You are on page 1of 11

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

net/publication/264496985

Fully developed turbulent flow pressure drop in circular and noncircular ducts

Article  in  Journal of Fluids Engineering · June 2012

CITATIONS READS

5 2,514

1 author:

Zhipeng Duan
Beijing Jiaotong University
71 PUBLICATIONS   708 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Zhipeng Duan on 19 September 2014.

The user has requested enhancement of the downloaded file.


Pressure Drop for Fully
Developed Turbulent Flow
in Circular and Noncircular
Zhipeng Duan
e-mail: zpduan@uwaterloo.ca Ducts
M. M. Yovanovich The objective of this paper is to furnish the engineer with a simple and convenient means
of estimating frictional pressure drop in ducts and the original physical behavior can be
Department of Mechanical clearly reflected. Fully developed turbulent flow frictional pressure drop in noncircular
and Mechatronics Engineering, ducts is examined. Simple models are proposed to predict the frictional pressure drop in
University of Waterloo, smooth and rough noncircular channels. Through the selection of a novel characteristic
Waterloo, Ontario, length scale, the square root of the cross-sectional area, the effect of duct shape has been
N2L 3G1, Canada minimized. The proposed models have an accuracy of 6% for most common duct shapes
of engineering practice and can be used to predict pressure drop of fully developed tur-
bulent flow in noncircular ducts. It is found that the hydraulic diameter is not the appro-
Y. S. Muzychka priate length scale to use in defining the Reynolds number to ensure similarity between
Faculty of Engineering and Applied Science, the circular and noncircular ducts. By using the Reynolds number based on the square
Memorial University of Newfoundland, root of the cross-sectional area, it is demonstrated that the circular tube relations may be
St. John’s, Newfoundland, applied to noncircular ducts eliminating large errors in estimation of pressure drop. The
A1B 3X5, Canada square root of the cross-sectional area is an appropriate characteristic dimension appli-
cable to most duct geometries. The dimensionless mean wall shear stress is a desirable
dimensionless parameter to describe fluid flow physical behavior so that fluid flow prob-
lems can be solved in the simple and direct manner. The dimensionless mean wall shear
stress is presented graphically and appears more general and reasonable to reflect the fluid
flow physical behavior than the traditional Moody diagram. [DOI: 10.1115/1.4006861]

Keywords: turbulent flow, pressure drop, friction factor, noncircular, length scale,
square root of flow area, hydraulic diameter

1 Introduction diameter does not necessarily yield equivalent circular tube fric-
tion factors. With the objective of improving turbulent flow pre-
Since the milestone works of Prandtl, von Karman, Nikuradse,
dictions, several other modified length scales have been proposed
and Blasius for circular tubes, it has been common practice in the
to replace the hydraulic diameter. Hodge [8] introduced the con-
fields of fluid mechanics to utilize the hydraulic or equivalent di-
cept of a hydraulically effective zone for calculating the fully
ameter in modeling turbulent flow pressure drop in noncircular
developed friction factors for isosceles triangular ducts. Ahmed
ducts. The standard procedure has been to calculate the hydraulic
and Brundrett [9] introduced a characteristic length suitable for
diameter and then use this as the characteristic length dimension
equilateral triangular and square ducts. Bandopadhayay and
in the Reynolds number to obtain circular tube equivalency.
Ambrose [10] introduced a generalized length dimension, which
For circular tubes, the choice of the length scale in the defini-
may be viewed as an average distance of the duct boundary from
tion of Reynolds number is obvious. However, for noncircular
the point of the maximum velocity in the ducts. The foregoing
ducts, the question always arises of what to use as the correct
examples illustrate how different and complex the proposed
length scale. Although it is customary to use the hydraulic diame-
length dimensions are. These substitute length scales provide
ter, for many years, this choice has been widely believed to be
improved friction coefficient for special noncircular ducts only
incorrect. White [1], Mironer [2], and Denn [3] (to mention a few)
and there is no general characteristic length scale that provides
question whether the hydraulic diameter is appropriate, but they
satisfactory predictions for all noncircular ducts.
stop short of recommending an alternative length scale. The hy-
Some researchers proposed to relate the turbulent friction coef-
draulic diameter of a given noncircular flow area is not numeri-
ficient to the laminar flow dimensionless parameters. Gunn and
cally equal to the diameter of a circular tube of the same area. The
Darling [11], Rehme [12], Malak et al. [13] proposed a few for-
choice of a length scale with noncircular ducts has been a peren-
mulas to calculate fully developed turbulent flow friction factors
nial and contentious issue. The use of hydraulic diameter for non-
using laminar flow friction factor Reynolds number product. Jones
circular ducts such as triangular ducts leads to unacceptably large
[14] examined frictional pressure drop in rectangular ducts and
errors, on the order of 30%, in turbulent flow friction factors
concluded that the hydraulic diameter is not the proper length
determined from the circular duct correlations. Large deviations
scale to define the Reynolds number in order to ensure similarity
were observed in narrow isosceles triangular ducts from the use of
between circular and noncircular ducts. For this reason, Jones
the hydraulic diameter rule by Eckert and Irvine [4,5], Carlson
introduced the laminar equivalent diameter. Later, Jones and
and Irvine [6], and Nan and Dou [7]. It indicates that the hydraulic
Leung [15] applied the concept of laminar equivalent diameter to
concentric annular ducts. Obot [16] proposed the critical friction
Contributed by Fluids Engineering Division of ASME for publication in the JOUR-
method in order to reduce the friction data for noncircular ducts
NAL OF FLUIDS ENGINEERING. Manuscript received January 24, 2011; final manuscript on the circular tube relations. The critical friction method indi-
received April 18, 2012; published online June 11, 2012. Assoc. Editor: Chunill Hah. cated that the modified hydraulic diameter is related to the ratio of

