Sei sulla pagina 1di 4

D.J.

Zigrang A Review of Explicit Friction


Department of Mechanical Engineering.
Factor Equations
N. D. Sylvester
A review of the explicit friction factor equations developed to replace the Colebrook
Department of Chemical Engineering.
equation is presented. Explicit friction factor equations are developed which yield a
The University of Tulsa, very high degree of precision compared to the Colebrook equation. A new explicit
600 South College, equation, which offers a reasonable compromise between complexity and accuracy,
Tulsa, Okla. 74104 is presented and recommended for the calculation of all turbulent pipe flow friction
factors for all roughness ratios and Reynold's numbers.

Introduction
The Colebrook equation for the friction factor [1], the sensitivity of the friction factor value to pipe roughness
together with a considerable uncertainty in the actual value of
t/D 2.51 1 the roughness may be the predominant source of uncertainty
v7=- 2l0g [ 3.7 RevJ J
(1)
in a friction factor value.
is implicit in / . Nevertheless, it generally has been used to It would seem that our real needs may be for more accurate
obtain friction factors for flow in closed conduits. The widely friction factor data upon which to base correlations and for
recognized Moody chart for friction factors [2] is based on it ways to reduce the uncertainty in pipe roughness rather than
as are others. At least until recently, such charts have fur- for more precise explicit versions of Colebrook's equation.
nished most of the friction factors used in pressure-drop However, the proliferation of explicit approximations
calculations for flow in closed conduits. prompts this review.
The earliest explicit approximation to Colebrook's
equation is due to Moody [3]. Since then, a series of in- Simple Explicit Equations
creasingly more precise explicit approximations have been
published [4-7], The most recent of these [8-11] have been Available friction factor charts cover both laminar and
based on numerical methods and can yield any desired degree turbulent flows. Pipe roughness does not have a measurable
of precision. Zigrang and Sylvester [8] and Shacham [9] as effect on pressure drop in laminar flow and this region is
reported by Olujic [10] have shown independently that adequately covered by equation (2) which is derived from the
starting with a reasonable approximation for / , two ad- Hagen-Poiseuille equation
ditional iterations will yield values for / which deviate 64
negligibly (0.02 percent average) from the fully converged /= (2)
Colebrook equation value, over a wide range of values for Re
Reynold's number and t/D. Williams [12] has pointed out It is usually accepted that flow can be either laminar or
that programmable hand-held calculators are so easy to use turbulent in the transition region 2000 < Re < 4000. Thus
and convergence is so rapid that there is little necessity to stop either equation (1) or equation (2) may actually be applicable.
short of full convergence. This review will examine explicit approximations to
In spite of these increasingly more accurate but increasingly Colebrook's equation down to Re = 2500 with the reservation
more complex explicit equations, J. J. J. Chen [13] has that they are applicable only to fully developed turbulent
suggested a very simple but admittedly inaccurate explicit flow.
equation. Equations demonstratably simpler than Chen's but Moody's equation [3] is included because it was the first of
considerably more precise relative to Colebrook's equation the explicit approximations to Colebrooks's equation.
are available. However, Chen points out that since
Colebrook's equation may contain significant error, there is /=0.0055[l+[20000(i) + -^-]1/3] (3)
little practical justification for increasing the precision of
explicit equations. Haaland [14] also has considered this Zigrang and Sylvester [8] showed that Moody's equation has
point, stating: ". . .Before simplicity is sacrificed for ex- an average error of 4.3 percent for their test cases; thus, it is
cessive accuracy, it is worth keeping in mind that the not considered for further evaluation. Another early explicit
Colebrook-White formula. . .itself may be 3-5 percent, if not equation, presented by Wood [4], is not considered for similar
more, in error compared to experimental data. . . . " Finally, reasons. Churchill's equation [6] was considered in an earlier
review by Zigrang and Sylvester [8] where it was shown to be
less accurate relative to equation (1) than other less complex
Contributed by the Petroleum Fluid Mechanics Committee of the Petroleum equations.
Division for publication in the JOURNAL OF ENERGY RESOURCES TECHNOLOGY.
Manuscript received by the Petroleum Division, July 23, 1984; revised Other simple explicit equations include those of Jain [5],
manuscript received February 8, 1985. Zigrang and Sylvester [8], Haaland [14], N. H. Chen [7],

