Professional Documents
Culture Documents
Background Discussion Document For EN 1991-4: List of Issues
Background Discussion Document For EN 1991-4: List of Issues
CEN TC250/SC1/PT4
This document has been prepared as a discussion paper indicating the basis of
changes and amendments to the ENV 1991-4 undertaken by the Project Team. It is
mostly a summary of discussions at the meetings in Copenhagen (June 2001) and
Brussels (November 2001), together with reactions to comments on the August 2001
draft received from several countries.
The meeting of the PT held in Copenhagen in June 2001 agreed a list of issues
(CEN TC250/SC1/PT4 – 33) that needed to be considered, and this document
addresses those issues.
List of issues
1. Conservatism of the rules
2. Principles and rules (what is compulsory and what is flexible)
3. Main text and annexes: follow CEN rules. Materials testing? What are we
allowed to keep in annexes?
4. Scope: may be OK for concrete silos
5. Reliability classes
6. Material variability
7. Table of material properties
8. Materials testing
9. Mass flow and funnel flow bounds
10. Terminology for maximum load magnifier
11. Value of the maximum load magnifier and physical phenomena it represents
12. Angle of internal friction (which one?)
13. Angle of repose (needed for squat silos)
14. The role of patch loads
15. Separation of filling and discharge eccentricities
16. Patch load rules for large eccentricities
17. High eccentricity flow and pressures
18. Extension of simple rules to wider range (don’t have the restriction of both
eccentricity of discharge and diameter limit to 5m)
19. Simplified rules 5m restriction too small
20. Internal ties in rectangular silos
21. Rectangular silo pressures
22. No distinction between concrete and steel: doesn’t the stiffness matter?
23. Planar silos
24. Coefficient for flat bottoms Cb: is 1.2 too low? Or should be omitted
25. Hopper loads
26. Corrugated walls
27. Pressure strengthening effect for buckling
28. Squat silos separated into squat and intermediate
29. Squat silo rules
30. Squat silos with high filling eccentricity
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
2
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
3
It has been suggested that the rules in ENV1991-4 are often more conservative than
those of DIN1055.
Unless the comparisons identify the chief source of the discrepancies, they do not
assist in determining whether the draft EN is generally too conservative, or whether
this is simply related to a difference of assumed material property value for a
particular solid.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
4
5. Reliability classes
The struggle between simplicity and the complexity required for safe design will
persist until designs are separated into categories of difficulty. The designers of
simple structures want and need a simple code to permit designs to be done quickly
by hand, but large complicated designs require computer calculations and realistically
complicated load cases should be specified to guarantee the safety of these structures.
This conflict is best met by introducing reliability classes. These allow simple
conservative rules to be used for simple cases, but demand more thorough treatment
for cases where either less conservatism is sought (as in a design to be used
repeatedly) or where the size, complexity and high risk mean that more design effort
is both required and justified.
Since the reliability classes are already defined within ENV1993-4-1, it would be
advantageous to adopt the same definitions. The class boundaries are defined using
boxed values, so modifications to these boundaries could also be introduced.
(1) Different levels of rigour should be used in the design of silo structures, depending on the
reliability of the structural arrangement and the susceptibility to different failure modes.
(2) The silo design should be carried out according to the requirements of one of the following three
classes of reliability used in this Part, which produce designs with essentially equal risk in the design
assessment and considering the expense and procedures necessary to reduce the risk of failure for
different structures: Classes 1, 2 and 3.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
5
Silos with capacity in excess of 200 tonnes in which any of the following
design situations occur:
a) eccentric discharge
b) local patch loading
c) unsymmetrical filling
Reliability Class 2 All silos covered by this Prestandard and not placed in another class
Reliability Class 1 Silos with capacity between 10 tonnes† and 100 tonnes
† Silos with capacity less than 10 tonnes are not covered by Eurocode 3 Part 4.1.
(3) A higher Reliability Class may always be adopted than that required in table 2.1.
(4) The choice of minimum Reliability Class should be agreed between the designer, the client and
the relevant authority.
(5) Reliability Class 3 should be used for local patch loading, which refers to a stored solids loading
case causing a patch load which extends round less than half the circumference of the silo, as defined in
ENV 1991-4.
NOTE: Appropriate provisions for Reliability Class 1 silos are set out in annex B.
6. Material variability
Different materials have very different variability. The variability of each material
should be explicitly defined, instead of having the same blanket range applied to
everything. A material that has a high variability places a greater demand on the
structure than one with low variability, since each part of the structure is heavily
loaded at a different property extreme, as shown in the following table.
