Vous êtes sur la page 1sur 10

Powder Technology 260 (2014) 68–77

Contents lists available at ScienceDirect

Powder Technology
journal homepage: www.elsevier.com/locate/powtec

Integration of a new shape-dependent particle–fluid drag coefficient law


in the multiphase Eulerian–Lagrangian code MFIX-DEM
Fabio Dioguardi ⁎, Pierfrancesco Dellino, Daniela Mele
Dipartimento di Scienze della Terra e Geoambientali, via E. Orabona 4, 70125, Bari, Italy

a r t i c l e i n f o a b s t r a c t

Article history: A new shape-dependent fluid–particle drag law has been added into the open source fluid dynamic software
Received 31 January 2014 MFIX-DEM (MFIX-Discrete Element Method), which is the Eulerian–Lagrangian version of the classic MFIX
Received in revised form 25 March 2014 Eulerian–Eulerian multiphase flow model. The drag law had been obtained by previous settling experiments of
Accepted 30 March 2014
volcanic pumices in a motionless Newtonian liquid (water or alcohol). Pumices are characterized by a highly
Available online 4 April 2014
irregular shape, which is much different from a sphere and drastically influences fluid drag. The new drag law
Keywords:
defines the particle–fluid drag coefficient as a function of both the fluid regime and particle characteristics, of
Fluid–particle drag which the shape factor is a compact descriptor that quantifies how the particle shape differs from a simple
Shape factor sphere. As a validation of the integration of the new drag law in the simulation software MFIX-DEM, the code
MFIX-DEM has been used to replicate the experiment results. The comparison with simulations performed with other
Eulerian–Lagrangian models formulas demonstrates that, by means of the new drag law, a significant improvement in the capability of the
Terminal velocity MFIX-DEM code to predict the terminal velocities of irregularly shaped particles is obtained. Thanks to this
Interphase momentum exchange implementation, MFIX-DEM should be used, from now on, for simulating fluid–particle flows in which the
particles are significantly different from simple spheres, as is usually the case of environmental flows like
explosive eruptions or ash and pollutant dispersal. Based on the results of this research, in the future an improved
version of MFIX-DEM will be also presented, with a drag law useful also in the case of mixtures to be treated with
a Eulerian–Eulerian multiphase model.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction momentum exchange due to particle–fluid drag. Usually, in the numer-


ical models the aforementioned drag correlations are functions of
In a fluid–particle flow, the interphase momentum exchange plays a the flow Reynolds number, particle size and density, void fraction, etc.
fundamental role in the dynamics of particle transportation and These laws are designed for spherical particles and do not take into
sedimentation [1–3]. The way in which this term is computed depends account the effect that different particle shapes can have on the mo-
on the chosen multiphase representation, which can be either Eulerian– mentum exchange between phases. However, in the recent multiphase
Eulerian (hereafter EE) or Eulerian–Lagrangian (hereafter EL). In the fluid-dynamics literature, many examples of shape-dependent fluid–
former, the different phases are assumed to be interpenetrating contin- particle drag correlations have been presented (e.g. [8–12]). These
ua, and for each phase a continuum description, consisting of a set laws are generally functions of both the Reynolds number and the
of conservation (mass, momentum and energy) and constitutive shape factor. For example, particle's sphericity, which is defined as the
equations (e.g. interphase momentum exchange), is solved. In the EE ratio of the surface area of the equivalent-volume-sphere to the actual
approach, drag correlations for a single particle, which are included in surface of the particle, is one of the most widely used parameters for
the interphase momentum exchange term of the equations, are gener- quantifying the particle shape [8,11]. Tran-Cong et al. [10] evaluated
alized to particle mixtures [4–6]. In the EL representation, a continuum the particle's shape as the ratio of the nominal diameter (the diameter
description is employed only for the carrier phase (fluid), while of the equivalent-volume-sphere) to the equivalent-sphere diameter.
the dispersed phase (e.g. solid particles) is treated as being composed There are many other shape descriptors, the most used are listed in
by discrete elements (single particles or clusters of particles) [7]; the comprehensive work of Nikku et al. [13].
the particle–fluid drag force is then computed directly from drag- The implementation of a shape-dependent fluid–particle drag law
coefficient relationships in order to couple the particle to the flow in a numerical code is a fundamental step towards more realistic simu-
field solution. The research here presented focuses on the interphase lations of multiphase flows in which the solid phase is represented by
irregularly-shaped particles. This is generally the case of environmental
⁎ Corresponding author. Tel.: +39 3403733928; fax: +39 0805442625. fluid–particle flows, e.g. pyroclastic density currents or eruptive
E-mail address: fabio.dioguardi@uniba.it (F. Dioguardi). columns in explosive eruptions, sand transportation in water and air,

http://dx.doi.org/10.1016/j.powtec.2014.03.071
0032-5910/© 2014 Elsevier B.V. All rights reserved.
F. Dioguardi et al. / Powder Technology 260 (2014) 68–77 69

and pollutants dispersal. In particular, explosive volcanic eruptions are which states that the volume fraction ε of all the phases (the subscripts f
characterized by the issue from the volcanic vent of a multiphase mix- and s refer to the fluid and the solid phase, respectively, m is the number
ture made of volcanic gases and solid particles [14–16]. This multiphase of the mth solid phase, when more than 1 solid phase is present) sums
flow forms in the volcanic conduit as the result of magma fragmentation to one.
[17–20], a process in which the magmatic melt is transformed into a The mass and momentum conservation equations for the fluid
mixture of volcanic gases, solidified magmatic particles, crystals (al- phase, in the absence of a phase change and chemical reactions, are
ready present in the magmatic melt) and country rock fragments. The
mechanical energy release of the eruption [20] leads to the acceleration ∂   
ε f ρf þ ∇  εf ρf vf ¼ 0 ð2Þ
of the gas–particle mixture in the conduit and the development of a ∂t
multiphase flow which, on issuing from the conduit vent, generates an
eruptive column [16,21]. Depending on vent conditions (mass flow ∂    X
M

rate, static and dynamic pressure, etc.), the column can evolve in ε f ρ f v f þ ∇  ε f ρ f v f v f ¼ ∇  S f þ ε f ρ f g− Ifm ð3Þ
∂t m¼1
many different scenarios (column collapse, convective plumes,
overpressured jets and all the transitional possibilities) [16], each one
responsible for a different type of impact over the territory. Understand- where ρ is the density, v is the velocity, S f is the fluid-phase stress
ing the flow dynamics of multiphase flows during explosive eruptions is tensor, g is the gravitational acceleration and Ifm is the interphase
a fundamental step towards hazard assessment. Numerical simulations momentum transfer between the fluid phase and the mth solid phase.
represent a widely used approach in this research field. Nowadays a The fluid phase can be modeled as a compressible fluid obeying to the
number of models of different complexity is available, from the simplest ideal fluid law:
pseudo-gas models, in which the gas–particle mixture is approximated
P f Mw
to a homogeneous fluid (e.g. [15]) to the more complex EE models ρf ¼ ð4Þ
RT f
(e.g. [22,23]) or EL models (e.g. [24]). For better capturing the dynamics
of explosive eruptions, it is fundamental to include realistic parameters
where Pf is the fluid pressure, Mw is the fluid molecular weight, R is the
into the numerical model, as for example particle-drag coefficient
universal gas constant and Tf is the fluid temperature. However MFIX al-
relationships that take into account the irregular shape of volcanic
lows the user to model the fluid phase as incompressible, by specifying a
particles.
constant value for the fluid density. In this work we model the free-fall
In this paper a relationship linking the drag coefficient directly to a
of a particle in motionless water or alcohol (the same liquids used in the
shape factor [25] has been included in the open-source fluid dynamic
experiments carried on for defining the drag law), thus the liquid can be
software MFIX (Multiphase Flow with Interphase eXchange [26]). The
considered as incompressible.
implemented fluid–particle drag relationship has been obtained by
The mth solid phase is represented by Npm spherical particles
settling experiments of volcanic particles in a cylinder filled of water
with each particle having diameter Dpm and density ρsm [7]. The solid
or alcohol [25], which were performed in a previous study. A wide
phases differ according to their radii and densities. For a total of M
range of particles' shapes and Reynolds number were investigated.
M
The implementation works with the EL version of the code, MFIX- solid phases, the total number of particles is equal to Np ¼ ∑ Npm .
DEM (MFIX-Discrete Element Method [7,27,28]), in which the solid m¼1

