Anda di halaman 1dari 6

Bernoulli Theorem, Minimum Specific Energy, and Water

Wave Celerity in Open-Channel Flow

Oscar Castro-Orgaz1 and Hubert Chanson2

Abstract: One basic principle of fluid mechanics used to resolve practical problems in hydraulic engineering is the Bernoulli theorem
along a streamline, deduced from the work-energy form of the Euler equation along a streamline. Some confusion exists about the
applicability of the Bernoulli theorem and its generalization to open-channel hydraulics. In the present work, a detailed analysis of the
Bernoulli theorem and its extension to flow in open channels are developed. The generalized depth-averaged Bernoulli theorem is
proposed and it has been proved that the depth-averaged specific energy reaches a minimum in converging accelerating free surface flow
over weirs and flumes. Further, in general, a channel control with minimum specific energy in curvilinear flow is not isolated from water
waves, as customary state in open-channel hydraulics.
DOI: 10.1061/共ASCE兲IR.1943-4774.0000084
CE Database subject headings: Open channel flow; Critical flow; Weirs; Flumes; Water waves.

Introduction Bernoulli Theorem

One of the most useful principles of fluid mechanics to solve The integration of the Navier-Stokes equations along a streamline
practical problems in hydraulic engineering is the Bernoulli theo- assuming that the flow is steady and the fluid is inviscid and
rem along a streamline, which is deduced from the work-energy incompressible, yields the Bernoulli equation along a streamline
form of the Euler equation along a streamline 共Rouse 1970兲. 共Rouse 1970; Liggett 1994; Chanson 2004, 2006兲
Many applications in open-channel hydraulics are based upon p V2
such a theorem that is only valid along a given streamline in first +z+ = const 共1兲
␥ 2g
instance. Some confusion exists about the applicability of the Ber-
noulli theorem, and its generalization to open-channel hydraulics. where p / ␥ = pressure head; z = vertical elevation of the fluid par-
Very few isolated studies 共Liggett 1993; Chanson 2006, 2008兲 ticle; and V = magnitude of velocity vector. The vorticity causes
have developed the Bernoulli equation to open-channel flow the constant in Eq. 共1兲 to change from one streamline to another.
problems. The extension of this principle to open-channel flows If both sides of Eq. 共1兲 are multiplied by the elementary discharge
provides a basic equation applicable to the calculation of the across a streamline, dQ = udA, with Q = discharge, A = flow cross
minimum specific energy and critical flow conditions, a physical section area, u = component of velocity vector normal to A, and
the resulting expression is integrated across a channel section, one
phenomenon that determines the head-discharge relationship in
obtains a constant quantity given by
control structures used for water measurement in irrigation and
sewage techniques, as flumes and weirs 共Fig. 1兲.
冕冉 冊
p V2
In the present study, a detailed and generalized extension of +z+ udA = const 共2兲
0 ␥ 2g
the Bernoulli theorem to open-channel flow is developed. Using
analytical and experimental results, the occurrence of minimum Both sides of Eq. 共2兲 may be divided by the total discharge Q, that
specific energy in open channels is reanalyzed, and general results is assumed a constant for all sections, from which it is obtained
for the critical flow depth in curvilinear flow are provided. Also, that the total head H of a cross section is conserved in the flow
some practical advice for the selection of gauging stations is high- direction

冕冉 冊
lighted in relation to wave motion at the section of minimum A
1 p V2
specific energy H= +z+ udA = const 共3兲
Q 0 ␥ 2g
Research Engineer, Dept. of Agronomy, Univ. of Cordoba, The total head H gives the total convective flow of energy across
c/Fernando Colón nº1, 3 izq., E-14002 Cordoba, Spain. E-mail: oscar@
A, as discussed in detail by Jaeger 共1956兲. A cross-sectional total
2 piezometric pressure coefficient Ke may be defined as 共Jaeger
Professor in Hydraulic Engineering, School of Civil Engineering,
Univ. of Queensland, Brisbane, Queensland 4072, Australia. E-mail: 1956兲

