Sei sulla pagina 1di 8

Particuology 9 (2011) 298305

Contents lists available at ScienceDirect

Particuology
journal homepage: www.elsevier.com/locate/partic

Experimental study of settling and drag on cuboids with square base


Jinsheng Wang, Haiying Qi , Junzong Zhu
Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Tsinghua University, Beijing 100084, China

a r t i c l e

i n f o

a b s t r a c t
The drag on non-spherical particles is an important basic parameter for multi-phase ows such as in biomass combustion, chemical blending, and mineral processing. Though there is much experimental research on such particles, there are few results for cuboids. This paper presents data for cuboids with a square base in static glycerinwater solutions of various volume concentrations. Complex motions were observed and characterized. A dimensionless expression is given for terminal velocity ut as a function of Archimedes number Ar which is used to develop an accurate correlation for friction factor CD . The accuracy of the correlation is 7.9% compared to experimental data in the literature. For both square plates and square rods, the terminal velocity per unit mass, ut /mp , was used to characterize the inuence of particle geometry on velocity, which was shown to be linear. 2011 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.

Article history: Received 12 July 2010 Received in revised form 12 October 2010 Accepted 22 November 2010 Keywords: Cuboid particles Unsteady settling behavior Archimedes number Terminal velocity Drag coefcient

1. Introduction Two-phase ow plays an important role in both technical and natural processes, such as combustion of pulverized coal, suspended bre ows in paper forming, chemical blending, oil drilling and mineral processing. To optimize processes and to design equipment, the terminal velocity and drag coefcient are signicant parameters. Early studies focused on spherical particles with accurate correlations for the drag coefcient and terminal velocity based on experimental data and theoretical studies (Clift, Grace & Webber, 1978). However, non-spherical particles have not been as well characterized even though they are common in nature and in industrial processes. In addition, unsteady sedimentation of nonspherical particles involves complex motions, such as swinging and rotation, for which drag coefcients for non-spherical particles are quite different from those for spherical particles. Non-spherical particles could be either regular or irregular in shape. Although most particles in industry and in nature are irregular, it is important rst to study particles with regular shapes for drag of non-spherical particles. Based on research of regular particles, the shape description and drag model could then be extended to irregular particles. As non-spherical particles, cuboids with a square base provide a typical shape for which experimental data could be expediently analyzed. However, not much experimental data for particles with such shapes are available in the literature as compared to disks

and cylinders. Based on study of particles of typical shapes, it is possible rst to develop suitable drag models for regular particles, which may be improved for irregular particles. The primary objective of this paper is to present a reliable correlation to predict the terminal velocity and drag coefcient for cuboids with a square base. The sphericity of these particles varies from 0.642 to 0.805, and aspect ratios from 0.3 to 5. The relationship between terminal velocity and particle geometry is subsequently analyzed, in order to enrich experimental research on non-spherical particles. 2. Previous work 2.1. Shape description The drag coefcients for spheres were initially obtained experimentally with sphere diameter, dp , as the characteristic length. Usually, studies on non-spherical particles use the volumeequivalent spherical diameter, dv , as the characteristic length. However dv does not uniquely describe the particle shape. Therefore shape description is a signicant issue in two phase ow studies with non-spherical particles. Generally, the aspect ratio, E, was used to describe the axisymmetric particles, such as cylinders and spheroids. This factor was dened as the ratio of the length projected on the axis of symmetry to the maximum diameter normal to the axis (Clift et al., 1978; Gabitto & Tsouris, 2008). Apart from aspect ratio various shape factors have been dened and used to describe shape effect of non-spherical particles, regular or irregular, including particle sphericity, , particle circularity, , Corey shape factor, , and

Corresponding author. Tel.: +86 10 62796036; fax: +86 10 62796036. E-mail address: hyqi@mail.tsinghua.edu.cn (H. Qi).