phase is represented by individual discrete particles, whose velocity Particles are represented in a Lagrangian frame of reference at a time t
n ðiÞ o
and trajectory are computed as a function of the solution of the by X ðt Þ; VðiÞ ðt Þ; ωðiÞ ðt Þ; DðiÞ ; ρðiÞ ; i ¼ 1; …; N , where X(i) is the ith
fluid-phase flow field. The effectiveness of the newly implemented
particle's position, V(i) and ω(i) are its linear and angular velocities,
code in performing simulations of multiphase flows has been tested
respectively, D(i) denotes its diameter and ρ(i) its density. If a particle
by comparing computational results with the terminal velocity
belongs to the mth solid phase, then its diameter and density are Dpm
measured directly by experiments. Results show that with this new im-
and ρsm, respectively. The particle's mass is the product of its density
plementation of the code it is possible to predict the behavior of multi-
and its volume (of the equivalent sphere):
phase flows with a better precision compared to other drag laws. It is
also possible to use the new drag law in the classic EE MFIX version, 3
ðiÞ πDðiÞ
ðiÞ
but the application of the law to mixture of particles is currently not m ¼ρ : ð5Þ
6
experimentally verified and is the object of current investigations.
In the following sections, the theory behind the EL model is succinct-
Its moment of inertia is:
ly presented, together with some of the constitutive laws that are used
in the new implementation in the fluid–particle drag law that we show 2
ði Þ mðiÞ DðiÞ
here. Afterward, results of the simulations are presented and a discus- I ¼ : ð6Þ
10
sion of the results and future possible improvements is carried out.
The position, linear and angular velocities of the ith particle change
2. Code description and implementation of the new drag law according to the Newton's laws as:

2.1. MFIX-DEM (EL) governing equations dXðiÞ ðt Þ ðiÞ


¼ V ðt Þ ð7Þ
dt
In MFIX-DEM, the fluid-phase equations are the same as those of the
EE version of MFIX, while the solid phase is modeled using discrete ðiÞ
ðiÞ dV ðt Þ ðiÞ ðiÞ ði∈k;mÞ ði Þ
particles, whose position, linear and angular velocity are obtained by m ¼ FT ðt Þ ¼ m g þ Fd ðt Þ þ Fc ðt Þ ð8Þ
dt
solving the Newton's laws ([7,27,28]).
The fundamental equation of a multiphase flow is:
ði Þ dωðiÞ ðt Þ ðiÞ
I ¼ T ðt Þ ð9Þ
dt
X
M
(i ∈ k,m)
εf þ ε sm ¼ 1 ð1Þ where F d is the total drag force (pressure and viscous) on the
m¼1 i th particle residing in the k th cell and belonging to the mth solid
70 F. Dioguardi et al. / Powder Technology 260 (2014) 68–77

Table 1
Experimental data and results.

Run Liquid ρf μf ρsm Dpm ψ Cd WtNewton Wtmeas Wtsim Wtsim_no_Dellino


(g cm−3) (poise) (g cm−3) (cm) (cm s−1) (cm s−1) (cm s−1) (cm s−1)

