You are on page 1of 10

J. agric. Engng Res.

(2001) 78 (3), 299}308


doi:10.1006/ jaer.2000.0640, available online at http://www.idealibrary.com on
SE*Structures and Environment

Static and Dynamic Silo Loads using Finite Element Models


F. Ayuga; M. Guaita; P. Aguado
E.T.S.I. Agro& nomos, Polytechnic University of Madrid, Ciudad Universitario s/n, 28040 Madrid, Spain;
e-mail of corresponding author: Ayuga@cvr.etsia.upm.es
E.P.S. University of Santiago de Compostela, Campus Universitario, 27002 Lugo, Spain; e-mail: guaita@lugo.usc.es
E.S.T.I. Agraria, University of Leo& n, Au, Portugal, 41.24071 Leon, Spain; e-mail: diapar@unileon.es

(Received 18 October 1999; accepted in revised form 29 August 2000; published online 24 October 2000)

Di!erent "nite element models for axisymmetric silo analysis, that simultaneously considers the behaviour of
the grain and the structure, are proposed in this paper. The use of commercial programmes and integration of
modern theories on pressures exerted by grain are the main premises of the research. The paper presents models
for both the static and dynamic conditions. In both cases, the in#uence of di!erent parameters is discussed.
A new method for emptying pressure determination is also proposed.
 2001 Silsoe Research Institute

1. Introduction and institutions are based, including recent international


standards such as Eurocode 1, Part 4 (ENV, 1995).
Storing of bulk materials in silos is essential in a large During the silo discharge, the vertical and horizontal
number of industries, and major investments are made in pressures can exceed the static results considerably. This
their construction both in the public and private sectors. has been known and investigated throughout the 20th
During the last 100 years, many accidents, explosions, century, but even today its cause, the factors that inter-
cracking and excessive deformations have been common vene and the predictable pressure values, all remain very
in silo structures due to the lack of knowledge about their uncertain. In order to overcome this di$culty, re-
structural behaviour. Such failures are costly in terms of searchers have proposed to modify the static pressures,
repair or replacement cost, loss of production and injury calculated by Janssen's theory, through the use of over-
or loss of life. Even after more than a century of research, pressure multiplication factors obtained in various re-
many uncertainties still exist in various areas of silo search and veri"ed in practice.
structural behaviour. Throughout the world, there are From the 1970s onwards, a large number of research
a large number of research teams investigating silo struc- teams have worked on the application of "nite element
tures, but the more advances that are made, the more analysis to silo problems (Manbeck & Nelson, 1975;
complex the problem appears. Mahmoud, 1975; Jofriet et al., 1977; Maeda & Ishiyaki,
Di!erent standards and rules can be used for silo 1979). At that time, models and programmes were ham-
design, all of which are based on theoretical and empiri- pered due to the limited capacity of computers and the
cal methods. Almost every standard uses Janssen's (1895) high cost of equipment. Furthermore, commercial pro-
theory which proposed equations based on a horizontal grammes for these purposes had not been developed. In
section of stored material that is in contact with the silo contrast today, it is possible to use computers whose
walls. In this theory, the e!ects of the stored material capacity and speed is continually increasing and a large
acting on the structure only depend on the silo hydraulic number of programs exist which not only calculate by
radius, the grain bulk density, the grain friction angles in means of "nite elements but also manipulate, analyse and
contact with the silo wall and the value of k (ratio present the results. A gap exists in the international
between horizontal and vertical pressures). In spite of its bibliography concerning the use of the potent commer-
limitations, and a century after its appearance, this re- cial programmes that are currently available for both
mains the theory upon which the standards of all countries silo research and for the development of "nite element
0021-8634/01/030299#10 $35.00/0 299  2001 Silsoe Research Institute
300 F. AY U G A E ¹ A ¸.

