Download

282 Cao and Zhao / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2017 18(4):282-305
Journal of Zhejiang University-SCIENCE A (Applied Physics & Engineering)

ISSN 1673-565X (Print); ISSN 1862-1775 (Online)
www.zju.edu.cn/jzus; www.springerlink.com
E-mail: jzus@zju.edu.cn
Buckling design of large steel silos with various slendernesses*
Qing-shuai CAO, Yang ZHAO†‡

(Space Structures Research Center, Zhejiang University, Hangzhou 310058, China)
†
E-mail: ceyzhao@zju.edu.cn
Received May 12, 2016; Revision accepted July 4, 2016; Crosschecked Mar. 9, 2017
Abstract: Large steel silos are typical kinds of thin-walled structure which are widely used for storing huge quantities of granular
solids in industry and agriculture. In the present analyses, the buckling design of large steel silos subject to Eurocode-specified
solid pressure is demonstrated. The finite element model is established using the commercial general purpose computer package
ANSYS. Six types of buckling analyses are carried out for the geometrically perfect and imperfect models with and without
consideration of material plasticity. The load cases of concentric discharge, discharge patch load, large eccentricity discharge, and
large eccentricity filling are considered. The buckling behavior of six example steel silos with capacities of 30 000–60 000 m3 is
investigated. The silos’ slenderness ranges from 4.77 to 0.35, comprising very slender, slender, intermediate slender, squat, and
retaining silos. The index called the ratio of capacity to steel consumption (RCS) is initially defined in the paper, which provides an
effective measure for the economical design of steel silos. It is validated that the RCS index increases rapidly with the decrease of
silo slenderness, and the storage efficiency of steel silo improved greatly as the slenderness changes from slender silo to retaining
silo. The effects of patch load reveal that the buckling modes in the case of discharge patch load are very different from those of
silos under concentric solid pressure, and the effect is unfavorable for buckling resistance of all levels of slenderness of the ex-
ample silos, but contributes a small decrease to the RCS index (less than 10%). The buckling deformations from both the linear and
nonlinear buckling analyses in large eccentric discharge are strongly asymmetrical arising from the circumferential and meridional
non-uniform distribution of the solid pressures. The buckling is mainly governed by the non-uniform distribution of the solid
pressure other than other influential factors such as the weld imperfection, geometrical and material nonlinearity, compared with
the load case of concentric discharge. The RCS index of example silos under large eccentric discharge is reduced substantially, and
is approximately half that of silos under concentric discharge. The linear and nonlinear buckling deformations in large eccentric
filling are also asymmetrical, deviating from the center to the side where the most friction locates to the highest wall contact. The
RCS index of example silos under large eccentric filling is also reduced substantially, and is approximately 70% that of silos under
concentric discharge. This reveals that the large eccentricity both in discharging and filling could result in a strong decrease of
storage efficiency of steel silos.
Key words: Steel silo; Shell, Slenderness; Buckling; Patch load; Nonlinear; Large eccentricity
http://dx.doi.org/10.1631/jzus.A1600369 CLC number: TH161.12
1 Introduction Depending on the geometry and capacity of steel

silos, they may be supported in two fundamental
Silos are usually constructed in a vertically manners, discrete support and continuous support.
circular form and are used as vessels for storing Discrete support is applicable for silos with small
particulate granular solids filled by gravity at the top. diameter constructed using a local bracket or column,
giving a limited number of narrow supports around
‡
Corresponding author the silo circumference, and continuous support refers
*
Project supported by the National Natural Science Foundation of to the silo with large diameter (such as silo diameter
China (No. 51378459)
larger than 15 m) which is supported exactly on the
ORCID: Qing-shuai CAO, http://orcid.org/0000-0002-9405-1858;
Yang ZHAO, http://orcid.org/0000-0001-6880-0504 bottom plate with the underpass beneath the flat bot-
© Zhejiang University and Springer-Verlag Berlin Heidelberg 2017 tom (BSI, 2002b; 2005; 2006; 2007a; 2007b). Large
Cao and Zhao / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2017 18(4):282-305 283
steel silos widely used for storing huge quantities of the distribution and magnitude of action, e.g., under
granular solids in industry and agriculture are usually loading shown by the design criteria in different
continuously supported on the ground and consist of a standards (ISO, 1995; Standards Australia, 1996; BSI,
rigid flat bottom, a cylindrical shell with stepped wall 2002b; 2005). There are two key issues which require
thickness, and a conical or dome roof on the top. to be examined before the design can be implemented,
At present, steel silo development is at a critical one is the determination of wall pressures induced by
point with a boom in capacity requirement and a the stored granular solids, and the other is connected
number of practical engineering developments in with the buckling investigation of a silo which is
recent years. On the one hand, there is a great need for subject to combined horizontal pressure and vertical
large steel silos with capacities of over 10 000 m3 friction. With the recent advances of the finite ele-
given the development of the world economy, and on ment method (FEM) and the numerical algorithm, it is
the other, there is a paucity of studies on the structural now reasonably convenient for the researchers to
behavior, especially buckling, under solid pressures. perform numerical analyses of steel cylindrical shells
There is a need for further study through many site (Li et al., 2005; Zhao et al., 2006) under complex
measurements and numerical analyses, even though loading patterns and boundary conditions. The buck-
buckling behavior of small capacity steel silos has ling behavior of cylindrical steel shells under axial
been reported (Sadowski and Rotter, 2011a; 2011b; compression has been the most commonly researched
2011c). Ever larger capacity steel silos are now being aspect (Greiner and Guggenberger, 1998; Kim and
developed with the aid of automation and mechaniza- Kim, 2002). The review of recent research advances
tion and this provides an improvement in productivity and trends in the area of stability of un-stiffened
and economic investment. Steel silos with the diame- circular cylindrical shells subject to general non-
ter of 60 m or even more than 100 m are also fre-
uniform axial compressive stresses was provided by
quently reported, whose contents may be more than
Song (2002). Design results of the finite element (FE)
100 000 m3. A practical ground-supported circular
investigation into the buckling strengths of steel silos
steel silo, which is constructed in Liaoning Province,
subject to code-specified pressures for eccentric
China with diameter of about 45 m and height of 36 m
discharge were presented by Song and Teng (2003),
(capacity 57 000 m3), is shown in Fig. 1.
which indicate that the effect of patch loads from
different codes on nonlinear buckling loads is small,
and the axisymmetric primary wall loads are shown to
control the buckling loads. The structural behaviour
of silos supported on discrete eccentric brackets was
presented by Gillie and Holst (2003), considering the
effects of bracket width, bracket height, and geomet-
ric imperfections. It is found that the bracket-
supported silos are generally imperfection insensitive,
and the strength is strongly but nonlinearly dependent
on the bracket width and height. Song (2004) also
studied the effects of patch loads on the structural
Fig. 1 A practical ground-supported circular steel silo behavior of circular flat-bottomed steel silos, and
concludes that patch loads have a great effect on the
The structural design of circular steel silos is stress states in the silo according to linear elastic
well known to be usually controlled by considerations analysis (LA). Geometrical nonlinearity and primary
of buckling under axial compressive stress with the pressures have a beneficial effect. The effects of patch
buckling deformation as the potential failure mode. loads on thin-walled steel silos were also evaluated by
The design is complicated, as illustrated by the rele- Gillie and Rotter (2002). The results show that the
vant current European standards (BSI, 2002a; 2002b; stresses set up are complex and that they could
2005). The design rules differ greatly depending on potentially lead to failure of the silo by either elastic
buckling or plastic collapse. The circumferential potentially lead to failure of the silo by either elastic
width of the applied load and the pressure distribution buckling or plastic collapse (Gillie and Rotter, 2002).
both strongly affect the form and magnitude of the Thin-walled steel silos were also found to be gener-
stresses produced in the silo wall. Most importantly, ally sensitive to imperfections (Rotter and Zhang,
weld imperfections have been reported to have a 1990); a detailed parametric study was presented into
detrimental effect on the buckling resistance of silos the behavior of a slender silo under asymmetrical
under axial load. The impact of circumferential weld pressures describing the action of an eccentric
imperfections was investigated using the FEM by parallel-sided pipe flow channel of varying cross
Pircher and Bridge (2001), and it was found that the section, which reveals that geometric nonlinearity is
influence on the buckling behavior depended on the of much greater significance for cylindrical shells
strake height in relation to the linear meridional under asymmetrical load patterns than under sym-
bending half-wavelength and the depth of the imper- metrical patterns, and that eigenmode-affine imper-
fection. The elastic buckling strengths of imperfect fections, which are deleterious under axisymmetric
cylindrical silos containing granular solids were also loading patterns, are instead beneficial to the buckling
examined by Rotter and Zhang (1990), with imper- strength of a silo under eccentric discharge, thus
fections in the form of a depression at a circumferen- making them unsuitable for use in any design for this
tial welded joint, and it was suggested that the restrain load condition (Sadowski and Rotter, 2012). The
effect of granular solid may be incorporated into the stability process in a silo is investigated using a static
buckling calculation. Based on the current European and dynamic FE analysis by taking both the geometric
standard, the buckling behaviour of small capacity and material nonlinearity into account during eccen-
cylindrical steel silos with stepwise wall thickness tric discharge. The differences between the results of
and aspect ratios varying from very squat to very static and dynamic analyses were comprehensively
slender was explored for both perfect and imperfect discussed (Iwicki et al., 2014).
models under concentric and eccentric discharge by This paper investigates the buckling behavior of
Sadowski and Rotter (2011b), which showed that the large steel silos with capacities of 30 000–60 000 m3
aspect ratio plays a deciding role in both the behav- through numerical analyses, with slenderness ranges
iour and design of silos. from 4.77 to 0.35, which we can characterize as very
Thin-walled circular silos with stepped wall slender, slender, intermediate slender, squat, and
thickness were studied for eccentric discharge pres- retaining silos. In this study, the details of geometry
sure by Sadowski and Roster (2011c), which showed and capacity of the example silos are first presented in
that a silo design that was found to be very safe under Section 2 together with their design conditions. Then,
concentric discharge pressures becomes very unsafe we describe the analysis for example silos including
under eccentric discharge. With the assumption of a the geometrically perfect and imperfect FE models
parallel-sided flow channel, Sadowski and Roster and the types of buckling analyses (Section 3). In
(2011a) explored further the buckling behavior of very Section 4, we give the definition of buckling re-
slender silos under eccentric discharge. These were sistance of steel silos. The numerical results for con-
examined using geometrically and materially non- centric discharge of the example silos with various
linear analyses. The mechanics of the nonlinear be- slendernesses are presented for both the perfect and
havior of a cylindrical silo under eccentrically dis- imperfect models in all proposed buckling analyses in
charge was explored using the analogy of a propped Section 5 and the assessments of the different con-
cantilever slice beam to describe the nonlinear be- sequences of different load cases are also made for the
havior of cylindrical shells (Sadowski and Rotter, buckling resistance and buckling mode. The buckling
2013). The effects of patch loads on thin-walled steel behaviors of the example silos under large eccen-
silos were also evaluated by Gillie and Rotter (2002) tricity discharge and large eccentricity filling are
and Song (2004), and the studies concluded that patch evaluated in Sections 6 and 7, respectively, for both
loads have a great effect on the stresses in the silo the perfect and imperfect models in all proposed
from the LA (Song, 2004) and that they could buckling analyses. At the end, some conclusions
expected to benefit the understanding of the buckling the above design conditions for buckling assessment
behavior of steel silos subject to solid pressures are are specified in Eurocodes. The appropriate design
obtained. conditions are summarized and listed in Table 2.
Table 2 Design conditions assumed for the numerical

