Documente Academic
Documente Profesional
Documente Cultură
com
Abstract
The drag of a non-spherical particle was reviewed and investigated for a variety of shapes (regular and irregular) and particle Reynolds
numbers (Rep). Point-force models for the trajectory-averaged drag were discussed for both the Stokes regime (Rep ≪ 1) and Newton regime
(Rep ≫ 1 and sub-critical with approximately constant drag coefficient) for a particular particle shape. While exact solutions were often available
for the Stokes regime, the Newton regime depended on: aspect ratio for spheroidal particles, surface area ratio for other regularly-shaped particles,
and min–med–max area for irregularly shaped particles. The combination of the Stokes and Newton regimes were well integrated using a general
method by Ganser (developed for isometric shapes and disks). In particular, a modified Clift–Gauvin expression was developed for particles with
approximately cylindrical cross-sections relative to the flow, e.g. rods, prolate spheroids, and oblate spheroids with near-unity aspect ratios.
However, particles with non-circular cross-sections exhibited a weaker dependence on Reynolds number, which is attributed to the more rapid
transition to flow separation and turbulent boundary layer conditions. Their drag coefficient behavior was better represented by a modified
Dallavalle drag model, by again integrating the Stokes and Newton regimes. This paper first discusses spherical particle drag and classification of
particle shapes, followed by the main body which discusses drag in Stokes and Newton regimes and then combines these results for the
intermediate regimes.
© 2007 Elsevier B.V. All rights reserved.
1. Drag of spherical particles The drag force arises from pressure and viscous stresses applied
to the particle surface and resists the relative velocity (w).
With respect to the velocities of the different phases, the Assuming that the particle and the far-field velocities are steady
particle velocity (v) is defined as the translational velocity of the and the flow is spatially uniform away from the particle,
particle center of mass (xp). The continuous-fluid velocity (u) is ∇ · u@p = 0, the magnitude of drag is primarily dictated by the
generally defined in all areas of the domain unoccupied by particle Reynolds number (Rep), defined as
particles. However, a hypothetical continuous-phase velocity
can be extrapolated to the particle centroid as u@p and termed qf jwjd
Rep u : ð2Þ
the “unhindered velocity”. The relative velocity of the particles lf
(w) is then based on the unhindered velocity, i.e., along a
particle trajectory In this expression d is the particle diameter, ρf is the
continuous-phase density, and μf is the continuous-phase
wðtÞuvðtÞ u@p ðtÞ: ð1Þ
viscosity. Stokes [1] derived the drag force for a sphere in the
It is important to note that u@p does not include the fluid limit of negligible convection terms (Rep ≪ 1) as
dynamic effects resulting from the presence of the particle itself.
FD ¼ 3kdlf w for Rep ≪1: ð3Þ
⁎ 306 Talbot Laboratory, 104 South Wright Street, Urbana, IL 61801-2935, This is often referred to as the Stokes' drag regime owing to his
United States. Tel.: +1 217 244 5581. derivation. For increasing particle Reynolds numbers, the flow be-
E-mail address: loth@uiuc.edu. hind the particle is remains an attached laminar wake for Rep b 22,
0032-5910/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2007.06.001
E. Loth / Powder Technology 182 (2008) 342–353 343
ratio (E). This is due to the increased surface area for the include non-continuum effects which are beyond the scope of
cylinder as compared to a spheroid with the same volumetric this article), then the net correction for oblate and prolate shapes
diameter (to be discussed in the next sub-section), so that a based on Eq. (15) and Table 1 becomes:
cylinder with a unity aspect ratio has a drag which is 14.4% pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
larger than that of a sphere. It can also be seen that variations in E1=3 1 E2
E from 0.01 to 100 result in only moderate changes in the h fE i ¼ for Eb1
cos1
pEffiffiffiffiffiffiffiffiffiffiffiffiffiffi
overall drag correction (less than four-fold). This arises because E 1=3 E2 1 : ð16Þ
increases in friction drag for longer bodies are approximately h fE i ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi for EN1
balanced by decreases in pressure drag, while the reverse is true ln E þ E2 1
for shorter bodies. As such, drag for particles whose aspect
ratios are near unity can often be reasonably estimated by As may be expected, the minimum drag shape for a given
simply using Stokesian drag and the volume-based diameter. particle volume when averaging over all orientations is a sphere
Generally, spheroids traveling at an initial orientation with (for which fE = 1). However, the minimum drag shape for a
no rotation will continue to fall at the same orientation, though a stationary orientation is a prolate spheroid with E = 1.955
broadside orientation tends to be more stable [2]. If the initial moving parallel to the axis of symmetry, where fE‖ = 0.9555
orientation is parallel or perpendicular to the relative velocity, (Fig. 3).
no torque exists since the drag acts in the opposite direction to
the gravitational force. However, it is possible for a spheroid to 3.2. Other regularly-shaped particles in creeping flow
be at an orientation which is neither parallel nor perpendicular.