models, which simultaneously work with the behaviour Sixteen "les were generated in order to represent the
of the stored material and that of the silo structure. various problems in silo analysis (static and dynamic
This article presents a "nite element model for the analysis, silos with #at bases, with hoppers, walls made
design of axisymmetric silos using the capacities of the of steel or concrete, etc.). With the help of these
ANSYS 5.3 (ANSYS, 1998) programme that simulta- models it has been possible to confront the di!erent
neously considers the behaviour of the grain and the problems which appear during the research, such as
structure at the same time. grain be-haviour, silo wall behaviour, contact prob-
The main structural problem for silos occurs when the lems, the "t of the mesh, and emptying and "lling the
greatest pressures are produced (during their discharge). silo.
The "rst disaster caused by overpressures during dis- Equal number of silos with #at bases and silos with
charge occurred in the 1930s when, with the intention of hoppers were studied. However, for discharge, silos with
reducing the factor of safety, structural calculation hoppers were chosen because of the focus of the research
methods were being re"ned and construction materials on the various #ow types that occur during this process,
were being improved. Before that period it had not been and in order to make use of the experimental data
easy to alter structural design, and large loads were used gathered by other authors, the majority of which corres-
for the calculation. Since then, extensive research has ponds to this type of silo.
been undertaken into the causes of overpressures during Two "nite elements were used, in accordance with
discharge, although it cannot be said that the problem ANSYS codes. A four-node element, having the possibili-
has been solved yet. ty of axisymmetry, for ensiled material and an element of
All current standards and suggestions for silo calcu- two nodes superimposed, having the possibility of taking
lations use coe$cients of overpressures during discharge into account friction and the loss of contact, for
and because of this they are unable to correctly evaluate grain}wall and grain}grain contact.
the complexity of the phenomenon.
The present research assumes the validity of, and is
principally based upon, the contributions of Jenike made 3. Structural behaviour of the silo in a static condition
during the 1960s (Jenike & Johanson, 1968). They clearly
de"ned the di!erent types of pressures against the walls In order to test the models and to study the in#uence
as a consequence of the change in the stress state in the of the principal parameters upon them, the silo was "rst
grain (passing from an active to a passive state and studied in a static condition, that is, with the grain at rest.
unbalancing the load in the process). This theory has An example of the models which were created, a diagram
been combined with Zhang's most recent contributions of a #at-bottom silo and steel walls, can be seen in Fig. 1.
(Zhang et al., 1994) relating to the decisive in#uence of The mesh is only indicative, as it was one of the para-
the dilatancy angle of the grain during the discharge meters analysed during the research. Similar models were
process and the consequent overpressure. created using the various combinations of concrete and
As a result, this article proposes a new method of rigid walls ("xed nodes) and the di!erent base shapes, #at
analysing silo discharge which combines both e!ects and and hopper.
implements them in a "nite element model. The results In order to simplify the calculations, the models
are compared with the previous research data and with developed using "nite elements are axisymmetrical.
those indicated by the various standards. Therefore, the simulated silo is cylindrical with a central
Using this model, conclusions can be drawn about the discharge. Silo sizes, whether in terms of height or of
di!erent in#uences of the mechanical parameters of grain diameter, could be varied as required, but for the pur-
on the silo design. poses of this research two sizes were selected (Fig. 2).
The material behaviour chosen is based on the
Drucker-Prager criterion of plasticity, one used when
2. Principles of the 5nite element models considering the e!ect of expansion on #ooring materials
(Drucker & Prager, 1952). The yield surface is a circular
The objective of this work was to produce results that cone with the material parameters chosen such that it
could easily be incorporated into the industry and be corresponds to the outer aspices of the hexagonal
used by silo-design engineers. A few commercial pro- Mohr}Coulomb yield surface.
grammes that ful"l these conditions already exist, based The chosen material was wheat - speci"cally, Camacho
on the theory of "nite elements. For this research, the wheat - since it is the most common material found in
university version of ANSYS 5.3 was used, a version that research conducted by other authors. The required char-
only di!ers from the commercial one in that the wave- acteristics of the wheat were obtained by means of labor-
front is limited to 800 degrees of freedom (DOF). atory tests. The values used are shown in Table 1. The
ST AT IC AN D DY NA MI C SI LO L O A D S 301