1 Water 1.00 1.02E−02 2.011 2.297 0.353 2.42 35.40 39.45 34.40 76.03
2 Water 1.00 1.02E−02 1.967 2.183 0.483 1.45 43.64 40.83 41.58 73.52
3 Water 1.00 1.02E−02 2.167 1.958 0.348 2.52 34.46 42.94 34.12 78.10
4 Water 1.00 1.02E−02 1.921 1.875 0.748 0.71 56.25 38.96 53.98 68.36
5 Water 1.00 1.02E−02 2.178 2.240 0.457 1.57 46.85 44.73 44.63 81.61
6 Water 1.00 1.02E−02 2.181 2.015 0.377 2.20 37.65 44.45 36.94 79.34
7 Water 1.00 1.02E−02 1.959 1.981 0.435 1.74 37.76 37.44 36.70 71.08
8 Water 1.00 1.02E−02 2.030 1.612 0.735 0.74 54.00 41.44 52.65 68.23
9 Water 1.00 1.02E−02 2.014 1.487 0.552 1.21 40.30 32.71 39.76 65.46
10 Water 1.00 1.02E−02 2.011 1.496 0.374 2.32 29.19 35.54 29.16 65.54
11 Water 1.00 1.02E−02 1.823 1.821 0.639 0.94 45.78 36.60 44.23 63.96
12 Water 1.00 1.02E−02 1.614 1.122 0.358 2.64 18.48 20.82 18.56 44.83
13 Water 1.00 1.02E−02 1.897 1.301 0.496 1.48 32.10 30.87 31.91 58.03
14 Water 1.00 1.02E−02 2.164 0.725 0.450 1.85 24.40 27.62 24.44 49.96
15 Water 1.00 1.02E−02 2.071 0.756 0.548 1.33 28.22 30.00 28.19 48.92
16 Water 1.00 1.02E−02 2.310 1.036 0.603 1.08 40.48 43.57 40.32 63.06
17 Water 1.00 1.02E−02 1.247 0.924 0.415 2.19 11.67 13.93 11.69 25.91
18 Water 1.00 1.02E−02 2.033 0.971 0.674 0.92 37.85 35.17 37.70 54.28
19 Water 1.00 1.02E−02 1.500 1.018 0.362 2.64 15.87 20.00 15.94 38.63
20 Water 1.00 1.02E−02 2.022 0.818 0.348 2.82 19.69 23.71 19.79 49.67
21 Water 1.00 1.02E−02 1.815 1.036 0.531 1.37 28.43 28.90 28.37 49.73
22 Water 1.00 1.02E−02 2.037 0.956 0.514 1.44 30.01 28.56 29.96 53.98
23 Water 1.00 1.02E−02 2.098 0.823 0.527 1.40 29.02 38.50 28.99 51.64
24 Water 1.00 1.02E−02 2.201 0.808 0.557 1.28 31.52 30.17 31.48 53.52
25 Water 1.00 1.02E−02 1.931 0.664 0.545 1.37 24.27 22.42 24.26 42.78
26 Water 1.00 1.02E−02 1.937 0.822 0.580 1.21 28.89 28.00 28.87 47.68
27 Water 1.00 1.02E−02 2.026 0.764 0.559 1.29 28.22 26.13 28.19 48.13
28 Water 1.00 1.02E−02 1.954 0.701 0.652 1.01 29.41 28.00 29.39 44.49
29 Water 1.00 1.02E−02 2.185 0.694 0.608 1.13 30.92 26.86 30.87 49.33
30 Water 1.00 1.02E−02 2.139 0.502 0.530 1.48 22.50 19.36 22.51 41.19
31 Water 1.00 1.02E−02 2.295 0.657 0.710 0.87 35.72 34.00 35.66 50.20
32 Water 1.00 1.02E−02 2.049 0.557 0.473 1.77 20.79 24.69 20.80 41.62
33 Water 1.00 1.02E−02 1.870 0.581 0.635 1.08 24.69 27.00 24.67 38.71
34 Water 1.00 1.02E−02 2.073 0.561 0.480 1.72 21.40 21.20 21.39 42.25
35 Water 1.00 1.02E−02 2.000 0.528 0.469 1.81 19.53 26.30 19.55 39.57
36 Water 1.00 1.02E−02 1.399 0.563 0.539 1.48 14.11 16.00 14.10 25.81
37 Water 1.00 1.02E−02 1.624 0.516 0.639 1.10 19.53 16.36 19.52 30.91
38 Water 1.00 1.02E−02 1.945 0.546 0.575 1.29 22.92 24.12 22.91 39.12
39 Water 1.00 1.02E−02 1.729 0.489 0.656 1.06 21.00 23.29 20.99 32.52
40 Water 1.00 1.02E−02 2.123 0.434 0.360 2.86 14.93 17.72 14.94 38.03
41 Water 1.00 1.02E−02 1.473 0.438 0.655 1.09 15.74 19.32 15.73 24.80
42 Water 1.00 1.02E−02 1.668 0.510 0.506 1.63 16.55 22.30 16.55 31.79
43 Water 1.00 1.02E−02 1.701 0.527 0.548 1.42 18.47 20.14 18.47 33.10
44 Water 1.00 1.02E−02 1.804 0.525 0.516 1.56 18.83 23.71 18.83 35.38
45 Water 1.00 1.02E−02 1.465 0.502 0.564 1.38 14.86 14.58 14.87 26.32
46 Water 1.00 1.02E−02 1.564 0.491 0.638 1.12 17.99 22.58 17.99 28.66
47 Water 1.00 1.02E−02 1.749 0.498 0.600 1.22 19.96 15.97 19.97 33.27
48 Water 1.00 1.02E−02 2.626 0.475 0.645 1.06 30.90 27.25 30.91 47.87
49 Water 1.00 1.02E−02 1.672 0.487 0.753 0.84 22.51 21.46 22.51 31.16
50 Water 1.00 1.02E−02 2.694 0.473 0.717 0.88 34.43 29.88 34.40 48.76
51 Water 1.00 1.02E−02 1.676 0.435 0.600 1.25 17.55 16.47 17.55 29.54
52 Water 1.00 1.02E−02 1.552 0.426 0.570 1.38 14.96 15.96 14.95 26.42
53 Water 1.00 1.02E−02 1.903 0.374 0.495 1.73 15.96 19.25 15.97 31.67
54 Water 1.00 1.02E−02 2.078 0.426 0.711 0.93 25.47 24.58 25.46 36.92
55 Water 1.00 1.02E−02 2.172 0.338 0.442 2.10 15.73 13.96 15.73 34.30
56 Water 1.00 1.02E−02 1.563 0.377 0.653 1.11 15.81 16.67 15.79 24.76
57 Water 1.00 1.02E−02 1.521 0.378 0.606 1.26 14.27 16.00 14.28 23.67
58 Water 1.00 1.02E−02 1.277 0.391 0.734 0.94 12.28 11.97 12.28 16.57
59 Water 1.00 1.02E−02 2.151 0.409 0.570 1.34 21.43 28.80 21.43 37.38
60 Water 1.00 1.02E−02 1.515 0.411 0.590 1.31 14.54 15.07 14.55 25.07
61 Water 1.00 1.02E−02 2.114 0.383 0.670 1.04 23.21 25.31 23.22 35.59
62 Water 1.00 1.02E−02 1.913 0.371 0.667 1.06 20.47 20.70 20.49 31.71
63 Water 1.00 1.02E−02 2.095 0.342 0.427 2.23 14.84 18.57 14.85 33.35
64 Water 1.00 1.02E−02 1.478 0.353 0.669 1.09 14.25 14.00 14.26 21.23
65 Water 1.00 1.02E−02 2.086 0.335 0.713 0.95 22.37 21.35 22.38 32.87
66 Water 1.00 1.02E−02 1.889 0.218 0.753 0.92 16.57 15.00 16.58 20.74
67 Water 1.00 1.02E−02 1.467 0.288 0.773 0.87 14.18 15.50 14.17 17.66
68 Water 1.00 1.02E−02 1.488 0.267 0.661 1.14 12.22 14.08 12.20 17.02
69 Water 1.00 1.02E−02 1.763 0.264 0.638 1.19 14.84 13.68 14.85 2.19
70 Water 1.00 1.02E−02 1.414 0.280 0.575 1.45 10.24 12.19 10.24 16.02
71 Water 1.00 1.02E−02 1.245 0.224 0.620 1.34 7.31 4.44 7.32 9.52
72 Water 1.00 1.02E−02 1.667 0.227 0.542 1.60 11.11 9.89 11.11 17.98
73 Water 1.00 1.02E−02 1.895 0.246 0.537 1.59 13.44 15.38 13.44 23.06
74 Water 1.00 1.02E−02 1.304 0.236 0.667 1.17 8.97 8.70 8.96 11.42
F. Dioguardi et al. / Powder Technology 260 (2014) 68–77 71

Table 1 (continued)
Run Liquid ρf μf ρsm Dpm ψ Cd WtNewton Wtmeas Wtsim Wtsim_no_Dellino
(g cm−3) (poise) (g cm−3) (cm) (cm s−1) (cm s−1) (cm s−1) (cm s−1)