Journal of Fluids Engineering Copyright V


C 2012 by ASME JUNE 2012, Vol. 134 / 061201-1

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
pffiffiffi
the critical Reynolds numbers for circular tube and noncircular A
ducts. He and Gotts [17] proposed the multichannel approach to
 ffiffiffi
sp A
¼ CpffiffiAffi (9)
P
predict friction factor by splitting a noncircular channel with sharp
pffiffiffi
corners into subchannels and isolating sharp corners from the Bejan [19] has also shown that the grouping P A is an impor-
main channel. tant geometric scaling factor.
The pressure drop is related to the mean wall shear stress or
2 Analysis of Internal Fluid Flow Friction friction factor and Reynolds number product in the following
2.1 Characteristic Length Scale. One of the most funda- manner:
mental problems in fluid dynamics is that of fluid flow in circular L ADp lw
 PL 
and noncircular channels under constant pressure gradient. Upon sL ¼ or Dp ¼ s (10)
l wPL
 L A L
obtaining the velocity distribution w(x, y) and mean velocity w, 
the friction factor and Reynolds number product may be defined The dimensionless mean wall shear stress has more physical
using the relations meaning and significance than the friction factor. There are certain
  common principles and laws that apply to the momentum and
A dp
 L energy transport processes. There are some examples that do point
sL P dz f ReL
sL ¼ ¼ ¼ (1) out that the analogies and similarities exist between the dimen-
lw
 lw 2 sionless mean wall shear stress (not the friction factor) and the
The factor 2 appears because the Fanning friction factor is dimensionless mean wall heat flux. It is well known that the
employed. If the Darcy friction factor is used, the factor becomes dimensionless mean wall heat flux (Nusselt number) can be
8. The Fanning friction factor will be used in this paper. The related to the dimensionless mean wall shear stress.
above grouping s is interpreted as the dimensionless mean wall Equation (10) is more appropriate and convenient to calculate
shear stress. The mean wall shear stress may also be related to the frictional pressure drop especially for laminar flow problems as s
pressure gradient by means of the force balance: is a constant. It is not necessary to obtain the value of friction fac-
tor f although it is customary to use Fanning or Darcy friction fac-
s PL ¼ DpA (2) tor to calculate pressure drop due to historical reasons.
In the fluid flow and heat transfer literature the convention is to
From this relation the mean wall shear stress is obtained: use the hydraulic diameter. For noncircular geometries, it is desir-
able to eliminate or reduce the effects of geometry such that the
Dp A
s¼ (3) general trends for all duct shapes may be easily modeled. It is bet-
L P ter to choose an appropriate characteristic length scale to nondi-
Using the method of scale analysis, we can examine the laminar mensionalize the fluid flow and heat transfer data.
momentum equation and consider the force balance between the Muzychka and Yovanovich [18] showed that the square root of
friction and pressure forces for a long channel: the cross-sectional area is more appropriate than the hydraulic di-
ameter for nondimensionalizing the laminar no-slip internal flow
w 1 Dp data. Duan and Muzychka [20,21] demonstrated that the same
2
 (4) conclusion can be extended to the more general slip flow regime.
L l L
Recently, Muzychka and Edge [22] reached the same conclusion
for non-Newtonian power law fluids. For laminar flow, it is well
Substituting the relation for s and the scale for w
 gives the follow- known that the use of circular tube correlations is not accurate for
ing scale for the dimensionless mean wall shear stress: different geometries. The fully developed friction factor and
Reynolds number product is a constant and the constant depends
A1
sL  (5) strongly on the geometry of the duct cross section. The friction
PL factor and Reynolds number product are independent of the Reyn-
olds number, unlike the turbulent flow case. Figures 1 and 2
which is purely geometric because it depends on the cross section [23,24] show that the square root of the cross-sectional area is a
area A, the perimeter P, and the arbitrary length scale L. more appropriate characteristic length scale than the hydraulic di-
For closure the geometric parameter CL is introduced so that ameter for nondimensionalizing the fully developed laminar flow
data. The theoretical and numerical data for some noncircular
A1
sL ¼ CL (6) ducts reported as f ReDh are plotted versus the effective aspect ra-
PL tio e in Fig. 1. Some data increase with increasing values of e
while other data decrease with increasing values of e. There is no
The geometric parameter CL is found to depend on the geometry general trend of the data. The definition of effective aspect ratio
of the cross section such as the shape, the aspect ratio, and the proposed by Muzychka and Yovanovich [18] and Duan and Yova-
choice of the length scale L. novich [23,24] is summarized in Table 1 for a number of geome-
The most frequently recommended length scale is the hydraulic tries. The aspect ratio for regular polygons (N  4) is unity. The
diameter defined as effective aspect ratio for most singly connected ducts is taken as
pffiffiffi the ratio of the maximum width to maximum length such that
4A pffiffiffi 4 A 0 < e < 1. For the trapezoid duct, double-trapezoid duct, triangle
L ¼ Dh ¼ ¼ A (7)
P P duct, and rhombic duct, simple expressions of effective aspect ra-
tio have been derived to relate the characteristic dimensions of the
For this length scale the dimensionless mean wall shear stress duct to a width to length ratio. A schematic diagram of the repre-
becomes sentative trapezoidal and double-trapezoidal cross sections is pro-
vided in Fig. 3. The next step in the comparisons is to convert all
sDh ¼
CDh
(8) data from f ReDh to f RepAffiffiffi and replot versus the effective aspect
4 ratio. When this is done as shown in Fig. 2, all data follow closely
a similar trend where the values decrease with increasing values
A novelplength
ffiffiffi scale proposed by Muzychka and Yovanovich [18] of e. It was found that the use of the hydraulic diameter in laminar
is L ¼ A for laminar flow. For this length scale the dimension- flow situations
pffiffiffi yields greater scatter in results as compared
pffiffiffi with
less mean wall shear stress becomes the use of A as a characteristic length scale. When A is used,

061201-2 / Vol. 134, JUNE 2012 Transactions of the ASME

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
the effect of duct shape is minimized, and all of the laminar flow
data can be predicted using a simple model based on the solution
for the rectangular duct Eq. (11). This means that the dimension-
less average wall shear stress can be made a weak function of duct
shape. It is clear that Eq. (11) characterizes the fully developed
laminar flow in noncircular ducts. The maximum deviation of
exact values occurs for very narrow noncircular ducts, which is
less than 6.5%. The difference is much smaller and within 3% for
most practical engineering configurations.