Fig. 2. Selected silo dimensions: (a) yat-bottom silo; and (b) silo
Fig. 1. Cylindrical silo with yat base and metal sides with hopper at a slope angle (a) of 453

range of values corresponds to those used to analyse the test for Poisson's ratio and Young's modulus from the
in#uence of these parameters on the results. edometric modulus using an edometer.
The entire development of these models can be consul- The results from the analyses are quite varied but, for
ted in Guaita (1995). The di!erent parameters that inter- the purposes of comparison and validation, the horizon-
vene were systematically analysed and the results were tal pressures on the walls, the vertical pressures on the
compared with those obtained by means of traditional walls and the relation to pressure on the silo wall k were
calculation theories and international standards. The chosen. Furthermore, for the analysis of the di!erent
parameters analysed were: Young's modulus, Poisson's parameters, results were obtained from the horizontal
ratio, internal friction angle, dilatancy angle, angle of and vertical pressures of the stored material, the zones of
friction with the wall and bulk density. The in#uence elastic and plastic behaviour within the stored material,
of the element size, wall thickness and the di!erent possi- and the distortions and tensions within the silo wall.
bilities of silo}hopper joint meshing, were also analysed.
It should be taken into account that traditional the-
ories and international standards only consider the inter- 4. Results of the static models
nal friction angle, the angle of friction with the wall and
the bulk density. Occasionally, the relationship between 4.1. Pressures against the walls
the horizontal and vertical pressures on the wall is used
as a parameter although it is not one in reality. The validity of the static models, for a #at-bottom silo,
The experimental procedures to measure the material becomes evident from Fig. 3(a). It can be observed that,
properties were: a direct shear test for the internal friction using parameters obtained by experimental tests shown
angle and the dilatancy, a direct shear test with a modi- in Table 1, the curve of the horizontal pressures on the
"ed box for the angle of friction with the wall, a triaxial wall compares very well with those obtained by means of

Table 1
Characteristics of the wheat with the range values used to analyse their in6uence on the results and the values obtained by means of
laboratory tests

Material parameter Range of values analysed Values obtained by


experimental tests

Elasticity module of stored grain E, kPa 5000 5000


Speci"c weight of grain o, kN m\ 7}9 9
Poisson ratio of grain l 0)2}0)4 0)3
Internal friction angle
,3 22}30 22
Wall}grain friction angle k,3 14)03 14)03
Dilatancy angle u,3 0}19 2)5
302 F. AY U G A E ¹ A ¸.

Fig. 3. (a) Horizontal pressures and (b) the ratio of horizontal to vertical pressures (k) under static conditions for the silo dimensions
shown in Fig. 2(a) and the material parameters shown in Table 1: , xnite element method; , Caquot and Kerisel (1956);
, Janssen (1895); , Reimbert and Reimbert (1980); , Eurocode (1995); , DIN 1055-6 (1987)

traditional theories and international standards. The 4.2.2. Poisson1s ratio


same occurs with the vertical pressures and with the ratio This is a key parameter in determining the grain pres-
of horizontal to vertical pressures k, although this ratio sures in a static position. Variation in this coe$cient
remains constant in all methodologies except in that of between 0)2 and 0)4 produced up to a 90% change in the
Reimbert and Reimbert (1980) and in the "nite elements pressures. This parameter is inextricably coupled to the
model [Fig. 3(b)]. horizontal and vertical pressures ratio k in the interior of
the grain. Nevertheless, it must not be forgotten that the
ratios obtained near the wall di!er from those which are
4.2. ¹he material parameters in-uence obtained in the grain interior, meaning, in e!ect, a value
of the ratio at rest k :

The e!ect of the parameters in the "nite element results
are obtained in the following sections. k
l"  (2)
1#k

4.2.1. >oung1s modulus
As had been expected, this parameter hardly a!ected Using "nite element models to simulate silos, obtain-
the values of the wall pressures, although it did a!ect the ing experimental values of Poisson's ratio by means of
total grain consolidation. Its value should be carefully triaxial tests becomes a necessity. The average values and
considered in cases where the study of the silo is under- ranges of variation for the various types of stored mater-
taken when it is being progressively "lled in layers be- ials ought to appear in the methods of calculation.
cause in such cases the pressures can be a!ected. This In addition, Poisson's ratio has a notable e!ect on the
value is determined by means of an edometric test in way in which the material behaves. With high values of
order to obtain the edometrical modulus E and Poisson's ratio, it is di$cult for the material to achieve
K a state of stress such that it reaches the yield surface and,
a triaxial test to measure Poisson's ratio l. With these
values Young's modulus E is obtained by means of as a result, the behaviour in the interior of the stored
material will be elastic. Therefore, the use of elastic}plas-
1!l!2l tic behaviour is no longer a necessity if the Poisson ratio
E"E (1)
K 1!l is high. The values at which such e!ect may be obtained
ST AT IC AN D DY NA MI C SI LO L O A D S 303

are di$cult to predict since it depends on the silo size and


the material speci"c weight as well as on the behaviour
models and their corresponding parameters. In the situ-
ation in which this research was carried out (speci"c size
of silo, wheat as material and Drucker}Prager as model)
with a value for l of 0)2, a complete plasti"cation of the
material is produced. In such a situation, horizontal
pressures can increase by up to 50% in comparison with
the elastic situation and, as a result, low values of
Fig. 4. Saw-tooth model for describing dilatancy angle in
Poisson's ratio must always use an elastic}plastic behav- particulate materials
iour model.

