Marine Structures 11 (1998) 185—204

Adhesively bonded steel corrugated core sandwich

construction for marine applications
E.M. Knox*, M.J. Cowling, I.E. Winkle
Glasgow Marine Technology Centre, University of Glasgow, James Watt Building, Glasgow G12 8QQ, UK

Received 31 July 1997; accepted 22 April 1998

Abstract
The aims of this research were to fabricate an adhesively bonded steel corrugated core
sandwich construction suitable for marine applications and to investigate the performance
characteristics and failure mechanisms in comparison with other methods of fabrication.
Construction of steel corrugated core sandwich structures for the marine industry are shown to
be viable using modern structural adhesives. These sandwich structures are also shown to be an
efficient form of panelling. Using both experimental techniques and numerical analyses, the
structural performance of adhesively bonded steel corrugated core sandwich beam elements
was investigated in bending transverse to the corrugations, under both static and fatigue
loading. The performance of these structures is shown to depend on the relative dimensions
of the core and face plate materials. In fatigue loading, an adhesively bonded corrugated sand-
wich demonstrated significantly improved performance over alternative fabrication methods.
 1998 Elsevier Science Ltd. All rights reserved.

Keywords: Adhesive bonding; Corrugated core; Sandwich structure; Fatigue

1. Introduction and background

The advent of modern structural adhesives has made many new design and material
combinations possible for both primary and secondary structural applications. In the
marine sector their full potential has not yet been fully realised, whereas in other areas
of industry, adhesive bonding technology is well established and more common place.
Despite the relatively low strength, low modulus and generally brittle behaviour of
most structural adhesives in their bulk state when compared with other structural
materials like steel, adhesives can offer a practical joining alternative. It is possible to

*Corresponding author. Tel.: 0141 330 6478; fax: 0141 330 4015; e-mail: [email protected].

0951—8339/98/$ - see front matter  1998 Elsevier Science Ltd. All rights reserved.
PII: S 09 5 1— 8 33 9 ( 98 ) 4 06 5 1- 9
186 E.M. Knox et al./Marine Structures 11 (1998) 185–204

design joints of high load-bearing capacity [1] that can be applied to a wide variety of
marine structures. The limiting strength of a bonded joint is largely a function of the
stress concentrations at the edges of the joint. Stiffened panels are fundamental
elements in many structures. Ships, in particular, are composed, primarily, of stiffened
panels that make up the hull, decks, bulkheads and superstructure. In fabricating
these panels a large number of man-hours may be involved in reworking those steel
components that have been distorted by the heat of conventional welding techniques.
In general, the problem stems from the relatively large size of double fillet weld beads
used in the attachment of stiffeners to thin plating (below 8 mm thick). Shipbuilders
would welcome a reduction in reworking of these parts which involve thousands of
man-hours. Thus, an early potential application emerged, as adhesives offered a prac-
tical solution to shrinkage and distortion problems [2]. In particular, the lack of
significant residual stresses and high stress concentrations offers real advantages
through the limitation of structural distortion and improved fatigue performance [3].
Traditional ship construction has used stiffened plate grillages to resist both global
axial bending and shear loading as well as local out-of-plane static and dynamic
loading. These grillages generally have lower stiffness in the transverse than in the
longitudinal direction and uneven stress distribution in bending. High stress concen-
trations are present that may give rise to relatively poor fatigue lives. Using solid steel
plates is structurally inefficient. For example, 50% of the weight of a solid plate
provides just 12% of the stiffness [4]. There are theoretical benefits from sandwich
construction applied to structural panels under axial compression or bending as
sandwich construction works on the principle of using materials only where they are
needed. Placing steel skins on the faces of a low density core enables maximum
advantage to be made of the high Young’s modulus of steel. Structural symmetry
assures more efficient use of the continuous plate members and allows for the
possibility of overall weight reduction and increased structural stiffness. The simplest
type of sandwich construction consists of two thin, stiff, strong face sheets separated
by a layer of low-density material that may be much less stiff and strong.
The core has several vital properties. To ensure that the face plates remain the
required distance apart, the core must be of adequate stiffness and strength in the
direction perpendicular to the face plates. The core must also have adequate shear
properties to ensure that in bending, the face plates do not slide over each other. This
also applies to the bond properties between the core and the face plates. If this does
not happen the face plates will act as two separate beams and the sandwich effect is
lost. To prevent local buckling under compressive loads the core must be stiff/strong
enough to keep the face plates flat. In the case of a core of discrete elements (e.g.
corrugations) face plate buckling in its own plane between the core elements may
occur. To avoid this problem the face plate must have sufficient inherent stiffness.
There are many areas of large marine structures in which structural performance
and weight are an important design issue. One of the fields in which sandwich
construction is thought to be the most relevant is in the fabrication of topside
components in offshore drilling and production rigs. These applications include decks,
bulkheads, helidecks and accommodation modules. The potential of hybrid sandwich
structures in the construction of safe havens consisting of fire and blast resistant
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 187