It would be helpful to include the different load cases defined in this table to aid the
designer in identifying the property extremes that will govern in different stress
resultant assessments and limit state verifications.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
6
Table 7.1 Values of properties needed for different wall loading assessments
Load case Purpose Property extreme to be adopted
Wall friction Lateral pressure Angle of internal
coefficient µ ratio λ friction φi
For the barrel wall
BN Maximum normal pressure on cylinder wall Lower Upper Lower
BF Maximum frictional traction on cylinder wall Upper Upper Lower
HF & HE Maximum vertical load on hopper Lower Lower Upper
The ENV rule insisted that the mean value of a property xm is reduced to 0,9xm and
increased to 1,15xm to allow for material variability.
It was agreed at the meeting of CEN TC250/SC1/PT4 that the principle of including
material variability should be maintained, but that a better calibration of the
variability of individual materials is needed. This calibration should allow for the
proper and appropriate values of the partial coefficients to be used in design.
The principles put in PrEN 1997-1 “Eurocode 7 Geotechnical design: Part 1 General
Rules” Sections 2.4.3 and 2.4.5 have been adopted as far as possible. Concerning
partial factors γF, Eurocode 7 offers 3 different methods for the placement of the
partial factor within the calculation. The one adopted here is Design Approach 2
(annex B), where the load is calculated in the basis of characteristic values of material
parameters and with the partial factor γF placed on the assessed load.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
7
solids are covered by this standard and it is important that the standard states clearly
which value of internal friction is to be used, and gives reasons in the Annex.
In the existing Annex B, the procedure to determine two internal friction angles, φ and
φc, is set out. The value of φ, used elsewhere in the standard corresponds to a soil
mechanics normally consolidated or loading value of internal friction, not the Jenike
construction for the effective angle of internal friction.
However, the notation (section 1.5) states that the value used in the standard is the
effective angle of internal friction.
It is recommended here that the notation section is changed to bring the value into line
with the Annex.
There should also be compatibility of notation with the rules for retaining walls given
in Eurocode 7.
It is also necessary that the values of internal friction angle in the table of material
properties should be amended to give the simple soil-mechanical “internal friction
angle” and not the “effective angle of internal friction”, since the use of this parameter
evaluating both the lateral pressure ratio λ or K and the discharge factor Co depend on
the simple internal friction angle.
The values in Table 5.1 can be used to show the variability of each solid. The
deduced variability (coefficient of variability) of the lateral pressure ratio for different
solids is shown in the figure.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
8
Coefficient of variation
0.30
0.25
0.20
0.15
0.10
0.05
0.00
Ba a
po ed
Fl ggr m
C clin t
at
C l: b r
oa : b k
nd at ize
la tz
W r
Li ime pe h
em C e ley
os M er
Sl Sa ars ck
Su s
on Fl at)
: g d : ry
et ro e
C ow n
C ke e: b ed
es yd ts
ya ry
w e
n
an
ga
oa ke
cr et ez
in
r ( at
C oal lac
L ore as
p row
m , h lle
nu a r
he
u
en me
So r, d
ag n e d
ne rat
o o
C ok der
Sa ph a
he
r
m
ou eg
on : p re
e a le
be
w
:c er
y
ra q u
lu
A
to
l:
Ph
Ir
C
The choice of a fixed value for the variability of these solids in the
ENV 1991-4 (1995) is quite unfortunate, since some solids are clearly much less
variable than this (e.g. Flyash) leading to rather conservative designs, and some are
much more variable (e.g. Coal) leading to an unsafe design procedure.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
9
First attempt
Table 5.1 Physical properties of bulk solids: characteristic values †
Type of Unit weight Angle Effective angle Lateral Wall friction angle ‡ Max. flow
