a considerable Notation
scatter in our findings, and this uncertainty will grow as the domain
The following symbols are used in this paper:
of application is enlarged. This has to be kept in mind when up-
a = air;
scaling the results from the present study.
co = combined;
Furthermore, as the size of the tank could not be changed during
cr = critical value;
the study, the authors were not able to analyze the influence of the
Ca = air concentration (-);
tank walls quantitatively. To minimize possible wall influences on
D = intake diameter (m);
the vortex formation, the tank was designed to have a distance of
Fco = combined Froude number (-);
4D between the side walls to the intake axis, a distance of 2D to the
FD = intake Froude number (-);
front wall, and 1D to the bottom wall, following the results regard-
g = gravity acceleration (m=s2 );
ing the critical submergence due to boundary distances of Denny
h = intake submergence, with h ¼ 0 at pipe axis (m);
and Young (1957). From top view, the vortex core was typically
h=D = relative intake submergence (-);
wandering within 1D to the entry of the intake pipe. With a crude
i, j = indices;
estimation of a vortex radius <2D (compare results of Keller et al.
max = maximum;
2014) it is assumed that the wall effects are negligibly small on the
Q = discharge (m3 =s);
vortex formation process, and, consequently, on the entrainment
RD = intake Reynolds number (-);
rates. However, also various other geometries effecting flow
Rf = measured air entrainment factor (-);
characteristics might result in slightly different air entrainment
t = time (s);
rates. For example, only horizontal-orientated sharp crested intakes
vD = intake velocity (m=s);
were tested herein, but the shape of the intake pipe and the sur-
VT = vortex type (-);
rounding near and far field geometry of walls may also influence
w = water;
the formation and duration of air entraining vortices. Consequently,
WD = intake Weber number (-);
an unbiased transfer to prototype is not appropriate under all
β = relative air entrainment rate (-);
circumstances.
β̄ = mean air entrainment rate (-);
v
β = mean air entrainment rate of VT6 phases (-);
Conclusions ν = kinematic viscosity (m2 =s);
ρ = density (kg=m3 ); and
Entrained air is in general negative as to the operation, safety, and σ = surface tension (N=m).
efficiency of pressurized waterway systems. This study closes a
knowledge gap concerning the amount of vortex-induced air en-
trainment at horizontal intakes and its relation to the governing flow References
conditions. Large-scale laboratory tests enabled to investigate air
entraining vortices that are almost undisturbed by boundary con- Chang, E. (1977). “Review of literature on the formation and modelling of
ditions, including tank walls near the vortex core, or superimposed vortices in rectangular pump sumps.” Technical Rep. TN1414, British
flow fields in the approach flow. Scale family tests proved the trans- Hydromechanics Research Association (BHRA), Cranfield, UK.
ferability of measured air entrainment rates to prototype scale. Chang, K. S., and Lee, D. J. (1995). “An experimental investigation of the
air entrainment in the shutdown cooling system during mid-loop oper-
However, under some circumstances, a simple upscaling may
ation.” Ann. Nucl. Energy, 22(9), 611–619.
not be appropriate. Daggett, L. L., and Keulegan, G. H. (1974). “Similitude in free-surface
Quasi-continous air entrainment rates of up to 0.8% have been vortex formations.” J. Hydraul. Div., 100(HY11), 1565–1581.
measured. By means of regression analyses a guideline has been Denny, D. F., and Young, G. A. J. (1957). “The prevention of vortices and
developed enabling the estimation of air entrainment rates at swirl at intakes.” Proc., 7th IAHR Congress, Lisbon, Portugal.
horizontal, sharp-crested prototype intakes [Eqs. (4) and (5)]. Gordon, J. L. (1970). “Vortices at intakes.” Water Power, 22(4), 137–138.

© ASCE 04015026-7 J. Hydraul. Eng.