panelling also demands serious consideration. This is especially so following the Piper
Alpha disaster [5]. It is recognised that if new methods can be found to join materials
reliably, then stiff lightweight metallic sandwich structures could offer advantages in
many marine structures which have to resist, primarily, loads normal to the face
plates. The corrugated core sandwich concept shown in Fig. 1 has the potential to
offer a wide range of attractive design solutions to operational shipboard problems. In
addition to weight savings, these can include space savings, fire resistance, noise
control and improved thermal performance. Until recently, the use of steel sandwich
construction for marine applications has been dismissed because of the difficulties of
fabrication with thin cores. The major problem of joining thin sandwich assemblies is
that there is always one joining face that is only accessible from inside. Through-
thickness laser welding has been considered as a solution [6—8]. There are, however,
practical disadvantages in this form of assembly, not least the need for an easily
accessible, portable, high-powered laser. Other disadvantages include a significant
volume of crevices formed between adjacent but intimately connected elements that
can promote crevice corrosion as well as the relatively high proportion of brittle
heat-affected zone material in the finished welds. The highly localised nature of joints
may also produce high-stress concentrations and thus lead to poor fatigue life.
Alternatively, others [9—12] have considered the use of spot welding as a possible
fabrication technique. However, the fatigue performance of spot welds has been found
to be poor by some investigators [13] and require further study in the context of
corrugated panels that would be subjected to fatigue loading [11].
In order to increase confidence of engineers in sandwich structures of the type
shown in Fig. 1, increased understanding of the behaviour characteristics of the
sandwich element is required. Information is required for directions both transverse
and parallel to the run of the corrugations and all the data must be documented
in an easy to use design handbook. This will enable understanding of both the
advantages and limitations of using such a corrugated sandwich panel. Progress in the

Fig. 1. All steel adhesively bonded (Araldite 2007) corrugated core sandwich beam.
188 E.M. Knox et al./Marine Structures 11 (1998) 185–204

development of the corrugated cores has been hindered by the lack of understanding
of their behaviour under load. Very few papers have been published [14] that deal
with the bending and buckling of sandwich panels with cores that are rigid enough to
make a distinct contribution to the bending stiffness of the panel, yet flexible enough
to permit significant shear deformations. If a corrugated core panel has the corruga-
tions running in the x-direction then the corrugated core is often stiff enough to make
a distinct contribution to the flexural rigidity in the zx plane but not in the yz plane.
Several workers [12,15] have demonstrated that simple beam theory gives adequate
indication of performance in the longitudinal direction of the stiffening. Transverse to
the corrugations a notation can be used that permits the corrugated core sandwich to
be treated as the special case of a general orthotropic sandwich panel. This work was
developed by Libove and Batdorf [16] and further developed by Libove and Hubka
[17]. These theories were developed specifically for corrugated panels used in aero-
space applications where the face sheets are considered to be ‘thin’ or ‘very thin’. Thus
no allowance is made for the local bending stiffness of the face plates about their own
centroidal axis. It is considered that for marine applications, where minimum plate
allowances are in force, the face plates on many sandwich panels are ‘‘thick’’ and
therefore must be considered in the analysis. The solution, therefore, has been to use
finite element stress analysis to describe the detailed stress distributions in a sandwich
construction.

