You are on page 1of 7

Flow Measurement and Instrumentation 22 (2011) 4349

Contents lists available at ScienceDirect

Flow Measurement and Instrumentation


journal homepage: www.elsevier.com/locate/flowmeasinst

Explicit solutions for critical and normal depths in channels with different shapes
Ali R. Vatankhah a, , Said M. Easa b
a

Department of Irrigation and Reclamation Engineering, University College of Agriculture and Natural Resources, University of Tehran, P.O. Box 4111, Karaj, 31587-77871, Iran

Department of Civil Engineering, Ryerson University, Toronto, ON, Canada M5B 2K3

article

info

Article history:
Received 19 August 2010
Received in revised form
1 November 2010
Accepted 12 December 2010
Keywords:
Critical and normal depths
Open channels flow
Explicit equations

abstract
Critical and normal depths are important for computing gradually varied flow profiles and for the design,
operation, and maintenance of open channels. A closed-form analytical equation for the normal depth
computation can only be derived for triangular channels. For exponential channels, it is also possible to
obtain such equations for the critical depth. This is not possible, however, for other geometries, such as
trapezoidal, circular, and horseshoe channels. In these channels, the governing equations are implicit and
thus the use of trial procedures, numerical methods, and graphical tools is common. Some channels have
explicit solutions for the critical and normal depths, while others do not. This paper presents new and
improved explicit regression-based equations for the critical and normal depths of open channels with
different shapes. A comparison of the proposed and existing equations is also presented. The proposed
equations are simple, have a maximum error of less than 1%, and are well-suited for manual calculations
and computer programming.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
The critical and normal depths play a major role in the design,
operation, and maintenance of open channels [1]. The critical
depth is a quantity of fundamental importance to understanding
the flow characteristics in open channels. If the actual depth is
less than the critical depth, the flow is considered supercritical,
which is fast-flow and is impacted by the upstream conditions.
If the actual depth is greater than the critical depth, the flow
is considered subcritical, which is slow-flow and is impacted by
the downstream conditions [2,3]. The occurrence of the critical
depth at the upstream end of a reach under supercritical flow
conditions or at the downstream end of a reach under subcritical
flow conditions provides a control section for gradually varied flow
computations [4].
The normal depth occurs in a steady uniform flow for a given
channel geometry, slope, and roughness, and a specified value of
the discharge. The normal depth is an important parameter for the
hydraulic design of open channels and has ramifications for flood
prediction. It is also important to control and make efficient use of
such channels.
For circular, trapezoidal, and horseshoe channels, the governing
equations for the critical and normal depths are implicit and
no analytical solutions exist. For these channels the critical
and normal depths are presently obtained by trial procedures,

Corresponding author.
E-mail addresses: arvatan@ut.ac.ir, alireza_vatankhah@yahoo.com
(A.R. Vatankhah), seasa@ryerson.ca (S.M. Easa).
0955-5986/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.flowmeasinst.2010.12.003

numerical and graphical methods, or explicit regression-based


equations. Specifically for the critical depth, explicit equations are
available for only trapezoidal [47] and circular channels [4,8,9],
but not for horseshoe channels. For the normal depth, explicit
equations are available for only rectangular [1013] and horseshoe
channels [14,15], but not for trapezoidal or circular channels.
In this paper, explicit solutions for the critical and normal
depths are presented for trapezoidal, circular, and horseshoe
channels using the curve fitting method. The proposed equations
either fill the current gap or improve upon existing equations in
terms of accuracy and simplicity. A comparison of the accuracy
of the proposed and existing solutions is also presented. Before
presenting the derivation of these explicit equations, it is necessary
to describe the geometric properties of the channel cross sections.
2. Geometric properties
2.1. Trapezoidal channels
Considering Fig. 1 for a trapezoidal channel section, the
applicable equations are as follows
A = (b + zy)y

(1)

P = b + 2y 1 + z 2

(2)

T = b + 2zy

(3)

where A is cross section area, b is bed width, z is horizontal distance


corresponding to 1 m vertical distance (side slope of the channel),
y is flow depth, P is wetted perimeter, and T is width of the channel
at the water surface.