280/Vol. 107, JUNE 1985 Transactions of the ASME

Copyright © 1985 by ASME


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/18/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Shacham [9], Serghides [11] and J. J. J. Chen [13]. These are Table 1 Comparison of simple friction factor equations
discussed in the forthcoming.
Jain's equation is based on the theoretical equations of von absolute
Karman and Prandtl together with curve fitting. deviation
Equation Complexity
no. Avg. Max. index Reference
- 2 1og[(e/Z>) + l^r] (4)
* - 1 0.0 0.0 -- Colebrook, 1939
Zigrang and Sylvester [8] substituted a value f o r / o f 0.04 4 1.0 4.1 35 Jain, 1976
5 1.9 7.4 24 Zigrang-Sylvester, 1982
into the right-hand side of Colebrook's equation to obtain 6 2.6 8.5 24 Shacham, 1980
7 0.6 2.3 35 Haaland, 1983
1 *, rt/D 13
1 (5) 10 4.9 12.6 41 N. H. Chen, 1979
Tf 11 1.7 8.9 24 Serghides, 1984
This is almost identical to Shacham's equation, except that
Shacham [9] used a value of 0.03 f o r / t o obtain Table 2 Comparison of friction factor equations of in-
1 e/D 14.5 termediate complexity
- 2 1og[ (6)
VJ " " " ° L 3.7 ' Re Percent
Haaland [14] reexamined the basis for Colebrook's absolute
deviation
equation in view of the simple friction factor relations of Equation Complexity
Prandtl and von Karman for smooth and rough pipes and no. Avg. Max. index Reference
developed the explicit relation 1 0.00 0.0 Colebrook, 1939
1 12 0.14 0.5 82 N. H. Chen, 1979
. „, [6.9 r e / f l l '•" 1 13 0.21 0.8 49 Zigrang-Sylvester, 1982
= (7)
v7 14
15
0.10
0.07
0.5
0.4
60
60
Zigrang-Sylvester, this work
Zigrang-Sylvester, this work
Haaland then generalized this relation to
1 1.8 ,logrr 6.9 " i "+ r e/D i u " n (8)
lLid h d J Numerically Derived Explicit Equations of Intermediate
For n = l, equation (8) becomes equation (7). Haaland Precision
suggested that n = 3 yields friction factors in consonance with The N. H. Chen [7] equation, from which equation (10) was
those recommended for use in gas transmission lines. extracted, is shown as equation (12).
J. J. J. Chen [13] suggested two relations which are 11098
5.0452.
" . . . simpler than any of those already reviewed
. . . (but) . . . not expected to be of high accuracy
1
= -2
"h:
e/D
7065 Re
W[«r(e/Z>)
2.8257
5.8506
Re 0 •]]
. . . " Chen's first equation is (12)
0.3 An equation comparable to Chen's in complexity and
/=0.3164[^ + 0.1lQ)] (9) accuracy was presented by Barr [16] but is not presented here.
Zigrang and Sylvester [8] combined equation (5) with
Zigrang and Sylvester [15] have shown that equation (5) is not equation (1) to get an equation of somewhat less complexity
only simpler than equation (9), but also considerably more than N. H. Chen's (see equation (13)). It is the first iteration
precise relative to Colebrook's equation. Chen's second of the numerical solution of equation (1) using the method of
equation is of similar accuracy. straight substitution and the value for / from equation (5) as
A simple explicit expression for friction factors can be the first guess.
obtained from N. H. Chen's [7] equation (see equation (12)). 1 e/D 5.02 e/D 13 "H
= - 2 log[ log[ Jj + ReJJ (13)
1 , IT e/D l i u [7.151»-»1 (10) 3.7 Re
2
vT- 108
LLT549-J +
h d J Zigrang and Sylvester [8] have shown that equation (12) is
Equation (10) is relatively inaccurate as shown in Table 1. slightly more precise than equation (13).
This is surprising because equation (12), from which it is However, two equations, which are more complex than
obtained, is quite good as is shown in Table 2. It appears that equation (13) but less complex than equation (12) and both of
while the form of N. H. Chen's equation comes from a which are simpler and more accurate than equation (12), can
numerical method, the values of the constants result from be obtained by combining equation (4) with equation (1) and
curve fitting. Thus, the extraction of equation (10) from by combining equation (7) with equation (1). The resulting
equation (12) may not be appropriate. equations are
Finally, Serghides [11] utilized an equation similar to
equations (5) and (6), except that the constant for the second 1
term in the parentheses is 12 rather than 13 or 14.5. „, r e/D 2.5i r r 21.25 m
lf~' - 2 1 O g L 3 . 7 + R e [ 1 - 1 1 4 - 2 1 O g k + R e 0 . 9 ] ] ]
1
:=-2 (14)
VT *HT*£]
-""I 3.7
Equations (4) through (7), (10) and (11) have been com- 1 „, [e/D 4.518, r 6.9 r e/D 1 '•'' 11 / l r N
pared to equation (1) for 2500<Re<10 7 and 4 x Tf- - 2 1 ° 4 3 . 7 - Re l 0 4 R e +
L 3.7 J JJ ( 1 5 )
1 0 5 <e/Z><0.05 as in reference [11]. The results are sum-
marized in Table 1. Included is the complexity index which is Equations (12) through (15), which are of intermediate
defined as the number of algebraic notation calculator key precision, were compared on the basis of error relative to
strokes required to solve the equation for Re = 105 and Colebrook's equation and on the basis of complexity. The
e/D = 0.001. Table 1 shows that equation (7) is the superior results are shown in Table 2. Table 2 shows that equation (15)
simple explicit equation. is the superior equation of intermediate precision.