2. Parametric study

As a preliminary study to investigate the performance of a corrugated core struc-

ture, a parametric study was undertaken to investigate the performance of a cor-
rugated sandwich structure in the longitudinal direction, compared with traditional
stiffened plating and other forms of discrete cored structures. This study included the
effect that various geometric details have on corrugated core performance in sandwich
construction when the face plate thickness is constant. Fig. 2 shows the influence of
various parameters on the structural efficiency of the steel corrugated core. The
measure of structural efficiency used is the sectional modulus-to-area ratio, Z/A, as
the structure is considered to be made wholly of steel of unit length over a unit width
of one metre. The section modulus is measured as the second moment of area divided
by the distance from the neutral axis to the outer surface of the face plate, I/y, over
a unit width of 1 m. Fig. 2 shows that core depth and web angle have the greatest
influence on the efficiency of the sandwich construction, whereas the flange width and
core thickness are weaker variables in the performance of corrugated cores. Fig.
3 shows the change in modulus and the efficiency with core angle for a given core
depth, plate thickness and flange size. This diagram indicates that the efficiency of the
section does not significantly increase if the web angle is greater than 30°, neither is
there a significant change in modulus of the section. Thus, a web angle of approxim-
ately 30° will give an efficient structural performance. If the angle is significantly less
than 30°, although there is an increase in structural performance, there is a decrease in
efficiency of the material used, for a given modulus.
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 189

Fig. 2. Influence of various parameters on the longitudinal structural efficiency of a corrugated core
sandwich.

Fig. 4 compares the efficiency of the steel corrugated core sandwich structure to that
of a single skin conventional stiffener arrangement, over a range of stiffener spacing.
This diagram indicates the improved efficiency of the sandwich construction over
a conventional stiffener arrangement if a suitable core geometry is chosen. The
envelope for the conventional stiffening is produced for a 10 mm plate thickness and
OBP (bulb plate) stiffener with web depth range of 120 to 430 mm at spacings of 500 to
1000 mm. The envelope for the sandwich construction is based upon two 5 mm face
plates and a depth range from 60 to 200 mm. When choosing a sandwich structure
significant savings in depth are possible. For example, a sandwich structure with
a specified modulus/metre of 1100 cm and depth of 200 mm has an efficiency of
8.5 cm, while the equivalent single skin panel with a modulus/metre of 1177 cm with
a reduced efficiency of 7.017 cm will have a panel depth of 350 mm. The depth of the
traditional single skin structure is thus 175% that of the sandwich structure. Further,
comparisons with other sandwich configurations, using alternative steel structural
sections as the core, are shown in Fig. 5. I-beam and channel sections proved the most
efficient cores from the range of structural sections surveyed. This figure shows that
the optimum corrugated core sandwich will also be as efficient as the best alternative
cores.
190 E.M. Knox et al./Marine Structures 11 (1998) 185–204

Fig. 3. Effect of core web angle on the longitudinal modulus and structural efficiency for a given core depth,
face plate thickness and flange width.

Fig. 4. Comparison of the longitudinal structural efficiency of the corrugated sandwich structure to that of
a single skin conventional stiffener arrangement.
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 191

Fig. 5. Comparison of the longitudinal structural efficiency of the corrugated sandwich structure with
alternative discrete structural sections as the core.