J. Hydraul. Eng., 2015, 141(11): 04015026


Haindl, K. (1959). “Contribution to air entrainment by a vortex.” Proc., Technology and Electrotechnology for the Mediterranean Countries,
8th IAHR Congress, Montreal, Canada. 1028–1031.
Hattersley, R. T. (1965). “Hydraulic design of pump intakes.” J. Hydraul. Ranga Raju, K. G., and Garde, R. J. (1987). “Modelling of vortices and
Div., 91(HY2), 223–249. swirling flows.” Swirling flow problems at intakes, IAHR hydraulic
Hecker, G. E. (1987). “Fundamentals of vortex intake flow, swirling flow structures design manual, J. Knauss, ed., Vol. 1, Balkema, Rotterdam,
problems at intakes.” IAHR hydraulic structures design manual, Vol. 1, Netherlands, 77–90.
Balkema, Rotterdam, Netherlands, 13–38. Stahel, W. A. (2008). Statistical data analysis, Vieweg, Wiesbaden,
Iversen, H. W. (1953). “Studies of submergence requirements of high- Germany (in German).
specific-speed pumps.” Trans. ASME, 75(4), 635–641. Suerich-Gulick, F., Gaskin, S., Villeneuve, M., and Parkinson, É. (2014a).
Jain, A. K., Ranga Raju, K. G., and Garde, R. J. (1978). “Vortex “Characteristics of free surface vortices at low-head hydropower
formation at vertical pipe intakes.” J. Hydraul. Div., 104(HY10), intakes.” J. Hydraul. Eng., 10.1061/(ASCE)HY.1943-7900.0000826,
1429–1445. 291–299.
Keller, J., Möller, G., and Boes, R. M. (2014). “PIV measurements of Suerich-Gulick, F., Gaskin, S., Villeneuve, M., and Parkinson, É. (2014b).
air-core intake vortices.” Flow Meas. Instrum., 40, 74–81. “Free surface intake vortices: Scale effects due to surface tension and
Downloaded from ascelibrary.org by ETH Zuerich - ETH-Bibliothek on 02/09/16. Copyright ASCE. For personal use only; all rights reserved.

Knauss, J. (1987). “Prediction of critical submergence, swirling flow viscosity.” J. Hydraul. Res., 52(4), 513–522.
problems at intakes.” IAHR hydraulic structures design manual, Vol. 1, Suerich-Gulick, F., Gaskin, S., Villeneuve, M., and Parkinson, É. (2014c).
Balkema, Rotterdam, Netherlands, 57–76. “Free surface intake vortices: Theoretical model and measurements.”
Kobus, H. (1984). “Local air entrainment and detrainment.” Technische J. Hydraul. Res., 52(4), 502–512.
Akademie Esslingen Symp. on Scale Effects in Modelling Hydraulic Tastan, K., and Yildirim, N. (2014). “Effects of Froude, Reynolds, and
Structures, H. Kobus, ed., International Association for Hydraulic Weber numbers on an air-entraining vortex.” J. Hydraul. Res., 52(3),
Research. 421–425.
Möller, G. (2013). “Vortex-induced air entrainment rate at intakes.” Wickenhäuser, M. (2008). “Zweiphasenströmung in Entlüftungssystemen
Mitteilung, R. M. Boes, ed., Vol. 220, VAW, ETH Zurich, Switzerland. von Druckstollen (‘Two-phase flow in de-aeration systems of pressur-
Möller, G., Detert, M., and Boes, R. M. (2012). “Air entrainment due to ized tunnels’).” Mitteilung, H.-E. Minor, ed., Vol. 205, VAW, ETH
vortices: State-of-the-art.” Proc., 2nd IAHR Europe Congress, Zurich, Switzerland (in German).
P. Rutschmann, M. Grünzner, and S. Hötzl, eds., Munich, Germany. Zhou, F., Hicks, F. E., and Steffler, P. M. (2002). “Transient flow in a
Padmanabhan, M. (1984). “Air ingestion due to free-surface vortices.” rapidly filling horizontal pipe containing trapped air.” J. Hydraul.
J. Hydraul. Eng., 10.1061/(ASCE)0733-9429(1984)110:12(1855), Eng., 10.1061/(ASCE)0733-9429(2002)128:6(625), 625–634.
1855–1859. Zhou, L., Liu, D., and Karney, B. (2013a). “Investigation on hydraulic
Papillon, B., Kirejczyk, J., and Sabourin, M. (2000). “Atmospheric air transients of two entrapped air pockets in a water pipeline.” J. Hydraul.
admission in hydroturbines.” Proc., HydroVision. Eng., 10.1061/(ASCE)HY.1943-7900.0000750, 949–959.
Pfister, M., and Chanson, H. (2014). “Two-phase air-water flows: Scale Zhou, L., Liu, D., Karney, B., and Wang, P. (2013b). “Phenomenon
effects in physical modeling.” J. Hydrodyn., 26(2), 291–298. of white mist in water rapidly filling pipeline with entrapped air
Poullikkas, A. (2000). “Effects of entrained air on the performance of pocket.” J. Hydraul. Eng., 10.1061/(ASCE)HY.1943-7900.0000765,
nuclear reactor cooling pumps.” Proc., Melecon 2000: Information 1041–1051.

© ASCE 04015026-8 J. Hydraul. Eng.

View publication stats J. Hydraul. Eng., 2015, 141(11): 04015026

You might also like