44

A.R. Vatankhah, S.M. Easa / Flow Measurement and Instrumentation 22 (2011) 4349

Notations
A
Cross section area
a, b, and c Coefficients
b
Bed width
D
Channel diameter
g
Gravitational acceleration
H
Height of the conduit
n
Mannings roughness coefficient
P
Wetted perimeter
Q
Discharge
R, r
Circle radius
sin() Longitudinal slope of the channel
S0
Longitudinal slope of the channel
t = R/r A characteristic parameter
T
Width of the channel at the water surface
y
Flow depth
yn
Normal flow depth
z
Side slope of the channel

Energy correction factor

Unit conversion constant

Water surface angle in radians


= y/D Dimensionless depth
c = Q 2 /[gD5 cos( )] Dimensionless discharge for critical
depth computations (circular cross section)
h = Q 2 /[gH 5 cos( )] Dimensionless discharge for critical
depth computations (standard horseshoe cross
section)
t = z 3 Q 2 /[gb5 cos( )] Dimensionless discharge for critical depth computations (trapezoidal cross section)
c = 2 cos1 (1 2cc ) For critical depth computations
n = 2 cos1 (1 2nc ) For normal depth computations
c = nQ /(D8/3 S0 ) Dimensionless discharge for normal
depth computations
(circular cross section)

h = nQ /(H 8/3 S0 ) Dimensionless discharge for normal


depth computations (standard horseshoe cross
section)

r = nQ /(b8/3 S0 ) Dimensionless discharge for normal


depth computations
(rectangular cross section)

t = nQ /(b8/3 S0 ) Dimensionless discharge for normal


depth computations (trapezoidal cross section)
cc = yc /D Dimensionless critical depth (circular cross section)
ch = yc /H Dimensionless critical depth (standard horseshoe cross section)
ct = zyc /b Dimensionless critical depth (trapezoidal cross
section)
nc = yn /D Dimensionless normal depth (circular cross
section)
nh = yn /H Dimensionless normal depth (standard horseshoe cross section)
nr = yn /b Dimensionless normal depth (rectangular cross
section)
nt = yn /b Dimensionless normal depth (trapezoidal cross
section)
Subscripts
c and n Denote critical and uniform flow conditions, respectively
c
Denotes circular cross section
h
Denotes standard horseshoe cross section
r
Denotes rectangular cross section
t
Denote trapezoidal cross section
I and II Denote Type I and Type II, respectively

Fig. 1. Cross section of a trapezoidal channel.

Fig. 2. Cross section of a partially filled pipe channel.

2.2. Circular channels


For a partially filled circular channel section, shown in Fig. 2, the
geometric elements are as follows
A=

D2

( sin )

8
1
P = D
2

T = D sin

(4)
(5)

= 2 cos1 (1 2)

(6)
(7)

where D is channel diameter, is water surface angle in radians,


and = y/D.
2.3. Standard horseshoe channels
A horseshoe cross section, shown in Fig. 3, consists of four arc
segments: a top arc (BC ) with radius r, a bottom arc (AD) with
radius R, and two lateral arcs (AB and DC ) with the same radius
R but with different circular centers [14]. Horseshoe cross sections
can be classified using the characteristic parameter t = R/r. For
t = 3 and t = 2, the cross sections are called standard Type I and
Type II horseshoe cross sections, respectively [14]. Note that when
t = 1, the horseshoe cross section becomes circular.
Fig. 3 shows a general horseshoe cross section and the
corresponding geometric symbols for three ranges of water depths
(y): (a) 0 y e, (b) e y r, and (c) r y 2r, where e is
the height of the bottom arc, which is given by e = 0.12917r ( =
0.294515 rad) for Type I, and e = 0.17712r ( = 0.424031 rad)
for Type II cross sections, respectively [15].
The required formulae for computing the geometric elements
of standard horseshoe cross sections (wetted perimeter and flow
area) for the three zones of flow depth were presented in [15].
These formulae are presented in Table 1. In addition, formulae for
computing the width of the channel at the water surface, which
are needed for developing the critical depth equation, are derived
in this study and are presented in the table.