3. Experimental details

Earlier research at Glasgow University established the feasibility and basic tech-
niques for bonding thick section steel panels for use in the offshore and marine sector
[2,18]. The procedures are simple and do not require ultra-clean laboratory condi-
tions and are as follows:
(1) The adherend surface is cleaned using shot blasting or abrasion techniques. The
debris from this process is removed by brushing or compressed air.
(2) The freshly prepared surface is then solvent degreased to remove any oils and
grease that contaminate the surface after machining and handling. A bonding
primer can be applied at this stage to increase durability performance in wet
environments.
(3) The adhesive is applied manually via a spatula. On a larger scale this may be done
using a mixer/dispenser machine.
(4) The prepared surfaces are brought together and clamped, jigged or positioned
using dead weights to maintain bondline thickness during cure. In the case of steel,
magnetic clamps may be used. Wire or shims may be used to control bondline
thickness accurately.
(5) For heat-curing adhesive systems the specimen is cured in an oven or using
heating element pads to supply local heat input. The specimen is cured according
to the adhesive manufacturer’s recommendations. Alternatively, two-part cold
curing adhesive systems are left to cure at ambient temperatures for the required
period. In this current study, a hot-cure one-part structural adhesive was used
192 E.M. Knox et al./Marine Structures 11 (1998) 185–204

(Araldite 2007 from Ciba-Geigy). This adhesive has a good track record and has
successfully been used on thick steel adherends previously [2,18]. The recommen-
ded cure schedule for this adhesive is 20 min at 180°C.
(6) After the recommended cure the specimen is allowed to cool in air before
declamping. Generally, the spew fillets (excessive adhesive squeezed out from the
adhesive during clamping) are left intact as they increase bond strength and
durability in a wet environment.
In this study four configurations of steel corrugated core steel sandwich structures
were examined:
Model A: A bonded half-scale equivalent to the laser-welded specimens tested at
Newcastle University [6]
Model B: A bonded specimen similar to that tested at DRA Dunfermline [8]
Model C: A bonded heavier gauge corrugated core that is currently used in industry
as a stiffened panel [19]
Model D: A single sample of a laser-welded construction [8]
Table 1 shows the experimental geometric dimensions and the terminology is
indicated in Fig. 6.
The performance of corrugated-core panels is affected by many design variables
and thus the mechanical response under load is complicated, particularly in the
direction transverse to the corrugations. In this study, the characteristics of the core
and the face plate were investigated and the effect of geometry changes such as web
angle were examined. In a realistic situation, such a corrugated panel may simply be
supported and subjected to combined bending and compression loads. This loading
configuration can be investigated in bending tests. Models A, B, C, and D were all

Table 1
Geometric dimensions of experimental corrugated beam elements

Three-point bend test units Model A Model B Model C Model D

Width (mm) 410 500 500 410

Length (mm) 100 100 100 100
Depth (mm) 60 75 86 80
Span between supports (mm) 350 350 330 370
Plate thickness (mm) 4 2.03 4 2.65
Core plate thickness (mm) 1.5 1.14 2.5 1.8
Flange width (mm) 5 5.25 40 12.5
Web length (mm) 53 74.4 90 77.6
Core height (mm) 52 69.9 78 75
Web angle (°) 18 20 30 25
Adhesive area (mm) 1000 1057 8000 x
Adhesive area/metre (mm) 16,400 20,019 32,000 x
Element width (mm) 50 52.8 250 130
Elements/beam — 8.2 9.47 2 3.15
Elements/metre — 20 18.94 4 7.69
Mass (kg) 3.83 3.015 4.58 2.725
Mass/metre width (kg/m) 9.34 6.03 9.16 6.65
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 193

Fig. 6. Corrugated core terminology diagram.

tested in static three-point bending in the direction transverse to the corrugations to

obtain the overall experimental stiffness value, failure mode and ultimate capacity.
These tests were carried out on an Instron 250 kN test frame at a constant cross-head
speed of 0.5 mm/min. In all cases, the support positions were at a corrugation crown.
A linear voltage displacement transducer was positioned below the mid-point of the
beam to record the central deflection for each specimen. The results are plotted in
Fig. 7 and the results analysed are summarised in Table 2.
All marine structures are subjected to varying loads under normal working condi-
tions. Cyclic loads arise as a result of wave action or from machinery or propeller
induced vibrations and may lead to fatigue failure. Underwater vehicles may experi-
ence a sufficient number of diving and resurfacing cycles to cause low cycle compres-
sive fatigue. Thus, the fatigue resistance of a marine structure is crucial. Limited results
were available for the fatigue performance of a laser welded construction of a cor-
rugated core sandwich beam in bending [6]. Model A is a half-scale equivalent of this
and, thus, it was possible to obtain a comparative fatigue performance. In both studies
the beams were loaded in bending transverse to the corrugations. The load was cycled
sinusoidally at 10 Hz in the range between 2 and 12 kN. The results of the two-test
series are compared in Fig. 8 on a S—N diagram.