2 Geometry of example silos and solid analyses
pressure Design condition Class
Action assessment AAC3
Large circular steel silos with capacities of Fabrication tolerance Class B
30 000–60 000 m3 as used in practice are chosen as quality (high: quality parameter
illustrative examples in this study, and the geometries Q=25)
Consequence class CC3
of the six silos are listed in Table 1. The examples
Wall surface D1 (slippery)
cover every category of silo slenderness specified in
EN 1991-4 (BSI, 2006), the very slender (slenderness
hc/dc>4.0, where hc is the height of vertical-walled The distribution patterns and magnitudes of
segment of silo from transition to equivalent surface, pressures on the silo wall exerted by the stored gran-
and dc is the dimension of inside of silo cross-section), ular solid have been a crucial problem in the design of
slender (2.0≤hc/dc≤4.0), intermediate slender (1.0< silos. In the currently published Eurocode (BSI, 2006),
hc/dc<2.0), squat (0.4<hc/dc≤1.0), and retaining principles for the determination of pressures on the
(hc/dc≤0.4) silos, and also cover a very wide range of vertical walls of silos are provided so that the wall
silo geometry, the diameter dc ranging from 20 m to pressures induced by stored bulk solids during the
60 m, and the height hc ranging from about 21 m to filling and discharge shall be evaluated according to
95 m. The geometries of steel silos which are in ser- the slenderness of the silo. The pressures on the ver-
vice are basically incorporated within this range. A tical wall of silo induced by the stored bulk solids
steel silo with diameter of 45 m and height of 36 m during concentric filling or discharge are composed
(capacity 57 000 m3) is shown in Fig. 1. of two parts, the symmetrical component and the
Buckling analyses for exploring the ultimate non-symmetrical component referred to as the patch
limit state of silos depend strongly on the following load in EN 1991-4 (BSI, 2006), which shall be taken
working conditions: the action assessment class (Ta- to act simultaneously on the wall of a silo. The
ble 2.1, BSI, 2006), the fabrication tolerance quality
symmetrical component consists of horizontal pres-
class (Table 5.1, BSI, 2007b), and the consequence
sure ph and vertical friction pw on the wall, which are
class (Table 2.1, BSI, 2007b), which take account of
given by Eqs. (1)–(4) for the filling stage, where the
the influence of the silo capacity, the supporting pat-
subscript f indicates filling stage.
terns, and the eccentricity during filling and discharge,
For 2.0≤hc/dc,
and the weld imperfection of silo wall (Rotter and
Teng, 1989), etc., as appropriate. In this study, large
  z 
steel silos constructed with plain rolled steel sheets phf ( z )   Kz0 1  exp     . (1)
are taken as the reference examples whose contents   z0  
are usually more than 30 000 t in normal service life,
as listed in Table 1. The associated requirements of For 0.4<hc/dc<2.0,
Table 1 Geometries of the example silos with capacities of 30 000–60 000 m3

No. Geometry of silo Diameter, dc (m) Height, hc (m) Slenderness, hc/dc Capacity (m3)
SILO1 Very slender 20 95.49 4.77 30 000
SILO2 Slender 25 81.49 3.26 40 000
SILO3 Slender 30 70.73 2.36 50 000
SILO4 Intermediate slender 35 51.97 1.49 50 000
SILO5 Squat 45 31.44 0.70 50 000
SILO6 Retaining 60 21.22 0.35 60 000
  z  h   n     z 
phf ( z )   Kz0 1   0
  1  . (2) phce   Kzoc 1  exp    , (7)
  z0  h0      zoc  
For hc/dc≤0.4, where zoc is the Janssen characteristic depth for a flow
channel. The horizontal pressure phse in the static zone
phf ( z )   K (1  sin r ) z. (3) is equivalent to phf in the concentric filling process,
and the pressure phae in the adjacent zone is deter-
The vertical friction pw is defined based on the mined by the simple relation phae=2phf−phce. The fric-
following common relation: tional traction in the zones is also defined according
to the common relation pw=μph. The eccentric flow
pwf ( z )   phf ( z ). (4) channel wall contact angle, which is proved to be
independent of slenderness of the silo but determined
In Eqs. (1)–(4), z is the depth below equivalent sur- by the radius ratio kc, is approximately equal to 11.06°,
face, γ and K are the unit weight and lateral pressure 18.64°, and 30.56° for kc=0.25, 0.40, and 0.60, re-
ratio of the particulate solid, respectively; μ is the wall spectively. For details of the distribution of large
friction coefficient for solid sliding on the vertical eccentric discharge pressure, please refer to EN
wall; z0=r/(2Kμ) for circular silos, and r is the radius 1991-4 (BSI, 2006).
of silo; h0 is the depth below the equivalent surface to In the case of large eccentricities of filling, the
the lowest point on the wall that is not in contact with effect of the asymmetry of the normal pressures in
the stored solid of the top pile, while h0=rtan(r)/3 for inducing vertical forces in the silo wall is considered,
a symmetrically filled circular silo with top pile. which is fulfilled by adding the vertical wall pressures
n=−(1+tanr)/(1−h0/z0), and r is the angle of repose to those evaluated for symmetrical filling with a fill
of the particulate solid of conical pile. level corresponding to filling symmetrically to the
The symmetrical discharge pressures phe and pwe highest wall contact. The expressions for determina-
are determined by the simple overpressure factors Ch tion of the asymmetry vertical pressure were given by
and Cw as follows: EN 1991-4 (BSI, 2006), and not cited here for brevity.
Before pressures on the vertical wall of a silo can
phe ( z )  Ch phf ( z ), (5) be evaluated, the relevant properties of the stored
pwe ( z )  Cw pwf ( z ), (6) particulate solid have to be made clear. For com-
monly encountered bulk solids in industry such as
where Ch and Cw are the discharge factor for hori- cement clinker, the characteristic values of material
zontal pressure and frictional traction, respectively. properties are: γ=18 kN/m3, K=0.38×1.31=0.498,
The factors are in the range between 1.0 and 1.15 μ=0.46×1.07=0.492, and r=47° (excerpted from
determined by the silo slenderness and action as- Table E.1, BSI, 2006). The characteristic values K
sessment class. and μ are determined by multiplying or dividing the
When it comes to large eccentricity discharge, corresponding mean values by the conversion factors
the wall pressures are calculated by the definition of a: K=aKKm and μ=aμμm. When a limit state verifica-
three zones called the flowing zone, the adjacent zone, tion is sensitive to the variability of a material prop-
and the static zone, which are based on a parallel- erty, upper and lower characteristic values of the
sided flow channel throughout the height of the silo material property should be taken into account ac-
wall. The size of flow channel is defined in terms of cording to BSI (2002a; 2002b; 2006). The following
the ratio of flow radius to silo radius by kc=rc/r, and two load cases are also proposed to be considered in
three specific channel radii are proposed, kc=0.25, the buckling assessment of silos by EN 1991-4 (BSI,
0.40, and 0.60, to account for the eccentricity of the 2006): (a) load case I, maximum frictional pressures
channel and related wall solid pressure. The hori- with corresponding horizontal wall pressures, plus
zontal pressure phce in the flowing zone is given by patch loads; (b) load case II, maximum horizontal
wall pressures with corresponding frictional pressures. realistic restraints of out-of-round deformation for the
It is suggested by the results from Song and Teng silo top. The symmetric boundary condition (Uy=0,
(2003) that load case I is slightly more detrimental Rotx,z=0) is correspondingly imposed in both merid-
than load case II. As a result, K and μ are chosen the ional cut edges of the FE model. The FE mesh and
upper values resulting in maximum friction with restraint condition for boundary edges of the silo wall
corresponding horizontal wall pressures. are shown in Fig. 2.
3 Numerical analysis procedure