12
f RepffiffiAffi ¼   p  (11)
pffiffi 192e
eð1 þ eÞ 1  5 tanh
p 2e

pffiffiffi Flow Models. It must be expected that the su-


2.2 Turbulent
periority of A to Dh should be applicable to turbulent flow.
Although given the flatter velocity profiles found in turbulent duct
flows, the friction factor Reynolds number product is a compara-
tively weaker function of duct shape compared with laminar flow.
The flatter velocity profiles are similar to the velocity distributions
in the very near entrance of ducts where the friction factor Reyn-
olds number product is nearly independent of duct shape [24].
The similarities may offer insight as to why using the Reynolds
number based on the square root of the cross-sectional area, the
circular tube relations may be applied to noncircular ducts thereby
eliminating large errors in estimation of pressure drop.
Following Prandtl’s boundary layer theory, Blasius [25] made a
critical survey of the numerous experimental results and arranged
them in dimensionless form in accordance with Reynolds’s law of
similarity. Blasius’s original dimensionless analysis of the varia-
bles involved led to his adoption of the form Re as the most signif-
icant grouping terms upon which f should depend. He established
Fig. 1 Fully developed f ReDh for noncircular ducts
the following classical empirical equation:

f Re0:25
D ¼ 0:0791 (12)

which is valid for smooth pipes of circular cross section for Reyn-
olds numbers from 4000 to 105.
If the
pffiffiffiBlasius formula, Eq. (12), for circular tubes is recast
using A as the characteristic length scale in Re, the following
relationship is obtained:

pffiffiffi ¼ 0:0767
f Re0:25
A
or f RepffiffiAffi ¼ 0:0767Re0:75
pffiffiffi
A
(13)

It is seen that the values of the Blasius parameter are 0.0791 and
0.0767
pffiffiffi when the length scales in the Reynolds number are D and
A, respectively.
When the hydraulic diameter is employed for noncircular cross
sections, then the Blasius parameter is no longer constant and it
 pffiffiffi
depends on the geometric characteristic aspect ratio P 4 A that
is related to the shape of the cross section. The form of the Blasius
equation based on hydraulic diameter is expressed as

0:0767
f Re0:25
Dh ¼   (14)
P 0:25
pffiffiffi
4 A
 pffiffiffi
The important geometric characteristic aspect ratio P 4 A relates
the hydraulic diameter to the square root of the cross-sectional
area and will be discussed later.
Prandtl [26], von Karman [27], and Nikuradse [28] contributed
to the development of an accurate relationship for fully developed
turbulent flow in smooth circular pipes. Employing the universal
velocity distribution law, Prandtl [26], von Karman [27], and
Nikuradse [28] proposed the following relation for fully devel-
oped turbulent flow in smooth pipes and it is fair to call it the
Fig. 2 Fully developed f RepffiffiAffi for noncircular ducts PKN equation after Prandtl, von Karman, and Nikuradse:

Journal of Fluids Engineering JUNE 2012, Vol. 134 / 061201-3

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 1 Definitions of effective aspect ratio

Geometry Effective aspect ratio

Regular polygons e¼1 N4


b
Simple singly-connected e¼
a
 
2b n a þ cn 1=n
Trapezoid e¼ þ n ¼ 1:7
aþc 2b
 n  n 1=
2b 2a n
Arbitrary triangle e¼ þ n ¼ 0:53
a b
2 0 1 31=
c n n
  1þ
6 aþc n B aC 7
e¼4 þ@ n ¼ 2:8
2a A 5
Double-trapezoid
2b
b
 n  n 1=
2b 2a n
Rhombus e¼ þ n ¼ 0:68
a b
1r
Annular sector e¼
uð1 þ rÞ

could be the roughness effects as seen in the Moody [32] chart


assuming the relative roughness is only 10 6. The PKN law does
not give f explicitly and thus requires an iterative solution. Cole-
brook [33], Filonenko [34], and Techo et al. [35] developed sev-
eral accurate explicit formulas; see Kakac et al. [36].
Similarly,
pffiffiffi if the PKN universal formula, Eq. (15), is recast
using A as the characteristic length scale in Re, the following
relationship is obtained:

1  pffiffiffi
pffiffiffi ¼ 4:0 log RepAffiffiffi f  0:1861 (17)
Fig. 3 Trapezoidal and double-trapezoidal ducts f

In the case of rough pipes, Colebrook and White [33,37] proposed


 the following equation by combining the relations developed for
1 pffiffiffiffiffi
pffiffiffiffiffi ¼ 2:0 log ReD 4f  0:8 (15) smooth and fully rough pipes:
4f  
1 D 18:7
pffiffiffiffiffi ¼ 1:74  2:0 log 2 þ pffiffiffiffiffi (18)
where the Darcy friction factor is four times the Fanning friction 4f D ReD 4f
factor. This is the PKN universal law of friction. It has been veri-
fied by Nikuradse’s experiments [29] up to a Reynolds number of This relation agrees with the two extremes of roughness and gives
3.4  106. The PKN universal equation agrees well with the Bla- values in good agreement with actual measurements on most
sius equation for Reynolds number up to 105. Deviations from the forms of commercial piping and usual pipe surfaces. This function
Blasius equation were noted for large Reynolds numbers. The ve- is asymptotic at one end to the smooth pipe line and at the other
locity distribution close to the wall followed essentially a power end to the horizontal lines of the fully rough pipe zone. By putting
law similar to Blasius, yet with an increasing exponent as the D=D!0, Eq. (18) reduces to Eq. (15) for smooth pipes and as
Reynolds number increases. Prandtl [30] also suggested that the Re!1 (very large Reynolds numbers), Eq. (18) is valid for fully
following equation is more accurate for large values of Reynolds rough pipes. Roughness plays an important role in channel flows
number: and the nature of surface roughness is still an active research area.
 pffiffiffiffiffi Afzal [38] proposed different expressions for various types of sur-
1 face roughness. Equation (18) requires an iterative process for the
pffiffiffiffiffi ¼ 1:95 log ReD 4f  0:55 (16)
4f calculation of f. Many explicit equations for calculating f in a
rough circular pipe have been developed by Wood [39]; Churchill
McKeon et al. [31] investigated the friction factor relationship for [40,41]; Swamee and Jain [42]; Chen [43]; Round [44]; Barr [45];
high Reynolds number turbulent pipe flow using two sets of data Zigrang and Sylvester [46]; Haaland [47]; Serghides [48]; Romeo
from the Princeton Superpipe. The range of Reynolds number is et al. [49]; and Avci and Karagoz [50], to mention but some.
from 3.1  104 to 3.5  107. McKeon et al. [31] indicated that the In traditional theory, the friction factor is independent of rela-
PKN law of friction was not accurate for high Reynolds numbers. tive roughness in the laminar region. The above conclusion is
The difference increases with Reynolds number: It is 1.7% at mainly based on the pioneering observation of Nikuradse’s sand
Re ¼ 106 and 3.2% at Re ¼ 108. McKeon et al. [31] slightly modi- grain roughness experiments. The sand roughness height was the
fied the constants of the PKN law from 2.0 and 0.8 to 1.93 and first successful attempt to measure roughness indirectly with a
0.537. One possible explanation for the difference could be the uniform sand layer glued onto a conduit surface. However when
entrance effect since the entrance effect increases with increased one looks at Nikuradse’s plot of friction factor versus Reynolds
Reynolds number. The entrance length could be about 40–100 number all laminar points lie above the theoretical friction factor
times the diameter at very high Reynolds numbers. Another possi- prediction. Furthermore, Nikuradse’s sand grain roughness distri-
ble explanation for the difference at very high Reynolds numbers bution is not dense and the roughness distribution density is low.