A.R. Vatankhah, S.M. Easa / Flow Measurement and Instrumentation 22 (2011) 4349

45

Fig. 3. Horseshoe cross section and its geometric symbols for three zones of flow depth: (a) 0 y e; (b) e y r; and (c) r y 2r = H.
Table 1
Formulae for computing geometric elements for three zones of flow depth of a horseshoe cross section.
Zones of flow depth
0 y e or 0

= cos1
A=

1 2
t H
4

t 2
t

1
2

ey

= sin1

sin(2)

P = tH
T = tH sin()

A=

or 0

H
2

1 2 2
t H
4

t 2
t

H
2

1
2

sin(2) +

2(t 1)
t

sin(2)

y H or 0

= 2 cos1 (2 1)

A = 14 H 2 Ct 2 + 12 ( + sin())

P = tH (2 )
T = H [1 t + t cos()]

P = H2 (4t + )

T = H sin 2

Note: C = 2 + 1 sin(2) cos(2), H = 2r = height of the tunnel and = y/H.

3. Computation of critical depth (governing equation)


The critical flow condition in an open channel is described by
the following relationship [16]

Q 2 T
=1
g cos( ) A3

(8)

where is the energy correction factor, sin( ) is the longitudinal


slope of the channel, and g is gravitational acceleration. Most of
the channels have slopes smaller than 1/100, thus it is reasonable
to assume cos( ) 1 for these channels. Using Eq. (8), explicit
equations for the critical depth are derived next for trapezoidal,
circular, and horseshoe channels.

about 2.5%. In such a case, the advantage of the presented equations


over his previous regression-based equation is rather questionable.
Based on the infinite series presented by Swamee and Rathie [7],
Srivastava [10] found that a fitted series would be more accurate
than a truncated one.
To further develop an improved explicit solution in the current
study, Eq. (9) is first numerically inverted using the curve fitting
method as follows
1/3

ct = t (1 + atb )c
(10)
where a, b, and c are coefficients. To determine these coefficients,
the percentage error (PE) of the dimensionless critical depth, ct , is
expressed as follows

PE =

3.1. Trapezoidal channels


Substituting for A and T from Eqs. (1) and (3) into Eq. (8) yields
the following dimensionless form

t = F (ct ) =

ct3 (1 + ct )3
1 + 2ct

(9)

where t and ct are dimensionless variables with t = z 3 Q 2 /


[gb5 cos( )] and ct = zyc /b, and the subscripts c and t denote
critical flow conditions and the trapezoidal section, respectively.
Using the curve fitting method, Swamee [4] obtained an explicit
solution for Eq. (9). The maximum error of that solution is less
than 2.2% in the practical range of 0 ct 3. Vatankhah and
Kouchakzadeh [5] improved Swamees solution by developing a
similar equation that has a maximum error less than 0.28% in
the practical range of 0 ct 15. Using the fixed point iteration method, Wang [6] obtained a very accurate but complicated
solution for the critical depth of trapezoidal sections. Subsequently, Swamee and Rathie [7] obtained analytical solutions of
the critical depth for trapezoidal sections in the form of converging
infinite series, based on Lagranges inversion theorem. When the
number of terms in the proposed series is limited to 45 for practical cases, the maximum relative error involved in the solution is

ct F (ct )1/3 (1 + aF (ct )b )c


ct

100

(11)

in which the dimensionless function F (ct ) is determined using


the geometry of the trapezoidal cross section according to Eq. (9).
Then, the sum of the squares of the PE values is minimized
as an objective function using the Solver toolbox of Microsoft
Excel. The resulting explicit equation in the practical range of the
dimensionless normal depth, 0 ct 3, is given by
1/3

ct = t (1 + 1.1524t0.347 )0.339
(proposed-trapezoidal channels).

