You are on page 1of 80

Accepted Manuscript

Review

A review on gravity flow of free-flowing granular solids in silos - Basics and


practical aspects

Khashayar Saleh, Shahab Golshan, Reza Zarghami

PII: S0009-2509(18)30597-9
DOI: https://doi.org/10.1016/j.ces.2018.08.028
Reference: CES 14443

To appear in: Chemical Engineering Science

Received Date: 29 December 2017


Revised Date: 5 August 2018
Accepted Date: 12 August 2018

Please cite this article as: K. Saleh, S. Golshan, R. Zarghami, A review on gravity flow of free-flowing granular
solids in silos - Basics and practical aspects, Chemical Engineering Science (2018), doi: https://doi.org/10.1016/
j.ces.2018.08.028

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A review on gravity flow of free-flowing granular solids in silos - Basics and practical

aspects

Khashayar Saleh a *, Shahab Golshanb, Reza Zarghamib


a
Sorbonne Universités/Université de Technologie de Compiègne, EA 4297 Transformations
Intégrées de la Matière Renouvelable, France
b
Process Design and Simulation Research Centre, School of Chemical Engineering, College
of Engineering, University of Tehran, P.O. Box 11155/4563, Tehran, Iran

Abstract
This paper provides a review on the flow of free-flowing coarse particles inside silos. Basic

principles and elementary mechanisms involved in gravity flow of granular materials are

discussed. The main focus is put on flow patterns, stress distribution, velocity profiles, mass

flowrate, Residence Time Distribution (RTD) and segregation inside the silo. A summary of

empirical correlations as well as a synthesis of more physical and mechanistic models related

to these aspects are presented. Computational and numerical methods used to describe the

granular flow of particles will be discussed in a separate paper.

Keywords: Granular solids, silo, hopper, discharge rate, Residence Time Distribution,

velocity profiles, flow pattern, segregation

1. Introduction
Gravity flow of powders and granular materials inside hoppers and silos have long been a

matter of interest for both academic and industrial communities. If this subject has intrigued

*
Corresponding author, E-mail address: khashayar.saleh@utc.fr, khashayar.saleh@gmail.com
many scientists and philosophers since antiquity, it has taken on a new dimension since the

industrial era when increasing quantities of granular material had to be stored as raw materials

or as end products (e.g. coal, ores, sugar, grains and cereals, floor, cement, etc.). Indeed, it is

difficult to find a granular product which has not been, at one time or another of its life cycle,

in a storage silo.

From a technological point of view, a silo might seem to be the simplest device that can be

imagined to store bulk solids. Several criteria however must be taken into account for a

successful operation of a silo. The ideal silo must be able to contain the required quantity of

product on a minimum floor area and ensure a regular flow (without chimney or arch

formation), with the desired flow rate. However, several problems are regularly encountered

during the storage or discharge of solids: funnel flow, arching, rat-hole and dead zone

formation, irregular flow and storage residence time, flooding, segregation, settling, caking,

self-heating, explosion, etc. (Conesa et al., 2004; Jenike, 1954c; Leturia et al., 2014; Marinelli

and Carson, 1992; Matchett, 2007; Pemberton, 1965; Prescott and Barnum, 2000; Purutyan et

al., 1998, 1999; Royal and Carson, 1991; Saleh and Guigon, 2012; Schulze, 2008; Shamlou,

1988). Although the majority of these troubles are specific to fine and cohesive powders, free-

flowing granular materials (subject of the present study) are also prone to some problems. In

fact, even for free-flowing particles and with steep hoppers, the mass flow is not always

guaranteed and an internal moving conical zone (chimney) could appear which results in

heterogeneity of storage times. There have been many studies carried out on granular flow

through various types of silos using a variety of different materials. An extensive review of

the literature on this topic is out of the scope of the present work but comprehensive and

detailed description of basic fundamentals can be found elsewhere (e.g. (Brown and Nielsen,

1998; Brown and Richards, 1970; Drescher, 1991; Jenike, 1954c, 1961c; Ketchum, 1911;

Khashayar and Pierre, 2009; Nedderman, 1992; Saleh and Guigon, 2009, 2012; Schulze,

2
2008; Shamlou, 1988)). Obviously, the most of attention has been paid to the flow of cohesive

powders because of difficulties that they cause but coarse and free-flowing particles present

also some challenging issues.

In the following, we limit our survey to the flow behaviour of granular solids in silos. The

term “granular solids” refers to coarse particles for which the gravity force is dominant with

respect to inter-particle interactions (e.g. van der Waals or electrostatic forces). They therefore

exhibit a free-flowing behaviour. Examples of this kind of product are cereal grains, polymer

pellets, catalyst beads, pharmaceutical tablets or pills, detergent or fertilizer prills, gravels,

etc. The points of special interest for such systems include the prediction and modelling of

flow patterns, wall and bulk stresses, velocity profiles, exit mass flow rates, the Residence

Time Distribution (RTD) and homogeneity of the composition.

2. Generality and Definitions


The flow patterns of granular materials (and then the mass flowrate and the Residence

Time Distribution) during the discharge depends tightly to the geometry of the hopper. This

geometry is generally defined based on a compromise between a high storage capacity and a

regular flow of particles. Obviously, for a given floor area, flat-bottom silos provide the

maximum capacity but are subject to funnel flow and dead zone formation whereas steep

hoppers improve the flow but lead to a lack of the storage volume.

A large variety of geometries of industrial silos have been reported. Figure 1 provides

some examples of the most common industrial silos. A first distinction could be made based

on the shape of the discharging region (Johanson, 1995; Purutyan et al., 1998; Schulze, 2008;

Shamlou, 1988). In this regard, silos could be classified as flat-bottom and inclined-bottom

silos. Different cross-sectional geometries could exist although the most commonly used are

circular or rectangular. Inclined-bottom silos are constituted of two superposed parts, a

3
converging inferior vessel, called hopper and an upper cylindrical vessel (bin). The hopper

could be asymmetric, plane-symmetric or axisymmetric with different shapes: e.g. conical,

pyramidal, wedged and Chisel, etc. Also, the hopper could be more or less deep and more or

less steep and different geometries of the exit aperture are possible too (circular, rectangular,

slot, etc.). Note that flat silos (2D silos) are also very popular for research purposes and have

been extensively used to simulate the flow inside axisymmetric 3D silos.

It is also well established that according to the silo geometry and flow properties of

powders, the flow could occur in two different modes (Figure 2): mass flow where all the bulk

material moves during discharge and the funnel flow where only a part of particles flow

through a central chimney surrounded by a dead zone (e.g. (Jenike, 1954b; Jenike, 1954a, c,

1961a, 1962a; Marinelli and Carson, 1992; Pemberton, 1965; Prescott and Barnum, 2000;

Purutyan et al., 2004; Purutyan et al., 1998, 1999; Saleh and Guigon, 2012; Schulze, 2008;

Shamlou, 1988; Tüzün and Nedderman, 1982)). Furthermore, the funnel flow can occur in

“parallel pipe flow”, “taper pipe flow” or “mixed flow” (Figure2, (Ooi et al., 1998)) but in

case of granular materials only the two latter modes are encountered. Note also that the term

“first-in/first-out” refers to a perfect plug flow, which is a special (ideal) case of the mass

flow.

3. Flow patterns
Velocity profiles and the RTD are closely related to the patterns formed during the

emptying of the silo. Many works carried out on this area have shown that the flow patterns of

granular solids differ significantly from those of fine cohesive powders (Brown and Nielsen,

1998; Brown and Richards, 1970; Shamlou, 1988).

For convergent channels (conical and wedge-shaped hoppers) in mass flow as well as for

funnel flow silos with a conical central moving zone, the radial flow model allows, to a first

4
order, a reasonable representation of the flow. This model, introduced first by Jenike (Jenike,

1961b) and Brown (Brown, 1961), is presented in Figure 3. According to this model, in

spherical coordinates (or circular coordinates for flat hoppers) particles keep their angular

position, , all along their trajectory. Obviously, for funnel flow hoppers this hypothesis

applies only to the central moving zone (funnel) if this latter has a conical shape (Fig. 3-b).

For cylindrical silos, the radial flow is no more valid. Numerous experimental

investigations have pointed out the presence of several zones during the discharge of free-

flowing coarse particles (Brown and Hawksley, 1947; Chatlyne and Resnick, 1973; Chen et

al., 2005, 2007; Chou et al., 2002; Cleary, 1999; Drescher and Ferjani, 2004; Gardner, 1966;

Gentzler and Tardos, 2009; Giunta, 1969; Hunt et al., 1999; Kvapil, 1965; Levinson et al.,

1977; Mullins, 1972, 1974; Nedderman and Tüzün, 1979; Nedderman et al., 1982; Ooi et al.,

1998; Sielamowicz et al., 2005; Steingart and Evans, 2005; Tüzün et al., 1982; Tüzün and

Nedderman, 1985). Indeed, fully mass flow rarely takes place and the presence of four

different zones (i.e. a slow motion zone, a dilated core, a dead zone and a free-fall zone) as

reported by Brown and Hawksley (Brown and Hawksley, 1947) and confirmed by other

experimental (e.g. (Chen et al., 2005, 2007; Chou et al., 2002; Drescher and Ferjani, 2004;

Gardner, 1966; Gentzler and Tardos, 2009; Giunta, 1969; Job et al., 2009; Kvapil, 1965;

Mullins, 1972, 1974; Nedderman and Tüzün, 1979; Nedderman et al., 1982; Ooi et al., 1998;

Sielamowicz et al., 2005; Steingart and Evans, 2005; Tüzün et al., 1982; Tüzün and

Nedderman, 1982, 1985) and theoretical investigations (e.g. (Balevičius et al., 2007;

Balevičius et al., 2008; Choi et al., 2005; Chou et al., 2002; Cleary, 1999; Drescher and

Ferjani, 2004; Ketterhagen et al., 2008, 2009; Langston et al., 1995; Meng et al., 1997;

Nguyen et al., 2009; Volpato et al., 2014)) are generally admitted. These zones are described

further in this section.

5
Nedderman and Tüzün (Nedderman and Tüzün, 1979) provided a detailed investigation of

the dynamics of flow. Indeed, all phenomena originate at the exit. When the aperture opens,

the particles at the immediate vicinity of the orifice fall out leaving the place for other

particles to move. The particles then move downward with a velocity that depends on their

position, on the stresses that they undergo and on the dissipative forces (e.g. friction,

stickiness) who oppose the flow. According to Nedderman and Tüzün, after the opening of

the aperture, first a transient flow regime occurs during which the bed voidage changes from

the initial packed state to that of the flowing state. It is only after this dilation period (a few

seconds) that the flow profiles become well established and distinct. Drescher and his co-

workers (Drescher, 1992; Drescher and Ferjani, 2004) brought experimental evidence of the

presence of the transient regime during the flow of cohesionless materials in flat-bottom bins

with centric apertures. According to these authors, following the opening of the outlet, a first

“confined flow stage”, referred as phase I, takes place during which a narrow plug of width d

with a conical upper boundary appears (Fig. 4-a) and grows laterally and vertically upward

until it attains the surface (Fig. 4-b). Next the first phase I, these authors distinguished an

"unconfined regime" including five further successive phases. However, phases III and IV

proposed by these authors show little or no difference and are grouped together in Fig. 4-c to

h. Once the upper part attains the surface of the bed, a V-shaped dip with an inclination equal

to the angle of repose appears (Fig. 4-c). During this phase, two processes contribute to the

flow, which are downward gravity flow and upward propagation of density shock. Phase II

ends when the propagation of the central plug is complete. At this stage, the width of V-

shaped interface is equal to that of the central plug (Fig. 4-d). During phase III, the V-shape

surface grows laterally (Fig. 4-e). In phase IV avalanches appear due to vertical steps left by

the movement of the central zone (Fig. 4-f). The last phase (V) begins when the edge of

concave surface attains the walls (Fig. 4-g). Avalanches continue, the shape of avalanching

6
zone at the top of the bed becomes stable whereas the height of plug zone decreases (Fig. 4-

h). Drescher et al. provided a model to describe the volume of each zone at different phases

(Drescher and Ferjani, 2004; Waters and Drescher, 2000). The presence of these different

phases have been confirmed by experimental works of Drescher et al. (Drescher and Ferjani,

2004) and those of Sielamowicz et al. (Sielamowicz et al., 2005). However, according to

some other investigators, not all these phases (especially phase V) exist necessarily for all

materials.

Moreover, during the unconfined regime, once the flow is well established, several flow

zones are distinguishable (Fig. 5). In case of flat-bottom hoppers or funnel flow, dead zones in

the lower corners of the silo appear. The transition between the lateral stagnant zones and the

central converging (wedged) zone at any given horizontal level takes place over a very thin

region across which there is a rapid change in velocity (from almost zero in the vicinity of the

wall to a maximum at the centre of the flowing zone). Commonly, this small band is

considered, as a first approximation, as a surface of rupture (called “surface of sliding”) at

which velocities and stresses are discontinuous. Generally, the moving zone propagates from

the edges of the opening to the top. The surface of sliding is not necessarily straight but more

often curved and concave. The width of the central moving zone increases with decreasing the

internal angle of friction and/or the half angle of silo (Gardner, 1966). The angle of the

surface (to the hopper’s wall) at the aperture edges is known as the “angle of approach”

(designated by  in Fig. 5). In the case of deep beds, the surface of sliding intersects the wall

at a given level and not necessarily on the top surface. Therefore, the flow proceeds in mixed-

flow (Fig. 2) and the flow pattern of particles above this level is vertical but not necessarily

uniform over the entire section of the bed. In the case of less deep beds, the surface of sliding

ends at the bed surface, which means that the height of the dead zone rises to the surface of

the bed and the discharge, occurs in funnel flow. In both cases, because of the higher
7
velocities of particles at the centre of the moving zone (e.g. symmetrical axis for conical

hoppers) the bed level drops faster in the centre. This results in a collapse in the centre of the

bed surface and formation of a V-shaped interface. If the angle of this interface (to the

horizontal) exceeds the angle of repose of the granular material, avalanches and rolling of

particles will occur. In this case, two juxtaposed mobile zones can be distinguished, the

former being delimited by the surface of sliding at the lower side and by the angle of repose at

the upper size (Zone B in Fig. 5) and the latter extending from this lower zone to the interface

of the bed (Zone C in Fig. 5). Note that in the case of conical or wedge-shaped hoppers with

steep walls (i.e. a half-angle with respect to the vertical smaller than the angle of approach, )

the stagnant zones disappear but the other zones persist. As for silo half-angles, , higher than

, the flow patterns are similar to those of flat-bottom hoppers except that the volume of

stagnant zones is smaller.

In addition, there exists a relatively small-dilated zone close to the aperture within which

the particles lose contact with each other. Consequently, particles are no more under shear and

fall freely. The surface below which this happens is assumed to have the shape of a dome and

is referred as the free-fall arch. Originally, the presence of such a zone was suspected from

the fact that the mass flow rate of exiting solids is independent of the pressure head (height of

filling). This makes plausible the idea of the existence of an arch supporting the weight of the

particles lying above this area. As a rough estimation, the height of this zone is such that the

particles reach only 1% of their terminal velocity at the exit of the hopper (Shamlou, 1988).

