You are on page 1of 14

ARTICLE IN PRESS

Biosystems Engineering (2005) 91 (3), 335–348


doi:10.1016/j.biosystemseng.2005.03.012
SE—Structures and Environment

Analysis of Dynamic Discharge Pressures in Cylindrical Slender Silos with a Flat


Bottom or with a Hopper: Comparison with Eurocode 1
Pablo Vidal1; Manuel Guaita2; Francisco Ayuga3
1
Escuela Ingenierı́as Agrarias, University of Extremadura, Crtra. de Cáceres s/n 06071 Badajoz, Spain; e-mail of corresponding author:
pvidal@unex.es
2
E.P.S. University of Santiago de Compostela, Campus Universitario, 27002, Lugo, Spain; e-mail: guaita@lugo.usc.es
3
E.T.S.I. Agrónomos, Polytechnic University of Madrid, Ciudad Universitaria s/n, 28040, Madrid, Spain; e-mail: ayuga@cvr.etsia.upm.es

(Received 26 April 2004; accepted in revised form 8 March 2005; published online 11 May 2005)

Few models simulate the discharge of flat-bottom silos by using finite elements. This paper presents a new
dynamic model for silo discharge using the Drucker–Prager plasticity model. The model was used to study
discharge in mixed flow and mass flow silos with rigid walls. Various silos were analysed, and pressures under
different flow patterns were compared. Results were also compared with Eurocode 1—part 4 on actions in
silos and tanks, which is the standard used in the European Union. Overpressure coefficients due to discharge
were obtained in all cases. In the case of flat-bottom silos, overpressures occurred in the lower part of the silo.
The values obtained for overpressures were lower than the values obtained in the case of mass flow hopper
silos. The effects of wall friction and outlet radius were also analysed.
r 2005 Silsoe Research Institute. All rights reserved
Published by Elsevier Ltd

1. Introduction mass flow, i.e. a discharge with total movement of the


stored content. Conversely, in flat-bottom silos, a part
Silos are important structures for bulk storage of a of the material at the bottom of the silo remains at rest
variety of produce. The silo wall must withstand the while the rest of the material flows. This type of flow is
pressures exerted by the material stored in the silo. The termed mixed flow.
main problem in silo design is that the load of the stored The tools used in silo research include experimental,
material must be precisely determined so that the analytical and numerical techniques. Experimental
appropriate wall thickness can be calculated. analyses are complex, and their main drawbacks are
The interaction between the silo structure and the high costs and difficulty to generalise results. The
stored material is of interest because the knowledge of application of analytical methods is very complex due
loads over the wall of silos is scarce, especially in flat to the nature of the problem being solved. Therefore,
bottom and eccentric silos. These loads include static numerical methods have become usual in research in
pressures, which act on silo walls when the material actions on silos. Moreover, these methods are cheap and
remains at rest; and discharge pressures, which are allow performance of parametric studies. In this work,
higher at the initial stages of discharge. the finite element method (FEM) is applied to the study
Circular section silos of central discharge are known of flat-bottom silos as compared with hopper silos.
to minimise the material used in silo manufacture. The
most widely used silos are silos with concentric hoppers,
and flat-bottom silos. Flat-bottom silos are easier to 2. Literature review
build. However, research in the last few years has
focused on cylindrical silos with conical hoppers, while Finite element analysis of silos requires the use of an
flat-bottom silos have remained in the background. appropriate law to describe the behaviour of the
Conical hoppers with an adequate inclination enable material. The use of elastic models is not recommended
1537-5110/$30.00 335 r 2005 Silsoe Research Institute. All rights reserved
Published by Elsevier Ltd
ARTICLE IN PRESS
336 P. VIDAL ET AL.

Notation

am variability parameter of the friction coefficient K stiffness matrix


dc characteristic dimension of inside silo cross-section M mass matrix
(diameter in circular sections), m ü nodal acceleration vector
f load vector u nodal displacement vector
h height of the silo, m f internal friction angle, deg
hc height of vertical-walled segment of silo from the
transition to the equivalent surface, m

