Stabilisation of Beams by Sandwich Panels
Stabilisation of Beams by Sandwich Panels
Stabilisation of Beams by Sandwich Panels
net/publication/44022049
CITATION READS
1 489
4 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Thomas Misiek on 02 June 2016.
CIB
WORKING
COMMISSION
W056
‐
SANDWICH
PANELS
PAPERS
AND
POSTGRADUATE
PAPERS
FROM
THE
SPECIAL
TRACK
HELD
AT
THE
CIB
WORLD
BUILDING
CONGRESS
2010,
10‐13
MAY
2010
THE
LOWRY,
SALFORD
QUAYS,
UNITED
KINGDOM
Selected
papers
from
the
Proceedings
of
the
18th
CIB
World
Building
Congress.
Proceedings
edited
by:
Professor
Peter
Barrett,
Professor
Dilanthi
Amaratunga,
Dr.
Richard
Haigh,
Dr.
Kaushal
Keraminiyage
and
Dr.
Chaminda
Pathirage
W056
Special
Track
Papers
(excluding
Postgraduate
Papers)
reviewed
by:
Professor
Jose
Amorim
Faria,
Professor
Jukka
Aalto,
Professor
Jörg
Lange,
Dr.
Lars
Heselius
and
Professor
J.
Michael
Davies
CIB
Publication
342
W056
‐
SANDWICH
PANELS
PAPERS
AND
POSTGRADUATE
PAPERS
FROM
THE
SPECIAL
TRACK
The
objectives
of
this
Commission
are
to
exchange
information
and
coordinate
research
programmes
in
all
technical
aspects
of
light‐weight
constructions
using
traditional
and
new
building
materials
and
to
prepare
synthesised
reports
on
matters
of
particular
interest.
CONTENTS
Papers
Stabilisation
of
Beams
by
Sandwich
Panels
‐
New
Regulations
and
Recent
1
Research
Results
‐
Misiek,
T.
Kapplein,
S.
Durr,
M.
Saal,
H.
A
Unified
Approach
for
the
Local
Buckling
of
Sandwich
Panels
and
Trapezoidal
Sheeting
14
Misiek,
T.
Hassinen,
P.
ECCS/CIB
Joint
Committee
on
Sandwich
Constructions,
Recent
European
27
Recommendations
on
Design
and
Testing
Davis,
J.M.
Hassinen,
P.
Heselius,
L.
Misiek,
T.
Methods
to
Measure
the
Durability
of
Structural
Sandwich
Panels
38
Hassinen,
P.
Pfeiffer,
L.
Reliable
Composite
Roofing:
Learning
from
Experience
51
Roberts,
K.
Postgraduate
Papers
Openings
in
Sandwich
Elements
61
Warmuth,
F.
Lange,
J.
Optimization
of
Geometry
and
Core
Materials
of
Sandwich
Panels
with
Metallic
Faces
73
Kurpiela,
A.
Lange,
J.
Berner,
K.
CIB
Brochure
85
Disclaimer
87
Stabilisation of Beams by Sandwich Panels
– New Regulations and Recent Research Results
Misiek, T.
Karlsruhe Institute of Technology, Karlsruhe, Germany
(email: [email protected])
Käpplein, S.
Karlsruhe Institute of Technology, Karlsruhe, Germany
(email: [email protected])
Dürr, M.
Montana Bausysteme AG, Villmergen, Switzerland
(email: [email protected]
Saal, H.
Karlsruhe Institute of Technology, Karlsruhe, Germany
(email: [email protected])
Abstract
Sandwich panels are modern pre-fabricated construction components used as cladding elements for
different types of buildings. Sandwich panels consist of an insulating core material covered by two
faces which are typically made of thin metal sheets. In standard applications, the panels are mounted
and fixed on a load-bearing substructure of beams or purlins. Sandwich panels can reduce the
problem of lateral torsional buckling of this substructure of beams or purlins by providing
stabilization either by shear stiffness or by torsional restraint. The new edition of the German
standard for the design of steel structures DIN 18800 gives formulae for the calculation of the
stiffness of the torsional spring for restraint of the substructure under vertical downward loading.
These new regulations are based on experimental investigations and parametric finite element
analyses. These formulae only apply for sandwich panels with steel facings and polyurethane and
mineral wool as core material. The paper explains the load-bearing mechanisms of the stabilisation
effect by the sandwich panel. A mechanical model is developed for extending the range of
application of the design formulae and to include the effects of creep and elevated ambient
temperature. It presents the new regulations of DIN 18800 and explains the tests on which these
regulations are based.The spectrum of applications not yet examined is investigated by tests and
accompanying numerical calculations within the framework of the EASIE project. As a result of
these investigations the torsional restraint of panels with facings made of aluminium and glass fibre
reinforced plastics (GFRP) and with cores made of EPS are dealt with. The load case wind suction is
discussed in addition. The increase of the torsional restraint obtained by fixing roof panels at the
upper flange (which is mainly with saddle washers) is also explained and quantified by these
investigations.
1
1. Stabilising effects on beams
Sandwich panels increase the resistance of substructures (beams, purlins) against lateral torsional
buckling by restraining the lateral displacements and rotations.
The high in-plane shear stiffness of sandwich panels can be used for stabilizing the lateral
displacement of the substructure and thus preventing lateral torsional buckling of the substructure.
This type of stabilization requires the exact knowledge of the in-plane shear stiffness. Special
considerations are necessary for the design of the fastenings because the flexibility of the connection
to the substructure and that of neighbouring panels to each other reduces the effective shear stiffness
significantly.
The torsional restraint is governed by the stiffness of the connection of the sandwich panel to the
substructure. Recent research carried out by Dürr (2008) showed that this stiffness significantly
depends on the load transferred by the sandwich panel to the substructure. Dürr (2009) gives
formulae for calculating the parameters of this moment-rotation-relation for sandwich panels under
deadweight loading and with two different core materials. So far only connections through the lower
flange of the outer face with two fasteners per element have been investigated. Other types of
connections (e.g. connection through upper flange of the outer face with saddle washers) and
different core materials are important yet unknown parameters of the moment-rotation-relation.
The focus of the present paper is on the stabilisation of beams by torsional restraint.
The torsional restraint by sandwich panels can be calculated by using the mechanical model of a
torsion spring with the spring stiffness c . This spring stiffness is a combination of the bending
stiffness of the attached panel c C, the stiffness of the connection c A and the distortional stiffness c B
of the beam to be stabilised. The stiffnesses c C and c B depend on the geometry of the sandwich
panels and type of beams used. They can be easily calculated. We will focus on the stiffness c A of
the connection between the sandwich panel and the subjacent beam because this is the weak link
dominating the value of the combined stiffness c of the chain of springs. In the following text, the
stiffness c A will be simply denoted as c to ease reading and to reduce the number of subscripts.
2
c
Figure 2 shows a generalised moment-rotation-relation for the spring stiffness of the connection of a
sandwich panel under deadweight loading.
m
c 2
mK
c 1
c 1
mK
c 2
0
In this generalised relation we assume that all of the fasteners are mounted on one side of the web as
shown in Figure 2. Here, the positive direction of rotation is defined as an anticlockwise rotation.
We can differentiate three parts of the moment-rotation-relation. For small rotations , there is the
value c 1. The load q acting on the panel is always transferred by contact from the inner face of the
panel to the upper flange of the beam. The rotational stiffness only depends on the width of the flange
and the indentation stiffness (Figure 3). This indentation stiffness is dominated by the compression
stiffness ECc of the core material. The rotational stiffness does not depend on the position of the
fasteners because the fasteners are not activated in this situation.
3
Figure 3: Mechanical model for c 1
The area of contact decreases, with increasing rotation until it is reduced to the final contact line with
the outer edge of the flange. At this stage, the restoring moment is the contact moment
q b
mK
2
When the deflecting moment to be stabilized exceeds the contact moment mK the value c 2 applies. At
this stage, tensile forces in the fasteners are activated. These tensile forces F t cause an indentation uw
of the fasteners’ heads and washers into the outer face of the sandwich panel. This additional
deformation decreases the stiffness significantly: The value c 2 is significantly smaller than the value
c 1.
bK
uw
The value c 2 depends on the indentation stiffness kw of the fastener and the indentation stiffness kf at
the line of contact at the outer edge of the flange. This stiffness depends on the direction of rotation
as defined in Figure 2 with regard to the position of the fastener and the distance bK of the fastener
from the contact line as defined in Figure 4. For positive rotations we still have a distinct value of
c 2, while for negative rotations c 2 is comparatively small because of the small distance bK and the
small corresponding contribution to the restoring moment. With an alternating fixing pattern the
values c 2 are the same for both directions of rotation, provided bK is the same for both directions of
rotation. However, due to the aforementioned influence of the indentation stiffness of the fasteners
head, the total number of fasteners has to be doubled. If not, the value c 2 reduces to half of the value.
4
2.3 Sandwich panels with uplift loading
Figure 5 shows a generalised moment-rotation-relation for the spring stiffness of the connection of a
sandwich panel subjected to uplift loading. The same assumptions regarding fasteners position and
direction of rotation apply as for deadweight loading.
c 2
c 1
0
c 1
0 K
For small values of the rotation , there is no torsional restraint of the beam by the sandwich panel.
This is due to the gap between the upper flange of the beam and the adjacent face of the panel
(Figure 6). This gap is caused by the indentation of the fastener’s head and washer when the panel is
subjected to uplift loading and the fasteners are therefore loaded with tensile forces. The tensile load
induces a torsional moment and (negative) rotation in the beam, giving a preferential direction for
lateral torsional buckling and enforcing the tendency for lateral-torsional buckling of the beam.
bK
uw
After reaching a rotation of = K, the stiffness increases. At this stage, the gap between the outer
edge of the flange of the beam and the adjacent face is closed. Therefore for this load case the
governing parameter is the contact rotation K
uw Ft
K
bK k w bK
5
with Ft being the tensile forces in the fasteners caused by the uplift loading and k w being the
indentation stiffness of the fastener. For negative rotations , the rotational stiffness remains
approximately zero because contact occurs only after very large rotations due to the small lever arm
bK.
bK
uw
A significant increase in stiffness can only be found for large positive rotations. Due to the direction
induced by the tensile load in the fasteners this is a rather theoretical case. The mechanical model is
the same as for deadweight loading (Figure 4), but with a different direction of rotation. In this case,
the values c 2 are the same for uplift and deadweight loading but can not be taken into account. In
practice, there is no torsional restraint with uplift loading for applications with all of the fasteners
mounted on one side of the web. The use of an alternating fixing pattern is possible, too.
With an alternating fixing pattern the values c 2 are the same for both directions of rotation.
However, due to the aforementioned influence of the indentation stiffness of the fasteners head, with
the equal partitioning of the fasteners to both sides of the web c 2 reduces to half of its value. With
the alternating fixing pattern the following disadvantage is avoided which occurs with the
arrangement of the fasteners on one side of the web: The tensile forces in the fasteners resulting from
the uplift loading induce a torsional moment enforcing the tendency for lateral-torsional buckling of
the beam.
Most recently, the German design code for steel structures, DIN 18800-2, was updated, now
including the possibility to use sandwich panels for the stabilisation of beams against lateral-torsional
buckling. These regulations are based on the investigations described in Dürr (2008) which received
financial support by the IFBS. The basic construction of the formulae (derivation of the influencing
parameters) for c 1 and c 2 was derived from an FE-analysis whereas the parameters c1 and c2 were
derived from a statistical evaluation of test results. These regulations are only given for downward
loading and for the core materials PUR and mineral wool. Also, the parameter range for E C and tK is
restricted.
A secant value of
6
mK
c
mK
can be taken into account, using the simplified moment-rotation relation shown in Figure 8. The
necessary values and parameters are given in the following tables.
Figure 8: Moment-rotation-relation
Table 2: Parameters
parameter according to
c1, c2
Table 3
2.0 N/mm² EC 6.0 N/mm² Young’s modulus of the core material
b [mm] width of the flange of the beam
0.42 mm tK 0.67 mm sheet thickness of the outer face layer
parameter depending on the pattern of fixings
= 1.0 alternating application of fixings
= 1.5 one-sided application of fixings
= 0.0 hidden fixings
design value of the downward load to be transferred
qd
from the panel to the beam
7
Table 3: Parameters c1 and c2
Two simplified mechanical models and an idealised moment-rotation-relation were used to model the
compliance between the beam and the sandwich panel. Figure 10 shows the models for the values c 1
and c 2.The model for c 1 consists of a single spring for taking account of the indentation at the edge
of the flange of the beam. Using the contact moment we obtain
b2
c 1 k 1
4
8
The spring stiffness k 1 has the unit force per square of length because k 1 is referring to the length of
the flange. k 1 is predominantly depending on the Young’s modulus EC of the core.
For the value c 2 a model with two springs is used. The second spring is the indentation of the screws
and the washers into the outer face. We obtain
b K2 b K2
c 2
1 1 1 1
k 2 k 1 n K 2 k 1
The indentation of the fasteners is the dominating effect and therefore we can simplify this equation
to
c 2 n K 2 bK2
While the spring stiffness k 2 also has the unit force per square of length, K 2 has the unit force per
length. k 2 is dominantly depending on the Young’s modulus EC of the core and the number n of
fasteners per length.
value c 1 value c 2
b b
2 2
k 2
= f (n, EC)
k = f (EC) k 1 = f (EC)
1
q w
w 1= w 1
k 1 2
b b
2 2
bK
Numerical investigations were performed to verify the mechanical model and to study the influence
of different parameters. These parameters were the essential dimensions and the Young’s modulus of
the core and the face material. We obtained the following results.
9
- For panels with two flat or lightly profiled faces (wall panels) both c 1 and c 2 depend on the
thickness D of the panels. c 1 and c 2 increase with the thickness D. They converge to the
value of the panel with a strongly profiled outer face (usually roof panels) with similar
arrangement of the fasteners.
- Both values c 1 and c 2 increase with the depth of profiling of the outer face due to the
support by this profiling. Beyond 10 mm of depth of this profiling no further increase of this
is possible such that c 1 and c 2 there attain their limit values for a panel with a profiled outer
face. This applies for the investigated geometry with a distance of ribs of 333 mm.
- Both c 1 and c 2 increase with Young’s modulus EC of the core material with the power of
0.9. The approximation by a linear function is justified.
- c 1 increases with the bending stiffness (EI)F2 of the inner face with the power of 0.1. The
influence of this stiffness can therefore be neglected for the common parameter range (faces
made of steel with thickness 0.38 mm tK 0.,71 mm, faces made of aluminium with 0.50
mm t 0.65 mm). For faces made of GFRP a reduction factor c F is required. As expected
there is no influence of (EI)F1 of the outer face on c 1 such that (EI)F1 can also be disregarded.
c 2 increases with the bending stiffness (EI)F1 of the outer face with the power less than 0.1 so
the same applies as for c 1. There is no significant increase of c 2 with increasing bending
stiffness (EI)F2 of the inner face. This justifies the mechanical model introduced above that
c 2 only depends on the core material and the type of profiling of the outer face.