75 Water 1.00 1.02E−02 1.658 0.233 0.685 1.08 13.62 13.07 13.60 18.23
76 Water 1.00 1.02E−02 1.600 0.218 0.724 1.00 13.09 13.18 13.08 16.27
77 Water 1.00 1.02E−02 1.477 0.244 0.674 1.12 11.66 13.12 11.65 15.55
78 Water 1.00 1.02E−02 2.409 1.708 0.335 2.70 34.12 51.33 34.12 81.97
79 Water 1.00 1.02E−02 2.683 1.801 0.554 1.15 58.65 43.50 56.93 91.15
80 Water 1.00 1.02E−02 2.567 0.705 0.713 0.85 41.14 39.83 41.11 57.17
81 Water 1.00 1.02E−02 2.367 0.624 0.581 1.22 30.18 34.12 30.18 50.27
82 Water 1.00 1.02E−02 2.467 0.669 0.791 0.72 42.09 35.00 42.05 53.91
83 Water 1.00 1.02E−02 2.367 0.634 0.481 1.68 26.01 35.31 26.04 50.67
84 Water 1.00 1.02E−02 2.200 0.607 0.406 2.24 20.63 21.60 20.64 46.46
85 Water 1.00 1.02E−02 3.033 0.553 0.654 1.00 38.25 40.28 38.26 57.74
86 Water 1.00 1.02E−02 2.600 0.659 0.595 1.16 34.45 36.25 34.45 55.88
87 Water 1.00 1.02E−02 2.400 0.757 0.703 0.87 39.92 35.69 39.86 55.97
88 Water 1.00 1.02E−02 2.567 0.773 0.580 1.19 36.46 38.25 36.47 59.83
89 Water 1.00 1.02E−02 2.400 0.488 0.783 0.77 34.12 30.30 34.11 45.02
90 Water 1.00 1.02E−02 2.233 0.445 0.674 1.00 26.79 26.35 26.77 40.35
91 Water 1.00 1.02E−02 2.367 0.475 0.705 0.92 30.42 32.25 30.41 43.89
92 Water 1.00 1.02E−02 2.167 0.451 0.778 0.79 29.52 27.50 29.53 39.52
93 Water 1.00 1.02E−02 2.100 0.412 0.845 0.70 29.18 26.75 29.15 36.68
94 Water 1.00 1.02E−02 2.300 0.466 0.837 0.69 33.83 35.37 33.81 42.40
95 Water 1.00 1.02E−02 2.167 0.458 0.639 1.09 25.28 26.30 25.29 39.83
96 Water 1.00 1.02E−02 2.167 0.497 0.593 1.23 24.86 23.90 24.88 41.48
97 Water 1.00 1.02E−02 1.933 0.526 0.669 1.01 25.26 22.67 25.27 38.15
98 Water 1.00 1.02E−02 2.167 0.556 0.605 1.17 26.96 29.60 26.94 43.86
99 Water 1.00 1.02E−02 2.400 0.526 0.660 1.01 30.87 33.06 30.87 46.74
100 Water 1.00 1.02E−02 2.270 0.315 0.716 0.94 23.52 21.83 23.54 34.47
101 Water 1.00 1.02E−02 2.260 0.378 0.935 0.59 32.45 29.17 32.43 37.60
102 Water 1.00 1.02E−02 2.305 0.380 0.809 0.75 29.37 31.06 29.37 38.37
103 Water 1.00 1.02E−02 2.448 0.388 0.647 1.08 26.07 25.39 26.05 40.84
104 Water 1.00 1.02E−02 2.356 0.382 0.716 0.92 27.16 25.78 27.15 39.22
105 Water 1.00 1.02E−02 2.818 0.374 0.862 0.67 36.46 32.75 36.46 44.93
106 Water 1.00 1.02E−02 2.478 0.353 0.752 0.85 28.30 24.52 28.29 39.36
107 Water 1.00 1.02E−02 1.975 0.344 0.848 0.71 24.83 22.58 24.81 31.56
108 Water 1.00 1.02E−02 2.119 0.341 0.828 0.74 26.02 21.39 26.01 33.66
109 Water 1.00 1.02E−02 2.073 0.395 0.535 1.50 19.23 21.58 19.22 35.47
110 Water 1.00 1.02E−02 2.512 0.315 0.563 1.40 21.10 26.44 21.11 37.61
111 Water 1.00 1.02E−02 2.446 0.337 0.845 0.71 30.04 26.00 30.03 38.04
112 Water 1.00 1.02E−02 2.082 0.272 0.887 0.68 23.85 22.60 23.84 28.15
113 Water 1.00 1.02E−02 2.010 0.302 0.860 0.71 23.77 24.58 23.77 29.43
114 Water 1.00 1.02E−02 2.137 0.306 0.758 0.86 22.93 18.36 22.93 31.97
115 Water 1.00 1.02E−02 1.957 0.245 0.801 0.82 19.35 19.82 19.37 37.41
116 Water 1.00 1.02E−02 1.897 0.289 0.942 0.61 23.54 21.54 23.52 26.41
117 Water 1.00 1.02E−02 2.039 0.314 0.851 0.71 24.45 21.27 24.44 30.91
118 Water 1.00 1.02E−02 2.008 0.276 0.742 0.91 19.95 18.64 19.96 27.29
119 Water 1.00 1.02E−02 2.005 0.306 0.847 0.72 23.57 19.42 23.58 29.66
120 Water 1.00 1.02E−02 1.889 0.258 0.604 1.30 15.18 16.33 15.17 23.90
121 Water 1.00 1.02E−02 2.114 0.271 0.712 0.98 20.11 17.86 20.11 28.57
122 Water 1.00 1.02E−02 2.150 0.299 0.943 0.60 27.31 22.25 27.30 31.58
123 Water 1.00 1.02E−02 2.188 0.268 0.570 1.41 17.17 20.73 17.17 29.44
124 Alcohol 0.81 1.72E−02 1.697 0.452 0.692 1.02 25.17 24.15 25.18 36.78
125 Alcohol 0.81 1.72E−02 1.140 0.396 0.736 0.98 14.71 12.50 14.71 17.94
126 Alcohol 0.81 1.72E−02 1.210 0.487 0.686 1.06 17.21 17.62 17.22 24.06
127 Alcohol 0.81 1.72E−02 1.292 0.509 0.616 1.25 17.79 19.17 17.79 27.98
128 Alcohol 0.81 1.72E−02 0.866 0.404 0.736 1.05 5.92 4.14 5.91 5.95
129 Alcohol 0.81 1.72E−02 1.209 0.453 0.695 1.05 16.71 17.71 16.69 22.61
130 Alcohol 0.81 1.72E−02 0.981 0.481 0.587 1.43 9.64 8.57 9.65 14.10
131 Alcohol 0.81 1.72E−02 0.865 0.391 0.769 0.98 5.96 6.05 5.96 5.70
132 Alcohol 0.81 1.72E−02 0.843 0.389 0.543 1.78 3.41 4.28 3.41 4.06
133 Alcohol 0.81 1.72E−02 0.986 0.488 0.711 1.03 11.59 13.22 11.58 14.53
134 Alcohol 0.81 1.72E−02 1.227 0.415 0.610 1.31 14.60 12.78 14.59 21.58
135 Alcohol 0.81 1.72E−02 1.189 0.468 0.779 0.86 18.22 18.25 18.20 22.52
136 Alcohol 0.81 1.72E−02 1.260 0.429 0.569 1.46 14.59 15.53 14.60 23.26
137 Alcohol 0.81 1.72E−02 0.937 0.306 0.634 1.34 6.83 7.75 6.84 7.79
138 Alcohol 0.81 1.72E−02 0.951 0.342 0.767 0.96 9.02 9.38 9.03 9.23
139 Alcohol 0.81 1.72E−02 0.927 0.321 0.690 1.16 7.22 4.10 7.22 7.72
140 Alcohol 0.81 1.72E−02 1.020 0.399 0.592 1.43 9.74 8.81 9.74 13.63
141 Alcohol 0.81 1.72E−02 1.379 0.346 0.726 0.99 17.92 16.89 17.90 22.39
142 Alcohol 0.81 1.72E−02 0.892 0.325 0.638 1.34 5.66 6.52 5.66 6.20

ðiÞ
phase, F(i)
c is the net contact force acting as a result of contact with equations (S f ; Fc and T(i)). In MFIX-DEM the total drag force is rep-
other particles, T(i) is the sum of all the torques acting on the ith par- resented as:
ticle and F(i)
T is the net sum of all forces acting on the i
th
particle. In
  β
ði∈kÞ
ν m   ði Þ  
this study we focus on the drag force between the solid particles ði∈k;mÞ
Fd
ði Þ
¼ −∇P f X ν m þ m v f X −V
ði Þ
ð10Þ
and the fluid, thus we omit the discussion on the other constitutive εsm
72 F. Dioguardi et al. / Powder Technology 260 (2014) 68–77

where Pf (X(i)) and vf (X(i)) are the fluid-phase mean pressure and Newtonian fluids of known density ρf and viscosity μf (distilled water
3
velocity at the particle location, ν m ¼ π D6 is the particle volume, and ethylic alcohol at 20 °C), which are reported in Table 1, together
(i ∈ k)
and βm is the local fluid-solid momentum transfer coefficient with particles data and simulation results (discussed in Section 3) for
∈ k)
for the particle i residing in the kth cell. β(i
m is a function of the each experimental run. During each experiment the particle motion
fluid and particle velocities and of the particle drag coefficient C d. has been recorded with a high-definition video-camera for measuring
The interphase momentum transfer between the fluid phase and the terminal velocity. For each particle, the density was measured by
the mth solid phase Ifm at the position xk is: standard Gay–Lussac pycnometers, which allow also the determination
of the diameter of the equivalent sphere (Dpm).
k 1 X
N m   Particle shape was evaluated through the shape factor Ψ, calculated
ði∈k;mÞ ðiÞ
Ifm ¼ F K Xm ; x k ð11Þ as:
ν k i¼1 d