because elastic models cannot consider the behaviour of In Rotter et al. (1998), a summary of the results of a
the granular material correctly. During the last 15 years, collaborative study on the discharging of a silo
in the field of silo analysis using numerical methods, the conducted by different authors was presented. The
Drucker–Prager model of behaviour (Drucker & Prager, study investigated the discharge of a silo with either a
1952) has been widely used (Link & Elwi, 1990; Dı́ez & hopper bottom or a flat bottom, with both rigid walls
Godoy, 1992; Meng et al., 1997; Ayuga et al., 2001a, and plane stress. The results for the flat-bottom silo
2001b; Martı́nez et al., 2002; Guaita et al., 2003; Wojcik showed considerable disparity. Two out of the three
et al., 2003; Goodey & Brown, 2004). results published showed a negative increase of silo
Different models of behaviour of the material were pressures during discharge in two-thirds of the silo wall,
used by other authors to study silo discharge. These and increases in the lower part of the silo. It should be
studies showed a large disparity in results (Rotter et al., regarded that maximum values occur at different heights
1998). However, the Drucker–Prager criterion some- only in the zone where the material is at rest, which
times offers the best approximations, if compared with suggests that the position of the failure boundary can
experimental results (Meng et al., 1997; Goodey et al., take different values.
2003). Sanad et al. (2001) expanded the analysis conducted
Studies on the analysis of flat-bottom silos are far less by Rotter et al. (1998) and provided more data about
common than studies on hopper silos. Only a few works silos with a flat bottom. This study compared the results
are published in international literature (Munch-Ander- obtained for pressures by using the discrete element
sen & Nielsen, 1990; Ragneau & Aribert, 1995; Nielsen, method (DEM) and the FEM. Differences in results
1998; Rotter et al., 1998; Sanad et al., 2001; Zhong et al., were attributed to the characteristics inherent to both
2001; Guines et al., 2001; Goodey & Brown, 2004). methods. The FEM studies conducted by various
Munch-Andersen and Nielsen (1990) showed the research groups used different models for the behaviour
filling process had no influence in the type of discharge of the material. The first research group in this study
flow, which was always mixed in the case of wheat in considered the stored material as a non-linear elastic
slender silos. Furthermore, mixed flow in slender silos porous solid with the Mohr–Coulomb failure criterion.
was found to produce higher pressures during discharge. The study analysed discharge by applying displacements
Marked differences were detected between test silos and to the material located at the outlet. The second research
real silos due to scale effects. group studied the emptying of these silos by applying
Ragneau and Aribert (1995) used models of beha- Boyce’s non-linear elastic model with the Drucker–Pra-
viour of the stored material taken from soil mechanics ger criterion for the plastic domain. Discharge was
and discharge was defined by a steady-state quasi-static defined by a steady-state quasi-static condition, pre-
condition. Firstly, Boyce’s hyperelastic model was viously determined by a failure boundary at 45 þ f=2 to
considered in the elastic domain, and Wilde plasticity the horizontal plane, which means 6251 in the case of
model was applied when elastic deformations were the material proposed by the study. The third group
overcome. This model was applied to silos with either considered a dynamic analysis using a viscoplastic
a flat bottom or a hopper bottom. In flat bottom silos, model for silo filling and emptying. All of the groups
the failure boundary was located at 45 þ f=21 above the reported an increase in pressures in the lower part of the
bottom of the silo, where f is the internal friction angle. silo wall. In Sanad et al. (2001), the discharge of a flat-
Nielsen (1998) stated that the failure boundary had a bottom silo was analysed by seven international research
stable position, but his study did not offer geometrical groups by using the DEM and following a mixed flow
values. pattern. The failure boundary had a mean value of
ARTICLE IN PRESS
ANALYSIS OF DYNAMIC DISCHARGE PRESSURES 337

inclination to the horizontal plane equal to 90-f; question the effects of a hopper or a flat bottom on the
specifically 551. This value ranged 601 to 481 for a cylindrical wall. Previous works by this research group
material with an internal friction angle f of 351. showed the influence of hopper eccentricity on static
An interesting study using plastic pellets (Zhong et al., pressures and silo discharge (Ayuga et al., 2001b; Guaita
2001) showed how the failure boundary remained stable et al., 2003; Vidal, 2003). The French standard AFNOR
when the angle of inclination to the vertical plane was P22 630 (1992) takes into consideration the influence of
equal to the internal friction angle. A scarcely clarifying a flat bottom or a hopper in the study of pressures
case was proposed for a material with a value for f of exerted on the wall.
401. The discharge height was larger than the silo
diameter.
Guines et al. (2001) used the non-linear Hujeux’s
3. Purpose
model (Aubry & Hujeux, 1982) for the elastic domain,
and the Cam-Clay plasticity model to simulate silo
The main purpose of this study is to model the
discharge under non-dynamic conditions. This model
dynamic discharge of a cylindrical mixed flow silo (silo
was applied for the calculation of a flat-bottom silo with
with a flat bottom), and of a cylindrical mass flow silo
the same failure boundary as Ragneau and Aribert
(silo with a hopper), by applying FEM. The surface-to-
(1995). At the failure boundary, external forces equiva-
surface model was considered for grain-wall contact,
lent to the forces at the other side of this boundary were
and the Drucker–Prager model was used for the
applied in order to transfer the stresses of the material,
behaviour of the material.
as shown in Guines et al. (2001). It should be noted that
Other objectives are: to analyse the results of static
in Ragneau and Aribert (1995) and Guines et al. (2001)
and dynamic pressures in both silos; to estimate the
the highest overpressures were obtained in the zone
overpressure coefficient; to compare the results with
where material is at rest.
those proposed by the last draft of the Eurocode; and to
In a recent study, Goodey and Brown (2004) analysed
analyse the influence of the friction coefficient and the
the effect of base boundary conditions of cylindrical
influence of the outlet radius.
silos to detect the origin of the increase in pressures at
the base of the silo. The analysis was conducted by
studying silo-layered fill, but it did not tackle the
problem of discharge. The material was simulated by 4. Simulation details
using ABAQUS software with the Drucker–Prager
plasticity criterion and surface-to-surface contact ele- In order to conduct the analysis, the reference silo
ments. The contact algorithm used was not specified. used was a cylindrical silo with a flat bottom, or with a
Another problem associated with FEM analysis of hopper with an inclination of 651, a radius of 2 m, a
flat-bottom silos is the grain-wall friction model. height of 14 m and an outlet radius of 06 m, as shown in
Friction models evolved from Jofriet et al. (1977) until Fig. 1.
present. Jofriet proposed a node-to-node friction model, Table 1 shows the parameters defined for the stored
widely used until recently (Meng et al., 1997). These material. All the values used in this study were analysed
models evolved into the model used for more modern experimentally by Moya et al. (2002) and Moya (2001),
contact elements: the surface-to-surface model. The using the Spanish Horzal wheat variety. In addition to
surface-to-surface model is much more realistic than the elastic parameters necessary for the elastic part of
the previous models and was developed in several the behaviour of the material—Young’s modulus and
research works (Cescotto & Charilier, 1992; Cescotto Poisson’s ratio—cohesion, internal friction angle, and
& Zhu, 1994; Simo & Laursen, 1992). dilatancy angle are required to define plastic behaviour
In addition, the present paper considers the proposal with the Drucker–Prager plastic model. These para-
of the last draft of Eurocode (prENV 1991-4. Eurocode meters were tested in Moya et al. (2002) and Moya
1, 2003), which estimates discharge pressures by apply- (2001). The results reported in the mentioned studies are
ing an overpressure coefficient to the analytic solution of included in this work. In addition to the parameters that
static pressures. According to the future Eurocode on describe the behaviour of the material, specific weight
actions in silos, a silo is considered a slender silo when and friction coefficient were analysed. In this case, the
the ratio of the height to the diameter of the vertical friction coefficient corresponds to friction between grain
section hc =d c X2; and a squat silo when 04ohc =d c p1: and a metal wall.
In this standard, solutions are offered for the calculation According to the mentioned dimensions and values
of pressures on cylindrical walls, and for the calculation for the material, and from the classification contained in
of hoppers or flat bottoms. Yet, the standard does not the Eurocode, the silo can be classified as a class 2
ARTICLE IN PRESS
338 P. VIDAL ET AL.
14 m