Although, the physical consistency of this heuristic model has been put in doubt by some

investigators (e.g. (Janda et al., 2012; Rubio-Largo et al., 2015)), the "free-fall arch" still

constitutes one of the main hypothesis in gravity flow of granular materials, especially for

mass flowrate estimation (see §5).

8
Finally, note also that some factors such as silo geometry (Kvapil, 1965), presence of

inserts (Kvapil, 1965; Ooi et al., 1998; Tüzün and Nedderman, 1985), multiple or eccentric

orifices (Chen et al., 2005, 2007; Chou et al., 2002; Kvapil, 1965; Nedderman and Tüzün,

1979), presence of table underneath the opening (Kvapil, 1965) or humidity (Purutyan et al.,

2004) modify significantly the flow pattern within the silo.

4. Stresses fields
Although the flow of granular materials is less influenced by the state of stresses within the

silo than fine powders, the knowledge of the distribution of stresses is however important for

structural and mechanical design purposes.

Roberts was the first who showed that the total pressure on the bottom of a silo was less

than that which should be expected due to the weight of the material contained in the silo

(Roberts, 1882, 1884). Since the work of Roberts in 1882, many models have been developed

for calculating stresses within silos. These models can be classified into two categories

(Khelil, 1989; Khelil and Roth, 1990):

- Those based on the calculation of the static pressures from the static equilibrium of the

granular material inside the silo (Airy, 1897; Janssen, 1895b; Koenen, 1896; Reimbert

and Reimbert-Auclair, 1982).

- Those based on the calculation of the total pressures from the static pressures to which

the overpressures during drainage are added. These latters could appear due to a variety

of causes such as sudden changes in flow mode, flow velocity, surface of the walls,

shocks due to breaks in arches, etc. (Caquot and Kerisel, 1956; Theimer, 1958; Walker,

1966; Walters, 1973).

9
Obviously, the majority of analytical models are derived from Janssen's model. A brief

reminder of the statics of granular media in general and of the Janssen’s model in particular

would then be useful to introduce the basic assumptions of these methods.

It is well known that stresses within granular media are neither hydrostatic nor isotropic.

Indeed, in those media, forces are transmitted from particle to particle through random contact

networks. In each contact point, a part (and only a part) of the normal force is then transmitted

laterally toward the walls (Figure 6). This results in a non-linear evolution of the normal and

lateral pressure profiles according to a decelerating exponential law with an asymptotic value

which is reached at several-fold bin diameter (§4.1).

The ratio between the lateral and normal stresses is denoted the lateral stress (or pressure)

ratio and designated by (or alternatively by K):

(1)

Typical values of  are between 0.3 and 0.6. Its value could be measured using a uniaxial

compression test. In special case of regular packing, could be calculated theoretically. For

example, for a triangular packing of equal-sized particles =0,58. For a coulomb solid,  can

be estimated from the relationship between minor and major principal stresses (Schulze,

2008):

(2)

where  is the angle of internal friction of the powder. An alternative equation recommended

in some silo codes is (Schulze, 2008) :

(3)

In absence of experimental data, for a rough estimation of this parameter a value of =0,4 is

generally acceptable (Roberts, 1990). Note that some experimental works have shown that 

varies between static (filling) and dynamic (emptying) states. For example, the experimental

10
works of Khelil and Roth (Khelil and Roth, 1990) on a full-scale silo showed values of

about 25% to 50% higher during drainage.

A large number of investigations has been carried out on the stress fields in silos.

Extensive and comprehensive presentations of the most relevant works are available in some

reference books (Drescher, 1991; Nedderman, 1992; Schulze, 2008; Shamlou, 1988). In this

section, we limit our discussion to a summary of the main highlights of these works.

Undoubtedly, one of the most important substantiated facts is that the distribution of

stresses inside silos are different during the filling (static state, passive walls) and the

discharge (dynamic state, active walls). Several experimental works have pointed out a sharp

increase in both lateral and vertical stresses during the discharge. For example, Khelil and

Roth (Khelil and Roth, 1990) reported a sharp raise about 50 to 100%. In addition, the stress

profiles vary differently in cylindrical upper part (bin) and in convergent part (hopper). These

cases must then be considered distinctly. Table 1 summarizes the most commonly used

correlations for both bin and hopper sections during the static and dynamic states. It should be

noted that these equations are given as an indication and provide only a rough estimate of the

constraints. Other more sophisticated models have been proposed by several researchers

(Schulze, 2008; Shamlou, 1988) but they remain just as approximate. Several parameters such

as surface imperfections and the presence of inserts or extractors (not considered in these

models) affect the state of stresses within a silo. In addition, the experimental validation of

these models as well as the estimation (and variations) of bulk properties (density, angles of

internal and wall friction, lateral stress ratio) are also subject to much uncertainty.

Consequently, the differences between predictions of different models remain inferior to

experimental errors (Shamlou, 1988).

11
4.1 Static stress fields (filling)

For the cylindrical section (bin) the variation of the average values of vertical and lateral

stresses with the depth of filling could be estimated by Janssen’s model. For an infinite

cylindrical column of powder at rest, the model of Janssen provides an acceptable estimation

of the vertical profile of bulk and wall stresses. Applying a force balance on a slice element of

volume of infinitesimal height, dh, and considering that both the bulk density and  are

constant and with the assumption that the vertical stress is acting evenly over the cross-

sectional area of the bed, Janssen (Janssen, 1895a) established the following equation for the

vertical stress:

(4)

whereas the lateral stress can be calculated from Eq. 1 ( ).

This equation shows that the vertical stress decreases exponentially with the depth of the

bed to reach its maximum value (i.e. ). This asymptotic value (limit of

as h tends to infinity) is almost reached for a value of the order of 4. Then, the

hydrostatic head of the stored powder above the discharge aperture is not entirely available

and this explains the experimental observation that for most silos the discharge rate is

practically independent of the bed height.

Note that originally, Janssen’s model has been established for infinitely deep silos as it

neglects any bottom effect as a boundary condition. Evesque and de Gennes (Evesque and de

Gennes, 1998) defined a characteristic height below which the Janssen model is not valid. For

shallow bins, the mean vertical pressure has been found to be higher than that predicted by

this model. Reimbert (Reimbert and Reimbert-Auclair, 1982) proposed an alternative

correlation to take account of this difference for circular bins (Table 1, Eq. T1.2). A

12
comparison between the model of Janssen and the correlation of Reimbert is presented in

Figure 7. Two almost limit cases have been considered: the first one corresponding to a small

wall friction and small lateral coefficient ( ) and the second to a high wall friction

and high lateral coefficient ( ). As can be seen, the difference between the two models

decreases strongly with the value of . For both models lead to similar results

beyond a h/D ratio of about 1.5. On the opposite case ( ), the difference becomes

more significant and spreads over a wider range of h/D ratios (up to 10).

Walters (Walters, 1973) brought a modification to Janssen's model by considering a non-

uniform distribution of the vertical stress over the cross-sectional area of the bed. This author

defined the coefficient as the ratio between the horizontal stress exerted at a given level and

the mean vertical stress. According to his analyses, is a function of the angle of internal

friction and the angle of wall friction. Depending on the values of , Walters defined the state

of static and dynamic stresses associated with the filling and emptying, respectively.

On the other hand, based on an idea of Freese (Freese, 1977), Sokol (Sokol, 1984)

introduced the variability of as a function of depth in Janssen's equations.

Finally, for non-circular bins, the Janssen’s model or Walter’s model could be applied by

replacing the bin radius, D/2, by its equivalent hydraulic radius defined as the ration of the

cross-sectional area to the perimeter of the silo.

Regarding the convergent section, a linear hydrostatic stress gradient has been suggested

by Jenike et al. and Walker (Walker, 1966). (Eq. T1.3).

The overall static stress field within the silo and on the wall can then be approximated by

juxtaposition of above-mentioned models. A qualitative representation is given in Figure 8-a.

13
4.2. Dynamic stresses (Discharge)
Until the middle of the 20th century, the methods used to predict the load on the silo walls

were based on static models, notably that of Janssen. However, these models found to be

unreliable for dynamic stresses because they considerably underestimate the loads on the

walls during the emptying of the silos. This inconsistency caused a number of accidents and

destruction of silos because the distribution of the stresses during the flow of the solid differs

radically from that given by the Janssen model for the static regime. Indeed, significant

overpressures in the order of 2 for granular materials have been measured experimentally (e.g.

(Khelil, 1989; Khelil and Roth, 1990)). This difference is particularly significant in hopper

section and arises from a fast shift from the static stress field (characterized by almost vertical

lines of major principal stress) to the dynamic field where the lines of major principal stress

become radial. The switch starts from the exit orifice and propagates promptly upward until it

is locked by the wall in an effective transition point, at a height, he. Above this point, the

stress field within the bin remains vertical and hence, there is a discontinuity on the stress

field at the transition point between the converging section and the vertical section. Note that

for mass flow silos, the transition happens at the intersection between the bin and the hopper

whereas in case of funnel flow, the effective transition point differs from the geometrical one

and corresponds to the intersection of the surface of sliding with the wall.

Several investigations had shown that for the vertical section in mass flow, the model of

Janssen could be conveniently applied provided the lateral walls are strictly straight and

without any surface irregularities and imperfections (due to welding for example) (Shamlou,

1988). However, this is rarely true for industrial silos and in this case, some corrections must

be brought to Janssen’s model. These involve either the direct overestimation of the stress by

applying a correction factor or to use an overestimated value for the lateral ratio (i.e. =1)

during the dynamic regime (Shamlou, 1988; Walters, 1973).

14
Walker (Walker, 1966) and Walters (Walters, 1973) established a model by applying the

elemental slice approach of Janssen in the converging section of the hopper (Eq. T1.4).

A qualitative representation of the dynamic stress field is given in Figure 8-b. Some other

qualitative examples for both static and dynamic states can be found in Ref. (Schulze, 2008).

Finally, the equations presented in the Table 1 have been established for silos in mass flow.

These models based on the equilibrium of the whole layer of material assume that the material

glides along the wall. For flat-bottom silos in funnel flow, Jenike (Jenike et al., 1973)

proposed a modified equation (Eq. T1.5) by using a correcting factor of lateral transmission

ratio and by replacing h in Janssen model by he (the height of powder above the effective

transition point between mass and funnel flow).

Other detailed models based on numerical resolution of fundamental laws of mechanics in

a local scale (rather than a slice) have been also reported in literature. Those models will be

discussed in the second part of the present work.

5. Velocity profiles
A large number of experimental and theoretical works had focussed on the flow fields and

velocity profiles during the discharge of silos. The experimental approach is usually based on

use of tracers that can be detected and tracked by appropriate in situ or ex situ probes. In 1982

Tüzün et al. (Tüzün et al., 1982) gave an extensive review on experimental techniques

reported hitherto. According to these authors, tracer methods that allow a direct observation of

the flow field could be classified into three categories, namely photographic, X-ray and -ray

and “flow-freezing” methods. As for the tracers, they may be coloured particles, radio-

isotopes, radio-pills or magnetic tracers. Specially designed probes (i.e. obstacle, capacitance

and fibre optic probes) have also found applications in this field. Since the review paper of

Tüzün et al. (Tüzün et al., 1982), the basic principles of these methods have not changed

15
much, although significant improvements in their resolution and response time as well as the

post-treatment of the data have been made. Important to note is that when using a tracer, it is

essential that the tracer follows the bulk flow as intimately as possible and that no significant

segregation due to size, shape or density differences between the tracer and particles could

happen.

Regarding the experimental study of the velocity profiles, two main operating modes have

been considered in literature:

- Steady-state mode where fresh material is supplied continuously to the top of the

hopper in a loose state to compensate the amount of solid which have been withdrawn.

According to Tüzün et al. (Tüzün and Nedderman, 1982) after a transient initial phase

during which the material moves and dilates from its original density to that

characteristic of the flowing material, the steady-state regime is achieved during which

the velocities at a given point remain stable. In order to perform experiments at steady-

state conditions, Nedderman and his co-workers (Nedderman and Tüzün, 1979) made

use of a large reservoir on the top of emptying silo so that the hopper could be kept full

of granular material for periods longer than 3 mean residence time. Obviously, under

those conditions, the avalanche region (zone C in figure 2) disappears and the velocity

profiles along the bed vary no more with time. If not doing so, according to these

authors the flow must be considered as a combination of transient and steady flow,

which make the interpretation of results difficult. On the other hand, most industrial

silos operate in transient mode and the transposition of data obtained at steady-state

conditions is not without difficulty either.

- Batch discharge where the material is filled once at the beginning of the operation and

then it is only withdrawn from the silo. This type of operation is the most encountered

in industry and is unavoidably transient in nature.

16
Let us focus first on the axial profile of the average velocity across the bed. In the absence

of side withdrawal, in Cartesian coordinates, the x-component of the average velocity must be

nil. The y-component of this velocity at any given level could be obtained if the geometry of

the moving zone and the average velocity at an arbitrary point is known. This reference point

is usually at the exit where the mass flow rate of particles could be measured quite easily. In

case of a straight surface of sliding, the moving zone is conical and the flow pattern is radial.

Assuming that the flowing media is incompressible (constant density), it can be easily shown

from the continuity equation ( ) that the velocity decays as a power law along the

vertical axis with exponent 2 (because , and ) (Nedderman, 1988). However, on

the one hand, the shape of the moving zone is not always known and on the other hand the

hypothesis of incompressible flow is irrelevant especially since this parameter is usually

measured at the output where the density corresponds to that of the free-fall (dilated) zone

(§3). In addition, the average velocity allows the determination of the space-time only (mean

residence time) but does not allow the determination of the RTD which requires the

knowledge of the velocity profiles in reference directions. On this matter, theoretical

approaches inspired by conventional fluid mechanics (e.g. Navier-Stokes) has been vain up to

date (especially due to the absence of rheological law) and more appropriate theories must be

established. So far, two theoretical approaches have been developed in literature. The first

one, based on “plasticity theory” involves the cohesion and flow properties of the bulk

material. These properties are relevant with regard of the flow of fine powders but the theory

fails for coarse granular materials for which the flow patterns have found to depend only on

the size of particles. Consequently, in which follow, only that concepts related to the flow of

coarse particles (d 0.5-1 mm) are considered.

17
Note that the majority of models established to predict the velocity profiles rely on the

equation of continuity. Considering the solids flow as incompressible, the continuity equation

in rectangular coordinates is:

(5)

Obviously, rectangular coordinates are especially suitable for silos with rectangular cross

sections. For other silo geometries, other coordinates, either cylindrical or spherical are more

appropriate. Table 2 (Bird et al., 1965) summarizes the general equations and their application

to some special cases that are of interest in general and for this work in particular.

In some special cases, due to the symmetry of silos, one of derivatives in these equations

could be put to zero (e.g. in rectangular coordinates – see also Table 2). However, it

is still necessary to express one of the remaining velocity components as a function of the

second one (i.e.: ). All the problem lies in the way this reliance is assumed.

