Beruflich Dokumente
Kultur Dokumente
ABSTRACT KEYWORDS
The particle shape is an important factor playing critical role in evaluation of the interactions between CFD–DEM method;
particles in high-concentration particle-fluid flows. In this paper, the well-known multisphere (MS) multisphere method;
approximation approach and the novel rolling resistance approach are utilized to examine their nonspherical particle;
particles transportation;
performance in order to simplify the generalized shaped particle’s interactions within the framework of
rolling resistance method
discrete element method (DEM) and computational fluid dynamics (CFD). The performance of two
approaches are compared with the perfect particle’s shape geometry, for the limited cases of cubic-
shaped and disk-shaped particle flows in a horizontal well drilling process as a reference scenario.
Deviation of the MS approximation shape from the perfect particle geometry is evaluated by comparison
of macroscopic properties of nonspherical particle. It is determined that the data on macroscopic
parameters yielded by the MS model tend to converge to those of the smooth particle with the
increasing number of the subspheres. Moreover, the effectiveness of rolling resistance method is
investigated by comparison of macroscopic properties of nonspherical particles with approximated MS
approach and approximated spherical particles subjected to the rolling friction. The results show that the
updated rolling resistance model can reduce the inaccuracy in the prediction of the particles deposit
originated from the spherical shape idealization fairly and can be considered where the cost of
computations is a restrictive issue.
CONTACT Siamak Akhshik Akhshik@mehr.sharif.ir Department of Mechanical Engineering, Sharif University of Technology, Azadi Avenue, Tehran P.O. Box
11155-9313, Iran.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/upst.
© 2016 Taylor & Francis
382 S. AKHSHIK ET AL.
particles shape effects, in one the hand, and the spherical shape Herschel–Bulkley model is used. In this model, the dynamic
idealization of the particles, on the other hand, is a desire. viscosity is considered as (Huilgol and You 2005):
For this purpose, a novel rolling resistant model is utilized (
l ¼ lyield when c_ < lsyield 0
;
in order to introduce the particles shape effect on the rolling n n ð4Þ
s0 þkðc_ ðs0 =lyield Þ Þ
motion of idealized spherical particles and update the l¼ c_ when c_ � ls0 ;
yield
traditional model for the nonspherical geometries. Finally,
the effectiveness of the presented approach is examined by where k is the consistency factor, n is the power law exponent,
an illustrative example. τ0 denotes the yield stress threshold, lyield means the yielding
viscosity, and c_ is the strain rate magnitude.
Mathematical formulations
Particle phase
A three-dimensional (3D) model is prepared to study the flow
of particles in a deviated well drilling based on the CFD and Multisphere method
DEM. In this model, the shapes of particles are generalized The rotation of spherical particles is restricted only by
as nonspherical in addition to common spherical type. The frictional forces between the particles and also with the other
fluid phase is considered as an incompressible fluid. The dis- surfaces, whereas for nonspherical particles, the rotation is
persed phase (particles phase) is treated as a collection of indi- affected by mechanical locking in addition to the frictional
vidual particles whose movements are governed by applying contact. For spherical particles, only the tangential forces lead
Newton’s second law. The continuum fluid phase and the dis- to the rotation of particles, and normal contact does not more
persed particles phase are strongly coupled via the interaction contribute to the moment and rotation, since they always act
forces such as the drag force, lift force, and pressure gradient through the center of the spherical particles. This is not the
force that are taken into account in the developed CFD–DEM case for nonspherical particles for which the rotation can be
model. Furthermore, the particles collisions are simulated by as a result of both normal and tangential forces. Therefore,
DEM. In the following sections, the governing equations spherical particles may not be a good representative for non-
of motion for the fluid and particles phases are presented, spherical particles. In order to consider the possible influence
followed by a description of the input data for CFD and of nonspherical shape of particles on the transportation
DEM simulations used in the present study. process, the MS approach is utilized for construction of
nonspherical particles.
