Powder Technology, 47 (1936) 83 - 86 a3

Short Communication

A Short Note on the Drag Correlation for plus four more, not given here are the best
Spheres available approximation to the standard drag
curve for Re < 106.
The need for such a complicated regression
R. TURTON and 0. LEVENSPIEL equation is questioned in this note. For this
Chemical Engineering Department, Oregon State purpose, the goodness of fit of the Clift et al.
Univereity, Corvallis, OR 97331 (U.S.A.) equation to the 408 data points reported in
(Received February 5.1986) the literature [ 13 - 311 was compared with
the goodness of fit of two very much simpler
equations. The first, eqn. (l), was originally
proposed by Gilbert et al. [5], while the
For the past 80 years or so, considerable second, eqn. (2), was proposed by Clift and
attention has been paid to the determination Gauvin [9]:
of terminal velocity and drag coefficient for
falling spheres, with numerous attempts at &=XReY+Z (1)
developing theoretically based expressions to
relate the drag coefficient with Reynolds &= $(l+AR@)+ ’ (2)
number. The application of such expressions l+DR#
is limited to the low Reynolds number range
(Re < 200). Outside of this range, empirical The constants for eqns. (1) and (2) were
relationships must be used. A comprehensive chosen so as to minimize the sum of squares
review of the available correlations [l - 111 is error Q, defined below:
given by Clift et al. [ 121, who present a new
correlation based on a critical review of Q = (lw,,G, - h&d2 (3)
published data. This correlation consists of six The resulting constants and values of Q for
polynomial equations with a total of 18 fitted eqns. (1) and (2) are given in Table 2. These
constants, and is reproduced in Table 1. Clift results are compared with those for the Clift
et al. [ 121 claim that this set of six equations, et al. correlation.

TABLE 1
Drag correlation from Clift et al. [ 121

Re range Correlation

0.01 < Re < 20 = -0.881 + 0.82w - 0.06~~

20<Re<260 = -0.7133 + 0.6306~

260 < Re < 1500 logloCD = 1.6435 - 1.1242~ + 0.1558w2

1600 < Re < 1.2 x lo4 l‘%OCD = -2.4571+ 2.5566w - 0.9295w2 + 0.1049w3

1.2 x lo4 < Re G 4.4 x lo4 l"k?lOCD = -1.9161+ 0.6370~ - 0.0636~~

4.4 x lo4 < Re G 3.38 x 10’ log&, = -4.3390 + i.58ogW - o.i646W2

where w = logloRe

0032-5910/86/$3.60 0 Elsevier Sequoia/Printed in The Netherlands

84

TABLE 2
Comparison of drag correlation equations

Model Best-fit parameters Least-squares error Q Percentage of data Percentage of data

(see eqn. (3)) within flO% of within +16% of
correlation correlation

Clift et al. [ 12 ] See Table 1 for the 0.283 90.2 97.8

18 constants
Eqn. (1) X = 27.2 0.640 72.3 92.9
Y = 0.827
2 = 0.427
Eqn. (2) A = 0.173 0.254 90.4 98.8
B = 0.657
c = 0.413
D = 16300.0
E = -1.09

The final best-fit equations are and the comparison with the data is displayed
in Figs. 1 and 2.
Cn = 27.2Re0*s2’ + 0.427 (4) It is evident from Fig. 1 that the three-
and constant equation does not correlate the data
very well at high Reynolds number. This is
.,
cn = g (1 + 0.173Re0*657) because this equation form will always be a
monotonically decreasing function, while the
experimental data show a minimum CD-value
0.413 at a Reynolds number around 5000 with Cn
(5)
+ 1 + 16300Re-‘*09 increasing for Reynolds numbers greater than

, Equation 4: Three constant expression

6
e, = mgg + 0.427
ne
,,,, ,I ,/,,,,, I, I
.
2 I I IIIIIII. I
\
Stokes law
I- 24
CD
Re
/II

2.