5.1. Convergent channels


The first models for velocity profile description were established for steady gravity mass

flow of solids in narrow angled converging channels (i.e. conical and wedged-shape hoppers).

Note that these models are also applicable to the moving part of cylindrical hoppers. Indeed,

for such systems it is reasonably admitted that the flow is radial.

Brown and Richards (Brown and Richards, 1970) were the first to establish a model for

such a system. This model is based on the concept of the “minimum energy theorem”.

Assuming a radial flow, the radial velocity decays as a power law along the vertical axis with

exponent n=2 for conical and n=1 for wedge-shaped hoppers. The velocity components in

spherical coordinates are then given by the following expressions (Cleaver and Nedderman,

1993; Fullard and Davies, 2016):

18
(6-a)

and

(6-b)

The minus sign in Eq. (6-a) denotes that the direction of the flow is toward the origin

(apex). The function takes account of the variation of the velocity with without

involving the radial distance, r. Assuming, to a first order, that the surface of “minimum

energy” corresponds to a radial arch with r=r0 (where r0 is the distance between the edge

points of the aperture and the fictive apex of the silo, Fig. 3) and considering, in addition, that

the discharge flow rate is constant, Brown (Brown, 1961) derived a relationship for

(Table 3, Eq. T3.1). This model was not tested experimentally by authors themselves. Further

attempts by other investigators to fit their results with this model was not successful and

showed too flat profiles predicted by the model with respect to experimental data. The model

of Mroz and Szymanski (Mroz and Szymanski, 1971) also based on energy considerations

was not more successful neither. Furthermore, Nedderman (Nedderman, 1988) showed that

the velocity profile is very sensitive to the suggested form of the free-fall arch.

Other researchers have borrowed this reasoning but proposed a different relationship for

. Jenike (Jenike, 1962b) considered the effect of stress fields in steady gravity flow of

frictional-cohesive solids. He established a general expression of (Eq. T3.2) for which

he provided graphical solutions for a number of initial conditions. Mroz and Szymanski

(Mroz and Szymanski, 1971) also proposed their own relation (Eq. T3.3) using different

boundary conditions. Generalov (Generalov, 1985) proposed a relation taking into account the

hopper half-angle (Eq. T3.4). Edelman, Smith (Nedderman, 1988) and Polderman (Polderman

et al., 1987) provided another relationship based on a simpler analytical solution (Eq. T3.5).

19
However, Nedderman showed that unlike Jenike’s equation, the solution of Mroz and

Szymanski contradicts the principle of coaxiality.

Nedderman (Nedderman, 1988) reported a synthesis of several series of experiments

carried out in a conical hopper with several granular solids. He found that among the above-

mentioned models, the one proposed by Jenike provides a reliable prediction as a first

approximation but only for small values of .

However, Mroz and Szymanski (Mroz and Szymanski, 1971) recognised the importance of

"dilation" accompanying granular flow. They developed continuum models based on

derivatives of soil plasticity but with addition of the methodology that give rise to "limited

dilation"/"confined dilation" which meant that the principle of "co-axiality" of the principal

stress and strain-rate vectors during flow can no longer be admitted. Later works incorporated

rotation as well as sliding motion of particles through their stress-strain coupling in flow

(Tüzün and Nedderman, 1982).

Nevertheless, Nedderman et al. (Nedderman, 1988) bowed out of continuum modelling of

hopper flows and recognised that for all intensive purposes, the "principle of co-axiality" in

coarse granular flows could only be reconciled with a radial flow field first developed by

Jenike but only in conical and wedge-shaped flows with very small hopper half-angles. The

kinematic models such as that proposed by Nedderman and Tuzun (Nedderman and Tüzün,

1979), presented in the next section) remains the most reliable method to predict velocity

fields in funnel-flow hoppers and flat-bottom bins due the presence of stagnant zones and

considerable flow dilation accompanying flow just before the discharge outlet.

20
5.2. Vertical silos
Obviously, the hypothesis of radial flow and the related models are only valid for the

convergent hoppers without the vertical bin. For vertical silos, with or without a convergent

hopper, the flow is not radial and more expounded models are required.

One of the most relevant models for the flow of coarse free-flowing particles is

undoubtedly the so-called “kinematic model”. The basic idea behind this stochastic model is

that particles could move only by falling into the spaces vacated by the departing particles in

their neighbourhood. This idea originated from the original work of Litwiniszyn (Litwiniszyn,

1972) who established a stochastic model in which the silo is splitted into a number of

fictional cells and the probability for a particle to fall into a cell as soon as it becomes vacant

is calculated (Figure 9-a). Mullins (Mullins, 1972) also proposed a similar model based on a

counter flow of voids. Since the particles are assumed to flow freely, the only property of

importance in these models is the size of cavities, which is of the same order of magnitude as

the particle size.

Inspired from the stochastic model of Litwiniszyn (Litwiniszyn, 1972), Nedderman and

Tüzün (Nedderman and Tüzün, 1979) established a deterministic kinematic model which

suppose that in an assembly of particles, during their descent, the upper particles tend to move

sideways toward the faster falling particles in the lower layer (Fig. 9-b). Hence, the horizontal

component of the velocity, is a function of the velocity gradient through the direction;

i.e. . They then expressed this dependency as the simplest possible type:

(7)

Their analyses, led to the same system of partial differential equations established by

Litwiniszyn or Mullins:

(8)

21
which must be solved using appropriate boundary conditions. This equation is formally

identical to the classical well-known two-dimensional diffusion provided that t in heat flow

equation is substituted by z. Mullins (Mullins, 1972, 1974) provided a series of solutions for a

number of special cases corresponding to different geometries or boundary conditions (Table

4).

Based on the similarity solutions, Nedderman and Tüzün (Nedderman and Tüzün, 1979)

obtained the following expression for vertical and horizontal velocity profiles in a flat-bottom

2D hopper for a narrow (point) opening in a wide hopper:

(9-a)

and (from Eq. 8):

(9-b)

where is the velocity at the centre line (x=0). Hence, it turns out that the proportionality

constant of Eq. 8, , called also the kinematic constant or the diffusion length is the only

experimental constant of this model. This empirical constant has the dimension of length and

could be obtained by linearization of Eq. 9-a (Nedderman and Tüzün, 1979):

(10)

Thus, can be obtained by linear regression of the plot of against . From their

experimental data, Tüzün et al. found the value of the empirical constant to be about

which is in good agreement with the experiments of Mullins who found . Since

these pioneer works, several other researchers have reported different values of . For

example, using monodisperse glass beads, Samadani et al. (Samadani et al., 1999) reported a

value of . Dosekun (Dosekun, 1980) carried out experiments in a steep walled silo

and found that is sensitive to the hopper angle. For their part, Medina et al. (Medina et al.,

22
1998) studied the velocity field using the PIV technique and found that B increases from

approximately to as the height increases. The same trend was reported by Choi et

al. (Choi et al., 2005).

Finally, note that rigorously, the kinematic model of Nedderman and Tüzün is only

applicable to the converging flow zone. However, the model was used quite successfully by

these authors to determine stagnant zones. Because there is no discontinuity in model’s

equations, they defined the surface of sliding as the limit beyond which the particles velocity

is less than one particle diameter per second. Other investigators (Magalhães et al., 2016;

Ramirez-Gomez et al., 2012) defined a local Mass Flow Index (I MF) as the ratio of the local

velocity to the maximum velocity (at the hopper centreline). According to these works, I MF

values higher than 0.3 are indicative of mass flow, otherwise the funnel flow with presence of

stagnant zones occurs.

It should be noted, however, that the so-called kinematic model described in this section is

purely stochastic and does not aim at stress-strain coupling. Hence, the so-called "flow

dilation" effect accompanying flow is modelled through the scaling of the kinematic constant

which scales as multiple of the particle size and is also a function of the "effective" particle

shape. These single particle properties are ignored in continuum models but could be

accounted for implicitly through the invoking of a "fabric tensor" in soil mechanics.

Note also that Sielamowicz et al. (Sielamowicz et al., 2006; Sielamowicz et al., 2015;

Sielamowicz and Czech, 2010; Sielamowicz et al., 2010, 2011a, b) have focused very

thoroughly on the velocity measurements in both concentric and eccentric discharge

scenarios. They presented a statistical analysis (Sielamowicz et al., 2015) through linear and

non-linear regression methods of the analysis of velocity data sets. They assumed that a

Gaussian law describes conveniently the velocity distribution and calculated the constant

23
values by the linearization or by using the non-linear regression. Their results showed that

much better description is obtained by non-linear regression.

Other investigations used numerical methods (e.g. (Watson and Rotter, 1996) who used

FEM method) to solve the set of equations 7 and 8 and stablish the velocity profiles.

Finally, the work of R. Jackson at Princeton (Jackson, 1986) is of particular importance.

Jackson pointed out the importance of "flow rheology" in granular flow and found that hopper

flows are essentially "transitional" between contact frictional and collisional stress-strain

coupling contributions. "Dilation" accompanying flow is the result of transition from sliding

to rotational friction between particles and with increasing acceleration under gravity. In

particular, Jackson highlighted a phenomenon called the "Jackson's Paradox" which points out

that for the mean flow field of the coarse granular matter to remain essentially incompressible,

the dilation effects must be confined to narrow shear zones (also termed as rupture zones)

whose size and periodicity depends on particle properties and particle shape and size. The

regions surrounding the shear zones must be critically consolidated to compensate and allow

for gross-shear to take place. However, such physics could only be studied with numerical

models and simulations (e.g. DEM and associated techniques). Further works of Koenders et

al. (Koenders et al., 2001) also helped to support the further development of post-evaluative

models of DEM simulations.

6. Exit mass flow rate


Several researchers have investigated the exit mass flow rate from silos. The first works

published on this focus date back to the very beginning of the 20 th century. Although the main

concern at that time was the determination of stress distributions for civil and mechanical

engineering purposes, the flow rate data appeared as a side-effect of those studies (Fowler and

Glastonbury, 1959).

24
For a wise exploitation of the literature, it should be kept in mind that the mass flow rate of

granular solids could differ sensibly from flat-bottom silos to conical silos and that the latter

has not been considered necessarily as a special case of the former by different investigators.

6.1. Bibliographic correlations for discharge flow rate


A summary of the most relevant correlations reported in literature is presented in Table 5.

As early as 1901, Ketchum and his team carried out a series of experiments on the flow of

wheat from the exit orifice of a flat-bottom silo (Ketchum, 1911). Their results showed that

the flow was independent of the head and varied as the cube of the orifice diameter (Table 5;

Eq. T5.1). Although these authors did not provide any correlation for the exit flow rate, their

founding was doubly intriguing for those who hitherto considered wheat (like dry sand) as a

model material. Indeed, considering those bulk materials as a pseudo-continuous medium

with an apparent bulk density, b, according to Bernoulli equation, the exit flow rate was

expected to be directly proportional to the square root of the head and to the cross-sectional

area of the opening (which is in its turn proportional to the square of the orifice diameter) (Eq.

T5.2). A similar conclusion can be obtained also based on conventional mechanics for free-

falling particles (Shamlou, 1988).

If the independency of the flow rate with respect to the head had been already revealed by

earlier investigations (i.e., Hagen 1852 (Hagen, 1852), Roberts in 1882 (Roberts, 1882,

1884)), the cube power dependency was not. These authors explained their results by the fact

that wheat grains in the bin tend to form a dome, which supports the weight above. The actual

surface of rupture (free flow) is assumed to be the lower surface of the dome and the grains

are in free-falling state only once they dropped from the space below the dome.

Note that further works carried out by some other investigators led to power law

dependencies close to that proposed by Ketchum and his co-workers. Deming and Mehring

25
(Deming and Mehring, 1929) were the first to undertake investigations on the discharge flow

rate of conical silos covering a large variety of materials and aperture geometries. Their

results confirmed also that the head had no practical effect upon the masse flow rate. Based

on dimensional analysis, they proposed a general equation to express the mass discharge (Eq.

T5.3), which for small ratios of leads to an exponent of 5/2 for d.

Newton et al. (Newton et al., 1945) proposed a correlation with but their results

suggested also a slight dependency on the head above orifice (Eq. T5.4). Also, the works of

Rausch (Rausch, 1948) resulted to an exponent of 2.75 (Eq. T5.5), those of Shirai (Shirai,

1952) to 2.5 (Eq. T5.6) and those of Franklin and Johanson (Franklin and Johanson, 1955) to

2.93 for the orifice diameter (Eq. T5.7), respectively. Tanaka and Kawai (Tanaka and Kawai,

1956) distinguished two cases according to the ratio of (Eq. T5.8). For ratios between 4

and 10, they proposed an exponent of 3 whereas for higher values an exponent of 2.7 was

recommended. Further works of Fowler and Glastonbury (Fowler and Glastonbury, 1959)

confirmed the latter. Indeed, using dimensional analyses, these authors found that, neglecting

the effect of the particle size, the exponent of d is equal to 2.7. Their experimental results

confirmed this postulate and led to a simple correlation for the exit mass flow rate of grains

(Eq. T5.9). This is also in good agreement with the correlation proposed by Rose and Tanaka

(Rose and Tanaka, 1959) (Eq. T5.10).

Luk’yanov et al. (Luk'yanov et al., 1960) suggested a polynomial dependency of the flow

rate on the orifice radius and on the particle mean size (Eq. T5.11). However, for small ratios

of , this equation reduces to a power law with an exponent of 2.5. This is consistent with the

results of Brown and Richards (Brown, 1961; Brown and Richards, 1970) that indicated also

the same value for the exponent (Eq. T5.12). Note that a power law with an exponent of 2.5

corroborates the free-falling arch hypothesis (Eq. T5.2) if one assumes that the height of the

26
free fall (e.g. the dome suggested by Ketchum et al.) is of the same order of magnitude as the

size of the orifice (Shamlou, 1988). The value of 2.5 for the exponent was also confirmed by

Beverloo et al. (Beverloo et al., 1961) whose analyses resulted in a quite different formula

(Eq. T5.13). Indeed, these authors found that their results fit better using ( ) instead of

. This was explained by assuming that along the perimeter of the orifice a zone whose the

size is proportional to dp is useless for the flow ("empty annulus" hypothesis). Note however

that the presence of such an inaccessible area at the edge of the orifice had been suggested

much earlier by Hagen (Hagen, 1852). In 1952 Weighardt (Wieghardt, 1952) suggested a

relation of that type also. Beverloo et al. (Beverloo et al., 1961) found that the coefficients in

their correlation are constant enough to use an average value of k=1.5 and C=0.58 regardless

of the particle type. Harmens (Harmens, 1963) investigated the gravity flow of some granular

solids in both flat-bottom and wedge-shaped hoppers with circular and non-circular orifices.

This author proposed a model based on the velocity profile of particles along the radius of the

orifice. Considering that the “free-fall arch” is a hypothetical conical surface and neglecting

the “empty annulus” effect, he established a model that confirmed again the exponent 2.5 for

small ratios (Eq. T5.14). One of the main drawbacks of this model is however that it makes

use of the base angle of the hypothetical cone, which is not known a priori. Furthermore,

Shinohara et al. (Shinohara et al., 1968) established another correlation based on the

continuity of the porous medium in a mass flow silo (Eq. T5.15).