- c 1 does not increase with the square of b but with the power of 1.3 (thin wall panels) to 1.7
(thick wall panels): The actual lever arm is smaller than b because of the indentation and
because of bending of the panel. c 2 increases with the square of b.
c 1 c1 EC b 2
c 2 c 2 n EC bK2
to determine c 2. The number n of fasteners per meter length and the distance bK depend on the fixing
pattern and the direction of rotation. c 2 should be set to zero unless bK 0.5 b. For double-symmetric
beams with one sided fixing at one fourth of the flange bK = 0.75 b or bK = 0 applies, depending on
the direction of rotation. For one-sided application of fixings this means that for one direction of
rotation c 2 is always zero. In principle the same is true for alternating fixing patterns: b K is always
10
the longer lever arm and for n only the number of fasteners per length corresponding to this lever arm
can be taken into account. Attention has to be paid to the units: c 1 is non-dimensional value whereas
c2 has the dimension meter because n as defined above has the dimension m-1. EC and bK are input
with their units.
Creep tests were performed and evaluated to consider the effect of duration of loading on Young’s
modulus EC of the core material. The extrapolation of the results of these tests according to EN 14509
to 2000 hours (representing snow loading) and 100000 hours (representing self-weight loading)
resulted in values C,t much higher than the value C,100000 = 1.0 given in Dürr (2008). This increase to
almost twice the value is mainly due to the statistical evaluation and the scatter of the test results.
Finally the values C,t are provided for use with the formula
EC
E C ,t
1 C ,t
The effect of temperature on Young’s modulus EC of the core material can be taken into account by
using the reduction factor k1 according to EN 14509:
E Ct , 80 C
EC , EC , k13 EC ,
E Ct , 20 C
From the tests with sandwich panels with faces made of GFRP additional reduction factors c F = 0.38
for c 1 and cF = 0.41 for c 2 were obtained. We recommend to use cF = 0.38 both for c 1 and c 2.
The final result is summarized in Table 4. The values listed for panels with a profiled outer face can
be used for panels fixed in the lower flange of the outer face or in the upper flange and also for fixing
with or without saddle washers as well because the differences of values found both in numerical and
experimental investigations were to small to be seriously quantified.
c 2 c 2 c F n E C ,t , bK2 0
EC EC E Ct , 80 C
EC,t, E C ,t , k13
1 C ,t 1 C ,t E Ct , 20 C
b
mK qd qd b
2
Table 5: Parameters
11
c1, c2 parameter according to Table 6
cF parameter depending on the pattern of fixing
cF = 1.00 face materials steel and aluminium
cF= 0.38 face material GFRP
12
5. Conclusion
Sandwich panels increase the resistance of substructures (beams, purlins) against lateral torsional
buckling by restraining the lateral displacements and rotations. The torsional restraint by sandwich
panels can be calculated by using the mechanical model of a torsion spring with the spring stiffness
c . This spring stiffness is a combination of the bending stiffness of the attached panel c C, the
stiffness of the connection c A and the distortional stiffness c B of the beam to be stabilised.
The new rules given in the German design code DIN 18800-2 for the calculation of the stiffness of
the connection c A which are based on investigations by Dürr (2008) are presented and their range of
application is extended. The investigations for this extension were performed within the framework
of the EASIE project.
Acknowledgements
This research has received financial support from the Industrieverband für Bausysteme im
Metallleichtbau (IFBS), the German association of the producers of sandwich panels and thin-walled
profiles. This paper also presents results obtained from the research of the EASIE project. The
EASIE project has received financial support from the European Community’s Seventh Framework
Programme FP7/NMP2-SE-2008 under grant agreement No 213302. We express our sincere gratitude
for this support.
References
Dürr M and Saal H (2009) “Die drehbettende Wirkung von Sandwichelementen beim
Biegedrillknicknachweis in der Neufassung der DIN 18800-2” Bauingenieur 84: 247-253.
DIN 18800-2:2008-11: Stahlbauten – Teil 2: Stabilitätsfälle – Knicken von Stäben und Stabwerken
EN 14509:2006: Self-supporting double skin metal faced insulating panels – Factory made products –
Specifications
13
A Unified Approach for the Local Buckling Of Sandwich
Panels and Trapezoidal Sheeting
Misiek, T.
Karlsruhe Institute of Technology, Karlsruhe, Germany
(email: [email protected])
Hassinen, P.
Faculty of Engineering and Architecture, Helsinki University of Technology, Espoo, Finland and
Pontek Consulting Engineers Ltd, Espoo, Finland
(email: [email protected])
Abstract
Failure of thin-walled building components like sandwich panels or trapezoidal sheeting is normally
initiated through a local buckling of the plane elements of the cross-section. For trapezoidal sheeting,
EN 1993-1-3 gives an equation for the determination of the effective width of these plane elements
and thus for the calculation of the load-bearing capacity of these components. The cross-sectional
parts of a lightly or strongly profiled facing of a sandwich panel can be regarded as elastically
supported plane elements, whereas the elastic support is provided by the core material. For the
determination of the load-bearing resistance of sandwich panels with lightly or strongly profiled
facings, a calculation procedure to determine the effective width of the elastically supported plane
elements is needed. Some approaches for the calculation of the effective width already exist. The
papers published so far are using a modified buckling coefficient for the calculation of the buckling
strength, following Winter’s approach such as given in EN 1993-1-3. Because no generally accepted
design procedures exist, the load-bearing resistance of sandwich panels is determined
experimentally. Based on the basic principles of structural stability, the buckling strength of the
elastically supported plane element can be calculated, taking into account the material properties of
the core material and the associated buckling wavelength for minimum buckling strength. Then the
design procedures of EN 1993-1-3 for thin plate buckling of trapezoidal sheeting can be used,
expanded by the procedures of EN 1993-1-5 for taking into account the column type buckling
behaviour for buckling wavelengths smaller than the total width of the plane element. Comparison of
the test results with different arrangements to the calculated values shows a good consistency. For
lightly profiled faces, depending on the depth of the profiling, failure will finally take place through a
plate buckling of the plane elements or by a column buckling of the stiffening profiles, whereas the
latter failure mode is looking similar as the wrinkling failure of a flat facing. Comparison of the
experimental results obtained with different test arrangements to the calculated values show the
present limits of the applicability of the proposed design procedure. The differences in failure modes
are discussed.
Keywords: sandwich panels, face layer, buckling, elastic foundation, elastic half-space
14
1. Introduction
The determination of the load-bearing capacity required for the design of sandwich panels is to a
large degree based on test results. In contrast, there is a large number of references for the
determination of the load-bearing capacity of the trapezoidal sheeting, documented in national
standards such as the Swedish StBK-N5 or the German DIN 18807 as well as international standards
such as EN 1993-1-3. This is astonishing because both building components are com-parable in
materials, geometry and load-bearing behaviour. Compared to trapezoidal sheeting the foam core
provides an additional elastic foundation to the plane thin-walled elements.
In the following we will show that the calculation procedures developed for trapezoidal sheeting can
be modified to be applied for sandwich panels. This will start with the basic module, of which all
cross-sections consist: a simple plane element, like a flange of a strongly profiled face of a sandwich
panel.
The local buckling of the plane cross-sectional elements shall be taken into account in the
determination of the load-bearing capacity of the thin-walled building components. Because of the
local buckling of the plates of a medium or high slenderness only an effective width, which is smaller
than the total width, can be taken into account.
x
b
EIF = f (EF, t, F)
c = f (EC, GC, C)
z
The calculation of the effective width is based on the critical buckling stress cr. An additional
support provided by an elastic foundation increases the critical buckling stress cr. This increase is an
15
addition to the local buckling stress of the plate without foundation calculated according to EN 1993-
1-3. The increase of the critical buckling stress leads to a decrease of the relative slenderness of the
plate and thus to an increase of effective width and load-bearing capacity. Up to now, most of the
present studies capture the effect of the elastic foundation by adjusting the buckling value k as
proposed by Davis, Hakmi and Hassinen (1991). This adjustment is based more on statistical
evaluations of test results. The determination of the critical buckling stress can be based on the elastic
potential, which is shown in Timoshenko and Woinowsky-Krieger (1959) for the plate without
foundation. The calculation of the stiffness of the Winkler’s foundation out of the material
parameters EC and GC of the core material is shown in Stamm and Witte (1974). We obtain the
following equation:
2 2 2 2
2
EF t 2 a m n
cr , p
12 1 F2 m a b
2 2 2
2 1 C EC a m n
1 C 3 4 C t m a b
with
EC EC GC 2 1 C
according to Stamm and Witte (1974) to take into account very approximately the anisotropy of the
material properties of core materials such as for example polyurethane and polystyrene foam.
The effective width of the plate can be calculated from the critical buckling stress and the relative
slenderness. EN 1993-1-3 gives the following expressions:
fy
p
cr , p
1,0
beff , p 0,673
1 0,22 for p
p
bp 2 p 0,673
p p
These equations derived by Winter assume pure plate-like behaviour. This can be found for ratios =
a/b > 1,0. Because of the elastic foundation we obtain a buckling wave pattern with an aspect ratio
a/b which is smaller than 1,0. This has to be taken into account when calculating the critical buckling
stress, leading to an increase of the buckling coefficient k . Unfortunately this also results in a minor
support of the mid-part of the plate by the supports at the longitudinal edges compared to buckling
patterns with higher aspect ratios. Therefore the load-bearing capacity is lower than the one
calculated according to Winter’s equation. Column-like buckling behaviour of the plate has to be
taken into account. This can be done by utilising the procedure given in EN 1995-1-5. In this case an
interpolation between the plate buckling curve according to Winter and the column buckling curve c
16
is done. This interpolation is based on the ratio of the elastic critical buckling stress of the plate and
the column:
c p c 2 c
with
0 1,0 where cr , p
1
cr , c
and
2 2
EF t 2 m 2 1 C EC a
cr , c
12 1 2
F a 1 C 3 4 C t m
fy
c
cr ,c
2
0,5 1 0,21 c 0,2 c
1
c 1,0
2 2
c
The presented approach for the determination of the load-bearing resistance of a flat plate on an
elastic foundation was verified using finite-element-calculations and comparisons with test results.
The plates were modelled using four-node shell elements with a linear-elastic ideal-plastic stress-
strain relationship. For the elastic foundation, spring elements with a linear elastic stress-strain
relationship were used. The parameters such as sheet thickness t, yield strength f y and stiffness of the
foundation c (via EC, GC and C) were varied. Figure 2 shows the results of the comparison between
the theoretically derived critical buckling stress and the buckling stress based on FE-calculations.
Using this FE-model and the initial imperfection corresponding to the first eigenvalue of the model,
the ultimate load has been computed. The initial imperfection was scaled to w0 = 0,1 t. The
comparison of the computed results with the results of the procedure based on the interpolation
between the plate buckling curve according to Winter’s equation and the column buckling curve c is
shown in Figure 3.
17
600
500
400
[N/mm²]
300
cr,p,FEM
200
100
0
0 100 200 300 400 500 600
cr,p,calc [N/mm²]
Figure 2: Comparison of critical buckling stress based on the finite-element-calculations and on the
analytical expressions
1,0
0,8
0,6
[-]c,FEM
0,4
0,2
0,0
0,0 0,2 0,4 0,6 0,8 1,0
c,calc [-]
To verify the calculation procedure, a series of tests was performed. These tests were done with
trapezoidal sheeting (five ribs) and with cross-sections cut out from the sheeting (one rib). The
trapezoidal sheeting was the upper face of a panel with one strongly profiled face (Figure 4). The
18
inner flat face was removed. Thus, the trapezoidal sheeting provided the only load-bearing
component of the specimen. Due to the remaining foam core, the plane elements of the face still were
supported by an elastic foundation. The normalized effective width c was back-calculated using the
calculation procedure for trapezoidal sheeting according to EN 1993-1-3, supplemented with the
procedures of EN 1993-1-5.
20 mm
tF1 = 0,47 mm
23 mm 114 mm
35 mm
61mm
no face
According to the calculation procedure, the aspect ratio of the buckling waves should be about =
0,3. Figure 5 shows the buckling waves in the upper flange confirming the calculated value.
An additional series of tests was done with the upper face layer without any foam material. The
geometry and material was identical to the one used in the tests described above. In this case no
elastic foundation for the plate is available and we have a cross-section of a standard trapezoidal
sheeting. Results of the tests are shown in Figure 6.
In addition, the tests results of Davis and Hakmi (1991) and the results of Pokharel and Mahendran
(2003) were used for comparison (Figure 6). Davis and Hakmi performed bending tests on C-shaped
19
members with a foam core. Pokharel and Mahendran glued flat sheets of different thicknesses and
steel grades to layers of plastic foam material.
1,0
0,8
0,6
c,test [-]
0,4
c,calc [-]
Figure 6: Comparison of calculation procedure with test results: normalised effective width c
All the test results show a good correlation with the calculated values. The calculated values make
the lower boundary to the test results. Interestingly the values with the highest discrepancy between
test and calculation being on the unsafe side are the ones for the standard trapezoidal sheeting
without any elastic foundation. This is a result of the very high slenderness of the upper flange, the
relative slenderness being of the order of 5,5 to 5,6.
It is interesting to discuss about the effect of the material properties used in the calculation procedure.
The first difference arises in the test method for the determination of the shear modulus G C. Davis
and Hakmi performed tests with small cubes of the core material which were loaded with a pure shear
load. No information is available about the tests of Pokharel and Mahendran. However, it can be
assumed that the tests were done in a similar way based on small cubes. Our tests for the
determination of the shear stiffness were performed with small short-spanning beams according to
EN 14509. The latter test method might lead to higher discrepancies between the measured value and
the actual material properties, resulting in a higher discrepancy between the calculated and
experimental values c,calc and c,test.
The elastic modulus EC of the core material is normally determined as an average value
corresponding to the total thickness of the specimen, determined separately in a compression and
tensile test. For plastic foam materials made in a continuous process the local values strongly deviate
from this mean value, depending on the relative position of the sample in the thickness direction of
the panel. For these sandwich panels, investigations on the distribution of the stiffness presented in
20
Dürr (2008) show that the elastic modulus near to the faces is higher than in the centre plane of the
panel. In addition, different values of the modulus can be obtained close to the faces depending on
their position (top or bottom) during the production. These effects have to be kept in mind when
looking for the results of our tests (which were done with real panels) and presumably also for the
tests of Davies and Hakmi. For the tests of Pokharel and Mahendran which were performed with
foam material glued to the face this effect can be neglected. In fact, when calculating the load bearing
resistance with the values from the tests, calculated values are on the safe side, because the higher
stiffness near the faces gives a stiffer elastic foundation than assumed by using the mean values over
the total core thickness.
In addition to the buckling tests with trapezoidal sheeting, tests with sandwich panels were also
performed. The panels had a similar geometry as the specimens introduced before. In these tests, the
inner face was not removed, so real sandwich-type load-bearing behaviour existed. The results of the
tests were back-calculated, too. Now, the lower flange of the outer face is also under compression
loading. This can be easily seen by comparing the pictures of this flange taken after the failure
(Figure 7).
