Professional Documents
Culture Documents
A Holistic Approach To Particle Drag Prediction: Teresa L. Thompson
A Holistic Approach To Particle Drag Prediction: Teresa L. Thompson
Abstract
Although perfect spherical particles are not often encountered in the practical engineering world, most
particle drag research to date has centered around spherical particles. Unfortunately, there has been
no other way to predict irregular particle drag accurately due to the lack of reliable shape descriptors
for rough particles. This paper presents a unified diagram (analogous to the Moody diagram for pipe
friction factors) to predict particle drag coefficients. The diagram developed presents particle drag as
a function of Reynolds number and a hydrodynamic particle shape descriptor, called scruple @), which
characterizes the shape, roughness, or irregularity, of a particle. Several methods of calculating scruple
are presented. It may be found from such shape descriptors as sphericity, aspect ratio or harmonic
persistence, since data exist to relate these descriptors to drag coefficient. Also, it may be presented
directly as an approximate value for some given ore, object or shape, as found from experimental
data. Future experimental work on drag coefficients will serve to expand our prediction of scruple
and extend this simple approach to the prediction of particle drag.
0.1 I I
TABLE 1. Constants a and b to use in eqn. (9) and sum then divided into the three categories mentioned
of least-squares error above, and a least-squares fitting program was used
to determine the relationship between the two in
Class a b Error No. of each category. The basic equation remains the same
data points in the form of
Isometric 152.6 -2.14 0.020 4 3=alo6” (9)
Thin disks 263.5 -4.17 0.019 4
Isometric 158.4 -2.15 0.781 8 with the constants a and b changing for the categories
and disks as shown in Table 1. Table 1 also presents the
resulting least-squares error summed over the data
shape and by using shape descriptors such as aspect set and defined by
ratios, harmonic persistences and sphericities.
Error = &(log&,,t - log,,a lob”)’ (IO)
Shape measurement
Figure 2 compares the actual values of scruple ob-
Sphericity tained by eqn. (8) and the calculated values obtained
The relationship between scruple and sphericity by eqn. (9) for the different categories.
was found to differ, depending on the fundamental A direct relationship between the particle drag
particle shape. In other words, sphericity alone was coefficient and scruple was desirable, while main-
insufficient as a descriptor for drag coefficient pre- taining the form of eqn. (5). However, solving eqn.
diction. Based upon Haider and Levenspiel’s data (9) for sphericity (in terms of scruple) and substituting
[4], three categories of particles were examined: that into eqn. (5) resulted in three separate equations
isometric particles, thin disks and a combined data for drag that were difficult to handle. Therefore, the
set for both disks and isometric particles. variables A, B, C and D of eqn. (5) were calculated
Given the sphericity of the particles, eqn. (5) was using the sphericities for which exact values of scruple
used to calculate the drag coefficient, from which had previously been determined. The relationship
scruple could be determined through eqn. (8). The between scruple and each of the variables was then
corresponding values of sphericity and scruple were found using the least-squares fitting program. These
150
s
C
r
; 100
I
e
50
0 I , I I I I I a , .... ., m. Jq
I I I I
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
sphericity
Fig. 2. Comparison of eqn. (9) with actual values.
61
Particles A B C D
TABLE 3. Suggested values of A,B,C and D for use with eqn. (5)
Shape A B C D
4
S
C
I
u 3
P
I
e
2
Ra
relationships are summarized in Table 2, while sug- error in each case was less than 0.9, and in most
gested values of A, B, C and D for particular shapes cases was of the order of lo-’ or 10e3.
based on exact values of scruple rather than calculated The drag coefficient can now readily be found by
values, can be found in Table 3. The least-squares reading from the diagram, given scruple and Reynolds
62
TABLE 4. Values of scruple for various regular geometric harmonics analysis was then performed on the image
smooth particles, estimated from sphericity, using eqns. in which an arc of chosen length (the step length)
(5) and (8). The values of scruple found directly from
stepped around the edge of the image several times,
Pettyjohn and Christiansen’s data are: cubeoctahedron,
cube, octahedron and tetrahedron forming a polygon within the image. The selection
of step length determined the harmonic of an image
where a second harmonic indicates two opposing
Particle Proportions Scruple
points. The second harmonic persistence is defined
Sphere 1.00 as the largest step length divided by the shortest
Cubeoctahedron 2.08 step length that would result in a second harmonic.
Cube 3.36 From this, eqn. (7) was derived for the relationship
Octahedron 2.93 between drag coefficient in Newton’s regime of set-
Tetrahedron 4.75
tling and the sum of the second persistences, P,;,,
Cylinder h=d 2.46
of each of the three views of a stone. Solving eqn.
h=2d 2.99
h=4d 4.38 (7) for the drag coefficient,
h = OSd 2.98
CD = (0.222 33P,“,)h.*89 3 (11)
h = 0.25d 4.97
Rectangular 1:l:l 3.25 and using the definition of scruple, the relationship
parallelpiped 1:1:2 3.81 between the sum of the second harmonic persistences
1:2:2 3.81
and scruple becomes
1:1:4 5.12
1:4:4 5.76 8 = (0.000 185 97P,,,)6~28g 3 (12)
1:2:4 5.12
Spheroid 1:1:2 1.74
1:2:2 1.85 Not roundness
1:1:4 3.67 Beddow et al. [lo] used the shape correction factor
1:4:4 4.82 of not-roundness when determining irregular particle
Ellipsoid 1:2:4 3.53 drag. However, the data for particle drag are based
Thin disks d=25t 21.77
on an effective diameter different from the equivalent
d=50t 43.07
d=75t 62.03
diameter that the present analysis is based upon and
d = 1OOt 78.27 the volume of Beddow’s particles was unavailable,
Right prisms Triangular 0.74 making it impossible to rework his data. Also, his
r=h=a Quadrangular 1.42 data were taken in the laminar/transitional regime
Pentagonal 2.20 of flow, so the assumption that the drag coefficient
Hexagonal 2.88 was taken in Newton’s regime would not be valid.