冕冉 冊 A
1 p
Note. This manuscript was submitted on May 26, 2008; approved on Ke = + y udA 共4兲
February 27, 2009; published online on March 5, 2009. Discussion period hQ 0 ␥
open until May 1, 2010; separate discussions must be submitted for indi-
vidual papers. This paper is part of the Journal of Irrigation and Drain- where h = flow depth and y = coordinate in the vertical direction
age Engineering, Vol. 135, No. 6, December 1, 2009. ©ASCE, ISSN above the channel bed, and the extended Coriolis coefficient ␣ for
0733-9437/2009/6-773–778/$25.00. curvilinear flow is 共Rouse 1970兲


Downloaded 24 Nov 2009 to Redistribution subject to ASCE license or copyright; see

冊 冉
dA − h + zb +
V2o dA
2g dx
=0 冊 共11兲

where Vo = free surface streamline velocity. The cross-averaged

mean head Hm is defined as

冕冉 冊
1 p V2
Hm = +z+ dA 共12兲
A 0 ␥ 2g
Fig. 1. Critical flow over 共a兲 round-crested weirs; 共b兲 Venturi It is worth pointing out that the definition of a mean value of the
channels energy head across a section was first defined by Rouse 共1932兲, in
relation to curvilinear flows in spillways.
Using Eq. 共12兲, Eq. 共11兲 is rewritten as

␣= V2udA 共5兲 dHm 共Ho − Hm兲 dA
U 2Q 0 = 共13兲
dx A dx
where U = Q / A = mean flow velocity. It may be remarked that
which is the form of the Bernoulli theorem in terms of the cross-
Eq. 共5兲 is a general expression for a Coriolis coefficient, in con-
sectional averaged mean head for all the streamlines Hm, with
trast to the widely used expression
Ho = zb + h + V2o / 2g = free surface streamline energy head. A first

A point that deserves attention is that the depth-averaged form of
␣= u3dA 共6兲 the Bernoulli theorem applied to an open-channel flow does not
U 2Q 0 imply a constant. Indeed, the head Hm共x兲 changes due to the vary-
ing flow area A共x兲 in the flow direction, as well as due to the local
that does not take into account all the velocity components, and,
difference between the energy head of the free surface Ho in
thus, is only accurate for flows with parallel streamlines. The total
relation to the mean value Hm across the depth.
head H may be rewritten, using Eqs. 共4兲 and 共5兲, as
A cross-sectional averaged piezometric pressure coefficient Km
can be defined as 共Rouse 1932; Chanson 2006兲
H = z b + K eh + ␣ = zb + E = const 共7兲

冕冉 冊
2g A
1 p
where zb = channel bed elevation and E = total specific energy. Km = + y dA 共14兲
hA 0 ␥
Eq. 共7兲 was developed by Jaeger 共1956兲 and discussed recently by
Castro-Orgaz 共2008兲. Using Eq. 共7兲, the equation of motion is and the “apparent” Boussinesq coefficient for curvilinear flow is
simply written as 共Chanson 2006兲

dH dzb dE dh

= + =0 共8兲 1
dx dx dh dx ␤= V2dA 共15兲
U 2A 0
Eq. 共7兲 was deduced from the application of the Bernoulli theo-
rem along a streamline, and can be viewed as the generalized Eq. 共15兲 is refereed to as an “apparent” Boussinesq coefficient,
Bernoulli theorem for open-channel flow in terms of the total as it contains the magnitude of the velocity vector V, which is
head H, and, hence, of the total flow of energy. Critical flow a scalar magnitude arising from the energetic nature of the
conditions, as given by the minimum specific energy dE / dh = 0, Bernoulli theorem along a streamline 共Rouse 1970兲. The
are deduced from Eq. 共8兲 when dzb / dx = 0: i.e., at the crest of a Boussinesq coefficient is a tensorial magnitude, defined as a vec-
weir 共Henderson 1966兲, with a continuous smooth free surface tor in a given direction, along which the conservation of momen-
共dh / dx ⬍ 0兲 and without any vertical flow profile 共Montes 1998兲 tum is applied 共Yen 1973兲. Therefore, the Boussinesq coefficient
as classically stated by Bélanger 共1828兲. is defined in the x direction as 共Yen 1973; Liggett 1993兲
If we are now interested in making a cross-sectional mean