4.2.5. ¹he wall friction coe.cient


4.2.3. ¹he internal friction angle
The wall friction coe$cient k is one of the parameters
This is one of the required parameters for the
employed in traditional theories and in international
Drucker}Prager behaviour model and one of those fre-
standards. This analysis has con"rmed previous "ndings
quently employed in traditional theories and standards.
that the greater the friction coe$cient, greater the hori-
With low values in this parameter, it is logical that
zontal pressure. There is a 30% increase when moving
plasti"cation of the stored material is produced more
from a coe$cient with a value of 0)25 to that of 0)5.
easily, something which implies a distinct increase in
Increasing the values of wall friction coe$cient, increases
horizontal pressures. Nevertheless, the values that lead
the level of plasti"cation and, as a result, accentuates the
to plasti"cation - around 223*are lower than usual in
di!erence in pressure by varying other parameters such
granular materials - around 253. This e!ect was observed
as the Poisson ratio or the internal friction angle.
using a value for Poisson's ratio of 0)3. Using lower
Poisson's ratio values, the material becomes plastic with
4.2.6. Bulk density
higher internal friction angle values.
No di!erences were found to exist between the results
of this research and those produced using traditional
4.2.4. ¹he dilatancy angle theories. The static pressures were proportional to the
The dilatancy angle is a new parameter that began to bulk density. However, the bulk density of the grain
be considered when behaviour models for grain of an could be varied during a simulated loading by using the
elastic}plastic type were considered in pressure cal- "nite element method.
culations of stored material. Dilatancy is de"ned as the
volumetric increment when a particulate material is sub- 4.2.7. Mesh size and distribution
jected to shearing. Dilation is related to the shear dis- It has been possible to demonstrate that with large
placement through the dilatancy angle u, which is the mesh sizes (of about 1 m) good results are obtained,
angle between the tangent of the shear plane and producing little di!erence compared to the "ner mesh
the horizontal shear surface (Fig. 4). Various authors sizes. However, the need for a "ner mesh in the joint
have related this phenomenon to that of overpressure between the silo and the hopper has been observed.
during discharge. Its value can vary between 03 and that Di!erences in pressure of up to 100%, using an element
of the internal friction angle, although in the bibliogra- of 1 m and another of 0)08 m, were discovered. In such
phy (Hardin et al., 1990; Zhang, 1994) values obtained cases, the use of a progressive mesh is recommended in
using agricultural products never exceeded 203. It can be the silo}hopper joint. Furthermore, it is essential to es-
deduced from the present research that there is practic- tablish that the friction at this point follows the direction
ally no in#uence of this value on vertical wall pressures in of the hopper, and not that of the silo wall, otherwise
the static position. The greatest static pressures were exaggerated distortion phenomena are produced (Fig. 5).
produced by an angle of 03, and with low values of
Poisson's ratio (0)3) and the angle of friction (223), the 4.2.8. =all thickness
pressure did not increase by more than 6%. Nonetheless, The results of the present research show that for the
its in#uence began to appear in hopper silos where in- purpose of pressure calculations, walls of reinforced con-
creases of up to 16% were produced in the horizontal crete behave exactly the same as would a perfectly rigid
pressures on the silo}hopper joint. In this case, the in- wall. Therefore, it is valid (in the application of the "nite
crease was produced wherever any dilatancy angle di!er- element method) to consider this assumption for the
ent from 03 was considered and varied little between the calculation of such pressures. It is not the same with
di!erent values that were tested. metal walls. A smaller wall thickness causes the stored
304 F. AY U G A E ¹ A ¸.