1674-2001/$ see front matter 2011 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.

doi:10.1016/j.partic.2010.11.002

J. Wang et al. / Particuology 9 (2011) 298305 Table 1 Shape description formulas available in literature for non-spherical particles. Shape description Aspect ratio Sphericity Circularity Corey shape factor Stokes shape factor Stokes Symbol E Denition Ratio of particle length projected on symmetry axis to maximum diameter normal to the axis Av = Ap = PA /Pp = dA /Pp = c/(ab)1/2 stokes =
18 f ut
2 ( p f )gdv

299

Source Clift et al. (1978) and Gabitto and Tsouris (2008) Wadell (1933) Tran-Cong, Gay, and Michaelides (2004) Tran-Cong et al. (2004) Pettyjohn and Christiansen (1948), Xie and Zhang (2001), Ganser (1993), and Bouwman, Bosma, Vonk, Wesselingh, and Frijlink (2004)

Av , surface area of volume-equivalent sphere; Ap , particle surface area; PA , projected perimeter of area-equivalent sphere; Pp , projected particle perimeter; a, b and c, the largest, intermediate and shortest particle axes (a > b > c); p and f , densities of particle and uid; g, gravitational acceleration; f , uid viscosity; dv , diameter of volume-equivalent sphere.

Table 2 Values of ai in Eq. (1). a0 K2 K3 K4 2.3288 4.905 1.4681 a1 6.458 13.8944 12.2584 a2 2.4486 18.42222 20.7322 a3 0.0 10.2599 15.8855

Table 3 Particle geometries (mm). No. P1 P2 P3 P4 Height (h) 3 5 8 10 10 15 20 15 20 25 30 4 4 4 6 6 6 8 8 8 Base length (a) 10 10 10 10 12 12 12 15 15 15 18 18 18 20 20 20

Stokes shape factor, Stokes , which are summarized in Table 1 along with the available reference sources. However, these methods to describe the particle shape are only suitable for one or several xed shaped particles. Thus, a more powerful shape factor is needed for a general drag model handling many kinds of non-spherical particles. 2.2. Drag coefcient for non-spherical particles There are many studies of drag coefcients for xedly, regularly shaped particles reported in the literature (Fan, Yang, Yu, & Mao, 2003; Guo, Lin, & Nie, in press; Heiss & Coull, 1952; Ku & Lin, 2008; Marchildon, Clamen, & Gauvin, 1964; Militzer, Kan, Hamdullpur, Amyotte, & Al-Taweel, 1989; Tripathi, Chhabra, & Sundararajan, 1994; Zhu, Lin, & Shao, 2000). These drag models are accurate only for the selected shapes and orientations. Holzer and Sommerfeld (2008) summarized the drag coefcients as a function of particle Reynolds number for various shaped particles, as shown in Fig. 1. The correlation proposed by Haider and Levenspiel (1989) for spherical and non-spherical particles based on more than 500 experimental data points is presented as follows: CD = 24 K3 (1 + K2 ReK1 ) + , Re 1 + K4 /Re Kj = exp
i=0

Panigrahi, and Muralidhar (2008) studied experimentally the ow past a square cylinder with different orientations, showing a minimum drag coefcient at an orientation of 22.5 . Ku and Lin (2008) found that cubic particles rotated faster than cylindrical particles with the same size in pipe ow. In addition, Delidis and Stamatoudis (2009) compared the velocities of cubes and spheres in the accelerating region for very high Reynolds number, showing that the velocity of spheres was always greater than that for the cubes of the equal volume. In the above mentioned studies, the particle shapes were limited to cubes or square cylinders, but, to the best of authors knowledge, no correlations applicable to drags for cuboid particles both plates and rods have yet been reported in the literature.

K1 = 0.0964 + 0.5565 ,

ai

, (j = 2, 3, 4) (1)

with the values of ai given in Table 2. Although this expression greatly improves predictions of particle sedimentation, it still cannot be applied in practice because of its poor accuracy. For cuboids, Albertson (1952), Alger and Simons (1968) provided data for settling of cubes in uid. Agarwal and Chhabra (2007) presented extensive data of cubes in Power law liquids. Dutta,

Fig. 1. Drag coefcients as a function of particle Reynolds number for different shaped particles (Holzer & Sommerfeld, 2008).