value for the energy head of all the streamlines, Hm, we cannot 1
␤xx = u2dA 共16兲
simply multiply Eq. 共1兲 by dA and then divide the result by A, as U 2A 0
the cross-sectional area A共x兲 varies along the flow according to
the local flow depth h = h共x兲. However, if the Bernoulli theorem Using Eqs. 共14兲 and 共15兲, the total mean head Hm is rewritten as
for a streamline, given by Eq. 共1兲, is differentiated

d p
dx ␥
=0 冊 共9兲
H m = z b + K mh + ␤
= zb + Em

where Em = depth-averaged specific energy, as given by Chanson


it is mathematically permissible to write 共2006兲. Expanding Eq. 共13兲, yields

冕 冉 冊
d p V2 dzb dEm 共Ho − Hm兲 dA
+z+ dA = 0 共10兲 + = 共18兲
0 dx ␥ 2g dx dx A dx
Using the Leibnitz rule, Eq. 共10兲 can be rewritten as For plane channel flow, Eq. 共18兲 can be written as


Downloaded 24 Nov 2009 to Redistribution subject to ASCE license or copyright; see
Fig. 2. Experimental data 共Fawer 1937兲 over cylindrical weir: 共䊏兲 Fig. 3. Experimental data 共Khafagi 1942兲 in a Venturi channel
dimensionless pressure head distribution p / ␥Eo共y / h兲, 共쎲兲 dimen- for Q = 22 l / s: 共䊏兲 dimensionless pressure head distribution
sionless total velocity distribution V / 共2gEo兲1/2共y / h兲, 共䉱兲 dimension- p / ␥Eo共y / h兲, 共쎲兲 dimensionless total velocity distribution
less total head distribution E / Eo共y / h兲 V / 共2gEo兲1/2共y / h兲, 共䉱兲 dimensionless total head distribution
E / Eo共y / h兲

dzb dEm dh Ho − Hm dh
+ = 共19兲 fluid flow equations in open-channel flow 共Yen 1973兲. Eq. 共23兲
dx dh dx h dx proved to be a particular case of the more general relation
which is the generalized depth-averaged Bernoulli theorem for 关Eq. 共19兲兴 developed herein.
open-channel flows. At the crest of a weir, dzb / dx = 0 and Eq. 共19兲 The test data of Fawer 共1937兲 with flow over round-crested
yields weirs and of Khafagi 共1942兲 in Venturi flumes were used to verify
Eq. 共21兲. Fig. 2 presents the data of Fawer 共1937兲 for a circular
dEm Ho − Hm weir of radius 3.25 and 30.3 cm width under a discharge of
= 共20兲
dh h 0.01525 m3 / s, a flow depth h = 5.37 cm and a specific energy
head over the weir of E = 7.68 cm. The test data show that the
Eq. 共20兲 implies that, strictly speaking, the depth-averaged mean flow is nearly irrotational, with a computed value dEm / dh
specific energy is only minimal 共dEm / dh = 0兲 at the crest of a weir = −0.012 using Eq. 共20兲. Figs. 3–5 show the data of Khafagi for a
when the flow is irrotational and, therefore, Ho = Hm. In a real flow Venturi channel of throat width 12 cm, inlet width 30 cm, radius
the vorticity causes a difference between Ho and Hm. However, in of channel sidewalls 54.5 cm and discharges of 22, 17.5, and 14
short transitions with accelerating converging streamlines, as in L/s, respectively. The test data yield dEm / dh = 0.00498, dEm / dh
the case of flumes and weirs, the flow is nearly irrotational and = 0.00547, and dEm / dh = 0.00796, for discharges of 22, 17.5, and
thus one obtains 14 L/s, respectively. Thus, theory and experiments support the
dEm Ho − Hm occurrence of minimum depth-averaged specific energy at chan-
= ⬇0 共21兲 nel controls.
dh h
from which
Minimum Specific Energy
Hm = zb + Kmh + ␤ = zb + Em ⬇ const 共22兲
2g The concept of critical flow was historically developed as
the singularity in the backwater equation for open-channel flows
Eq. 共22兲 was successfully verified by Chanson 共2006兲 with test
共Bélanger 1828; Chanson 2004兲. Later, it was concluded by
data on round-crested weirs, computing the coefficients ␤ and Km
Bakhmeteff 共1932兲 what the conditions are at which the specific
using flow net diagrams. Montes 共1998兲 estimated the coefficients
␤ and Km of Eq. 共22兲 with a Boussinesq-type approach, and com-
pared successfully the results with test data on parabolic weirs. In
flows with nearly parallel streamlines, the flow is “gradually var-
ied,” and the pressure is hydrostatic 共Km = 1兲, the vertical velocity
is negligible 共u ⬇ V兲 and the free surface slope is very small
共dh / dx ⬇ 0兲. Under these conditions, the generalized depth-
averaged Bernoulli theorem 关Eq. 共19兲兴 gives
zb + h + ␤xx = const 共23兲
which was derived by Liggett 共1993兲. It is of interest to remark
that the result of Liggett 共1993兲 was obtained from the depth-
averaged form of the Euler equations for flows with parallel
streamlines and a velocity shape almost invariant with distance Fig. 4. Experimental data 共Khafagi 1942兲 in a Venturi channel
共␤xx ⬇ const兲. The depth-averaging process of the momentum for Q = 17.5 l / s: 共䊏兲 dimensionless pressure head distribution
equation yields an equation with a Boussinesq coefficient p / ␥Eo共y / h兲, 共쎲兲 dimensionless total velocity distribution
关Eq. 共23兲兴 rather than with a Coriolis coefficient 关Eq. 共7兲兴, in V / 共2gEo兲1/2共y / h兲, 共䉱兲 dimensionless total head distribution
agreement with the full integration of the energy and momentum E / Eo共y / h兲