061201-4 / Vol. 134, JUNE 2012 Transactions of the ASME

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
In addition, the uncertainty observed by Nikuradse was approxi-
mately 3–5% in the turbulent region and it is definite that the
uncertainty in the laminar region is higher due to smaller pressure
drops given that manometers were used to determine pressure
drops. The laminar flow studies by Nikuradse and other early
investigators were conducted on the same experimental setups
used in turbulent studies. The focus of Nikuradse’s experiments
was largely on turbulent flow, which is more practically important
in hydraulics. The uncertainty in the laminar region greatly over-
shadows the experimental measurements. The effect of wall
roughness on the laminar flow is less well understood and further
experimental study with accurate instrumentation is invaluable in
helping to understand the role of roughness. The earlier conclu-
sion that roughness has no effect on laminar friction factor is
questionable. It is believed that the frictional pressure drop
depends on relative surface roughness, roughness shape, and dis-
tribution although the surface roughness effects in the laminar
region are not as so significant as in the turbulent region. Duan
and Muzychka [51–53] reviewed a vast amount of experimental
and numerical simulation results and analytically indicated that Fig. 4 Modified Blasius and PKN models for smooth pipes
the friction factor depends on the relative roughness of the walls
of channels in the laminar region and as such the relative rough-
ness should not be neglected for various types of surface rough-
ness. The roughness does not influence the friction factor in the It can be demonstrated that Eq. (22) reduces
pffiffiffi to an explicit equa-
laminar region only when relative roughness is small and rough- tion of Eq. (17) for smooth ducts as D A!0:
ness distribution is not dense.
As the friction factor in the fully rough regime is independent " !#2
6:115
of Re for circular pipes, it must be expected that this conclusion f RepffiffiAffi ¼ 3:6 log RepAffiffiffi (23)
should be applicable to noncircular ducts. The expectation that in RepAffiffiffi
the fully rough regime the law of frictional pressure drop should
be independent of the duct shape has been validated for two dif- The above simple and accurate relation can be used as an approxi-
ferent duct geometries: scalene triangle [54] and rectangle [55]. mation for the modified PKN equation. If greater accuracy is
The friction factor Reynolds number product is a weak function of desired, exact explicit equations for Colebrook–White equation
duct shape for high Reynolds number and large relative rough- and PKN equation are given in the Appendix.
ness. In the fully rough regime the law of resistancepshould
ffiffiffi be of a
nearly universal function for all ducts. Employing A as the char-
acteristic length scale, the Colebrook–White equation becomes 3 Discussion of Results
!
The modified Blasius and PKN models for smooth circular
1 D 8:286
pffiffiffi ¼ 3:48  4:0 log 1:772 pffiffiffi þ pffiffiffi pffiffiffi (19) cross section are presented as f RepffiffiAffi and they are compared with
f A Re A f the extensive data collected by Schlichting [56] as shown in Fig.
4. The laminar flow and transition from laminar to turbulent flow
The four relatively simple explicit equations proposed by Church- regimes are also shown. With over 120 years of research on the
ill [40], Swamee and Jain [42], Round [44], and Haaland [47] transition process since the classical experiment of Reynolds, we
have the following general form: can still present only limited results since the transition problem
     appears to be quite complicated and not well understood. The
1 D m B3 n
pffiffiffi ¼ B1 log þ (20) transition mechanism and a definitive picture of the transition pro-
f B2 D ReD cess is still an active and quite difficult research problem. Between
Reynolds numbers of 2000 and 3000 or 4000, the conditions
where B1, B2, B3 and exponents m and n are determined by accu- depend upon the initial turbulence due to such extraneous factors
rate experimental data. These explicit equations provide compara- as sudden changes in sections, obstructions, or a sharp-edged en-
ble accuracies in similar ranges of the Reynolds number and trance corner; and the conditions are probably also affected by
relative surface roughness. pressure waves initiating instability and roughness. In the transi-
The following Haaland [47] equation is presented in fluid tion region, the available experimental data are scattered. There is
mechanics texts and; therefore, it will be used to develop a general considerable uncertainty about the exactitude of the available data
explicit equation for turbulent flow in noncircular channels: from different sources. The laminar to turbulent transition region
  ! should be considered a metastable region. We propose a simple
1 D 1:11 6:9 curve fit model using the experimental data collected by Schlicht-
pffiffiffiffiffi ¼ 1:8 log þ (21)
4f 3:7D ReD ing [56] to approximately estimate f RepAffiffiffi in the transition regime:

f RepAffiffiffi ¼ 0:01381RepAffiffiffi  13:44 (24)


On the basis of simplicity and good agreement with the predic-
tions of Colebrook–White equation, the following general explicit
equation based on Haaland’s work [47] is proposed for practical The Reynolds number (RepffiffiAffi ) range is from 2000 to 3500. The
pressure drop calculations. The maximum deviation from the agreement between the experimental data and the model predic-
Colebrook–White model is 1.2% for Reynolds numbers from tion has a maximum difference of 7.7% and a rms difference of
4000 to 108. 4.3%. Due to the large variation in the available data, higher order
approximations offered no additional advantage.
"  10=9 !#2 The modified Blasius equation based on the hydraulic diameter
D 6:115
 pffiffiffi 0:25
f RepffiffiAffi ¼ 3:6 log 0:2047 pffiffiffi þ pffiffiffi RepAffiffiffi (22) includes the geometric characteristic parameter P 4 A that
A Re A
explains why the turbulent data obtained for noncircular ducts can

Journal of Fluids Engineering JUNE 2012, Vol. 134 / 061201-5

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 2 Numerical values for geometric characteristic aspect ratio and modified Blasius parameter

P  
pffiffiffi P 0:25
Cross section pffiffiffi
4 A 4 A

Circle 0.886 0.970


Square 1 1
Trapezoid 1.082 1.020
Rectangle 1: 3.5 1.203 1.047
Scalene right triangle (30 deg,60 deg,90 deg) 1.271 1.062
Scalene triangle with 2 two round corners (26 deg,40 deg,114 deg) 1.610 1.126
Circle with single rectangular Indentation 1.120 1.029
Circle with twin rectangular indentations 1.220 1.051
Isosceles triangle (apex Angle angle 2h ¼ 4.01 deg) 2.767 1.290
Isosceles triangle (2h ¼ 7.96 deg) 2.032 1.194
Isosceles triangle (2h ¼ 12.0 deg) 1.713 1.144
Isosceles triangle (2h ¼ 22.3 deg) 1.370 1.082
Isosceles triangle (2h ¼ 38.8 deg) 1.190 1.044
Isosceles triangle (2h ¼ 60.0 deg) 1.140 1.033
Isosceles triangle (2h ¼ 90.0 deg) 1.207 1.048
Isosceles triangle (2h ¼ 99.5 deg) 1.255 1.059
Isosceles triangle (2h ¼ 114.7 deg) 1.366 1.081
Isosceles triangle (2h ¼ 127.9 deg) 1.511 1.109
Isosceles triangle (2h ¼ 147.1 deg) 1.880 1.171