From the 70’s, the physical consistency of the heuristic hypotheses (i.e. "free-fall arch"

and "empty annulus") of previous models became a subject of controversy (e.g. (Janda et al.,

2012; Rubio-Largo et al., 2015)) and other approaches based on force or energy balance

(Brown and Richards, 1970; Carleton, 1972; Madrid et al., 2017), classical mechanics

(Davidson and Nedderman, 1973; Williams, 1977) or fluid mechanics (Staron et al., 2012)

27
began to emerge. In 1970, Brown and Richards (Brown and Richards, 1970) proposed a

model based on application of Bernoulli’s equation. Assuming that the total energy of solids

decreases progressively to reach its minimum value at the exit (minimum energy theorem),

they established a theoretical equation for both circular and slot openings (Eq. T5.16). This

model was however criticized by Nedderman et al. (Nedderman et al., 1982) because it does

not take into account the pressure term. Overestimations by a factor 2 was reported by these

authors. Carleton (Carleton, 1972) proposed a model for mass flow in conical silos based on a

force balance on a moving fine slice of solids. This balance allows estimating of the mean

velocity at the exit, which in turn allows the determination of the mass flowrate if the cross-

sectional flow area and the bulk density are known (Eq. T5.17). Davidson and Nedderman

solved the equations of motion in radial direction for frictionless hopper walls. Assuming that

the material is subjected to a radial force body, , (which is valid only for small values of

), their analyses led to two simple equations for the discharge mass flow rate (Eq. T5.18).

Williams (Williams, 1977) developed a model using a similar approach including the friction

at walls for mass flow discharge of free-flowing particles through conical hoppers. His

analyses resulted in a set of equations for the upper and the lower limits of the mass flow rate

but only the equation of the upper limit (Eq. T5.19) is proper for practical applications.

Applying the concept of the empty periphery around a slot, Myers and Sellers (Myers and

Sellers, 1977) found a similar correlation (Eq. T5.20) than Brown and Richards for

rectangular openings.

Another category of model concerns those based on equation of continuity. If the velocity

profile of solids is known, the exit mass flow rate can be obtained by integrating the velocity

over a given cross-sectional area of flow, generally that of the exit. The simplest case is the

radial flow for which the velocity profiles were presented in previous section. Using Eq. 6, the

exit masse flow rate in spherical coordinates is given by:


28
(11)

This equation is valid for any radial position but practically it is more relevant to apply it at

the exit where (see Fig. 3). Hence, if is known (for example by one of equations

given in Table 3) , the mass flow rate can be deduced by solving the previous equation.

Finally, it is worth to note that since the emergence of numerical tools, other investigations

based on application of fluid dynamics (CFD, FEM) or of mechanics of contacts (DEM) have

been reported. These approaches have the advantage of giving a detailed description of the

flow in the silo. With regard to the discharge flowrate, we can mention for example recent

works of Staron et al. (Staron et al., 2012) who using the Navier–Stokes equation applied to

plastic fluids with an effective friction law, found an equation closed to that of Beverloo.

Also, Madrid et al. (Madrid et al., 2017) using the energy balance, coupled with a constitutive

equation for the rheology of the granular solids established a simple differential equation for

the flow rate. Their analyses resulted in a very similar equation than Beverloo and led to a

proportionality constant which is very close to the experimental fitted value

of 0.58 proposed by him.

6.2. Effect of the exit geometry


A quick review of bibliographic correlations gathered in Table 5 shows that the most

widely investigated geometry of the opening is the circular shape. However, some studies on

other geometries have been carried out as well. Fowler and Glastonbury (Fowler and

Glastonbury, 1959) were the first who tested different aperture shapes (circular, triangular,

rectangular, hexagonal, pentagonal, ellipsoidal). They concluded that, independently of the

geometry of the opening, the mass flow rate was proportional to the product of the hydraulic

diameter ( ), raised to a power, and the cross-sectional area of the aperture, (Eq. T5.9).

29
Later, results of Myers and Sellers (Myers and Sellers, 1977) who used rectangular openings

showed a good agreement with this conclusion. More generally, Nedderman et al.

(Nedderman et al., 1982) stated that for non-circular orifices, it would seem more convenient

to use a generic form of Beverloo equation including the effective area and the hydraulic

mean diameter of the space remaining after an empty zone of width has been

removed (Eq. T5.21):

(12)

where for Beverloo correlation. This equation took rise from a combination of the

original idea of Fowler and Glastonbury (Fowler and Glastonbury, 1959) and that of Beverloo

et al. (Beverloo et al., 1961).

Furthermore, Al-Din and Gunn mentioned that this approach could lead to a physical

inconsistency for long slots because it suggest that the flow rate becomes independent of slot

length. They then proposed a correlation to overcome this inconsistency (Eq. T5.22).

6.3. Effect of the hopper half-angle


In case of conical hoppers with steep angles, particles slide or roll over the hopper wall and

this will affect the discharge flow rate. So far, no general rule has been established yet for the

form of this dependency. One of the first investigations taking into account the effect of the

hopper angle is that of Deming and Mehring (Deming and Mehring, 1929) (Eq. T5.3).

However, Franklin and Johanson (Franklin and Johanson, 1955) found that the correlation

proposed by these authors is only applicable for half-angles smaller than ( ). They

proposed then an alternative correlation including a correction factor taking into account the

effect of the hopper angle, , and of the angle of internal friction, (Eq. T5.7).

30
One of the most significant works on this subject has been carried out by Rose and Tanaka

(Rose and Tanaka, 1959) who proposed an alternative correlation (Eq. T5.10). These authors

introduced a multiplying correction factor, F, to take account of the effect of hopper half-

angle. They distinguished two cases:

- For large hopper half-angles the flow proceeds in funnel mode and a stagnant zone

appears. If the angle of the stagnant boundary zone with respect to the horizontal is

designated by , the funnel flow occurs when . In this regime, the wall

inclination has a very small effect on the flow rate and the correction factor is equal to

unity:

- In case of mass flow ( ), the correction factor proposed by Rose and Tanaka is

given by

6.4. Effect of other parameters


According to the literature, the following parameters could influence in a lesser extent the

flow:

- the particles size and shape. In Beverloo correlation this effect is taken into account by

k which is about 1.5 for spherical particles and somewhat larger for irregular shaped

particles. Indeed, the main problem for angular particles is the proper choice of the

characteristic dimension of particles. From 1940’s to 1960’s a series of correlations of

the general dimensionless form:

(13)

was established introducing the ratio of to take into account the effect of the particle

size (e.g. Rausch (Rausch, 1948), Eq. T5.5; Shirai (Shirai, 1952), Eq. T5.6); Tanaka

31
and Kawai (Tanaka and Kawai, 1956), Eq. T5.8; Fowler and Galstonbury (Fowler and

Glastonbury, 1959), Eq. T5.9, Shinohara et al. (Shinohara et al., 1968), Eq. T5.15).

Besides, some other correlations (Rose and Tanaka (Rose and Tanaka, 1959), Eq.

T5.10; Luk’yanov et al. (Luk'yanov et al., 1960), Eq. T5.11; Beverloo et al. (Beverloo

et al., 1961), Eq. T5.13;Harmens (Harmens, 1963), Eq. T5.14 as well as Al-Din and

Gunn (Al-Din and Gunn, 1984), Eq. T5.22) take account of the particle size using more

complex functions for . However, for all these correlations, for small ratios of , the

effect of the particle size becomes negligible.

- the angle of friction or cohesion. For free-flowing granular materials, the cohesion is

low and the angle of friction does not play an important role. Some investigators

however took into account this parameter. For example, the correction factor in

correlation of Rose and Tanaka (Rose and Tanaka, 1959) includes the powder

cohesion, c. However, according to these authors and many others (e.g. Deming and

Mehring, (Deming and Mehring, 1929); Franklin and Johanson, 1955 (Franklin and

Johanson, 1955); Harmens (Harmens, 1963); Kotchanova, (Kotchanova, 1970)) this

effect could be reasonably neglected for this kind of particles.

- silo diameter. This is generally admitted that the silo diameter does not affect the mass

flow rate as long as it remains large with respect to the orifice diameter and the particle

size. According to Ketchum (Ketchum, 1911) and Brown and Richards (Brown, 1961)

this limit is while for Franklin and Johanson (Franklin and Johanson, 1955)

the criteria is given by . According to Harmens (Harmens, 1963) the

discharge rate is not affected by silo diameter provided that .

32
6.5. Concluding remarks on the discharge flow rate
Modeling of the discharge flow rate has involved a wide variety of approaches, ranging

from the most empirical to the advanced numerical methods (DEM, FEM, CFD), including

dimensional analysis, fundamental laws of mechanics, energy and momentum conservation

equations as well as the classical mechanics of fluids.

Among all existing models, the one proposed by Beverloo et al. is so far the most popular

and the most widely used. This is for the most part due to its ease of use and its robustness as

well as its quite reliable physical basis. However, this model has two main disadvantages:

- it requires the use of the apparent bulk density whereas this parameter could vary

significantly with time and space. Actually, Beverloo et al. considered the initial

density (the density achieved after filling) but the representative bulk density with

regard of the application seems to be the density at the exit, which cannot be measured

properly. For example, Huntington and Rooney (Huntington and Rooney, 1971)

recommended the use of a “flowing density” defined as the ratio of the mass flow rate

to the volumetric flow rate calculated from the descent velocity of the top surface of

the bulk. Obviously, it is not practical (neither always possible) to determine this

density.

- due to variation ranges of C (about 20% as 0.55<C<0.65) and those of the bulk density

which can achieve upto 20-25%, the accuracy of this correlation (like all others) for

predictive uses is not good enough (about ±25%). However, very accurate results are

generally obtained by fitting the experimental data using C and k as adjustable

parameters. In that context, the problem of the proper choice of the bulk density

becomes also less important as its effect is corrected by fitted value of C factor.

33
7. Residence Time Distribution
Despite its importance, very few studies, either experimental or theoretical, have been

conducted on the Residence Time Distribution (RTD) of granular solids in silos. However, the

RTD is an information of the utmost importance by itself. In addition, RTD experiments are

among the rare methods that can be used to study the flow pattern of solids in 3D silos.

Generally, the residence time is defined as the time a particle takes to move from its initial

position to the outlet.

In this field, Polish Academy of Sciences- IPPT-PAN have made many and varied

contributions over the years as well as the pioneer works of Litwiniszyn (Litwiniszyn, 1972),

Mroz and Szymanski (Mroz and Szymanski, 1971). Later, Sielamowicz et al. (Sielamowicz et

al., 2005; Sielamowicz et al., 2006; Sielamowicz et al., 2015; Sielamowicz et al., 2010,

2011a, b) contributed significantly to the study of RTD in both centric and eccentric silos.

Cleaver and Nedderman (Cleaver and Nedderman, 1993) established an expression for the

residence time in mass flow conical hoppers. Considering a radial flow, the velocity profile is

given by Eq. 6. The residence time, T, of a particle with an original position is given

by:

(14)

where r0 refers to the radius of the orifice. According to this equation, if the hypothesis of

radial and incompressible flow is valid, a plot of versus time must be linear and the

function can be deduced from the slope of this line.

The experimental work of Cleaver and Nedderman showed a good agreement between the

predictions of this model and the experimental data obtained for coarse free-flowing particles

(Kale seeds and Polypropylene granules) inside conical silos operating in steady-state mode.

34
Recently, in an exclusively theoretical work, Fullard and Davies (Fullard and Davies,

2016) applied this model to the case of wedge-shaped mass flow silos with both flat and

heaped superposed layers. Assuming a radial flow within converging part and a plug flow

within the cylindrical upper part, they showed that there exists a specific hopper geometry that

minimises the spread of the RTD for a particular heap angle.

Mullins (Mullins, 1972) was the first to establish analytical solutions for the exit time of

particles inside vertical silos in steady-state flow. His analyses were based on the velocity

profiles as described by the stochastic kinematic model (Table 4). Mullin obtained the

following relationships for steady flow of particles in a semi-infinite bed:

(15-a)

for a point orifice and:

(15-b)

for a finite orifice.

The case of batch operating silos is more complicated. Able et al. (Able et al., 1996)

measured experimentally the exit time distribution during the batch discharge of a flat

bottomed cylindrical bunker. They established a model in which the velocity field is presented

by the kinematic model of Nedderman et al. (which is parabolic rather than radial). Because

both a flow zone and a stagnant zone exist together within the silo, the model considers that

the discharge of particles takes place in two stages i.e. a first stage during which the particles

could be stagnant followed by a second stage when they move toward the exit. For a particle

at (Z,r) position, the latency time passed in the stagnant zone is given by :

(16)

35
where corresponds to the difference between the density of the stagnant zone, , and that

of the moving zone, . After this stationary step, the particle will join the parabolic

streamline supposed by the kinetic model. The particle will reach the exit at a time:

(17)

Hence, the total residence time for a particle which is initially at (Z,r) position is:

(18)

The experimental results of Able et al. showed a good agreement with predictions of the

model during the early stages of discharge. This agreement fades over time and significant

discrepancies were observed at the end of the operation. The particles exit sooner than

predicted by the model. This is attributed to the emergence of a fast-moving central core at the

centreline, which displaces the particles at large distances much higher than the diffusion

scale, B. The problem of modelling of this last phase remains unsettled.

Also, the case of axisymmetric flow in cylindrical flat-bottom silos deserves attention.

Analytical solutions for this case are only available for radial flow. For parabolic profiles as

those expected by the kinematic model, despite the velocity profiles are known, the use of this

profile to calculate the residence time of particles is not straightforward. Indeed, using

cylindrical coordinates, the residence time in this case is given by:

(19)

However, depends on also and integrating this equation needs an additional equation.

Zhang (Zhang, 1997) proposed the following set of equations:

(20)

36
which can be solved by numerical methods with corresponding initial conditions (i.e. at

, and . Note that and are given by equations of Table 4.

Zhang (Zhang, 1997) used these equations to develop an engineering model to describe the

residence time distribution of full scale silos. He divided the silo into four distinct zones with

different flow patterns:

- A top flow zone with a finite thickness where all particles moves toward the centre

along the angle of repose,

- A central flow zone of relatively fast velocity toward the outlet

- A feeding zone surrounding the central flow zone

- A stagnant zone

The predictions of the model were in good agreement with experimental data for the first

stages of the operation but significant deviations appeared for the last stages of the discharge.

In conclusion, the RTD in silos of different geometries has not yet been elucidated and

constitutes a large field of research for future works.

8. Segregation
Segregation is one of the main concerns with heterogeneous populations of particles (i.e.

different size, shape, density or surface properties). It has long been known that when a

mixture of granular materials of different properties are handled, a segregation phenomenon

may occur even if the mixture is originally homogeneous (Brown, 1939; Engblom et al.,

2012a; Fan et al., 2017). Segregation can take place during both filling and discharging of

silos. For instance, when a mixture of particles is poured onto the top of a heap or a silo, the

larger (or denser) particles with greater inertia will have a greater tendency to roll over the

slope at the top of the heap and to be in the periphery (Brown, 1939; Carson et al., 1986;

Cizeau et al., 1999; Goyal and Tomassone, 2006; Hogg, 2009; Mosby et al., 1996; Shinohara

37
et al., 1972; Standish, 1985; Williams, 1963). Conversely, fine particles will be more easily

stopped by the heap and will be located in the centre of the silo. This spontaneous segregation

leads to a distribution in strata of the two populations of fines and large particles referred

sometimes as Christmas tree (Combarros Garcia et al., 2016). Other mechanisms exist

however resulting in different types of segregation.