65º

65º
0.6 m 2m 65º
Fig. 1. Geometry and dimensions of the silo

Fig. 2. Meshes of axisymmetric elements used to simulate


discharge
Table 1
Parameters of the material

Parameter Value
Grain–wall friction follows the classical Coulomb
Specific weight 84 kN/m3 model, which is sufficiently valid for this analysis. The
Young’s modulus 51 kPa contact element used was a one-dimensional, two-node
Poisson’s ratio 03 element, paired with another two-node contact element,
Grain–wall friction coefficient 02
Internal friction angle 251 with two integration points at the grain-wall contact
Apparent cohesion 3 kPa lines of the elements. These two elements were the
Dilatancy angle 1761 ANSYS elements Targe169 and Conta171, which are in
accordance with the characteristics of the contact
proposed by Cescotto and Charilier (1992).
In the analyses, the resolution algorithm used for the
slender silo with a steep hopper. Silos of class 2 have a contact was ‘penalty function+Lagrangian multiplier’.
capacity exceeding 100 000 kg. Results showed no differences between the calculations
An axisymmetrical FEM analysis was used to make made by applying this method or by applying just
calculations, considering a rigid wall. The elements that penalty function.
represent the granular material are four-node elements To solve silo discharge, a transient dynamic analysis
with two degrees of freedom—axial displacement and was conducted, based on the following equation:
radial displacement. Figure 2 shows the finite element
mesh for both cases, generated with four-node elements. M u€ þ Ku ¼ f (1)
ARTICLE IN PRESS
ANALYSIS OF DYNAMIC DISCHARGE PRESSURES 339

where: M is the mass matrix; K is the stiffness matrix; ü


is the nodal acceleration vector; u is the nodal
displacement vector; and f is the load vector.
A non-linear analysis was performed, divided into two
stages. The first stage was silo filling, conducted by
applying a static analysis. The second stage was carried
out after eliminating the constraints at the silo outlet, by
applying a transient dynamic analysis.
The discharge of all silos was studied for 15 s. In this 45º
interval, discharge pressures were higher than static 35º
pressures, gradually increasing until a maximum value
was reached. From that moment, discharge pressures
started decreasing. The time of discharge analysis was (b) (c)
similar to the intervals used by Wojcik et al. (2003). The
results presented for dynamic pressures corresponded to
the calculation of the envelope of the different discharge 65º
curves obtained for different times of analysis.
To solve the analysis, the ANSYS programming 55º
language, a finite element software was used. The use
of these systems enables easier diversification of results
among scientists around the world, and includes the
improvements in numerical methods in a faster and
more efficient manner. (d) (e)
In the hopper silo, variations in the calculation of
static and dynamic pressures only modify the boundary
conditions. The elimination of constraints at the bottom
of the hopper resulted in a displacement of all the (a)
material towards the silo outlet, as explained above.
The discharge of the flat-bottom silo was studied Fig. 3. Models for flat-bottom silos and angles to horizontal
based on the results obtained by Jenike and Johansson plane: (a) axisymmetric view of different failure boundaries;
(b) failure boundary at 351; (c) failure boundary at 451; (d)
(1968), considering that a failure boundary occurs failure boundary at 551; (e) failure boundary at 651
during discharge. The failure boundary adopts different
positions at different stages of the discharge process.
Moreover, according to the mentioned studies (Munch- stored wheat. These models fully represent the possible
Andersen & Nielsen, 1990; Ragneau & Aribert, 1995; positions of the failure boundary. Different inclination
Nielsen, 1998; Rotter et al., 1998; Sanad et al., 2001; angles were considered in the failure boundary: 351 [Fig.
Zhong et al., 2001; Guines et al., 2001), two zones can be 3(b)] 451 [Fig. 3(c)] 551 [Fig. 3(d)] and 651 [Fig. 3(e)]. The
distinguished in the silo: a zone where the material is in proposed values were considered sufficient if compared
motion, and a zone where the material is at rest, divided with the value of the internal friction angle f of 251.
by the failure boundary. This failure boundary appears As in the case of the hopper silo, the material stored in
not to vary greatly in position, at least in flat-bottom the flat-bottom silo was defined assuming friction with
silos. Moreover, the location of this boundary cannot be the silo wall. The wall was considered rigid, which is a
shown in a precise manner because there are few studies common simplifying hypothesis that helps to decrease
on the discharge of flat-bottom silos, and great the convergence time of the results.
disparities are found among authors.
In order to consider this effect, several models of flat-
bottom silos were developed in this study with a failure 5. Results and discussion
boundary inside the granular material, represented by a
contact zone, Fig. 3(a). The contact zone represented the The results presented below analyse the variation in
friction between static and dynamic parts during static and discharge pressures, and the overpressure
discharge. The separation between these two parts of coefficient between static and dynamic pressures.
the granular material was defined by using surface-to- All the values for pressures correspond to horizontal
surface contact elements with a coefficient of friction pressures and are expressed in kPa. The results are
equal to the tangent of the internal friction angle of the compared in all cases with the last draft of Eurocode 1.
ARTICLE IN PRESS
340 P. VIDAL ET AL.