Bulk solid of of internal pressure φw pressure
γ repose friction Ratio (µ = tan φw) multiplier
φr φi K or λ Co
γl γu φr φil φiu Kl, λl Ku, λu Wall type D1 Wall type D2 Wall type D3
Lower Upper Lower Upper Lower Upper Lower Upper Lower Upper Lower Upper
kN/m3 kN/m3 degrees degrees degrees degrees degrees degrees degrees degrees degrees
Default material * 6.0 20.0 40 25 55 0.35 0.65 10 30 12 35 14 45 1.45
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
10
Revised version of
Table 5.1 Physical properties of bulk solids: characteristic values †
Type of Unit weight Angle Effective angle Lateral pressure Wall friction coefficient ‡ Max. flow
bulk solid of of internal Ratio µ pressure
γ repose friction K or λ (µ = tan φw) multiplier
φr φi Co
γl γu φr φil φiu Km, λ m aλ Wall Wall Wall aµ
type type type
D1 D2 D3
Lower Upper Lower Upper Mean Factor Mean Mean Mean Factor
KN/m3 KN/m3 Degrees degrees Degrees
Default material * 6.0 20.0 40 25 55 0.35 1.6 0.32 0.39 0.50 1.60 1.45
Alumina 10.0 12.0 27 25 40 0.50 1.20 0.41 0.52 0.64 1.10 1.40
Barley 7.0 8.5 20 26 33 0.56 1.08 0.35 0.41 0.54 1.20 1.35
Cement 13.0 16.0 28 40 50 0.32 1.18 0.41 0.48 0.59 1.10 1.40
Cement clinker 15.0 18.0 33 42 52 0.29 1.19 0.41 0.52 0.64 1.10 1.40
Coal: black 8.5 11.0 35 40 60 0.24 1.46 0.35 0.57 0.63 1.25 1.45
Coal: brown 7.0 9.0 33 45 65 0.18 1.56 0.39 0.57 0.84 1.15 1.45
Coal: powdered 6.0 9.0 38 40 50 0.32 1.18 0.40 0.49 0.56 1.10 1.45
Coke: breeze 7.0 8.0 40 35 45 0.39 1.16 0.45 0.56 0.61 1.10 1.45
Coke: petroleum 6.5 7.5 38 37 47 0.36 1.16 0.50 0.61 0.66 1.10 1.45
Concrete aggregate 17.0 19.0 34 30 40 0.46 1.14 0.35 0.43 0.57 1.25 1.40
Flour (wheat) 6.5 7.0 40 23 30 0.61 1.08 0.34 0.50 0.57 1.20 1.45
Fly ash 8.0 14.0 22 30 35 0.51 1.06 0.39 0.57 0.70 1.15 1.45
Iron ore pellets 19.0 22.0 35 35 45 0.39 1.16 0.41 0.56 0.61 1.10 1.40
Lime, hydrated 6.0 8.0 40 35 45 0.39 1.16 0.39 0.55 0.70 1.15 1.40
Limestone powder 11.0 13.0 30 40 60 0.24 1.46 0.35 0.50 0.67 1.20 1.45
Maize 7.0 8.5 30 28 33 0.54 1.06 0.35 0.46 0.54 1.20 1.40
Phosphate rock 16.0 19.0 27 35 55 0.31 1.39 0.40 0.48 0.54 1.15 1.40
Sand: coarse dry 14.0 17.0 30 30 40 0.46 1.14 0.35 0.46 0.57 1.20 1.40
Sand: quartz 15.0 17.0 30 35 40 0.43 1.07 0.33 0.33 0.42 1.20 1.40
Slag: granular, dry 10.5 12.0 40 35 38 0.45 1.04 0.34 0.41 0.49 1.10 1.40
Soya beans 7.0 8.0 23 25 32 0.57 1.08 0.22 0.26 0.30 1.20 1.30
Sugar 8.0 9.5 29 33 38 0.46 1.07 0.35 0.50 0.63 1.25 1.40
Wheat 7.5 9.0 20 26 32 0.57 1.07 0.35 0.43 0.54 1.25 1.30
† Where this table does not contain the material to be stored, testing should be undertaken. The “default material” is offered as a substitute for situations where the cost implications of using a wide property range for the
design are minor, so that it is difficult to justify the cost of testing.
* Properties for a “default material” are shown here, for conditions where no data is available. For small installations, these properties may be adequate, but they will lead to very uneconomic large silos: testing is always
preferable.
‡ For Wall Type D4 (corrugated wall), see Appendix D4.
NOTE: the lateral pressure ratios K for flyash and wheat were incorrect in the original table (taken from Rotter (2001)) and have now been corrected.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
11
The new version for the EN is therefore proposed with the relationships:
Max K = aK Km … (5.2)
Min K = Km / aK … (5.3)
Max µ = aµ µm … (5.4)
Min µ = µm / aµ … (5.5)
where Km is the mean value of K and µm is the mean value of µ.
It is supposed that:
i) the value of aK depends on the material;
ii) the value of µm depends on the wall surface type (thus several values of µm for a
given solid);
iii) the value of aµ is proposed to be taken as a single value for all walls (this is not
a necessary assumption, but it simplifies the table and the process).
The values offered below are derived from upper and lower tabulated values given in
the Australian Standard (AS 3774-1996) assuming that the latter are 5%ile and 95%ile
values (as indicated in that standard) but that the EN wishes to adopt 10%ile and
90%ile values. The distributions were all assumed to be log-normal.
