CHALLENGE JOURNAL OF STRUCTURAL MECHANICS 5 (2) (2019) 72–79

Research Article

Arch effect in silos on discrete supports - Is it a myth or reality?

Lyubomir A. Zdravkov *
Department of Metal, Wood and Plastic Structures, University of Architecture, Civil Engineering and Geodesy (UACEG), Sofia 1046, Bulgaria

ABSTRACT ARTICLE INFO

Steel silos are interesting, complicated facilities. In order to assure its complete emp- Article history:
tying by gravity they are often placed on supporting frame structure above the Received 10 March 2019
ground. Values of stresses in joints between thin walled shell and supporting frame Revised 3 May 2019
elements are very high. It can cause the local buckling in the shell. The simplest way Accepted 8 June 2019
to design steel silos is to divide hypothetically the cylindrical shell into two parts -
Keywords:
ring beam, supported in some points and shell above, uniformly supported. This con-
Steel silo
ception is accepted by European Standard EN 1993-4-1. The particular moment is Meridional stress
that the ring beam and cylindrical body above it are separated. Actually the two ele- Ring beam
ments are jointed and work together in the same time. Considering the last results of Arch effect
Zeybek, Topkaya and Rotter from 2019, and as well as his own research, the author Vertical stiffening
asks the question if it is true that the transferring of discrete base reactions to the Buckling
cylindrical body is done by bending work of the ring beam, which is the conception
in EN 1993-4-1? Or the vertical reaction forces are actually redistributed on the
height based on the work of the cylindrical shell under compression as an arch. Using
the contemporary capabilities of the programs for spatial analysis of building struc-
tures the author will try to find the answer of this question.

1. Introduction

Usually the steel silos are elevated above the ground

facilities, placed on supporting structure. The purpose is
to assure the easy and complete emptying of the stored
product by gravity. The supporting structure is different
for every facility because it depends on the real condi-
tions of exploitation. The most used are the two types -
built by horizontal girders and columns or only by col-
umns. Both type of frame structures cause concentrated
meridional forces in the cylindrical body of the silo. As a
result the thin wall shell could buckles.
The simplest way to design steel silos is to divide hy-
pothetically the cylindrical shell into two parts - dis-
cretely supported ring beam and uniformly supported
shell above it, see Fig. 1. This conception is accepted by
the European standard EN 1993-4-1. Obviously, to as-
sure uniform supporting on the whole circumference of
the circular shell, bending stiffness of the ring beam
should be high. Unfortunately the EN 1993-4-1 does not
say what should be the recommended stiffness of the Fig. 1. Traditional design model for silos on discrete
ring beam. supports.

* Corresponding author. Tel.: +359-885-081-305 ; E-mail address: [email protected] (L. A. Zdravkov)

ISSN: 2149-8024 / DOI: https://doi.org/10.20528/cjsmec.2019.02.004
 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79 73

Rotter (1985) suggested that a value of ratio ψ = 0.25 For simplification, the Eq. (6) could be represented by
might be suitable for adoption in design, where: two simple relations:
𝐾shell 𝜂
ψ= (1) , when 𝐻 ≤ 𝐻𝑐𝑟
𝐾ring 𝑓𝑠 = { 3 (10)
1.0, when 𝐻 > 𝐻𝑐𝑟
in which:
Kshell is stiffness of cylindrical shell; where Hcr is critical height of cylindrical shell. It could be
Kring is stiffness of ring beam. determined by formula:
Based on English translation of study of Vlasov (1961)
4
about of curved beams, stiffness of ring beam Kring is ex- 𝐻𝑐𝑟 =
3 √3 𝑅
√ 𝑅 (11)
pressed as: 𝑛√(𝑛 2 −1) 𝑡