5.1. Silo with failure boundaries at 651, 551, 451 and 351 occurs in the zone of the failure boundary (the last 3 m
to the horizontal plane of the silo wall). This increase reaches a value of
8354 kPa, then decreases to about 64 kPa. A strong
First, the static and discharge pressures for a flat- increase in the dynamic pressure occurs at the joint of
bottom silo with a failure boundary at 651 to the the wall and the bottom of the silo. The maximum
horizontal plane are presented. The model developed in pressure is obtained at the silo bottom in accordance
this study shows great stability and ease of convergence. with Ragneau and Aribert (1995), Rotter et al. (1998),
Figure 4(a) shows both static and dynamic pressures. and Guines et al. (2001), Goodey and Brown (2004).
The silo is represented on the left part of the figure. Static The silos used in the mentioned studies were flat-bottom
pressures show a similar distribution to the distribution silos of rectangular section, not circular, except in
established by Janssen equation. The values presented are Goodey and Brown (2004). It should be noted that
lower than the values proposed by Eurocode for static pressures in rectangular silos are different at the edges
pressures in the cylindrical part of the silo. Static than at the centre of the wall; on the contrary, pressures
pressures produce stresses on the material, which serve in axisymmetric cylindrical silos are constant around the
as the reference to consider the initial moment of wall.
discharge and eliminate constraints at the outlet. The overpressure coefficient is analysed at different
By analysing displacements of the stored material heights of the silo [Fig. 4(b)]. This coefficient takes a
under the failure boundary, it was observed that— mean value of 125 above the failure boundary, while the
unlike the rest of the stored material—this part of the mean value is 2 under the failure boundary, which
material remained at rest. corresponds to the material at rest. The overpressure
Within discharge pressures, two important zones coefficient locally increases up to a value of 262 in the
divided by the effective transition can be distinguished. zone of the effective transition, and to a value of 314 at
The upper part, in which all the material contacting the the silo bottom.
wall is in motion, and the lower part, where the highest In Rotter et al. (1998), the overpressure coefficients
horizontal discharge pressures occur. A strong increase can be estimated indirectly because there are no values

14 14

13 13

12 12

11 11

10 10

9 9
Height h, m

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

0 0
0 −20 −40 −60 −80 −100 −120 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Horizontal pressure, kPa (b) Overpressure coefficient
Fig. 4. (a) Horizontal pressures in kPa in a flat-bottom silo with a failure boundary at 651 versus height: —, static pressure; TT,
dynamic pressure; (b) overpressure coefficient versus height
ARTICLE IN PRESS
ANALYSIS OF DYNAMIC DISCHARGE PRESSURES 341

published for the overpressure coefficient in this type of observed in discharge pressures in Fig. 4(a), and in the
silos, at least as far as the authors know. The analysis of silo fill reported by Goodey and Brown
overpressure coefficients reach values lower than 1 in (2004). Moreover, if the overpressure coefficients
the zone above the failure boundary, which appears to obtained for the non-dynamic situation in Fig. 5(b) are
contradict experimental results (Munch-Andersen & compared with the dynamic situation in Fig. 4(b),
Nielsen, 1990). These results reveal scarce knowledge notably lower values are observed for the non-dynamic
of this type of silos. model than for the dynamic model.
In agreement with Goodey and Brown (2004), The failure boundary of the second dynamic model
Fig. 4(a) shows a pressure peak in horizontal pressures was located at an inclination of 551 to the horizontal
at the end of the silo wall. If the silo discharge is plane. The zone with the material at rest reached a
analysed by applying displacements to the stored height of 2 m on the wall.
material at the outlet, the pressure peak disappears The analysis of pressures in this silo [Fig. 6(a)] shows
[Fig. 5(a)]. Goodey and Brown (2004) make an that the values for static pressures are very similar to the
interesting analysis of the origin of such pressure peak pressures obtained in the previous case because the
and conclude that granular materials with high values of failure boundary does not have any effect on static
Poisson’s ratio show pressure peaks at the silo bottom pressures. The results were also verified by comparing
because vertical pressure is high. From among the the stresses of the stored material during silo filling with
materials analysed, wheat is the material that generates the same silo without the failure boundary.
the highest pressure peak at the silo bottom. Without The shape of discharge pressures in a silo with the
conducting experimental analyses, the comparison of effective transition at 551 to the horizontal plane [Fig.
results in Fig. 4(a) and in Fig. 5(a) could suggest that this 6(a)] is similar to the shape of the pressures obtained in
pressure can be more easily detected when the strain the situation described above. The same zones can be
history of the material is considered. This hypothesis differentiated. In the upper part, the results are similar
would account for the fact that no pressure peak occurs in shape to static results, but the values are higher than
in filling pressures in Fig. 4(a), while the pressure peak is the values obtained at rest. During discharge, pressures

14 14

13 13

12 12

11 11

10 10

9 9
Height h, m

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

0 0
0 −20 −40 −60 −80 −100 −120 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Horizontal pressure, kPa (b) Overpressure coefficient
Fig. 5. (a) Horizontal pressures in kPa for a non-dynamic analysis in a flat-bottom silo with a failure boundary at 651 versus height:
—, static pressure; TT, discharge pressure; (b) overpressure coefficient versus height
ARTICLE IN PRESS
342 P. VIDAL ET AL.