Downloaded 24 Nov 2009 to Redistribution subject to ASCE license or copyright; see
Fig. 5. Experimental data 共Khafagi 1942兲 in a Venturi channel
for Q = 14 l / s: 共䊏兲 dimensionless pressure head distribution
p / ␥Eo共y / h兲, 共쎲兲 dimensionless total velocity distribution
V / 共2gEo兲1/2共y / h兲, 共䉱兲 dimensionless total head distribution
E / Eo共y / h兲

Fig. 6. Discharge curve of flow in open channels 共a兲 共—兲 Eq. 共26兲,
energy reaches a minimum value. The developments herein prove 共䊊兲 experimental data Chanson and Montes 共1997, 1998兲 flow over
that, in curvilinear flow, it is possible to define the concept of cylindrical weirs, 共䉱兲 experimental data Khafagi 共1942兲 flow in
critical flow either using a convective energy flux total specific Venturi channels; 共b兲 共—兲 Eq. 共26兲, 共쎲兲 experimental data Fawer
head E or a depth-averaged specific head for all streamlines 共1937兲 flow over round-crested weirs, 共䊐兲 experimental data Blau
共Montes 1998兲 flow over parabolic weirs; 共c兲 共—兲 Eq. 共26兲, 共䉭兲
across the depth, Em. If the flow is irrotational, it is also permis-
experimental data Kindsvater 共1964兲 flow over trapezoidal-profile
sible, and even simpler, to write
weirs; and 共d兲 共—兲 Eq. 共26兲, 共䉭兲 experimental data Gonzalez and
V2o Chanson 共2007兲 flow over broad-crested weirs, 共䊊兲 experimental data
Em ⬇ Eo = h + 共24兲 Chanson 共2005兲 in near critical flows
where Eo = free surface specific energy
Thus, the depth-averaged specific energy is accurately repre- depth at the crest, with up to three real values of the flow depth-
sented by the specific energy of the free surface streamline. This energy ratio for any product of the square of discharge coefficient
approach avoids the use of depth-averaging coefficients, and per- and fictitious Coriolis coefficient. For the establishment of critical
mits one to represent the head-discharge relationship with only flow conditions, two simultaneous conditions are required. First,
one parameter, a “fictitious” Coriolis coefficient ␣o defined as an extreme in the channel geometry 共maximum elevation in a