The resulting calculated variation factors aK and aµ for different solids are shown
below.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
12
1.4
1.3
1.2
ENV value
1.1
1.0 ne ted
lu e
Ce Ce y
or ash
m , hy ets
s
Iro Fl )
ya e
r
Co l, b k
Co ke, ed
et ze
ph ize
nd rse
r
at
ow n
nd ck
a
, f rtz
tc t
r( m
en en
de
at
Co inke
ga
an
t
So nac
in
rle
c
, p ow
ga
he
u
e
r
S a e ro
Co la
he
Sl qua
l
ra
Sa coa
a
m
w
Co de
Su
Fl role
be
ke bre
el
Ba
M
re
,b
W
r
ur
d
po
w
l
Li e p
gg
at
al
,
A
,
a
A
ag
,p
ou
e
to
os
m
n
m
al
es
Ph
Li
Source BPSTab16.XLS
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
13
1.20
1.10
1.05
1.00
ne ated
lu e
Li re p h
e s yd s
s
Iro Fl t)
ya e
al er
ck
Co e, red
et ze
ph ize
nd rse
r
at
ow n
nd ck
a
, f rtz
tc t
r( m
en en
de
ga
an
e, llet
t
So rnac
in
rle
s
w
a
ga
he
Co ink
a
e
S a e ro
Co la
he
Sl qua
Sa coa
os Ma
m
al bro
m
w
Co de
Su
Fl role
be
ke bre
Ba
r
re
e
,b
W
po
w
Ce Ce
u
gg
at
,
A
,
Co al,
h
A
,p
ag
k
,p
o
ou
to
m
n
m
Ph
m
Li
Source BPSTab16.XLS
1.20
1.10
1.05
1.00
ne ated
A ate
Li re p h
s
e s yd s
Iro Fl t)
ya e
al er
ck
Co e, red
et ze
ph ize
nd rse
r
at
ow n
nd ck
a
, f rtz
tc t
r( m
en en
de
ga
an
e, llet
So rnac
in
rle
s
w
he
Co ink
a
e
S a e ro
Co la
he
Sl qua
g
Sa coa
os Ma
m
al bro
m
w
Co de
Su
Fl role
be
ke bre
Ba
r
re
e
,b
W
po
w
lu
Ce Ce
u
gg
at
,
,
Co al,
h
A
,p
ag
k
,p
o
ou
to
m
n
m
Ph
m
Li
Source BPSTab16.XLS
For most solids, use of this data will increase the variability of solids assumed in
design. Thus the maximum design pressure will rise. If this is adopted, the overall
conservatism of the complete calculation, involving discharge pressures, should be
checked. In particular, the Project Team should make sure that the adopted value of
the discharge factor Co is adequately reduced to return the complete calculation to
about the same outcome as before.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
14
General remarks:
Course grained materials (wheat)
Has densities controlled mainly from method of filling
seem not very sensitive to test principle
Powders (cement)
has densities controlled mainly from consolidation (pressure level)
seem very sensitive to test principle (soil mechanics is used for the standard). Very
higt values of internal friction in the draft, version D are considered unrealistic.
CA-Silo ref.:
Munch-Andersen J., L.O.Nielsen and J. Nielsen: ”Comparison of load parameters for
stored materials”, Department of Structural Engineering and Materials, Technical
University of Denmark, Report S, no 1, 1997.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
15
Wheat
Unit weight Internal friction Lateral pressure Wall friction
ratio
Draft, version D 7.5-9.0 26-32 0.57 0,35, 0.43, 0.54
* or / by 1.07 * or / by 1.25
DIN 9.0 0.60 0.25, 0.40, 0.60
CA-silo 7.7-8.7 24-30 0.55-0.65 Steel: 0.28-0.40
Conc. : 0.40
Proposal 8.0-8.8 24-30 0.55-0.65 Steel: 0.25- 0.40?
Conc: 0.30 – 0.45
Cement
Unit weight Internal friction Lateral pressure Wall friction
ratio
Draft, version D 13-16 40-50 0.32 0,41, 0.48, 0.59
* or / by 1.18 * or / by 1.10
DIN 16 0.65 0.40, 0.45, 0.50
CA-silo 16.0-16.8 34-39 0.40-0,49 Steel: 0.52-0.55
Proposal 13-17 34-39 0.40- 0.50 Steel:?
Conc: ?
Coal
Unit weight Internal friction Lateral pressure Wall friction
ratio
Draft, version D 8.5-11.0 40-60 0.24 0,35, 0.57, 0.63
* or / by 1.46 * or / by 1.25
DIN 10.0 0.60 0.45, 0.50, 0.60
CA-silo 7.1-11.2 26-42 0.36-0.61 Steel: 0.35-0.45
Conc. : 0.36
Proposal 7.5-11.5 26-45 0.36-0.61 Steel: 0.35- 0.45?
Conc: 0.35 – 0.50?
9. Material testing
The testing of bulk solids should be always encouraged for designs in higher
Reliability Classes. The EN should provide an economic advantage in design to the
designer or client who does undertake materials testing. It is important that a smaller
variability can be deduced from the tests than the variability required to be used when
values are taken from the table of properties.
The variability of the material should be a major focus of the testing, and an
appropriate statistical technique should be defined for deducing the characteristic
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
16
values from the measured values. Such a method is given in an Annex in the
Australian Standard and in Rotter’s Guide (2001).
The ENV sought to clarify the fact that the value Co is only valid for slender silos by
terming it the “maximum” load magnifier. Since this need remains, it is proposed
here that the lower value for squat and intermediate silos be termed instead the
“reduced discharge factor”, which can be as low as 1.0.