14 14

13 13

12 12

11 11

10 10

9 9
Height h, m

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

0 0
0 −20 −40 −60 −80 −100 −120 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Horizontal pressure, kPa (b) Overpressure coefficient
Fig. 6. (a) Horizontal pressures in kPa in a flat-bottom silo with a failure boundary at 551 versus height: —, static pressure; TT,
dynamic pressure; (b) overpressure coefficient versus height

to the wall increase and take a mean value of 73 kPa, Static pressures are identical to the pressures obtained
and two local values of 9235 and 100 kPa. in previous cases, as expected. The values for discharge
The overpressure coefficient takes different values pressures at the part of the silo located above the failure
[Fig. 6(b)] because dynamic pressures increase and static boundary are similar to the values obtained in previous
pressures remain the same as in the previous case. In the models. Under the effective transition, the values for
zone of the wall, above the failure boundary, this pressures in the intermediate zone under the failure
coefficient takes a mean value of 13, which is slightly boundary are similar among them, with a mean value of
higher than in the previous case. In the dead zone of the around 80 kPa, and local peak pressures of 100 kPa.
silo, the overpressure coefficient increases, and reaches In the previous model, dynamic pressures showed a
values higher than 2 in all cases. A mean value of 213 is tendency to increase. In accordance with this tendency,
obtained between a height of 05 and 15 m [Fig. 6(b)]. the overpressure coefficient [Fig. 7(b)] increases, reach-
The values for local increases in pressure are higher than ing values of 235 at the intermediate zone, and
in the previous case, as revealed by the coefficients of maximum local values of 3 and 33. These values are
overpressure obtained at these points, with values of mainly caused by the low values of static pressures in
288 and 33. these zones of flat-bottom silos.
The third dynamic model studied has an inclination The last dynamic model had an inclination of the
of the failure boundary of 451 [Fig. 7(a)], which effective transition of 351 to the horizontal. Only the
corresponds to a height of 14 m on the wall. This first 07 s of discharge are presented. As expected, a
model does not show difficulties of convergence, but model with this inclination in the failure boundary
more calculation time is required to reach a solution shows great difficulties for slipping of particles. From
than in the previous two cases. The calculation time for the first 07 s onwards, the values obtained for pressures
discharge pressures is also 15 s. This value is sufficient must be disregarded. In addition, convergence is
because the value of pressures decreases from that difficult, and great computational efforts are required
moment. to reach a discharge time of 13 s. This result is justified
ARTICLE IN PRESS
ANALYSIS OF DYNAMIC DISCHARGE PRESSURES 343

14 14
13 13
12 12
11 11

10 10

9 9
Height h, m

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

0 0
0 −20 −40 −60 −80 −100 −120 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Horizontal pressure, kPa (b) Overpressure coefficient
Fig. 7. (a) Horizontal pressures in kPa in a flat-bottom silo with a failure boundary at 451 versus height: —, static pressure; TT,
dynamic pressure; (b) overpressure coefficient versus height

by the value of the internal friction angle of wheat, reveals the overpressures caused by different effective
which is 251. transitions in the studied silos. The values of horizontal
static and dynamic pressures are presented in compar-
ison with the results of the calculation of horizontal
5.2. Envelope of results of flat-bottom silos static and discharge pressures using Eurocode 1 part 4
or Rotter (2001). These results were obtained taking into
The proposed models show two local increases in the consideration only the cylindrical wall. Eurocode 1 part
pressure exerted on the wall, at the effective transition 4 does not consider any influence on the part of the silo
and at the silo bottom. Literature review revealed wall located under the cylindrical wall.
proposals to avoid local pressure increases at the
silo–hopper joint (Keiter & Rombach, 2001). Other
studies, such as couple FEM-DEM study using discrete 5.3. Results of mass flow silo and comparison with flat-
elements at the transition (Lu et al., 1997; Jofriet et al., bottom silo
1997), take into consideration this increase in pressure.
No fully verified criterion is currently available In order to verify the models generated in this work,
according to which real silos and test silos coincide in the results obtained for flat-bottom silos are compared
the position of the failure boundary. The failure with the results obtained for a mass flow silo with a
boundary seems to adopt an inclination close to 90-f hopper at an inclination of 651 to the vertical. Accord-
to the horizontal plane, but there is no certainty about ing to previous models, the wall is completely rigid.
the truth and fixedness of this value during discharge. In Figure 9(a) shows the results of horizontal static and
order to consider these effects, Fig. 8 shows the result of dynamic pressures compared with horizontal pressures
the discharge envelope for silos with an inclination of of Eurocode 1 part 4. Walker’s theory (Walker, 1966)
the effective transition of 451, 551, and 651. The result was used to estimate pressures in the hopper. Like in all
ARTICLE IN PRESS
344 P. VIDAL ET AL.