3.1 FE model
Ux,y=0
The buckling behavior of steel silos subject to
Uy=0, Rotx,z=0
Uy=0, Rotx,z=0
solids pressure is investigated using the commercial
general purpose FE computer package ANSYS
(ANSYS, 2008). The eight-node quadratic isopara-
metric shell element SHELL93 involving both the
bending and membrane effect is used to discretize the
shell wall. Each FE node has six degrees of freedom u
(DOFs), including translations in the nodal x, y, and z

directions and rotations about the nodal x, y, and z v
w
axes. The FE model is defined in the cylindrical Ux,y,z=0, Rotx,y,z=0
coordinate system (i.e., x-radial direction, y-
circumferential direction, and z-meridional direction), Fig. 2 FE half model of circular steel silo
including the definition of silo geometries and output
results, unless it is specifically stated otherwise. The The material of the silo wall is assumed to be
results are from the top layer of the shell element, in isotropic elastic-perfectly plastic with properties
which the meridional displacement u and the cir- typical of steel (grade S355): an elastic modulus E of
cumferential displacement v are specified as positive 2.1×105 N/mm2, a Poisson’s ratio ν of 0.3, and a
in the coordinate positive directions z and y, and the characteristic value of yield strength fyk of 355 N/mm2
radial displacement w as positive outwardly normal to (BSI, 2007a).
the shell surface. The symmetric distribution of solid Another geometrical parameter which has been
pressure permits half of the silo to be modeled (Song proved to have a considerable effect on the buckling
and Teng, 2003; Sadowski and Rotter, 2011b). As a behavior of a circular steel silo is the weld depression
result, the FE half model, obtained by cutting the shell imperfection arising from the fabrication process of
wall vertically across a diametrical plane, is used in the silo wall. In the present analysis, the assumed
the following analysis. shapes of weld depression (type A) are adopted and
The base strake of the silo wall is commonly introduced to the FE model, such as are proposed by
anchored to the rigid bottom plate in most practical Rotter and Teng (1989). The height of a normal strake
situations. For simplification, the bottom plate is not is assumed to be 3.0 m. For the example silos, such as
included here in the numerical model, and all trans- SILO4 in Table 1, 17 circumferential weld depression
lational and rotational degrees of freedom (Ux,y,z=0, imperfections along the wall height are modeled, in
Rotx,y,z=0) at the bottom edge of silo wall are re- which areas in the vicinity of the weld depression are
strained as boundary conditions. The roof and wind discretized with denser meshes.
girder at the top of silo are also omitted in the nu-
3.2 Types of buckling analysis
merical model, so only a cylindrical shell is modeled.
The radial and tangential translational restraints (Ux,y The limit state of buckling should be taken as the
=0) are applied to the silo top to simulate the influ- condition in which all or part of the structure suddenly
ence of the roof and the wind girder, as they provide develops large displacements normal to the shell
surface, caused by loss of stability under compressive The load-displacement curves, which are calcu-
membrane or shear membrane stresses in the shell lated by selecting the monitoring node at the critical
wall, leading to inability to sustain any increase in the buckling point, may be obtained for various buckling
stress resultants, possibly causing total collapse of the analysis types. There are also two types of funda-
structure. For the determination of load carrying ca- mental equilibrium path (Fig. 3) on which the critical
pacities when checking the buckling limit state, the buckling factor λcr is based: (1) the maximum load
following six types of buckling analysis, which are factor λcr1: load displacement curves predict a distinct
also recommended by EN 1993-1-6 (BSI, 2007a), are maximum load factor followed by a descending path;
carried out for the numerical investigation: (2) the load factor λcr2 corresponding to the largest
1. LBA—linear elastic bifurcation analysis of tolerable deformation, where this occurs during the
the perfect silo; loading path before reaching a maximum load factor.
2. GNA—geometrically nonlinear elastic analy- The tolerable global deflection w (radial displace-
sis of the perfect silo; ment) for a steel silo is recommended to be taken as
3. MNA—geometrically linear and materially H/50 (BSI, 2007b), where H is the height of the
nonlinear analysis of the perfect silo; structure measured from the foundation to the roof.
4. GMNA—geometrically and materially non- As the shell structure always buckles accompanying
linear analysis of the perfect silo; the geometric imperfections or plasticity, only non-
5. GNIA—geometrically nonlinear elastic anal- linear buckling results are adopted to estimate the
ysis of the imperfect silo; buckling strength of the example silos. The load
6. GMNIA—geometrically and materially non- carrying capacity of buckling is characterized by the
linear analysis of the imperfect silo. buckling resistance factor λR, which is proposed by
For buckling analyses, the recommended partial
EN 1993-1-6, when checking the buckling limit state.
factor γf of solids pressure is taken as 1.5 by EN
As a result, the critical buckling factor from the minor
1991-4, and the partial safety factor γM1 for the re-
values of λcr1 and λcr2 is taken as the buckling re-
sistance of the shell wall to stability is recommended
sistance factor λR of the example silos, which is a
for silos as 1.1 (BSI, 2007b), giving an overall safety
dimensionless parameter indicating the ratio of the
amplification factor 1.5×1.1=1.65 applied to the de-
designed load on silo structure to buckling strength.
sign values for the relevant load case. This safety
It is well known that buckling strength of a silo
factor of 1.65 provides a valuable reference for the
structure is closely correlated with the mesh size of
load proportionality factor against which nonlinear
the silo wall in the FE model. The lowest linear crit-
numerical analyses may be estimated to obtain a re-
ical buckling load factor λcr of the example silos is
alistic measure of safety of the designed structure.
obtained using different numbers of elements in both
circumferential and axial directions, as is plotted in
4 Determination of buckling resistance of Fig. 4. When the size of shell elements of the silo wall
silo structure decreases from 4.0 m×4.0 m (circumferential size and
axial size) to 1.0 m×1.0 m, the critical buckling factor
For the buckling analyses of shell structures, two gradually converges at the constant load factor for all
kinds of buckling are defined according to the struc- example silos. The discrimination of critical buckling
tural equilibrium path. These are well known as the factors between element sizes 4.0 m×4.0 m and
linear bifurcation buckling and nonlinear snap- 1.0 m×1.0 m is about 3%–6% for the example silos,
through buckling. The linear buckling analysis pre- and that is reduced to 1%–2% for results between
dicts the theoretical buckling strength (the bifurcation element sizes 2.0 m×2.0 m and 1.0 m×1.0 m, while
point) of an ideal linear elastic structure, while in the the results from element sizes 1.5 m×1.5 m and
nonlinear buckling analysis, the critical buckling is 1.0 m×1.0 m make no difference. It is concluded that
obtained by following the structural load-displacement the quadratic element type SHELL93 with midside
full course responses for both geometrically perfect nodes is applicable and highly effective in modeling
and imperfect models. the buckling behavior of silo structures subject to
solid pressure. It is certainly preferable to keep the discontinuity of wall stiffness compared to the
element sizes to less than 1.5 m×1.5 m in order to stepped wall thickness. For the tapered wall thickness,
obtain the accurate buckling factors. the equivalent thickness teq is determined by the wall
thickness at the uppermost ttop and at the downmost
tbot, and in the form of teq=(ttop+tbot)/2.
5 Buckling behavior under concentric
5.1 Stress and deformation distribution
discharge
The meridional compressive stress has been
For various reasons, buckling analyses of silo recognized as the main cause of many buckling fail-
walls are carried out in a diversity of forms known as ures of a silo structure. It is very helpful to make clear
the uniform thickness, the stepwise thickness, and the the distribution of stress and deformation on the silo
tapered thickness. Although the stepped wall thick- wall for a better understanding of buckling behavior.
ness is commonly adopted in practical fabrication of a In this section, the results of stress and deformation
silo wall for ease of construction, the tapered wall from the LA are presented for the example silos sub-
thickness for the example silos is also preferred in this ject to concentric discharge pressure, and are shown
paper because of its convenience for numerical mod- in Figs. 5 and 6. The vertical axes in the plotting
eling and better structural performance, which would represent the depth z from the bottom normalized by
bring about an effectively continuous taper and avoid the height hc of the silo wall, while the horizontal
1.0
Type 1: critical point
SILO1
0.8
SILO2
SILO3
Load factor λ
0.6
SILO4
z/hc
SILO5
0.4
SILO6
Type 2: critical point
0.2
0.0
Deflection w
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2
Tolerable deflection wmax σz/fyk
Fig. 3 Determination of critical buckling factor λcr of silo Fig. 5 Distribution of meridional stress of example silos
structure along the wall height
88 1.0
SILO1
77 0.8 SILO2
SILO1 SILO2 SILO3