300

J. Wang et al. / Particuology 9 (2011) 298305

2.3. Terminal velocity of non-spherical particles The terminal velocity, ut , can be calculated based on the force balance on particles: ut = 2mp g(
p

f)

p f Ap CD

(2)

where mp is the particle mass and g the gravitational acceleration. Since CD is a function of terminal velocity, the determination of ut requires iteration. Terminal velocity ut can also be calculated in terms of other dimensionless numbers such as the Archimedes number dened as: Ar =
3 dv f( p 2 f f )g

(3)
Fig. 2. Schematic geometry of the particles to be studied.

Terminal velocities for various minerals have been correlated in this way (Ganguly, 1990; Tsakalakis & Stamboltzis, 2001). 3. Experimental The present experiments used a rectangular parallelepiped test vessel, 1500 mm 300 mm 300 mm, with 87 cuboids made of aluminum (2703.2 kg/m3 ), PVC (1510.3 kg/m3 ) and PMMA (1170.8 kg/m3 ) having different base lengths, a, and heights, h, as listed in Table 3 and shown in Fig. 2. The particles are divided into four categories marked P1P4, each with either a or h varied once a time. Six glycerinwater solutions of different concentrations were used as the test uids as shown in Table 4. During experiments, temperature uctuated from 18 to 21 C, small enough to be considered negligible. When a particle reaches its terminal velocity, the gravitational force, buoyancy and drag are balanced as: FD = Fg Fb = 6
3 dv g( p

f ),

(4)

where dv , f , and p are chosen from Tables 3 and 4, so that CD can be calculated from Eq. (4) once the terminal velocity, ut , is known. The terminal velocity was measured with a high-speed camera in the experiments, which recorded the time lapse for the particle falling a certain distance. The terminal velocities were measured in two adjacent regions, both located sufciently far from the end for neglecting end effects. The two measurements are considered acceptable only when the difference between the two was less than 5%, and ut is the average of the two values. Wall effect was eliminated by choosing only the data of particles with settling oscillation amplitude less than 10% of the crosssectional dimension. Each particle was dropped three times with repeatability of results of about 2%. The system was calibrated with a glass sphere ( = 2539.8 kg/m3 ) of 16.17 mm in diameter with the experimental drag coefcients within 2.3% of the expected values (Clift et al., 1978).

Table 4 Properties of glycerin aqueous solution at 20 C. No. Concentration C (vol%) Density f (kg/m3 ) Dynamic viscosity f (103 Pa s) F0 0 971 1.523 F20 20 1028.8 3.064 F40 40 1086.6 5.203 F60 60 1144.4 14.87 F80 80 1198.4 82.77 F100 100 1260 872.8

Fig. 3. Typical unstable settling patterns. (a) Too close to the wall (a = 4 mm, h = 10 mm, p = 2703.2 kg/m3 , F0); (b) periodic movements with a small swing for a square plate. (a = 10 mm, h = 3 mm, p = 1510.3 kg/m3 , F40); (c) periodic movements with a small swing for a square rod (a = 10 mm, h = 15 mm, p = 1510.3 kg/m3 , F40).

J. Wang et al. / Particuology 9 (2011) 298305

301

Fig. 4. Motion stability diagram. (\ means no sedimentation occurred because the particle density was less than uid density.).

Fig. 5. Inuence of particle geometry on ut for F60.

Fig. 6. Relationship between Rep and Ar for all cases.

Fig. 7. Relationship between CD and Rep for all the small-oscillation cases as compared with that for spheres.

302

J. Wang et al. / Particuology 9 (2011) 298305

Fig. 8. Comparison between the present model and data in literatures (Albertson, 1952; Alger & Simons, 1968).