(12)

The maximum error involved in Eq. (12) is less than 0.06% in


the practical range of 0 ct 3. A summary of the existing
explicit solutions of the critical depth for trapezoidal channels and
their maximum errors is shown in Table 2. As noted, the proposed
solution is preferable to other solutions in terms of both accuracy
and simplicity.
3.2. Circular channels
Substituting for A and T from Eqs. (4) and (6) into Eq. (8) yields
the following dimensionless form

c = h1 (cc ) =

(c sin c )3
83 sin

c
2

(13)

46

A.R. Vatankhah, S.M. Easa / Flow Measurement and Instrumentation 22 (2011) 4349

Table 2
Summary of proposed and existing explicit equations for critical depth in trapezoidal and circular channels.
Equation reference

Proposed formulae

Application range

Maximum relative error (%)

ct = (t0.7 + 0.747t0.42 )0.476


[

1/2 ]
1/3
1/3
1/3
1/3
ct = 12 1 + 1 + 4t {1 + 4t [1 + 4t (1 + 4t )1/5 ]1/6 }1/6
[
]

0.4P 1/P
1/3
1/3
1 + t / 2
ct = t
P = 2.129t0.01565

0 ct 3

2.20

None

0.015

0 ct 15

0.28

0 ct 0.347

0.021

ct > 0.347

0.014

0 ct 3

0.06

0.1 cc 0.85
0.02 cc 1

5.8 (D 0.25 m)
1.46

0.9584c0.25
(1+0.0106c0.26 0.0132c1.863 )10.022

0 cc 0.92

0.25

cc = (1 + 13.6c2.1135 13c2.1 )0.1156

0.01 cc 1

0.27

Trapezoidal channels
Swamee [4]
Wang [6]
Vatankhah and
Kouchakzadeh [5]

1/3
t

ct =

Srivastava [10]

0.33662

2/3
t

+ 0.13529t1.075

ct1 = (2t )0.2 + 0.37913t0.4 + 0.071776t0.6 0.02409t0.75


1/3
t

ct =

Proposed

0.347 0.339

(1 + 1.1524t

Circular channels
Straub [8]
Swamee [4]

cc = 1.01 D0.01 c0.25


cc = (1 + 0.77c3 )0.085

Vatankhah and
Bijankhan [9]
Proposed

cc =

There are no explicit equations available in the literature for the horseshoe cross section, and the proposed equations are presented in the paper.

where c and cc are dimensionless variables with c = Q 2 /


[gD5 cos( )], c = 2 cos1 (1 2cc ), and cc = yc /D.
Straub [8] proposed a semi-empirical equation for the critical
depth in circular open channels as follows

cc = 1.01 D0.01 c0.25 .

(14)

As noted, Eq. (14) is dimensionally inhomogeneous. This equation


is not very accurate and its error depends on the diameter of the
channel. For example, for D 0.25 m, the maximum error is 5.8%.
Swamee [4] numerically inverted Eq. (13) for the dimensionless
critical depth, cc , using the curve fitting method as follows

cc = (0.77c + 1)
3

0.085

0.9584c0.25

(1 + 0.0106

0.26
c

0.0132c1.863 )10.022

(16)

The maximum error of Eq. (16) is less than 0.25% in the practical
range of 0 cc 0.92.
In the current study, the following regression-based equation is
proposed for computing the critical depth

cc = (13.6c2.1135 13c2.1 + 1)0.1156


(proposed-circular channels).