granular material. It was thought that the full silo in


a static position would respond according to Rankine's
active state of equilibrium and that once the grain began
to move, it would pass to a passive state of equilibrium
(Jenike & Johanson, 1968). Some authors later modi"ed
this simpli"cation and they took into account the
grain}wall friction and also the types of #ow during
discharge (Rotter et al., 1997; Aguado et al., 1997).
More recently, this overpressure has been attributed
to the dilatancy phenomenon which is seen as the cause
of the grain not displacing parallel to the rupture line
that provokes an increase in volume and, as a conse-
quence, an increase in the lateral pressures (Zhang et al.,
1994).
On the other hand, Jenike and Johanson attributed the
overpressures to the types of #ow during discharge
(Jenike & Johanson, 1968). Attaching strain gauges demon
strated that in the case of mass #ow the pressure peak
occurred at the silo}hopper joint the same as occurred in
the case of a static silo. However, in the case of funnel
#ow, the pressure peak was produced exactly where the
funnel for grain #ow started. No variation was observed
below this point owing to the fact that the grain against
the wall remained at rest in this area. This line of reason-
ing has not led to a method of predicting overpressures
during discharge.
The results of this research show that all these phe-
Fig. 5. Pressures with diwerent mesh sizes: , Eurocode; nomena are intimately related to overpressure during
, xnite element method with regular mesh and size of 0)5 m; discharge.
, progressive meshing The prime objective of this research was to analyse the
phenomenon of silo discharge using the "nite element
material to exert less pressure, due to radial de#ection. method as a tool. Models were developed based on those
Di!erences in pressure of up to 25%, using a #at-bottom previously created for static stored material incorporat-
silo model with an internal friction angle of 303 and ing a series of modi"cations derived from the following
a wall thickness of 1 mm, were discovered. However, if three basic hypotheses.
the material behaviour is elastic}plastic and total plasti"-
cation of the material is produced in the zone adjacent to (1) Maximum pressures
the wall pressures are not a!ected by wall material or Maximum pressures are produced during the initial mo-
wall thickness. ments of discharge (Drescher, 1978; Sugita, 1972). Dis-
charge is simulated in the present analysis by a small
displacement of the grain in the silo outlet. At the same
5. Structural behaviour of the silo during discharge time, a vertical cylindrical surface of contact elements
with "xed grain}grain friction was included in the model
It is known that the main structural problem of silos is in order to simulate the movement of granular material
the increase in horizontal pressures when discharge within the silo (Fig. 6).
starts. In earlier silo designs overpressures were not con- (2) ¹he existence of di+erent -ow types (mass or funnel)
sidered, even though as early as 1950s, it was fairly well This implies that the rupture lines within the granular
recognized that overpressures occur during emptying material can take di!erent directions. Thus, in models
(Caquot & Kerisel, 1956). This pressure increase has been developed for mass #ow, the material descends creating
measured both in model and full-scale silos. Nevertheless, friction against the silo wall. In models for funnel #ow,
disagreement still exists between the magnitude of these a new line of rupture is formed creating an angle, called b,
overpressures and their causes. with the horizontal axis (Fig. 6), and which, due to the
Traditionally, the increase in pressures has been axisymmetric character of the model, produces a funnel
attributed to a change in the state of equilibrium of the of stored material.
ST AT IC AN D DY NA MI C SI LO L O A D S 305

is produced. The uncertainty about the slide line position


in the latter case is due to the scarcity of research and the
still limited development of discrete element methods
(Rong, 1994; Rotter et al., 1997; Jofriet et al., 1997). By
this method the behaviour of the material is simulated
grain by grain and, as a result, it is not necessary to force
its movement.
Owing to this uncertainty about the slide line position,
it was agreed to retain this ambiguity, developing
a model which would solve the problem of di!erent
angles of inclination along the rupture line in the grain b,
record the data and, "nally, obtain a curve of the max-
imum pressures reached at each sectional height of the
silo wall.