In this approach, a nonspherical particle is constructed by
Fluid phase
rigidly connected spheres (Kruggel-Emden et al. 2008). The
The 3D governing equations of motion of unsteady, viscous, contact is detected if the distance between the centers of the
and incompressible fluid phase are described by the local two neighboring spheres is equal to the sum of their radii.
averaged Navier–Stokes equations on a computational cell The advantage of this method is that it provides an approxi-
scale with source terms. The mass conservation equation is mation of the actual nonspherical particle while maintaining
expressed as computational efficiency and simplicity of the spherical
geometry. It must be noted that this advantage comes at the
@ðaqf Þ
þ r � ðaqf uf Þ ¼ 0; ð1Þ expense of increased total number of spheres, which increase
@t the computational cost of the simulations.
where uf is the fluid velocity, ρf is the fluid density, and α is In this paper, four different shapes of particles are prepared
the volume fraction of the fluid phase. The momentum to investigate their behavior on the particles transport process.
conservation equation is given as These particle samples are shown in Figure 1, where S1 is
a common sphere model (where S denotes to “Sphere”), S6
@ðaqf uf Þ
þ $ � ðaqf uf uf Þ ¼ a$p þ a$ � s Sf þ aqf g; ð2Þ is a nonsmooth spherical particle that consists of six spherical
@t elements, D48 uses 48 spherical elements which are located on
where p is the fluid pressure, τ is the viscous stress tensor, and circular paths and construct a disk-shaped particle, and finally,
Sf is the volume-averaged (on a cell) interaction forces.
The source term Sf for a particular computational cell was
calculated by summing the fluid interaction forces on all par-
ticles present within that cell, and dividing by the volume of
the fluid computational cell:
!
X M
Sf ¼ Ff ;i IVcell ; ð3Þ
i¼1
�
Lift force can transmit a torque Tpr not exceeding a maximum value
The lift forces including the shear lift force (Saffman) and of lrReq|Fn|, where lr is the rolling friction coefficient and
the rotational lift force (Magnus) are perpendicular to the Req is the radius of a sphere with the same volume of
direction of the relative velocity between the particles and the nonspherical particle (i.e., equivalent radius). Relative
fluid. The shear lift force (Saffman) exerted on pth particle is rotation �between two particles in contact is allowed if
given by (Saffman 1965; Mei 1992; Sommerfeld 2000). Tpr ¼ Tpr max ¼ lr Req jFn j.
qf p 3 � � � The contribution of the particle shape on the rolling
Fs ¼ CLS d p uf up � x f ; ð26Þ resistance torque is due to the effect of an offset between the
8
normal line of the surface at the contact point and the mass
where ωf is the curl of the fluid velocity (ωf ¼ ∇ � uf) and center of nonspherical particle, as illustrated in Figure 4.