Figure 7: Upper and lower flange of the specimens after failure: trapezoidal sheeting with elastic
foundation (left) and sandwich panel (right)
The effective cross-section of the outer face can be calculated with the introduced procedure. At first,
the effective cross-section of the outer face under compression has to be determined by strictly
following the calculation procedure introduced in chapter 2 (approach A). In fact, there will be
always a difference between the calculated effective width and the test results. These differences in
the determination of the effective width will cause further differences in the calculation of the load-
bearing capacity of the sandwich cross-section. In our case, we have also the possibility to use the
effective width of the upper flange from the tests on tests described in chapter 3 (approach B) to
minimize the effect of the former differences. The use of both the both approaches allows us to
compare the results. The calculation has to be done iteratively. For the first iteration we assume a
21
constant compression stress distribution over the whole cross-section of the outer face. We obtain the
following results:
20 mm
23 mm 114 mm
11,1 mm 35 mm
Figure 8: Effective cross-section of the outer face after the first iteration
Table 1: load-bearing capacity of the sandwich panel after the first iteration
According to this calculation, the compression stress in the lower flange reaches a value of = 148
N/mm² which is below the yield strength. The effective width in the lower flange is larger than
assumed at first. Same is true for the lower part of the webs. Therefore the effective cross-section is
larger than calculated in this first step. The accuracy of the calculated load bearing capacity can be
improved iteratively. For the lower flange, this can be done by calculating the effective width with
= 148 N/mm² instead of fy = 400 N/mm². Special considerations are required for the web. We can use
the simple assumption given in EN 1993-1-3 that the effective width at the lower edge is 1,5-times
the one at the upper width were = fy. After the first iteration, the stresses in the lower flange are
further decreasing, and the stresses in the inner face increase. Load-bearing capacity is increasing
with every iteration. However, the values are converging very soon. Stress distribution based on the
fifth iteration can be accepted to represent the ultimate load:
20 mm
23 mm 114 mm
11,1 mm 35 mm
Figure 9: Effective cross-section of the outer face after the fifth iteration
22
Table 2: Load-bearing capacity after five iterations
This is a considerable improvement compared to the results of the first iteration. However, in the
example we have a difference between the values of the pure calculational approach A and the test
result. The difference is in the same order as for the simple flat plate element.
A local buckling phenomena can also be observed in tests with panels having a slightly profiled face.
Figure 10 shows the face of a lightly profiled sandwich panel (depth of lining approx. 2,0 to 2,5 mm)
in a bending test before reaching the failure load. Finally, failure will occur by global buckling or
wrinkling of the face over the complete width of the panel, depending on the support conditions at the
longitudinal edges of the compressed face.
Figure 10: Local buckling of the slightly profiled face of a sandwich panel
For an extension of our approach to lightly profiled faces on an elastic support, the results of the
buckling tests of Pokharel and Mahendran (2005) were used. Pokharel and Mahendran (2005) varied
the thickness of the faces and the width b of the plane elements of the profiling. The total dimensions
of the faces were 400 mm and 1200 mm in the width and length directions, respectively. All the
edges were fully supported. The effective width of the plane elements between the stiffeners can be
calculated according to the presented calculation procedure. After incorporating the restrictions in
EN 1993-1-3 requiring some reductions in width also the cross-sectional values of the stiffeners itself
can be calculated. The part of the critical buckling stress provided by the stiffeners can now be
calculated using
23
Is t
cr , s 1,8 E
b 2 be3
Assuming a global buckling failure of the stiffened plate and recalculating the tests with the
equations introduced in section 2 of this paper, the results shown in Figure 11 were obtained.
40,0
4,25 mm 4,25 mm
13 mm b
32,0
1,0 mm
24,0
Nu,test [kN]
16,0
8,0
b = 78,5 mm
b = 28,5 mm
0,0
0,0 8,0 16,0 24,0 32,0 40,0
Nu,calc [kN]
It can be shown that for the geometry investigated in Pokharel and Mahendran (2005) the effects of
the stiffeners can be neglected and the support at the longitudinal edges as well as the elastic
foundation of the core material are the dominating stabilising effect. In this case, failure occurred by
local buckling of the entire stiffened plate. The overall width dominated against the width of the
lining between the stiffeners.
Typical flat or lightly profiled faces of sandwich panels do not have supports at the longitudinal
edges. The wrinkling stress of these panels is usually calculated using the so-called Plantema-
equation.
cr 0,82 2 EF EC GC
The coefficient 0,82 is in most cases modified to take into account the effects of imperfections etc.
By doing so the transition form the elastic critical buckling tress to the ultimate buckling stress is
done. Further modifications of this coefficient, sometimes denominated as a wrinkling factor, can be
justified by taking into account the geometry of the faces, leading to a value derived from tests.
Further adjustments of this equation are discussed in Pokharel and Mahendran (2005).
24
This problem concerning the flat or lightly profiled faces of typical sandwich panels without a
support on the longitudinal edges could also be calculated within the framework of EN 1993-1-3. The
wrinkling failure can be interpreted as a buckling failure of the stiffeners of the face. After
calculating the elastic critical buckling load of the stiffener, a buckling curve for the stiffener
available in EN 1993-1-3 could be used. The relative slenderness and the buckling curve are given by
fy
d
cr , s
and
1,0 d 0,65
1,47 0,723 d
for 0,65 d 1,38
d
0,66 d 1,38
d
This buckling curve was developed for trapezoidal sheeting using the assumption that the elastic
foundation of the stiffener is provided by the support of the neighbouring longitudinal edges of the
stiffened element.
Further investigations have to be done before applying the design procedures of EN 1993-1-3 for the
design of lightly profiled sandwich panels. At present, the observations can be used to make a
differentiation between lightly and strongly profiled faces of sandwich panels. When calculating the
critical buckling load of the stiffener, the width bP = sW of the web has to be fully effective, thus c =
1,0. As a second criteria, column-buckling of this stiffener must not be the significant failure mode,
so d = 1,0 for column buckling of the stiffener. Then, the face can be regarded as strongly profiled
and the load-bearing capacity of the panel can be calculated as presented in the previous chapters.
6. Conclusion
The determination of the load-bearing capacity required for the design of sandwich panels is to a
large degree based on test results. In contrast, there is a large number of references for the
determination of the load-bearing capacity of trapezoidal sheeting. Because both building
components are comparable in materials, geometry and load-bearing behaviour, a calculation
procedure for sandwich panels which relies on the design principles developed for trapezoidal
sheeting seems to be possible, by taking into account the effects of the elastic foundation provided by
the core layer.
The basic principles of the design model are introduced and compared with Finite-Element-
calculations as well as with test results from different sources, showing the applicability of the
procedures. Apparently this calculation procedure can be applied to strongly profiled faces of
25
sandwich panels as well as for trapezoidal sheeting with bonded plastic foam insulation. The last-
mentioned building components have found increasing dissemination on the market in the last years.
Finally, an overview to the design of lightly profiled sandwich panels was made. Comparison with
test results showed the present limits of the applicability of the design procedures of EN 1993-1-3 if
applied to sandwich panels. Lightly profiled faces show a complex behaviour with interactions from
plate buckling and column buckling failure. This can not be captured on a sufficient safety level at
the moment.
Acknowledgements
ThyssenKrupp Steel Europe AG provided specimens for the experimental investigations. We express
our sincere gratitude for this support.
References
Timoshenko S P Woinowsky-Krieger S (1959) Theory of Plates and Shells, New York, McGraw-
Hill.
Davies M J, Hakmi M R and Hassinen P (1991) “Face buckling stresses in sandwich panels”
Contributions 1991 ECCS Nordic Steel Colloquium, p. 99-110.
Davies M J and Hakmi M R (1991) “Postbuckling behaviour of foam-filled thin-walled steel beams”
Journal of Constructional Steel Research 20: 75 - 83.
Pokharel N and Mahendran M (2003) “Experimental investigations and design of sandwich panels
subject to local buckling effects” Journal of Constructional steel research 59: 1533-1552.
Pokharel N and Mahendran M (2005) “An investigation of lightly profiled sandwich panels subject to
local buckling and flexural wrinkling effects” Journal of Constructional steel research 61: 984-1006.
EN 1993-1-3: Eurocode 3: Design of steel structures – part 1-3: General rules - Supplementary rules
for cold-formed members and sheeting
EN 1993-1-5: Eurocode 3: Design of steel structures – part 1-5: Plated structural elements
26
ECCS/CIB Joint Committee on Sandwich
Constructions: Recent European Recommendations on
Design and Testing
Davis, J.M.
83 Park Road Hale, Altrincham Cheshire WA15 9LQ UK
(email: [email protected])
Hassinen, P.
Helsinki University of Technology, Finland
(email: [email protected])
Heselius, L.
LHH Consulting Oy Ab Ltd
(email: [email protected])
Misiek, T.
Karlsruhe Institute of Technology, Karlsruhe, Germany
(email: [email protected])
Abstract
Light-weight sandwich panels, made of two metal faces separated by an insulating core material, are
modern pre-fabricated construction components. The design, manufacture and use of these
composite structural elements have required continuous development of regulations and standards.
The Joint Committee is the combination of two working groups; CIB W056 “Lightweight
constructions” and ECCS TWG 7.9 “Sandwich panels and related structures”. It therefore provides a
powerful forum for the consideration of international research and development of the design, testing
and end-use of sandwich panels. This paper introduces the Joint Committee and its work by outlining
its history and its most important publication, “European Recommendation for sandwich panels”.
The significant influence of the Committee on the recently published European Product Standard EN
14509 is also described. Focus is then directed to the latest publications of the Joint Committee,
namely the “Preliminary European Recommendations for Testing and Design of fastenings for
sandwich panels” and the “State of the art report for Design of Sandwich Panels with Openings”.
The latter is still under preparation and will probably be published in 2010.
27
1. The joint committee on sandwich constructions
The first sign of the existence of the CIB Steering Group S56 was at the International Symposium on
Low-Rise Lightweight Constructions held in Budapest in April, 1971. A significant product of this
era was CIB Publication 59 “Recommendations for the Structural Design of Lightweight Sandwich
Panels” which was published in 1978. This document gives principles for design with regard to static
out-of-plane and in-plane actions, impact loads and dynamic loads. It also gives recommendations
for design for environmental effects and fire. Finally, it discusses the properties of connections and
construction systems including the principles of the quality control. The report was based on ISO
standard 2394-1973 which introduces the principles of the modern limit state design. The report,
which was far ahead of its time, was written as a framework for national regulations and it probably
provided the first formalised recommendations for the design of sandwich panels. Nowadays, the
emphasis in Europe has moved away from national guidelines and towards harmonized standards for
the whole continent. However, historically, “Recommendations” produced by European-wide
Committees of experts have led to a similar unified outcome.
The name of Steering Group S56 was changed to Working Commission W056 and the membership
reconstituted. It held its first meeting in Espoo, Finland in May, 1984. Initially, developments in
lightweight constructions were introduced and discussed without limitation as to materials or types of
construction. Later, the subjects of W056 became focused more towards mineral wool cored
sandwich panels because of the need for guidelines for this new sandwich panel product. The
coordinators of W056 during this period were Professor J M Davies in 1986 - 89 and Lars Heselius in
1990 - 2006. Thus, W056 extended the Preliminary European Recommendations for Sandwich
Panels written originally by ECCS TWG 7.4 (see 1.2) and the outcome was CIB Publication No 148
which was published in 1993 and reprinted in 1995.
The European Convention for Constructional Steelwork (ECCS) has a long and distinguished record
of pre-standardisation work and its Technical Committee TC7 “Cold-formed thin-walled sheet steel
in buildings” has been one of its most successful working committees and, since its inception in
1974, it has produced numerous significant documents. The workload of TC7 quickly increased to
the point where a number of separate Working Groups were required and in 1983, noting the
increasing interest of the market in light-weight metal sheet faced sandwich construction, it was
decided to form a new technical working group TWG 7.4 “Sandwich Panels” to produce European
Recommendations for design, testing and good practice in the use of sandwich panels. The work of
TWG 7.4 started in 1983 under the chairmanship of Dieter Stemman taking full advantage of his
experience with German “Zulassung” documents.
28
TWG 7.4 produced “Preliminary European Recommendations for Sandwich Panels” in two parts.
Part I “Design” was published in 1991 and relates to the analysis, design and testing for the load-
bearing capacity of the sandwich panels. This significant document has provided the basis for a
number of subsequent Recommendations and Standards. Part II “Good Practice” was published in
1990 and introduces subjects such as building physics, fire, installation and erection work.
In the late nineties needs to update the European Recommendations became obvious and ECCS TC7
founded a new Technical Working Group TWG 7.9 “Sandwich Panels and related Subjects”.
Because of the mutual interest, in 1998, ECCS TWG 7.9 and CIB W056 combined to form the Joint
Committee on Sandwich Construction, which is still active today. The first two years of work
resulted in the final manuscript of the “European Recommendations for Sandwich Panels, Part 1:
Design”, which was published as CIB Publication No 257 in 2000 and ECCS Publication No 115 in
2001. The authors pay tribute to the work of Antti Helenius who acted as secretary of W056 and the
Joint Committee for more than ten years.
Based on the work of the CIB W056 Working Commission, a group of members combined together
to write a book covering all aspects of sandwich panel design and construction. Although individual
members took responsibility for individual chapters, this book is unique in that the authors took joint
responsibility for the whole and each chapter was discussed in detail at a series of meetings. The
contributions were edited by Professor J M Davies and the result was published in 2001 entitled
“Lightweight Sandwich Construction”.
The above Recommendations have formed the basis of the European product standard for self-
supporting double skin metal faced insulating panels, EN 14509, which was published in December
2006 and cited in the Journal of CEN in December 2008. Drafting of the product standard was in the
hands of CEN Committee TC128/SC11, which includes most members of the Joint Committee. The
product standard extends the European Recommendations to include all of the “Essential
Requirements” of the Construction Product Directive in accordance with formal mandates.
Because it is a European “product” standard, EN 14509 defines the requirements and the methods of
verification of the essential properties of the factory made sandwich panels. The standard does not
and cannot specify requirements for operations carried out after manufacture such as cutting, fixing
and installation. Consequently, in 2004, the reactivated Joint Committee identified a number of
subjects in which further direction and harmonization is needed. The first items to be considered
were the fastening and openings in sandwich panels.
29
2. The recent European recommendations
2.1.1 Background
The product Standard EN 14509 does not give any requirements regarding the joints and fastenings
of the sandwich panels. In order to remedy this, the Joint Committee has produced preliminary
recommendations which discuss the experimental determination of the load-bearing resistance of
fastenings, the evaluation of the test results and the principles of the design of fastenings. These
preliminary recommendations, now available as ECCS Publication No 127 and CIB publication No
320 in 2009, are based on the earlier versions of the European Recommendations for Sandwich
Panels ECCS & CIB (2000) updated to accord with current knowledge and experience.
Figure 1: Fixings of sandwich panels: typical application and fastener (right hand Figure courtesy of
Würth).
Because of the inherent rigidity of the sidelap, seam fasteners are not generally used in sandwich
construction. Attention is therefore concentrated on the connections to the supporting structure.
These may be concealed in the sidelap or may be visible and pass through the complete panel.
Whichever solution is adopted, self-tapping or self-drilling screws are generally used. Both of these
are able to cut their own threads into the substructure, however self-tapping screws require pre-
drilling. The screws have a formed drilling bit that allows the screwing and drilling in a single
30
operation. In order to produce a rainproof joint, sealing washers with a vulcanised EPDM layer are
used between the screw head and the face. The washers of the screws fixed to the upper flange of the
trapezoidally profiled facing, may take the form of saddle washers to tighten and to support the
profile.