冉 冊 2
weir, minimum width in a flume, Henderson 1966兲 is needed
Vo to create a potential section for the appearance of critical flow.
␣o = 共25兲
U Second, in the extreme section, a sufficient condition given by
the derivative dEo / dh = 0 is required to provide a unique relation
that represents the quotient between the free surface and the mean
between Eo and h for a given Q, avoiding two of the solutions of
velocities. From Eq. 共24兲, the generalized channel flow relation is,
Eq. 共26兲 that do not imply critical flow conditions. A comparison
using Eq. 共25兲 共Castro-Orgaz 2008兲
of test data with the whole curve 关see Figs. 6共a–c兲兴 simply shows

冉 冊冉 冊
−1 +
=0 共26兲
that the upper branch is the critical flow solution of the three
possible roots. However, although the critical points can only lie
in the upper branch, other possible types of flows can also lie
where Cd = discharge coefficient= Q / 关b共gH3o兲1/2兴 and b = channel there. Subcritical flow over the whole weir profile implies higher
width. tailwater levels than the modular limit of the weir 共Dominguez
Eq. 共26兲 and the experimental data of Chanson and Montes 1959; Montes 1998兲. Under these conditions, the relation h / Eo at
共1997, 1998兲, series QIIA, for flow over circular-crested weirs, the weir crest increases above the values for free flow. As the
the data of Fawer 共1937兲 of flow over round-crested weirs, tailwater level increases for a given upstream head Eo, the flow
the data of Blau of parabolic weirs 共Montes 1998兲, the data of that passes over the weir is reduced. Extreme submergence con-
Kindsvater 共1964兲 共model 2, free flow conditions兲 on flow over ditions imply h / Eo = 1, and, consequently, ␣oC2d / 2 = 0, which is
trapezoidal-profile weirs, and the data of Khafagi 共1942兲 for the initial point of the upper branch.
circular-shaped inlet flume of rectangular cross section are plotted The geometry of a broad-crested weir does not allows for an
in Figs. 6共a–c兲. As shown in Figs. 6共a–c兲, all the data of critical extreme in the channel geometry, given by a channel bottom el-
flow in flumes and weirs collapse in the upper branch of evation or a width contraction, and thus, critical depth and its
the curve. Then, of the two branches of the discharge curve, only position on the weir are governed mainly by frictional effects and
the upper branch has a physical meaning of critical flow at a streamline curvature 共Rouse 1932兲. It is then futile to attempt to
weir crest and at a flume throat, corresponding to relations h / Eo define the discharge characteristics of the broad-crested weir try-
⬎ 2 / 3 for curvilinear flows. In parallel flows, h / Eo = 2 / 3 and, ing to locate the real critical depth section, which, in the more
consequently, the particular point ␣oC2d / 2 = 4 / 27 in Fig. 6 is ob- general case, is not necessarily equal to the hydrostatic pressure
tained. However, the lower part of the curve was not close to any critical depth, a case in which computations become complex
experimental data. This is because Eq. 共26兲 is a mathematical 共Castro-Orgaz 2008兲. Then, although in the strictest sense one
relation between the discharge, the specific energy, and the flow cannot find the critical depth section on a broad-crested weir with


Downloaded 24 Nov 2009 to Redistribution subject to ASCE license or copyright; see
Fig. 7. Minimum specific energy and celerity of shallow-water wave
over a round-crested weir