(𝑛 2 −1) 𝐸𝐼𝑟 1
2 Hcr represents the height of shell which is effective of
𝐾ring = 𝑅4 𝑓𝑟
(2) redistributing of discrete forces from supports and
equalizing of axial normal stresses. When height of shell
where: H ≤ Hcr, entire shell resists axial loads from supports.
n is number of uniformly spaced supports; When H > Hcr, only that part between bottom of shell and
E is modulus of elasticity; critical height Hcr is effective in redistributing of vertical
Ir is moment of inertia about a radial axis; reactions from discrete columns.
R is radius of ring beam centroid. In their researches Topkaya and Rotter (2011a)
(2011b) conducted extensive finite element analyses for
𝑓𝑟 = 1 + 𝑛2𝐾𝑟
𝐸𝐼
(3) verification of Rotter’s criterion about stiffness of ring
𝑇 beam. With 1,280 separate finite-element analyses
(FEA), covering two different types of ring sections, var-
in which: ious heights and radii of cylindrical shells, the authors
𝐸𝐶𝑤
checked validity of suggested by Rotter (1985) ratio ψ =
𝐾𝑇 = 𝐺𝐽 + 𝑛2 𝑅2
(4) 0.25. On basis of done FEA they concluded, when a stiff-
ness ratio ψ ≤ 0.1, axial stresses will not deviate more
where: than 25% from the uniform support assumption.
G is shear modulus; Research of Zeybek, Topkaya and Rotter (2019)
J is torsional constant; shows that the equations, based on the theory of Vlasov
Cw is warping constant for an open sections. (1961) for a curved beam provide results with accepta-
Semi-membrane theory of shells, proposed by Vlasov ble accuracy when the girder is separated from the cylin-
(1964), gives an expression of stiffness of cylindrical drical body. When the ring beam and cylindrical shell are
shell, as follow: jointed, the received through finite elements analysis
values are considerably different from the analytical re-
𝐸 𝑡 3/2 1 sults in closed form. The differences going high with in-
𝐾shell = 𝑛√(𝑛2 − 1) 4 ( ) (5) crease of the thickness of the cylindrical shell.
√3 𝑅 𝑓𝑠
It should be noted that all above mentioned re-
where t is thickness of the cylindrical shell. searches are conducted on the smooth steel shells with-
out vertical stiffeners on them. On other side, common
(𝑒 η )2 −2𝑒 η sin(η)−1
𝑓𝑠 = (𝑒 η )2 (6) practice in design of steel structures is to place stiffening
−2𝑒 η cos (η)+1 elements on the point, where are applied concentrated
loads. In our case, the vertical stiffeners should be placed
in which: above the discrete supports, see Fig. 2.
2𝜋𝐻
η= (7)
μ

where:
H is height of cylindrical shell;
μ is expressed by Calladine (1983) long wave bending
half-wavelength:

4
2𝜋 √3 𝑅
𝜇= √ 𝑅 (8)
𝑛√(𝑛 2 −1) 𝑡

Based on Eqs. (2) and (5), stiffness ratio ψ will look

like as:

𝐾shell 0.76 (𝑅𝑡)2 𝑅 𝑛2 𝑓𝑟 Fig. 2. Stiffening elements above discrete supports of

𝜓= 𝐾ring
= 𝐼𝑟
√ √(𝑛2
𝑡 −1)3 𝑓𝑠
(9) the shell.
74 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79

In his research Zdravkov (2017a, 2018) shows that f) The heights of the openings ho are different, depending
vertical stiffening elements increase the height of the on the height of the stiffeners. They are calculated ac-
critical zone, where the vertical reactions of discrete cording to the simple formula:
supports are redistributed. Considering the last results
of Zeybek, Topkaya and Rotter (2019), as well as his own ℎ𝑜 = 𝐻𝑠 − 100 mm (12)
research, the author put the question if it is true that re-
distributing of the separate reactions of supports is done where Hs is the height of vertical stiffeners, see Fig. 6.
by the bending of ring beam which is the conception in
EN 1993-4-1, see Fig. 1. Or vertical forces in discrete sup-
ports are transferred on the height by work of the cylin-
drical shell above the supports as a compressed arch, see
Fig.3.
In this article the author will try to find answer of
these questions.

Fig. 4. Numerical models – dimensions and loading.

Fig. 3. Compression forces in the cylindrical shells
above the supports.

2. Finite Element Analysis

For the purpose of research, three steel cylindrical

shells are modelled, using software ANSYS. Their param-
eters are as follow:

а) Dimensions: а) shell without opening

- shell 1 – diameter D = 3 m, height H = 6 m;
- shell 2 – diameter D = 4 m, height H = 8 m;
- shell 3 – diameter D = 5 m, height H = 10m.