Load factor λcr
66 SILO3 SILO4 0.6 SILO4

z/hc
SILO5 SILO6 SILO5

55 0.4 SILO6
44 0.2
33 0.0
4.0×4.0 3.0×3.0 2.0×2.0 1.5×1.5 1.0×1.0
0.0 0.3 0.6 0.9 1.2 1.5
Element size (m) w/teq
Fig. 4 Variation of critical load factor with mesh density Fig. 6 Distribution of radial displacement of example silos
of example silos along the wall height
axes represent the meridional stress component σz various slendernesses of example silos in each buck-
normalized by the material yield stress fyk and the ling analysis. This shows that the GMNIA analysis
radial displacement w normalized by the equivalent gives out the minimum buckling factors from all
thickness teq of the silo wall, respectively. proposed buckling analysis types for all example silos.
It is shown in Fig. 5 that the meridional stress is The critical buckling factor λcr in GMNIA analysis is
compressive along the wall height and increases slightly larger than and satisfies the requirement of
gradually from top to bottom, while the inflection the overall safety factor 1.65, which is consequently
point is induced at a critical height zcr due to the re- taken as the buckling resistance factor λR of the ex-
straint effect of rigid base. The stress magnitude σz,cr ample silos. The designed wall thickness distribution
at critical height zcr is significantly reduced by the corresponding to the critical buckling factor λcr in
negative moment, which is much more obvious for a Table 3 and used for buckling analysis below is given
small slenderness silo and tensile meridional stress is in Table 4.
even discovered in the zone adjacent to critical height The load-displacement curves are plotted in
zcr for retaining silo SILO6. The critical height zcr and Fig. 7 for various types of buckling analysis of ex-
the meridional stress σz,cr vary significantly with the ample silos. The curves are calculated by selecting the
aspect ratio of example silos and the compressive monitoring node in the maximum wall deflection at
stress reaches its maximum value σz,max at the bottom the critical buckling point. The vertical axis is the
of the silo wall. The critical height zcr is discovered to dimensionless load factor λ being applied to the de-
increase with the decrease of silo slenderness, which sign value of solids pressures, while the horizontal
is about 0.03hc for example SILO1 and 0.08hc for axis represents the out-of-plane radial displacement
example SILO6. The stress ratio σz,cr/σz,max is found to normalized by the equivalent thickness teq of the silo
increase with the increase of silo slenderness, which wall listed in Table 4. The load-displacement curves
is about 0.62 for example SILO1 and −0.14 for ex- are highly nonlinear which means the principle of
ample SILO6. It is concluded that the meridional superposition appropriate for the linear structure turns
stress σz is highly unevenly distributed along the out to be inapplicable for the thin-walled steel silos
height of the silo wall due to the pressure distribution under concentric solid pressure. Furthermore, the
and boundary effect, especially in the area adjacent to load-displacement curves predict a distinct maximum
the wall bottom. load followed by a descending path, in which the
The radial displacement plotted in Fig. 6 is maximum load is taken as the critical buckling point
shown to increase from top to bottom, and reaches its λcr for the equilibrium path.
maximum value wmax at the critical height zcr and Fig. 7 shows that the load factor λ initially in-
vanishes at the bottom. The maximum radial dis- creases linearly with the increasing radial displace-
placement is closely dependent on the slenderness of ment in the range of a certain critical wcr before the
example silos, and increases significantly with the critical point λcr is reached, and the critical radial
decrease of silo slenderness. The distribution is ex- displacement wcr corresponding to the critical buck-
actly consistent with that of meridional stress shown ling factor λcr varies greatly for different slender-
in Fig. 5. In addition, the circumferential distribution nesses of example silos, in which the critical dis-
of the meridional stress and radial displacement are placement wcr increases apparently with the decrease
axisymmetric, as can be easily concluded from the of the slenderness of silo. For instance, the critical
symmetry of silo structure and wall pressure. displacement wcr for SILO1 (very slender silo) is
about 0.4 times (12 mm) the even thickness teq, but the
5.2 Critical buckling factor and load displacement
value is approximately equal to 2.1 times (58.8 mm)
full course
teq for SILO6 (retaining silo) in the GMNIA analysis
In this section, buckling behavior in concentric (Fig. 7d). In other words, the slope to the vertical axis
discharge is first investigated for the case of steel silos of the load-displacement curves in the linear stage is
with tapered wall thickness. Both linear bifurcation much larger for small slenderness of silo than that for
and nonlinear buckling analyses are carried out. The large slenderness of silo. While it is even more ap-
critical buckling load factors are listed in Table 3 for parent in other types of buckling analysis, the critical
Table 3 Critical buckling factors for example silos from proposed buckling analysis types under concentric discharge
Critical buckling factor, λcr
No. Geometry of silo
LBA GNA MNA GNIA GMNA GMNIA
SILO1 Very slender 3.87 4.34 2.23 3.59 2.12 1.77
SILO2 Slender 3.78 4.13 2.22 3.72 2.03 1.73
SILO3 Slender 4.22 4.37 2.38 3.98 2.10 1.79
SILO4 Intermediate slender 4.04 4.17 2.11 3.96 1.91 1.68
SILO5 Squat 4.71 4.62 1.93 4.47 1.82 1.71
SILO6 Retaining 7.43 7.22 1.82 7.05 1.69 1.66
Table 4 Designed wall thickness and RCS index of example silos under concentric discharge
Diameter, Height, Slenderness, ttop tbot teq Capacity Steel consump-
No. RCS
dc (m) hc (m) hc/dc (mm) (mm) (mm) (m3) tion (m3)
SILO1 20 95.49 4.77 6.0 54.0 30.0 30 000 180.00 166.67
SILO2 25 81.49 3.26 6.0 56.0 31.0 40 000 198.40 201.61
SILO3 30 70.73 2.36 6.0 60.0 33.0 50 000 220.00 227.27
SILO4 35 51.97 1.49 6.0 56.0 31.0 50 000 177.14 282.26
SILO5 45 31.44 0.70 6.0 44.0 25.0 50 000 111.11 450.01
SILO6 60 21.22 0.35 4.0 52.0 28.0 60 000 112.00 535.71
RCS: ratio of capacity to steel consumption
5.0 8.0
SILO5
4.0
6.0
SILO6
Load factor λ
Load factor λ
3.0
4.0
2.0 SILO1
SILO2
2.0
1.0 SILO3
SILO4
0.0 0.0
0.0 0.5 1.0 1.5 2.0 0 20 40 60 80 100 120
w/teq w/teq
(a) (b)
8.0 1.8
7.0 1.5
6.0
SILO1 SILO2 1.2
Load factor λ
Load factor λ
5.0 SILO3 SILO4

4.0 SILO5 SILO6 0.9
3.0
0.6
2.0
0.3 SILO1 SILO2
1.0 SILO3 SILO4
SILO5 SILO6
0.0 0.0
0 5 10 15 20 25 30 35 40 0.0 1.0 2.0 3.0 4.0 5.0
w/teq w/teq
(c) (d)
Fig. 7 Typical load-displacement curves of example silos for various types of buckling analysis
(a) & (b) GNA; (c) GNIA; (d) GMNIA
displacement wcr for SILO1 in the GNA analysis is with the deformations evaluated by the LA analysis in
about 0.81 times (24 mm) the even thickness teq, and Fig. 6. In such a case, the deformation of thin-walled
the value is amplified to be about 13.6 times (380 mm) example silos is plastic buckling at the bottom in-
teq for SILO6 (Figs. 7a and 7b). It indicates that the duced by the meridional compressive stress due to the
silo with smaller slenderness is superior in structural vertical friction acting on the internal surface of the
ductility and plasticity to that of the one with larger silo wall.
slenderness, and the retaining silo or squat silo is The buckling modes from the GNIA and
preferred in structural design for practical purposes. GMNIA where the weld imperfection is taken ac-
When the critical buckling point λcr is passed, the count of are also similar, and show other different
sudden descent of the load factor is clearly found with buckling deformations from those of geometrically
a large radial deflection increase, which indicates the perfect models. The results from the GMNIA are
shift of the equilibrium path from the prebuckling to plotted in Fig. 9 for the example silos studied, which
the postbuckling stage. reveal that the structural deformations mainly occur
In other types of nonlinear buckling analyses in the form of harmonic waves in the axial direction
such as MNA and GMNA, the equilibrium paths throughout the whole height of the silo wall and the
shown by the load-displacement curves are also number of harmonic waves is almost identical to that
characterized by one of the two curves plotted in of the weld imperfections. In addition, the modes
Fig. 3, and similar conclusions can also be drawn, so are also axisymmetrical plastic buckling and the
they are not given here for brevity. elephant-foot deformations at the bottom part of the
shell wall also occur in spite of the interactive effect
5.3 Buckling modes
with the multi weld depressions. It suggests that the
The buckling modes of example silos for LBA geometric weld imperfections induced during fabri-
analysis are plotted in Fig. 8 corresponding to the cation should be taken into consideration for the
lowest buckling eigenvalue, which shows that the structural design of steel silos.
structural deformations are axisymmetrical and
mainly occur in the form of harmonic waves in the
axial direction in the middle region of the silo wall.
The number of harmonic waves caused by the me-
ridional compression is closely related to the slen-
derness of the silo, i.e., the number increases with the
increase of the silo slenderness and decreases with the
decrease of the silo slenderness. The deformations
from linear buckling analysis are far different from
that predicted by the LA, in which the outward radial
deflections increase with the depth from the top, and
obtain their maximum value at the height adjacent to
the silo bottom.
The nonlinear buckling deformations which are
taken from the critical buckling point are very dif-
SILO1 SILO2 SILO3 SILO4
ferent from the buckling mode given by the linear
bifurcation (LBA) result. The buckling deformations
predicted by the GNA, MNA, and GMNA analyses
are similar, where the radial deflections increase with
the depth and the maximum deflections occur just
adjacent to the bottom of the silo wall. The buckling
modes from the GNA, MNA, and GMNA analyses
SILO5 SILO6
shift into the well-known elephant-foot deformations
at the bottom part of the shell wall, which agree well Fig. 8 Buckling modes of the example silos in LBA
earity is strong and detrimental to buckling behavior

of all slendernesses of example silos, resulting in a
decrease of buckling resistance for both perfect and
imperfect models. The conclusion agrees well with
that from Iwicki et al. (2014) and that from Song and
Teng (2003). The detrimental effect of material non-
linearity brings about a decrease of buckling load
factor by about 50%–80%, which is much more un-
favorable for steel silos with small slenderness than
those with large slenderness. For instance, the buck-
ling load factor reduced by about 51% for a very
slender silo (SILO1) arising from the effect, but the
factor reduced by about 78% for the retaining silo
(SILO6). It is indicated that the effect of material
SILO1 SILO2 SILO3 SILO4
plasticity is extremely powerful and plays a decisive
role in nonlinear buckling analyses for every slen-
derness of large steel silo. As a result, material with
higher yield strength is recommended in the buckling
design of large steel silos, where the current conclu-
sions show a considerable discrepancy with those
SILO5 SILO6 drawn from investigations of small capacity of steel
silos (Sadowski and Rotter, 2011b).
Fig. 9 Buckling modes of the example silos in GMNIA
5.4 Effects of nonlinearity and imperfection

The nonlinear effect proves to have much in-
fluence on the buckling behavior of steel silos. The
geometrical nonlinear effects on various slender-
nesses of example silos are plotted in Fig. 10 by the
ratio of GNA load factor λcr1 to LBA load factor λcr2
listed in Table 3. This reveals that geometrical non-
linearity is beneficial for steel silos with large slen-
derness but somewhat harmful for steel silos with
small slenderness. The buckling factor λcr for a very Fig. 10 Geometrical and material nonlinear effects in
slender silo (SILO1) increases by about 12% for the buckling analysis
beneficial effect of geometrical nonlinearity, but it
decreases by about 3% for retaining silo (SILO6) for The weld imperfection effect is shown in Fig. 11
the harmful effect of geometrical nonlinearity. As by the ratio of GNIA or GMNIA load factor λcr1 to
steel silos are typical kinds of thin-walled structure, GNA or GMNA load factor λcr2 listed in Table 3,
and large deflections always occur with the buckling which reveals that weld imperfection is detrimental to
of a structure, geometrical nonlinearity should buckling behavior of all slendernesses of example
therefore be taken into account in buckling analysis of silos. The detrimental effect is more serious for
steel silos. slender silos than squat silos, where the buckling load
The material nonlinear effect is also plotted in factor reduced by about 17% for the very slender silo
Fig. 10 by the ratio of MNA, GMNA or GMNIA load (SILO1), but the factor reduced only by about 3% for
factor λcr1 to LBA, GNA or GNIA load factor λcr2 retaining silo (SILO6). The reasons are explained in
listed in Table 3, which reveals that material nonlin- what follows: since the strake height for construction
of the vertical wall of the silo is assumed to be the The wall thicknesses satisfying the buckling
invariant value of 3 m, the number of circumferential requirement of silo design stipulated by the Eurocode
weld imperfections simulated in the slender silo is are listed in Table 4 together with the RCS index. The
much greater than that in the squat silo. For instance, top thickness of the tapered wall is commonly en-
there are 31 circumferential weld imperfections along countered as a thin 4 mm or 6 mm, but the bottom
the wall height modeled for the very slender silo thickness is much greater, and that for the retaining
(SILO1), whereas six merely weld imperfections are silo wall can even be as much as 60 mm, which un-
modeled for the retaining silo (SILO6). The interaction doubtedly adds difficulty to the fabrication and flex-
between the neighboring circumferential weld im- ure of the silo wall. For the example silos above, the
perfections further reduced the buckling strength and steel in a silo wall ranges from about 112 m3 to 180 m3
this is more obvious for slender silos with more weld with the variation of slenderness from retaining to
imperfections. It suggests that the weld imperfections very slender, equivalent to a mass between approxi-
should be incorporated in the numerical models and the mately 880 t and 1400 t.
magnitude be reduced in the fabrication of a silo wall The variation of the RCS index with the slen-
to improve the buckling resistance of the structure. derness of steel silo is plotted in Fig. 12, which shows
that the RCS index increases rapidly with the decrease
of silo slenderness. In other words, the storage effi-
ciency of a steel silo improves greatly with the slen-
derness varying from slender silo to retaining silo. For
instance, the RCS index for the very slender silo
(SILO1) is about 167, but the value for the retaining
silo (SILO6) is about 536, a factor of approximately
3.2. Similarly, for steel silos with a capacity of
50 000 m3, the RCS index for SILO5 with a slender-
ness of 0.7 is approximately 1.6 times that of SILO4
with slenderness of 1.49. It suggests that silos with
small slenderness such as squat and retaining silos are
Fig. 11 Weld imperfection effects in buckling analysis to be preferred in steel silo structural design.
5.5 Designed wall thickness and steel consumption