any accuracy, it is relevant to analyze data on flow over broad- Fig. 8. Water wave celerity and velocity profiles, 共——兲 normalized
crested weirs. The test data of Gonzalez and Chanson 共2007兲 for velocity profile V / 共gh兲1/2共y / h兲 for E / R = 1, 共– –兲 normalized velocity
a large broad-crested weir are also plotted in Fig. 6共d兲, corre- profile V / 共gh兲1/2共y / h兲 for E / R = 2.5, 共쎲兲 dimensionless celerity
sponding to flow depths in gauging stations on the first half of c / 共gh兲1/2共␭ / h兲 of linear Airy waves 关Eq. 共28兲兴
broad-crested weir models. As seen in Fig. 6共d兲, the flow over a
broad-crested weir may be properly called transcritical, rather Similar conclusions were outlined by Castro-Orgaz 共2008兲, who
than critical, as the flow changes between the two real branches of improved the celerity c by incorporating the nonuniform velocity
Eq. 共26兲 without a definite flow pattern in terms of a critical depth and nonhydrostatic pressure effects. Therefore, only in the de-
relationship h / Eo. In this regard, it is also interesting to plot data scending branch of the weir, where the flow is parallel, one can
with near critical flows 共Chanson 2005兲 in Fig. 6共d兲. In those find sections isolated from shallow-water waves. The flow feature
cases, all the experiment points lie in the upper branch, but, in discussed is relevant in such irrigation works as the flow dividers
most cases, the relationship h / Eo is exceedingly high 共⬇0.85兲 for analyzed by Dominguez 共1959兲.
representing critical flows, even when including streamline cur- The conclusions of this discussion demand caution in engi-
vature effects. Note that flow over round-crested weirs 共Fawer neering practice when trying to define whether the flow is sub-
1937兲 imply h / Eo around 0.7. It proves again that although critical or supercritical by means of causing surface waves in a
curved streamline critical flows lie only in the upper branch, with, channel and observing the direction in which these waves travel.
typically, h / Eo ⬇ 0.7 as a mean, other types of curvilinear flows It was proved that the section of minimum specific energy
can also be found there, as near critical flows. 共i.e., critical flow section兲 is in general not a section which equals
the shallow-water wave celerity. For the test data of Fawer
共Fig. 2兲 the dimensionless critical depth at the weir crest is
Water Wave Celerity h / Ho = 0.699, and the dimensionless celerity is c / 共2gEo兲1/2
= 共0.699/ 2兲1/2 = 0.5912. This value is very near to the dimension-
In the previous section, a set of expressions for the Bernoulli less velocity at the free surface 关Vo / 共2gEo兲1/2 = 0.548, see Fig. 2兴,
theorem in an open-channel flow were developed, each of which so, in this case, nearly the whole flow section is isolated
could be used to define critical flow based on the concept of from “shallow-water waves.” More questionable is the fact that
minimum specific energy. Recently, it was shown that in general Eq. 共27兲 is the “computational velocity of floods” when the
critical flow is not single magnitude when it is defined with the method of characteristics is applied to the Saint-Venant gradually
energy and momentum principles 共Castro-Orgaz 2008兲. A simple varied unsteady shallow-flow equations, but a two-dimensional
and relevant case can be explained in relation to the weir flow 共2D兲 water wave celerity may differ from Eq. 共27兲 共Montes 1998兲
case, and the Bernoulli theorem. Consider the weir flow drawing when streamline curvature is important. Moreover, when nearly
of Fig. 7. At the weir crest, the streamlines are curved and sloped, the whole section is isolated from shallow-water waves, as in the
and the velocity distribution increases from the free surface to the case of the experiments of Fawer 共1937兲, it does not give any
channel bottom. According to the Bernoulli theorem for a stream- guarantee of a similar performance under a real 2D water wave
line, an increase in the velocity head causes a drop in the pres- motion. This flow feature deserves some discussion. Hager 共1991兲
sure, which is no longer hydrostatic across the depth. The increase showed that the velocity distribution over round-crested weirs can
in the velocity causes an increase in the discharge for a given be described using the free vortex approach V 共R + y兲 = const, with
head, and the test data of Fawer 共1937兲 and the computations of R = curvature radius at channel bed. Normalized velocity profiles
Chanson 共2006兲 proved this flow feature. Critical flow can also be V / 共gh兲1/2共y / h兲 for two typical values of the dimensionless head,
defined using momentum considerations, in relation to the E / R = 1 and 2.5, are plotted in Fig. 8. At the elevation y / h where
shallow-water wave celerity of a one-dimensional 共1D兲 flow V / 共gh兲1/2 = 1 the velocity profile equals the shallow-water wave
共Montes 1998; Chanson 2004兲 celerity. Fig. 8 shows that the incipient shallow-flow celerity ap-
pears at y / h = 0.71 and 0.8 for E / R = 2.5 and 1, respectively, indi-
c = 共gh兲1/2 共27兲 cating that in general, as a mean, the upper 25% of the flow depth
with c = shallow-water wave velocity of a 1D flow. Eq. 共27兲 was is not isolated from shallow-water waves. Some additional data
first proposed by Joseph-Louis Lagrange 共1781兲. Fig. 7 shows can be added to Fig. 8, considering the general dimensionless
that, under critical flow conditions, only a particular streamline celerity c / 共gh兲1/2共␭ / h兲 of linear Airy waves 共Montes 1998兲
has a velocity equal to the celerity c. As a result, the flow region
above the section of minimum specific energy 共weir crest兲 is not
isolated from shallow-water waves. This shows that minimum
= 冋
册 1/2