A summary of investigations on the segregation during filling and emptying of silos is

provided in Table 6. The earliest attempts to describe the phenomenon were purely qualitative

and involved heterogeneous segregation in heaps and vibrated beds. One of the first (if not the

first) papers published on this topic is due to Brown (Brown, 1939) who discussed the types

of segregation encountered during the handling of coal, especially when it is vibrated or

poured into heaps. Brown distinguished two types of size segregation: bottom-top segregation

due to vibration and, stratification due to pouring. Donald and Roseman (Donald and

Roseman, 1962) brought experimental evidence of segregation in a horizontal mixer drum.

Later, Williams (Williams, 1963; Williams, 1965) gave some explanations on the effect of the

difference in particle size on the segregation due to vibration and pouring of particulate solids.

Since then, a series of experimental and modelling investigations has been carried out and

reported. In 1976, Williams provided a review on the segregation of particulate materials

(Williams, 1976). Also, in 2002, Prof. Shinohara in Japan brought together international

works on heap and hopper flow segregation which gave place to a special issue of the Chem.

Eng. Sci, "Solids Flow Mechanisms and Their Applications" (Shinohara and Tuzun, 2002).

Lately, Gray et al. (Gray et al., 2015) reviewed advances on the modelling of the basic

segregation processes in a single avalanche. Very recently, Fan et al. (Fan et al., 2017)

provided a general review of previous works carried out on this subject and presented the state

of knowledge on physical mechanisms behind segregation. Both theoretical and experimental

aspects were covered in their works.

38
Consequently, in this section, we give only a brief reminder of the basic mechanisms of

segregation and limit our survey to those works that specially deal with the experimental

study of the segregation of free-flowing granular materials in silos.

8.1. Segregation mechanisms


Several investigators have discussed the basic mechanisms responsible for segregation

(Brown, 1939; Carson et al., 1986; Cizeau et al., 1999; Donald and Roseman, 1962; Dyrøy et

al., 2001; Engblom et al., 2012c; Fan et al., 2017; Hogg, 2009; Ketterhagen et al., 2008;

Mosby et al., 1996; Rhodes, 1998; Salter et al., 2000; Stavans, 1998; Tang and Puri, 2004;

Williams, 1963; Williams, 1965). The first classifications of segregation were somewhat

confused and imprecise because there was an amalgam between the elementary mechanisms

of segregation, their causes and their consequences. For example, Tang and Puri (Tang and

Puri, 2004) listed 13 patterns of segregation reported in preceding works, namely: trajectory,

rolling, displacement, percolation, sieving, air current, fluidization, agglomeration,

concentration-driven displacement, push-away, impact/bouncing, embedding and angle of

repose. Furthermore, they highlighted that some of these mechanisms do not take place

commonly and in addition, some of them cannot be considered as a mechanism strictly

speaking. Based on a more detailed review, these authors identified four fundamental

segregation mechanisms that are trajectory, sieving, fluidization, and agglomeration. In fact, a

more thorough analysis of the literature shows that a consistent classification of segregation

must be necessarily multi-criterion.

In order to properly identify the relevant criteria, it is important to have a closer look on the

segregation process. Obviously, for segregation to occur two conditions must be met:

- A relative motion between particles. In this respect, the segregation process can be

classified by the type of motion (rotating, tumbling/avalanching/sliding,

39
suspending/free-fall, aerated and fluidized), or by the nature of the energy input

(e.g. work) causing the motion, which can be mechanical (vibration, rotation),

potential (free fall during charging and discharging), kinetic/inertia or pneumatic.

- A difference in properties. Basically, any factor that could make some particles to

move differently from the others or through a preferential path could cause

segregation. In this regard, the main properties are size, density, shape, resilience

and surface charges or forces (i.e. electrostatic, magnetic, van der Waals and

capillary).

Furthermore, the combination of these two conditions could give rise to different

segregation mechanisms and segregation paths. With regards of the silos, the main

mechanisms occurring during filling or emptying are presented below.

8.1.1. Segregation due to vibration

Some industrial silos are equipped with a vibrating device to initiate or improve the flow.

However, the vibration can lead to the separation of the particles. In case of vertical vibration,

fine particles can move downward through the interstitial space between larger particles. In

addition, at each jump, the space beneath the coarser particles is filled by fine particles and

the coarse particles do not return to the same level. This mechanism known as sieving/sifting

or percolation leads to a top-to-bottom segregation (Tang and Puri, 2004). Another

mechanism mentioned by several authors (Carson et al., 1986; Dyrøy et al., 2001; Hogg,

2009; Mosby et al., 1996; Salter et al., 2000) is the push away effect during the vibration

which takes place when larger and/or dense particles fall down and hit finer particles and push

them away. This mechanism lead to a side-to-side segregation.

8.1.2. Segregation due to aeration and fluidization

40
To improve the flow or avoid clogging of the product, some silos are aerated or even

fluidized. The airflow could then, in some cases, push the finer and/or less dense particles

towards the surface. This kind of process results in top-to bottom segregation.

8.1.3. Segregation by elutriation

This mechanism is similar to aeration except that the airflow is created when filling the silo

with the powder itself. When the silo is loaded, the particles displace an amount of air

equivalent to their own volume creating a more or less turbulent airflow. This air flow can

suspend the finest (and/or less dense) particles which will settle down later on the sides of the

heap. A solid flow rate or a low filling height can remedy this problem.

8.1.4. Segregation due to trajectories

This phenomenon can be encountered when the silo is loaded by a belt or pneumatic

conveyor for example (Carson et al., 1986). In this case, the particles are projected rather

horizontally within the silo. The distance traveled by the particles increases with the square of

their diameter (Stokes law). Particles of dissimilar size will have a different path because the

larger particles go farther than the fines. The same reasoning is also valid for a vertical (free

fall) loading of the particles where coarse particles accelerate more than fines.

8.2. Influencing parameters affecting the segregation inside silos


According to Mosby et al. (Mosby et al., 1996) the predominant mechanism of segregation

depends on two categories of parameters:

- Particle properties such as particle size distribution, density, shape, resilience, friction

coefficient, surface texture, cohesivity and adhesion

41
- Process variables such as fall height, rate of feed, size of heap, moisture and mixing

ratio.

Among all these parameters, particle size distribution, density and density distribution as

well as shape and shape distribution are the most important factors.

8.2.1. Influence of particle size distribution and size ratio


Undoubtedly, the particle size is the most important parameter causing the segregation. Its

influence is much more pronounced than that of density, which is considered to be the other

key factor in segregation. This can be explained by a greater influence of the size on mobility

of particles on the one hand, and by a faster increase in particle inertia with their size, on the

other hand. For example, increasing the size ratio by a factor 2 gives rise to an increase in

particle weight by a factor 23. In addition, the particle size may vary over a much wider range

than the density. Particle size variations of several orders of magnitude (3-4) are quite

common whereas the density ratio could rarely exceed one order of magnitude. It is therefore

normal that the majority of the work of the literature is devoted to the study of this parameter.

Several authors have studied the influence of the particle size on the segregation in silos

(Artega and Tüzün, 1990; Bagster, 1996; Chu, 2008; Cizeau et al., 1999; D’Arco et al., 2006;

Engblom et al., 2012a, b, c; Engblom et al., 2012d; Goyal and Tomassone, 2006; Karolyi et

al., 1999; Ketterhagen et al., 2006; Ketterhagen et al., 2008; Ketterhagen et al., 2005;

Ketterhagen et al., 2007; Salter et al., 2000; Samadani et al., 1999; Seil et al., 2012; Shinohara

and Golman, 2002; Shinohara et al., 2001; Shinohara et al., 1970, 1972; Standish, 1985; Wu

et al., 2013). Generally, it appears that the segregation occurs mainly in case of free-flowing

powders. There is, however, no consensus or general rule about the size limit or size ratio

above which segregation occurs. For example, it is generally admitted that the size

segregation happens for free-floing particles having a mean particle size larger than 100µm.

42
However, the segregation by agglomeration mentioned by Tang and Puri (Tang and Puri,

2004) as one of the main mechanisms takes place in case of very fine particles. Shinohara et

al. (Shinohara et al., 1970, 1972) studied the mechanisms of segregation during filling and

discharging of particles. These authors analysed the process on the basis of screen (or

hypothetical)-hopper model where small particles pass through the space between large

particles. They then developed a model to estimate the critical size ratio beyond which the

segregation could take place. Samadani et al. (Samadani et al., 1999) conducted discharge

experiments with various size ratios and reported segregation occurrence at even small size

ratios (1.2). Nevertheless, they observed that the extent of segregation increased with the

size ratio. Salter et al. (Salter et al., 2000) investigated the segregation of binary mixtures

during the filling of a 2D hopper. These authors has noticed that there is an increase in the

segregation patterns (i.e. composition at outlet versus time) as the particle size ratio increases

from 1.4:1 to 3.1:1. These results, however, seem to contradict those of Shinohara et al.

(Shinohara et al., 2001) that investigated the segregation of multi-sized mixtures during filling

and found that the segregation diminishes with the increase of the size ratio of the smallest

particles. For their part, Goyal and Tomassone (Goyal and Tomassone, 2006) using a two-

dimensional silo found that the transition from complete segregation to partial segregation

during pouring of bulk solids appears as the size ratio of the species falls below a critical

value of about 1.4. Ketterhagen et al. (Ketterhagen et al., 2007) studied the effect of the size

ratio during discharge of wedge-shaped hoppers using DEM simulations and found that the

extent of segregation significantly increases for size ratios greater than 1.9. Engblom et al.

(Engblom et al., 2012d) reported an accentuation of segregation with the particle size ratio

which is in good agreement with numerical study (DEM simulation) of Seil et al. (Engblom et

al., 2012d).

43
8.2.2. Influence of particle density
Despite of its importance, so far the effect of the particle density has not been much

studied. Only a few works have dealt with this subject (Engblom et al., 2012a; Seil et al.,

2012; Shinohara and Golman, 2003). Shinohara et al. (Shinohara and Golman, 2003) brought

experimental evidence of density segregation during emptying after filling of silos. Engblom

et al. (Engblom et al., 2012d) found an enhancement of segregation with increasing particle

density ratio (heavy/light). Accordingly, Seil et al. (Seil et al., 2012) using DEM simulations

found that the extent of segregation varies almost linearly with the particle density ratio.

8.2.3. Influence of the bulk composition (mass fraction of fines)

The composition of the bulk material has a considerable influence on segregation. The

effect of this parameter has been essentially investigated for binary mixtures of particles

(Artega and Tüzün, 1990; Bagster, 1996; Ketterhagen et al., 2007; Salter et al., 2000;

Samadani et al., 1999; Shinohara and Golman, 2003; Shinohara et al., 1968; Standish, 1985;

Williams, 1963; Wu et al., 2013) and only a few studies have been devoted to

multicomponent mixtures (Engblom et al., 2012a, b, c; Engblom et al., 2012d; Shinohara and

Golman, 2002; Shinohara and Golman, 2003; Shinohara et al., 2001). In general, these works

leads to the conclusion that the segregation extent diminishes with increasing the mass

fraction of fine particles.

8.2.4. Other influencing parameters

Besides the size ratio, the density ratio and the bulk composition, other important factors in

regard of segregation which have been investigated and reported in the literature are:

- solid feed rate (Artega and Tüzün, 1990; Salter et al., 2000; Samadani et al., 1999;

Shinohara and Golman, 2002; Shinohara et al., 1972; Williams, 1963; Zigan et al.,

44
2008). Later works by Smith and Tuzun (Smith and Tüzün, 2002a, b) attempted to

establish the effects of feed rate and particle properties in binary granular heap flows.

- free-fall distance (Engblom et al., 2012a, c; Ketterhagen et al., 2005; Shinohara et al.,

1972; Standish and Jones, 1984)

- silo geometry (Engblom et al., 2012b; Engblom et al., 2012d; Ketterhagen et al., 2005;

Ketterhagen et al., 2007; Shinohara and Golman, 2003; Wu et al., 2013)

- number of components (Shinohara and Golman, 2002; Shinohara et al., 2001)

- friction properties (Károlyi et al., 1998; Ketterhagen et al., 2008), moisture (Anand et

al., 2010; Bagster, 1996), particles elasticity (Windows-Yule and Parker, 2015) and

shape (Tao et al., 2013; Yu and Saxén, 2014)

Engblom et al. (Engblom et al., 2012a) studied the effects of the free fall distance,

intermittent filling/discharge and the discharge rate on segregation at filling using a 0.4 m3

silo and two ternary mixtures. They concluded that side-to-side segregation with

accumulation of fine particles to the silo walls increases with the free fall distance.

Segregation was aggravated in case of intermittent filling/discharge. They attributed this

observation to the change of the shape of the powder bed’s surface when a portion of the silo

contents is withdrawn.

Engblom et al. (Engblom et al., 2012d) studied the segregation of powder mixtures in silos of

different scale using binary and ternary mixtures. Based on dimensional analyses they

proposed empirical correlations (using four dimensionless group) for describing the

concentration of fine particles at the walls as a result of silo filling and segregation towards

the end of complete discharge.

Shinohara and Golman (Shinohara and Golman, 2002; Shinohara et al., 2001) proposed

segregation indices to describe the intensity of segregation of multi-sized particle mixtures

during filling a two-dimensional silo. They reported that the segregation increases with a

45
lower feed rate and lower concentration of the smallest component, and exhibit a maximum

with increasing numbers of components.

Ketterhagen et al. (Ketterhagen et al., 2007) investigated the segregation of granular

materials during discharge from a hopper by DEM simulations and experimental runs using a

small silo allowing a full scale simulation. These authors reported that the hopper cross-

sectional shape and hopper angle as well as the filling method and characteristics of the

granular material are the important parameters affecting the segregation during discharge.

Garcia et al. (Combarros Garcia et al., 2016) also studied the underlying mechanisms of

segregation in heaps and silos using experiments, DEM simulations and a continuum-based

model. Their experiments were carried out with a bimodal mixture of differently sized

particles discharging from a pilot scale silo with a belt conveyor at the outlet. The results

showed that in case of mass flow, a remix at the silo outlet occurs and the segregation is not

significant. In case of funnel flow, the segregation exists and the fines are more abundant in

the beginning of discharge.

9. Conclusion
This paper tends to provide a large and comprehensive review of gravity flow of granular

solids. The points of interest are the flow pattern, the stress distribution, the velocity profile,

the exit flow rate and the Residence Time Distribution (RTD) of particles inside silos.