Φ
where K(X(i)m , xk) is a generic kernel determining the influence of the Ψ¼ ð13Þ
X
particle force at X(i)
m on a grid node located at xk, vk is the volume of
the kth grid cell. where Φ is the sphericity and X the circularity of the particle. Sphericity
Here we introduce in the code a new law for Cd, which, as being a is defined by the ratio Asph/Ap, in which Asph is the surface area of the
function of the particle shape factor, is suitable for particles with diverse equivalent sphere, calculated by Dpm, and Ap is the surface area of the ac-
shapes, and performs well even with highly irregular shapes [25] as tual particle, calculated by approximating particles to scalene ellipsoids:
those of rock particles or natural ash fragments (volcanic powder).
 1
ðdl =2Þz ðdm =2Þz þ ðdl =2Þz ðds =2Þz þ ðdm =2Þz ðds =2Þz z
2.2. The new drag law implemented in the code Ap ¼ 4π ð14Þ
3

The particle drag law proposed by Dellino et al. [25] was obtained via where dl, dm and ds are the long, medium and small axes (measured by
an experimental study consisting of measuring the terminal velocity of photo analysis on particles mounted under a stereomicroscope) and z is
volcanic pumice particles in a vertical cylinder (height of 1.5 m and equal to 1.6075. Fig. 1 shows an example of the determination of the
inner radius of 0.5 cm) filled with water or alcohol. three axes. The circularity X is defined by the ratio Pmp/Pp, in which
In multiphase-flow physics the terminal velocity, which is the Pmp is the measured particle perimeter and Pp is the perimeter of the
equilibrium velocity at which a particle settles through a static, viscous circle equivalent to the maximum projection area Amp. This shape factor
fluid, is the fundamental quantity for constraining the particle hydraulic is a compact descriptor of the irregular shape of particles [25]. The closer
behavior. For a Newtonian fluid, the terminal velocity of a particle is Ψ is to 1, the closer the particle is to a spherical shape and vice-versa. In
defined by the so-called Newton's impact law: Fig. 2 some volcanic particles and their shape factors are displayed.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi Typically pyroclastic particles cover a wide range of shape factor values,
u
u4gDpm ρsm −ρ f which is always way lower than 1. This is due to the smaller scale irreg-
t
Wt Newton ¼ ð12Þ ularities distributed over the pumice outline (e.g. bubbles), which are
3C d ρ f
typical of volcanic particles.
Upon combining Eq. (12) with the definition of the particle's
Cd is a function of the particle Reynolds number and there is a well-
Reynolds number:
defined correlation between Cd and Re for spherical particles. In partic-
ular, at 103 b Re b 105 Cd is quite constant and, at Re = 104, it is about ρ f WtDpm
0.44. However, for irregularly shaped particles, Cd can be significantly Re ¼ ð15Þ
μf
different from that of a sphere, thus experiments are needed for
defining specific Cd–Re relationships. the quantity CdRe2 was obtained:
Dellino et al. carried out an experimental study on the size, shape
 
and density dependence of Cd and Re of natural pumices [25]. A set of
2
4gDpm ρ f ρsm −ρ f
142 pumices of different size, shape and density were chosen for mea- C d Re ¼ ð16Þ
suring the terminal velocity of individual particles settling through 3μ 2

0.3 cm
0.3 cm dm ds
dl

dl

Fig. 1. Example of the determination of the three axes dl, dm and ds with the photo analysis. The selected pumice particles are mounted under a stereomicroscope on a goniometric universal
stage and oriented in order to get the three major axes.
F. Dioguardi et al. / Powder Technology 260 (2014) 68–77 73

which, being independent of terminal velocity, is particularly By rearranging Eqs. (12), (16) and (17), the following equation for
convenient. This quantity was multiplied by the shape factor Ψ and the drag coefficient was obtained:
data points were plotted in a diagram Re vs. CdRe2 Ψ; the least square  
function was constrained to give a value of 0.44 for the drag coefficient 0:69gD3pm ρ f 1:33ρsm −1:33ρ f
Cd at Re = 104, a procedure that allowed to obtain an exponent of 1.6 Cd ¼  1:0412 : ð18Þ
gψ1:6 D3pm ρ f ðρsm −ρ f Þ
for the shape factor. The function is a power law with a correlation μ 2f μ2
f
coefficient of 0.99:
This law takes into account both fluid regime characteristics and
  particle properties, including particle shape. Although in the formula
2 1:6 0:5206
Re ¼ 1:0387 C d Re Ψ ð17Þ there is not an explicit reference to Re, its dependency on the Reynolds
number can be easily inferred by looking Eqs. (15)–(17), from which
Eq. (18) was derived. The drag law replicates quite well the terminal
from which the dependency of Cd from Re and Ψ is evident. velocity in a wide range of particle Reynolds number, namely between

Fig. 2. Display mount of different pyroclastic particles coming from the eruption deposits of Somma Vesuvius, Campi Flegrei and Vulcano Island (Italy). The shape factor Ψ of each particle is
reported, together with the scale.
74 F. Dioguardi et al. / Powder Technology 260 (2014) 68–77

a b c

Fig. 3. Display mount of three different frames of a video footage of a free-fall experiments. The dashed line marks the theoretical vertical trajectory for a spherical particle. The discrepancy
between the pumice particle trajectory and the theoretical one is evident and is responsible for the difference between Wtmeas and WtNewt.

10 and more than 105, which accounts for most of the regimes of natural the default drag law already present in the MFIX-DEM code. As for de-
and industrial multiphase gas–particle flows [25,29]. fault drag law already present in MFIX, there are different possible
In this paper, the drag law of Eq. (18) has been implemented in the choices, which are quite similar in their algebraic form. We chose, as a
MFIX source code. In particular, a new external function C_DS_DEL has reference for the comparison with our new drag law, the well-known
been added to the subroutine DRAG_GS (in the file drag_gs.f), which formula of Wen and Yu [4], which is suitable for diluted multiphase
calculates the interphase drag force of both the EE and the EL versions flow [31]:
of MFIX. The external function contains Eq. (18). In MFIX-DEM, the
fluid–particle drag force (second term of Eq. (10)) is calculated by the 3εsm ε f ρ f  
 −2:65
β¼ C d vsm −v f ε f ð19Þ
subroutine DES_DRAG_GP, which is stored in the file drag_fgs.f (in the 4Dpm
model/des folder). This subroutine calls the same external function
C_DS_DEL. For this implementation, it has been necessary to define 3 where the solid-phase drag coefficient is:
new variables: 1) a logical variable CD_DELLINO, which by default is 
24  0:687
set to FALSE. If the user sets it to. TRUE., Eq. (18) is employed in place Cd ¼ 1 þ 0:15 ε f Res ð20Þ
of the default particle drag relationships. 2) PSI_s, which stores the ε f Res
particle shape factor value Ψ. 3) DES_PSI_s, which is the MFIX-DEM
counterpart of PSI_s. Details on the implementation of the new drag in which Res is the relative particle Reynolds number:
law are listed in the file Cd_Dellino_implementation, which is available
as Supplementary material. ρ f Dpm jvsm −v f j
Res ¼ : ð21Þ
μf