Second harmonic persistence where C is the particle’s longest diameter and A,B
In a recent paper [ll], smooth pebble and rough are the particle’s longest diameters orthogonal to
gravel samples were analyzed using image analysis both each other and C.
techniques on three orthogonal views of the stones These three diameters correspond to the three
to define the edge of the stone’s image. A polygonal orthogonal views taken of smooth pebbles and rough
TABLE 5. Scruple values for cuboids extrapolated from McKay and McLain’s data [16]
Particle size Particle shape Sphericity Velocity Diameter Reynolds Drag Scruple
(mm/s) (mm) number coefficient
12.5 x 12.5 x 12.5 Cuboid 1:l:l 0.81 148 15.51 1281.0 1.76 4.43
12.5 x 12.5 x 19.0 Cuboid 1:1:1.52 0.79 158 17.83 1 572.2 1.77 4.43
12.5 x 12.5 x 29.0 Cuboid 1:1:2.32 0.75 175 20.53 2 005.0 1.67 4.13
12.5 x 12.5 x40.0 Cuboid 1:1:3.2 0.71 190 22.85 2 422.9 1.57 3.86
12.5 x 12.5 x 60.0 Cuboid 1:1:4.8 0.65 250 26.16 ,3 649.8‘ 1.04 2.60
12.5 x 12.5 x 80.0 Cuboid 1:1:6.4 0.60 304 28.79 4 884.3 0.774 1.94
63
Cuboids
6
7 --
6 --
5 --
4 --
3 --
2 --
1 --
I
0.6 0.7 0.0 0.9 1 1.1
Fig. 4. Comparison of scruple inferred from McKay and McLain [16] with eqn. (17) shown as a dotted line and scruple
exact (solid line).
64
(a) 0 I
0 4.049 452
D
@I 4.502 182
0
(c) 4.833 626
(e)
0 0 2.790 526
(0
0
4.155 78
TABLE 8. Summary of relationships between scruple and other shape descriptors for various particles
aspect
Cuboid 1:1:x a= -27.30+75.20+-44.40#
ratio
Rough rocks,
smooth pebbles 3.0 <P,, < 6.0 8 = (0.000 185 97P,,)6.~9’
Thin disks 8 =263.469*10-4.‘69”
Isometric particles 8 = 152.625*10-2~‘4’+
Isometric particles
3 = 158.435*10-2.‘s4$
and thin disks
Rough rocks, 8 = - 3.85 + 4.83R,
1.0 <R, < 2.0
smooth pebbles
values in Table 5 are found directly experimentally, based on equivalent spherical diameter, actually drops
they are to be favored over the values in Table 4, relative to a cube. For all of McKay and McLain’s
which were inferred through the medium of sphericity data, the best fit parabola was
using eqns. (5) and (8). Figure 4 shows the relationship
3 = - 27.30 + 75.20+ 44.403 (15)
between ly and 8 for McKay and McLain’s cuboids,
and gives the line of eqn. (8) as a reference. Values Table 6 provides additional values for scruple.
are credible for compact cuboids, but it is obvious Some of the data used in the paper by Clark et al.
that the agreement is very poor at high aspect ratios. [ll] was used to find average values of 3 for smooth
For high aspect ratios, the drag coefficient, when pebbles, rough rocks (similar to road or railroad
6.5
7 W. W. Willmarth, N. E. Hawk and R. L. Harvey, Whys. 13 E. J. Wasp et al., Solid-Liquid Flow Sluny Pipeline
Fluids, 7 (1964) 197. Transportation SeriesonBulkMaterials Handling, 1(1975/
77) No. 4, 39.
8 R. Turton and 0. Levenspiel, Powder Technol., 47
14 J. J. Comish III, in A. Avallone and T. Baumeister
(1986) 83. III (eds.), Marks’ Standard Handbook for Mechanical
9 E. B. Christiansen and D. H. Barker, AIChE J., II Engineers, McGraw-Hill, New York, 9th edn., 1987, pp.
(1965) 145. 1 l-76.
10 J. K. Beddow, A. D. Ah Chin, J. Portz, M. Ward and 15 W. F. Lindsey, E. A. John and W. L. Haberman (eds.),
A. F. Vetter, Powder Technol., 48 (1986) 59. Introduction to Fluid Mechanics, Prentice Hall, Engle-
wood Cliffs, NJ, 1988, p. 294.
11 N. N. Clark, P. Gabriele, S. Shuker and R. Turton,
16 G. McKay and H. McLain, Trans. Am. Inst. Chem.
Powder Technol., 59 (1989) 69. Eng., 58 (1980) 107.
12 T. Allen, Par?icle Size Measuremenf, Chapman and Hall, 17 D. Geldart, Gas Fluidization Technology, Wiley, New
New York, 1981. York, 1986.