Journal of Energy Resources Technology JUNE 1985, Vol. 107/281

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/18/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1.00
I—V Table 3 Comparison of friction factor equations of highest
precision
Percent
Eq.(l3) absolute
deviation
Equation Complexity
Ul 0.75 no. Avg. Max. index Reference
o
DC 1 0.0000 0.0000 Colebrook, 1939
UJ 16 0.0300 0.1400 74 Zigrang-Sylvester, 1982
Q.
17 0.0002 0.0023 124 Serghides, 1984
(E Eq.(l4) Eq.(l2) 18 0.0170 0.1980 99 Serghides, 1984
O
CC 0.50 O 19 0.0040 0.0300 99 Zigrang-Sylvester, this work
DC
UJ

s Eq.(l5)Q It is of interest to compare the accuracies of equations (16)


5 through (19) with the best of the equations of intermediate
X precision to see if the added precision is worth the additional
< 0.25 - effort. The comparison shown in Table 3 is from Serghides
5 [11] with the addition of the results for equations (16) and (19)
Eq.(l6)
O and the complexity index. The accuracies of the equations
Eq.(IT) compared in Tables 2 and 3 are shown relative to their
Eq.(l9)(? ^ ^ complexity in Fig. 1. The trend line shows that equations (15),
(16), and (19) offer an edge in accuracy relative to complexity.
50 75 100
COMPLEXITY INDEX
Accuracy Requirements
Fig. 1 Equation accuracy versus complexity Index
The explicit friction factor equations considered have
maximum errors relative to Colebrook's equation as great as
Numerically Derived Explicit Equations of Highest 13 percent and as low as 0.0023 percent for the ranges of Re
Precision and e/D covered. Tables 1, 2, and 3 and Fig. 1 all show that
accuracy and complexity are correlated. The effort required
Equation (16) is the second iteration of equation (13) ob- to use an equation is proportional to its complexity. In
tained by Zigrang and Sylvester [8]. general, one would like to use the simplest equation which
1 „, [e/D 5.02, [e/D 5.02, [e/D provides "reasonable" accuracy.
v7 If the errors in the explicit equations relative to Colebrook's
equation were random and if the 3 to 5 percent error in
Colebrook's equation was random, the task of deciding
£]]] (16) whether a given explicit equation is "reasonably" accurate
would be relatively easy. In this case, the magnitude of the
Equations (13) and (16) imply that additional iterations are combined errors would be the root-sum-square value.
possible. However, the errors in the explicit equations are not random
Serghides [11] has taken equation (16) and incorporated and the errors in Colebrook's equation relative to the ex-
Steffenson's accelerated convergence technique to make the perimental data are not expected to be random. Since the
resulting solution more accurate than that which would result explicit and Colebrook errors are of both signs the errors will
from several additional straight iterations. Serghides' add in some cases and cancel in other cases.
equation is
Consider the effect of using an explicit equation with a
(B-A)2 1-2 maximum error of one percent. Solving the Colebrook
f -[' C-2B+A1
(17a)
equation for the friction factor might give a maximum error
of five percent. Using the relatively inaccurate equation
where
would, in the worst case, add an additional one percent error
„ , [e/D 121 for a total of six percent. On the other hand, if the errors were
A = = (Mb)
of different sign, the maximum error would be reduced from
five percent to four percent.
e/D 2.5L41 Now consider the effect of using an explicit equation having
B -21og[- (17c)
a maximum error of one percent under the assumption that
the errors are random. In this case, it is assumed that the