4. Finite element analysis

Finite element analyses were undertaken to determine the detailed stress distribu-
tion in the Model A corrugated beam during the experiments. Owing to the symmetry
only half the beam was modelled. The analysis was undertaken using SESAM
assuming the linear elastic material properties given in Table 3. The model was made
of three-dimensional, 20 node brick elements and the adhesive is incorporated as
194 E.M. Knox et al./Marine Structures 11 (1998) 185–204

Fig. 7. Load/displacement curves for experimental sandwich beam elements tested in static three-point
bending transverse to the corrugations.

Table 2
Experimental and calculated corrugated beam element results

Model A B C D

Plate stiffness (mm) 6.28E#05 5.25E#05 1.35E#06 7.99E#05

Web stiffness (mm) 1.68E#04 3.45E#04 1.14E#05 5.76E#04
Longitudinal unit stiffness (mm) 6.66E#05 6.24E#05 2.23E#06 1.05E#06
Longitudinal stiffness/metre (mm) 1.33E#07 1.18E#07 8.90E#06 8.08E#06
Web buckling stiffness (mm) 28 12 130 49
Web critical buckling load (kN) 21 5 33 17
Transverse experimental rigidity
(Nmm) 3.91E#10 7.04E#09 1.57E#09 1.90E#09
Transverse experimental stiffness
(mm) 1.96E#05 3.52E#04 7.85E#03 9.50E#03
Failure mode Bond Web Face plate Face plate
Failure load (kN) 25 24.55 14.61* 12.5*
Failure displacement (mm) 1.853 3.35 8.28 9.74

*Test stopped due to large deformation.

a 0.2 mm thick layer and modelled as three layers of six elements. The axial surface
stress distribution across the surfaces of the lower and upper face plates of the
sandwich beam is shown in Fig. 9 for a nominal load. The detailed stress distribution
gives an indication of the complex behaviour of the sandwich element. The maximum
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 195

Fig. 8. Fatigue experimental results for sandwich beam elements tested in three-point bending transverse to
the corrugations.

Table 3
Material property data for input into finite element stress
analysis

Material Young’s Poisson’s ratio

modulus (N/mm)

Mild steel 210,000 0.3

Epoxy adhesive
Araldite 2007 5000 0.37

stress occurs at one bond distance from the central load position and the distribution
alternates from compression to tension across the width of the face plate. Maximum
stresses occur at the bond edges and there are high stress differentials across the
adhesive bond.

5. Discussion of results

The parametric study showed the structural benefits of using a corrugated core
sandwich construction with the corrugations oriented in the longitudinal direction.
196 E.M. Knox et al./Marine Structures 11 (1998) 185–204

Fig. 9. Finite element transverse in plane stress distribution across the face plate surfaces of a bonded steel
corrugated core sandwich beam.
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 197