3. Test on the code ability to predict terminal velocity by means of In this formula, the drag coefficient is only a function of the Reynolds
the new drag law number and fluid-phase volume fraction. Particle shape is not consid-
ered, so we expect that the Wen and Yu [4] formula should lead to a sig-
In order to evaluate the reliability of the new drag law (Eq. (18)) as nificant underestimation of the drag force in Eq. (10). As a consequence,
implemented in the simulation code, the multiphase flow code MFIX- settling velocity calculation should give quite different results when
DEM was used to calculate the terminal velocity of the 142 particles compared to those obtained by the application of our new drag law.
previously used in the experiments. Numerical simulations were To judge the precision of calculation based on the two different drag
performed both with the newly implemented drag law and also with laws, numerical results were compared with experiment measure-
60
ments. In addition, calculations have been performed also by directly
using the Newton impact law formula of Eq. (12), by using as drag coef-
50 ficient that resulting from Eq. (18). This is useful to judge results against
y = 0.86x + 3.26 an analytical reference.
Wt meas (cm s -1 )

r = 0.92
40 As the Cartesian grid tool has not been tested with MFIX-DEM [30],
and as MFIX-DEM does not work with cylindrical coordinates [7], the
30
domain has been approximated by a vertical 1.5 m high parallelepiped
20
with a squared base of 10 cm side length. The domain has been
discretized by a Cartesian uniform mesh consisting of 2 cm cubic cells.
10 Particle data used for the simulation are reported in Table 1, namely
they are: density ρsm, diameter Dpm and shape factor Ψ. The terminal ve-
0 locity measurements and simulation results are also listed in Table 1,
0 10 20 30 40 50 60
they include: 1) the terminal velocity, WtNewton, as calculated with the
Wt Newton (cm s -1 ) Newton's impact law (Eq. (12)) and obtained by considering as Cd the
one calculated by Eq. (18); 2) the experimentally measured terminal
Fig. 4. Wtmeas vs. WtNewt plot. Each point represents an experimental run. The thick line is velocity Wtmeas; 3) the terminal velocity Wtsim, calculated with the
the linear regression line; the equation and the correlation coefficient r are also reported.
The thick dashed line is the equality line. Although the points are dispersed along the
MFIX-DEM numerical simulation code by means of the implementation
equality line, a discrepancy is visible especially for the higher terminal velocities, which of our new drag law (Eq. (18)); 4) the terminal velocity Wtsim_no_Dellino,
is due to the irregularities of the fall trajectory (Fig. 3). calculated with the MFIX-DEM numerical simulation code by means of
F. Dioguardi et al. / Powder Technology 260 (2014) 68–77 75

a With Cd Dellino et al. (2005) b With Cd Dellino et al. (2005)


150 50 150 50
45 45
145 145
40

v (cm s-1)
40
35 WtNewton 35
140 140
30 30

y (cm)
y (cm)

Wtmeas

v (cm s-1)
135 25 135 25 WtNewton
20 20 Wtmeas
130 130
15 15
10 10
125 125
5 5
120 0 120 0
0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
t (s) t (s)

Without Cd Dellino et al. (2005) Without Cd Dellino et al. (2005)


150 50 150 50
45 45
145 145
40 40

v (cm s-1)
35 WtNewton 35
140 140
30 30
y (cm)

y (cm)
Wtmeas
135 25 v (cm s-1) 135 25 WtNewton
20 20 Wtmeas
130 130
15 15
10 10
125 125
5 5
120 0 120 0
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
t (s) t (s)

Fig. 5. Plot showing the time evolution of the particle vertical position (dashed line) and vertical velocity (thick line) for two different experiments (a: experiment no. 50; b: experiment no.
100, see Table 1) each one simulated with the new law of Dellino et al. [25] (Eq. (18), upper plot) and the default law (Eq. (20), lower plot). Wtmeas and WtNewt are also reported on the right
y axis. The terminal velocity is reached when the velocity curve becomes a line parallel to the time x-axis, i.e. when the vertical velocity becomes constant. The simulated terminal velocity
is always closer to Wtmeas and WtNewt when the law (Eq. (18)) is used.

a the default drag law (Eq. (20)). Some frames of an experimental run are
100 displayed in Fig. 3. It is worth noting that, due to the irregular shape of
the pumice particles, the fall trajectory sometimes deviates from a
80 straight vertical line (dashed line). These secondary motions, ranging
Wtsim (cm s-1)

from small oscillation and rotation to tumbling and chaotic motions,


60
y = 0.97x + 0.73 are widely discussed in the literature (e.g. [32,12]). For this reason a
r = 0.92 scatter between the measured terminal velocities Wtmeas and those cal-
40
culated with the Newton's impact law (Eq. (12)) WtNewton appears in
Fig. 4, especially for the higher terminal velocities. The value of the ter-
20
minal velocity as resulting from numerical simulations have been ob-
tained by checking in the output files the vertical velocity at different
0
0 20 40 60 80 100 times and marking the time instant at which the vertical velocity be-
Wtmeas (cm s-1) came constant. An example of the terminal velocity determination is
shown in Fig. 5, where the vertical velocity (solid line) and the vertical
b 100 position in the conduit (dashed line) are plotted as a function of time
for two different experiments (Fig. 5a and b, respectively), each one
simulated by taking into account both our new drag law (Eq. (18),
80 y = 1.7x-2.77
Wtsim_no_Dellino (cm s-1)

r = 0.94 upper plot) and the default Wen and Yu drag law (Eq. (20), lower
plot). On the right axes WtNewton and Wtmeas are reported. The simulated
60
terminal velocity approaches WtNewton and Wtmeas much better when
40
our new drag law (Eq. (18)) is used. The simulation performed by
using the default Wen and Yu law results in a significant overestimation
20 depending on an underestimation of the drag force. The simulation re-
sults are closer to WtNewton than to Wtmeas. This is, again, an effect of
0 the irregular trajectory of particles, which is not accounted for neither
0 20 40 60 80 100
in the Newton impact law, nor in the simulation calculations. Anyway,
Wtmeas (cm s-1) simulation with our new drag law give always much closer results. In
Fig. 6a the simulated terminal velocity Wtsim, as obtained by means of
Fig. 6. a. Wtsim vs. Wtmeas plot. b. Wtsim_no_Dellino vs. Wtmeas plot. Each point represents an ex- our new drag law, is plotted against the measured terminal velocity
perimental run. The thick line is the linear regression line; the equation and the correlation
coefficient r are also reported. The thick dashed line is the equality line. It is evident that
Wtmeas. Despite the aforementioned scatter, results prove that, with
the implementation of (Eq. (18)) in MFIX-DEM allows to reproduce the measured termi- the new drag law, MFIX-DEM is able to replicate the experimentally
nal velocity (a), otherwise the trend is significantly different from the equality line (b). determined terminal velocities quite correctly. Even if the error is larger
76 F. Dioguardi et al. / Powder Technology 260 (2014) 68–77

a 100
the momentum exchange acts via a drag force, which depends on a
particle–fluid drag coefficient Cd. Although for spherical particles the
drag coefficient is solely a function of the Reynolds number and the
80
functional dependency is well established, for non-spherical particles
the application of a Cd–Re relationship is not straightforward. For this
Wtsim (cm s-1)