The values used in the ENV were all empirical and are deemed to depend on the
material being stored. For the TEN materials listed in the table of materials, values of
Co are given.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
17
a
Discharge factor or Load magnifier C o
1.45
1.40
1.35
1.30
1.25
Barley Cement Cement Dry sand Flour Fly ash Maize Sugar Wheat Coal
clinker
For materials not listed in the table, section 7.4 gives the rule:
(3) For materials not listed in table 7.1, the maximum wall load magnifier may be obtained using:
For φ ≤ 30o, C0 = 1,35 … (7.4)
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
18
large load magnifier (that is only weakly related to physical phenomena) together
with substantial material variability and substantial partial factors.
In relation to (a) and (b) above, the values of Co in the ENV material property table
(taken from the ENV) are compared below with the values derived from the above
Eqs 7.4 and 7.5, using φ determined using K = 1.1(1 – sinφ). The equations 7.4 and
7.5 were derived from this data, but the fit is poor, and either the K or the Co values
should be modified to make them match.
Comparison of discharge factors
2.00
ENV tabulated Co
1.90
Co from ENV equation
1.80 Co from ENV equation with AS3774 phi
Discharge factor Co
1.70
1.60
1.50
1.40
1.30
1.20
28 30 32 34 36 38 40 42 44 46
Deduced internal friction angle phi (degrees)
If these equations are used with mean values of φ taken from the Australian Standard
AS3774, the fit is very bad. This may be partly caused by the AS3774 values of φ
being effective angles of internal friction derived from Jenike tests (see below). This
choice of internal friction angle tends to increase the apparent φ for powders and
cohesive solids, leading to a higher derived Co and a lower deduced K.
Since most of the testing done in recent years has been undertaken by Jenike-style
labs, it is difficult to obtain the correct data. Moreover, since most earlier testing was
interpreted with slender reinforced concrete silos in mind, the earlier published values
of K were not mean values but maxima and the corresponding φ values minima. It is
not easy to obtain reliable mean values of the soil mechanical internal friction angle
for a large range of different solids. The data on which to base the defined mean
values of the material properties remains a significant problem for the PT.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
19
0.50
0.45
0.40
0.35
Barley Cement Cement Dry sand Flour Fly ash Maize Sugar Wheat Coal
clinker
In squat silos, where the diameter is large and the wall height small, the total storage
capacity is much affected by the angle of repose. The point of highest contact
between the solid and the wall is also generally governed by the angle of repose, so
the wall pressure rule requires this to be evaluated carefully. A determination of the
equivalent surface in all silos also strictly needs the angle of repose to be defined first.
It is therefore desirable that this material property be included in the table of
properties.
It may be necessary to define two angles of repose, one for conical piles and the other
for wedge-shaped piles, but these may be close enough for the conical pile value to be
sufficient.
The angle of drained repose is generally not required as part of a silo pressure
assessment for worst case loading conditions.
Silos that are either filled or discharged eccentrically to the vertical axis are subject to
unsymmetrical pressures relative to that axis. These pressures are caused by
asymmetries of solids packing and of solids flow.
Silos that are concentrically filled and discharged are also subject to unsymmetrical
pressures, though these are generally smaller than those noted above. The asymmetry
arises from geometric imperfections in the silo walls and minor asymmetries of
packing that can cause disproportionately large asymmetries in the pressure pattern.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
20
Where there are no eccentricities of the inlet or outlet, the patch load must serve to
provide a minimum level of asymmetrical load, to ensure that the silo is able to resist
minor deviations from symmetry.
Where eccentricities exist, the patch load was also used to represent the increasing
asymmetrical loads, though the loading pattern of the patch load is rather different
from that due to real eccentricities. For this reason, the ENV restricted the use of this
treatment to eccentricities that were less than 0.25dc, where dc is the characteristic
horizontal dimension of the silo.
In the changes between the ENV and EN versions of this standard, the patch load
should be adopted for those roles for which it is particularly suited, and other methods
used where it is less appropriate.
It should be noted that powders are commonly aerated when they are placed in the
silo, leading to a level, or almost level top surface and a relatively homogenous
particle placement structure within the stored solid. For this reason, the EN permits
the designer to ignore the filling patch pressure for powder solids.
The filling patch load is always superseded in design importance by the discharge
patch load, except in those silos that do not need to be designed for discharge
pressures or the discharge patch load. This means that squat silos, silos with
guaranteed internal flow and silos with mechanical discharge systems will be affected
by the filling patch pressure, but others will not.
The ENV used a rather heavy filling patch load because the variability of the
properties of stored solids was considerably underestimated. With a more realistic
estimate of this variability now adopted into the EN, it has been possible and
appropriate to reduce the magnitude of the filling patch load. The revised rule
demands a 10% increase in pressure when the eccentricity of filling is zero, and a
20% increase if solids strikes the wall during filling.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
21
considerable discrepancies would persist between the requirements of DIN and those
of the EN. It therefore seems appropriate to return to the more complicated
formulation used in DIN, now calibrated differently to account for the material
property variability that is properly recognised in the EN.