The depth and web angle of the core were shown to be the most influential parameters
on structural efficiency of the corrugated core for a given thickness of core plate, if face
plate thickness remains constant. A web angle of 30° will produce a sandwich of good
structural efficiency and high stiffness. This study also shows that flange size and core
plate thickness are relatively weak variables in sandwich design for high efficiency.
Several other independent studies have determined the structural benefits of cor-
rugated core sandwich construction. An experimental study at Manchester University
[11] concluded that there are several apparent advantages of panels of this type,
including high strength-to-weight and high stiffness-to-weight ratios. In a computa-
tional optimisation study on three types of panelling (a conventional stiffened panel,
a corrugated panel and a corrugated sandwich panel) Sen et al. [20] found that, if
low-structural volume is important, corrugated sandwich structures should be
seriously considered. Corrugated panels offer attractive combinations of low cost and
low weight. This could have important implications for space limited ships such as
RoRo vessels, where plane surfaces can also be an advantage. Conventional stiffened
plating tends to be bulky and cost reduction is incompatible with weight saving. This
study has also shown that for a given stiffness, a corrugated sandwich approaches half
the depth of a conventional stiffener and plate arrangement. Wiernicki et al. [21] also
conclude that a lightweight metallic corrugated core sandwich construction has the
potential to become a more attractive alternative to a conventional plate stiffener steel
structure, when all performance characteristics are considered on a total system basis.
The corrugated core is highly directional and is generally stiff enough to make
a distinct contribution to the flexural rigidity in the longitudinal direction but is more
compliant in the transverse direction. Design tables and software for corrugated core
sandwich are commercially available from ESDU International Ltd [22]. A particular
case that ESDU sheets do not cover is bending in the direction transverse to the
corrugations [15], probably due to the complex nature of the analysis. This study has,
therefore, concentrated on investigating performance characteristics in the transverse
corrugation direction. Fig. 7 is a load-displacement plot for the four beams tested in
bending transverse to the corrugations demonstrating that overall elastic deformation
occurs in the transverse direction as if it were a continuous medium. The calculated
experimental rigidity values from these tests are shown in Table 2 and assume that the
beams initially deform as a global structure. This response was also found by other
researchers [23] who tested corrugated panels under dynamic and impact loading in
bending transverse to the corrugations. However, the core structure is a series of plate
elements and, as such, permits local instabilities in both core and face components
[23]. This is reflected in the detailed stress distribution found by finite element stress
analysis. The axial stress distribution across the surfaces of the face plates of the
sandwich beam resulting from loading within the elastic limit is shown in Fig. 9. It is
clear that the stress variation across the plates does not follow simple beam theory.
The maximum stress does not occur at the centre of the beam (excluding the local
effect due to the load point) but at one corrugation wavelength away from the centre.
The bending stresses on the surface of the face plates alternate from tension to
compression along the width of the structure. The axial stress is obviously being
disturbed by the alternating presence of the single and double thickness of the plate
198 E.M. Knox et al./Marine Structures 11 (1998) 185–204

due to the presence of the core. In a similar study involving a finite element stress
analysis of a corrugated sandwich panel, Tan et al. [12] found that the locations of the
troughs have an influence on the stress and strains transverse to the corrugations, but
very little influence on the stress patterns parallel to the corrugations. The positions of
maximum stress occur at the bond edges while across the bonds there are high stress
differentials. The transverse stress distribution across the lower and upper plate
surface demonstrates that each face plate section, between the crowns of the cor-
rugated core, acts as a separate entity within the global structure and is liable to local
buckling instabilities. Across the surface of the lower face plate the stress alternates
between compression and tension, unlike in-beam theory where the lower plate
should be wholly in tension. Despite the detailed variation in stress, the general trend
for stresses in the lower plate is predominantly tensile (at a higher tensile stress level
than those in the upper plate) and for the upper plate surfaces are compressive and
low tensile stresses as might be expected from the curvature developed in bending.
Thus, in general, the finite element stress analysis indicates that buckling of an element
in the upper face plate is more likely. Tan et al. [12] report a similar result from finite
element analysis on corrugated sandwich panels. They found that the strain contours
transverse to the corrugations showed a ‘‘zig-zag’’ pattern with the peaks lying in the
regions in contact with double plate thickness due to the core. Similarly, high
membrane stresses occur in these regions. They also suggest that local bending can
occur in the wide unsupported plate between the corrugation crowns.
Several modes of failure can occur in sandwich beams subjected to flexural loading,
depending on sandwich geometry and core and skin properties [24]. There are two
basic failure modes of a corrugated sandwich construction: (a) face failure, (b) core
failure.
Failure of the faces occurs when the initial transverse stress is insufficient to cause
failure of the core, i.e. where the corrugated core is strong due to small web angle, low
pitch, etc. If the integrity of the core is maintained, and the faces are maintained the
correct distance apart throughout bending, very large face plate stresses are generated.
Two important modes of failure, tensile or compressive, of the skin material can occur
at the beam surface where stresses in flexure are the highest [24], as was indicated by
the current experimental and numerical results. If the inner face plate is relatively thin,
local instabilities will cause buckling of the compressive face due to the axial stresses.
This can happen between the crowns of the corrugation. Tan et al. [12] found similar
evidence, experimentally and numerically, for local bending occurring in the wide
unsupported plates between stiffeners. If the faces are relatively thick, as in this
experimental study, and failure of the core has not occurred, overall global bending of
the beam element will occur beyond the yield stress of the plate, to cause plastic
permanent deformation of the beam. Thus, the type of face plate failure is governed by
the relative thickness of the face plate material. Generally, buckling of the face plate
will not be expected in marine-type structures as the face plate material must be of
a minimum thickness to resist impact damage such as dropped objects, debris in the
sea and severe corrosion/erosion. Neither, therefore, will a tensile face plate failure be
expected. None of these local failure modes were seen in the present study as the face
plate materials were all relatively thick. Models C and D both demonstrated a face
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 199