(17)

The maximum error of Eq. (17) is less than 0.27% in the practical
range of 0.01 cc 1. Table 2 presents a summary of the proposed and existing explicit equations for circular cross sections.
Clearly, the proposed solution offers both simplicity and accuracy
compared with other solutions.
3.3. Standard horseshoe channels

A3ch
H 5 Tch

chII

(1.7h10.7 + 0.000006h8.877 )0.02799


(1 + 1.6h2.4 )0.11

(proposed-standard Type I horseshoe channels)


(3.37h12.74 + 0.000053h10.954 )0.02275
=
(1 + 1.07h2.15 )0.14
(proposed-standard Type II horseshoe channels)

(19)

(20)

where the subscripts I and II denote Type I and Type II, respectively.
The maximum error of Eq. (19) is less than 0.65% in the practical
range of 0.01 ch 0.988 and that of Eq. (20) is less than 0.55%
in the practical range of 0.01 ch 0.99.
The proposed explicit equations of the dimensionless critical
depth for trapezoidal, circular, horseshoe channels, Eqs. (12), (17),
(19) and (20), are depicted graphically in Fig. 4. The actual data used
for estimating these equations using regression are also shown. The
(almost) perfect match between the actual and proposed critical
depths is evident.
4. Computation of normal depth (governing equation)
The uniform flow condition in an open channel is described by
the following Mannings formula [17]
Q =

S0 A5/3

(21)
n P 2/3
in which is the unit conversion constant, 1.0 (SI), 1.49 (CU), S0 is
the longitudinal slope of the channel, and n is Mannings roughness
coefficient.
4.1. Trapezoidal channels

Substituting for A and T from Table 1 into Eq. (8) yields the
following dimensionless form

h = h2 (ch ) =

chI =

(15)

The maximum error of Eq. (15) is less than 1.46% in the practical
range of 0.02 cc 1.
Recently, Vatankhah and Bijankhan [9] used the curve fitting
method to obtain a more accurate equation for the critical depth of
a circular channel by inverting Eq. (13) as follows

cc =

As previously mentioned, there is currently no explicit equation


for calculating the critical depth of horseshoe channels. Therefore,
using the curve fitting method the following explicit equations
were developed for computing the critical depth for standard
horseshoe channels

(18)

in which h and ch are dimensionless variables with h = Q 2 /


[gH 5 cos( )], ch = yc /H , H is height of the conduit, and the subscript h denotes the standard horseshoe cross section.

Substituting for A and P from Eqs. (1) and (2) into Eq. (21), the
Mannings equation becomes
5/3

nt (1 + z nt )5/3
(22)

(1 + 2nt 1 + z 2 )2/3

where nt = yn /b, yn is the normal depth, t = nQ /(b8/3 S0 )


t =

and the subscript n denotes normal flow conditions.

A.R. Vatankhah, S.M. Easa / Flow Measurement and Instrumentation 22 (2011) 4349

47

The maximum error involved in Eq. (26) is less than 0.06% in the
range of 0 nr 100. Note that from an engineers viewpoint,
it would be more useful to have a simple and reasonably accurate
expression for the practical range rather than a complicated and
more accurate expression for a much wider range. Eq. (26) is more
accurate over a larger range, but Eq. (25) is simpler to use, less
computationally intensive, and reasonably accurate. It is applicable
for the entire practical range of nr .
4.2. Circular channels
Substituting for A and P from Eqs. (4) and (5) into Eq. (21) the
following equation is obtained

c = h3 (nc ) =

(n sin n )5/3
2/3

213/3 n

(27)

in which c = nQ /(D8/3 S0 ), n = 2 cos1 (1 2nc ) and


nc = yn /D.
Using the curve fitting method, the following regression-based
equation is proposed for computing the normal depth for circular
channels,
(0.55c1.1 14.55c4.136 +0.4645)

nc = 1.025c

(proposed-circular channels).

Fig. 4. Comparison of actual and proposed dimensionless critical depth: (a)


trapezoidal channels and (b) circular and standard horseshoe channels.