6. Results obtained using discharge models

6.1. Pressures against the walls

For mass #ow, the pressure peak coincided with


silo}hopper joint [Fig. 7(a)]. Pressure distribution ob-
tained using funnel #ow models conformed with the
Fig. 6. Schematic discharge model; b, rupture angle
experimental observations by Jenike and Johanson
(1968) that the pressure peak coincided with the junction
(3) ¹he dilatancy e+ect of the rupture line in the grain and the vertical wall of the
This is simulated by means of the change in the friction silo [Fig. 7(b)].
direction (depending on the dilatancy angle) in the con- A pressure envelope curve was developed which rep-
tact element. It is assumed that the material is separated resents the maximum values of the di!erent curves ob-
following the line of rupture thus provoking an increase tained in the di!erent #ow conditions (Fig. 8). Di!erences
in the lateral pressures. in pressure of up to 100%, using an element of 1 m
Depending on the geometric characteristics of the silo (regular mesh) and another of 0)08 m (progressive mesh)
and the type of stored material, either mass or funnel #ow in the silo}hopper joint, were also discovered (Fig. 8).

Fig. 7. (a) Mass yow (regular mesh) pressures and (b) funnel yow pressures for a rupture angle (b) of 553, an internal friction angle of
223, a dilatancy angle of 193 and a bulk density of 9 kN m!3: , static model; , discharging model
306 F. AY U G A E ¹ A ¸.

Fig. 8. Final discharge model for an internal friction angle of 223,


a dilatancy angle of 193 and a bulk density of 9 kN m!3: ,
static model; , discharge for b"453; , discharge
for b"503; , discharge for b"553; , discharge
for b"603; , discharge for b"653; ,
discharge for b"703; , discharge for b"753; , en- Fig. 9. Eurocode and proposed model pressures comparison for
veloping curve for regular mesh; , enveloping curve for an internal friction angle of 223, a dilatancy angle of 193 and
progressive mesh a bulk density of 9 kN m!3: , proposed model using progress-
ive meshing; , Eurocode for mass yow; , Eurocode for
funnel yow

The values obtained using a progressive mesh agree with


those obtained by means of Eurocode 1, Part 4, except in 7. Conclusions
the hopper wall where they were lower (Fig. 9).
The ratio of horizontal to vertical pressures k was 12% Finite element models capable of simulating the be-
greater than in the static example, and was similar (ex- haviour of silo-stored granular material were developed,
cept in the hopper wall) to that of the reduced value both for the static example and for that of centralized
proposed by Eurocode 1, Part 4. discharge.
All of this justi"es the focus of the research and the The in#uence of the type of wall and the geometrical
validity of the models used. structure of the silo, as well as the di!erent parameters of
silo-stored granular material, were analysed with regard
to their e!ect on pressures, not only in the case of mater-
6.2. ¹he material parameters in-uence ial at rest but also while being discharged.
By means of these model values for pressure against
It is interesting to examine the e!ect on wall pressures the silo walls were obtained in mass #ow and funnel #ow.
of Poisson's ratio and of the dilatancy angle with regard Values obtained in mass #ow were similar to those pro-
to the discharge models. posed by other authors and calculations using the Euro-
The Poisson ratio a!ects the lateral pressures, al- code. Nevertheless, by using these models it is possible to
though notably less than in the static example. An in- get a more detailed analysis and better adaptation to
crease from 0)2 to 0)4 increase overpressures by 30% on particular situations. The models also allow an expla-
the hopper wall, 20% in the silo}hopper joint and 10% nation of the pressure distribution in funnel #ow, pheno-
on the vertical wall [Fig. 10(a)]. mena observed in the experimental trials of other
Unlike the static example, in the discharge model, an authors.
increase in the dilatancy angle from 5 to 193 produces an In these models, parameters such as that of dilatancy
increase of 30% in the lateral pressures. As a conse- angle or the Poisson ratio, which exert a great in#uence
quence, there is an increase in the coe$cient k [Fig. on lateral pressures, are taken into account. Dilatancy
10(b)]. angle acts during the discharge while the Poisson ratio
ST AT IC AN D DY NA MI C SI LO L O A D S 307

Fig. 10. (a) Poisson+s ratio inyuence on pressures for regular mesh, an internal friction angle of 223, a dilatancy angle of 193 and a bulk
density of 9 kN m!3: , Poisson+s ratio"0)2; , Poisson+s ratio"0)3; , Poisson+s ratio"0)4: (b) Dilatancy angle
inyuence on pressures for regular mesh, an internal friction angle of 223, a Poisson+s ratio of 0)3 and a bulk density of 9 kN m!3:
, dilatancy"53; , dilatancy"103; , dilatancy"153; , dilatancy"193