index
Steel silos with large capacity are usually used
for storage of huge quantities of granular solids, and
RCS
the consumption of materials in the construction of

some silo walls can be as large as thousands of tons.
The amount of material in steel silos may differ
greatly depending on the designers. The index called
the ratio of capacity to steel consumption (in volume)
(RCS) is initially defined by Eq. (8), which provides a SILO1 SILO2 SILO3 SILO4 SILO5 SILO6
useful parameter in the design of steel silos.

Fig. 12 Relationship of RCS index with slenderness of
example silos
Capacity of steel silo
RCS  . (8)
Steel consumption of silo wall
5.6 Discharge patch load effect
This RCS index of a steel silo is a dimensionless It is specified (BSI, 2006) that the patch load
parameter indicating the material efficiency of the shall be used to represent accidental asymmetries of
storage of steel silos. loading during discharge. From the viewpoint of safe
design, the patch load position hp should be placed silos is similar to that obtained in the load case of
anywhere over the height of the silo wall, and buck- concentric discharge, and thus is not given again.
ling of the silo should be based on the most unfa-
5.7 Effect of boundary condition
vorable position of the patch load. The effect of patch
load position on buckling behavior of example silos is In the preceding analysis, the base strake of silo
investigated, and this shows that the patch load is wall is assumed to be anchored to the rigid bottom
unfavorable to buckling and results in a decrease of plate and the clamped boundary condition (Ux,y,z=0,
about 5%–25% of the buckling factor. The most un- Rotx,y,z=0) is used at the lower edge of the silo wall.
favorable position of the patch load is approximately However, the restraints of rotational degrees of free-
0.6 time of the height hc for the example silos, which dom (Rotx,y,z=0) at the bottom edge of the silo wall
is adopted for the evaluation of patch load effect in could not be effectively achieved for many practical
this section. considerations, such as the limited rigidity of the
The effect of patch load on buckling resistance in bottom plate, and what is more important, full welds
the GMNIA is plotted in Fig. 13 by the ratio of between the base strake and the bottom plate are es-
buckling resistance λcr1 in the case of patch load to λcr2 pecially difficult to achieve. The simply supported
under concentric solid pressure, which shows that the boundary condition (Ux,y,z=0) is considered and the
effect is unfavorable for buckling resistance of all buckling results are compared with those from the
slendernesses of example silos. Nevertheless, the clamped boundary condition.
detrimental effect is much more serious for silos with The effect of boundary condition on buckling
large slenderness than those with small slenderness. resistance in the GMNIA is plotted in Fig. 14 by the
The buckling resistance λcr1 is reduced by about 9% ratio of buckling resistance λcr1 under simply sup-
for the very slender silo (SILO1), whereas the detri- ported to λcr2 under a clamped boundary condition. It
mental effect is very small for the retaining silo reveals that the variation of boundary condition from
(SILO6), with only a decrease of about 1% compared clamped to simply supported has a limited influence
to λcr2. The recommendation by EN 1991-4 that the on buckling strength of example silos, but is more
effect of patch load on squat and retaining silos can be marked for squat silos than for slender silos. The
neglected is shown to be reasonable according to the variation of boundary condition from clamped to
numerical evaluation. In such cases, the RCS index is simply supported results in a decrease of buckling
also slightly reduced by less than 10% for slender strength by about 9% for SILO6, but only induces a
silos due to the limited effect of patch load, even in decrease of about 1% for SILO1. It is concluded that
case of the largest eccentricity of 0.25dc, and it re- the effect of boundary condition should be incorpo-
mains almost invariable for squat and retaining silos. rated in the numerical models for buckling analysis,
The variation of RCS with slenderness of example especially for the squat and retaining silos.
1.0
0.8
0.6
λcr1/λcr2
λcr1/λcr2
0.4
0.2
0.0
SILO1 SILO2 SILO3 SILO4 SILO5 SILO6 SILO1 SILO2 SILO3 SILO4 SILO5 SILO6
Fig. 13 Effect of patch load on buckling resistance in Fig. 14 Effect of base rigidity on buckling resistance of
GMNIA example silos
5.8 Stepwise thickness of silo wall For thickness distribution in groups A, B, and C, the
buckling deformations initially occur in the top sev-
The preceding investigation is carried out by
eral strakes of the silo wall which is relatively thin
assuming that the wall of the steel silo is tapered from
and vulnerable to deflections compared with other
the bottom to the top. This assumption is theoretically
regions. Thus, in such cases the buckling strength of
ideal, whereas silo strakes with different thicknesses
the wall strakes at the lower part is not fully utilized,
are joined with welds to construct the complete
at about 60%–70% to that of the tapered wall thick-
structure in most practical circumstances, forming the
ness, as is shown in Figs. 15 and 16a. For the uniform
so-called stepped wall thickness. Therefore, it is of
wall thickness in group E, the buckling deformation
great importance to perform a buckling analysis of
plotted in Fig.16c occurs first at the vicinity of the
silo structures with stepped wall thickness. The
vulnerable bottom, resulting in a decrease of about
buckling behavior of the retaining silo (SILO6) with
half of the buckling resistance compared to that of the
stepped wall thickness and with the minimum number
tapered wall thickness. For thickness distribution in
of weld depressions is taken as the representative of
group D, the buckling deformation is shown in
the example silos and evaluated in this section. The
Fig. 16b and arises in wall strakes at the top and
results will also be compared with that from the ta-
bottom simultaneously, and the buckling resistance
pered and uniform wall thickness under the circum-
increases by about 5% due to the favorable wall
stances that the three types of wall thickness distri-
thickness distribution. As a result, the thickness ar-
bution share the same equivalent thickness teq of
rangement of group D is the optimal one of those
28 mm as determined by the former investigations. In
listed in Table 5. It is concluded that the wall thick-
the case of the stepped wall silo, the equivalent
ness type and its distribution have the most important
thickness teq is defined in the following form:
influence on the buckling modes and buckling
strength of steel silos. Under the condition of the same
teq 
ht
i i
, (9) amount of steel used in construction, different
h i schemes for wall thickness arrangement are suggested
in buckling design of steel silos to achieve the optimal
where hi and ti is the height and thickness of the ith buckling strength.
wall strake, respectively. For the convenience of
comparison, four groups of thickness for stepped wall 1.2
are designated, as are noted groups A to D in Table 5 1.05
1.0
together with the distribution of uniform and tapered
wall thickness. 0.8 0.73
The effects of different types of silo wall thick- 0.64
λcr1/λcr2
0.59
ness on buckling resistance are demonstrated in 0.6 0.52
Fig. 15 by the ratio of λcr1 of the silo wall thickness in 0.4
groups A–E to λcr2 of the tapered thickness in group F
predicted in the preceding studies. It indicates that the 0.2
distribution of wall thickness has much influence on
0.0
the buckling resistance and buckling mode of the silo Group A Group B Group C Group D Group E
structures. The buckling resistance may decrease Fig. 15 Effect of thickness distribution on buckling re-
greatly due to irregular distribution of wall thickness. sistance of silos
Table 5 Thickness groups of the retaining silo wall (SILO6) (units: m for hi; mm for ti)
Group h1×t1 h2×t2 h3×t3 h4×t4 h5×t5 h6×t6 h7×t7 teq
A 3.22×4.0 3.0×12.0 3.0×24.0 3.0×30.0 3.0×36.0 3.0×42.0 3.0×50.0 28.0
B 3.22×4.0 3.0×10.0 3.0×20.0 3.0×28.0 3.0×36.0 3.0×44.0 3.0×56.0 28.0
C 3.22×4.0 3.0×8.0 3.0×16.0 3.0×24.0 3.0×36.0 3.0×50.0 3.0×60.0 28.0
D 3.22×6.0 3.0×14.0 3.0×21.0 3.0×28.0 3.0×36.0 3.0×43.0 3.0×50.0 28.0
E Uniform thickness: ti=28.0 mm
F Tapered wall thickness: ttop=4.0 mm; tbot=52.0 mm; teq=28.0 mm
and deformation are highly unevenly distributed not