b) All shells are with constant thickness t = 5 mm;

c) All shells are supported by six immovable supports

with dimensions in plane 125×125 mm, see Fig. 4.
b) shell with openings
d) In Fig. 4 every support are placed two vertical steel
plates with section 8x100 mm and with different height. Fig. 5. Vertical stiffeners on the cylindrical shell.
On their upper end exists an intermediate ring with a
section L100x8 mm, see Figs. 4 and 5. g) In order to strengthen the shells in radial direction, on
50mm above the lower edge and on 50mm below the up-
e) Cylindrical body in three models of the shells is con- per edge are placed rings with section L100x8 mm,
tinuous, see Fig. 5(а). In the other three models are made welded as is shown on Fig. 7.
openings between the vertical stiffeners, see Fig. 5(b).
On this way the “skirt” of the silo cannot works as a ring h) The stored in the facilities product varies. For each
beam. The support’s forces will be transferred up on the shell it is as follow:
cylindrical body only through work of the cylindrical - shell 1 – cement;
shell above the supports on compression, as an arch, see - shell 2 – lime;
Fig. 3. - shell 3 – sand.
 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79 75

- at ideal position of the intermediate stiffening ring on

the shell.
Topkaya and Rotter (2014) determined the ideal po-
sition of intermediate stiffening rings on the shell. They
expect a ring, placed at this ideal position, can effectively
remove all circumferential nonuniformity in the axial
membrane stress above it. The simple expression of
ideal location HI is:
𝑅
𝐻𝐼 = √12(1 + 𝜈) (14)
𝑛

where ν is a coefficient of Poisson.

Fig. 6. Opening in the base of the cylindrical shell. - using an average value of distribution of discrete
forces FR from supports α = 30°. The height H30 should be
calculated by the formula:
𝜋𝑅
𝐻30 = tan(90° − 𝛼°) (15)
𝑛

- the length of the stiffeners HL is equal to distance be-

tween the supports. It is calculated according to the for-
mula:
2𝜋𝑅
𝐻𝐿 = 𝑛
(16)

Fig. 7. Shape of the intermediate stiffening ring. j) Material of elements is steel S235, with a properties
according to European standard EN 10025-2:2004.
Necessary stiffness of intermediate stiffening rings is
Every product causes vertical load Pwf due to the fric- determined by Zeybek et al. (2015). Stiffness ratio χ
tion between the stored material and the shell. Its values could be expressed as:
are determined for every particular product according to
standard EN 1991-4. All loads are uniformly distributed 𝐾shell 𝑅𝑡(𝐴𝑅 2 +𝐼𝑥 𝑛 2 (𝑛 2 −1))
and applied as a surface pressure on the shell. They are 𝜒=𝐾 = 12√3(1+𝜈)3/2𝐴𝐼 2 2 (17)
stiffener 𝑥 𝑛(𝑛 −1)
applied to internal surface of the shells, see Fig. 4.
where:
i) Shells 1, 2 and 3 are analysed for four different heights Kshell is circumferential stiffness of the shell;
Hs of vertical stiffeners above supports. Kstiffener is circumferential stiffness of circular ring;
The heights reached by the stiffening plates are deter- A is cross sectional area of the stiffening ring;
mined as follows: Ix is moment of inertia of the stiffening ring about verti-
- using an average value of distribution of discrete cal axis "x-x".
forces FR from supports α = 45°, see Fig. 8. The height H45 The results in research of Zeybek et al. (2015) indicate
is determined with the expression: that ratios below about χ < 0.2 provide a satisfactorily
uniform axial membrane stress distribution above the
𝜋𝑅
𝐻45 = (13) intermediate ring stiffener, so this limit is recommended
𝑛
for practical design. In his later research Zeybek et al.
where R is radius of cylindrical shell. (2017) confirmed, that correlation smaller than χ < 0.2
are sufficient even when the rings are placed under their
ideal position.
The steel angle section L100x8 and a part of the cylin-
drical shell form an intermediate stiffening ring with a
shape as is shown on Fig. 7.
Effective width l of the steel sheets over and below the
joint is calculated according to the standard API 650, by
the expression:

𝑙 ≤ 13.4√𝐷𝑡 (18)

where:
D is a diameter of the cylindrical shell, m;
Fig. 8. Average angle α of distribution of the t is thickness of the cylindrical shell, mm.
compressive forces on height.
76 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79