Fig. 5 Comparison with experimental data based on hydraulic Fig. 6 Comparison with experimental data based on square
diameter root of area

be predicted accurately when the hydraulic diameter is used. sure drop in isosceles triangular ducts is not adequately described
 pffiffiffi
 pffiffiffi 0:25
Table 2 shows the numerical values of P 4 A and P 4 A by the use of circular tube methods through the use of the hydrau-
for several noncircular cross sections that have been studied by lic diameter concept. The hydraulic diameter is not the correct
several researchers. The hydraulic diameter is used for turbulent length scale to obtain geometric similarity between round and tri-
flow because it is a convenient length scale and it leads to accepta- angular ducts.
ble accuracies for friction and pressure drop
pffiffiffibetween the circular Figure 6 shows the comparison between the proposed modified
and noncircular cross sections when P 4 A is not too large as Blasius model and the experimental data for isosceles triangular
seen in Table 2. ducts. It is found that the model predictions agree with all experi-
Carlson and Irvine [6] conducted experimental research on fully mental data within 5.8%. The use of the modified Blasius model
developed turbulent flow friction factors in isosceles triangular provides remarkable representation of the data.
ducts with apex angles 2h ¼ 4.01 deg, 7.96 deg, 12.0 deg, 22.3 All available air and water experimental data for noncircular
deg, and 38.8 deg. The use of the hydraulic diameter in the Bla- cross sections
pffiffiffi are compared with the modified Blasius model
sius equation yields friction factor values that are 6.2% to 27% based on A in Fig. 7. All data lay close to the simple model. The
higher than the experimental values as seen in Fig. 5. The fully exceptional point corresponds to the very narrow isosceles trian-
developed turbulent flow friction factors for an equilateral triangu- gle of 2h ¼ 4.01 deg. The maximum difference of about 5.8%
lar duct were measured by Schiller [57]. The experimental data occurs for the narrow isosceles triangular channels. Again, it is
were lower than those predicted by the Blasius equation using the seen that the square root of the cross-sectional area is the more
hydraulic diameter. Nan and Dou [7] measured fully developed appropriate length scale to obtain geometric similarity between
turbulent flow friction factors in large apex angle isosceles trian- circular and noncircular ducts.
gular ducts with 2h ¼ 99.5 deg, 114.7 deg, 127.9 deg, and 147.1 The simple model presented as f RepffiffiAffi , Eq. (13), can be applied
deg. It is seen from Fig. 5 that the friction factors are lower by to other common geometries. Figure 8 presents the comparison
20% than those values calculated from the Blasius equation for a between the proposed modified Blasius model of Eq. (13) and the
round tube using the hydraulic diameter. This is consistent with available experimental data of some common noncircular ducts.
the results of Carlson and Irvine [6]. These data show that pres- The model predictions are in agreement with all the available

061201-6 / Vol. 134, JUNE 2012 Transactions of the ASME

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Fig. 9 f ReDh comparison for experimental data
pffiffiffiffi
Fig. 7 Modified Blasius model based on A for noncircular
ducts
The corresponding experimental
pffiffiffi data in Fig. 8 are plotted in
Fig. 9 by switching from A to hydraulic diameter as the charac-
teristic length scale. The agreement between the experimental
data and the classical Blasius equation or the PKN equation pre-
diction has a maximum difference of 27%. It is seen that the use
of the hydraulic diameter in turbulent flow situations yields
greater
p ffiffiffi scatter in experimental data as compared with the use of
A as a characteristic length scale. The square root of the cross-
sectional area is a more appropriate characteristic dimension ap-
plicable to noncircular ducts eliminating large errors in estimation
of pressure drop.
Friction factors spanning the practical roughness were plotted
by Moody [32], constructed using the Colebrook–White equation.
The Moody chart appears in every fluid mechanics text. However,
it is obvious that the friction factor f is not a desirable dimension-
less parameter to describe fluid flow behavior as f is unnecessary
and superfluous especially when dealing with laminar flow prob-
lems. It is not necessary to obtain the value of f. Furthermore, it is
noted that the friction factor f decreases with an increase of Reyn-
olds numbers and approaches zero as Re!1 [59]; however, the
dimensionless average wall shear stress (s or fRe) should increase
with an increase in the Reynolds number and is a desirable dimen-
sionless parameter. It is seen that pressure drop increases with an
increase in the dimensionless average wall shear stress from Eq.
(10). The values of f RepffiffiAffi for turbulent flow in smooth and rough
circular and noncircular channels based on the modified Blasius
equation, the modified PKN equation, and the modified
Colebrook–White equationpffiffiffi are plotted in Fig. 10. The effect of
Fig. 8 f RepffiffiAffi comparison for experimental data
surface roughness D A is represented by five curves. Figure 10
is proposed to calculate the dimensionless mean wall shear stress
and pressure drop so that fluid flow problems can be solved in a
turbulent flow data within 6.0%. It is seen that the model is sur- simple and direct manner. This figure is more reasonable to reflect
prisingly successful. the original physical behavior. Figure 10 is more general and valid
Obot and Adu-Wusu [58] reported measured fully developed for circular and most noncircular ducts. It is seen that the dimen-
turbulent flow friction factors in a smooth scalene triangular duct sionless mean wall shear stress (s or fRe) increases with an
with two rounded corners. The experimental data are about 17% increase in the Reynolds number and relative roughness from this
lower than the Blasius equation, Eq. (12), predictions. However, figure. The smooth pipe data collected by Schlichting [56] and the
the experimental data are only 2.8% lower than the modified Bla- recent experimental data from McKeon et al. [31] on smooth pipes
sius model of Eq. (13) predictions as seen in Fig. 8. Good agree- are included in this figure.
ment between predictions employing the modified Blasius model It is clear that Eq. (13) is valid for smooth ducts and Reynolds
and experimental data is achieved. number up to 105 and general explicit relation Eq. (22)