0.1 ’
I 2 4 6 IO 2 4 6 100 2 4 6 lo3 2 4 6 104 2 4 6 105 2

dsph “t p9
Re= cc
Fig. 1. Experimental data for drag coefficients of spheres as a function of Reynolds number and least-squares
correlation for the three-parameter equation (eqn. (1)).
 85

IO expression
7 6 Equation 5: Five constant

u) NC
“2 4

$
-0 2

II I

0” 6

0. I
I 2 4 6 IO 2 4 6. 100 2 4 6 lo3 2 4 6 lo4 2 4 6 lo5 2

dsph ut pg
Re= ~
Fig. 2. ,ExperimentaI data for drag coefficients of spheres as a function of Reynolds number and least-squares
cokelakon for the five-parameter equation (eqn. (2)).

this. Thus, the three-constant equation gives a CD experimentally reported drag

systematic error trend at Reynolds numbers coefficient of a sphere,
greater than 1000. 8. = %-kph(/-% - P&3P,ut2
The five-constant equation, on the other CD least-squares estimate 6f the
hand, is capable of showing a minimum CD- drag coefficient of a sphere
value, and it can be seen from Fig. 2 that this Q sum of squares error, defined
equation correlates the data extremely well in eqn. (3)
throughout the whole Reynolds number Re Reynolds number at the ter-
range. It shows no systematic error trends minal velocity of the sphere,
and gives a lower Q-value than the six-segment = drphut~glcc
18constant expression recommended by Clift x9 y,z fitted constants used in cor-
et al. [12]. Equation (5) has the added ad- relation eqn. (1)
vantage of converging to the Stokes equation
at low Reynolds number. In conclusion,
therefore, it is felt that eqn. (5) is the simplest
and best correlation for the available data in REFERENCES
the subcritical regime (Re < 2 X 10s).
L. Schiller and A. Z. Nauman, Ver. Deut. Ins., 77
(1933) 318.
ACKNOWLEDGEMENT C. E. Lapple, Particle Dynamics, Eng. Res. Lab.,
E.I. DuPont de Nemours and Co., Wilmington,
Work for this project was carried out under Delaware, 1951.
NSF Grant # CBT-8420034. I. Langmuir and K. B. Blodget, U.S. Army Air
Force Tech. Report No. 5418, 1948.
H. S. Allen,Philoe. Mug., 50 (1900) 323.
LIST OF SYMBOLS M. Gilbert, L. Davies and D. Altman, Jet Propul.,
25 (1955) 26.
A, B, C, D, E fitted constants used in cor- H. Kurten, J. Raasch and H. Rumpf, Chem-lng-
relation eqn. (2) Tech., 38 (1966) 941.
86

7 F. F. Abraham, Phys. Fluids, 13 (1970) 2194. 19 W. Miiller, Phys. Z., 39 (1938) 57.
8 F. Ihme, H. Schmidt-‘Daub and H. Brauer, Chem- 20 C. N. Davies, Proc. Phys. Sot. London, 57 (1945)
Zng-Tech., 44 (1972) 306. 259.
9 R. Clift and W. H. Gauvin, Proc. Chemeca ‘70, 1 21 C. Wieselberger, Phys. Z., 23 (1922) 219.
(1970) 14. 22 C. Wieselberger, Ergeb. Aero. Vers., Gottingen, 2
10 H. Brauer and D. Mewes, Chem-Zng-Tech., 44 (1923) 28.
(1972) 865. 23 R. G. Lunnon,Proc. Roy. Sot. A, 118 (1928)
11 Z. Tanaka and K. Iinoya, d. Chem. Eng. Jpn., 3 680.
(1970) 261. 24 R. Gunn and G. D. Kinzer, J. Meteor., 6 (1949)
12 R. Ciift, J. R. Grace and M. E. Weber,‘Bubbles. 243.
Drops and Particles, Academic Press, New York, 25 K. L. Goin and W. R. Lawrence, AZAA J., 6
1978. (1968) 961.
13 F. W. Roes and W. W. Wiiimarth, AZAA J., 9 26 G. A. Shakespear, Philos. Mag., 28 (1914) 728.
(1971) 285. 27 H. S. Allen,Philos. Mag., 50 (1900) 519.
14 H. Liebster,Ann. Phys. (Leipzig), 82 (1927) 541. 28 A. B. Bailey and J. Hiatt, AZAA J., 10 (1972)
15 M. Viajinac and E. E. Covert, J. Fluid Mech., 54 1436.
(1972) 385. 29 T. Maxworthy, J. Fluid Mech., 23 (1965) 369.
16 E. S. Pettyjohn and E. B. Christiansen, Chem. 30 K. V. Beard and H. R. Pruppacher, J. A tmos. Sci.,
Eng. Prog., 44 (1948) 157. 26 (1969) 1066.
17 H. D. Arnold, Philos. Mag., 22 (1911) 755. 31 T. H. Nguyen, Technical Paper, Dept. Chem.
18 J. Schmiedel, Phys. Z., 29 (1928) 593. Eng., McGill Univ., Montreal (1973).