The fasteners between the sandwich panel and the substructure may be loaded by tensile and shear
forces as well as by bending moments (head deflection) due to the thermal movements of the faces.
eC e3
F
e1 e2 direction of span
Figure 2: Test arrangement for pull-through resistance based on small-scale specimens at an end
support (ECCS & CIB, 2009).
eC e3
F
e4 e4
direction of span
Figure 3: Test arrangements for pull-through resistance with small-scale specimens at an intermediate
support (ECCS & CIB, 2009).
The tensile resistance of a fastening represents the minimum value of the pull-through resistance and
the pull-out resistance. The load-bearing capacity of the screw itself does normally not play an
important role in the resistance of the fastenings of sandwich panels. The Recommendations ECCS
& CIB (2009) deal with the pull-through resistance only. For determining the pull-out resistance
reference is made to ECCS (2008).
31
The distance between the end of the panel and the fastener has an influence on the tensile resistance
of the fastening. Therefore, two separate test arrangements have been developed to take into account
the different nature of the fastening at an end support and at an intermediate support as shown in Figs
2 and 3. Figure 4 shows the principal set-up of a test for the determination of the shear resistance of
a screw fastening. Since the influence of the external face decreases with the increasing thickness eC
of the core layer, the tests are to be performed with the largest envisaged panel thickness. As an
alternative, the sole direct load transmission between the internal face and the substructure may be
investigated.
t1 F
eC e1 u
Figure 4: Shear test assembly for screws passing through the panel (ECCS & CIB, 2009).
For the determination of the resistance of a screw to the deflection of the screw head, the shaft of the
fastener is subjected to a repeated deflection of „u‟, where u is the maximum lateral displacement of
the head. The deflection spectrum is as follows: 20.000 cycles with a deflection of 4/7 u, 2.000
cycles with a deflection of 6/7 u and 100 cycles with a deflection of u. This load spectrum is based
on the assumption of a service life of 50 years in a location in central Europe. During the test, the
screw shall not fail and, after cyclic loading, the screw has to achieve at least 80 % of the mean value
of the pull-out resistance without the repeated load.
2.2.1 Background
In the majority of buildings, functional requirements necessitate openings in the wall and roof
cladding. The current state of the art with regard to openings within the sandwich panel wall and roof
construction generally requires the addition of reinforcement in the form of additional framing to
replace the load-bearing capacity which has been removed by the opening (Fig 5). Recent research
results indicate that this additional framing is not always needed. The state of the art report of the
Joint Committee will introduce the possibility of designing panels with openings without the need for
any additional strengthening.
The cross-section of the panel which remains after cutting an opening may be able to carry the
applied loads. In this case, the opening is classified as „small‟. The evaluation of the remaining
32
resistance is evidently the primary design task and this includes verification of the resistance of the
net section to the design bending moments and shear forces at the critical points in the vicinity of the
opening taking account of any stress concentrations.
section A - A
A
Figure 5: Openings requiring additional supporting members (ECCS & CIB, 2010).
remaining cross-section
gross cross-section
Figure 6: Remaining cross-section in the vicinity of an opening (ECCS & CIB, 2010).
The ECCS & CIB report (2010) gives design formulas to calculate the remaining resistance of a panel
with openings (Fig 6).
If the remaining cross-section of an individual panel is not able to carry the applied loads, the load-
bearing resistance of the neighbouring panels can be activated. Due to the difference of the stiffness
between the panels with and without openings, the whole or a part of the loads applied directly to the
panel with openings will be transferred via the longitudinal joints to the adjacent panels (opening „A‟
in Figure 7). The most severe case is the load transfer from a completely cut sandwich panel
33
(opening „B‟ in Figure 7). While in case A the panel might have sufficient capacity to withstand the
load, case B relies entirely on the load transfer. In both cases, the neighbouring panels will receive
additional loads which can be calculated on the basis of the compatibility of the deflections in the
longitudinal joints (Figure 8). These loads are transferred by these joints and therefore, the
assessment of the strength and stiffness of the joints is essential. Furthermore, the load transfer will
result in an eccentric line load applied to the adjacent panel without an opening thus activating its
torsional rigidity and causing additional shear stresses due to the torsional moment.
Figure 7: Small and full-width openings in sandwich panel walls (ECCS & CIB, 2010).
Figure 8: Line loads on the longitudinal joint between neighbouring sandwich panels with different
bending and shear stiffness (ECCS & CIB, 2010).
The intensity of the load, which can be transferred through the longitudinal joints to adjacent panels,
depends on the bending, shear and torsional rigidity of the complete panels and, in addition, on the
shear rigidity and shear resistance of the longitudinal joint. Simple framework-software can be used
to compute the internal forces and deflections. ECCS & CIB (2010) presents the basic principles of
the design procedure. A test set-up for the determination of the stiffness and resistance of the
longitudinal joints is presented as well as formulas to determine the required bending and torsional
stiffnesses of the panels.
34
2.2.4 Large-sized openings
In some cases load transfer from the panel with openings is not possible due to the limited shear
resistance of the longitudinal joint or the limited load-bearing resistance of the neighbouring
sandwich panels. An additional frame has then to be designed to carry the whole load of the sandwich
panel with openings. This frame may be placed within the longitudinal joints of the panel (Figure 9).
longitudinal beam
Fc Fc
F1,c
F1,c
Fc cross-beam Fc
Ft Ft
longitudinal beam
Figure 9: Frame-construction within the panel using special aluminium profiles with web parts made
of plastic sections (ECCS & CIB, 2010).
The Joint Committee is the birthplace of the European research Project “Ensuring Advancement in
Sandwich Construction through Innovation and Exploitation” (EASIE). The „EASIE‟ project
commenced in October 2008 and is funded to approximately 4 million Euro with financial support
from the European Community's Seventh Framework Programme FP7/ NMP2-SE-2008 under grant
agreement No 213302.
EASIE provides support to small and medium sized enterprises to develop the design, manufacturing
and use of products related to sandwich constructions. The Joint Committee first defined subjects and
tasks of the project reflecting the needs and interests of the industries were collected. In addition,
EASIE will also study subjects relevant to the updating of the product standard EN 14509. It
includes a number of innovative subjects such as the performance of sandwich panels subject to in-
plane shear and axial resistance.
The second goal of EASIE is to disseminate existing and new information for use in practice. This
will be carried out through seminars and the production of practical guidelines and e-learning
modules. The subject areas include the principles of the design and use of sandwich panels, fastening
and fixing, properties in extreme situations and the information given by the research results of
EASIE. Practical guidelines and seminars will help and broaden the correct and safe use of sandwich
panels in Europe and ICPC.
35
4. Future tasks of the Joint Committee
The earlier work of the Joint Committee has produced independent technical harmonised guidelines
and relevant background documents for the European product standard. The technology of sandwich
panels is still evolving and the Committee seems still to have a useful role as author of technical
guidelines based on new and reliable information without pressure from commercial or other
interests. Because of this continuous development, keeping the available documents up to date is an
important task of the Committee. Future tasks are likely to include the preparation of new guidelines
on the basis of the technical information emerging from projects such as EASIE. Finally, questions
arising from environmental issues in buildings and building products may open new technical tasks
requiring a measure of European collaboration.
References
CIB (1978) Recommendations for the structural design of lightweight sandwich panels. Rotterdam:
CIB - International Council for Research and Innovation in Building Construction, S56 Lightweight
Constructions.
International Organization for Standardization (1973) General principles for the verification of the
safety of structures. ISO 2394.
ECCS & CIB (2000) European Recommendations for Sandwich Panels – Part 1: Design.
Brussels/Rotterdam: ECCS - European Convention for constructional steelwork, TWG 7.9 Sandwich
panels and related structures & CIB - International Council for Research and Innovation in Building
Construction, W056 Lightweight Constructions. ECCS Publication No 115 and CIB Publication No.
257.
ECCS (2008) The Testing of Connections with Mechanical Fasteners in Steel Sheeting and Sections.
Brussels: ECCS - European Convention for constructional steelwork, TWG 7.10 Connections in
Cold-formed Steel Structures. ECCS Publication No 124.
ECCS & CIB (2009) Preliminary European Recommendations for Testing and Design of Fastenings
of Sandwich Panels. Brussels/Rotterdam: ECCS - European Convention for constructional steelwork,
TWG 7.9 Sandwich panels and related structures & CIB - International Council for Research and
Innovation in Building Construction, W056 Lightweight Constructions. ECCS Publication No 127
and CIB Publication No. 320.
ECCS & CIB (2010) State of the art report for Design of Sandwich Panels with Openings.
Brussels/Rotterdam: ECCS - European Convention for constructional steelwork, TWG 7.9 Sandwich
panels and related structures & CIB - International Council for Research and Innovation in Building
Construction, W056 Lightweight Constructions. (will be published in 2010?)
36
EN 14509 (2006). Self-supporting double skin metal faced insulating panels - Factory made products
- Specification. Brussels: CEN - Comité Européen de Normalisation.
37
Methods to Measure the Durability of Structural
Sandwich Panels
Hassinen, P.
Aalto University Department of Structural Engineering and Building Technology, Finland
(email: [email protected])
Pfeiffer, L.
ThyssenKrupp Steel Europe AG Profit Center Color/Construction, Germany
(email: [email protected])
Abstract
Determination of the intended length of the service life is an increasingly important question and may
be soon a crucial parameter in the development and design of civil engineering structures. Length of
the service life may be limited by a reduction of the resistance or by an unsatisfactory function at the
serviceability state. To determine the service life, knowledge about the deterioration mechanism and
their influence on the structural behaviour is needed. In some cases, information about the
development of the mechanical and environmental loads during the service life is important, too.
European standard EN 14509 introduces experimental methods to measure the changes of the self-
supporting sandwich panels and further, to classify the panels to be fit for the use in external walls
and roofs. The methods are based on an indirect parameter by measuring the changes of the cross
panel tensile strength and not the properties, which are directly used in the analysis and in design.
Ageing of the cross panel tensile test specimens is based on accelerated ageing histories using a high
temperature and moisture. The method is practical and relatively easy to use, however, it does neither
tell about the changes caused by a real natural environment nor the real changes of the ultimate
resistance or the function at the serviceability limit state.
The contribution introduces the parameters which are essential in structural design and then, studies
the possibilities to measure the changes of the parameters in long-term use. The essential parameters
are the wrinkling strength of a face under compressive stresses and the shear strength of the core and
bond. These are followed by a number of other failure modes, such as the resistance at the supports
to pressure and suction loads. The information is based on tests with small-scale specimens, the
results of which will be compared to a limited number of results on full-scale specimens having been
in service and exposed to natural ageing conditions. The contribution concludes the promising
possibilities and draws outlines for development of methods for the use in practice.
This publication was prepared within the scope of the EASIE research project. The project is
sponsored in EU 7th framework programme.
38
1. Introduction
Sandwich panels in the context of this report are composite structures, consisting of a minimum of
two deck layers and a core. The term “panel” is derived from the Latin, French and Dutch word
“paneel” and means “a flat piece of construction” (Brockhaus, 1991, a). This describes the general
shape of a sandwich panel. The word “sandwich” describes generally a combination of different
layers (Brockhaus, 1991, b). A typical sandwich layer in construction consists of three layers. A
rather thick rigid core material is laminated between two thin faces. There are many possible
combinations of facing and core materials, depending on the intended use of the sandwich panel. This
work, however, deals primarily with the most common form of a sandwich panel used currently in the
building construction: factory made engineered elements, consisting of thin metallic facings of a
thickness between 0.4 and 0.75 mm and rigid foam or inorganic wool core.
The individual layers of sandwich panels by themselves have almost no load-bearing capacity. In a
sandwich panel the layers no longer act separately but are connected in a process of adhesion. Only
this rigid connection makes the structure a sandwich panel and increases the load-bearing capacity
substantially.
While the metal sheet facing alone is almost without bending capacity as is the core by itself, the
connection of the three layers creates a completely new composite structure with widely enhanced
capacities. The composite sandwich panel possesses substantially higher load-bearing capacities with
regard to bending, shear and torsion impact than the sum of its individual components.
Compared to the metal faces, the core in a sandwich panel generally possesses very limited tension or
compression stiffness. The bending moment is therefore distributed to the two faces. For a single
span, three-layer sandwich panel loaded by a distributed load, this leads to compression and tension
in the two faces that are kept apart at a fixed distance by the core.
Based on the properties of the layers and the distribution of the internal stresses, the core carries
through almost all of the shear force. Additionally, the core provides foundations to the faces of the
sandwich panel. Because of that, the properties of the core have a significant influence on the over all
performance of the sandwich panel and need to be carefully evaluated in sandwich panel design.
Losses in the stiffness and strength of the core have immediate influences on the performance of the
whole panel. Such losses can, for example, be caused by durability related degradation.
39
1.3 Durability of construction products
A survey by Sarja (2002) on the typical share of civil engineering products for European countries
shows that:
These figures emphasise the enormous economical importance of civil engineering products for the
societies. Due to the importance and costs of these products, it becomes obvious that there are high
demands to the service ability during the life span of the investment.
Economical effects are dominated by the fact that buildings generally require large investments. The
expectations in the investment can only be fulfilled if the building keeps its functionality over a
certain time period. The estimated time period depends on the type of building. For an industrial
building like a warehouse, for example, it is generally assumed that the building is designed for a
service period of 25 to 30 years. After this period a substantial refurbishment of the building is
generally required.
The most important technological influence for a civil engineering product is the guarantee of the
safe use over a time. Building materials need to be designed in a way, which guarantees an adequate
structural performance. If a structural building material cannot resist against the received load
impacts, the building is damaged or may even collapse. In a worst case scenario, this may lead to a
loss of lives and properties. Therefore, the main task in the regulation is to ensure an adequate
structural performance of the building materials throughout the whole life cycle of the structure.
Legally binding design procedures, which are very much harmonized across Europe, are the base to
guarantee a safe structural performance. For sandwich structures, such procedures are defined in the
European standard on sandwich panels EN 14509, where also requirements towards the durability of
the sandwich panels together with experimental evaluation procedures are described.
In practice, requirements to the fitness for use and to the withstanding the extreme actions during the
use are set down to a structure. The previous requirement defines the serviceability limit state and the
latter one the ultimate limit state. The structure shall maintain the properties in its environment
during its design working life, which defines the requirements on the durability of the structure. The
durability requirements may be achieved by choosing correct materials and manufacturing
techniques. As a result, the deterioration of the structure will stay small enough or the function and
resistance will be guaranteed by regular inspections and refurbishment. In constructions, the first
principle is used in the major cases of structures.
The available durability tests give information about the changes of the strength on a macroscopic
level without explaining the physical and chemical mechanisms affecting the deterioration of the
material. It is believed, that the durability test is able to classify if the product is to fit for the purpose
40
or not. The durability test methods and the loading histories are therefore strongly dependent on the
application.
The European standard, EN 14 509, makes in chapter 5.2.3 a statement concerning durability;
“Durability and other long term effects.” Detailed information on testing scenarios are written down
in the normative annex B, “Durability testing method for sandwich panels.”