(Sommerfeld 2000) Obviously, this offset is zero for the ideal case of spherical
p
4:1126 geometry. This eccentricity, de , leads to an extra torque (i.e.,
CLS ¼ f ðReHB ; Res Þ; ð27Þ
Re0:5
s
in addition to the hysteresis effect) about the center of mass
f ðReHB ; Res Þ ¼ related to the normal force within the contact as
( �
ð1 0:3314 b0:5 Þe ReHB =10 þ 0:3314 b0:5 ReHB � 40 Tpr eccentricity ¼ rp � Fn ; ð32Þ
;
0:0524ðb � ReHB Þ0:5 ReHB > 40 � � �
� � p
where � Tpr eccentricity � ¼ de jFn j. In order to idealize the
ð28Þ
nonspherical particles with the spherical ones with the same
in which β ¼ 0.5Res/ReHB (0.005 < β < 0.4) and Reynolds volume and concurrently, to consider the shape effect, this
number for shear flow is Res ¼ qf dp2 jxf j=l. extra torque is substituted by an extra rolling friction force
The rotational lift force (Magnus) exerted on pth particle is at the contact point with the value of μshape Req|Fn| equal to
calculated as (Oesterleé and Bui Dinh 1998; Sommerfeld 2000) p
de jFn j. Therefore, the rolling friction coefficient should be
� � updated as
p 2 X � ðuf up Þ
FM ¼ dp qf CLM juf up j ; ð29Þ
8 jXj lr ¼ lr þ lshape ; ð33Þ
where CLM is the coefficient of rotational lift (Sommerfeld
p
2000), given by where lshape ¼ de =Req :
� � Due to the dependency of the equivalent rolling friction
Rer 0:4 0:3
CLM ¼ 0:45 þ 0:45 e 0:568 Rer ReHB : ð30Þ coefficient to the position of contact point, different estimation
ReHB for it exists. Therefore, four different values between the
minimum and the maximum estimated values are selected
for simulations. The nominal radius of approximated spherical
Pressure gradient force
particles is calculated from the following:
The pressure gradient force is expressed as
rffiffiffiffiffiffi
Fp ¼ Vp $p; ð31Þ 3 3V
Req ¼ ; ð34Þ
where Vp indicates the volume of the pth particle and ∇p is the 4p
gradient of the static pressure of the fluid phase at the location
where V represents the volume of a nonspherical particle.
of particle.
Figure 7. Particles concentration versus annulus inclination (angle) for selected drill pipe rotation speed and fluid inlet velocity.
Particle deposit along the well hole The effects of particle shape on concentration profile
Figure 8 demonstrates the process of particles deposit through The simulation of steady particle flow patterns and their
the well hole. It is seen that the particles gradually accumulate concentration profile at a same position along the well hole are
and the bed grows up to a specific height that the velocity of shown in Figure 9 for spherical, cubic, and disk-shaped particles.
fluid flow reaches a critical value. At this point, the fluid velo- As it is seen, the particles with the disk shape demonstrate
city is so high that the particles cannot deposit and the hori- higher concentration in comparison with cubic and spherical
zontal particle transport is guaranteed. ones. It is due to the fact that although the particles are
As the time passes, the phenomenon of particle deposit approximately in the same size, but nonspherical particles
continues along the total length of well hole. Ultimately, a are not sliding easily on each other. In general, it can be
steady-state condition is achieved and the bed height becomes inferred that as the shape of particles are deviated from
constant throughout the entire length of the hole. Similar the ideal spherical (i.e., their sphericity decreases), their
behavior has been reported by experimental work of Tomren, concentration increases.
Iyoho, and Azar (1986).
Figure 11. Effect of particle shape on the particles concentration for different levels of accuracy of multisphere approach.
Figure 12. Effect of particle shape on the particles concentration for spherical particles with rolling resistance model.
PARTICULATE SCIENCE AND TECHNOLOGY 389
Figure 13. Effect of rolling friction coefficient related to disk-shaped particle on particles concentration in different hole inclination and different drill
pipe rotation.
Figure 14. Effect of rolling friction coefficient related to cubic-shaped particle on particles concentration in different hole inclination and different drill
pipe rotation.
390 S. AKHSHIK ET AL.
reduced number of component spheres allowed the simulation shape idealization of the particles, on the other hand, is a
to run about five times as fast. desire. For this goal, a rolling friction method has been utilized
in order to simplify the geometry of the nonspherical particles
as the idealized spherical ones, taking into account the shape
The rolling friction updated model and the shape
effects, by implementation of an extra torque at the contact
idealization of particles
points in addition to the common hysteresis torque effect.