4. Results and discussion 4.1. Unstable settling patterns During experiment, various patterns of settling behavior were observed as shown in Fig. 3. Once released and within a few seconds, the particles tended to adjust themselves with their maxi-

mum projected areas normal to the direction of motion. In general, most of the particles did not fall along the center line of the settling column while swinging and rotating, especially in low-viscosity uids. In some extreme cases, the particles even ran up against the wall. Such behaviors were found to be independent of initial particle orientation. In low-viscosity uids such as F0, F20, F40 and F60, particles of different shapes behaved differently at the terminal velocity. Movements with small oscillations were periodic: the square plates spiraled down with a sharp angle between the normal to the particle base and the center line of the column as shown in Fig. 3(b). Square rods zigzagged down with the principal axis perpendicular to the vertical as shown in Fig. 3(c). Cubes rotated almost continually following a complex helical path. However, in high-viscosity uids F80 and F100, the high viscosity caused all particles to settle in a stable manner with little oscillations. Motion of particle settling is classied into three categories on the basis of the extent of swing, as stable settling, oscillations less than 30 mm and oscillations larger than 30 mm, as shown in Fig. 4. The repeatability of the data in the rst two categories is well, and they were referred to as limited oscillation. The diagram shows an obvious demarcation between cases of small and large oscillation. The oscillations were dependent on uid viscosity and particle density, size and shape. Large oscillation occurred for low viscosity uids and large particles with plates

Fig. 9. Inuence of particle geometry on ut /mp for F60.

J. Wang et al. / Particuology 9 (2011) 298305

303

were more stable than rods and cubes. This diagram distinguishes all the cases and provides guidance for further studies. The mechanism for the different behavior is an important topic for further study. 4.2. Terminal velocity The terminal velocity was calculated for the small oscillation cases. The inuence of geometry on the terminal velocity was studied for the cases using F60 as examples. Fig. 5 uses either a or h as the abscissa to show two zones for plates and for rods respectively, divided in between by the cubes of h/a = 1. The data in Fig. 5 for F60 show that ut is larger for the larger particle densities, but as the particle size increases, ut rst increases and then remains almost constant. The turning point occurs at the cube geometry. One reason is that both the projected area and the mass change with the geometry, but they have opposite effects on the velocity. Another reason is that the spiral path is longer than the zigzag path, so the velocity in the gravitational direction along the spiral path is less than the one along the zigzag path. Since both uid and particle characteristics affect the terminal velocity, using one expression to describe all conditions is difcult. The particle Reynolds number, Rep , dened by Eq. (5), is the ratio of the inertia force to the viscous force which can represent a dimensionless velocity. Rep =
f dv ut f

(5)

The Archimedes number, Ar, represents the ratio of the buoyant force to the viscosity force, which includes particle geometry and density as well as uid density and viscosity as dened in Eq. (3). The plot of Rep versus Ar in Fig. 6 shows that all data are well represented by a single curve. As a result, the data can be well correlated by Eq. (6) with R2 = 0.997. The coefcients in Eq. (6) are listed in Table 5. Rep = exp(k1 Ar 1/k2 + k3 Ar 1/k4 + k5 ) 4.3. Drag coefcients Fig. 7 shows the relationship between CD and Rep for all the small-oscillation cases. The relationship is similar to but slightly above the curve for spheres. The data for CD deviate from that for spheres for large Rep , because the effect of the particle shape on the ow eld increases and oscillation also increases with Rep . For Rep > 100, CD remains almost constant at 2 as the difference from the spherical data increases. There are also larger differences compared with the spherical correlation in the region of small Rep . For Rep < 0.2, the difference gradually increases, different from the reported results in the literature. Additional experiments with spherical particles indicated that the measured CD for spheres was only 2.3% higher than that reported in the literature. This implies that the differences for the non-spherical particles resulted from the shape characteristics of the rectangular prisms. A force balance can relate CD with Rep and Ar as: CD = 4 dv ( 3
p

Fig. 10. Relationship between ut /mp and 1/a or 1/(ah) for F60.

(6) Combining Eqs. (7) and (8) then gives the drag coefcient as: CD =
1/B 1/B 4 exp(A1 Rep 1 + A2 Rep 2 + C) . 2 3 Rep

(9)

f )g

2 f ut

4 Ar . 3 Re2 p

(7)

Thus, CD can be correlated with Rep through the relationship of Ar and Rep . The data in Fig. 8 can be used to express Ar as a function of Rep with R2 = 0.994:
1/B 1/B Ar = exp(A1 Rep 1 + A2 Rep 2 + C),

(8)

with the coefcients A1 , A2 , B1 , B2 and C listed in Table 6.