In squat silos, they have a very large effect because the difference between highest
contact points on the two sides of the silo diameter is great and the solid effectively
applies a large overturning moment on the silo. The nature of this loading is not
represented at all well by a patch load.
The second effect of filling eccentricities is quite different in slender silos: the solid
that lies immediately beneath the impacts of the filling process develops a closer
packing structure, anisotropy and a higher local density. If this denser material is
centrally located, there is little loss of symmetry. However, when the denser material
occurs close to the wall at points well below the solid surface, highly unsymmetrical
pressures may develop on the wall. The patch load provides a good representation of
this phenomenon, but the eccentricity that matters is not the eccentricity of the top of
the pile of solids when the silo is full, but the maximum filling eccentricity that occurs
during the filling process.
θc
surface
ec
profile
for full
condition
et
silo
centre
line
ef
α
eo
For this reason the Project Team decided to divide the filling eccentricity into two
separate factors governed by two differently assessed eccentricities (Fig. 1.1b). The
first, et, defines the centre of the top pile of solid: it is useful for squat silo eccentric
filling problems. The second, ef, is the maximum eccentricity of the top of the heap
during the entire filling process. It is useful for assessing the patch load to represent
unavoidable asymmetries in the loads.
These two eccentricities can only be determined by a calculation based on the solids
trajectory into the silo, and for this reason, these eccentricities are related to the chute
placement and end slope.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
22
Discharge eccentricities have a very large effect on very slender silos, where a
channel of flowing material against the wall causes a serious local pressure reduction,
leading to bending of the wall and large membrane stresses. Many failures due to
eccentric discharge have occurred in recent years in both metal and concrete silos.
The patch load does not represent this loading well.
In squat silos, moderate discharge eccentricities (eo < 0.25dc) have a rather small
effect because the flow channel is internal. This eccentricity is therefore not so
important in squat silos.
In view of the above, the PT divided the effects of filling eccentricities into two
categories and separated it from the discharge defined separate rules for each
situation. The patch load is retained where it represents the silo pressure phenomena
well.
The effect of the eccentricity, either for filling or for discharge, is clearly dependent
on the aspect ratio of the silo. It is therefore suggested here that the patch load be
used wherever it represents the pressure pattern phenomenon well, and that special
rules are developed specifically for filling eccentricities and for discharge
eccentricities. These special rules should be particularly intended for use on silos in
Reliability Class 3, where the structural analysis requirements are more demanding,
and the silos have been placed in this class because of their size and the difficulties of
design.
For silos in Reliability Class 1, no special treatment should be necessary and no patch
is needed.
For silos in Reliability Class 2, the patch load, and a simple treatment of highly
eccentric discharge pressures should be used.
For silos in Reliability Class 3, the full treatment of patch loads with properly
calculated eccentric flow channel pressures should be used.
The special rule for filling eccentricities should focus on squat silos, with a rule that
causes differential pressures across the diameter, leading to an overturning of the
whole silo, and increased vertical compression on the side where the filling pile of
solid lies.
The special rule for discharge eccentricities should focus on slender silos, with a
special load case representing a channel of flowing solid down one side. A sample
procedure of this kind was devised by the American Concrete Institute committee,
and similar rules can be developed. Large eccentricities in silos of significant size
require a finite element analysis (see ENV1993-4-1), so there is no advantage in
making this load definition simple and suited for hand analysis.
If these two features are adopted, there is no need to restrict the eccentricity of the
outlet or of the filling spout.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
23
A proposal has been made for the EN by Rotter, based on the method given in the
Guide for the Economic Design of Circular Metal Silos and originally proposed in
1986.
For silos of moderate overall size (Reliability Class 2), simple rules can be used for
concentric filling and discharge cases, but silos with highly eccentric discharge outlets
should be treated as more complicated.
For silos of great size and complexity (Reliability Class 3), finite element analysis is
required for design by ENV 1993-4-1 and there is no benefit in simplifying the
loading case down to a rule designed for hand analysis.
Discussion should also focus on where the bounds on sizes should be between the
individual classes.
Recent research has shown that stiff-walled rectangular silos experience higher
pressures at midside than in the corner, inducing larger bending moments in the wall.
The pressure pattern should probably be amended to reflect this detrimental effect.
Much recent experimental and theoretical research on flexible walled rectangular silos
has shown that the pressures at the midside of the wall are much smaller than those in
the corners. This means that the bending moments developing in the walls and
corners are smaller than a uniform pressure design would suppose. If advantage is
taken of this effect, the walls of flexible silos can be made lighter, making them even
more flexible, and considerable improvement in design can be achieved.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
24
Thus, for silos in Reliability Class 3, where considerable design effort is required,
there should be two alternative design methodologies with different pressure regimes
in a rectangular silo:
a) stiff walled (concrete and stiffened metal silos)
b) flexible-walled (unstiffened metal silos)
Patch loads
Recent research on the effect of patch loads on rectangular silos suggests that these
loads make only a minor difference to the design, and that the complexity of design
that they introduce is not justified by the difference in outcome. It is not clear that
patch loads are needed if the pressure distribution is made non-uniform according to
the proposal outlined above.