Comparison between the MS approximation method and the The main advantage of the rolling friction-based model updat-
rolling resistance method for disk-shaped and cubic-shaped ing is due to the computational cost-effectiveness. Considering
particles are shown in Figure 12. Based on various contact some numerical examples, it has been shown that the rolling
points and eccentricity distances, various frictions coefficients resistance model can compensate the errors of the prediction
are obtained. The maximum rolling friction coefficients for the of the particles hold-up, originated from the spherical
disk-shaped and cubic-shaped particles are computed as 0.95 shape idealization in traditional simulations, to some extent.
and 0.32, respectively. Four rolling friction coefficients are Obviously, the introduced model must be updated
considered in the simulations. As seen from this figure, as further, especially for the more complex shapes of particles.
the rolling friction coefficient increases, the particle volume The developed model may be considered as a reliable
concentrations approach the results of the MS method (i.e., prediction tool and cost-effective approach with respect to
the true simulation of the nonspherical particles). This trend the MS method.
is repeated for all selected fluid inlet velocities. Particularly,
the time required for the simulation of particle transportation
using rolling resistance model is one tenth of the time required References
using MS method.
Akhshik, S., M. Behzad, and M. Rajabi. 2015a. CFD–DEM approach
Figures 13 and 14 examine the particle concentration for to investigate the effect of drill pipe rotation on cuttings transport
selected inclination angles (0°, 20°, 40°, 60°, 80°) and different behavior. Journal of Petroleum Science and Engineering 127 (1):229–244.
inlet velocity (cases (a) 0.58, (b) 0.87, (c) 1.16 m/sec) at con- doi:10.1016/j.petrol.2015.01.017.
stant rotation velocity ω ¼ 50 rpm. For lowest inlet velocity Akhshik, S., M. Behzad, and M. Rajabi. 2015b. CFD–DEM model
for simulation of non-spherical particles in hole cleaning process.
for case (a), the particle concentration faces with a jump
Particulate Science and Technology 33 (5):1–10. doi:10.1080/02726351.
(remarkable increase) at inclination angles near 40° which this 2015.1010760.
happening has been reported experimentally by Tomren, Alexandrou, A. N., P. Le Menn, G. Georgiou, and V. Entov. 2003. Flow
Iyoho, and Azar (1986). Clearly, with decreasing inclination instabilities of Herschel–Bulkley fluids. Journal of Non-Newtonian
from 80° to 40°, where the particle deposit phenomenon Fluid Mechanics 116 (1):19–32. doi:10.1016/s0377-0257(03)00113-7.
Chien, S.-F. 1994. Settling velocity of irregularly shaped particles. SPE
comes up further, the differences between the results of the
Drilling & Completion 9 (4): 281–289. doi:10.2118/26121-pa.
rolling resistance model and the MS method are increased. Di Renzo, A., and F. P. Di Maio. 2004. Comparison of contact-force
However, by increasing the inlet fluid velocity, the differences models for the simulation of collisions in DEM-based granular flow
are decreased. Particularly, for the case of the cubic particles, codes. Chemical Engineering Science 59 (1):525–541. doi:10.1016/
as it is observed in Figure 14, the results of the MS method j.ces.2003.09.037.
Favier, J. F., M. H. Abbaspour-Fard, M. Kremmer, and A. O. Raji. 1999.
are followed by the rolling friction updated model.
Shape representation of axi-symmetrical, non-spherical particles
Ultimately, it can be concluded from Figures 13–14 that the in discrete element simulation using multi-element model
simulation of the hole-cleaning process enhanced by the roll- particles. Engineering Computations 16 (4):467–480. doi:10.1108/
ing friction model, and cannot exactly follow the true model 02644409910271894.
considering the general nonspherical particles where many Ferellec, J.-F., and G. R. McDowell. 2010. A method to model realistic
particle shape and inertia in DEM. Granular Matter 12 (1):459–467.
complex phenomena may be occurred due to the shape effects;
doi:10.1007/s10035-010-0205-8.
but the rolling resistance model can compensate the errors of Hogue, C. 1998. Shape representation and contact detection for discrete
the prediction of the particles hold-up, originated from the element simulations of arbitrary geometries. Engineering Computations
spherical shape idealization in traditional simulations, to some 15 (3):374–390. doi:10.1108/02644409810208525.