Swamee and Rathie [11] proposed two infinite series solutions


that help in evaluating the normal depth for trapezoidal cross
sections. When two different solutions are given for a cross section,
a range of applicability should be determined. However, the choice
of either solution for the most widely used nt range depends
on the nt -value and the side slope of the channel, and this fact
complicates their applications in practice [12].
To overcome this limitation, an explicit equation is developed
here for the normal depth using the fixed point iteration scheme.
Rearranging Eq. (22), then

nt

(proposed-trapezoidal channels)

(23)

in which nt0 is an initial guess for the normal depth which is


proposed as follows

nt0 =

1 + 0.856

3/5
t

(1 + z 1.263 )(1 0.0585z


3/5

+ 1.945z

(24)

nr = r3/5 (1 + 2r3/5 + 1.712r6/5 )2/5


(proposed-rectangular channels)
(25)

8/3
where nr = yn /b, r = nQ /(b
S0 ). The maximum error
involved in Eq. (25) is less than 0.08% in the practical range, 0
nr 3.
In comparison, Srivastava [10] proposed an expression for the
normal depth of rectangular channels, based on truncation of the
iterative algorithm and the curve fitting method, as follows

nr = r

1 + 2.404r0.6321 (1 + 2.030r0.9363 )

0.3929 2/5

Substituting for A and P from Table 1 into Eq. (21) yields the
following dimensionless form
5/3

h = h4 (nh ) =

Anh

(29)

2/3

H 8/3 Pnh

The maximum error involved in Eq. (23) is less than 0.7% in the
practical range of 0 z 3 and 0 nt 1.
For rectangular channels (z = 0), Eq. (21) becomes

3/5

The maximum error involved in Eq. (28) is less than 0.35%


in the practical range (0.005 nc 0.82). There is a rationale
for selecting the upper value of the preceding practical range.
Channels with a closing top-width can be designated as channels
of the second kind [16]. In these channels, the discharge is not
a single-value function of the normal depth beyond a certain
range of the flow depth. In this range, a small disturbance in the
water surface may cause it to seek alternate normal depths, thus
contributing to the instability of the water surface. In practice, it
is usual to restrict the flow depth to be below this range to avoid
this double normal depth phenomenon. For circular channels, this
instability range occurs when the flow depth is greater than 0.82
of the channel height. Thus, in practice it is sufficient to restrict the
flow depth to be less than this ratio.
4.3. Standard horseshoe channels

3/5
t (1 + 2nt0 1 + z 2 )2/5
=
1 + z nt0

3/5
t

(28)

(26)

in which h = nQ /(H 8/3 S0 ), nh = yn /H , H is height of the


conduit, and the subscript h denotes a standard horseshoe cross
section. It is important to note that for nh > 0.82, the discharge is
not a single-valued function of the normal depth for the standard
Type I and Type II horseshoe cross sections. Thus, in practice it
is sufficient to restrict the depth of flow to be less than 0.82 of
the channel height in both types. In addition, it is not necessary
to compute the normal depth for very small water depth [13].
Therefore, the minimum value of the dimensionless normal depth
is suggested as nh = 0.05.
Liu et al. [15] developed general formulae for the direct
computation of the normal depth for all types of horseshoe cross
sections (for three zones) as follows,

nh = (0.057t 2 0.3738t + 1.3849)h(0.00216t


[for Q Q (e)]

2 0.0126t +0.4806)

(30)

48

A.R. Vatankhah, S.M. Easa / Flow Measurement and Instrumentation 22 (2011) 4349

Table 3
Summary of proposed and existing explicit equations for the normal depth in rectangular and standard horseshoe channels.
Equation reference

Proposed formulae

Application range

Maximum relative error (%)