Journal of Fluids Engineering JUNE 2012, Vol. 134 / 061201-7

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Moody chart. The present models takes advantage of the selection
of a more appropriate characteristic length scale square root of
flow area to develop simple models. The accuracy of the devel-
oped models was found to be within 6% for most common ducts.
Given that some of the deviation is due to experimental error
rather than to failure of the relations to represent the physical
behavior, both Eqs. (13) and (22) are probably reliable for most
noncircular channels. They are, therefore, recommended as more
convenient and accurate for design calculations of turbulent flow
frictional pressure drop in noncircular channels.
The general exact explicit equations of the Colebrook–White
equation and the PKN equation have been presented through the
use of the Lambert W function. The exact explicit equations pro-
vide a more accurate choice to calculate frictional pressure drop
when computer algebra systems are available.
As for turbulent flow, no models or graphical and tabulated
data exist for most noncircular geometries. These developed mod-
els can be used to predict pressure drop of turbulent flow in non-
circular channels such as rectangular, trapezoidal, double-
trapezoidal, triangular, rhombic, hexagonal, octagonal, circular
sector, circular segment, and annular sector ducts. It has been
demonstrated that the hydraulic diameter is inappropriate for
accurate estimation of turbulent flow pressure drop in noncircular
ducts. It has also been shown that the use of the square root of the
cross-sectional area to define a Reynolds number yields excellent
Fig. 10 Friction factor Reynolds number product for turbulent similarity between circular and noncircular ducts in laminar and
flow valid for circular and most common noncircular ducts turbulent flow. The influence of the shape of the cross section is
negligible for turbulent flow if the section is not too elongated.
characterizes the fully developed turbulent flow in most common The square root of the cross-sectional area is a more appropriate
noncircular channels. The maximum deviation of experimental length scale than the hydraulic diameter for noncircular ducts.
values is less than 6% for most common channels. The turbulent It may be pointed out that most of the above statements, formu-
flow frictional pressure drop may be predicted from Eq. (13) and las, and charts are valid only for “long” ducts. For short ducts, the
Eq. (22). rules controlling turbulence may be different. Also, the entrance
For very narrow smooth and rough noncircular geometries such effect cannot be neglected for relatively short ducts. The pressure
as those with effective aspect ratios less than 0.1, there are no reli- drop in short ducts will be investigated by the authors in the future
able experimental data to validate the proposed models. Based on work.
the available limited experimental data in the literature, it is noted
that different results have been reported in the previous investiga- Acknowledgment
tions with identical or similar small aspect ratios. Some experi-
mental results from different sources are even contradictory. The authors acknowledge the anonymous reviewers who made
There is uncertainty about the available data from various sources. valuable suggestions for improvements. The authors also would
It is quite possible that the effect of aspect ratio becomes signifi- like to thank Professor Richard Culham for his support to the
cant when the shape of the cross section is very elongated. The work.
authors presently propose the following relation for very narrow
smooth noncircular ducts:
Nomenclature
  A ¼ flow area, m2
ffiffiffi ¼ C pP 0:25
0:25
f Rep ffiffiffi (25) a ¼ major semiaxis of ellipse or rectangle, m
A
4 A a ¼ base (maximum) width of a trapezoidal, triangular,
double-trapezoidal, or rhombic duct, m
The value of C is a weak function of the small effective aspect ra- b ¼ minor semiaxis of ellipse or rectangle, m
tio and close to 0.085 based on a survey of the available limited b ¼ height of a trapezoidal, triangular, double-trapezoidal, or
data collected by Jones [14] for narrow rectangular ducts and by rhombic duct, m
Jones and Leung [15] for large radii ratio concentric annuli. It is c ¼ short side of a trapezoidal or double-trapezoidal duct, m
clear that the effect of small aspect ratio needs to be investigated D ¼ diameter of circular duct, m
in detail in future work. Dh ¼ hydraulic diameter ¼ 4A/P,
m
f ¼ Fanning friction factor ¼ s= 12qw
2
L ¼ channel length, m
4 Conclusion L ¼ arbitrary length scale of cross section, m
Fully developed turbulent flow in noncircular channels was N ¼ number of sides of a polygon
investigated. This paper was intended for applications to normal n ¼ correlation parameter
conditions of engineering practice. Simple models were developed P ¼ total wetted perimeter, m
P ffiffiffi
for predicting the dimensionless average wall shear stress or the p ¼ slenderness parameter
4 A 
friction factor Reynolds number product in noncircular channels Dp ¼ pressure drop, N m2
for turbulent flow. The dimensionless mean wall shear stress (s ) Re ¼ Reynolds number ¼ wL= 
is a more desirable dimensionless parameter to describe fluid flow r ¼ dimensionless radius ratio ¼ ri/ro
behavior so that fluid flow problems can be solved in a simple and ri ¼ inner radius of a concentric duct, m
direct manner. The dimensionless mean wall shear stress is pre- ro ¼ outer radius of a concentric duct, m
sented graphically, and it is more general and rational to represent w ¼ velocity, m/s
the original fluid flow physical behavior than the traditional w ¼ average velocity, m/s