Due to the linearity of the drag in creeping flow conditions, the Regularly-shaped non-spheroidal particles such as cones,
net drag force may be obtained based on a simple combination cubes, etc. (Fig. 2) do not typically have analytical solution for the
of the individual components [2,4], i.e. drag even in the creeping flow limit. As a first approximation,
their shapes and corresponding drag corrections may be
FD ¼ 3klf ðwO fEO þ w⊥ fE⊥ Þ for Rep ≪1: ð13Þ approximated as ellipsoids by determining an effective aspect
ratio (E). As shown in Fig. 3, this approach works well for
In this expression, w‖ and w⊥ are the components of relative cylinders since their shape is quite similar to that of a spheroid.
velocity which are parallel and perpendicular to the particle's However, additional accuracy may be obtained in terms of a shape
axis of symmetry (and whose vector sum is w). If both velocity correction ( fshape), defined similar to Eq. (12). As with the
components are non-zero, FD will not be parallel to w so that spheroidal shape factor fshape is inversely proportional to the
the particle will have a glide angle with respect to g [4]. One change in terminal velocity (for d = const.) since the drag
may recast the force of Eq. (11) into a drag component in the dependence is linear in creeping flow:
relative velocity direction and then define the remaining fluid
j
dynamic force as lift. However, herein we will consider the CD;shape Wterm;sphere
force of Eq. (11) as drag and define lift to be the result of non- fshape u ¼ : ð17Þ
CD;sphere Rep ≪1 & const:vol:
Wterm;shape
zero shear associated with the surrounding fluid or non-zero
relative rotation of the particle.
To estimate the shape factor for non-spheroidal regularly-
Similarly, the drag force for ellipsoids with aspects ratios E1,
shaped particles, it is common to consider two dimensionless
E2 and E3 (and other orthotropic particles) will be based on the
area parameters: the surface and the projected area ratios. Each
Stokes correction factor in each direction and the component of
of these areas can be normalized by the surface area of a sphere
relative velocity associated with each axis direction
which has the same volume, i.e.
FD ¼ 3klf ðw1 fE1 þ w2 fE2 þ w3 fE3 Þ for Rep ≪1: ð14Þ
Asurf Aproj
A⁎surf u ; A⁎proj u 1 2 : ð18Þ
In Brownian motion, all orientations can be assumed to be kd 2 4 kd
equally possible since molecular interaction is random. In this
case, one may consider an average correction factor by integrating The surface area ratio will always be greater than one since
over all possible orientations. For particles which are regularly the spherical geometry has the minimum surface area for a
given volume, i.e., Asurf⁎ ≥ 1. The inverse of the surface area
shaped about three axes, the drag for many realizations (or long
times) is based on the average of the inverse correction factors, i.e.: ratio is more commonly defined as the “sphericity ratio” [8], the
“sphericity” [2], or the “shape factor” [9]. For a cylinder with a
3 1 1 1 length to diameter ratio (Ecyl), ratio (Ecyl), geometry relations
¼ þ þ for Rep ≪1: ð15Þ
h fE i fE1 fE2 fE3 give the surface area ratio and equivalent volume diameter as:
Since spheroids have two axes which are equal in length and Lcyl ⁎ 2Ecyl þ 1 3Ecyl 1=3
Ecyl u ; Asurf ¼ ; d ¼ dcyl : ð19Þ
perpendicular to the axis of symmetry, this yields fE1 = fE‖ and dcyl 2 Þ1=3
ð18Ecyl 2
fE2 = fE3 = fE⊥. If one assumes that Stokesian drag is appropriate
which is reasonable for most small particles in a liquid (small The projected area ratio will depend on the orientation of the
particles in air undergoing Brownian motion will generally particle as well as its shape. For example, a long cylinder will have
346 E. Loth / Powder Technology 182 (2008) 342–353
However, this fit does not take into account inertia effects
which can be significant at low aspect ratios and high density
ratios. In particular, disks at high Reynolds numbers can
undergo significant pitching and/or tumbling depending on a
dimensionless inertia parameter⁎
(I⁎), which in turn is propor-
⁎ kq dO
tional to density ratio I u : As a result, measured drag
64d⊥
coefficients for disks falling in air [43] indicate a somewhat
reduced drag for a given area ratio.