For all these aspects, existing reliable models were presented and discussed in terms of

their main hypotheses and their reliability. For simple silo geometries, the existing models

provide a good estimation of the parameters of interest except the RTD and the extent of

segregation. It should be noted that despite the considerable amount of information available

in the literature, these two phenomena are still poorly understood. Although the main

mechanisms are now known qualitatively, the quantitative prediction of this phenomenon has

46
not yet entered in the field of engineering and presents a big challenge. In 1982 in the

conclusion of his paper, Standish specified "Fundamental study of size segregation

phenomena is still in its infancy". We must recognize that this is still the case for the

segregation and more generally, for flow of granular materials in silos; the infant does not

grow up fast enough.

One of the questions that needs to be answered is scaling (down and up) of silos. This is

important in two respects. First, the current computational capacities do not make it possible

to treat large size silos. Their simulation by numerical techniques therefore requires reducing

the number of particles to be treated. On the other hand, the experimental validation of the

results is also heavy and can hardly be done on industrial silos. It is therefore necessary to

reduce the size of the silos. This can be achieved by two means, a reduction of the dimensions

of the 3D silos by respecting the homothetic rules. However, this method does not allow

respecting the ratio between the size of the particles and the dimensions of the silo and must

be avoided. Another solution is to work with a cut of the silo (2D silo) respecting the diameter

and height of the silo. This method seems more relevant but requires studying the rules of

similarity between the two configurations. This is one of the main lines of research currently

being developed by the authors.

Another problem of interest (currently investigated by authors) is the coupling of the flow

and multiphisic phenomena that may occur during storage (e.g. material transfer, heat

transfer, chemical or biological reaction, etc.). This should allow a better control of certain

risks related to grain storage, in particular self-heating and fire.

Besides the RTD and segregation, other challenges in this field concern the flow of fine

(cohesive) powders and the silos of complex geometries for which the numerical methods

could be very useful or even indispensable.

47
Nomenclature
Cross-sectional area of the silo
Cross-sectional area of the opening
Effective area of the opening (once the width of empty zone is removed)
Width of rectangular opening
Constant in Eq. 7
Beverloo’s constant
Diameter of circular opening
Particle mean size
Silo diameter
Hydraulic diameter
Effective hydraulic diameter (once the width of empty zone is removed)
g Gravitational acceleration
Height (head) of granular solids in silo
see table 1
Height of effective transition point
Height of the silo
Beverloo’s constant
Lateral stress ratio
Length of a rectangular slot
Length of silo for planar silos
Mass flow rate
M Mass
n (in Eq. 15-a) particle density (number of particles per unit volume)
Q Volumetric flow rate
r (coarse to fine) size ratio
R Radius
t Time
T Residence Time
Tp Residence Time for a point orifice
Tf Residence Time for a finite orifice
Perimeter of silo
velocity

Greek symbols
Hopper half-angle
Angle of approach
Angle of repose
(or φ) Angle of internal friction
Angle of repose
(or K) Lateral stress ratio
Coefficient of sliding friction between wall and bulk
Bulk density
Lateral (horizontal) stress
Vertical stress (pressure)
Wall stress
Latency time in stagnant zone

48
Angle of wal friction
Angle between the wall and the major principal plane or shape factor in Table 5

References
Able, R.M.G., Othen, S.M., Nedderman, R.M., 1996. The exit time distribution during the
batch discharge of a cylindrical bunker. Chemical Engineering Science 51, 4605-4610.
Airy, W., 1897. The pressure of grain. Proc. Institution of Civil Engineers 131.
Al-Din, N., Gunn, D.J., 1984. The flow of non-cohesive solids through orifices. Chemical
Engineering Science 39, 121-127.
Anand, A., Curtis, J.S., Wassgren, C.R., Hancock, B.C., Ketterhagen, W.R., 2010.
Segregation of cohesive granular materials during discharge from a rectangular hopper.
Granular Matter 12, 193-200.
Artega, P., Tüzün, U., 1990. Flow of binary mixtures of equal-density granules in hoppers-
size segregation, flowing density and discharge rates. Chemical Engineering Science 45, 205-
223.
Bagster, D., 1996. Studies on the Effect of Moisture Content and Coarse and Fine Particle
Concentration on Segregation in Bins. KONA Powder and Particle Journal 14, 138-143.
Balevičius, R., Kačianauskas, R., Mróz, Z., Sielamowicz, I., 2007. Microscopic and
macroscopic analysis of granular material behaviour in 3d flat-bottomed hopper by the
discrete element method, Arch. Mech., Warszawa pp. 231-257.
Balevičius, R., Kačianauskas, R., Mróz, Z., Sielamowicz, I., 2008. Discrete-particle
investigation of friction effect in filling and unsteady/steady discharge in three-dimensional
wedge-shaped hopper. Powder Technology 187, 159-174.
Beverloo, W.A., Leniger, H.A., van de Velde, J., 1961. The flow of granular solids through
orifices. Chemical Engineering Science 15, 260-269.
Bird, R.B., Stewart, W.E., Lightfoot, E.N., 1965. Transport Phenomena. John Wiley and
Sons, Inc, New York.
Brown, C.J., Nielsen, J., 1998. Silos: fundamentals of theory, behaviour and design. CRC
Press.
Brown, R.L., 1939. The fundamental principles of segregation. J. Inst. Fuel 13, 15.
Brown, R.L., 1961. Minimum energy theorem for flow of dry granules through apertures.
Nature.
Brown, R.L., Hawksley, P.G.W., 1947. The internal flow of granular masses. 26, 159.
Brown, R.L., Richards, J.C., 1970. Principles of powder mechanics-Essays on the packing and
flow of powders and bulk solids. Pergamon Press, Hungary.
Caquot, A., Kerisel, J., 1956. Traité Mecanique des Sols. Imprimeur-Librairie, Paris.
Carleton, A.J., 1972. The effect of fluid-drag forces on the discharge of free-flowing solids
from hoppers. Powder Technology 6, 91-96.
Carson, J.W., Royal, T.A., Goodwill, D.J., 1986. Understanding and Eliminating Particle
Segregation Problems. Bulk solids handling 6, 139-144.
Chatlyne, C.J., Resnick, W., 1973. Determination of flow patterns for unsteady-state flow of
granular materials. Powder Technology 8, 1707.
Chen, J.F., Rotter, J.M., Ooi, J.Y., Zhong, Z., 2005. Flow pattern measurement in a full scale
silo containing iron ore. Chemical Engineering Science 60, 3029-3041.
Chen, J.F., Rotter, J.M., Ooi, J.Y., Zhong, Z., 2007. Correlation between the flow pattern and
wall pressures in a full scale experimental silo. Engineering Structures 29, 2308-2320.
Choi, J., Kudrolli, A., Bazant, M.Z., 2005. Velocity profile of granular flows inside silos and
hoppers. Journal of Physics: Condensed Matter 17, S2533-S2548.

49
Chou, C.S., Hsu, J.Y., Lau, Y.D., 2002. The granular flow in a two-dimensional flat-bottomed
hopper with eccentric discharge. Physica A: Statistical Mechanics and its Applications 308,
46-58.
Chu, B.-H., 2008. Experimental investigation of particle segregation in hopper discharge
University of florida.
Cizeau, P., Makse, H.A., Stanley, H.E., 1999. Mechanisms of granular spontaneous
stratification and segregation in two-dimensional silos. Physical Review E - Statistical
Physics, Plasmas, Fluids, and Related Interdisciplinary Topics 59, 4408-4421.
Cleary, P.W., 1999. The effect of particle shape on hopper discharge, Second International
Conference on CFD in the Minerals and Process Industries, Melbourne, Australia, pp. 71-76.
Cleaver, J.A.S., Nedderman, R.M., 1993. Measurement of velocity profiles in conical
hoppers. Chemical Engineering Science 48, 3703-3712.
Combarros Garcia, M., Feise, H.J., Strege, S., Kwade, A., 2016. Segregation in heaps and
silos: Comparison between experiment, simulation and continuum model. Powder
Technology 293, 26-36.
Conesa, C., Saleh, K., Thomas, A., Guigon, P., Guillot, N., 2004. Characterization of Flow
Properties of Powder Coatings Used in the Automotive Industry. KONA Powder and Particle
Journal 22, 94-106.
D’Arco, A., Donsì, G., Ferrari, G., Montesano, M., Poletto, M., 2006. Discharge of Size-
Segregated Powders from a 2D-aerated Silo. KONA Powder and Particle Journal 24, 104-118.
Davidson, J.F., Nedderman, R.M., 1973. Trans. Inst. Chem. Engrs 51.
Deming, W.E., Mehring, A.L., 1929. The Gravitational Flow of Fertilizers and Other
Comminuted solids. Industrial and Engineering Chemistry, 661-665.
Donald, M., Roseman, B., 1962. Mechanisms in a horizontal drum mixer. British Chem. Eng
7, 749-753.
Dosekun, R., 1980. The flow of granular materials. . University of Cambridge.
Drescher, A., 1991. Analytical methods in bin-load analysis. Elsevier.
Drescher, A., 1992. On the criteria for mass flow in hoppers. Powder Technology 73, 251-
260.
Drescher, A., Ferjani, M., 2004. Revised model for plug/funnel flow in bins. Powder
Technology 141, 44-54.
Dyrøy, A., Karlsen, M., Enstad, G.G., de Silva, S., 2001. A system for the reduction of air
current segregation in silos, Handbook of Powder Technology, pp. 623-630.
Engblom, N., Saxén, H., Zevenhoven, R., Nylander, H., Enstad, G.G., 2012a. Effects of
process parameters and hopper angle on segregation of cohesive ternary powder mixtures in a
small scale cylindrical silo. Advanced Powder Technology 23, 566-579.
Engblom, N., Saxén, H., Zevenhoven, R., Nylander, H., Enstad, G.G., 2012b. Segregation of
construction materials in silos. Part 1: Experimental findings on different scales. Particulate
Science and Technology 30, 145-160.
Engblom, N., Saxén, H., Zevenhoven, R., Nylander, H., Enstad, G.G., 2012c. Segregation of
construction materials in silos. Part 2: Identification of relevant segregation mechanisms.
Particulate Science and Technology 30, 161-178.
Engblom, N., Saxén, H., Zevenhoven, R., Nylander, H., Enstad, G.G., 2012d. Segregation of
powder mixtures at filling and complete discharge of silos. Powder Technology 215–216,
104-116.
Evesque, P., de Gennes, P.-G., 1998. Sur la statique des silos. Compte Rendu Acad. Sci. Paris
2 b, 761-766.
Fan, Y., Jacob, K.V., Freireich, B., Lueptow, R.M., 2017. Segregation of granular materials in
bounded heap flow: A review. Powder Technology 312, 67-88.

50
Fowler, R.T., Glastonbury, J.R., 1959. The flow of granular solids through orifices. Chemical
Engineering Science 10, 150-156.
Franklin, F.C., Johanson, L.N., 1955. Flow of granular material through a circular orifice.
Chemical Engineering Science 4, 119-129.
Freese, B., 1977. Druckverthaltnisse in zylindrischen Silozelen. Universität Karlsruhe.
Fullard, L., Davies, C., 2016. Minimising the spread of residence-time distribution for flat and
heaped powders in a wedge-shaped planar hopper. Particuology.
Gardner, G.C., 1966. The region of flow when discharging granular materials from bin-
hopper systems. Chemical Engineering Science 21, 261-273.
Generalov, M.B., 1985. Discharge of bulk materials from apparatus. Theor. Found. chem.
Engng 19, 51.
Gentzler, M., Tardos, G.I., 2009. Measurement of velocity and density profiles in discharging
conical hoppers by NMR imaging. Chemical Engineering Science 64, 4463-4469.
Giunta, J.S., 1969. Flow patterns of granular materials in flat-bottom bins. J. Engng Ind.
ASME 91, 406-413.
Goyal, R.K., Tomassone, M.S., 2006. Power-law and exponential segregation in two-
dimensional silos of granular mixtures. Physical Review E - Statistical, Nonlinear, and Soft
Matter Physics 74.
Gray, J.M.N.T., Gajjar, P., Kokelaar, P., 2015. Particle-size segregation in dense granular
avalanches. Comptes Rendus Physique 16, 73-85.
Hagen, E., 1852. Druck und Bewegung des trockenen Sandes, Bericht über die zur
Bekanntmachung geeigneten Verhandlungen der Königlich Preußischen Akademie der
Wissenschaften zu Berlin.
Harmens, A., 1963. Flow of granular material through horizontal apertures. Chemical
Engineering Science 18, 297-306.
Hogg, R., 2009. Mixing and Segregation in Powders: Evaluation, Mechanisms and Processes.
KONA Powder and Particle Journal 27, 3-17.
Hunt, M.L., Weathers, R.C., Lee, A.T., Brennen, C.E., Wassgren, C.R., 1999. Effects of
horizontal vibration on hopper flows of granular materials. Physics of fluids 11, 68-75.
Huntington, A.P., Rooney, N.M., 1971. Discharge of granular materials from hoppers, Project
Report, , Department of Chemical Engineering, University of Cambridge.
Jackson, R., 1986. Some Features of the Flow of Granular Materials and Aerated Granular
Materials. Journal of Rheology 30, 907.
Janda, A., Zuriguel, I., Maza, D., 2012. Flow rate of particles through apertures obtained from
self-similar density and velocity profiles. Phys Rev Lett 108, 248001.
Janssen, H.A., 1895a. Z. Ver. Dtsch. Ing. 39, 1045-1049.
Janssen, H.A., 1895b. Versuche über getreidedruck in silozellzn. Ztg. Ver. Dt. Ing. 39, 1045-
1049.
Jenike, A.W., 1954a. Flow of solids in bulk handling systems. The American Society of
Mechanical Engineers.
Jenike, A.W., 1954b. Flow of solids in bulk handling systems, AIME, Minerals Benefication
Division Meeting, San Fransisco.
Jenike, A.W., 1954c. How Bulk Behaves. Flow 10, 74-126.
Jenike, A.W., 1961a. Gravity flow of Bulk solids. ASME.
Jenike, A.W., 1961b. Gravity flow of bulk solids, in: Engng. Exp. Station, U.U. (Ed.), Salt
Lake City.
Jenike, A.W., 1961c. Gravity flow of bulk solids., in: Engng. Exp. Station, U.U., Salt Lake
City (Ed.), USA.