061201-8 / Vol. 134, JUNE 2012 Transactions of the ASME

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
x, y ¼ Cartesian coordinates in cross section, m The above exact equations provide a more accurate choice to cal-
z ¼ coordinate in flow direction, m culate frictional pressure drop when it is handy to use computer
algebra systems.
Greek Symbols
D¼ average height of surface roughness element, m References
e¼ effective aspect ratio [1] White, F. M., 1974, Viscous Fluid Flow, McGraw-Hill, New York.
h¼ half angle of isosceles triangle apex, deg [2] Mironer, A., 1979, Fluid Mechanics, McGraw-Hill, New York.
l¼ dynamic viscosity, N s m2 [3] Denn, M., 1980, Process Fluid Mechanics, Prentice-Hall, Upper Saddle River,
NJ.
¼ kinematic viscosity, m2/s  [4] Eckert, E. R. G., and Irvine, T. F., 1956, “Flow in Corners of Passages With
s¼ mean wall shear stress, N m2 Noncircular Cross Section,” Transactions of ASME, 78, pp. 709–718.
s ¼ dimensionless mean wall shear stress ¼ sL=l w [5] Eckert, E. R. G., and Irvine, T. F., 1960, “Pressure Drop and Heat Transfer in a
u¼ half angle of annular sector, rad Duct With Triangular Cross-Section,” ASME J. Heat Transfer, 82, pp.
125–138.
[6] Carlson, L. W., and Irvine, T. F., 1961, “Fully Developed Pressure Drop in Tri-
Subscripts angular Shaped Ducts,” ASME J. Heat Transfer, 83, pp. 441–444.
pffiffiffi [7] Nan, S. F., and Dou, M., 2000, “A Method of Correlating Fully Developed Tur-
A ¼ based upon the square root of flow area bulent Friction in Triangular Ducts,” Trans. ASME J. Fluids Eng., 122, pp.
Dh ¼ based upon the hydraulic diameter 634–636.
L ¼ based upon the arbitrary length L [8] Hodge, R. I., 1961, “Frictional Pressure Drop in Noncircular Ducts,” ASME J.
Heat Transfer, 83, pp. 384–385.
[9] Ahmed, S., and Brundrett, E., 1971, “Characteristic Lengths for Non-Circular
Ducts,” Int. J. Heat Mass Transfer, 14, pp. 157–159.
Appendix [10] Bandopadhayay, P. C., and Ambrose, C. W., 1980, “A Generalized Length
Dimension for Noncircular Ducts,” Lett. Heat Mass Transfer, 7, pp. 323–328.
Computer algebra systems such as MAPLE, MATLAB, and MATHE- [11] Gunn, D. J., and Darling, C. W. W., 1963, “Fluid Flow and Energy Losses in
MATICA can be used to calculate very accurate values of the friction Non-Circular Conduits,” Trans. Inst. Chem. Eng., 41, pp. 163–173.
[12] Rehme, K., 1973, “Simple Method of Predicting Friction Factors of Turbulent
factor for turbulent flow in channels. These computer algebra sys- Flow in Non-Circular Channels,” Int. J. Heat Mass Transfer, 16, pp. 933–950.
tems have special functions such as the Lambert W function [60]. [13] Malak, J., Hejna, J., and Schmid, J., 1975, “Pressure Losses and Heat Transfer
The Lambert W function has been widely used in some fields for in Non-Circular Channels With Hydraulically Smooth Walls,” Int. J. Heat Mass
many years, but awareness of the function was not as high as it Transfer, 18, pp. 139–149.
[14] Jones, O. C., 1976, “An Improvement in the Calculation of Turbulent Friction
should have been. The Lambert W function is defined as the solu- in Rectangular Ducts,” Trans. ASME J. Fluids Eng., 98, pp. 173–181.
tion of the following transcendental equation: [15] Jones, O. C., and Leung, J. C. M., 1981, “An Improvement in the Calculation of
Turbulent Friction in Smooth Concentric Annuli,” Trans. ASME J. Fluids Eng.,
W ð xÞ þ ln½W ð xÞ ¼ lnð xÞ (A1) 103, pp. 615–623.
[16] Obot, N. T., 1988, “Determination of Incompressible Flow Friction in Smooth
Circular and Noncircular Passages: A Generalized Approach Including Valida-
The general implicit Colebrook–White equation is expressed as tion of the Century Old Hydraulic Diameter Concept,” Trans. ASME J. Fluids
Eng., 110, pp. 431–440.
  [17] He, S., and Gotts, J. A., 2004, “Calculation of Friction Coefficients for Noncir-
1 C4
pffiffiffi ¼ C1  C2 ln C3 þ pffiffiffi (A2) cular Channels,” Trans. ASME J. Fluids Eng., 126, pp. 1033–1038.
f Re f [18] Muzychka, Y. S., and Yovanovich, M. M., 2002, “Laminar Flow Friction and
Heat Transfer in Non-Circular Ducts and Channels: Part I-Hydrodynamic Prob-
lem,” Compact Heat Exchangers, A Festschrift on the 60th Birthday of Ramesh
where C1 to C4 are constants and their values depend on the length K. Shah, Grenoble, France, pp. 123–130.
scale chosen. The constant C2 can be obtained by passing from or- [19] Bejan, A., 2000, Shape and Structure: From Engineering to Nature, Cambridge
dinary logarithms to natural logarithms (log x ¼ ln x=ln 10). The University Press, Cambridge, England.
foregoing equation has an exact solution that can be expressed in [20] Duan, Z. P., and Muzychka, Y. S., 2007, “Slip Flow in Non-Circular Micro-
channels,” Microfluid. Nanofluid., 3, pp. 473–484.
terms of the Lambert W function. Using MAPLE, MATLAB, or MATHE- [21] Duan, Z. P., and Muzychka, Y. S., 2007, “Compressibility Effect on Slip Flow
MATICA we obtain the following general explicit form for the Fan- in Non-Circular Microchannels,” Nanosc. Microsc. Therm., 11, pp. 259–272.
ning friction factor: [22] Muzychka, Y. S., and Edge, J., 2008, “Laminar Non-Newtonian Fluid Flow in
Noncircular Ducts and Microchannels,” Trans. ASME J. Fluids Eng., 130(11),
1 p. 111201.
f ¼ 2 0   1 32 (A3) [23] Duan, Z. P., and Yovanovich, M. M., 2009, “Pressure Drop for Laminar Flow
C1 in Microchannels of Arbitrary Cross-Sections,” Semi-Therm 25 Semiconductor
exp Re   Thermal Measurement and Management Symposium, San Jose, CA, pp.
6 B C3 Re C 7
C22 6 B C2 C  C3 Re7 111–120.
4W @ C2 C4 exp
C2 C4 A C2 C4 5 [24] Duan, Z. P., and Muzychka, Y. S., 2010, “Slip Flow in the Hydrodynamic En-
trance Region of Circular and Noncircular Microchannels,” Trans. ASME J.
Fluids Eng., 132(1), p. 011201.
pffiffiffi [25] Blasius, H., 1913, “Das Ahnlichkeitsgesetz bei Reibungsvorgangen in
Therefore, employing A as the characteristic length scale, the Flussigkeiten,” VDI Forschungsheft, 131, pp. 1–12.
modified exact explicit Colebrook–White equation can be [26] Prandtl, L., 1935, “The Mechanics of Viscous Fluids,” Aerodynamic Theory,
Vol. III, W. F. Durand, ed., Springer, Berlin, pp. 142.
expressed as [27] von Karman, T., 1934, “Turbulence and Skin Friction,” J. Aeronaut. Sci., 7, pp.
1–20.
f RepffiffiAffi [28] Nikuradse, J., 1932, “Gesetzmassigkeit der turbulenten Stromung in glatten
Rohren,” VDI Forschungsheft, 356, pp. 1–36.
0:3314
i2 RepAffiffiffi
[29] Nikuradse, J., 1930, “Untersuchungen uber turbulent Stromung in nicht kreis-
¼h     formigen Rohren,” Ing.-Arch, 1, pp. 306–332.
Dffiffiffi pffiffi
ffi pffiffi
ffi Dffiffiffi p
W 0:5150exp 0:1231 A Re A Re A 0:1231 A Re A
p p ffiffi
ffi [30] Prandtl, L., 1952, Essentials of Fluid Dynamics, Hafner Publishing Company,
New York, NY.
(A4) [31] McKeon, B. J., Zagarola, M. V., and Smits, A. J., 2005, “A New Friction Factor
Relationship for Fully Developed Pipe Flow,” J. Fluid Mech., 538, pp.
429–443.
Also,
pffiffiffithe exact form of the modified PKN equation becomes as [32] Moody, L. F., 1944, “Friction Factors for Pipe Flow,” Transactions of ASME,
D A!0 66, pp. 671–684.
[33] Colebrook, C. R., 1939, “Turbulent Flow in Pipes With Particular Reference to
0:3314 the Transition Region Between Smooth and Rough Pipe Laws,” J. Instit. Civil
f RepffiffiAffi ¼ h  i2 RepAffiffiffi (A5) Eng., 11, pp. 133–156.
W 0:5150RepffiffiAffi [34] Filonenko, G. K., 1954, “Hydraulic Resistance in Pipes,” Teploenergetika, 1,
pp. 40–44.