Currently, the durability performance is only evaluated qualitatively. This means that the loss in cross
panel tensile strength under an artificial, durability accelerating climate must stay within certain
boundaries over a time. The cross panel tensile strength is seen as an indicator for over all panel
performance. Depending on the type of degradation pattern (Fig 1), a core material is either fit for the
application or not. The boundaries chosen for an acceptable loss of the strength are, however, lacking
the scientific foundation. Even a loss of 60% of the initial tensile strength can be acceptable in
accordance with the standard. In particular, such a loss in the tensile strength is allowed without any
impact on other relevant design parameters. Because it is not possible to determine the remaining
wrinkling stress after the artificial ageing of the tensile strength, a new small-scale test to study the
development of the wrinkling stress is developed.
The wrinkling of the face under compression is a typical and very important failure mode for
sandwich panels with membrane-like faces (Fig. 2). As stated before, in a loaded sandwich panel, one
41
face is in tension while the other one is loaded by compressive stresses. A membrane under
compression has, on its own, no stability, but it fails immediately through buckling. In a sandwich
panel, the core provides a lateral support for the thin face and prevents the early buckling failure. The
lateral support is activated when the face deforms in a wave-like pattern and induces stresses in the
core material. The loss of the tensile strength of the core or the bond results in an immediate
wrinkling failure of the face. Wrinkling is one of the most critical failure modes in a sandwich panel
construction and it very often determines the ultimate limit state.
The wrinkling stress is defined as the ultimate stress, a face can take in compression before it fails in
buckling. The compression in the panel face is, in practice, induced by a bending moment that can, be
caused by wind loads or temperature loads in multi span panels. Determination of the wrinkling stress
of a sandwich panel generally requires full scale testing. The full scale test may be carried out by
subjecting a simply supported beam to four line loads.
In a test setup according to figure 3 the ultimate bending moment (Mu) and the wrinkling stress ( w)
for a lightly profiled panels are given by:
FBu L Mu
Mu w
8 and eC A1 (1), (2)
where Fu is ultimate load including self weight of panel and loading equipment
L span of the test specimen
eC distance between the centroids of the faces
A1 cross-section area of the face in compression
In order to be able to determine wrinkling stress after an artificial ageing, it is necessary to develop a
small scale wrinkling test set-up which allows the determination of the wrinkling stress or more
precisely the time dependent change of the wrinkling stress. The small sample can then fit into a
climate chamber and tests on the aged samples can be performed.
The mechanical parameters necessary for the design may be obtained on the basis of the small scale
testing. An exception is the wrinkling strength. For a durability research this poses the problem that it
is not possible to determine the wrinkling strength after an artificial ageing. Therefore, a small scale
test setup determining wrinkling strength has to be developed. When developing a small scale
42
wrinkling test for sandwich panels, it is the general idea to no longer induce compressive stresses in
the face through a bending moment but to cut out a small panel area and expose it to a direct axial
load in the direction of the plane. Baehre (1988) and Pfeiffer (2000) have previously tried to establish
small scale wrinkling tests for the design of sandwich panels. However, it was not possible to
determine the exact wrinkling capacity of a sandwich panel through small scale testing. The main
problem in the earlier work was in finding a way to introduce the in-plane loads without damaging
the thin face locally, a problem that does not occur in a full scale test. Special load application
devices help to overcome the problem. During the test series it has been found that the full-scale
bending test loads the sandwich panel twofold. In the upper face, the in-plane bending stress meets
the deformation caused by the line loads, which evoke the actual bending moment and compressive
stresses in the face. For a direct comparison between six point bending test and a small scale
wrinkling test, this load combination must not be neglected. The results gathered in this project
suggest that the research conducted by Baehre (1988) would have had a positive outcome if the local
deformations were taken into consideration. With the results on hand, it was not possible to find a
small scale test defining the wrinkling strength in the same way as a full scale test does.
Figure 3. Arrangements of a full-scale and a small-scale wrinkling test with descriptions of the
distribution of the bending moment and compressive stress.
The new test combines the two previously used test methods. Instead of causing compression in the
upper face through a bending moment, the small scale test loads directly the face in a vertical, in-
plane direction (Fig. 3). While the authors in their earlier work used a wooden T-shaped load
application device, which was glued to a rectangular specimen with completely flat faces causing
compression in one face, Baehre (1988) used a micro-profiled specimen with T-shaped load
application devices made of steel. The load application device was not glued to the specimen but
rather clipped onto it. In both cases the T-shaped device was chosen to gradually introduce the load to
the specimen without causing local failures at the point of the loading (Fig. 4). Baehre failed to find a
43
small-scale test, giving similar results as obtained in a full-scale bending test. The reasons for this
were the local deformations in the compressed face under the loading points in the full scale test.
Such deformations weaken the observed wrinkling strength as they cause an additional local moment
and additional stresses in the face. For a small scale test comparable to a full scale test this effect
must be accounted for.
Figure 4. Test set-up of a small-scale wrinkling test from the left to right; Baehre (1988), Pfeiffer
(2000) and Pfeiffer (2004).
Figure 5. Introduction of the load in a small-scale wrinkling test. For illustration purposes, the
specimen is separated from the load introduction device. The total load (F) is transferred to the
specimen directly (Fdirect) and through the shear force (Fshear).
For this work, a combination of the two previously described small-scale tests, determining the
wrinkling stress, was chosen. The profiled specimen from Baehre was enlarged to a bone-shaped
sample, again using T-shaped load application devices at both ends of the specimen. This time the
application devices were both glued and clamped to the samples. Only the combination of gluing and
44
clamping together with the bone shape hinders the specimen from failing early through local
crushing. Gluing the load application device to the samples allows the load to be introduced to the
panel faces over a larger surface area, reducing concentrated load effects. The load is introduced
through a shear force in the glue area and directly to the top and bottom ends, where the specimen
touches the T-shaped device (Fig 5).
To study the effect of the local deformation at the load introduction areas in the full scale test,
samples with and without local deformations were studied. A three step approach was taken. First,
samples with completely undisturbed faces were tested, giving the maximum wrinkling stress that can
be obtained. In a second step, the faces were dented prior to executing the small scale test. The depth
of the imprinted deformations was of the same dimensions as observed in the full scale test. In a third
step, the effect was simulated through a bi-axial loading in the small scale test. An additional load
device was clamped to the specimen loading it in a perpendicular direction. All obtained results were
compared to the results of the full scale bending test. The last procedure resulted in the most accurate
results.
Preparing of the bone-shaped specimens requires careful machining of the two metal sheets and the
core. To simplify the preparation work, a test set-up has been developed, which base on a rectangular
specimen. To avoid the failure at the end and to create a wrinkling failure in the mid-part of the
specimen, the specimen is loaded by an additional transversal load (Fig. 6).
a) b) c) d)
Figure 6. a) Rectangular small-scale test specimen, b) initial displacement caused by the transversal
load (initial deformation equal to the face thickness, w0 tF), c) displaced mesh and d) displacement
caused by the transversal and axial load (wu 3 tF).
45
Transversal load causes an imperfection in the face and compressive stresses in the core layer. The
level of the transversal load has to be high enough to cause an initial imperfection to the wrinkling
failure and on the other hand, lower than the elastic limit of the core layer (Fig. 7). Experiments on
polyurethane-foam and mineral wool cored specimens have shown the test set-up to give acceptable
modes of wrinkling failures.
1
axial stress in relation to
0,8
wrinkling stress
0,6
Fdef.1
Fdef.2
0,4
Fdef.3
0,2 Fdef.4
0
0 0,0005 0,001 0,0015 0,002 0,0025 0,003
axial compressive deformation
Figure 7. Influence of the transversal load on the stress-deformation behaviour of the face in the test
described in Fig. 6. Transversal loads Fdef.i cause the transverse stress of 83, 63, 42 and 21 kN/m2 in
the core. The axial stress of the face is compared to an analytically determined ideal wrinkling stress
of 125 N/mm2 in the case.
In order to calibrate the test setup, a test series, comparing the large-scale and small-scale test
arrangements of the wrinkling stress was undertaken. All specimens in this series were taken from the
same production batch. The full-scale panels were first tested as simply supported beams as described
previously. The test panels had a PUR core between the two steel faces. The wrinkling failure
obtained in the full scale tests resulted in an average wrinkling stress of 180.94 N/mm² (individual
test results between min. 180.75 N/mm² and max. 181.27 N/mm²). The wrinkling failure, obtained
from the small scale test without the additional effect of the transversal load, averaged to 228.73
N/mm² (individual test results between min. 223.06 N/mm² and max. 234.76 N/mm²).
46
Figure 8. Local deformation in the loading point causing an early failure in the six point bending test.
The discontinuity (d) occurs under all loading points but interacts with maximum bending moment at
the indicated point.
Comparing the results given by the two methods, a difference of 26.4 % can be found. This difference
is not acceptable if the determination of the accurate wrinkling stress is required. The reason for the
observed difference lies in the test setup of the full scale bending test. In a six point bending test the
load is transferred to the panel in four line loads (Fig. 8). At the point of the load introduction, this
results in deformations of the face causing a failure at the inner point of the loading beam, where
maximum local deformations and the maximum bending moment interact. In order to achieve the
same results from the small scale testing as determined through full scale testing, this deformation
effect needs to be taken into account. An additional loading device generating a transverse load, was
introduced in the test. This additional device loads the panel perpendicular to the faces, causing the
same local deformation as obtained in the full scale test. The load is applied through two steel
profiles, having the same dimensions as the ones used in the full scale test. The loading is controlled
with the help of two screws and measured in a load cell. At an average of 178.79 N/mm² (individual
test results between min. 177.07 N/mm² and max. 180.50 N/mm²) the results obtained from the setup
show good correlation with the results of the full scale bending test. When comparing the average
results, 98.8 % of the strength determined in the full scale bending tests could be reached.
The results indicate clearly that it is possible to use the small scale testing to determine the accurate
wrinkling strength of a sandwich panel. It is, however, necessary to simulate the full-scale test
accurately and to take into account even the small details, like the local deformations at the load
introduction points.
When lacking knowledge of the maximum load obtained in the full scale test (Fmax), which is the case
when no full scale test has been carried out on a particular panel, it is necessary to gradually increase
47
the deformation load relative to the load parallel to the face. The load needed on the deformation
device can be calculated on the basis that the stress parallel to the face shall be equal in the both tests.
This means, that a certain load in the small scale test, corresponds to a certain load in the full scale
test, both causing the same stresses parallel to the surface. At the same time this load determines the
load on the additional load device and can be calculated as described in the following:
The membrane stress in the flat face in the six point bending test can be written by
F6 P L
f 6P (3)
8 eC A f 16 P
where σf6P is the stress in face in 6 point bending test
F6P total load on panel in 6 point bending test
Af16P: cross sectional area of the face in compression in 6 point bending test
FSC
f SC (4)
A f 1SC
where σfSC is the stress in the face in the small scale wrinkling test
FSC total load parallel to the face in the small scale wrinkling test
Af1SC cross sectional area of the compressed face in a small-scale test
A simulation of the full scale test through the small scale test requires an equal stress parallel to the
face (wrinkling stress). Equating the equation 3 and 4 gives the load F6P
8 FSC eC b6 P F6 P 2 FSC eC b6 P
F6 P which can also be written as (5)
L bSC 4 L bSC
Local stresses in perpendicular direction to the face in the load introduction point are
F6 P
def .6 P full-scale test and (6)
4 b6 P B
Fdef
def .SC small-scale test (7)
bSC B
48
For comparison of the two tests, the local transverse stresses need to be equal, which results in
F6 P bSC
Fdef (8)
4 b6 P
The load in the additional deformation device can now be calculated, depending on the axial loading
of the sample, the distance between the centroids of the two panel faces, and the span length of the
static system in the six point bending test.
4. Conclusions
The results presented in the paper indicate that the small-scale wrinkling test can be used in the
evaluation of the wrinkling stress. The test requires extensive preparation work such as bone shape or
rectangular cutting and gluing and fixing of the load application devices. Laboratories dealing with
sandwich panel technology on a regular base can adopt the test with some additional work. For
further work on the durability related degradation of the wrinkling stress, it is sufficient to consider
only the durability related changes and thus the relative values. In future research work the test
arrangements without the additional deformation device will be studied. Also a simplified rectangular
wrinkling test set-up will be developed further.
The shear and compression strength of the core is determined in most cases on the basis of the small-
scale tests. Thus, the durability of the shear and compressive strength can be studied by exposing the
specimens to accelerated deterioration processes in a climate chamber. These studies will contribute
further information to the evaluation of the durability of the structural sandwich panels.
References
Sarja A (2002) „Neue Anforderung an die Dauerhaftigkeit von Konstruktionen“ (New requirements
towards durability of constructions), Der Prüfingenieur 20: 46-53.
EN 14 509 “Self-supporting double skin metal faced insulating sandwich panels – Factory made
products – Specification”
Baehre R (1988) Experimentelle Ermittlung der aufnehmbaren Knitterspannung von ebenen und
leicht profilierten Stahldeckschichten, Research Report, University of Karlsruhe
49
Pfeiffer L (2000) Der Einfluss der Zugfestigkeit auf die Knitterspannung bei Sandwichelementen,
Final Diploma Thesis, Fachhochschule Mainz (D) and Häme Polytechnic (FIN)
50
Reliable Composite Roofing: Learning From
Experience
Roberts, K.
Roberts Consulting, UK
(email: [email protected])
Abstract
In reviewing the findings of roof inspections there are often common issues that recur. When the
problem is repeated for the same roof system on different projects there is an opportunity to learn
from the experience and to change practice. The paper asks the question ‘how should this
information be shared for the common good?’
Investigations are often commercially sensitive and occasionally the subject of litigation. If
information is to be shared through a public forum it is essential that it is fair, balanced and
objective. The paper examines previous means in which common construction faults have been
recognised and publicised in an anonymous and constructive way.
The paper refers to the CIB / RILEM Roofing Materials and Systems Task Group that was
established in 2007 to improve our understanding of the reliability of roofing systems and
specifically to identify recommendations that can improve reliability, such as by ‘learning from
experience’.
The findings are described from two investigations into the weathertightness of composite panel roof
systems laid on shallow slopes. Sliding movement has been observed at end laps, highlighting the
need for the fixing and sealant details to be designed to accommodate such movements.
By developing appropriate means to share our experience in a constructive way we can improve the
reliability of the roofs we design and build.
51
1. Introduction
Building owners, their professional advisers and main contractors are becoming increasingly
concerned at the number of recently completed buildings where there are recurring rainwater leaks
through the roofs several years after completion. The problem appears to be widespread in the UK
and occurs on both pitched and flat roofed buildings. For building owners and facilities managers the
reliability of their roofs is an important topic.
The author is a civil and structural engineer who specialises in roofing and cladding, working
throughout the UK and Ireland. Over the past nine years the author has inspected more than 120
different roofs of different types in all parts of the UK and Ireland. The reason for undertaking 80%
of the surveys was because of reports of water ingress through the roofs. The surveys were generally
commissioned after the roofing contractor and design team had carried out some remedial works in an
attempt to cure the leakage, indicating that the roofs were not readily repairable. Not all roofs of new
buildings leak although there is clearly the potential for improvement. This paper offers constructive
feedback.