14 overpressure coefficient is higher than on the wall, and


takes values ranging 15–2.
13 Figure 10 shows the comparison between horizontal
pressures in a silo with a 651 hopper and the same silo
12 with a flat bottom and the failure boundary at 651. This
figure compares pressures under mass flow and mixed
11 flow conditions. In the vertical wall of the hopper silo,
the values for static and dynamic pressures are similar.
10
The highest differences among static pressures are
9
recorded at the hopper. In contrast, the values are
closer during discharge, and pressures in the flat-bottom
8 silo are lower than pressures in the hopper silo. The
Height h, m

situation is different in the lower part of the silo, in


7 which a decrease in pressures occurs at the outlet of the
hopper silo.
6

5 5.4. Influence of the friction coefficient

4 The friction coefficient is a very important parameter


in the calculation of loads on silos. Although this
3 parameter is usually considered constant, it varies with
wall roughness throughout the useful life of the silo.
2
Therefore, the analysis of the influence of this parameter
is relevant.
1 Annex E of prENV 1991-4. Eurocode 1, (2003) and
Rotter (2001) include the properties of the different
0
0 −20 −40 −60 −80 −100 −120 materials usually stored in silos. A modification factor is
Horizontal pressure, kPa adopted depending on the variability of each parameter,
according to a normal distribution (Rotter, 2001). The
Fig. 8. Envelope of horizontal discharge pressures in a flat- modification factor allows the calculation of the
bottom silo compared to pressures according to Eurocode 1 in
kPa versus height: —, static pressure according to finite element maximum and minimum values of each parameter by
method (FEM); TT, dynamic pressure according to FEM; multiplying and dividing the mean value by this factor.
, static pressure according to Eurocode 1; , dynamic For wheat, the variability parameter am takes a value of
pressure according to Eurocode 1 116. As the friction coefficient assumed in this case is
02, the influence of the friction coefficient is studied for
a maximum value of 0232 and a minimum value of
0172, obtained from applying am to the coefficient of
the previous cases presented in this work, an interval of friction of Spanish Horzal wheat.
15 s of discharge is analysed. Within this interval, Figure 11 shows the results obtained from modifying
discharge pressures reach a maximum and then the this parameter in the envelope of pressures for silos with
pressures decrease. Figure 9(a) shows that the values for failure boundaries at 651, 551 and 451. In calculations,
horizontal static and dynamic pressures are always the coefficient of friction was maintained constant at the
lower than the values proposed by Eurocode 1 part 4. failure boundary, which depends on the internal friction
Pressures increase at the transition, as in the majority of angle of Spanish Horzal wheat.
the models that use FEM. The mean discharge pressure As was expected according to Janssen law, when the
in the upper half of the hopper obtained a value of friction coefficient decreases from 02 to 0172, the value
80 kPa. of static and dynamic horizontal pressures increases at all
Figure 9(b) shows the overpressure coefficient. As in heights of the silo. Consequently, the highest pressures
experimental studies, discharge pressures are higher are obtained with this value of the friction coefficient.
than filling pressures at all heights of the silo, suggesting The largest difference in dynamic pressures is obtained in
an overpressure coefficient higher than 1. The over- the zone of the wall above the failure boundaries,
pressure coefficient takes a mean value of 12 between a considered with average pressure increases of 488%.
height of 4 and 11 m on the silo wall. In the hopper, the These values contrast with the average increase of 365%
ARTICLE IN PRESS
ANALYSIS OF DYNAMIC DISCHARGE PRESSURES 345

14 14

13 13

12 12

11 11

10 10

9 9
Height h, m

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

0 0
0 −20 −40 −60 −80 −100 −120 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Horizontal pressure, kPa (b) Overpressure coefficient
Fig. 9. (a) Horizontal pressures in a silo with a 651 hopper by using finite element method (FEM) and Eurocode 1 in kPa versus
height: , horizontal static pressure according to FEM; , horizontal dynamic pressure according to FEM; ,
horizontal static pressure according to Eurocode 1; ,horizontal dynamic pressure according to Eurocode 1; (b) overpressure
coefficient according to FEM versus height

in the lower part of the silo, where the different failure affects the quantity of material stored under the failure
boundaries of the silo are located. The dependence of boundary. If the outlet is wide, a larger quantity of
pressures on the friction coefficient is weaker in the lower material flows through it and a smaller quantity of
part of the silo because friction at the failure boundary material remains at rest during silo discharge. In this
remains unaltered, as confirmed by the fact that static section, the influence of the outlet radius on static and
pressures vary uniformly at all heights of the silo. dynamic pressures in the reference flat-bottom silo is
The analysis of dynamic pressures for an increase in analysed. Three outlet radius values were analysed: 05,
the friction coefficient from 02 to 0232 reveals a 06 and 07 m.
decrease in pressures. Like in the previous case, the zone As the size of the outlet aperture varies, and because
above the material at rest shows an average decrease of different inclinations were assumed for failure bound-
396%, while the zone under the failure boundary shows aries of 651, 551 and 451, the height under the failure
an average decrease of 318%. In this case, friction at the boundary is different for each outlet radius. The largest
failure boundary is constant. Therefore, the influence of difference between the height of the effective transition
the friction coefficient in the zone where the highest in the silo with outlet radius 05 m and the height of the
pressures occur is not as important as in the case of effective transition in the silo with outlet radius 07 m is
conical hopper silos, in which the variation of grain–wall not relevant, specifically 044 m. For better understand-
friction affects the whole height of the silo. ing results, Fig. 12 plots pressures versus the height of
the ‘reference silo’, without considering the mentioned
difference in height. Figure 12 shows static and dynamic
5.5. Influence of the outlet radius pressures for the three values of the outlet radius
analysed.
Outlet aperture determines the quantity of stored The effect of absorption of horizontal pressures by the
material discharged. In this type of silo, outlet aperture material at rest increases as the outlet radius decreases.
ARTICLE IN PRESS
346 P. VIDAL ET AL.