60
y = 0.98x + 0.44 reason we implemented in the open-source multiphase code MFIX a
r=1 drag relationship linking Cd to a particle shape factor obtained by a
40
number of settling experiments of irregularly-shaped pumice particles
in a column of water or alcohol. By using MFIX-DEM, the EL version of
20 MFIX, we simulated all the experimental runs with both the new and
the default particle–fluid drag laws. Results show a significant im-
0 provement in the prediction of the terminal velocity of non-
0 20 40 60 80 100
spherical particles if the new drag law is used. Thanks to this imple-
WtNewton (cm s-1) mentation, MFIX-DEM can now be successfully used for the simula-
tion of fluid–particle flows with non-spherical particles, which is
b 100 usually the case of environmental flows (ash and pollutant dispers-
al, explosive volcanic eruptions, etc.). In the future another
y = 1.52x + 1.55 development would be the extension of the here presented drag
80
Wtsim_no_Dellino (cm s-1)

r = 0.9 law to a multi-particle system in which particle interactions are


accounted for. This is the object of a research which is actually
60
underway and will consist in the derivation of a multi-particle
drag coefficient formula starting from the single-particle C d rela-
40 tionship and velocity–voidage correlations. This is similar to the
approach followed by Syamlal and O'Brien [33], who used alterna-
20 tively the Richardson and Zaki [34] and the Garside and Al-
Dibouni [35] equations. The suitability of this approach, which
0 was originally developed for spherical particles, needs an experi-
0 20 40 60 80 100 mental verification.
WtNewton (cm s-1) We think that this is a fundamental step towards more realistic two-
fluid model simulations, which are still the most feasible approach for
Fig. 7. a. Wtsim vs. WtNewt plot. b. Wtsim_no_Dellino vs. WtNewt plot. Each point represents an ex- simulating multiphase flows in which a large number of solid particles
perimental run. The thick line is the linear regression line; the equation and the correlation are involved.
coefficient r are also reported. The dashed line is the equality line. It is evident that the im-
plementation of (Eq. (18)) in MFIX-DEM allows to reproduce the theoretical terminal ve-
locities recalculated with the Newton's impact law (Eq. (12)) (a), otherwise the simulated
terminal velocities differ significantly from WtNewt (b).
Symbol notation

at higher terminal velocities, still the values are dispersed around the
equality line (dashed line). This is further confirmed by the linear Latin
correlation law that has an angular coefficient close to 1 (0.967) and
Symbol Description Units
an intercept slightly larger than 0 (0.729). The improvement of MFIX-
DEM terminal velocity prediction with the implementation of Eq. (18) S Stress tensor dyn cm−2
Ifm Interphase momentum exchange dyn cm−3
is even more evident by looking at Fig. 6b, in which the terminal velocity
X(i) Position vector of the ith particle cm
simulated with the default drag law of Wen and Yu, Wtsim_no_Dellino V(i) Velocity vector of the ith particle cm s−1
(Eq. (20)), is plotted against the measured terminal velocity Wtmeas. Al- ω(i) Angular velocity vector of the ith particle rad s−1
though the correlation coefficient is slightly larger than in the plot of D(i) Diameter of the ith particle cm
Fig. 6a (0.94 vs. 0.92), the correlation law follows a completely different ρ(i) Density of the ith particle g cm−3
m(i) Mass of the ith particle g
trend from the equality line, with an angular coefficient of 1.7 and an in- I(i) Moment of inertia of the ith particle g cm2
tercept value equal to −2.77. The improvement of MFIX-DEM with the F(i
d
∈ k,m)
Total drag force acting on the ith dyn
new drag law implementation is further confirmed by the plots in Fig. 7. particle in the kth cell
In Fig. 7a Wtsim is plotted vs. WtNewton: there is a complete agreement F(i)
c Net contact force acting on the ith particle dyn
T(i) Sum of the torques acting on the ith particle g cm2 rad s−2
(r = 1) between the simulated terminal velocities and those calculated
Ap Surface area of the actual particle cm2
with the Newton's impact law, as the regression line lies on the equality Asph Surface area of the equivalent sphere cm2
line. On the other hand, no agreement exists between the terminal ve- Cd Fluid–particle drag coefficient –
locity simulated with the default drag law Wtsim_no_Dellino and WtNewton dl Long axis of the particle cm
(Fig. 7b). This means that with non-spherical particles the default drag dm Medium axis of the particle cm
ds Small axis of the particle cm
law is not able to correctly predict the terminal velocity. It can be con- Dp Particle mean diameter cm
cluded that, thanks to the implementation of the shape-dependent K Kernel function for the calculation of –
drag law of Dellino et al. [25], MFIX-DEM is now able to satisfactorily the interphase drag
calculate the terminal velocity of non-spherical particles in a wide g Gravity acceleration cm s−2
i Number of the ith particle –
range of shape factors (from 0 to 1).
M Number of solid phases –
Mw Average molecular weight of the fluid –
4. Conclusions N Number of chemical species –
Np Number of particles –
The interphase momentum exchange is a fundamental quantity in P Pressure dyn cm−2
R Universal gas constant erg K−1 mol−1
the transport process of multiphase flows. In a fluid–particle mixture
F. Dioguardi et al. / Powder Technology 260 (2014) 68–77 77