51
Jenike, A.W., 1962a. Gravity Flow of Solids. Transactions of the Institution of Chemical
Engineers 40, 264-271.
Jenike, A.W., 1962b. Stress and velocity fields in gravity flow of bulk solids, Utah Engng stn
Bull.
Jenike, A.W., Johanson, J.R., Carson, J.W., 1973. Bin loads. Trans. ASME, J. of Eng. for Ind.
95, 1-16.
Job, N., Dardenne, A., Pirard, J.-P., 2009. Silo flow-pattern diagnosis using the tracer method.
Journal of Food Engineering 91, 118-125.
Johanson, J., 1995. Flow indices in the prediction of powder behavio, Pharmaceutical
Manufacturing International. Sterling Publications
Károlyi, A., Kertész, J., Havlin, S., Makse, H.A., Stanley, H.E., 1998. Filling a silo with a
mixture of grains: Friction-induced segregation. Europhysics Letters 44, 386-392.
Karolyi, A., Kertesz, J., Makse, H., Havlin, S., Stanley, H.E., 1999. Filling a silo: density
profiles and friction induced segregation. Computer Physics Communications 121–122, 672.
Ketchum, M.S., 1911. The design of walls bins and grain elevators _ The Engineering News
Publishing Company, New York
Ketterhagen, W., Curtis, J., Wassgren, C., 2006. Modeling granular segregation during hopper
discharge via discrete element methods, AIChE Annual Meeting, Conference Proceedings.
Ketterhagen, W.R., Curtis, J.S., Wassgren, C.R., Hancock, B.C., 2008. Modeling granular
segregation in flow from quasi-three-dimensional, wedge-shaped hoppers. Powder
Technology 179, 126-143.
Ketterhagen, W.R., Curtis, J.S., Wassgren, C.R., Hancock, B.C., 2009. Predicting the flow
mode from hoppers using the discrete element method. Powder Technology 195, 1-10.
Ketterhagen, W.R., Curtis, J.S., Wassgren, C.R., Kong, A., Narayan, P.J., 2005. Segregation
during hopper discharge: A DEM and experimental study, AIChE Annual Meeting,
Conference Proceedings, p. 3078.
Ketterhagen, W.R., Curtis, J.S., Wassgren, C.R., Kong, A., Narayan, P.J., Hancock, B.C.,
2007. Granular segregation in discharging cylindrical hoppers: A discrete element and
experimental study. Chemical Engineering Science 62, 6423-6439.
Khashayar, S., Pierre, G., 2009. Caractérisation et analyse des poudres Propriétés physiques
des solides divisés. Techniques de l'ingénieur Principes de formulation base documentaire :
TIB489DUO.
Khelil, A., 1989. Etude du champs de vitesses et de contraintes dans les silos métalliques.
INPL de Nancy.
Khelil, A., Roth, J.C., 1990. spécification des charges et des écoulements dans les silos
métalliques. Rev. Franç. Géotech, 11-25.
Koenders, M.A., Gaspar, N., Tüzün, U., 2001. The physical effects of structures formation in
granular materials. Physics and Chemistry of the Earth, Part A: Solid Earth and Geodesy 26,
75-82.
Koenen, M., 1896. Berechnung des Seiten und Bodendrucks in Silozellen. Centralblatt der
Bauverwaltung 16, 446–449.
Kotchanova, I.I., 1970. Experimental and theoretical investigations on the discharge of
granular materials from bins. Powder Technology 4, 32-37.
Kvapil, R., 1965. Gravity flow of granular materials in hoppers and bins. International Journal
of Rock Mechanics and Mining Sciences & Geomechanics Abstracts 2, 25-41.
Langston, P.A., Tüzün, U., Heyes, D.M., 1995. Discrete element simulation of internal stress
and flow fields in funnel flow hoppers. Powder Technology 85, 153-169.

52
Leturia, M., Benali, M., Lagarde, S., Ronga, I., Saleh, K., 2014. Characterization of flow
properties of cohesive powders: A comparative study of traditional and new testing methods.
Powder Technology 253, 406-423.
Levinson, M., Shmutter, B., Resnick, W., 1977. Displacement and velocity fields in hoppers.
Powder Technology 16, 29-43.
Litwiniszyn, J., 1972. Stochastic Methods in Mechanics of Granular Bodies. International
Centre for Mechanical Sciences.
Luk'yanov, P.I., Gusev, I.V., Nikitina, N.I., 1960. Khim. Tekh. Top. Mas. 5, 45.
Madrid, M.A., Darias, J.R., Pugnaloni, L.A., 2017. A differential equation for the flow rate
during silo discharge: Beyond the Beverloo rule, Powders & Grains 2017. EPJ Web of
Conferences 140, 03041 (2017), Montpellier - France.
Magalhães, F.G.R., Atman, A.P.F., Moreira, J.G., Herrmann, H.J., 2016. Analysis of the
velocity field of granular hopper flow. Granular Matter 18, 33.
Marinelli, J., Carson, J.W., 1992. Solve Solids Flow Problems in Bins, Hoppers, and Feeders.
Chemical Engineering Progress.
Matchett, A.J., 2007. The shape of the cohesive arch in hoppers and silos — Some theoretical
considerations. Powder Technology 171, 133-145.
Medina, A., Córdova, J.A., Luna, E., Treviño, C., 1998. Velocity field measurements in
granular gravity flow in a near 2D silo. Physics Letter A 250, 111-116.
Meng, Q., Jofriet, J.C., Negi, S.C., 1997. Finite Element Analysis of Bulk Solids Flow. Part 2:
Application to a Parametric Study. J. Agric. Engng Res. 67.
Mosby, J., de Silva, S.R., Enstad, G.G., 1996. Segregation of Particulate Materials –
Mechanisms and Testers. KONA Powder and Particle Journal 14, 31-43.
Mroz, Z., Szymanski, C.Z., 1971. Gravity flow of a granular material in a converging channel.
Arch. of Mech 23, 897.
Mullins, W.W., 1972. Stochastic Theory of Particle Flow under Gravity. Journal of Applied
Physics 43, 665-678.
Mullins, W.W., 1974. Experimental evidence for the stochastic theory of particle flow under
gravity. Powder Technology 9, 29-37.
Myers, M.E., Sellers, M., 1977. Rate of discharge from wedge-shaped hoppers, in: report, F.-
y.p. (Ed.), Master's thesis Department of Chemical Engineering, Cambridge.
Nedderman, R.M., 1988. The measurement of the velocity profile in a granular material
discharging from a conical hopper. Chemical Engineering Science 43, 1507-1516.
Nedderman, R.M., 1992. Statics and kinematics of granular materials. Cambridge University
Press, Great Britain.
Nedderman, R.M., Tüzün, U., 1979. A kinematic model for the flow of granular materials.
Powder Technology 22, 243-253.
Nedderman, R.M., Tüzün, U., Savage, S.B., Houlsby, G.T., 1982. The flow of granular
materials—I: Discharge rates from hoppers. Chemical Engineering Science 37, 1597-1609.
Newton, R.H., Dunham, G.S., Simpson, T.P., 1945. Trans. Amer. Inst Chem. Engrs 41, 215.
Nguyen, V.D., Cogné, C., Guessasma, M., Bellenger, E., Fortin, J., 2009. Discrete modeling
of granular flow with thermal transfer: Application to the discharge of silos. Applied Thermal
Engineering 29, 1846-1853.
Ooi, J.Y., Chen, J.F., Rotter, J.M., 1998. Measurement of solids flow patterns in a gypsum
silo. Powder Technology 99, 272-284.
Pemberton, C.S., 1965. Flow of imponderable granular materials in wedge-shaped channels.
Journal of the Mechanics and Physics of Solids 13, 351-360.
Polderman, H.G., Boom, J., De Hilster, E., Scott, A.M., 1987. Solids flow velocity profiles in
mass flow hoppers. Powder Technology 42, 737.

53
Prescott, J.K., Barnum, R.A., 2000. On Powder Flowability. Pharmaceutical Technology, 60-
84.
Purutyan, H., Carson, J.W., Troxel, T.G., 2004. Improve Solids Handling During Thermal
Drying. Chemical Engineering Progress, 26-30.
Purutyan, H., Pittenger, B.H., Carson, J.W., 1998. Solve Solids Handling Problems by
Retrofitting. Chemical Engineering Progress.
Purutyan, H., Pittenger, B.H., Carson, J.W., 1999. Six Steps to Designing a Storage Vessel
that Really Works. Powder and Bulk Engineering 13, 56-68.
Ramirez-Gomez, A., Gonzalez-Montellano, C., Gallego, E., Ayuga, F., 2012. Three
dimensional discrete element models for simulating the filling and emptying of silos:
analysis of numerical results. Computers & Chemical Engineering 40
22-32.
Rausch, J.M., 1948. Gravity flow of solid beds in vertical towers. Princeton University.
Reimbert, M., Reimbert-Auclair, A.M., 1982. Silos, théorie et pratique. Eyrolles.
Rhodes, M.J., 1998. Introduction to Particle Technology. John Wiley & sons, New York.
Roberts, A.W., 1990. Modern concepts in the design and engineering of bulk solids handling
systems. TUNRA Ltd., The Univ. of Newcastle, N.S.W., Australia.
Roberts, I., 1882. On the pressure of wheat stored in elongated cells or bins. Engineering 34,
399.
Roberts, I., 1884. Determination of vertical and lateral pressures of granular substances.
Proceed. Royal. Soc. of London 36, 225-240.
Rose, H.E., Tanaka, T., 1959. Rate of discharge of granular materials from bins and hoppers.
Engineers, London 208, 465.
Royal, T.A., Carson, J.W., 1991. Fine Powder Flow Phenomenoa in Bins, Hoppers and
Processing Vessels, Bulk 2000: Bulk material Handling Towards the Year 2000, London.
Rubio-Largo, S.M., Janda, A., Maza, D., Zuriguel, I., Hidalgo, R.C., 2015. Disentangling the
Free-Fall Arch Paradox in Silo Discharge. Phys. Rev. Lett. 114.
Saleh, K., Guigon, P., 2009. Caractérisation et analyse des poudres Propriétés
comportementales des solides divisés. Techniques de l'ingénieur Principes de formulation
base documentaire : TIB489DUO.
Saleh, K., Guigon, P., 2012. Mise en oeuvre des poudres - Stockage et écoulement des silos,
in: l'Ingénieur, T.d. (Ed.), Techniques de l'Ingénieur, Paris, pp. 1-29.
Salter, G.F., Farnish, R.J., Bradley, M.S.A., Burnett, A.J., 2000. Segregation of binary
mixtures of particles during the filling of a two-dimensional representation of a hopper.
Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical
Engineering 214, 197-208.
Samadani, A., Pradhan, A., Kudrolli, A., 1999. Size segregation of granular matter in silo
discharges. Phys. Rev. E60, 7203-7209.
Schulze, D., 2008. Powders and bulk solids : behavior, characterization, storage and flow.
Springer, Berlin ; New York.
Seil, P., Gómez, J.O., Pirker, S., Kloss, C., 2012. Numerical and experimental studies on
segregation patterns in granular flow from two hoppers, ECCOMAS 2012 - European
Congress on Computational Methods in Applied Sciences and Engineering, e-Book Full
Papers, pp. 5389-5402.
Shamlou, P.A., 1988. Handling of bulk solids : theory and practice. Butterworths, London ;
Boston.
Shinohara, K., Golman, B., 2002. Segregation indices of multi-sized particle mixtures during
the filling of a two-dimensional hopper. Advanced Powder Technology 13, 93-107.

54
Shinohara, K., Golman, B., 2003. Density segregation of a binary solids mixture during batch
operation in a two-dimensional hopper. Advanced Powder Technology 14, 333-347.
Shinohara, K., Golman, B., Nakata, T., 2001. Size segregation of multicomponent particles
during the filling of a hopper. Advanced Powder Technology 12, 33-43.
Shinohara, K., Idemitsu, Y., Gotah, K., Tanaka, T., 1968. Mechanism of gravity flow of
particles from a hopper. Ind. Eng. Chem. Process Design and Development 9, 2.
Shinohara, K., Shoji, K., Tanaka, T., 1970. Mechanism Of Segregation And Blending Of
Particles Flowing Out Of Mass-Flow Hoppers. Industrial and Engineering Chemistry Process
Design and Development 9, 174-180.
Shinohara, K., Shoji, K., Tanaka, T., 1972. Mechanism of Size Segregation of Particles in
Filling a Hopper. Industrial and Engineering Chemistry Process Design and Development 11,
369-376.
Shinohara, K., Tuzun, U., 2002. Solids flow mechanisms and their applications. Chemical
Engineering Science 57, 213-214.
Shirai, 1952. Powder Orifice Nomograph. Kagaku Kikai 16.
Sielamowicz, I., Blonski, S., Kowalewski, T.A., 2005. Optical technique DPIV in
measurements of granular material flows, Part 1 of 3—plane hoppers. Chemical Engineering
Science 60, 589-598.
Sielamowicz, I., Błoñski, S., Kowalewski, T.A., 2006. Digital particle image velocimetry
(DPIV) technique in measurements of granular material flows, Part 2 of 3-converging
hoppers. Chemical Engineering Science 61, 5307-5317.
Sielamowicz, I., Czech, A., Kowalewski, T.A., 2015. Comparative analysis of empirical
descriptions of eccentric flow in silo model by the linear and nonlinear regressions. Powder
Technology 270, 393-410.
Sielamowicz, I., Czech, M., 2010. Analysis of the radial flow assumption in a converging
model silo. Biosystems Engineering 106, 412-422.
Sielamowicz, I., Czech, M., Kowalewski, T.A., 2010. Empirical description of flow
parameters in eccentric flow inside a silo model. Powder Technology 198, 381-394.
Sielamowicz, I., Czech, M., Kowalewski, T.A., 2011a. Empirical analysis of eccentric flow
registered by the DPIV technique inside a silo model. Powder Technology 212, 38-56.
Sielamowicz, I., Czech, M., Kowalewski, T.A., 2011b. Empirical description of granular flow
inside a model silo with vertical walls. Biosystems Engineering 108, 334-344.
Smith, L., Tüzün, U., 2002a. Life after computer simulations: towards establishing bulk
evolution rules based on discrete granular dynamics. Chemical Engineering Science 57, 253-
264.
Smith, L., Tüzün, U., 2002b. Stress, voidage and velocity coupling in an avalanching granular
heap. Chemical Engineering Science 57, 3795-3807.
Sokol, L., 1984. Bin loads in vertical silos. International journal of bulk solids in silos 1.
Standish, N., 1985. Studies of size segregation in filling and emptying a hopper. Powder
Technology 45, 43-56.
Standish, N., Jones, J.J., 1984. Tracer study of discharge segregation from a paul wurth
hopper, Symposia Series - Australasian Institute of Mining and Metallurgy, 36 ed, pp. 279-
285.
Staron, L., Lagrée, P.-Y., Popinet, S., 2012. The granular silo as a continuum plastic flow:
The hour-glass vs the clepsydra. Phys. Fluids 24
Stavans, J., 1998. Axial Segregation of Powders in a Horizontal Rotating Tube. Journal of
Statistical Physics 93, 467-475.