2. Reliability studies
Over the past three decades there has been a growth in reliability engineering studies, particularly in
the aerospace, vehicle production and electronics industries, improving the in-service performance of
the assemblies. Consumers are acutely aware of the problem of less than perfect reliability in
domestic products such as televisions and cars, and have now come to expect these products to work
first time and continue until they become obsolete. These studies have been highly developed in
Japan where quality and reliability were adopted as national priorities. Building owners expect the
same levels of service from their roofs, in particular meeting the basic requirement of providing a dry
internal space.
Reliability has been defined by O’Connor (2002) as ‘the probability that an item will perform a
required function without failure under stated conditions for a stated period of time’. A crude
measure of the reliability of a roof is the number of times the roofing contractor needs to be called
back to site to resolve a problem.
There are many reasons why a roof might leak and result in a need for remedial works. Knowing as
far as is practicable the potential causes of failures is fundamental to preventing them, although it is
rarely possible to anticipate all of the causes and a level of uncertainty needs to be taken into account.
The CIB W83 / RILEM Joint Committee on Roofing Materials and Systems comprises of roofing
specialists drawn from more than 15 countries. In 2007 a Task Group was established to develop our
understanding of the reliability of roofing systems, and specifically to identify and prioritise practical
actions that can deliver improvements.
52
Introducing element redundancy is encouraged at the design stage such that if one layer is not perfect
and does not perform there is a secondary back up. The double layer roof, with a primary weathering
outer face and a secondary underlayer that drains freely out of the building, was described in a paper
presented by Roberts (2007).
Some recommendations are no more than common sense, such as engaging competent tradesmen with
the appropriate training and experience to carry out the works, supported by active field inspections
and testing to check the quality of materials and work. A recurring theme in the Task Group
discussions is the retirement of experienced personnel, losing the knowledge built up over years.
This will create foreseeable problems as construction output increases when the economy grows
again.
Learning from experience is another recommendation that can lead to the avoidance of problems.
Constructive feedback on the completion of a project can result in product developments. In a paper
written by Dr Mann (2008) the importance of learning from failures at the interfaces within a
construction project was highlighted. The paper reviewed a wide range of problems and summarised
the findings of subsequent enquiries. These studies identified poor communications between the
parties as a principal failing, recognising that the human interface within any engineering project is
perhaps the riskiest of all. The paper concluded that ‘there are many lessons to be learned but
corporate memories are weak and it is incumbent on every engineer in each generation to study
failures and gain wisdom from them.’
The choice of composite panels for low slope roofing applications has become popular in the UK and
Ireland over the past decade. The sandwich panels comprising of an insulated core and thin metal
faces, have been successful because they are light in weight, energy efficient and can be easily
handled and erected. Members of CIB Commission W56 have contributed to a book edited by
Professor Davies (2001). This offers an authoritative manual on many aspects of composite panel
construction, including thermal, acoustic and fire performance, structural behaviour and mechanical
testing. The European Convention for Constructional Steelwork has also published European
recommendations for sandwich panels. Based on this work a European Standard was approved in
June 2006 for the specification of self supporting double skin metal faced insulating panels, BS EN
14509.
The author regularly carries out independent inspections of roofs and is generally commissioned by
building owners, their professional advisors and general contractors. Over the past four years there
have been twenty buildings inspected with composite panel roofs laid at shallow slopes. The
common complaint has been intermittent rainwater leakage into the buildings. At the time of the
inspections the age of the roofs varied between 12 months and 10 years. The sizes of the roofs were
between 500 m2 and 40,000 m2. The panels had been manufactured by three different companies and
installed by different roofing contractors.
53
Within chapter 3 of the Lightweight Sandwich Construction book there is only a limited treatment of
the watertightness issues at joints, with no reference to end laps, fixings and ancillary penetrations.
A test method for watertightness is offered which could be of assistance, although it is understood
that this is generally applicable to external wall systems with few manufacturers using the test for
shallow pitch roofing applications. BS EN 14509 offers a specification for the panels themselves and
clause 5.2.6 states that ‘where required, the water permeability (resistance to driving rain) of a
complete assembly of sandwich panels shall be assessed.’ This includes site applied seals, flashings
and fixing methods, and recognises that the watertightness of the assembly is a function of its
installation.
Leaks through roofs have been the cause of disputes, sometimes resulting in litigation. Each party
instructs a different surveyor to carry out detail examinations and prepare written reports. It is
common in the UK for the experts to be directed to meet, to discuss their findings and to seek
agreement on causation. These exercises can produce information that would be useful to the wider
industry if shared through an appropriate forum. It is difficult to obtain a true assessment of the
national scale of the water ingress problems with shallow pitch composite roof panels, with no
independent survey results publically available. Informal discussions with roofing contractors
indicate that they also come across such problems, suggesting that the cases inspected by the author
are not unique and that further examination would be worthwhile. There is an opportunity to learn
from recent experiences.
During heavy rain there were more than 500 active leak positions within the food processing plant,
with the extent becoming progressively worse over the five year period since completion. Each roof
slope comprised of two long composite panels with an end lap detail at mid slope. More than 95% of
the active leaks were directly below the composite panel end laps.
54
Figure 1: typical composite panel end lap
On closer examination it was found that the end laps were 150mm long with two lines of 9x3mm
butyl sealant tape, see Figure 1. There were no tail stitching screws originally installed. The sealant
was found to have perished with small pieces crumbling away. On the day of the inspection there
were cold weather conditions and it was seen that there was contraction in the composite panels, with
the end lap joint opening by up to 5mm, see Figure 2. Additional end lap stitcher screws had been
installed as a remedial measure, although after a season these had worked loose.
Evidence of movement of the composite panels relative to fixed structural steel penetrations was also
observed, with cracks in rigid reinforced coatings. Elsewhere a lack of fit between the in-plane
rooflights and the metal faced composite panels contributed to the water ingress. It is noteworthy that
55
the building was situated on high ground in a severely exposed location with the design roof slope at
a minimum for profiled sheet roof cladding with pierce fixings.
The case study observed the sideways movement that can occur at the ends of long lengths of
unrestrained composite panels. The recommended position of the primary fixing screws at end laps
has subsequently changed to pass through overlapping panels, as shown in the Metal Cladding and
Roofing Manufacturers Association Technical Paper 16 (2004). This will offer the benefit of
restraining the ends of the panels. These fixing screws will be subjected to significant thermally
induced shear loads, particularly in cold winter conditions. In addition the recommended sealant size
has changed from 9x3mm to 6mm diameter.
More than 40 active leak positions were reported below the aluminium faced composite panel roofs
of an educational building laid to very shallow slopes in the severely exposed location. One surveyor
remarked that if ever there was to be a case study on how a composite panel roof should not be
designed and built, then this was it.
In strong windy conditions the aluminium snap-on cappings became detached resulting in water
ingress at side laps. The large number of flue penetrations at the bottom of the shallow roof slopes
created long term ponding upslope, with wind driving water up and under the crown fixed rooflights
with inadequate kerb heights.
56
Figure 3: general view of the composite panel roof
The roof became the subject of a legal action which was eventually resolved by mediation.
Subsequently all of the aluminium faced composite panel roofs were encapsulated using a polyester
reinforced elastomeric liquid coating, with new upstands to support the barrel vault rooflights. It is
notable that it took six years for the matter to be resolved after practical completion of the building.
The consequential financial costs were significant. It is experiences such as these that are a motivator
for building owners and contractors alike to seek improvements to the reliability of the roofs we
design and build.
Figure 4: large number of flue penetrations at the foot of the roof slope
57
6. Sharing information for the common good
Investigations are often commercially sensitive and occasionally the subject of litigation. If
information is to be shared through a public forum it is essential that it is fair, balanced and objective.
The names of parties and products should not be given.
In the first instance when there is a difficulty with a specific product or system it would be reasonable
to expect the manufacturer to identify the problem and the means of resolution. This information
should then be circulated to the manufacturer’s own network of approved installers. This is a
procedure that some manufacturers do follow and should be encouraged.
With time and further experience it would be appropriate for the manufacturer to update their
published technical literature, advising specifiers and designers of the known product limitations.
This does not always happen.
At one time the BRE was a publicly funded organisation that undertook much of the government
sponsored research into the performance of buildings. Where there were common problems repeated
then concise Defect Action Sheets were prepared, circulated and used as effective training material.
One of the roles of the BRE is to offer an Advisory Service to the construction and property industry.
Over the years an enormous amount of information was gathered about the behaviour of buildings,
including the performance of roofing and cladding elements. Much of this knowledge and experience
was retained by members of the BRE staff, many of whom have since retired or moved on to
alternative employment.
The findings from BRE studies on roofs were collated by Harrison (1996). The book was addressed
to building surveyors and other professionals, such as architects and builders who maintain, repair,
extend and renew the national building stock. Within the preface Harrison recognised that there is no
shortage of written guidance on roofing ‘as BRE has consistently said for many years. The problem
is that people do not use the guidance that exists.’
One of the roles of a trade association is often to collate and circulate findings from problems
reported by members, seeking to promote best practice. The Building Employers’ Confederation used
to provide a technical advisory service to the building industry with technical bulletins issued
quarterly between 1974 up until April 2000. Sadly this service is no longer offered. Another
organisation that has also recently closed is the Flat Roofing Alliance, which has been producing
information sheets since 1988.
58
Contractor in-house advisory services
Many of the large national general contractors used to maintain a central engineering support service
that undertook internal investigations and then shared the findings within their company. This proved
to be an effective way in which the company could learn from their own experiences, to avoid
problems on future projects and thus improve efficiency and reputation. However, this central
service came at a financial cost which, unfortunately, has been seen as non-essential by some and has
fallen by the wayside as a result of budget cuts.
Trade magazines publish technical articles, summarising findings from investigations and offering
constructive advice. This includes the long running series of technical notes published in the roofing
magazine RCI and circulated to 7,500 people within the cladding industry.
Conferences such as CIB ’10 are also an opportunity for delegates to listen and learn from the
experiences of others, although again financial restrictions may limit attendance at such meetings and
can be a barrier to this means of knowledge transfer.
Looking to the future, further growth is seen in the use of the internet as a medium for transferring
knowledge. In particular more widespread use of question & answer forums is foreseen which can
give outline feedback and reports on problems encountered, often sharing overseas experiences. The
challenge will be in moderating the information, attempting to check that the facts are fair, balanced
and objective.
7. Conclusions
There have been reports of intermittent rainwater leakage through laps and penetrations in composite
panel roof systems laid to shallow falls and particularly on exposed sites. Site observations have
identified sliding movement at end laps. The fixing and sealant details should be designed to
accommodate these movements. The satisfactory performance of standard laps should be
demonstrated for shallow pitched roofs using the water permeability tests set out in BS EN 14509.
Penetrations such as rooflights and vents further increase the risks of unwanted water entry and can
be minimised by introducing proper upstands.
There is an opportunity to learn from these experiences. In the first instance it should remain the
responsibility for the system manufacturer to maintain and keep up to date technical information
regarding the installation and performance of their systems. Historically independent feedback on
building defects has been provided by the BRE and trade associations, although this work is
diminishing. Technical articles in journals and papers presented at conferences can be effective
means to transfer knowledge.
59
In the future discussion forums on the internet are likely to be used for sharing information, although
the need for independent moderation is recognised. By developing appropriate means to share
feedback in a constructive way we can learn from experience and improve the reliability of the roofs
we design and build.
References
O’Connor PDT (2002) Practical reliability engineering, Fourth Edition, Chichester, Wiley.
Roberts K (2007) ‘Reliable roofing: the double layer roof’, Proceedings of the International
Conference on the Building Envelope and Systems Technology, March 2007, pages 151-156, Bath,
UK.
Mann AP (2008) ‘Learning from failures at the interface’, Proceedings of Institution of Civil
Engineers, November 2008, London, UK.
Report 62 (1990) ‘Preliminary European Recommendations for Sandwich Panels: Part II Good
Practice’, European Convention for Constructional Steelwork, Brussels.
Report 115 (2001) ‘European Recommendations for Sandwich Panels: Part I Design’, European
Convention for Constructional Steelwork, Brussels.
BS EN 14509 (2006) ‘Self supporting double skin faced insulating panels – factory made products –
specification’, British Standards Institute, London, UK.
Technical Paper 16 (2004) ‘Guidance for the effective sealing of end lap details in metal roofing
constructions’ Metal Cladding and Roofing Manufacturers Association, Wirral, UK.
Harrison HW, Trotman and Saunders (2009) Roofs and roofing: performance, diagnosis,
maintenance, repair and avoidance of defects, Third Edition, Building Research Establishment Press,
Garston, UK.
60
Openings in Sandwich Elements
Warmuth, F.
Institut für Stahlbau und Werkstoffmechanik, Technische Universität Darmstadt
(email: [email protected])
Lange, J.
Institut für Stahlbau und Werkstoffmechanik, Technische Universität Darmstadt
(email: [email protected])
Abstract
Sandwich panels with flat or lightly profiled faces often form the façade of a building. Due to
different requirements like windows, doors or ducts, it is necessary to cut openings into the face of a
building. At present, openings in sandwich panels require an additional substructure, which transfers
the loads to the main structure.
The aim of the project is to find a possibility to avoid this kind of substructure. Basically, there are
two possibilities:
For small openings in a panel, it seems to be obvious to look at the single element only. If the panel
is not able to carry the load, it has to be strengthened. Therefore one solution could be a supporting
frame, which is integrated in the opening and diverts the loads around it. Also additional profiles –
integrated in the joints - could be used to strengthen the element stiffness of the panel and to reduce
the stress concentration in the corners of the openings.
For larger openings it is necessary to include the neighbouring elements in the load transfer.
Therefore, detailed information on the rigidity of the longitudinal joint and the torsional rigidity of
the elements is needed. A calculation model shall be developed to describe the load transfer.
First test results for a single element and an element formation with window openings are presented.
These tests are part of the European research project EASIE (www.easie.eu). In addition,
comparative calculations with existing calculation proposals for openings without window frames are
shown.
The long-term objective is to generate a calculation model for sandwich elements with different kind
of openings. This model should be only dependent on parameters which can be derived from some
basic tests.
The EASIE project has received financial support from the European Community’s Seventh
Framework Programme FP7/NMP2-SE-2008 under grant agreement No 213302.
61
1. Introduction
Sandwich panels often form the outer shell of buildings. As façade elements they perform different
tasks at the same time. Sandwich panels do not only seal the building, they also transfer the loads to
any kind of substructure and form a very good thermal insulation.Sandwich elements consist of two
thin covering sheets, which enclose the core material. The sheets are usually made of galvanized
steel. As core material, different materials with good insulating properties are used. The most
common core materials are polyurethane (PUR) foam and mineral wool. Depending on the estimated
use, the panels have thicknesses between 40 and 200 mm and a length up to 24 m. Due to different
requirements like windows, doors or ducts, it is necessary to cut openings into the face of a building.
These openings always result in an attenuation of the sandwich panel. At present, openings in
sandwich panels require a substructure, which transfers the loads to the main structure. These
replacements lead to an additional effort of erection and are visually not attractive. The aim of the
presented project is to find a possibility to avoid this kind of substructure. In this paper, first test
results of sandwich elements with windows are presented. Two different kinds of openings have been
tested. In the first case, the opening has been cut in one panel and the window frame has been bonded
in the opening. In the second case, the width of a complete panel has been replaced by a window. The
load had to be transferred to the neighbouring elements.