specific energy considerations are not necessarily in agreement where ␭ = wavelength. For shallow flows, ␭ / h → ⬁ and Eq. 共28兲
with momentum concepts when defining critical flow conditions. equals Eq. 共27兲. As seen in Fig. 8 the celerity of linear waves


Downloaded 24 Nov 2009 to Redistribution subject to ASCE license or copyright; see
support substantively the previous conclusions, proving that, in ␤ ⫽ apparent Boussinesq coefficient 共-兲;
general, the section of minimum specific energy is not isolated ␤xx ⫽ Boussinesq coefficient 共-兲;
from water wave motion. ␥ ⫽ specific fluid weight 共N m−3兲; and
␭ ⫽ wavelength 共m兲.

A detailed analysis of the Bernoulli theorem and its extension to
flow in open channels has been developed. From the analytical Bakhmeteff, B. A. 共1932兲. Hydraulics of open channels, McGraw-Hill,
results of the extension of the Bernoulli principle to open-channel New York.
flow, the generalized depth-averaged Bernoulli theorem is pro- Bélanger, J. B. 共1828兲. Essai sur la solution numérique de quelques
problèmes relatifs au mouvement permanent des eaux courantes,
posed, extending the earlier works of Liggett 共1993兲 and Chanson
Carilian-Goeury, Paris 共in French兲.
共2006,2008兲. From the new depth-averaged Bernoulli equation, it
Castro-Orgaz, O. 共2008兲. “Critical flow in hydraulic structures.” Ph.D.
was shown that the depth-averaged specific energy reaches a thesis, Univ. of Cordoba, Spain.
minimum in converging accelerating free surface flow over weirs Chanson, H. 共2004兲. The hydraulics of open channel flows: An introduc-
and flumes. A generalized open-channel flow diagram based on tion, Butterworth-Heinemann, Stoneham, Mass.
the Bernoulli theorem was used to show analytically the critical Chanson, H. 共2005兲. “Physical modelling of the flow field in an undular
depth relationships in curvilinear flows. A comparison with ex- tidal bore.” J. Hydraul. Res., 43共3兲, 234–244.
periment data of round-crested weirs and Venturi channels dem- Chanson, H. 共2006兲. “Minimum specific energy and critical flow condi-
onstrated the validity of the analytical findings. In general, a tions in open channels.” J. Irrig. Drain. Eng., 132共5兲, 498–502.
channel control with minimum specific energy in curvilinear flow Chanson, H. 共2008兲. “Closure to ‘Minimum specific energy and critical
is not isolated from water waves. Hence, any method for produc- flow conditions in open channels.’” J. Irrig. Drain. Eng., 134共6兲,
ing waves in water is usually not appropriate for deciding whether 883–887.
the flow is subcritical or supercritical. Chanson, H., and Montes, J. S. 共1997兲. “Overflow characteristics of cy-
lindrical weirs.” Research Rep. No. CE 154, Dept. of Civil Engineer-
ing, Univ. of Queensland, Brisbane.
Chanson, H., and Montes, J. S. 共1998兲. “Overflow characteristics of cir-
Notation cular weirs: Effects of inflow conditions.” J. Irrig. Drain. Eng.,
124共3兲, 152–162.
The following symbols are used in this paper: Dominguez, F. J. 共1959兲. Hydraulics, Editorial Nascimiento, Santiago de