Journal of Fluids Engineering JUNE 2012, Vol. 134 / 061201-9

Downloaded 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
[35] Techo, R., Tickner, R. R., and James, R. E., 1965, “An Accurate Equation for [49] Romeo, E., Royo, C., and Monzon, A., 2002, “Improved Explicit Equations for
the Computation of the Friction Factor for Smooth Pipes From the Reynolds Estimation of the Friction Factor in Smooth and Rough Pipes,” Chem. Eng. J.,
Number,” ASME J. Appl. Mech., 32, pp. 443–446. 86, pp. 369–374.
[36] Kakac, S., Shah, R. K., and Aung, W., 1987, Handbook of Single-Phase Con- [50] Avci, A., and Karagoz, I., 2009, “A Novel Explicit Equation for Friction Factor
vective Heat Transfer, Wiley, New York, NY. in Smooth and Rough Pipes,” Trans. ASME J. Fluids Eng., 131(6), p. 061203.
[37] Colebrook, C. R., and White, C. M., 1937, “Experiments With Fluid Friction [51] Duan, Z. P., 2007, “Analysis of Slip Flow in Microchannels,” Ph.D. disserta-
Roughened Pipes,” Proceedings of the Royal Society of London, Series A, 161, tion, Memorial University, St. John’s, Newfoundland, Canada.
pp. 367–381. [52] Duan, Z. P., and Muzychka, Y. S., 2008, “Effect of Corrugated Roughness on
[38] Afzal, N., 2007, “Friction Factor Directly From Transitional Roughness in a Developed Laminar Flow in Microtubes,” Trans. ASME J. Fluids Eng., 130(3),
Turbulent Pipe Flow,” Trans. ASME J. Fluids Eng., 129, pp. 1255–1267. p. 031102.
[39] Wood, D. J., 1966, “An Explicit Friction Factor Relationship,” Civil Eng., 36, [53] Duan, Z. P., and Muzychka, Y. S., 2010, “Effects of Axial Corrugated Rough-
pp. 60–61. ness on Low Reynolds Number Slip Flow and Continuum Flow in Microtubes,”
[40] Churchill, S. W., 1973, “Empirical Expressions for the Shear Stress in Turbu- ASME J. Heat Transfer, 132(4), p. 041001.
lent Flow in Commercial Pipe,” AIChE J., 19(2), pp. 375–376. [54] Obot, N. T., Esen, E. B., and Adu-Wusu, K., 1987, “Pressure Drop for Rib-
[41] Churchill, S. W., 1977, “Friction Factor Equation Spans All Fluid Flow Roughnessed Scalene Triangular Duct Having Two Rounded Corners,” Int.
Regimes,” Chem. Eng., 84, pp. 91–92. Comm. Heat Mass Transfer, 14, pp. 11–20.
[42] Swamee, P. K., and Jain, A. K., 1976, “Explicit Equations for Pipe Flow Prob- [55] Han, J. C., Glicksman, L. R., and Rohsenow, W. M., 1979, “An Investigation of
lems,” J. Hydraulic Eng. ASCE, 102, pp. 657–664. Heat Transfer and Friction for Rib-Roughnessed Surfaces,” Int. J. Heat Mass
[43] Chen, N. H., 1979, “An Explicit Equation for Friction Factor in Pipe,” Ind. Transfer, 21, pp. 1143–1156.
Eng. Chem. Fund., 18, pp. 296–297. [56] Schlichting, H., 1979, Boundary Layer Theory, 7th ed., McGraw-Hill, New
[44] Round, G. F., 1980, “An Explicit Approximation for the Friction Factor Reyn- York.
olds Number Relation for Rough and Smooth Pipes,” Can. J. Chem. Eng., 58, [57] Schiller, L., 1923, “Uber den Stromungswiderstand von Rohrenverschiede-
pp. 122–123. nenq Uerschitts und Rauhigkeitsgrades,” Zeit. Ang. Math. und Mech., 3, pp.
[45] Barr, D. I. H., 1972, “New Forms of Equations for the Correlation of Pipe Re- 2–13.
sistance Data,” Proc. Inst. Civil Eng. London, 53(2), pp. 383–390. [58] Obot, N. T., and Adu-Wusu, K., 1985, “The Flow Pattern in a Scalene Triangu-
[46] Zigrang, D. J., and Sylvester, N. D., 1982, “Explicit Approximation to the Solu- lar Duct Having Two Rounded Corners,” Trans. ASME J. Fluids Eng., 107, pp.
tion of Colebrook’s Friction Factor Equation,” AICHe J., 28, pp. 514–515. 455–459.
[47] Haaland, S. E., 1983, “Simple and Explicit Formulas for the Friction Factor in [59] Panton, R. L., 2005, Incompressible Flow, 3rd ed., John Wiley and Sons, New
Turbulent Pipe Flow,” Trans. ASME J. Fluids Eng., 105, pp. 89–90. York.
[48] Serghides, T. K., 1984, “Estimate Friction Factor Accurately,” Chem. Eng., 91, [60] Corless, R. M., Gonnet, G. H., Hare, D. E. G., Jeffrey, D. J., and Knuth, D. E.,
pp. 63–64. 1996, “On the Lambert W Function,” Adv. Comput. Math., 5, pp. 329–359.

061201-10 / Vol. 134, JUNE 2012 Transactions of the ASME

Downloaded
View publication stats 12 Jun 2012 to 129.97.58.107. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

You might also like