The revised rule now states that if the wall is subject to first order bending effects
when the uniform Janssen pressures are applied, the patch load can be ignored. Since
the patch load was only intended to induce a limited amount of bending, it can be
safely ignored in this case.
A new section has been added as “6.5 Loads in rectangular silos with internal ties” to
provide rules for this case. These rules should be carefully reviewed by the PT.
The patch load was devised in the context of stiff-walled concrete silos. The direct
application of this rule to thin-walled steel silos has not been demonstrated, and the
different form of patch load is intended to represent the effects of asymmetry in metal
silos quite well.
The above discussion of rectangular silos also suggests that the stiffness of the
structure is particularly important in rectangular silos. The rules should be amended
to reflect this difference.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
25
For circular silos, it seems likely that the horizontally uniform component of pressure
at any level will be similar for stiff and flexible walled silos, because the
circumferential stiffness is very high in both designs. However, the stiffness of the
response to non-uniform or asymmetrical pressures is very different for thick-walled
and thin-walled silos, and the patch loading rules should reflect this. The patch load
defined in ENV1991-4 to be applied to a circular steel silo (pressure pattern pocosθ) is
less demanding than the discrete rectangular patches used for stiff-walled silos to
address this concern.
24. Coefficient for flat bottoms Cb: is 1.2 too low? Or does material
variability cover this phenomenon so that Cb should be omitted
The earliest standards for silos were all primarily concerned with defining the
maximum normal pressure against the wall of a tall concrete silo. This maximum
pressure occurs when the wall is as smooth as possible and the lateral pressure ratio is
as high as possible. The values of material parameters given in the tables reflected
these extreme material property values.
However, a disadvantage of using these extremes (min µ and max K) was that the
vertical stress in the solid was always underestimated. As a result, the total load that
would regularly be applied to the hopper was underestimated. This was overcome by
retaining the fixed (extreme) values of material properties relevant to the vertical
walls, but increasing the load on the bottom by using a bottom loading coefficient Cb.
This coefficient raised the calculated bottom pressures to values seen in tests.
The justification given for this bottom load coefficient was often different, suggesting
that the increase in bottom load was caused by falling solid or impact, but there is
little evidence that static equilibrium was ever seriously in error, and it seems more
likely that the underestimate arose chiefly from the choice of values for the material
properties.
There are a few solids for which falling solid, collapsing arches or impact are
potential events. These solids and the phenomenon concerned should be separately
identified and the rule made specific to them. Solids in this category might include
highly cohesive solids in slender silos and mechanically interlocking solids such as
cement clinker. A draft rule has been suggested for these cases.
One bad feature of the bottom load multiplier Cb is that it makes no sense to have a
single value that is used on all bottoms. Where the silo has a moderately squat
vertical-walled section above the hopper, wall friction on the vertical wall does not
support much of the total weight of solids. The calculated vertical stress in the solid
at the transition is then close to the total weight of solid. If this vertical stress is then
multiplied by a factor of 1,2 or more, it can easily exceed the total weight of solids,
suggesting that more load is applied to the hopper than can arise from the weight of
solids. If some justification of this idea is attempted in terms of collapsing arches, it
may be noted that arching is least likely to occur in squat vertical wall geometries, and
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
26
most likely in very slender (tall) silos, so the justification is applied to entirely the
wrong silo geometries.
By introducing the variability of material properties, the main problem of bottom load
magnifiers is overcome. Two load cases are examined: one gives the maximum
normal pressure against the vertical wall; the other the maximum vertical stress in the
silo just above the hopper. The equivalent bottom load magnifier that arises from
these two load cases using the material variability factors of 0,9 and 1,15 in the ENV
is approximately 1,3. There is therefore no real case for introducing a bottom load
magnifier for all but highly cohesive and mechanically interlocking solids.
Perhaps the DIN 1055 hopper method could be attached as an Annex to permit its
continued use, but the weight of negative comments on this method from several
different countries suggests that this is not a good choice to make for the EN.
The ENV 1993-4-1 standard for silo structures has the paragraph:
(6) The elastic pressurised imperfection factor αpe should be based on the smallest local internal
pressure (a value that can be guaranteed to be present) at the location of the point being assessed, and
coexistent with the axial compression:
p̄
αpe = αo + (1 − αo) 0,3
... (5.14)
p̄ + α
o
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
27
with:
pr
p̄ =
t σxRc
where:
p is the minimum reliable local value of internal pressure.
Thus, it is vital that EN 1991-4 defines the value of this “minimum reliable local
value of internal pressure” for thin-walled silos.