Thus, Eq. (9) can be used to represent the drag coefcient for cuboids with square base for Rep from 0.05 to 6300. Other experimental data in the literature (Albertson, 1952; Alger & Simons, 1968) were used to verify the applicability of the drag correlation in the present study, as shown in Fig. 8. The mean relative error is 7.9%, indicating that the drag correlation is suitable for cuboids beyond their current experimental range. As shown in Fig. 5, particle mass and particle volume have opposite effects on terminal velocity. To eliminate the effect of particle mass, the terminal velocity per unit mass, ut /mp , was used to correlate the data with the particle geometry as shown in Fig. 9 for F60, indicating that ut /mp decreases as particle size increases yet with a gradually decreasing gradient. The curves for the different heights in Fig. 9(d) are approximately parallel. Thus, when ut /mp is plotted against 1/a as shown in Fig. 10(a), ut /mp is found to be approximately inversely proportional to a. Fig. 11(b) shows a nearly linear relationship between ut /mp and 1/(ah), with the slope increasing with particle density, p . This relationship is suitable for both plates and rods. Thus, there is essentially no difference between the terminal velocities for the different shapes when the shape varies in any one dimension.

304 Table 5 Coefcients in Eq. (6). k1 0.03568 k2 5.043

J. Wang et al. / Particuology 9 (2011) 298305

k3 15.32

k4 10.72

k5 10.51

Table 6 Coefcients in Eq. (8). A1 0.00136 B1 0.5656 A2 2.938 B2 12.32 C 4.308

(1) Once released into a uid, and within a few seconds, particles adjusted themselves with their maximum projected areas oriented normal to the descent direction. Various complex motions occurred during sedimentation, including oscillation, rotation and twirling. Square plate particles tended to follow a spiral path while square rod particles followed a zigzag path. These motions were independent of initial particle orientation. (2) The extent of particle oscillation was plotted to classify the different behaviors. Such regime diagram serves to identify the various motion regimes to provide guidance for further studies. (3) The Archimedes number, Ar, was found to be related exponentially to the Reynolds number, Rep . A dimensionless relationship was developed to predict ut . A correlation between CD and Rep was developed from the relationship between Ar and Rep . This correlation ts the experimental data of other researchers with mean relative error of 7.9%. (4) Even at low Rep , there is still obvious difference in CD for cuboids and for spheres, indicating that shape characteristics led to larger pressure drag for cuboids than for spheres. (5) The parameter, terminal velocity per unit mass, ut /mp , highlights the effect of particle geometry. Analysis resulted in a linear relationship between ut /mp and particle geometry, indicating essentially no difference in terminal velocity between different shapes while any one dimension is varied.

Acknowledgement This work was supported by the Major Program of the National Natural Science Foundation of China with Grant No. 10632070.

References
Fig. 11. The same relationship between Rep and Ar for plates and rods when having the same base. Agarwal, N., & Chhabra, R. P. (2007). Settling velocity of cubes in Newtonian and Power law liquids. Powder Technology, 178, 1721. Albertson, L. (1952). Effect of shape on the fall velocity of gravel particles. In Proceeding of the 5th Hydraulic Conference Iowa City, (pp. 243261). Alger, G. R., & Simons, D. B. (1968). Fall velocity of irregular shaped particles. Journal of Hydraulics Engineering Division, 94, 721737. Bouwman, A. M., Bosma, J. C., Vonk, P., Wesselingh, J. A., & Frijlink, H. W. (2004). Which shape factor(s) best describe granules. Powder Technology, 146, 6672. Clift, R., Grace, J. R., & Webber, M. E. (1978). Bubbles, drops and particles. New York: Academic Press. Delidis, P., & Stamatoudis, M. (2009). Comparison of the velocities and the wall effect between spheres and cubes in the accelerating region. Chemical Engineering Communications, 196, 841853. Dutta, S., Panigrahi, P. K., & Muralidhar, K. (2008). Experimental investigation of ow past a square cylinder at an angle of incidence. ASCE Journal of Engineering Mechanics, 134, 788803. Fan, L., Yang, C., Yu, G. Z., & Mao, Z. S. (2003). Empirical correlation of drag coefcient for settling slender particles with large aspect ratio. Journal of Chemical Industry and Engineering (China), 54, 15011503 (in Chinese). Gabitto, J., & Tsouris, C. (2008). Drag coefcient and settling velocity for particles of cylindrical shape. Powder Technology, 183, 314322. Ganguly, U. P. (1990). On the prediction of terminal settling velocity of solids in liquidsolid systems. International Journal of Mineral Processing, 29, 235247. Ganser, G. H. (1993). A rational approach to drag prediction of spherical and nonspherical particles. Powder Technology, 77, 143152.