0 nr 100

0.06

0 nr 3

0.08

Rectangular channels
3/5

1 + 2.404r0.6321 (1 + 2.030r0.9363 )0.3929

Srivastava [10]

nr = r

Proposed

nr = r (1 + 2r

3/5

3/5

2/5

6/5

+ 1.712r )2/5

Standard horseshoe channels

nh = (0.057t 2 0.3738t + 1.3849)h(0.00216t


Liu et al. [15] (Types I and II)

0 < Q Q (e) and 0.05 nh

0.10

nh = (0.1997t 2 + 1.124t 0.9243)h1.2 + (0.0956t 2


0.5401t + 1.8479)h0.6 + (0.002t 2 0.0166t + 0.0524)
nh = (0.0111t 2 + 0.0501t + 2.8232)h1.2 + (0.0381t 2
0.2063t 0.3739)h0.6 + (0.0001t 2 0.0063t + 0.3789)

Q (e) Q Q (0.5H )

0.35

Q (0.5H ) Q and nh 0.82

1.35

nhI = 0.5h

0.05 nh 0.82

0.63

0.05 nh 0.82

0.63

(1.722h0.892 17.6h4.2 +0.3956)

Proposed (Type I)

(1.5h1.14 182.46h6.9 +0.4414)

nhII = 0.75h

Proposed (Type II)

2 0.0126t +0.4806)

There are no proper explicit equations available in the literature for circular and trapezoidal cross sections, and the proposed equations are presented in the paper.

nh = (0.1997t 2 + 1.124t 0.9243)h1.2


+ (0.0956t 2 0.5401t + 1.8479)h0.6
+ (0.002t 2 0.0166t + 0.0524)
[for Q (e) Q Q (0.5H )]

(31)

1.2
h
0.6
h

nh = (0.0111t + 0.0501t + 2.8232)


+ (0.0381t 2 0.2063t 0.3739)
+ (0.0001t 2 0.0063t + 0.3789)
[for Q (0.5H ) Q ].
2

(32)

To apply these formulae, the limits of the discharge Q should be


determined. Then, the equation corresponding to the flow range is
used. In practice, however, it is preferable to have a single normal
depth equation for the entire range of flow (0.05 nh 0.82).
Therefore, the following regression-based equations were developed for computing the normal depth of horseshoe channels,
(1.722 0.892 17.6 4.2 +0.3956)

h
h
nhI = 0.5h
(proposed-standard Type I horseshoe channels)

(33)

(1.5h1.14 182.46h6.9 +0.4414)

nhII = 0.75h
(proposed-standard Type II horseshoe channels)

(34)

where the subscripts I and II denote Type I and Type II, respectively.
The proposed approximations of the dimensionless normal depth
of Eqs. (33) and (34) work very well over the entire practical range
of depth (0.05 nh 0.82) with a maximum percentage error
less than 0.63%. Table 3, presents a summary of the proposed and
existing explicit equations for rectangular and Standard Horseshoe
cross sections. Clearly, the proposed solution offers both simplicity
and accuracy compared with other solutions.
The proposed explicit equations of the dimensionless normal
depth for trapezoidal, circular, and horseshoe channels, Eqs. (23),
(28), (33) and (34), are depicted graphically in Fig. 5, along with the
actual data. Again, the proposed critical depth equations (almost)
perfectly match actual data.
5. Conclusions
The critical and normal depths are important elements in
the design, operation, and maintenance of open channels. The
calculation of these elements is traditionally performed using trial
procedures, numerical/graphical methods, or explicit regressionbased equations. Explicit solutions for the critical depth are
available in the literature for trapezoidal and circular channels,
but not for horseshoe channels. For the normal depth, explicit

Fig. 5. Comparison of actual and proposed dimensionless normal depth: (a)


trapezoidal channels and (b) circular and standard horseshoe channels.

equations are available for only rectangular and horseshoe


channels, but not for trapezoidal or circular channels.
This paper has presented explicit solutions of these elements
for three types of channels: trapezoidal, circular, and horseshoe.
For the critical depth, new explicit equations for horseshoe
channels and improved explicit equations for trapezoidal and
circular channels are presented. For the normal depth, new explicit
equations for circular and trapezoidal channels and improved
solutions for horseshoe channels are presented.
Dimensionless variables of the governing equations (Eqs. (8)
and (21)) are very powerful tools for developing general regression
equations without the need for using actual data. The explicit
equations were developed using regression analysis based on the
practical ranges of the corresponding dimensionless variables.