14 14

13 13

12 12

11 11

10 10

9 9
Height h, m

Height h, m
8

7 7
6
6
5
5
4
4
3
3
2
2
1
1
0
0 −20 −40 −60 −80 −100 −120 0
Horizontal pressure, kPa 0 −20 −40 −60 −80 −100 −120
Horizontal pressure, kPa
Fig. 10. Horizontal pressures in kPa for a silo with a 651 hopper
or a flat-bottom silo with a failure boundary at 651 to the Fig. 11. Influence of the friction coefficient on the envelope of
horizontal plane versus height: , horizontal static horizontal pressures of the flat-bottom silo in kPa: J, horizontal
pressure in a hopper silo; , horizontal dynamic pressure static pressure with friction coefficient 0172; , horizontal
in a hopper silo; —, horizontal static pressure in flat-bottom silo; dynamic pressure with friction coefficient 0172; —, horizontal
TT, horizontal dynamic pressure in flat-bottom silo static pressure with friction coefficient 02; TT, horizontal
dynamic pressure with friction coefficient 02; W, horizontal
static pressure with friction coefficient 0232; , horizontal
dynamic pressure with friction coefficient 0232

In the zone immediately under the effective transi-


tion, differences in horizontal pressures are scarce.
Differences are larger near the outlet, because the 6. Conclusions
width of the material at rest increases when the
outlet radius is 05 m, and decreases when the outlet This study presents an analysis of the results of
radius is 07 m. In the zone under the effective dynamic silo discharge, considering flat-bottom silos,
transition, and for an outlet radius of 06 m, pressures the Drucker–Prager model of behaviour of the stored
show an average decrease of 721% because the material, and grain–wall surface-to-surface contact. The
outlet radius is 01 m smaller. However, if the outlet analysis was performed by using ANSYS software.
radius is 01 m larger, pressures show an average In flat-bottom silos, the values for static pressures are
increase of 543%. In accordance with Goodey and distributed according to a shape similar to the shape
Brown (2004), a larger outlet radius—07 m— deduced from Janssen theory.
causes higher vertical pressures because a larger In the case of cylindrical silos with a flat bottom,
quantity of material is displaced. Under these circum- increases in pressures occur during discharge. These
stances, an increase in pressures at the silo bottom can overpressures occur mainly in the lower part of the silo.
be expected. During discharge of a cylindrical silo with a flat bottom,
ARTICLE IN PRESS
ANALYSIS OF DYNAMIC DISCHARGE PRESSURES 347

14 15–2. In flat-bottom silos, the value for the overpressure


coefficient on the wall is 15, while in the hopper, the
13 maximum mean value of the coefficient is 23, with local
increases that reach values of 3 and 33.
12 Discharge pressures in a flat-bottom silo that behaves
in mixed flow are lower than discharge pressures in a
11 mass flow hopper silo, considering that the failure
boundary in the flat-bottom silo is equal to the
10
inclination of the hopper. These results are induced by
9
the effect of the material that is at rest.
The decrease in the coefficient of friction between the
Height h, m

8 stored material and the wall causes increases in


discharge pressures that, according to the variation
7 range reported in Eurocode 1, must not exceed 488%.
Increases in such coefficient produce decreases in
6 maximum pressure of 396%. These differences were
observed at the effective transition.
5 The increase in the size of the outlet radius causes
increases in horizontal pressures because the effect of the
4 material at rest is weaker and the quantity of the
material in movement is larger. Maximum pressures of
3 120 kPa were obtained at the bottom of the silo for an
outlet radius of 07 m.
2

1
Acknowledgements
0
0 −20 −40 −60 −80 −100 −120 The authors are grateful to the CICYT (Spanish
Horizontal pressure, kPa Research and Technology Commission) for funding this
Fig. 12. Influence of the outlet radius on the envelope of project (AGL2002-02973).
horizontal pressures of the flat-bottom silo in kPa versus height: Pablo Vidal is grateful to the training and mobility
J, horizontal static pressure with outlet radius 05 m; , program of the Polytechnic University of Cartagena for
horizontal dynamic pressure with outlet radius 05 m; —, granting his stay at the University of Guelph (Canada).
horizontal static pressure with outlet radius 06 m; TT, horizontal
dynamic pressure with outlet radius 06 m; W, horizontal static
pressure with outlet radius 07 m; , horizontal dynamic
pressure with outlet radius 07 m References

AFNOR P22 630 (1992). Construction métallique: silos en


acier. [Metallic construction: steel silos.] Association Fran-
increases in pressures of up to 100 kPa were obtained at çaise de Normalisation, Saint-Denis La Plaine Cedex,
the effective transition and at the silo bottom by using France
Aubry D; Hujeux J C (1982). A double memory model with
FEM. The studied silo was 14 m high, had a 2 m radius, multiple mechanisms. International Symposium on Numer-
and stored Spanish Horzal wheat. ical Models in Geomechanics, Zurich, 13–17 September
However, discharge pressures do not correspond to Ayuga F; Guaita M; Aguado P (2001a). Static and dynamic silo
the Janssen law if it is weighted by a coefficient, as loads using finite element models. Journal of Agricultural Engi-
proposed in Eurocode 1 part 4, due to the influence of neering Research, 78(3), 299–308, doi:10.1006/ jaer.2000.0640
Ayuga F; Guaita M; Aguado P; Couto A (2001b). Discharge
the silo bottom and to the type of flow in this zone. In and the eccentricity of the hopper influence on the silo wall
such cases, the type of silo bottom may be a criterion to pressures. Journal of Engineering Mechanics ASCE,
consider in order to calculate actions on the walls. The 127(10), 1067–1074
French standard AFNOR P22 630 takes this criterion Cescotto S; Charilier R (1992). Frictional contact finite
into consideration. elements based on mixed variational principles. Interna-
tional Journal for Numerical Methods in Engineering, 36,
In silos with a hopper bottom, the overpressure 1681–1701
coefficient has an approximate value of 12 on the wall. Cescotto S; Zhu Y Y (1994). Large strain dynamic analysis
In the hopper, this coefficient takes values ranging using solid and contact finite elements based on a mixed
ARTICLE IN PRESS
348 P. VIDAL ET AL.