Basically, it can be distinguished between different kinds of openings. The most important
differentiation is the size of an opening. We can distinguish between little openings like penetrations
for supply lines and large openings for windows and doors. A second criterion is the location of the
opening. From central openings in one panel to openings across longitudinal joints everything is
possible. In recent years, some research projects on openings in sandwich panels were conducted.
Especially for openings within one sandwich panel calculation proposals exist. In 1998
Courage/Toma (1998) published a possibility to calculate the stress peaks around openings and the
consequences on the bearing capacity of the sandwich panel.
Proposal Courage/Toma:
If 0,1 0,8 :
62
Marc Böttcher (2005) proposed a more simple approach, in which he also included openings, which
are not arranged at midspan.
Proposal Böttcher:
In both publications, the wrinkling stress is reduced depending on the relation between the width of
the opening and the width of the panel. Depending on the size of the opening, the results of the two
calculation models are more or less similar. For little openings, these calculation models make sense.
For larger openings, the formulas given above lead to a very high loss of bearing capacity.
For large openings it is not useful, to confine the load transfer only to the weakened element.
Especially at openings with the width of one panel (complete cut-out) there has to be an interaction
with the neighbouring panels. The loads have to be transferred over the longitudinal joint and the
unweakened neighbouring panels can participate in the load transfer. The German association for
light metal structures IFBS (2006) published a paper, how to calculate these kind of openings. In this
calculation model it is assumed, that the load of the opening directly affects a torsional moment in the
neighbouring element. For the verification of the system, the bearing capacity of the neighboured
element and of the joint has to be proved.In all the publications mentioned above, only openings
without any kind of strengthening were discussed. If windows are built into a sandwich façade, there
exist always window frames, which almost certainly have a stiffening effect on the whole panel and
especially on the covering sheet in the region of the window corners. That probably improves the
load transfer. To check this assumption, panels with window frames have been tested.
3. Experimental investigations
3.1.1 Test-set-up
In a first test series sandwich panels with a window at midspan have been tested in a six point
bending test. The width of the panels was 1 m, the window opening in each test 70 x 70 cm. Two
different thicknesses of panels were tested. Four panels had a thickness of 60 mm and a total length
of 4000 mm (test 1 to 4), two panels had a thickness of 120 mm and a total length of 6000 mm (test 5
and 6). In figure 1 the test set up is shown. Half of the panels were tested in a positive position
(equates to wind pressure), the other half in a negative position (equates to wind suction). The
important difference between these tests is the geometry of the window frame, which has some
influence on the failure criteria.
63
Figure 1: Experimental set-up for test series 1
The deflections were measured at 5 different locations. Three displacement transducers were fixed at
midspan (in the middle and left and right of the window frame), two at the corner of the window
frame. In some of the tests, additional strain gauges have been applied around the window corner to
get detailed information about the load transfer around the opening.
3.1.2 Results
Displacements
Load-displacement diagrams registered only very little differences of the displacement at all
measured points. One of the load-displacement diagrams is exemplary shown in figure 2. The
uniform displacement shows the stiffening effect of the window frame, which avoid different
deflections around midspan. The displacement was nearly linear until the failure of the panel.
3,5
load of the hydraulic cylinder in kN
2,5
1,5 midspan 1
1
midspan 2
w indow 1
0,5
w indow 2
0
0 5 10 15 20 25 30 35 40 45
deflection at m idspan ( x = 0 m m ) and in the
corner of the w indow ( x = 400 m m ) in m m
Figure 2: Load-displacement-diagram of one test at midspan and in the corners of the window
64
Comparing the displacements of the test panels with the calculated displacements of an unweakened
panel result in the following table:
All the panels failed by wrinkling in the corner region of the opening. Depending on the geometry of
the window frame the location of wrinkling was a bit different (see figure 3). The very stiff frame
leads to wrinkling directly at the frame corner, where the panel is still complete over the whole width.
In the case of the less stiff frame the face sheet wrinkles close to the corner of the opening in the
panel. For the same panel thicknesses, the load levels by failure are quite similar for each panel.
65
Table 2: Comparison of failure loads with and without opening
Test Thickness Failure load Calculated failure load Deviation
of the panel in kN of the complete in %
panel in kN
Allocation of stresses:
Strain gauges have been applied in the region around the window corner to get detailed information
about the load transfer around the opening. Exemplary one test-set- up is shown in figure 5.
0,00
-50,00
Stress in N/mm²
-100,00
1kN
3 kN
-150,00
5 kN
5,8 kN
-200,00
-250,00
-300,00
0 100 200 300 400 500
66
In figure 4 the allocation of stresses along the width of the panel directly in front of the window
frame is shown for different load levels. In the diagram, only half of the panel width is presented.
Two interesting things can be pointed out.
1) The openings with a size of 70 cm and the window frame of 80 cm lead to explicit stress peaks in
the region around the window corner. With increasing loads, the peaks become more definite.
2) In the middle of the panel the stress is increasing with higher loads, but has still a very low level.
That means, in spite of the high stiffness of the window frame, the main part of the load is transferred
through the covering sheet to the outer continuous part of the panel.
3.2.1 Test-set-up
In a second test series an interconnection system with 3 sandwich panels and a window in the middle
panel has been tested. The width of the panels was 1m, the window opening was a complete panel
width (1 m) and 1 m length. Like in test series 1, two different thicknesses of panels were tested in a
positive and a negative position. In figure 6 the test set up is shown.
3.2.2 Results
Displacements
Whole systems with windows lead to much higher deflections than sandwich panels without openings
or windows. For the tests with 60 mm thick panels a comparison of the calculated displacements
without window and the measured displacements with window is shown in figure 8. In some tests,
67
particularly the negative position, the window frame slipped out of the system. That explains the
different deflections between neighbouring points of measuring in figure 7.
0
0 0,1 0,2 0,3 0,4 0,5
-5
-10
-20 60_1
60_2
-25 60_3
-30 60_4
-35
-40
-45
Length / Total length
-10
w ithout w indow
Deflection in mm
-20 60_1
60_2
-30 60_3
60_4
-40
-50
-1500 -1000 -500 0 500 1000 1500
Distance from the longitudinal axis in m m
Figure 7: Deflections for the load level 4 kN, 60 mm panels and a span of 3900 mm
All the panels failed by an overstressing of the longitudinal joint and a following collapse of the
complete system including wrinkling of the outer panels. In some cases the window frames slipped
out of the joint. The slipping of the frames did not lead to an immediate collapse of the system, but
certainly had an influence on the failure load. As the slipping is a question of an exact fabrication of
68
the window it can be avoided. Probably according to the inexactnesses of the windows sizes, the
failure load levels were different for each system.
69
Allocation of stresses:
Strain gauges have been applied to get detailed information about the load transfer around the
window. Exemplary one test-set- up is shown in figure 8.
20
0
0 200 400 600 800 1000 1200
-20
Stresses in N/mm²
-40 60_1
60_2
-60
60_3
-80 60_4
-100
-120
-140
Distance from the longitudinal axis in m m
In figure 9 the allocation of stresses along the width of the panels directly in front of the window
frame is shown for the load level 4 kN. In the diagram, only half of the test set-up width is presented.
In each test, the characteristics of the stresses are similar. In the medial panel the stresses are nearly
zero, all the load is transferred to the outer panels. There, a concentration of stress can be identified
in the region close to the longitudinal joint.
70
4. Comparison with existing methods of calculation
As described in chapter 2, different calculation proposals exist for single sandwich panels with
openings. In table 4, the failure loads of the tests are compared to calculation results for openings
without frames according to calculation proposals of Böttcher (2006) and Courage/Toma (1998).
In IFBS (2006) a proposal exists for calculating a 3 panel system with an opening. According to this
proposal, there are three possible failure modes: Shear failure in the outer panel, wrinkling of the
outer panel and failure of the longitudinal joint. By recalculating the tests according to the IFBS
paper, one asserts that the longitudinal joint is always the critical failure mode. In table 5 the degree
of capacity utilization for the different failure modes is shown for a load level of 4 kN per panel.
For the bearing strength of the joint, in IFBS (2006) only a rough estimate is given. That could be the
reason for significant deviations between some of the test results and the recalculation, like shown in
table 6.
71
Table 6: Comparison between tests and calculations
5. Conclusions
Test results of sandwich elements with windows have been presented in this paper. Two different
kinds of openings were tested. The results of test series 1 show, that window frames have a positive
effect on the load bearing capacity of single sandwich elements with openings. The failure load can
be clearly increased in comparison to panels with frameless openings. Nevertheless, for larger
openings the loss of bearing capacity is very high. In the second series, the width of a complete panel
has been replaced by a window. The entire load was transferred to the outer panels. The maximum
loads in the second series varied within a wide range. The main reason for that is evidently the
slipping of the window frames out of the joints. An interesting point is the bearing capacity of the
longitudinal joints. Further tests will show, if a more exact knowledge about the bearing capacity of
the joint bring better correlations between the tests and the calculations. At the end of the project,
there shall exist a calculation model for sandwich elements with different kind of openings. This
model should be only dependent on parameters which can be derived from some basic tests.
References
Böttcher M. (2005) Wand-Sandwichelemente mit Öffnungen, Publication of the institute for steel
construction and material mechanics, TU Darmstadt
Courage, Toma (1998) Structural detailing of opening in sandwich panels, Publication of the
European Communities
72
Optimization of Geometry and Core Materials
of Sandwich Panels with Metallic Faces
Kurpiela, A.
Institute of Steel Construction and Material Mechanics, Technische Universität Darmstadt
(email: [email protected])
Lange, J.
Institute of Steel Construction and Material Mechanics, Technische Universität Darmstadt
(email: [email protected])
Berner, K.
Institute for Sandwich Technology, University of Applied Sciences Mainz
(email: [email protected])
Abstract
Due to their numerous advantages, sandwich panels are increasingly used. However, a scientifically
sound optimization of sandwich panels has not yet been done. Sandwich producers have made
selectively optimization works of their products but mainly with regard to the manufacturing
procedures. The aim of the optimization in the current project is an adjustment of the geometry of
metal faces together with the properties of the core material in such a way that the load bearing
capacity and the costs of sandwich panels are optimal. The production technology should be
considered as well. This topic is very extensive mainly due to the numerous parameters which
determine the load bearing capacity of sandwich panels.
The optimal geometry of the metal sheets and the optimal mechanical properties of the core material
have to be developed on the basis of theoretical investigations and checked in following mechanical
tests. In the first step the sets of achievable mechanical properties of the used materials are
determined. After this, a combination of the mechanical properties and the geometry of the sandwich
panels can be created in order to obtain the largest span. In this regard it is important to achieve a
high load bearing capacity of the panel and high utilization factors of all mechanical properties at the
same time. The conducted investigations demonstrate possibilities for the optimization of sandwich
panels. An achievement of larger spans without a significant increase of costs is possible only by
changing the core properties. Furthermore, a significant potential for the optimization of the metal
sheets was recognised. There is a possibility to diversify both the quality and the geometry of the
steel sheets. The numerous parameters which determine the properties of sandwich panels turn the
optimization into a complex procedure. Yet, on the other hand, they also provide numerous
alternatives in the optimization processes.
Keywords: optimization, sandwich panel, core material, metal face, load bearing capacity
73
1. Introduction
Sandwich panels used in civil engineering consist typically of a thick core with low density between
two thin high density faces. The materials can be configured in many possible combinations. For
example, for the cover layers thin metal faces, timber based plates or glass fibre reinforced plastics
can be used. The insulating core layer in most cases is made of structured foams like polyurethane
(PUR), polystyrene (PS) or of mineral wool (MW).
Sandwich panels are used as light-weight roofs and wall claddings in industrial and commercial
buildings. Usually the structures are loaded by permanent loads, like self-weight, snow and wind
loads. However, additional loads like temperature differences between external and internal metal
faces or creep of the core must be taken in account for statical calculations for sandwich panels.
The high load bearing capacity of sandwich panels is the result of a rigid connection between the
core material and the cover layers. The bending moment is distributed to the two faces (e.g. for
panels with flat faces in the form of axial forces) and the shear loads are borne by the core layer.
The optimization of sandwich panels in respect to the load bearing behaviour means to adjust the
mechanical properties of the core material and metal faces in such way that the largest possible spans
can be reached. Other important structural properties of sandwich panels (e.g. insulating properties)
or the manufacturing conditions should be considered at the same time.
The current optimization was done on sandwich panels with cover layers made of lightly profiled
steel sheets and a PUR core. In this case the thickness of the core can be between 40 and 300 mm.
The thicknesses of the steel faces vary between 0.4 and 1.0 mm.
Saving money and resources is very important in almost every industry sector. In this sense,
application-oriented research has the role to find the optimal exploitation of materials in the
manufacturing process. Compared with steel, concrete or timber constructions, sandwich panels are
relatively new components. The first applications of sandwich construction were implemented in the
seventies of the last century. That is the reason why sandwich constructions were previously not
investigated as deeply as the other construction types mentioned above. This is also the reason why
no investigations into the optimization of sandwich panels have been done before.
In this paper a possible method for the optimization of sandwich panels in respect to their load
bearing capacity is described.
74
There are two main assumptions concerning the optimization of sandwich panel load bearing. Firstly:
all mechanical properties of sandwich panels shall be adjusted in such way that the largest possible
spans can be reached. Second: all the mechanical properties shall be maximally exploited. These two
assumptions, together with the several statical systems and different possible load cases that possibly
occur, turn the optimization of sandwich panels into a complicated process. The several statical
systems which can be used, and the different loads that can occur, generate different failure modes as
being decisive for the determination of the span widths.
According to the sandwich theory, the calculation method for sandwich panels shapes the
background for the optimization calculations. In this respect the following main properties of
sandwich structures describe the load bearing behaviour. These are the crucial factors in the
calculation method:
Additionally the shear strength and shear modulus at high temperature and the shear strength for long
term behaviour can be expressed as a function of the adequate main values, which were measured at
room temperature.
For a determination of these properties, experimental tests on each panel type have to be done. The
exact relations between the several mechanical properties are not known. Furthermore, there are no
constitutive equations that describe how the achievable wrinkling stress of the metal face depends on
the properties of the used materials and the face geometry. It makes the determination of the load
bearing capacity only in an analytical way (without tests) to an impossible process. The achievable
wrinkling stress of sandwich panels depends on many parameters. These are not only the used
materials with defined properties. A very important point is the bond between the core layer and the
cover sheets, which can be influenced by many factors. For example the production procedures or the
adhesive that is used in the case of adhesively bonded panels can be crucial for the final properties of
each single panel type.
75
Because of numerous different factors, which impact the mechanical behaviour of sandwich panels
and additionally because of the lack of constitutive equations describing the behaviour, the existing
mathematical optimization methods cannot be rationally used at the moment. As an approach, in the
first step a parameter analysis according to the calculation method for sandwich panels shall be done.
The aim of the current project is the development of an optimization method that will consider both
the core material and the geometry of the faces in only one step. However, due to the mentioned
technical hitches, the optimization of sandwich panels will be split into the two steps of an
optimization of the core and an optimization of the metallic faces.