The data of Stringham et al. [18] indicate that inertial effects
become important especially once the Reynolds number based
on the disk diameter becomes larger than 2000. As such, the
above drag correction expression is only approximate.
For cylinders and prolate spheroids (E N 1), secondary
motion has also been found to be significant only at extreme
aspect ratios. For example, Jayaweera and Mason [19] found
that unsteady motion such as fluttering was present in liquids at
high Reynolds numbers up to Ecyl b 25, whereas Gorial and
O'Callaghan [20] found that cylinders in air were generally Fig. 6. Newton-drag correction for regularly-shaped particles based on data of
stable for Ecyl N 4. However, both studies (and those they Table 3, Pettyjohn and Christiansen [29], Stringham et al. [18], and Haider and
reviewed) indicated that a drag coefficient based on broadside Levenspiel [23].
E. Loth / Powder Technology 182 (2008) 342–353 349
CD
CD⁎ u
Cshape
: ð28Þ
⁎ Cshape Rep
Rep u
fshape
Fig. 8. Free-fall experimental data for various particle shapes in terms of normalized drag coefficient and Reynolds number with fits for both approximately circular
and non-circular cross-sections. The inset plot shows the conventional relationship between CD and Rep without normalization.
data (with Ecyl ≤ 0.05) indicate “non-circular cross-section” cylinders and even oblate spheroids with E N 0.5. As such, Eq.
characteristics. Other simulations by Comer and Kleinstreuer for (30) can be referred to as the “circular C/S Clift–Gauvin fit”.
Rep ranging from 40 to 120 indicate that spheroids with E =0.2 lie For moderate particle Reynolds numbers, a normalized
approximately in the middle of the circular and non- circular cross- Schiller–Naumann expression may be similarly defined by
section curves. Note that, the data from List and Schemanauer [22] modifying Eq. (11), i.e.
indicate that the cone-hemisphere shape has an fshape of 1.5 and a
Cshape of 2, while the dendritic snow flake (with a thickness to outer f ¼ fshape ½1 þ 0:15ðRe⁎p Þ0:687 for Re⁎p b800 &c circular C=S:
diameter ratio of 4%) has an fshape of 3.1 and a Cshape of 25. ð30Þ
Use of a dimensionless Clift–Gauvin expression based on
(10) yields Other expressions for spherical particles could also be
normalized in this fashion. For the second group of particles
24 h i whose relative cross-section is non-circular, a modified version
CD⁎ ¼ ⁎ 1 þ 0:15ðRep Þ
⁎ 0:687
of the Clift–Gauvin formula similar to that proposed by Ganser
Rep
0:42 can be obtained based on the best fit to the data as
þ for c circular C=S: ð29Þ
1 þ ðRe
42;500
⁎ 1:16
Þp
CD⁎ ¼
24
Re⁎p
½
1 þ 0:035ðRe⁎p Þ0:74 ð31Þ
24
CD⁎ ¼ þ 0:4 for non circular C=S: ð32Þ
Re⁎p
Fig. 11. Drag coefficient dependence on turbulence levels including data from 4. Conclusions
Gauvin and Clamen [41].
This study is a comprehensive review of particle drag associated
are normalized by its diameter and the free-stream turbulence with non-spherical shapes. The focus was on trajectory-averaged
′ ) are normalized by the relative velocity:
levels (urms drag coefficients based on particles in free-fall. Drag coefficients
for regular and irregular shapes have been collected herein for all
V
V⁎ urms available experimental and numerical data, including recent results
urms u ð33aÞ
w which are important to establishing robust models. For both very
high and very low Reynolds numbers, it was found that aspect ratio
V
V⁎ drms (E) was the best shape parameter for spheroids, surface area ratio
drms u : ð33bÞ
d (Asurf ⁎) was the best for regular particles, and max–med–min area
ratio (Ammm⁎) was the best for irregular particles.