formulation; application to metalforming. Journal of agricultural materials for simulation by using numerical
Materials Processing Technology, 45, 657–663 methods.] PhD Thesis, Universidad Politécnica de Madrid,
Dı́ez M; Godoy L A (1992). Viscoplastic incompressible flow of Spain
frictional-cohesive solids. International Journal of Mechan- Moya M; Ayuga F; Guaita M; Aguado P (2002). Mechanical
ical Sciences, 34(5), 395–408 properties of granular agricultural materials. Transactions
Drucker D C; Prager W (1952). Soil mechanics and plastic of the ASAE, 45(5), 1569–1577
analysis on limit design. Quarterly of Applied Mathematics, Munch-Andersen J; Nielsen J (1990). Pressures in slender grain
10(2), 157–165 silos CHISA 1990. 2nd European Symposium on Stress and
Goodey R J; Brown C J (2004). The influence of the base Strain in Particulate Solids, Prague
boundary condition in modelling filling of a metal silo. Nielsen J (1998). Pressures from flowing granular solids in
Computers and Structures, 82, 567–579 silos. Philosophical Transaction of the Royal Society of
Goodey R J; Brown C J; Rotter J M (2003). Verification of a 3- London. Philosophical Transaction Mathematical, Physical
dimensional model for filling pressures in square thin-walled and Engineering Sciences, Series A, 356, 2667–2684
silos. Engineering Structures, 25(14), 1773–1783 prENV 1991-4. Eurocode 1 (2003). Basis of design and actions
Guaita M; Couto A; Ayuga F (2003). Numerical simulation of on structures—part 4: actions on silos and tanks. Final PT
wall pressure during discharge of granular material from draft (Stage 34) 5 March 2003. European Committee for
cylindrical silos with eccentric hoppers. Biosystems Engineer- Standardization, Brussels
ing, 85(1), 101–109, doi:10.1016/S1537-5110(03)00037-0 Ragneau E; Aribert J M (1995). Silo pressure calculation: from
Guines D; Ragneau E; Kerour B (2001). 3D finite-element a finite element approach to simplified analytical solutions.
simulation of a square silo with flexible walls. Journal of Bulk Solids & Handling, 15(1), 71–84
Engineering Mechanics, 127(10), 1051–1057 Rotter J M (2001). Guide for the Economic Design of Circular
Jenike A W; Johansson J R (1968). Bins loads. Journal of Metal Silos. Spon Press, London
Structure Division, Proceeding of ASCE, 94(ST4), Rotter J M; Holst F G; Ooi J Y; Sanad M (1998). Silo
1011–1041 pressures predictions using discrete-element and finite-
Jofriet J C; Lelievre B; Fwa F (1977). Friction model for finite element analyses. Philosophical Transaction of the Royal
element analyses of silos. Transactions of the ASAE, 20(4), Society of London. Philosophical Transaction Mathemati-
735–740 cal, Physical and Engineering Sciences. Series A, 356,
Jofriet J C; Negi S C; Lu Z (1997). A numerical model for flow 2685–2712
of granular materials in silos—part 3: parametric study. Sanad M A; Ooi J Y; Holst J M F G; Rotter J M (2001).
Journal of Agricultural Engineering Research, 68(3), Computations of granular flow and pressures in a flat
237–246, doi:10.1006/jaer.1997.0198 bottom silo. Journal of Engineering Mechanics, 127(10),
Keiter T W R; Rombach G (2001). Numerical aspects of FE 1067–1074
simulations of granular flow in silos. Journal of Engineering Simo J C; Laursen T A (1992). An augmented Lagrangian
Mechanics, 127(10), 1044–1050 treatment of contact problems involving friction. Computers
Link R A; Elwi A E (1990). Incipient flow in silo–hopper and Structures, 42(1), 97–116
configurations. Journal of Engineering Mechanics, 116, Vidal L P (2003). Modelos avanzados para simulación
172–188 numérica por elementos finitos de acciones en el vaciado
Lu Z; Negi S C; Jofriet J C (1997). A numerical model for flow de silos cilı́ndricos con tolva centrada y excéntrica.
of granular materials in silos—part 1: model development. [Advanced models for finite element numerical simulation
Journal of Agricultural Engineering Research, 68(3), of actions in the discharge of cylindrical silos with concentric
223–229, doi:10.1006/jaer.1997.0196 and eccentric hoppers.] PhD Thesis, Universidad de
Martı́nez M A; Alfaro I; Doblaré M (2002). Simulation of Santiago de Compostela, Spain
discharging processes in metallic silos. Analysis of the Walker D M (1966). An approximate theory for pressures and
induced pressure distribution and comparison with different arching in hoppers. Chemical Engineering Science, 21(11),
standards. Engineering Structures, 24(12), 1561–1574 975–997
Meng Q; Jofriet J C; Negi S C (1997). Finite element analysis of Wojcik M; Enstad G G; Jecmenica M (2003). Numerical
bulk solids flow—part 1: development of a model based on a calculations of wall pressures and stresses in steel cylindrical
secant constitutive relationship. Journal of Agricultural En- silos with concentric and eccentric hoppers. Particulate
gineering Research, 67(2), 141–150, doi:10.1006/jaer.1997.0161 Science and Technology, 21(3), 247–258
Moya M (2001). Determinación de parámetros fı́sicos de Zhong Z; Ooi J Y; Rotter J M (2001). The sensitivity of silo
materiales agrı́colas para su simulación por métodos flow and wall stresses to filling method. Engineering
numéricos. [Determination of physical parameters of Structures, 23(7), 756–767

You might also like