extent. Obviously, the introduced model must be updated Huilgol, R. R., and Z. You. 2005. Application of the augmented
Lagrangian method to steady pipe flows of Bingham, Casson and
further to minimize the existed errors, especially for the more
Herschel–Bulkley fluids. Journal of Non-Newtonian Fluid Mechanics
complex shapes of particles where the physics behind the 128 (1):126–143. doi:10.1016/j.jnnfm.2005.04.004.
shape effects is much more complicated than what is approxi- Kruggel-Emden, H., S. Rickelt, S. Wirtz, and V. Scherer. 2008. A study
mated by the rolling resistance method. on the validity of the multi-sphere discrete element method. Powder
Technology 188 (2):153–165. doi:10.1016/j.powtec.2008.04.037.
Kruggel-Emden, H., E. Simsek, S. Rickelt, S. Wirtz, and V. Scherer. 2007.
Conclusions Review and extension of normal force models for the discrete element
method. Powder Technology 171 (3):157–173. doi:10.1016/j.powtec.
When implementing more complex body shapes for particles 2006.10.004.
such as disk-shaped or cubic-shaped and MS cluster into a dis- Kuang, S. B., A. B. Yu, and Z. S. Zou. 2009. Computational study of flow
crete element code, multiple contact points may occur during regimes in vertical pneumatic conveying. Industrial & Engineering
Chemistry Research 48 (14): 6846–6858. doi:10.1021/ie900230s.
a collision. Considering the long times of computations due to
Laín, S., and M. Sommerfeld. 2008. Euler/Lagrange computations of
the complexities in contact predictions of nonspherical particle pneumatic conveying in a horizontal channel with different wall
in MS approach, introducing a model that takes into account roughness. Powder Technology 184 (1):76–88. doi:10.1016/j.
the particles shape effects, in the one hand, and the spherical powtec.2007.08.013.
PARTICULATE SCIENCE AND TECHNOLOGY 391
Latham, J. P., and A. Munjiza. 2004. The modelling of particle systems Saffman, P. G. 1965. The lift on a small sphere in a slow shear flow. Journal
with real shapes. Philosophical Transactions of the Royal Society of of Fluid Mechanics 22 (1):385–400. doi:10.1017/s0022112065000824.
London, Series A: Mathematical Physical and Engineering Sciences Saito, F., M. Kamiwano, and R. Aoki. 1984. Free settling of a non-
362 (1822):1953–1972. doi:10.1098/rsta.2004.1425. spherical particle in a stagnant liquid. Particulate Science and
Mei, R. 1992. An approximate expression for the shear lift force on Technology 2 (3):247–58. doi:10.1080/02726358408906409.
a spherical particle at finite Reynolds numbers. International Shook, C. A., and M. C. Roco. 1991. Slurry flow: Principles and
Journal of Multiphase Flow 18 (1):145–147. doi:10.1016/0301-9322 practice. 3rd ed. Boston, MA: Butterworth-Heinemann.
(92)90012-6. Sommerfeld, M. 2000. Theoretical and experimental modelling
Mezhericher, M., T. Brosh, and A. Levy. 2011. Modeling of particle of particulate flows. Technical Report Lecture Series, von Karman
pneumatic conveying using DEM and DPM methods. Particulate Institute for Fluid Dynamics, 20–23.
Science and Technology 29 (2):197–208. doi:10.1080/02726351003 Tomren, P. H., A. W. Iyoho, and J. J. Azar. 1986. Experimental study
792914. of cuttings transport in directional well. SPE Drilling Engineering
Oesterleé, B., and T. Bui Dinh. 1998. Experiments on the lift of a spinning 1 (1):43–56. doi:10.2118/12123-pa.
sphere in a range of intermediate Reynolds numbers. Experiments in Wadell, H. 1933. Sphericity and roundness of rock particles. Journal
Fluids 25 (1): 16–22. doi:10.1007/s003480050203. of Geology 41 (1):310–31. doi:10.1086/624040.
Copyright of Particulate Science & Technology is the property of Taylor & Francis Ltd and
its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.