plate failure caused by plastic deformation of the whole structure, as shown in Fig. 10.
Table 2 shows that the calculated web stiffness of models C and D are the highest of
the four beams, indicating a relatively stiff strong core. In both cases, the thickness of
the face plate is such that local buckling of the inner face plate does not occur. In
summary, Models C and D have a relatively strong core, thus high face plate stresses
are developed. The face plate is thick enough to prevent a buckling failure or a tensile
failure and therefore, the overall sandwich beam becomes permanently deformed.
Core failure will occur if the core is weak relative to the structural performance of
the face plates. The failure of the core can also occur in two different modes. One mode
of collapse results in interfacial slip. When this occurs the bonds between the face plate
and the core fail. The other mode of collapse is caused by web buckling. As soon as
a critical stress is exceeded, the core buckles in one or more of its elements. This type of
failure was also reported by Clark [15] in an experimental and numerical analysis of
a corrugated core bonded aluminium structure intended for automobile applications.
These types of failure occur when the geometry of the core is relatively easy to crush,
e.g. large corrugation pitch and thin core plating. Model B demonstrates a core failure
by web buckling, as shown in Fig. 11. Model B has a weak core as indicated by the low
calculated buckling load of the web (see Table 2). Model A failed by interfacial slip
causing shear failure of the bonds, as shown in Fig. 12, enabling the face plates to work
as two separate beams so the sandwich effect is lost. The stiffness of this core is also
relatively low compared with the plate stiffness that was used.
The failure of the corrugated sandwich panel in bending transverse to the corruga-
tions is governed by a combination of face and core geometry. There is no single
predominant factor. The core and face plate failures are mutually exclusive since the
mechanism of face failure requires that the core integrity remains intact to develop
very high axial stresses in the faces [23]. This can only occur at moderate curvatures

Fig. 10. Model C demonstrating failure by permanent plastic deformation of the beam element after
loading transverse to the corrugations in bending.
200 E.M. Knox et al./Marine Structures 11 (1998) 185–204

Fig. 11. Model B demonstrating failure by web buckling of the core after loading transverse to the
corrugations in bending.