only along the wall height but also in the circumfer-
ential direction. The maximum meridional stress
along the circumferential occurs in the flowing zone,
where the vertical distribution of meridional stress σz
(a) is plotted in Fig. 17 for example silos.
1.0
dc=20
SILO1
0.8
dc=25
SILO2
0.6 dc=30
SILO3
(b)
z/hc
dc=35
SILO4
0.4 dc=45
SILO5
SILO6
dc=60
0.2
0.0
-3.0 -2.0 -1.0 0.0 1.0 2.0 3.0
(c) σz/fyk
Fig. 16 Buckling modes for various groups of wall thick- Fig. 17 Distribution of meridional stress of example silos
ness distribution along the wall height
(a) Groups A, B, and C; (b) Group D; (c) Group E
It shows that the meridional stress is compres-
sive in the vertical coordinate z above 0.2hc and be-
6 Buckling behavior under large eccentricity comes tensile for coordinate z below 0.2hc which
discharge results from the inflecting bending moment due to the
asymmetric distribution of solid pressure. The max-
In the process of discharging with large eccen- imum stress is closely correlated with the slenderness
tricity, the pressures on the wall are determined by of silos, where the maximum compressive stress σz,max
assuming a parallel flow channel with circular sec- occurs approximately at the critical height zcr of 0.6hc,
tion. Three typical zones distinguishing the circum- and the maximum tensile stress σz,0 occurs at the
ferential variation of solid pressure are specified and bottom of the silo wall. The maximum stress in-
these are called the flowing zone, the adjacent zone, creases significantly with the increasing slenderness
and the static zone. The eccentric flow channel radius of example silos, where the maximum compressive
rc is taken as 0.6 times the silo radius in the following stress σz,max and tensile stress σz,0 for SILO1 are about
numerical investigation, which has been widely ac- −2.20fyk and 2.88fyk, respectively, and they are about
cepted to be the most detrimental size of eccentricity, −0.58fyk and 2.16fyk for SILO6, respectively. For
and also adopted by other researchers (Sadowski and meridional stress in the adjacent zone, the distribution
Rotter, 2011c). The eccentric flow channel wall con- is quite distinct from that in the flowing zone shown
tact angle θc has the most impact on distribution zones in Fig. 17. The critical height zcr is about 0.55hc, and
of solid pressures, with a central angle of about 30.56° the inflection point is about 0.3hc. The meridional
for kc of 0.6 specified in EN 1991-4 (BSI, 2006). stress is tensile at a height range from the top to the
inflection point while it is compressive at a height
below the inflection point. The maximum compres-
In this section, the results of stress and defor- sive and tensile stresses in the adjacent zone are about
mation from a LA are presented for the example silos −2.96fyk and 1.65fyk for SILO1, respectively, and they
subject to large eccentric discharge pressure, and are are about −1.56fyk and 0.61fyk for SILO6, respectively.
shown in Figs. 17 and 18. It is indicated that the stress The stress distribution in the static zone is similar to
that shown in Fig. 5 and is not given here. It is con- (SILO1), and the deformed area shifts into its adja-
cluded that the stress distribution of steel silos subject cent zone gradually with the decrease of the silo
to eccentric discharge pressure is highly non-uniform slenderness, such as the retaining silo (SILO6). The
in both circumferential and meridional directions, deformation in the static zone is nearly invisible
which is quite different from the load condition of compared to that in the flowing zone or adjacent zone.
concentric discharge. In addition, the deformation transfers to the top area
The distribution of radial displacement also for small slenderness of silo such as the squat and
shows that the deformation of steel silos is retaining silos, in the form of local buckling, indi-
non-uniform in the circumference from the flowing cating the distinction of buckling modes between the
zone to the static zone, where the center of the flow- slender silos and the squat silos.
ing zone is designated as the circumferential angle θ
of −180 in Fig. 18. The maximum radial displace-
ment occurs at about the angle of −130–−140, and
increases with increasing slenderness of silo. The
circumferential variation of deformation is charac-
terized by several waves due to the interval of out-
ward and inward displacements, and the number of
waves is dependent on the slenderness of silo which
increases with the decrease of slenderness. The vari-
ation of waves number is consistent with the cir-
cumferential variation of stress on the silo wall.
25.0
20.0 SILO1 SILO2
15.0 SILO3 SILO4
SILO5 SILO6
10.0 SILO1 SILO2 SILO3 SILO4
5.0
w/teq
0.0
-5.0
-10.0
-15.0
-20.0
-25.0
-180 -150 -120 -90 -60 -30 0
Circumferential angle θ ()
Fig. 18 Distribution of radial displacement of example

silos along the circumference SILO5 SILO6
Fig. 19 Buckling modes of the example silos in LBA