Latin (continued)
(continued) [7] R. Garg, J. Galvin, T. Li, S. Pannala, Documentation of open-source MFIX-DEM soft-
ware for gas–solids flows, Computer Science and Mathematics Division, Oak Ridge
Symbol Description Units National Laboratory, Oak Ridge, TN, 2012.
Re Reynolds number – [8] G. Ganser, A rational approach to drag prediction of spherical and nonspherical
Res Relative Reynolds number – particles, Powder Technol. 77 (1993) 143–152.
[9] R.P. Chhabra, L. Agarwal, N.K. Sinha, Drag on non-spherical particles: an evaluation
t Time s
of available methods, Powder Technol. 101 (1999) 288–295.
T Temperature K
[10] S. Tran-Cong, M. Gay, E.E. Michaelides, Drag coefficients of irregularly shaped particles,
v Phase velocity m s−1
Powder Technol. 139 (2004) 21–32, http://dx.doi.org/10.1016/j.powtec.2003.10.002.
Vrm Terminal velocity correlation for – [11] A. Hölzer, M. Sommerfeld, New simple correlation formula for the drag coefficient of
particle mixtures non-spherical particles, Powder Technol. 184 (2008) 361–365, http://dx.doi.org/10.
Wt Particle terminal velocity cm s−1 1016/j.powtec.2007.08.021.
x Position vector in the computational domain cm [12] G.H. Bagheri, C. Bonadonna, I. Manzella, P. Pontelandolfo, P. Haas, Dedicated vertical
z Coefficient for the scalene ellipsoid area – wind tunnel for the study of sedimentation of non-spherical particles, Rev. Sci.
Pmp Measured particle perimeter cm Instrum. 84 (2013) 054501, http://dx.doi.org/10.1063/1.4805019.
Pp Perimeter of the circle equivalent to cm [13] M. Nikku, P. Jalali, J. Ritvanen, T. Hyppänen, Characterization method of average gas–
the maximum projection area solid drag for regular and irregular particle groups, Powder Technol. 253 (2014)
284–294, http://dx.doi.org/10.1016/j.powtec.2013.11.035.
[14] L. Wilson, R.S.J. Sparks, G.P.L. Walker, Explosive volcanic eruption — IV. The control
of magma properties and conduit geometry on eruption column behavior, Geophys.
Greek J. R. Astron. Soc. 63 (1980) 117–148.
[15] A.W. Woods, The fluid dynamics and thermodynamics of eruptions columns, Bull.
Symbol Description Units Volcanol. 50 (1988) 169–193.
[16] P. Dellino, F. Dioguardi, B. Zimanowski, R. Büttner, D. Mele, L. La Volpe, R. Sulpizio,
ε Phase volume fraction – D.M. Doronzo, I. Sonder, R. Bonasia, S. Calvari, E. Marotta, Conduit flow experiments
ρ Density g cm−3 help constraining the regime of explosive eruptions, J. Geophys. Res. 115 (2010)
ν Volume cm3 B04204, http://dx.doi.org/10.1029/2009JB006781.
β Local fluid–solid momentum transfer g cm−3 s−1 [17] R. Büttner, P. Dellino, L. La Volpe, V. Lorenz, B. Zimanowski, Thermohydraulic explo-
Ψ Particle shape factor – sions in phreatomagmatic eruptions as evidenced by the comparison between
Φ Particle sphericity – pyroclasts and products from molten fuel coolant interactions experiments, J.
X Particle circularity – Geophys. Res. 107 (2002), http://dx.doi.org/10.1029/2001JB000511.
[18] R. Büttner, P. Dellino, H. Raue, I. Sonder, B. Zimanowski, Stress-induced brittle frag-
μ Viscosity poise
mentation of magmatic melts: theory and experiments, J. Geophys. Res. 111 (2006)
B08204, http://dx.doi.org/10.1029/2005JB003958.
[19] U. Kueppers, B. Scheu, O. Spieler, D.B. Dingwell, Fragmentation efficiency of explo-
Subscripts and superscripts sive volcanic eruptions: a study of experimentally generated pyroclasts, J. Volcanol.
Geotherm. Res. 153 (2006) 125–135, http://dx.doi.org/10.1016/j.jvolgeores.2005.
08.006.
Symbol Description [20] T. Dürig, F. Dioguardi, R. Büttner, P. Dellino, D. Mele, B. Zimanowski, A new method
f Fluid phase for the determination of the specific kinetic energy (SKE) released to pyroclastic par-
s Solid phase ticles at magmatic fragmentation: theory and first experimental results, Bull.
m mth solid phase Volcanol. 74 (2012) 895–902, http://dx.doi.org/10.1007/s00445-011-0574-9.
[21] Y.J. Suzuki, T. Koyaguchi, 3-D numerical simulations of eruption column collapse:
l lth solid phase
effects of vent size on pressure-balanced jet/plumes, J. Volcanol. Geotherm. Res.
k Cell number
221–222 (2012) 1–13, http://dx.doi.org/10.1016/j.jvolgeores.2012.01.013.
Newton Terminal velocity calculated with [22] S. Dartevelle, G.A. Valentine, Transient multiphase processes during the explosive
the Newton's impact law eruption of basalt through a geothermal borehole (Namafjall, Iceland, 1977)
meas Measured terminal velocity and implications for natural volcanic flows, Earth Planet. Sci. Lett. 262 (2007)
sim Simulated terminal velocity with the new drag law 363–384, http://dx.doi.org/10.1016/j.epsl.2007.07.053.
sim_no_Dellino Simulated terminal velocity with the default drag law [23] T. Esposti Ongaro, A.B. Clarke, B. Voight, A. Neri, C. Widiwijayanti, Multi-
phase flow dynamics of pyroclastic density currents during the May 18, 1980 lateral
blast of Mount St. Helens, J. Geophys. Res. 117 (2012), http://dx.doi.org/10.1029/
2011JB009081.
[24] D.M. Doronzo, J. Martí, R. Sulpizio, P. Dellino, Aerodynamics of stratovolcanoes dur-
ing multiphase processes, J. Geophys. Res. 117 (2012) B01207, http://dx.doi.org/10.
Acknowledgments 1029/2011JB008769.
[25] P. Dellino, D. Mele, R. Bonasia, G. Braia, L. La Volpe, R. Sulpizio, The analysis of the
The MFIX staff and support greatly helped in our understanding of influence of pumice shape on its terminal velocity, Geophys. Res. Lett. 32 (2005)
L21306, http://dx.doi.org/10.1029/2005GL023954.
the details of the code and in the implementation of the new drag law.
[26] M. Syamlal, W. Rogers, T.J. O'Brien, MFIX documentation: theory guide, Tech. Rep.,
This work was supported by INGV-DPC 2013, Project V1: Probabilistic DOE/METC-95/1013, NTIS/DE95000031, National Energy Technology Laboratory,
volcanic hazard evaluation. Department of Energy, Morgantown, WV, 1993.
[27] R. Garg, J. Galvin, T. Li, S. Pannala, Open-source MFIX-DEM software for gas–solids
flows: part I — verification studies, Powder Technol. 220 (2012) 122–137, http://
Appendix A. Supplementary data dx.doi.org/10.1016/j.powtec.2011.09.019.
[28] T. Li, R. Garg, J. Galvin, S. Pannala, Open-source MFIX-DEM software for gas–solids
flows: part II — validation studies, Powder Technol. 220 (2012) 138–150, http://
Supplementary data to this article can be found online at http://dx. dx.doi.org/10.1016/j.powtec.2011.09.020.
doi.org/10.1016/j.powtec.2014.03.071. [29] P. Dellino, M.T. Gudmundsson, G. Larsen, D. Mele, J.A. Stevenson, T. Thordarson, B.
Zimanowski, Ash from the Eyjafjallajökull eruption (Iceland): fragmentation pro-
cesses and aerodynamic behaviour, J. Geophys. Res. 117 (2012) B00C04, http://dx.
References doi.org/10.1029/2011JB008726.
[30] J.F. Dieteker, Multiphase Flow with Interphase eXchanges Cartesian Grid User Guide,
[1] G.V. Middleton, J.B. Southard, Mechanics of Sediment Movement, 2nd ed. Society of National Energy Technology Laboratory, Morgantown, WV, 2013.
Economic Paleontologists and Mineralogists, Tulsa, OK, 1984. [31] ANSYS FLUENT Theory Guide, Release 14.5, ANSYS Inc., Canonsburg, PA, 2012.
[2] A. Prosperetti, G. Tryggvason, Computational Methods for Multiphase Flow, 1st ed. [32] A. Chow, E.E. Adams, Prediction of drag coefficient and secondary motion of free-
Cambridge University Press, 2007. falling rigid cylindrical particles with and without curvature at moderate Reynolds
[3] C.T. Crowe, J.D. Schwarzkopf, M. Sommerfeld, Y. Tsuji, Multiphase Flows with Drop- number, J. Hydraul. Eng. 137 (2011) 1406–1414, http://dx.doi.org/10.1061/(ASCE)
lets and Particles, 2nd ed. CRC Press, Boca Raton, FL, 2011. HY.1943-7900.0000437.
[4] C.Y. Wen, Y.H. Yu, Mechanics of fluidization, Chem. Eng. Prog. Symp. Ser. 62 (1966) [33] M. Syamlal, T.J. O'Brien, The derivation of a drag coefficient formula from velocity–
100–111. voidage correlations, Tech. Report, US Department of Energy, 1987.
[5] M. Syamlal, T. J. O'Brien, A generalized drag correlation for multiparticle systems, [34] J.F. Richardson, W.N. Zaki, Sedimentation and fluidization. Part 1, Trans. Inst. Chem.
Unpublished report, U.S. Department of Energy (1987). Eng. 32 (1954) 35–53.
[6] D. Gidaspow, R. Bezburuah, J. Ding, Hydrodynamics of circulating fluidized beds, [35] J. Garside, M.R. Al-Dibouni, Velocity–voidage relationships for fluidization and sedi-
kinetic theory approach in fluidization VII, Proceedings of the 7th Engineering mentation, Ind. Eng. Chem. Process Des. Dev. 16 (1977) 206–214.
Foundation Conference on Fluidization, 1992, pp. 75–82.

Vous aimerez peut-être aussi