Effective width l for the shells with the smallest diam- Axial normal stresses are accounted by the height of
eter, D = 3 m, is l = 51.9 mm. The author accepts to have shell, in the middle between two supports and above the
effective width l = 50 mm for all shells. It is on way of supports. After that are determined the values of ratio
safety. σx,m/σx,s , where:
The geometric characteristics of the obtained stiffen- σx,m is meridional normal stress by height of the cylinder,
ing ring are: in the middle between two supports;
a) Area - A = 20.5 cm2; σx,s is meridional normal stress by height, above the sup-
b) Moment of inertia about vertical axis „х-х“- Ix=358.4 ports.
cm4. The idea is that where the ratio σx,m/σx,s =1.0 , is the
For different shells, the ratio of the stiffness’s χ, calcu- upper border of the critical zone in the shell, in which are
lated according to the Eq. (17), has the values as follow: redistributed vertical reactions of supports. Above that
- shell 1 – χ = 0.042; border circumferential nonuniformity in the axial mem-
- shell 2 – χ = 0.0764; brane stresses does not exists and the shell is continu-
- shell 3 – χ = 0.130. ously supported.
The maximum value of the ratio χ = 0.130 < 0.2, so it On the second stage, in the used program ANSYS is ac-
could be expected that the stiffness of the intermediate tivated the option “Buckling Analysis”. Through this op-
ring will be sufficient to equalize the meridional stresses tion it is possible to calculate the reserve of bearing ca-
in the shell above it. pacity k of the cylindrical shell before that it losses sta-
The shells are modeled by 2D quad elements shell181 bility, completely or partially. The reserve k gives a quan-
with a maximum length of side 50 mm. The method of tity assessment of the influence of the made openings on
their creation is “All quad”. Element’s midside nodes are the bearing capacity of the shell.
controlled by the program.
Thin shell structures are sensitive for effect of
changes of geometry during loading. On that reason geo- 3. Results and Discussion
metrically nonlinear analyses (GNA) are used, according
to the recommendations of EN 1993-1-6. The graphics below, see Figs. 10-12, show the changes
ANSYS’s option “symmetry” is activated to reduce a of ratio σx,m/σx,s. by height, calculated using numerical
calculation time. In analysis is used a quarter of silo only, methods for analysis.
see Fig. 9.

а) without openings in the ring beam

b) with openings in the ring beam

Fig. 10. Change of ratio σx,m/σx,s by the height
Fig. 9. Quarter of silo, used in numerical analysis. of the cylindrical shell 1 (D = 3 m, H = 6 m).
 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79 77

а) without openings in the ring beam а) without openings in the ring beam

b) with openings in the ring beam b) with openings in the ring beam
Fig. 11. Change of ratio σx,m/σx,s by the height Fig. 12. Change of ratio σx,m/σx,s by the height
of the cylindrical shell 2 (D = 4 m, H = 8 m). of the cylindrical shell 3 (D = 5 m, H = 10 m).

Obviously removing of material from the space be- a) shell 1 – diameter D = 3 m, height H = 6 m
tween the supports, see Fig. 5 and 6, has its influence. At Buckling reserve capacity k
least the diagrams showing the changes of the ratio Height of
shell without shell with
σx,m/σx,s by the height of shell have different shape. More stiffeners
openings openings
important is that the values of the stresses σx,m and σx,s in H45 = 785 mm 31.416 30.675
models with openings become equal on the bigger HI = 987 mm 36.381 34.703
height. It means that the ways of transfer of compression H30 = 1360mm 42.644 41.416
stresses by the height are different for the continual bod- HL = 1 571 mm 46.33 44.558
ies and for the bodies with openings.
Moreover, if on the models shown on Figs. 5(b) and 6 b) shell 2 – diameter D = 4 m, height H = 8 m
the base reactions obviously are redistributed on the Buckling reserve capacity k
height only through the work on compression of the cy- Height of
shell without shell with
stiffeners
lindrical shell as an arch, this is not the same for the shell openings openings
without openings. From where it can be concluded that H45 = 1047mm 24.943 23.468
the concept shown in EN 1993-4-1 has a grain rational- HI = 1317 mm 28.809 26.897
ity. H30 = 1814mm 34.478 34.021
On the Figs. 10-12 we can see ratio σx,m/σx,s > 1.0. It HL = 2095mm 38.277 36.243
means that in the part of the shell meridional stresses in
the middle, between supports, are bigger than the c) shell 3 – diameter D = 5 m, height H = 10m
stresses above the supports. The similar phenomena Buckling reserve capacity k
Height of
was observed in the previous research of Zdravkov shell without shell with
stiffeners
(2017a) and (2017b). openings openings
The results of carried out Buckling Analysis, reporting H45 = 1309mm 5.434 5.363
buckling above the vertical stiffeners and stiffening ring HI = 1646 mm 6.463 6.306
above them, see Fig. 13(а), for every one shell are as fol- H30 = 2267mm 7.951 7.671
low: HL = 2618 mm 8.857 8.374
78 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79

It gives impression that the reserve of the bearing ca- It is important to notice that when the cylindrical
pacity k in the tight shell is always bigger. Which is an- shells without openings are researched, the first form of
other proof that the redistributing of discrete base reac- buckling is always caused by shearing. The area of buck-
tions on the height is not only achieved by compression ling is in both sides of the vertical stiffeners, see Fig.
forces as it is shown on Fig. 3. 13(a).