This question is only covered by a note in Section 7 of the ENV 1991-4 which reads:
NOTE: For shell structures minimum (support) loads may be the unfavourable loads
The best arrangement would be to move Section 7 towards the front of the standard,
and to close this new section with a statement of the different property extreme
combinations that must be used, together with a clear statement that each set of
property extremes should constitute a separate load case for structural design.
a) Slender silos
A silo in which the overall aspect ratio of the stored solid hb/dc exceeds 1.50.
c) Squat silos
A silo in which the overall aspect ratio of the stored solid hb/dc is less than 1.00.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
28
ph
The approximate rule in ENV 1991-4 looks fine for normal pressures, and using hand
calculations, but this is not the only load case of importance. The rule presents
considerable problems when the vertical load in the wall is to be evaluated.
Unfortunately the intersection point hLJ cannot be defined using a simple equation:
the distance hLJ is the solution of:
hLJ -h /z ho
+ e LJ o = 1 +
λ zo λ zo
Whilst this may not seem to be a problem when normal pressures are being
considered, it leads to a complicated evaluation for the vertical force developing in the
silo wall, since the expression for the vertical force depends on a depth which cannot
be written explicitly and is not easily evaluated.
A better treatment of the wall pressures in squat silos is to use the semi-empirical
expression of the Reimbert theory [1976], suitably generalised to accommodate a
more general expression for the natural lateral pressure ratio λ of the solid [Rotter,
1983b, 2000b].
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
29
The above leads to simple expressions for the hoop and vertical forces in the wall:
z−ho -2
nθ = r pho 1− +1
zo−ho
(z − ho)2
nx = − µ pho
( z + zo − 2ho)
The above generalised Reimbert equation leads to the same result as Janssen at
greater depths, but generalised Reimbert provides a most useful conservative estimate
of pressures near the surface in squat silos, whilst satisfying all the equilibrium
requirements and boundary conditions.
et
Highest wall
contact with solid
zs
dc
Figure 5.5 Filling pressures in an eccentrically filled squat or intermediate
slenderness silo
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
30
in which:
γ A γr
pho = = … (5.78)
µ U 2µK
zs
Z =B … (5.79)
r
B= – ho … (5.80)
2µK
-0.05
-0.10
-0.15 Approx
Nz/Nzo
Nz/Nzo
-0.20
-0.25
-0.30
-0.35
-0.40
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Dimensionless eccentricity of fill: tan(phi)*Re/A
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
-0.02
Coefficients a1-a3
-0.04
-0.06 a1 y = -0.248x
a2
-0.08
a3
-0.1 Linear (a3)
y = -0.2892x
Linear (a1)
-0.12 Linear (a2)
-0.14
Value of T
Reference:
Rotter, J.M. (1983) “Structural Effects of Eccentric Loading in Shallow Steel Bins”, Proceedings,
Second International Conference on the Design of Silos for Strength and Flow, Stratford upon Avon,
Nov., pp 446-463.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
31
The key consideration in the development of the proposed rule was that it should
blend smoothly with the rules for retaining walls given in EN 1997, and with the rules
for squat silos. The adopted rule gives a linear pressure variation with depth that is in
agreement with the result that would be obtained by application of soil mechanics
theory. It is in accordance with EN 1997 Section 9.5.2.
Two problems are known in this regard: the shrinkage of a thin metal silo structure
onto a stiff thermally inert stored solid following a sudden fall in ambient
temperature; and the temperature differential associated with the filling of a silo with
hot solids, leading to severe local bending of the wall near the solid surface. Both
problems are given a brief treatment, and it is hoped that these rules will be improved
with time.
The rule for shrinkage of a thin metal silo onto a thermally inert solid is taken from
the Australian Standard (AS 3774), which adopted the analytical treatment of
Andersen (1966). Whilst this treatment does not account for the effects of changes in
the mobilisation of wall friction, which can have a drastic effect, it is satisfactory for
relatively squat silos with relatively small thermal differentials.
Andersen, P.F. (1966) “Temperature Stresses in Steel Grain Storage Tanks”, Civil Engineering, ASCE,
Vol. 36, No. 1, pp 74-76.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
32
The rule for silos filled with hot solids remains a principle at the present time, until a
quantitative rule can be developed.
b) For rectangular silos, the PT thought that the effect of the pressure distribution
should be addressed. The comments outlined in the section on rectangular silos above
deals with this question.
2. Dust explosions
The PT considered that it is too difficult to set out complete rules for dust explosions,
but warnings and advice should be included within the EN. Following a contribution
from Prof. Eibl and Dr Ruckenbrod, an annex has been added for special explosive
aspects in relation to silos.
3. Silage silos
The PT considered that it is too difficult to set out complete rules for silage silos, but
warnings and advice should be included within the EN.
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc
33
c:\aa\profesnl\codes\cen-sc1\en-work\versionf\bkgrnd15.doc