- 2 l 0 4 ^ + -^rJ
„, [e/D 2.515-1 Colebrook equation errors also are random. Here the root-
c = = (\ld)
sum-square procedure applies and the resulting error would
beV5 2 + 1 2 = 5.10 percent. The increase in friction factor
Serghides also presents the simpler equation
uncertainty by 0.1 percent would certainly be regarded as
(/4-4.781) 2 I"2 negligible.
/ = [4.781 -
5-2.4+4.781 -](18)
The actual effect of the use of an explicit equation with a
maximum error of one percent lies between these extremes.
where A and B are defined as in equation (17).
On the basis of this discussion it is suggested that the high
Finally, the next iteration of equation (16) is of the same accuracies of equations (16) through (19) are unnecessary in
level of convergence as equation (17), and can be written practice.
1 [e/D 5.02 [e/D 5.02 [e/D
- - 2 log ^ — — i r r l o g [ T T — R ~ 8
1 T7 Conclusions and Recommendations
Based on this review and, particularly, in view of the
5.02 inaccuracy of the Colebrook equation, the following con-
*•(£•£)]]] (19)
clusions and recommendations are presented:

282/Vol. 107, JUNE 1985 Transactions of the ASME

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/18/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1 Until more accurate experimental data is available or a 3 Moody, L. F., "An Approximate Formula for Pipe Friction Factors,"
more accurate correlation of existing data is obtained, the TRANS. ASME, Vol. 69, 1947, p. 1005.
4 Wood, D. J., "An Explicit Friction Factor Relationship," Civil Engineer-
Colebrook equation should continue to serve as the standard. ing, Dec. 1966, pp. 60-61.
2 Additional explicit friction factor equations are not 5 Jain, A. K., "Accurate Explicit Equation for Friction Factor," Journal of
needed. Hydraulics Division, ASCE, Vol. 102, HY5, May 1976.
3 Numerical solutions for friction factors based on the 6 Churchill, S. W., "Friction-factor Equation Spans all Fluid Flow
Regimes," Chemical Engineering, Vol. 91, November 7, 1977.
Colebrook equation can be obtained by iteration to any 7 Chen, N. H., "An Explicit Equation for Friction Factor in Pipe," In-
desired degree of precision. dustrial Engineering Chem. Fundamentals, Vol. 18(3), 1979, pp. 296-297.
4 For many applications, the simple explicit expressions of 8 Zigrang, D. J., and Sylvester, N. D., "Explicit Approximations to the
equations (5), (6), (7) or (11) will suffice. Solution of Colebrook Friction Factor Equation," AIChE Journal, Vol. 28,
No. 3, 1982.
5 If more precision is required, any of equations (12) 9 Shacham, M., Industrial Engineering Chemistry, Fundamentals, May
through (19) are adequate for turbulent flows. 1980, pp. 228-229.
6 If the application is exclusively in the region of higher 10 Olujic, Z., "Compute Friction Factors Fast for Pipe Flow," Chemical
Reynold's numbers (Re > 10000) and higher roughness ratios Engineering, December 14, 1981.
11 Serghides, C. K., "Estimate Friction Factors Accurately," Chemical
(e/Z?>.004), equation (5) is recommended. In this region it is Engineering, March 5, 1984, pp. 63-64.
nearly as accurate as equation (7) and substantially simpler. 12 Williams, C. L., Electric Power Research Institute, personal communica-
7 Equation (15), which offers a reasonable compromise tion, June 17, 1982.
between complexity and accuracy, is recommended for 13 Chen, J. J. J., "Technical Note 400: A Simple Explicit Formula for the
Estimation of Pipe Friction Factor," Proceedings of the Instn. of Civil
calculation of all friction factors for turbulent flows for all Engineers, Part 2, Vol. 77, Mar. 1984, pp. 49-55.
roughness ratios and Reynold's numbers. 14 Haaland, S. E., "Simple and Explicit Formulas for the Friction Factor in
Turbulent Pipe Flow," ASME Journal of Fluids Engineering, Vol. 105, 1983, p.
References 89.
15 Zigrang, D. J., and N. D. Sylvester, "Comments on Technical Note
1 Colebrook, C. F., "Turbulent Flows in Pipes, With Particular Reference 400: A Simple Explicit Formula for the Estimation of Pipe Friction Factor,"
to the Transition Region Between Smooth and Rough Pipe Laws," Journal of Proceedings of the Instn. of Civil Engineers, Vol. 79, Mar. 1985, pp. 215-221.
Inst, of Civil Engineering, London, Vol. 11, 1939, pp. 133-156. 16 Barr, D. I. H., "Solutions of the Colebrook-White Function for
2 Moody, L. F., "Friction Factors for Pipe Flow," TRANS. ASME, Vol. 66, Resistance to Uniform Turbulent Flow," Proceedings of the Instn. of Civil
1944, p. 641. Engineers, Part 2, Vol. 71, 1981, pp. 529-535.

Journal of Energy Resources Technology JUNE 1985, Vol. 107/283

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/18/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Potrebbero piacerti anche