A ⫽ cross-sectional area 共m2兲; Chile, Chile 共in Spanish兲.
b ⫽ channel width 共m兲; Fawer, C. 共1937兲. “Étude de quelques écoulements permanents à filets
Cd ⫽ discharge coefficient 共-兲; courbes.” Ph.D. thesis, Université de Lausanne, La Concorde, Lau-
c ⫽ velocity of a shallow-water 1D wave 共m s−1兲; sanne, Switzerland 共in French兲.
E ⫽ total specific energy of flow, also specific energy of Gonzalez, C. A., and Chanson, H. 共2007兲. “Experimental measurements
of velocity and pressure distributions on a large broad-crested weir.”
a streamline 共m兲;
Flow Meas. Instrum., 18, 107–113.
Em ⫽ mean specific energy of flow 共m兲;
Hager, W. H. 共1991兲. “Experiments on standard spillway flow.” Proc.,
Eo ⫽ specific energy of free surface streamline 共m兲;
ICE, Vol. 91, Institution of Civil Engineers, London, UK, 399–416.
g ⫽ acceleration of gravity 共m s−2兲;
Henderson, F. M. 共1966兲. Open channel flow, McMillan, New York.
H ⫽ total energy head of flow 共m兲; Jaeger, C. 共1956兲. Engineering fluids mechanics, Blackie and Son, Edin-
Hm ⫽ mean energy head of flow 共m兲; burgh, U.K.
Ho ⫽ free surface streamline energy head 共m兲; Khafagi, A. 共1942兲. Der venturikanal: Theorie und anwendung. versuch-
h ⫽ flow depth 共m兲; sanstalt für wasserbau, eidgenössische technische hochschule zürich,
Km ⫽ cross-sectional averaged piezometric pressure mitteilung 1, Leemann, Zürich, Switzerland.
coefficient 共-兲; Kindsvater, C. E. 共1964兲. “Discharge characteristics of embankment
Ke ⫽ total head piezometric pressure coefficient 共-兲; shaped weirs.” Geological survey water supply paper 1617-A, U.S.
p / ␥ ⫽ pressure head 共m兲; Department of the Interior, Denver.
Q ⫽ discharge 共m3 s−1兲; Lagrange, J. L. 共1781兲. “Mémoire sur la théorie du mouvement des flu-
R ⫽ curvature radius at channel bed 共m兲; ides.” 关Memoir on the theory of fluid motion兴, Oeuvres de Lagrange,
U ⫽ mean flow velocity= Q / A 共m s−1兲; Vol. 4, Gauthier-Villars, Paris, France, 695–748 共in French兲.
u ⫽ component of local velocity normal to A 共m s−1兲; Liggett, J. A. 共1993兲. “Critical depth, velocity profile and averaging.”
V ⫽ magnitude of velocity vector 共m s−1兲; J. Irrig. Drain. Eng., 119共2兲, 416–422.
Vo ⫽ free surface streamline velocity 共m s−1兲; Liggett, J. A. 共1994兲. Fluid mechanics, McGraw-Hill, New York.
x ⫽ streamwise distance 共m兲; Montes, J. S. 共1998兲. Hydraulics of open channel flow, ASCE, Reston,
y ⫽ coordinate in the vertical direction above the
Rouse, H. 共1932兲. “The distribution of hydraulic energy in weir flow in
channel bed 共m兲; relation to spillway design.” MS thesis, MIT, Boston.
zb ⫽ elevation of the channel bed 共m兲; Rouse, H. 共1970兲. “Work-energy equation for the streamline.” J. Hydraul.
z ⫽ vertical elevation of a fluid particle 共m兲; Div., 96共5兲, 1179–1190.
␣ ⫽ Coriolis coefficient 共-兲; Yen, B.C. 共1973兲. “Open-channel flow equations revisited.” J. Engng.
␣o ⫽ fictitious Coriolis coefficient 共-兲; Mech. Div., 99, 979–1009.


Downloaded 24 Nov 2009 to Redistribution subject to ASCE license or copyright; see