does so both during the discharge and in the static state. Caquot A; Kerisel J (1956). Traite Mecanique des Sols.
In spite of their great in#uence, these two parameters are [Treatise on soil mechanics.] Gauthier-Villars Editeur-
not taken into account in current guidelines. This is due Imprimeur-Librairie. Paris
DIN 1055 Part 6 (1987). Design loads for buildings. Loads in
to the fact that they were simply not considered in tradi- silo bins
tional methods of calculation and that, furthermore, very Drescher A (1978). Kinematics of the mass #ow of granular
few experimentally obtained values exist for agricultural material through a plane hopper Geo. 28(1), 27}42
materials. Drucker D C; Prager W (1952). Soil mechanics and plastic
analysis on limit design. Quarterly Applied Mathematics,
10(2), 157}165
Eurocode 1, ENV 1991-4. (1995). Basis of design and actions on
Acknowledgements structures. Part 4: Actions on silos and tanks
Guaita M (1995). CreacioH n de Modelos para la SimulacioH n de
The authors are grateful to the CICYT (Spanish Re- Silos por el MeH todo de los Elementos Finitos y AnaH lisis de
search and Technology Commission) for funding this los Empujes EstaH ticos del Material Almacenado. [Model
project (AGF97-1141). development for silo simulation using the "nite element
method and analysis of the static pressures of the stored
materials.] Tesis Doctoral. Universidad PoliteH cnica de Madrid
Hardin B O; Hardin K O; Ross I J; Schwab C V (1990). Triaxial
References compression, simple shear, and strength of wheat en masse.
Transactions of the ASAE, 33(3), 933}943
Aguado P J; Ayuga F; Guaita M (1997). Comparative evalu- Janssen H A (1895). Versuch uK ber Getreidedruck in Sillozellen.
ation of numerical methods for predicting #ow and stress [Experiments on grain loads in silo cells.] Zeischrift des
"elds in silos. Filling of a Silo. EPSRC DEMFEM Inter- Vesein Deutscher Ingenieure, 39, 1045}1049
national Collaboration CA-SILO Collaborative Action: Jenike A W; Johanson J R (1968). Bins loads. Journal of
WG5. Problem 1. Universidad PoliteH cnica de Madrid Group Structure Division, Proceeding of ASCE, 94(ST4), 1011-1041
Contribution Jofriet J C; Lelievre B; Fwa T F (1977). Friction model for
ANSYS (1998). User's Manuals for Revision 5.3. Vols. I, II, III, "nite element analyses of silos. Transactions of the ASAE,
IV and V. Swanson Analysis Systems Inc., Houston, USA 20(4), 735}744
308 F. AY U G A E ¹ A ¸.

Jofriet J C; Negi S C; Lu Z (1997). Parametric study of bin Rong G (1994). Discrete element modelling for #ow of partic-
loads using a hybrid numerical model. ASAE Meeting Pre- ulate materials in bins PhD Thesis. University of Guelph,
sentation. Paper no 97-4102. Minneapolis Canada
Maeda Y; Ishizaki S (1979). Analysis of cylindrical shells for Rotter M; Ooi J; Holst M; Zhong Z (1997). Comparative
design of steel silos. Journal Civil Engineering Design, 1(4), evaluation of numerical methods for predicting #ow and
325}354 stress "elds in silos. EPSRC DEMFEM International
Mahmoud M H (1975). Silage}silo interaction using material Collaboration CA-SILO Collaborative Action: WG5.
characterization and "nite element analysis. PhD Thesis. http://oats.civ.ed.ac.uk/research/silo/demfem/
Department of Civil Engineering, Ohio State University, Sugita M (1972). Flow and pressures of non-cohesive granu-
Columbus, OH lar materials in funnel #ow bin. ASME No. 72-MH-20,
Manbeck H B; Nelson G L (1975). Three dimensional constitut- pp 1}8
ive equations for wheat en masse. Transactions of the ASAE, Zhang Q; Britton M G; Xu S (1994). Using dilatancy angle to
18(6), 1122}1127 predict dynamic loads during discharge in bulk solid storage
Reimbert M; Reimbert A (1980). Pressures and overpressures in structures. Proceeding of Powder and Bulk Solids'94,
vertical and horizontal silos. International Conference on Rosemont, IL 383-390
Design of Silos for Strength and Flow, Powder Advisory Zhang Y (1994). Mechanical properties of soyabean and corn.
Centre, London MS Dissertation, University of Guelph