In this paper the investigation into optimization of one of the mostly used core materials –
polyurethane – will be shown. The optimization of the core material means to define the mechanical
values – the strengths and the modulus (like in 1.2) – which are needed for getting an optimal core
material.
In the first step it is possible to execute the optimization of core materials for existing panel
geometries. In this case any existing panel with a given geometry can be taken and the mechanical
properties of the core can be optimized in such a way that the given geometry and the new core
material bring an optimal panel regarding to the load bearing capacity.
The mechanical properties are included in certain sets. In this case the optimization of core material
means to search for combinations of mechanical properties from defined sets. The created
combinations should lead to a high load bearing capacity of the panel and maximal utilization factors
of all mechanical properties by limitation of the spans. First of all the sets of mechanical properties of
core material (PUR) were determined according to known material properties. The background for
this procedure is that the material combinations should be set up of material properties which are
possible to manufacture.
For instance the shear modulus depending on the density of the core material is shown in Figure 1a.
After all relevant mechanical properties were collocated in this way the boundaries for achievable
mechanical properties of the core were defined (see Figure 1b). Depending on the chosen density, the
possible scope of mechanical properties is known and the raster for creation of new combinations can
be defined. In this case a raster of five values was chosen.
76
8,0 8,0
6,0 6,0
5,0 5,0
4,0 4,0
3,0 3,0
2,0 2,0
1,0 1,0
0,0 0,0
35 40 45 50 35 40 45 50
Density PUR in kg/m³ Density PUR in kg/m³
a) b)
Figure 1a: Values of shear modulus of PUR depending on the core density; Figure 1b: Example for
taken value raster for creating the combinations of core properties, density of PUR 40 kg/m³;
The known scopes for all relevant mechanical values can be determined:
For creating the properties combinations for all values scopes a raster of five was chosen
Table 1: Creation of combinations of mechanical core properties for PUR, density 40 kg/m³
The combination of the main values (marked columns in Table 1) gives 625 compositions of
mechanical properties.
77
625 sandwich panels with the same face geometry and different core materials were generated. By
determination of the spans for all 625 fictive panels one of them can result as optimal. The optimal
panel should give the largest span and the mechanical properties of the core should be exploited
maximal.
Example:
Sandwich panel with given
geometry of faces and a PUR core;
Core density 40 kg/m³ q (wP /wS)
in kN/m²
Thickness of the panel D = 80 mm
External face, steel, micro lined profile, tF1 = 0,60 mm;
Interior face, steel, lined profile, tF2= 0,50 mm;
L
Statical system:
Vertical wall panel, one–span beam
Table 2: Mechanical properties of the core; basic panel and optimal combination
Load-span curve
6 6
5 5
fCv=0,189 N/m m ²
4 4
Load in kN/m2
basic element
3 fCv=0,157 N/m m ² 3 element with optimal core
for defined case
2 2
fCv=0,092 N/m m ²
1 1
0 0
0 2000 4000 6000 8000
Span length l v in mm
Figure 2: Load-span curve, example for basic panel with given core and a panel with optimal core
As a means to simplify the analysis of the calculations results, the achievable spans were shown in
dependency on the increasing loads, like in Davies (2001), see Figure 2.
78
The results of the investigations show that the increase of the span is possible even only due to an
improvement of the core properties. But in the most relevant cases the wrinkling stress or the
deflection limitation are decisive for the determination of the span of sandwich panel. This can be
observed in the area of low load (between 0,25 and 1,5 kN/m²).
The wrinkling stress depends on the core material, but even more on the properties of the metal faces.
That is the reason for a deeper investigation into optimization of the geometry of metallic faces in
respect to getting higher wrinkling stress.
3.1 Background
The optimization of metallic faces means to adjust the geometry and the quality of the metal face in
such a way that the achievable wrinkling stress of the face will be maximal. Many types and forms of
metal faces are produced today. There are flat faces and lightly profiled faces for wall elements and
strongly profiled faces for roof elements. This paper regards lightly profiled faces only. The
optimization of metal faces is shown on example of the ribbed profile.
bP1 bR bP2 bR
tF
hR
During the mechanical tests done on numerous types of sandwich panels for their certification it was
observed that the achievable wrinkling stress changes depending on different geometries. It can be
supposed that deeper ribs (hR in the Figure 3) cause higher wrinkling stress. But, the exact connection
between the rib depth and the achievable wrinkling stress is not known. Furthermore, which influence
have other geometry factors, e.g. the plate widths (bP) or the face thickness (tF) on the achievable
wrinkling stress? Beside the influence of the geometry the elastic bedding on the core through its
mechanical properties should be considered as well.
The influence of the geometry of metal sheets on the achievable wrinkling stress was done both in
theoretical investigations and experimental tests.
At the moment in optional calculation of wrinkling stress, the lightly profiled faces are regarded as
flat faces. In this way their explicit higher load capacity is not regarded. To achieve an optimization
of the geometry, a new analytical model for an assessment of the wrinkling stress of lightly profiled
79
faces depending on the geometry of the profiling was compiled. The model is based on the existing
theories by Plantema (1966) and Stamm/Witte (1974).
ax
x x
w0
x
z c=K· K(EC,GC, C)
Figure 4: Stresses and deformation on flat, constantly bedded sandwich panel structure
To the differential equation for flat plates a term for consideration of the bedding of the core was
added.
1 t 3F
Where BF 2
stiffness of the plate (2)
12 (1 F )
By consideration of the behaviour conditions, like deformation of flat plate on elastic bedding and the
failure mechanism, the differential equation can be reduced to the searched axial force in the plate:
2 c
Fx BF x 2
(4)
x
2 (1 C )
and K EC is the bedding constant (6)
(1 C ) (3 4 C )
80
For consideration of the shear modulus in the elastic bedding due to the core, the relation between
shear modulus and young modulus for isotropic material was used
EC 2 (1 C ) GC (7)
2 (1 ) 2 G C EC
K C
(6a)
(3 4 C) (1 C )
2 K
Fx BF x (8)
x
The lowest value for Fx is the buckling load of the flat plate Fki:
dFki
0 (9)
d x
3
Fki 3
2 K 2 BF (10)
2
In the case of a lightly profiled face, the failure mechanism is similar to a flat face. Nevertheless, the
ribs should be additionally considered as stiffeners. For this reason the lightly profiled face was
divided in the areas of stiffeners (ribs with effective widths of plates1) and flat plates between
stiffeners like in Figure 5. The lightly profiled face is considered as a configuration of beams and flat
strips on elastic bedding.
The buckling load for each beam with the partial cross sections like in Figure 5 can be taken like in
(10):
3
Fki,i 3
2 K 2 Bi (11)
2
1
Effective widths of plates were determined according to DIN 18807-1, 1987-06
81
bP1 bR bP2 bR ReH
tF hR
c=K· K(EC,GC, C)
beff1/2 beff1/2 beff2/2 beff2/2
Lv c
Figure 5: Lightly profiled face of sandwich panel; fragmentation of ribbed profile into beam on
elastic bedding with cross section Ai
By consideration of stressed cross-section area Ai the wrinkling stress of each single part of the face
equals:
Fki,i
w ,i (12)
Ai
3
w,i
3 2 K 2 Beff ,i (13)
2t D
The partial wrinkling stresses were spread in the complete cross-section area and the total wrinkling
stress of a lightly profiled ribbed face is:
w,i
Ai
w (14)
Ai
For investigation of the influence of the depth of ribs by lightly profiled faces on the bending moment
capacity full scale tests on simply supported panels were done. The tests were made according to EN
14509 (2006), A.5. Sandwich panels with steel faces and a core of PUR were tested. The specimens
were taken from one production batch. This provided the possibility to keep all parameters constant.
Only the depth of the ribs was varied.
b1 = b2 = 50 mm, b1 b2 bR bR =
hR
2,65 mm; tF =
0,60 mm
82
Table 3: Test results for specimens with varied depth of the ribs
Type of Nominal Tested face geometry Depth of the ribs Wrinkling stress Increasing of
panel thickness of wrinkling stress by
panel varied depth of the
ribs
Wall d = 80 mm 1.1. Ribbed profile hR = 0,63 mm 170,4 N/mm² -
According to the test results the achievable wrinkling stresses (average values) of the metal faces
were determined. The results show a clear connection between the depth of the ribs by lightly
profiled metal faces and the achievable wrinkling stress. There is the possibility to increase the
wrinkling stress considerably by insignificantly deepening the profiling. With regard to the
optimization processes is it important that the increasing of the wrinkling stress is possible even by
using the same mass of steel and core material.
The wrinkling stress for the tested panels was calculated according to (14). In Figure 7 the calculated
wrinkling stresses and the corresponding test results are presented. There is the varied parameter -
depth of the ribs - presented on the abscissa and the achievable wrinkling stress values are shown on
the ordinate.
300
250
Wrinkling strength in N/mm²
Calculation
200
Test
150
100
50
0
0,0 1,0 2,0 3,0 4,0
Depth of the rib hR in mm
Figure 7: Wrinkling stress depending on the depth of the profiling by ribbed profile
4. Conclusions
A method for optimization of sandwich panels with thin metallic faces in respect to their load bearing
behaviour was shown. There is the possibility to optimize both the core material and the geometry of
83
the metallic faces. The connection of those two methods will give a possibility for optimizing
sandwich panels in only one step.
It has been found that very often, for the relevant cases of loading, the wrinkling stress or deflection
limit is decisive for limitation of the span widths. These two values can not be significantly improved
by only changing the core material. For this reason the improvement of the wrinkling stress and the
reduction of the available deflection should be taken as important points for optimizing the cover
layers.
The changes on the face geometry can lead to a higher wrinkling stress. Additionally the selection of
optimal strength and stiffness values of the core material can increase the load bearing capacity.
The numerous factors which influence the mechanical behaviour of sandwich panels provide many
options for an improvement of the load bearing capacity. As a practical effect, the producers are able
to implement selective changes for getting better load bearing of their products.
References
Plantema F J (1966) Sandwich Construction, London, John Wiley and Sons, Inc.
EN 14509, Self-supporting double skin metal faced insulating panels - Factory made products -
Specifications; November 2006
84
International Council for Research and
Innovation in Building and Construction
CIB’s mission is to serve its members through encouraging and CIB Commissions (August 2010)
facilitating international cooperation and information exchange TG58 Clients and Construction Innovation
in building and construction research and innovation. CIB is en- TG59 People in Construction
gaged in the scientific, technical, economic and social domains TG62 Built Environment Complexity
related to building and construction, supporting improvements TG63 Disasters and the Built Environment
in the building process and the performance of the built envi- TG64 Leadership in Construction
ronment. TG65 Small Firms in Construction
TG66 Energy and the Built Environment
CIB Membership offers: TG67 Statutory Adjudication in Construction
• international networking between academia, R&D TG68 Construction Mediation
organisations and industry TG69 Green Buildings and the Law
• participation in local and international TG71 Research and Innovation Transfer
CIB conferences, symposia and seminars TG72 Public Private Partnership
• CIB special publications and conference TG73 R&D Programs in Construction
proceedings TG74 New Production and Business Models in Construction
• R&D collaboration TG75 Engineering Studies on Traditional Constructions
TG76 Recognising Innovation in Construction
Membership: CIB currently numbers over 400 members origi- TG77 Health and the Built Environment
nating in some 70 countries, with very different backgrounds: TG78 Informality and Emergence in Construction
major public or semi-public organisations, research institutes, TG79 Building Regulations and Control in the Face of Climate
universities and technical schools, documentation centres, Change
firms, contractors, etc. CIB members include most of the major TG80 Legal and Regulatory Aspects of BIM
national laboratories and leading universities around the world TG81 Global Construction Data
in building and construction. W014 Fire
W018 Timber Structures
Working Commissions and Task Groups: CIB Members W023 Wall Structures
participate in over 50 Working Commissions and Task Groups, W040 Heat and Moisture Transfer in Buildings
undertaking collaborative R&D activities organised around: W051 Acoustics
• construction materials and technologies W055 Construction Industry Economics
• indoor environment W056 Sandwich Panels
• design of buildings and of the built environment W062 Water Supply and Drainage
• organisation, management and economics W065 Organisation and Management of Construction
• legal and procurement practices W069 Housing Sociology
W070 Facilities Management and Maintenance
Networking: The CIB provides a platform for academia, R&D W077 Indoor Climate
organisations and industry to network together, as well as a W078 Information Technology for Construction
network to decision makers, government institution and other W080 Prediction of Service Life of Building Materials and
building and construction institutions and organisations. The Components
CIB network is respected for its thought-leadership, information W083 Roofing Materials and Systems
and knowledge. W084 Building Comfortable Environments for All
W086 Building Pathology
CIB has formal and informal relationships with, amongst oth- W089 Building Research and Education
ers: the United Nations Environmental Programme (UNEP); W092 Procurement Systems
the European Commission; the European Network of Building W096 Architectural Management
Research Institutes (ENBRI); the International Initiative for W098 Intelligent & Responsive Buildings
Sustainable Built Environment (iiSBE), the International Or- W099 Safety and Health on Construction Sites
ganization for Standardization (ISO); the International Labour W101 Spatial Planning and infrastructure Development
Organization (ILO), International Energy Agency (IEA); Inter- W102 Information and Knowledge Management in Building
national Associations of Civil Engineering, including ECCS, fib, W104 Open Building Implementation
IABSE, IASS and RILEM. W107 Construction in Developing Countries
W108 Climate Change and the Built Environment
Conferences, Symposia and Seminars: CIB conferences W110 Informal Settlements and Affordable Housing
and co-sponsored conferences cover a wide range of areas of W111 Usability of Workplaces
interest to its Members, and attract more than 5000 partici- W112 Culture in Construction
pants worldwide per year. W113 Law and Dispute Resolution
W114 Earthquake Engineering and Buildings
Leading conference series include: W115 Construction Materials Stewardship
• International Symposium on Water Supply and Drainage for W116 Smart and Sustainable Built Environments
Buildings (W062) W117 Performance Measurement in Construction
• Organisation and Management of Construction (W065)
• Durability of Building Materials and Components
(W080, RILEM & ISO)
• Quality and Safety on Construction Sites (W099)
• Construction in Developing Countries (W107)
• Sustainable Buildings regional and global triennial conference
series (CIB, iiSBE & UNEP)
• Revaluing Construction
• International Construction Client’s Forum
PAGE 1
International Council for Research and
Innovation in Building and Construction
Publications: The CIB produces a wide range of special CIB Annual Membership Fee 2010 – 2013
publications, conference proceedings, etc., most of which are Membership will be automatically renewed each calen-
available to CIB Members via the CIB home pages. The CIB dar year in January, unless cancelled in writing 3 months
network also provides access to the publications of its more before the year end
than 400 Members.
Fee Category 2010 2011 2012 2013
FM1 Fee level 11837 12015 12195 12378
FM2 Fee level 7892 8010 8131 8252
FM3 Fee level 2715 2756 2797 2839
AM1 Fee level 1364 1384 1405 1426
AM2 Fee level 1133 1246 1371 1426
IM Fee level 271 275 279 283
All amounts in EURO
PAGE 2
DISCLAIMER