6.2 Buckling modes
The results reveal that the buckling deformations The nonlinear buckling deformations corre-
from both the linear and nonlinear buckling analyses sponding to the critical buckling point occur over
are asymmetrical arising from the circumferential and nearly all the height of the silo wall, but obtain the
meridional non-uniform distribution of the solid maximum value in the middle height for both the
pressures. The buckling deformations from the LBA perfect (GNA, MNA, and GMNA) and imperfect
analysis shown in Fig. 19 occur taking the form of (GNIA amd GMNIA) models. The deformations in
several meridional waves with assorted magnitudes the flowing zone are concave to the center and convex
mainly in the middle part of the flowing zone for silos in the adjacent zone deviating from the center. Several
with large slenderness as the very slender silo circumferential waves with small magnitude are also
seen in the static zones. The “elephant’s failure” which is a remarkable distinction from that subject to
modes, which have been the signs of plastic defor- the concentric solid pressure.
mation of silos under eccentric solid pressure, are not
6.3 Critical buckling factor
clearly observed in the GMNA and GMNIA with
consideration of material nonlinearity. The buckling The load-displacement equilibrium paths in a
modes of example silos in the GMNIA are plotted in large eccentricity discharge are also highly nonlinear,
Fig. 20. and the critical buckling load factors λcr are listed in
Table 6 for various slendernesses of example silos in
each buckling analysis. The results are very distinct
from those of silos under concentric solid pressure,
and reveal that the GMNA or GMNIA, which takes
account of the effect of geometrical and material
nonlinearity, give out the least value for all example
silos from all proposed buckling analysis types. The
minor value from critical buckling factor λcr in the
GMNA or GMNIA satisfies the requirement of the
overall safety factor 1.65 which is consequently taken
as the buckling resistance factor λR of the example
silos in a large eccentricity discharge.
It is also indicated in Table 6 that the geometrical
effect is advantageous for the critical buckling factor
λcr for all slendernesses of example silos. There is an
increase in buckling factor by 2.5–3.5 times com-
SILO1 SILO2 SILO3 SILO4 pared with that under concentric solid pressure. The
effect of weld imperfections in a silo under large
eccentric discharge shows a distinct variation of crit-
ical buckling factor λcr from that under concentric
solid pressure. It is worse for large slenderness of
silos and better for small slenderness of silos. For
instance, the detrimental effect of weld imperfections
decreases the critical buckling factor λcr by 19% and
21% for the very slender silo (SILO1) and slender silo
(SILO2), respectively, while the beneficial effect of
weld imperfections increases the critical buckling
SILO5 SILO6
factor λcr by 10% and 4% for the squat silo (SILO5)
Fig. 20 Buckling modes of the example silos in GMNIA and retaining silo (SILO6), respectively. The com-
plicated effect of weld imperfections on the buckling
behavior of silo results from the type of weld depres-
In addition, the distribution of weld imperfec- sion, the magnitude, the separation and number of
tions have not as much impact on the buckling modes imperfections, and most important of all, the asym-
of the numerical models in the GNIA and GMNIA as metrical distribution of solid pressure during large
that under concentric solid pressure. The harmonic eccentric discharge. The axisymmetrical weld im-
waves along the meridioanal direction, which have perfection type, which has been shown to result in a
been clear signs of imperfective silo structure under significant loss of buckling load factor in concentric
concentric pressure, are not yet observed for the im- discharge, seems not to be an appropriate choice for
perfect models (GNIA and GMNIA). The buckling investigation of buckling of silos under large eccen-
deformations are governed mainly by the distribution tric discharge. In addition, the material nonlinearity is
of the circumferential and meridional non-uniformity detrimental, and has the most detrimental effect and
of the solid pressure in a large eccentricity discharge, makes a loss of above 3/4 of the buckling factor λcr.
6.4 Designed wall thickness and steel consumption circumferentially uniform in spite of the non-uniform
index distribution of solid pressure. This is based on the
assumption that the possibility of eccentricity during
The designed wall thickness distribution corre-
sponding to the critical buckling factor λcr in Table 6 discharging is identical in every direction, except that
and used for buckling analysis is given in Table 7 the particular operation is arranged to assure a fixed
together with the RCS index. It shows that both the eccentricity direction.
top and bottom thickness of the tapered wall silo The variation of RCS with the slenderness of
under large eccentric discharge are much larger than example silos under large eccentric discharge is
that under concentric solid pressure, and the bottom plotted in Fig. 21 together with that under concentric
thickness in some example silos are even as thick as solid pressure for ease of comparison. This shows that
88 mm, which is about 1.5 times that of silos under the variation of RCS index of the two groups of
the concentric solid pressure. It indicates, on the one column is similar and increases rapidly with the
hand, that the non-uniform distribution of solid
pressure in a large eccentric discharge has the most
detrimental effect on wall thickness and its buckling
resistance; on the other, it shows that the effect is over
all the height of the silo wall. Designs for silos under
large eccentric discharge and concentric discharge are
proposed to be carried out independently. The amount
of steel used in the example silos ranges from about
204 m3 to 360 m3 with the variation of slenderness
from retaining to very slender, equivalent to a mass
from approximately 1700 t to 2800 t, which is about
twice that of silos under concentric discharge. In such
cases, the fabrication and flexure of the silo wall be-
come even more complicated during construction. In Fig. 21 Relationship of RCS index with slenderness of
addition, the thickness in Table 7 is designed to be example silos
Table 6 Critical buckling factors for example silos from proposed buckling analysis types under large eccentricity
discharge
SILO1 Very slender 1.51 5.31 2.73 4.31 2.09 1.66
SILO2 Slender 1.33 4.63 2.49 4.08 1.91 1.68
SILO3 Slender 1.24 4.67 2.54 4.79 1.69 1.74
SILO5 Squat 1.47 5.64 2.64 6.22 1.68 1.84
SILO6 Retaining 1.16 6.71 1.92 6.96 1.69 1.74
Table 7 Designed wall thickness and economic index of example silos under large eccentricity discharge
Diameter, Height, Slenderness, ttop tbot teq Capacity Steel con-
No. RCS
dc (m) hc (m) hc/dc (mm) (mm) (mm) (m3) sumption (m3)
SILO1 20 95.49 4.77 40.0 80.0 60.0 30 000 359.99 83.34
SILO2 25 81.49 3.26 40.0 86.0 63.0 40 000 403.21 99.20
SILO3 30 70.73 2.36 40.0 88.0 64.0 50 000 426.63 117.20
SILO4 35 51.97 1.49 36.0 88.0 62.0 50 000 354.29 141.13
SILO5 45 31.44 0.70 32.0 70.0 51.0 50 000 226.68 220.57
SILO6 60 21.22 0.35 32.0 70.0 51.0 60 000 203.99 294.13
decrease of silo slenderness. Nevertheless, the RCS ranging from 0.25dc to 0.50dc, with a typical eccen-
index of example silos under large eccentric discharge tricity of 0.50dc which is deemed as the most detri-
is reduced substantially, at approximately half that of mental size of eccentricity and adopted in the fol-
silos under concentric discharge. For instance, the lowing numerical evaluation.
RCS index for the very slender silo (SILO1) is about
167 for concentric discharge, but the value decreases
to about 83 for large eccentric discharge. Similarly, In this section, the results of stress and defor-
the RCS index for the retaining silo (SILO6) is about mation from the LA are presented for the example
536 for concentric discharge, but the value decreases silos subject to large eccentricity filling pressure, and
to about 294 for large eccentric discharge. Moreover, are shown in Figs. 22 and 23. They indicate that the
for steel silos with a capacity of 50 000 m3, the RCS stress and deformation are highly unevenly distrib-
index for SILO5 is approximately 1.6 times that of uted in both the circumferential and meridional di-
SILO4 for concentric discharge. Nevertheless, the rections. The maximum meridional stress along the
value for SILO5 is about 1.57 times that of SILO4 for circumference occurs in the zone below the point of
large eccentric discharge. The squat or retaining silo the highest wall contact, where the vertical distribu-
with small slenderness has the greatest storage effi- tion of meridional stress σz is plotted in Fig. 22 for
ciency from the economical design viewpoint. The example silos. The meridional stress is compressive
small slendernesses of silos such as squat and re- along the wall height and increases gradually from the
taining silos are strongly proposed in practical struc- top to bottom, while the inflection point is induced at
tural design for both concentric and large eccentric
discharge. 1.0
SILO4
0.8 SILO45
7 Buckling behavior under large eccentricity SILO5
0.6
filling
z/hc
0.4
The diameter of some large steel silos can be as
much as 60 m or even larger. They are usually ec- 0.2
centrically filled for technical reasons, such as the
limitation of the machinery operation. Even for a 0.0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
concentrically filled silo, the top surface pile during σz/fyk
filling or when the silo is full may be formed with a
large eccentricity, resulting in asymmetrical solid Fig. 22 Distribution of meridional stress of example silos
along the wall height
pressure on the wall and the asymmetrical meridional
stresses. The effect of large eccentric filling should be
1.0
taken into consideration for the intermediate and
squat silos when the filling eccentricity is greater than
0.8
0.25 times the silo diameter, as is specified by EN
1991-4. Another silo, with a capacity of 50 000 m3 0.6 SILO4
designated as SILO45 in the following analysis, is SILO45
z/hc
SILO5
added to the group of example silos. Its size is: a 0.4
diameter of 40 m, height of 39.79 m, and slenderness
0.2
of 0.99. As a result, the buckling behavior of three
example silos including SILO4, SILO45, and SILO5
0.0
are evaluated for the intermediate and squat silos 0.00 0.10 0.20 0.30 0.40 0.50
under large eccentricity filling. This indicates that the w/teq
magnitude of the asymmetrical friction increases Fig. 23 Distribution of radial displacement of example
almost linearly with the increase of filling eccentricity, silos along the wall height
critical height zcr due to the restraint effect of rigid addition, the buckling modes corresponding to the
base. The distribution of stress is similar to that of critical buckling point are also plastic combined with
example silos subject to concentric discharge except the “elephant’s failure” deformations at the wall bot-
that the meridional stress is compressive along the tom in the GMNIA on account of material plasticity,
whole height of the silo wall. There is no tensile stress as is shown in Fig. 25.
induced even if the inflection point and negative
moment exist. It is concluded that the meridional
stress σz is highly unevenly distributed along the
height of the silo wall and this is validated to be the
main cause of structural buckling of silo.
The radial displacement plotted in Fig. 23 is
shown to increase from the top, which reaches its first
extremum value w1 at the critical height zcr1 and then
decreases slightly until reaches its second extremum SILO4 SILO45 SILO5
value w2 at the critical height zcr2 and vanishes rapidly
Fig. 24 Buckling modes of the example silos in the LBA
at the bottom. The maximum radial displacement
occurs at the critical height zcr2 and is closely de-
pendent on the slenderness of the example silos, and
increases significantly with the decrease of silo
slenderness.
7.2 Buckling modes
The results show that the buckling deformations
from both the linear and nonlinear buckling analyses
are asymmetrical, deviating from the center to the SILO4 SILO45 SILO5
side where the most friction locates to the highest wall Fig. 25 Buckling modes of the example silos in the
contact. The linear buckling modes from LBA plotted GMNIA
in Fig. 24 show that the deformations take the form of
several meridional waves with assorted magnitudes in
7.3 Critical buckling factor
the single side. The number of meridional waves is
reduced with the decrease of silo slenderness. The The load-displacement equilibrium paths in
nonlinear buckling deformations for geometrically large eccentricity filling are also highly nonlinear, and
perfect models in the GNA, MNA, and GMNA are the critical buckling load factors λcr in each type of
similar and occur only at the wall area adjacent to the buckling analysis are listed in Table 8 for the evalu-
silo bottom, forming so-called plastic “elephant’s ated example silos. The table shows that the critical
failure” modes with consideration of material non- load factor from the GMNIA has the least value for all
linearity in the GMNA. example silos, and the critical buckling factor λcr in
The nonlinear buckling deformations for geo- the GMNIA just satisfies the requirement of the
metrically imperfect models in the GNIA and overall safety factor 1.65 which is consequently taken
GMNIA are also similar and are characterized by a as the buckling resistance factor λR of the example
certain number of harmonic waves in the axial direc- silos in large eccentricity filling. It is also indicated in
tion throughout the whole height of the silo wall, and Table 8 that the geometrical effect is favorable for the
the number of harmonic waves is almost the same as critical buckling factor λcr for all slendernesses of
that of the weld imperfections. Note that the harmonic example silos making about a 5% improvement of the
deformations are also asymmetrical with a much load factor, whereas the material nonlinearity and
larger amplitude of deformation on the side where weld imperfection in the wall are detrimental. The
most friction locates compared to the other side. In material nonlinearity has the most detrimental effect,
and accounts for a loss of about 30%–50% of the substantially, which is approximately 70% that of
buckling factor λcr. silos under concentric discharge.
For instance, the RCS index for the intermediate
7.4 Designed wall thickness and steel consumption
slender silo (SILO4) is about 282 for concentric dis-
index
charge, but the value decreases to about 182 for large
The designed wall thickness distribution corre- eccentric filling, a reduction of about 35%. Similarly,
sponding to the critical buckling factor λcr in Table 8 the RCS index for the squat silo (SILO5) is about 450
and used for buckling analysis is given in Table 9 for concentric discharge, but the value decreases to
together with the RCS index. It shows that the de- about 312 for large eccentric filling, a reduction of
signed bottom thickness of the tapered wall silo under about 30%. Furthermore, for the squat silo (SILO45),
large eccentricity filling is about 1.5 times that of silos the RCS index is 238 which is approximately 1.31
under the concentric solid pressure, while the de- times that of SILO4 and 0.76 times that of SILO5 for
signed top thickness remains the same. It indicates large eccentric filling. The conclusion can be drawn
that the large eccentric filling pressure has consider- that the squat or retaining silos with small slenderness
able impact on the buckling resistance of the silo, have the best storage efficiency from the economical
whose effect is mainly centered around the lower part viewpoint. The small slendernesses of silos such as
of the wall. The amount of steel used in example silos squat and retaining silos are also preferred in large
ranges from about 160 m3 to 274 m3 with the variation eccentric filling.
of slenderness from squat to intermediate slender,
equivalent to a mass of between approximately 1200 t
and 2200 t, which is about 1.5 times that of silos under
concentric discharge. In such cases, the fabrication
and flexure of the silo wall also become even more
complicated during construction.
The relative relation of economical RCS index is
plotted in Fig. 26 for example silos under large ec-
centric filling together with that under concentric
discharge for ease of comparison. It shows that the
two groups of RCS index share the similar variable
tendency which increases rapidly with the decrease of
silo slenderness. Nevertheless, the RCS index of Fig. 26 Relationship of RCS index with slenderness of
example silos under large eccentric filling is reduced example silos
Table 8 Critical buckling factors for example silos from proposed buckling analysis types under large eccentricity
filling
SILO45 Squat 3.45 3.55 3.01 2.94 1.96 1.73
SILO5 Squat 2.87 3.05 2.24 2.44 1.88 1.67
Table 9 Designed wall thickness and economic index of example silos under large eccentricity filling
Diameter, Height, Slenderness, ttop tbot teq Capacity Steel con-
No. RCS
dc (m) hc (m) hc/dc (mm) (mm) (mm) (m3) sumption (m3)
SILO4 35 51.97 1.49 6.0 90.0 48.0 50 000 274.29 182.29
SILO45 40 39.79 0.99 6.0 78.0 42.0 50 000 210.01 238.09
SILO5 45 31.44 0.70 6.0 66.0 36.0 50 000 160.02 312.48
8 Conclusions the solid pressures other than other influential factors