A.R. Vatankhah, S.M. Easa / Flow Measurement and Instrumentation 22 (2011) 4349

The functional form of the regression equation is determined


by evaluating different mathematical functions and selecting
the one that minimizes the maximum relative error. This error
also depends on the number of data points selected within the
ranges of the dimensionless variables. In addition, visualization
of the response surface of the governing equation was found to
greatly help in selecting the appropriate mathematical functions
for evaluation. These factors and the skills of the analyst could
result in different regression equations for the same dimensionless
variables.
The maximum relative error of the proposed regression
equations is less than 1%. The proposed equations exhibit both
simplicity and accuracy. It is hoped that the efficient computational
tools presented in this paper will be useful in the design and
analysis of open channels.

[6]
[7]
[8]
[9]

[10]

[11]
[12]

[13]

References
[14]
[1] Raikar RV, Shiva Reddy MS, Vishwanadh GK. Normal and critical depth
computations for egg-shaped conduit sections. Flow Measurement and
Instrumentation 2010;21(3):36772.
[2] Jain SC. Open channel flow. New York: John Wiley and Sons; 2001.
[3] Chaudhry MH. Open-channel flow. New York: Springer; 2006.
[4] Swamee PK. Critical depth equations for irrigation canals. Journal of Irrigation
and Drainage Engineering 1993;119(2):4009.
[5] Vatankhah AR, Kouchakzadeh S. Discussion of exact equations for critical

[15]

[16]
[17]

49

depth in a trapezoidal canal. Journal of Irrigation and Drainage Engineering


2007;133(5):508.
Wang Z. Formula for calculating critical depth of trapezoidal open channel.
Journal of Hydraulic Engineering 1998;124(1):901.
Swamee PK, Rathie PN. Exact equations for critical depth in a trapezoidal canal.
Journal of Irrigation and Drainage Engineering 2005;131(5):4746.
Straub WO. A quick and easy way to calculate critical and conjugate depths in
circular open channels. Civil Engineering 1978;701.
Vatankhah AR, Bijankhan M. Choke-free flow in circular and ovoidal channels.
Proceedings of the Institution of Civil Engineers, Water Management 2010;
163(4):20715.
Srivastava R. Discussion of exact solutions for the normal depth problem by
Prabhata K. Swamee and Pushpa N. Rathie. Journal of Hydraulic Research 2006;
44(3):4278.
Swamee PK, Rathie PN. Exact solutions for normal depth problem. Journal of
Hydraulic Research 2004;42(5):5417.
Kouchakzadeh S, Vatankhah AR. Discussion of exact solutions for normal
depth problem by Prabhata K. Swamee and Pushpa N. Rathie. Journal of
Hydraulic Research 2007;45(4):56771.
Zhao YF, Lu Q, Zhang KD. An approximate formula for calculating water depth
of uniform flow in circular cross section. Journal of Northwest Agriculture and
Forestry (Natural Science Edition) 2008;36(5):2258.
Lv HX, Xin QC, Hua LF. Calculation on normal depth of horseshoe cross section
by iterative method. Journal of Yangtze River Scientific Research Institute
2001;18(3):710.
Liu J, Wang Z, Fang X. Iterative formulas and estimation formulas for
computing normal depth of horseshoe cross-section tunnel. Journal of
Irrigation and Drainage Engineering 2010;136(11):78690.
Subramanya K. Flow in open channels. New Delhi (India): Tata McGraw-Hill;
1986.
Chow VT. Open-channel hydraulics. New York: McGraw-Hill; 1959.

You might also like