The roughness effect is shown in Fig. 10a for a relatively low As suggested by Ganser, the key to determining the drag
′⁎ up to 1.5%, which leads to a four-fold
turbulence level for drms coefficient for non-spherical particles at intermediate Reynolds
reduction in Rep,crit. Conditions with very high surface irregu- numbers is to employ a normalized expression for the drag
′⁎ on the order of 15% or higher) have been studied
larities (drms coefficient based on the Stokes and Newton regimes. The
in the context of irregular particles as discussed in previous resulting drag coefficient models can be obtained from the
section. While very high Rep studies for such conditions are collected equations, figures and tables for the generally broadside
scare, there is no evidence of any drag coefficients below 0.3 for condition in terms of fshape and Cshape. However, it was found that
such particles. Furthermore, such particles are not likely to have the overall behavior also depends on the cross-sectional shape
a drag-crisis at all since the highly irregular surfaces will cause relative to the fall velocity. It should be kept in mind that the
the particles to have turbulent flow over the particle at modest correlations are based on mean free-fall characteristics and that
Rep values for which the measured CD is approximately con- the area ratios give only a gross generalization of the shape
stant (e.g. non-circular cross-section particles in Fig. 8). Unfor- characteristics, e.g. regularly-shaped particles of different shapes
tunately, there is insufficient quantitative data at intermediate but the same sphericity ratio can have significantly different
surface roughness levels (0.015 b drms ′⁎ b 0.15) to determine the Newton-based drag coefficients (that are not predicted by the
behavior of the transition. above functions). Thus, it is always best to obtain CD based on
The effect of turbulence is qualitatively equivalent to that of experiments with that same particle shape and Reynolds number,
small roughness levels on the particle surface. This effect is shown especially for shapes that are substantially non-spherical.
′⁎ b 10− 4)
in Fig. 10b for particles that are effectively smooth (drms However, the theoretical models and empirical correlations put
for which Rep,crit reduces to values as low as 20,000 at turbulence forth all have limits of for which they may be considered
levels of 20%. On the other hand, turbulence levels less than 1% reasonable. Furthermore, additional data is required to improve
have little effect on critical Reynolds number. The combined their accuracy and increase their ranges of applicability.
effect of roughness and turbulence intensity is also shown in
Fig. 10b, where it can seen that turbulence has a more profound Acknowledgement
effect and that the combination gives even larger reductions in
Rep,crit. Correspondingly, roughness becomes less important at The author would like to especially thank Mr. Vladimir
high turbulence intensities. An empirical expression which ap- Kudinok for creating the figures as well as assisting in the data
proximately correlates the combined effects of the data in Fig. 10a collection and analysis.
and b is given by
References
V⁎ 0:03
log10 Rep;crit ¼ 4:4ðdrms þ 0:002Þ
[1] G.G. Stokes, On the effect of the inertial friction of fluids on the motion of
V⁎ 0:55
1:9ðurms V⁎
Þ ðdrms þ 0:002Þ0:07 : ð34Þ pendulums, Trans. Camb. Phil. Soc. 9 (part II) (1951) 8–106.
E. Loth / Powder Technology 182 (2008) 342–353 353
[2] R. Clift, J.R. Grace, M.E. Weber, Bubbles, Drops and Particles, Academic [24] R.P. Chhabra, L. Agarwal, N.K. Sinha, Drag on non-spherical particles: an
Press, New York, 1978. evaluation of available methods, Powder Technol. 101 (1989) 288–295.
[3] J.M. Dallavalle, Micrometrics, Pitman Publishing, New York, 1948. [25] N.S. Cheng, Simplified settling velocity formula for sediment particle,
[4] F.M. White, Fluid Mechanics, McGraw-Hill, New York, 1991. J. Hydraul. Eng. 123 (1997) 149–152.
[5] R. Clift, W.H. Gauvin, Proc. CHEMECA '70, vol. 1, Butterworth, [26] G.V. Madhav, R.P. Chhabra, Drag on non-spherical particles in viscous
Melbourne, 1970, pp. 14–28. fluids, Int. J. Miner. Process. 43 (1995) 15–29.
[6] L. Schiller, A.Z. Naumann, Über die grundlegenden Berechungen bei der [27] H.-Y. Xie, D.-W. Zhang, Stokes shape factor and its application in the
Schwerkraftaufbereitung, Ver. Deut. Ing. 77 (1933) 318–320. measurement of sphericity of non-spherical particles, Powder Technol. 114
[7] A. Oberbeck, Crelles J. 81 (1876) 62. (2001) 102–105.
[8] H. Wadell, The coefficient of resistance as a function of Reynolds number [28] J.K. Comer, C. Kleinsteuer, A numerical investigation of laminar flow past
for solids of various shapes, J. Franklin Inst. 41 (1934) 310–331. non-spherical solids and droplets, ASME J. Fluids Eng. 117 (1995)
[9] C.T. Crowe, M. Sommerfeld, Y. Tsuji, Multiphase Flows with Droplets and 170–175.
Particles, CRC Press, Boca Raton, FL, 1998. [29] E.S. Pettyjohn, E.B. Christiansen, Chem. Eng. Prog. 4 (1948) 157–172.
[10] D. Leith, Drag on non-spherical objects, Aerosol. Sci. Tech. 6 (1987) [30] L. Albertson, Effect of shape on the fall velocity of gravel particles, Proc.