4.4. Universal drag model for non-spherical particles The above discussion indicates that particles with a square base have the same relationship between Ar and Rep for both plates and rods. Additional experimental data and the data from literature (Gogus, Ipekci, & Kokpinar, 2001) showed that this rule is suitable for particles having a different base such as circular or triangular as shown in Fig. 11, thus, promising drag coefcient correlations for particles with the corresponding bases to lead to corresponding drag models. Such results will be discussed in a further study. 5. Conclusions An experimental study of the sedimentation characteristics of cuboids with a square base showed that:

J. Wang et al. / Particuology 9 (2011) 298305 Gogus, M., Ipekci, O. N., & Kokpinar, M. A. (2001). Effect of particle shape on fall velocity of angular particles. Journal of Hydraulic Engineering-ASCE, 127, 860869. Guo, X. H., Lin, J. Z., & Nie, D. M. (in press). New formula for the drag coefcient of cylindrical particles. Particuology. Haider, A., & Levenspiel, O. (1989). Drag coefcient and terminal velocity of spherical and nonspherical particles. Powder Technology, 58, 6370. Heiss, J. F., & Coull, J. (1952). The effect of orientation and shape on the settling velocity of non-isometric particles in a viscous medium. Chemical Engineering Progress, 48, 133140. Holzer, A., & Sommerfeld, M. (2008). New simple correlation formula for the drag coefcient of non-spherical particles. Powder Technology, 184, 361365. Ku, X. K., & Lin, J. Z. (2008). Motion and orientation of cylindrical and cubic particles in pipe ow with high concentration and high particle to pipe size ratio. Journal of Zhejiang University Science, 9, 664671. Marchildon, E. K., Clamen, A., & Gauvin, W. H. (1964). Drag and oscillatory motion of freely falling cylindrical particles. Canadian Journal of Chemical Engineering, 42, 178182.

305

Militzer, J., Kan, J. M., Hamdullpur, F., Amyotte, P. R., & Al-Taweel, A. M. (1989). Drag coefcient for axisymmetric ow around individual spheroidal particles. Powder Technology, 57, 193195. Pettyjohn, E. S., & Christiansen, E. B. (1948). Effect of particle shape on free settling rates of isometric particles. Chemical Engineering Progress, 44, 157172. Tran-Cong, S., Gay, M., & Michaelides, E. E. (2004). Drag coefcients of irregularly shaped particles. Powder Technology, 139, 2132. Tripathi, A., Chhabra, R. P., & Sundararajan, T. (1994). Power law uid ow over spheroidal particles. Industrial and Engineering Chemistry Research, 33, 403410. Tsakalakis, K. G., & Stamboltzis, G. A. (2001). Prediction of the settling velocity of irregularly shaped particles. Minerals Engineering, 14, 349357. Wadell, H. (1933). Sphericity and roundness of rock particles. Journal of Geology, 41, 310331. Xie, H. Y., & Zhang, D. W. (2001). Stokes shape factor and its application in the measurement of sphericity of non-spherical particles. Powder Technology, 114, 102105. Zhu, Z. F., Lin, J. Z., & Shao, X. M. (2000). An experimental study of the force and torque on cylindrical particle in simple uid. Journal of Zhejiang Institute of Science and Technology, 2, 116119 (in Chinese).

Potrebbero piacerti anche