such as weld imperfection and geometrical and ma-
This paper presented a comprehensive study of terial nonlinearity, compared with the load case of
the buckling behavior of various slendernesses of concentric pressure. The linear and nonlinear buck-
large circular steel silos with capacity of 30 000– ling deformations in large eccentric filling are also
60 000 m3, under the load cases of concentric dis- asymmetrical, deviating from the center to the side
charge, large eccentricity discharge, and large eccen- where the most friction locates to the highest wall
tricity filling. The studies are demonstrated in ac- contact.
cordance with Eurocode, by undertaking six types of 6. Large eccentricity both in discharge and filling
buckling analysis on account of the geometrical and would result in a strong decrease of storage efficiency
material nonlinearity and weld imperfection. The of a steel silo. The RCS index of example silos under
following conclusions can be drawn: large eccentricity discharge and filling is reduced
1. The load-displacement curves for nonlinear substantially, and is approximately half and 70% that
buckling analysis are highly nonlinear and predict a of silos under concentric discharge, respectively.
distinct maximum load followed by a descending path.
The silo with smaller slenderness is superior in References
structural ductility and plasticity to that with a larger ANSYS, 2008. ANSYS User’s Manual. Swanson Analysis
slenderness, and the retaining silo or squat silo is Systems Inc., Houston, USA.
BSI (British Standards Institution), 2002a. Basis of Structural
preferred in practical structural design.
Design, BS EN 1990:2002. BSI, UK.
2. The material nonlinearity is strong and det- BSI (British Standards Institution), 2002b. Actions on Structures-
rimental to buckling behavior of all slendernesses of Part 1-1: General Actions–Densities, Self-weight, Imposed
example silos, resulting in a decrease of buckling Loads for Buildings, BS EN 1991-1-1:2002. BSI, UK.
resistance for both perfect and imperfect models. The BSI (British Standards Institution), 2005. Design of Steel
Structures-Part 1-1: General Rules and Rules for Build-
effect of geometrical nonlinearity and weld imper-
ings, BS EN 1993-1-1:2005. BSI, UK.
fection on the buckling behavior of steel silos is BSI (British Standards Institution), 2006. Actions on Structures-
somewhat complicated, but is closely correlated with Part 4: Silos and Tanks, BS EN 1991-4:2006. BSI, UK.
silo slenderness, the distribution of solid pressures on BSI (British Standards Institution), 2007a. Design of Steel
silo wall, etc. Structures-Part 1-6: Strength and Stability of Shell Struc-
3. The economic index called RCS provides an tures, BS EN 1993-1-6:2007. BSI, UK.
BSI (British Standards Institution), 2007b. Design of Steel
effective measure for the design of steel silos. The Structures-Part 4-1: Silos, BS EN 1993-4-1:2007. BSI, UK.
results show that the RCS index increases rapidly Gillie, M., Rotter, J.M., 2002. The effects of patch loads on
with the decrease of silo slenderness, where the thin-walled steel silos. Thin-Walled Structures, 40(10):
storage efficiency of a steel silo improved greatly 835-852.
with the slenderness varying from slender silo to http://dx.doi.org/10.1016/S0263-8231(02)00028-9
Gillie, M., Holst, J.M.F.G., 2003. Structural behaviour of silos
retaining silo. supported on discrete, eccentric brackets. Journal of
4. The wall thickness type and its distribution Constructional Steel Research, 59(7):887-910.
have the most important influence on the buckling http://dx.doi.org/10.1016/S0143-974X(02)00078-0
modes and buckling strength of steel silos. Under the Greiner, R., Guggenberger, W., 1998. Buckling behaviour of
condition of the same amount of steel used in con- axially loaded steel cylinders on local supports-with and
without internal pressure. Thin-Walled Structures, 31(1-3):
struction, different schemes for wall thickness ar- 159-167.
rangement are suggested in buckling design of steel http://dx.doi.org/10.1016/S0263-8231(98)00011-1
silos to achieve the optimal buckling strength. ISO (International Organization for Standardization), 1995.
5. The buckling deformations from both the Basis for Design of Structures-Loads Due to Bulk Mate-
linear and nonlinear buckling analyses in large ec- rials, ISO 11697:1995. ISO.
Iwicki, P., Tejchman, J., Chroscielewski, J., 2014. Dynamic FE
centric discharge are strongly asymmetrical arising
simulations of buckling process in thin-walled cylindrical
from the circumferential and meridional non-uniform metal silos. Thin-Walled Structures, 84:344-359.
distribution of the solid pressures. The buckling is http://dx.doi.org/10.1016/j.tws.2014.07.011
mainly governed by the non-uniform distribution of Kim, S.E., Kim, C.S., 2002. Buckling strength of the
cylindrical shell and tank subjected to axially compres- http://dx.doi.org/10.1016/j.tws.2004.05.009

sive loads. Thin-Walled Structures, 40(4):329-353. Song, C.Y., Teng, J.G., 2003. Buckling of circular steel silos
http://dx.doi.org/10.1016/S0263-8231(01)00066-0 subject to code-specified eccentric discharge pressures.
Li, B.R., Wang, X.Y., Ge, H.L., et al., 2005. Study on ap- Engineering Structures, 25(11):1397-1417.
plicability of modal analysis of thin finite length cylin- http://dx.doi.org/10.1016/S0141-0296(03)00105-6
drical shells using wave propagation approach. Journal of Standards Australia, 1996. Loads on Bulk Solids Containers,
Zhejiang University-SCIENCE A, 6(10):1122-1127. AS 3774-1996. Standards Australia, Sydney, Australia.
http://dx.doi.org/10.1631/jzus.2005.A1122 Zhao, Y., Cao, Q.S., Xie, X.Y., 2006. Floating-roof steel tanks
Pircher, M., Bridge, R.Q., 2001. Buckling and post-buckling under harmonic settlement: FE parametric study and de-
behaviour of silos and tanks under axial load-some new sign criterion. Journal of Zhejiang University-SCIENCE
aspects. Journal of Structural Engineering, 127(10): A, 7(3):398-406.
1129-1136. http://dx.doi.org/10.1631/jzus.2006.A0398
http://dx.doi.org/10.1061/(ASCE)0733-9445(2001)127:10
(1129)
Rotter, J.M., Teng, J.G., 1989. Elastic stability of cylindrical
shells with weld depressions. Journal of Structural En-
中文概要
gineering, 115(5):1244-1263.
http://dx.doi.org/10.1061/(ASCE)0733-9445(1989)115:5 题目：多种长细比的大型钢筒仓结构的屈曲设计
(1244) 目的：大型钢筒仓结构在正常装卸料过程中极易发生各
Rotter, J.M., Zhang, Q., 1990. Elastic buckling of imperfect
种失稳破坏。本文旨在分析多种长细比的大型钢
cylinders containing granular solids. Journal of Structural
Engineering, 116(8):2253-2271. 筒仓结构在散料荷载作用下的屈曲性能，为结构
http://dx.doi.org/10.1061/(ASCE)0733-9445(1990)116:8 的稳定设计提供理论依据及技术支持。
(2253) 创新点：1. 提出散料荷载作用下大型钢筒仓结构屈曲分析
Sadowski, A.J., Rotter, J.M., 2011a. Buckling of very slender 的 4 种典型工况：轴对称卸料荷载工况、小偏心
metal silos under eccentric discharge. Engineering 卸料荷载工况、大偏心卸料荷载工况和大偏心装
Structures, 33(3):1187-1194.
料荷载工况；2. 建立了适用于不同屈曲分析类型
http://dx.doi.org/10.1016/j.engstruct.2010.12.040
Sadowski, A.J., Rotter, J.M., 2011b. Steel silos with different 的多种数值模型，研究了结构的稳定性能，并分
aspect ratios-I: behaviour under concentric discharge. 析了各种非线性、初始几何缺陷、仓壁厚度的分
Journal of Constructional Steel Research, 67(10):1537- 布和基底嵌固刚度等对结构稳定性能的影响。
1544. 方法：1. 根据不同工况下散料荷载分布的特点和屈曲分
http://dx.doi.org/10.1016/j.jcsr.2011.03.028 析类型的差异，通过非线性有限元方法，研究结
Sadowski, A.J., Rotter, J.M., 2011c. Steel silos with different
构在四种散料荷载工况下的稳定性能；2. 通过研
aspect ratios-II: behaviour under eccentric discharge.
究结构在各种非线性屈曲分析下的荷载-位移全
Journal of Constructional Steel Research, 67(10):1545-
1553. 过程响应，确定结构的屈曲临界荷载和结构的稳
http://dx.doi.org/10.1016/j.jcsr.2011.03.027 定承载力；3. 比较各种荷载工况下结构的屈曲临
Sadowski, A.J., Rotter, J.M., 2012. Structural behavior of 界荷载、屈曲模态及仓壁厚度分布特点，分析结
thin-walled metal silos subject to different flow channel 构用钢量指标的变化。
sizes under eccentric discharge pressures. Journal of 结论：1. 非线性屈曲分析的荷载-位移曲线是高度非线性
Structural Engineering, 138(7):922-931.
的，不同长细比钢筒仓的屈曲平衡路径之间有较
http://dx.doi.org/10.1061/(asce)ST.1943-541X.0000530
Sadowski, A.J., Rotter, J.M., 2013. Buckling in eccentrically 大差异。2. 钢筒仓结构的屈曲延性随长细比的减
discharged silos and the assumed pressure distribution. 小而显著增大。3. 材料非线性对结构的稳定性极
Journal of Engineering Mechanics, 139(7):858-867. 为不利；几何非线性和初始几何缺陷对结构稳定
http://dx.doi.org/10.1061/(asce)EM.1943-7889.0000525 性能的影响与结构的长细比密切相关。4. 壁厚的
Song, C.Y., 2002. Buckling of un-stiffened cylindrical shell 分布对钢筒仓的稳定承载力和屈曲模态影响很
under non-uniform axial compressive stress. Journal of 大；设计时应考虑仓壁壁厚的不同组合形式，并
Zhejiang University-SCIENCE, 3(5):520-531.
确定使结构取得最大稳定承载力的仓壁最优分
http://dx.doi.org/10.1631/jzus.2002.0520
Song, C.Y., 2004. Effects of patch loads on structural behavior 布形式。5. 装料或卸料过程中的偏心使结构的储
of circular flat-bottomed steel silos. Thin-Walled Struc- 存效率（容耗比指标）显著降低。
tures, 42(11):1519-1542. 关键词：钢筒仓；壳体；偏心卸料；屈曲稳定；非线性

Download

Uploaded by

Copyright:

Available Formats

Download

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Download

Uploaded by

Copyright:

Available Formats

282 Cao and Zhao / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2017 18(4):282-305

Journal of Zhejiang University-SCIENCE A (Applied Physics & Engineering)

Buckling design of large steel silos with various slendernesses*

Qing-shuai CAO, Yang ZHAO†‡

1 Introduction Depending on the geometry and capacity of steel

Table 2 Design conditions assumed for the numerical

Table 1 Geometries of the example silos with capacities of 30 000–60 000 m3

3 Numerical analysis procedure

(DOFs), including translations in the nodal x, y, and z

SILO1 SILO2 SILO3

66 SILO3 SILO4 0.6 SILO4

SILO5 SILO6 SILO5

5.0 SILO3 SILO4

earity is strong and detrimental to buckling behavior

5.4 Effects of nonlinearity and imperfection

5.5 Designed wall thickness and steel consumption

the consumption of materials in the construction of

useful parameter in the design of steel silos.

and deformation are highly unevenly distributed not

Fig. 18 Distribution of radial displacement of example

Fig. 19 Buckling modes of the example silos in LBA

8 Conclusions the solid pressures other than other influential factors

cylindrical shell and tank subjected to axially compres- http://dx.doi.org/10.1016/j.tws.2004.05.009

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.