153–161. 5th Hydr. Conf., Studies in Engineering, University of Iowa, Iowa City,
[11] G.H. Ganser, A rational approach to drag prediction of spherical and non- Iowa, 1952.
spherical particles, Powder Technol. 77 (1983) 143. [31] G.R. Alger, Simons, J. Hyd. Eng. Div. ASCE 94 (1968) 721.
[12] J.A. Jimenez, O.S. Madsen, A simple formula to estimate settling velocity [32] M. Gogus, O.N. Ipecki, M.A. Kokpinar, Effect of particle shape on fall
of natural sediments, J. Waterw. Port Coast. Ocean Eng. (2003) 70–78. velocity of angular particles, J. Hydraul. Eng. 127 (10) (October 2001)
[13] M. Dressel, Dynamic shape factors for particle shape characterization, 860.
Part. Charact. 2 (1985) 62–66. [33] J. Baba, P.D. Komar, Measurements and analysis of settling velocities of
[14] S. Tran-Cong, M. Gay, E.E. Michaelides, Drag coefficients of irregularly natural quartz sand grains, J. Sediment. Petrol. 51 (1981) 631–640.
shaped particles, Powder Technol. 139 (2004) 21–32. [34] D.A. Smith, K.F. Chueng, Settling characteristics of calcareous sand,
[15] R. Hill, G. Power, Extremeum principles for slow viscous flow and the J. Hydraul. Eng. (June 2003) 479–483.
approximate calculation of drag, Q. J. Mech. Appl. Math. 9 (1956) [35] N.N. Clarke, P. Gabriele, S. Shuker, R. Turton, Powder Technol. 59 (1989)
313–319. 69.
[16] T.L. Thompson, N.N. Clark, A holistic approach to particle drag [36] K.V. Beard, H.R. Pruppacher, A determination of the terminal velocity and
prediction, Powder Technol. 67 (1991) 57–66. drag of small water drops by means of a wind tunnel, J. Atmos. Sci. 26
[17] I.A. Lasso, P.D. Weidman, Stokes drag on hollow cylinders and (1969) 1066–1071.
conglomerates, Phys. Fluids 29 (1986) 3921–3934. [37] L.B. Torobin, W.H. Gauvin, Can. J. Chem. Eng. 38 (1960) 142–153.
[18] G.E. Stringham, D.B. Simons, H.P. Guy, The behavior of large parti- [38] E. Achenbach, Experiments on the flow past spheres at very high Reynolds
cles falling in quiescent liquids, Prof. Pap. US Geol. Surv., vol. 562-C, numbers, J. Fluid Mech. 54 (Part. 3) (1972) 565–575.
1969. [39] E. Achenbach, The effects of surface roughness and tunnel blockage on the
[19] K.O.L.F. Jayaweera, B.J. Mason, The behavior of freely falling cylinders flow past spheres, J. Fluid Mech. 65 (Part 1) (1974) 113–125.
and cones in viscous fluid, J. Fluid Mech. 22 (1965) 709–720. [40] R.S. Neve, F.B. Jaafar, The effect of turbulence and surface roughness on
[20] B.Y. Gorial, J.R. O'Callaghan, Aerodynamic properties of grain/straw the drag of spheres in thin jets, Aeronaut. J. 86 (1982) 331–336.
materials, J. Agric. Eng. Res. 46 (1990) 275–290. [41] A. Gauvin, W.H. Clamen, Effect of turbulence on the drag coefficients of
[21] E.B. Christiansen, D.H. Barker, The effect of shape and density on the free spheres in a supercritical flow regime, AIChE J. 15 (1969) 184.
settling of particles at high Reynolds number, AIChE J. 11 (1965) [42] Ahmadi G., Clarkson University, personal communication, 2005.
145–151. [43] W.W. Willmarth, N.E. Hawk, R.L. Harvey, Steady and unsteady motions
[22] R. List, Schemenauer, Free-fall behavior of planar snow crystals, conical and wakes of freely falling disks, Phys. Fluids 7 (1964) 197–208.
graupel and small hail, J. Atmos. Sci. 28 (1971) 110–115. [44] R.M. Turian, F.-L. Hsu, K.S. Avrarnidis, D.-J. Sung, R.K. Allendorfer,
[23] A. Haider, O. Levenspiel, Drag coefficient and terminal velocity of Settling and rheology of suspensions of narrow-sized coal particles, AIChE
spherical and non-spherical particles, Powder Technol. 58 (1989) 63–70. J. 38 (1992) 969–987.