55
Steingart, D.A., Evans, J.W., 2005. Measurements of granular flows in two-dimensional
hoppers by particle image velocimetry. Part I: experimental method and results. Chemical
Engineering Science 60, 1043-1051.
Tanaka, T., Kawai, S., 1956. Gravitational Flow of Particles From Hopper Funnels. Chemical
Engineering 20, 144-147.
Tang, P., Puri, V.M., 2004. Methods for Minimizing Segregation: A Review. Particulate
Science and Technology 22, 321-337.
Tao, H., Zhong, W., Jin, B., Ren, B., 2013. DEM simulation of non-spherical granular
segregation in hopper, AIP Conference Proceedings, pp. 720-726.
Theimer, O.F., 1958. On the storage of raw cocoa beans in silo compartments. International
Chocolate Review V Jahr. XIII.
Tüzün, U., Houlsby, G.T., Nedderman, R.M., Savage, S.B., 1982. The flow of granular
materials—II Velocity distributions in slow flow. Chemical Engineering Science 37, 1691-
1709.
Tüzün, U., Nedderman, R.M., 1982. An investigation of the flow boundary during steady-
state discharge from a funnel-flow bunker. Powder Technology 31, 27-43.
Tüzün, U., Nedderman, R.M., 1985. Gravity flow of granular materials round obstacles—I:
Investigation of the effects of inserts on flow patterns inside a silo. Chemical Engineering
Science 40, 325-336.
Volpato, S., Artoni, R., Santomaso, A.C., 2014. Numerical study on the behavior of funnel
flow silos with and without inserts through a continuum hydrodynamic approach. Chemical
Engineering Research and Design 92, 256-263.
Walker, D.M., 1966. An approximate theory for pressures and arching in hoppers. Chemical
Engineering Science 21, 975-997.
Walters, J.K., 1973. A theoretical analysis of stresse in silos with vertical walls. Chemical
Engineering Science 21, 13-21.
Waters, A.J., Drescher, A., 2000. Modeling plug flow in bins/hoppers. Powder Technology
113, 168–175.
Watson, G.R., Rotter, J.M., 1996. A finite element kinematic analysis of planar granular
solids flow. Chemical Engineering Science 51, 3967-3978.
Wieghardt, K., 1952. Über einige Versuche an Strömungen in Sand. Ing-Arch. 20, 109–115.
Williams, J.C., 1963. The segregation of powders and granular materials. Univ. Sheffield Fuel
Soc. J. 14, 29−34.
Williams, J.C., 1965. Cause and effects of segregation in powders. chemical processing.
Williams, J.C., 1976. The segregation of particulate materials. A review. Powder Technology
15, 245-251.
Williams, J.C., 1977. The rate of discharge of coarse granular materials from conical mass
flow hoppers. Chemical Engineering Science 32, 247-255.
Windows-Yule, K., Parker, D., 2015. Density-Driven Segregation in Binary and Ternary
Granular Systems. KONA Powder and Particle Journal 32, 163-175.
Wu, S., Kou, M., Xu, J., Guo, X., Du, K., Shen, W., Sun, J., 2013. DEM simulation of particle
size segregation behavior during charging into and discharging from a Paul-Wurth type
hopper. Chemical Engineering Science 99, 314-323.
Yu, Y., Saxén, H., 2014. Segregation behavior of particles in a top hopper of a blast furnace.
Powder Technology 262, 233-241.
Zhang, K., 1997. Flow of Particulate Solids in Silos The University of Edinburgh
Zigan, S., Thorpe, R.B., Tuzun, U., Enstad, G.G., Battistin, F., 2008. Theoretical and
experimental testing of a scaling rule for air current segregation of alumina powder in
cylindrical silos. Powder Technology 183, 133-145.

56
Table 1: Correlations for vertical stress and wall stress in static and dynamic states
Mass flow
section Eq. N° Model Remarks
Janssen (1895)
T1.1 Infinitely deep bin
vertical section,
Static field – filling or at rest

Reimbert (1982)
Bin

T1.2 Shallow bin

Jenike (1962) and Walker (1966): : the vertical stress at the top of the hopper
section, hopper
Convergent

For Jenike:
T1.3
For Walker:

Janssen model: Eq. (T1-1)


Bin

- Only for perfectly straight lateral walls


Dynamic field – Discharge

Walker (1966) and Walter (1973)


Hopper

T1.4

: angle between the wall and the major principal plane


Funnel flow

Dynamic field
T1.5 Janssen model:
Discharge

58
Coordinates Corresponding equation of continuity Further specific assumptions
General:
Any coordinates -
Incompressible:
Cartesian square silo :
z General:

Incompressible:
Flat (2D) silo :
y
o

Cylindrical General:
No angular velocity :

Incompressible: Then:

y
o

x

Spherical z
General:
No angular velocity :

r Incompressible: Then:

y
o
j
59
x
Table 2: Equation of continuity and specific cases for different silo geometries

60
Table 3: Different correlation for reported in litterature

Eq. n° Ref. Expression of Remark

T3.1 Brown
(1961)

T3.2 Jenike A is an unknown constant


(1962)
T3.3 Mroz & Szymanski
(1971)

T3.4 Generalov
(1985)
T3.5 Polderman et al.
(1987)
with :

61
Table 4: Analytical solutions for the velocity profile for some different geometries of the orifice

Eq. n° Particular case Analytical solution

steady state:
T4.2
Point orifice draining a semi-infinite bed
Transient (time dependent):
T4.3

T4.4 Finite orifice draining a semi-infinite bed

T4.5 point orifice draining a vertical pipe

62
Table 5: Correlations for discharge mass flow rate from hoppers
Eq. Ref. Correlation Remarks

T5.1 Ketchum et al. Flat-bottom square silo: (0.3mx0.3mx2.44 m)
(1911) Circular orifice : 5 cm
T5.2 Free falling Any silo geometry
(e.g. Shamlou, 1988) Circular orifice
: discharge coefficient
T5.3 Deming and Mehring Conical silo, circular orifices
(1929) : angle of repose
: half-angle of hopper
: 15°; 30°; 45°
B : 1; 2; 3; 5; 10 mm
T5.4 Newton et al. Flat-bottom silo, circular orifice
(1945)
T5.5 Rausch (1948) C : Coefficient of Rausch’s equation

T5.6 Shirai (1952)

T5.7 Franklin and Johanson Flat-bottom silo, Circular orifices


(1955) : in inches
: density of material in lb/ft3
With : : in lb/min
F: correction factor for conical hoppers
T5.8 Tanaka and Kawai Not valid for mean particle sizes less than 150 micron
(1956) for

for
T5.9 Fowler and Glastonbury Flat-bottom silo, Circular orifices
(1959) Various geometry and size of orifices
: shape factor of particles

T5.10 Rose and Tanaka for


1959
for

63
: the included semi-angle of the flow channel
T5.11 Luk’yanov et al. Flat-bottom silo, Circular orifices
(1960) Original equation based on volume
T5.12 Brown and Richards Flat-bottom silo, Circular orifices
(1961)
T5.13 Beverloo et al. C = 0.58 in SI units
(1961) k =1.3 – 2.9
flat-bottom and wedge-shaped silo, circular orifices
T5.14 Harmens : base angle of hypothetical conical discharge surface
(1962) if
if

T5.15 Shinohara et al.


(1968)

Flat-bottom silo, Circular and slot orifices


Circular opening:
T5.16 Brown & Richards
(1970) Slot opening :
T5.17 Carleton Mass flow in conical silos
(1972)
T5.18 Davidson and Conical and chisel-shaped hoppers,
Cylindrical silos: circular and rectangular orifices
Nedderman
(1973) Chisel-shaped silos:
T5.19 Williams Conical hopper, circular orifices
(1977)
T5.20 Myers and Sellers Flat-bottom silo, slot orifices
1977
T5.21 Nedderman et al. C’ 2.35 for Beverloo Eq.
(1982)
Conical silo, rectangular orifices
T5.22 Al-Din & Gunn Long rectangular orifices,
(1984) A: the flow area
Fo and Fp: orifice and particle constants

64
Fo=1.0 for circular and half square triangular orifices
Fo=1.4 for rectangular and elliptical orifices
Fp = 1 for round particles, 0.75 for sharp-edge isomeric
particles, 0.32 for irregular shape particles

65
Table 6: A summary of significant works on segregation in hoppers

Material Process Approach Subject of the study-parameters studied


Ref.

General/review

Particle density
Experimental
Other parameters Remarks

Fundaments-
powder silo

Mechanisms

composition
modelling
Emptying

size ratio
Size and
Filling
Brown 1939 - Heaps and bunkers x x x x General article but focused on
segregation of coal
Williams 1963 Binary mixture of spherical
Perspex box The effect of
particles (d=2.88mm and x x x Filling by pouring
L24” x H18” x 1” T Vibration
0.55 mm)
Shinohara, Shoji, Transparent, Mass-flow
Binary mixture of particles
and Tanaka 1970 (d=1.70 mm and 0.120 mm) =15°, variable slit x x x x Marginally deals with the
size, effect of the density
Shinohara, Shoji, Transparent, Height of the free-
Binary mixture of particles
and Tanaka 1972 =15°, variable slit x x x fall
(d=1.70 mm and 0.120 mm)
size, Feed rate
Standish 1985 Ternary mixtures of ore
Height of the free- Studied also the Residence
(r=3500) and coke particles Model P-W model
x x x x x fall Time Distribution vs the
(800 kg.m-3) (d=1, 4 and hopper
initial position
6,15 mm)
Standish and Kilic Mode of
Ferrous sinter of 8 different
1985 Model P-W model discharge: Blast furnace in stop-start
size fractions within the x
hopper continuous vs mode
range 6-0.25 mm
stop-start
Carson, Royal, and General article including
x x x x
Goodwill 1986 several case studies
Artega and Tüzün Binary mixtures of four
Effect of the
1990 types of materials of near Perspex cylindrical
microstructure of
spherical shape : ABS, bunker of 112.5 mm x x x x x x
the bed and the
acrylic beads, turnip and diam.
flowrate
radish seeds

66
Bagster 1996 Binary mixtures of coarse
sand (d50%1mm) and four
The effect of Dimensions of the silo are not
grades of fine sand Flat bottom 2D silo x x x x
moisture content specified
(d50%=145, 235 and 330
µm)
Mosby, de Silva, Presents some segregation
x x x
and Enstad 1996 testers
Cizeau, Makse, and Presents a model for
Binary mixture Flat bottom 2D silo x x x
Stanley 1999 stratification
Karolyi et al. 1999 The effect of Granular Media Lattice Gas
Binary mixtures 2D silo x x x
friction properties (GMLG) model
Samadani, Pradhan, Flat bottom 2D silo
Glass beads (d=0.2, 0.5, Void filling model of
and Kudrolli 1999 L89cm x H45 cm x x x x x Feed rate
0.6, 0.7 and 1.1 mm) segregation
W1.27
Salter et al. 2000 Binary mixtures of various
size ratio (1.4:1, 1.85:1, Segregation test facility x x x x x Feed rate Pilot scale 2D silo
2.36:1 and 3.1:1)
Dyrøy et al. 2001 Develops a system for the
Alumina mixtures Industrial silos x x x reduction of air current
segregation
Shinohara, Golman, Multicomponent mixtures Transparent 2D silo Number of Provides a correlation for
x x x x
and Nakata 2001 of glass beads (L0.5m x H0.75 m) components segregation index
Shinohara and Multicomponent mixtures
Number of
Golman 2002 of glass beads (0.174, Transparent 2D silo Introduces a segregation index
x x x x components
0.358, 0.507, 0.784, 1.660 (L0.5m x H0.75 m) for multicomponent mixtures
Feed rate
mm)
Shinohara and Binary mixtures of different
Golman 2003 density : glass beads (2520 Transparent 2D silo Feed rate and silo Makes use of particles of the
x x x x
kg.m-3) and lead shots (L0.5m x H0.75 m) angle same size (165 µm)
(11300 kg.m-3)
Tang and Puri 2004 Describes methods for
x
minimizing segregation
Ketterhagen et al. Hopper geometry:
Spherical, bi-disperse
2005 Different geometries x X x x angle, aperture Uses DEM simulation
mixtures
size, fill height,
D’Arco et al. 2006 2D aerated silo:
FCC and glass powders H800mm x W180mm x x x x Aeration flow rate Image acquisition
10mm T

67
Goyal and acrylic beads of size
Heap segregation when filling
Tomassone 2006 500–600 µm and acrylic or
from a corner
glass beads (90–130, 180–
2D silo: 22cm long x x X x x Analytical model based on the
125, 850–1000, 360–425,
minimal model proposed by
675–775, and
BdG
400–450 µm)
Ketterhagen, Curtis, Spherical, bi-disperse 3D wedge-shaped and
x x X x Uses DEM simulation
and Wassgren 2006 mixtures cylindrical hoppers
Ketterhagen et al. Spherical, bi-disperse Methods of filling
2007 mixtures, glass beads Interaction
3D wedge-shaped x x X x x Uses DEM simulation
(0.521, 1.16, 1.28 and 2.24 properties
mm) Silo geometry
Ketterhagen et al. Methods of filling
2008 3D wedge-shaped x X x Friction coef. Uses DEM simulation
Silo geometry
Zigan et al. 2008 Provides a scaling rule for air
Industrial cylindrical Feed rate
Alumina x x x current segregation
silo Air extraction rate
Fluent simulation
Hogg 2009 Gives specific examples of
x segregation in flow over
surfaces
Anand et al. 2010 Particle cohesion
Cohesive powders Flat bottom 2 D silo x X due to liquid Uses DEM simulation
bridges
Engblom et al. Ternary mixtures: sand (0.8 Cylindrical silo with
Effect of hopper
2012a and 1.5 mm) limestone conical hopper: 600mm
x x x x x angle and free-fall Funnel flow and mass flow
(0.15 and 0.45 mm) and Diam., 200 mm Diam.
distance
cement (0.015mm) Aperture,
Engblom et al. Cylindrical silo with
Makes use of dimensionless
2012d Binary and ternary mixtures conical hopper: 600mm
x x x x x x x free-fall distance analysis to reduce the number
of Cement, Lime and Glue Diam., 200 mm Diam.
of experiments
Aperture,
Engblom et al. Cylindrical silo with
2012b, 2012c sand limestone and cement conical hopper of x x x x x x Silo geometry Use of large scale silos
different size
Seil et al. 2012 Spherical particles (2.65,
Flat bottom 2Dsilo x x X x x DEM simulation
3.15, 4 and 6 mm)
Tao et al. 2013 Non-spherical particles Wedge-shaped x x Shape DEM simulation

68
rectangular silo Non-spherical particles
Wu et al. 2013 Sinters (80, 200 and 400 Paul-Wurth type DEM simulation
x x x Hopper slope
mm) hopper
Yu and Saxén 2014 DEM simulations
Four different shape of Top hopper of a blast Non-spjherical particles
x x Shape
particles furnace Comparison with Standish’s
experiments [165]
Windows-Yule and Vibro fluidized bed of Use of Positron Emission
Spherical 3mm particles x x Particle elasticity
Parker 2015 particle Particle Tracking (PEPT)
Bertuola et al. 2016 Data from literature-Glass Taken from Model based on continuum
beads (521, 1160,2240 µm), Ketterhagen et al Eulerian approach
x
Acrilic (605 µm) and radish (2007) and Arteaga and Comparison with data from
seed (2410 µm) Tuzun (1990) literature
Combarros Garcia Flat-bottom and Comparison between
et al. 2016 Sand (0.26 and 1.6 mm) wedge-shaped silo (90 x x x Method of filling experiments, DEM simulation
kg) and continuum modelling
Sakai 2016 x x DEM simulation
Fan et al. 2017 3 different continuum
x approach to model
segregation in bounded heaps
Spherical particles (3, 5, 6,
Xu et al. 2017 7 mm) with different
Wedged-shape hopper x x X DEM simulation
densities (1050, 2100 and
4200 kg.m-3)

69
Graphical abstract

You might also like