а) buckling due to shearing

b) buckling due to compression

Fig. 13. Modes of buckling in the cylindrical shells.

4. Conclusions

The current research, made for six cylindrical shells dividing the cylindrical body of the silo into discretely
on discrete supports, shows that the continual shells supported ring beam and cylindrical body there is some
have different behaviour than the shells with openings truth. For that reason the body of silo above the supports
between the supports. should be checked for:
From where could be concluded that the reactions of  buckling in the area above vertical stiffeners, caused
discrete supports are not distributed on the height only by meridional (axial) forces;
by work of the cylindrical shell on compression above  buckling to left or right of vertical stiffeners, due to
them, as an arch. In the conception of EN 1993-4-1 for shearing forces.
 Zdravkov / Challenge Journal of Structural Mechanics 5 (2) (2019) 72–79 79

REFERENCES

ANSYS 17 (2016). Ansys Inc., Canonsburg, PA, USA.

API Standard 650 (2013). Welded Tanks for Oil Storage, Twelfth Edi-
tion. American Petroleum institute.
Calladine CR (1983). Theory of Shell Structures. Cambridge University
Press, Cambridge, U.K.
EN 10025-2 (2004). Hot rolled products of structural steels - Part 2:
Technical delivery conditions for non-alloy structural steels. Euro-
pean Committee for Standardization, Brussels.
EN 1991-4 (2006). Eurocode 1 - Actions on structures - Part 4: Silos
and tanks. European Committee for Standardization, Brussels.
EN 1993-1-6 (2007). Design of steel structures - Part 1-6: Strength and
stability of shell structures. European Committee for Standardiza-
tion, Brussels.
EN 1993-4-1 (2007). Design of steel structures - Part 4-1: Silos. Euro-
pean Committee for Standardization, Brussels.
Rotter JM (1985). Analysis and Design of Ringbeams. In “Design of Steel
Bins for Storage of Bulk Solids”, J. M. Rotter, ed., University of Sydney,
Sydney, Australia, 164 –183.
Topkaya C, Rotter JM (2011a). Ring beam stiffness criterion for column
supported metal silos. ASCE Journal of Engineering Mechanics, 134,
846-853.
Topkaya C, Rotter JM (2011b). Stiffness of silo supporting ring beams
resting on discrete supports. 6th International Conference on Thin-
Walled Structures, Timisoara, Romania.
Topkaya C, Rotter JM (2014). Ideal location of intermediate ring stiff-
eners on discretely supported cylindrical shells. Journal of Engi-
neering Mechanics, 140(4), 04013001, 1-10.
Vlasov VZ (1961). Thin-Walled Elastic Beams. National Science Foun-
dation, Washington, DC.
Vlasov VZ (1964). General theory of shells and its applications in engi-
neering. NASA Technical Translation, TTF-99, Washington, DC.
Zdravkov LA (2017a). Vertical stiffeners and internal pressure - influ-
encing factors on distribution of meridional stresses in steel silos
on discrete supports. Challenge Journal of Structural Mechanics,
3(3), 123-128.
Zdravkov LA (2017b). Influence of intermediate rings and height of
skirt on effective width of compression zone in junction column -
cylindrical shell of steel silo. International Jubilee Scientific Confer-
ence "75th Anniversary of UACEG", Sofia.
Zdravkov LA (2018). Influencing factors on effective width of com-
pressed zone in joint column - cylindrical shell of steel silo. Chal-
lenge Journal of Structural Mechanics, 4(1), 1-8.
Zeybek Ӧ, Topkaya C, Rotter JM (2015). Strength and stiffness require-
ments for intermediate ring stiffeners on discretely cylindrical
shells. Thin-Walled Structures, 96, 64-74.
Zeybek Ӧ, Topkaya C, Rotter JM (2017). Requirements for intermediate
ring stiffeners placed below the ideal location on discretely sup-
ported shells. Thin-Walled Structures, 115, 21-33.
Zeybek Ӧ, Topkaya C, Rotter JM (2019). Analysis of silo supporting ring
beams resting on discrete supports. Thin-Walled Structures, 135,
285-286.