provided the skins remain a large distance from the neutral axial. The switch between
the two failure types is subtle and thus a detailed understanding of the core geometry
and localised deformation effects is necessary to enable design for a specific mode of
failure. Further evidence for this was also found in a numerical study by Clark [15]
who carried out a sample parametric study indicating how the properties of a panel
change if its weight is kept constant but the ratio of the skin gauge to corrugation
gauge is varied from 1 : 1 to 3 : 1. He found that for ratios 1 : 1, 1.5 : 1 and 2 : 1 the failure
was by local buckling in the skin. For the skins with ratio 2.5 and 3 times the thickness
of the corrugation core plate, the thin webs buckled first. For Clark’s particular case
the optimum method to distribute weight was to have face plates of twice the
thickness of the core plate thickness. He concluded that this is not a universal law, but
depends on the specific case. Furthermore, he concluded that by using finite element
analysis he could predict failure loads of this kind to within 20% but that design is still
largely a case of trial and error.
For marine applications the skin thickness should be kept relatively high to allow
for corrosion resistance and guard against accidental penetration damage from local
impact. This may limit the attainable weight efficiency of a sandwich panel. As Sen
et al. [20] reported, weight efficiency is critically dependent on the permissible
minimum plate thickness and this form of panel will only offer significant weight
saving if the minimum thickness can be reduced. Similarly, for automobile applica-
tions, optimum sandwich skin thicknesses are too thin [24] and hence are susceptible
to damage or corrosion. However, weight savings are still possible with sandwich
panels having thicker skins than the optimum but in such cases failure of the skins is
relatively unlikely and the geometry of the core will be the important design para-
meter.
Ideally, more experiments on thick section corrugated core sandwich panels should
be undertaken to verify optimum design characteristics of the detailed core geometry,
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 201

Fig. 12. Model C demonstrating shear bond failure allowing the face plate to slip after loading transverse to
the corrugations in bending.

bond and face plates. The manufacture of a continuous corrugated core of the
required dimensional tolerance was outside the scope of the University workshops
and many commercial fabricators. Thus experimental testing has been limited to the
types of core that were available.
Static three-point bend tests did not highlight any substantial benefits (nor disad-
vantages) in terms of static strength and stiffness from using adhesive bonding or laser
welding as a joining method, apart from the practical fabrication methods already
mentioned. It must be noted that the specimens tested were not specifically designed
to increase performance through using adhesive bonding i.e. the bond area could be
further optimised to maximise strength and stiffness of the structure. Furthermore,
due to the intimately connected adherends when using adhesive bonding, there is less
202 E.M. Knox et al./Marine Structures 11 (1998) 185–204

likelihood of crevice corrosion than in a laser welded construction. However, the true
benefits from adhesive bonding are shown under fatigue loading. The fatigue tests on
the Model A beam (see Fig. 8) show the improved fatigue performance of the
adhesively bonded beam relative to a laser welded beam. The calculated gradient of
the SN line through the limited experimental data points for the laser welded
construction is 2.7. Most experimental data for crack growth in structural steels gives
a gradient [25] (exponent) of 3. In metals, from fracture mechanics theory of fatigue, it
is known that, if the crack initiation phase is negligible, the slope parameter of the SN
curve is identical to the crack growth power law. This parameter depends mainly on
the material property and hence all SN curves for a given material should be
essentially parallel. For the experimental adhesively bonded model the calculated
gradient, using regression analysis, was 8.067. This value is high compared with
fatigue crack growth in steel. Other workers have found similar steep gradients for
bulk adhesives and bonded joints. Luckyram and Vardy [26] calculated, for a cold
cure adhesive, a Paris power law gradient of approximately 4. While the gradient
calculated for a bonded composite cracked lap shear joint was found to be in the
range of 4—4.5 [27], in the modified Paris law equation incorporating the strain
energy release rate parameter. The relatively steep slope shown in the experimental
data here, 8.067, means that small changes in applied loads and hence stress range
cause a large change in the fatigue life. Thus, the failure propagation rates in these
components are more sensitive to errors in design load than in typical crack growth
rate in metallic welded structures. Because of this sensitivity the design of bonded
components for a relatively short ‘finite life’ may be unacceptable with regard to
reliability. An effective ‘infinite life’ (say'10 cycles) may be a viable alternative. Then
the crack growth threshold value may be an important material property. Of course,
this relates to an adhesively bonded corrugated sandwich where fatigue failure occurs
within the adhesive bond. As demonstrated in the static tests, failure mode will depend
on the relative dimensions of the core and the face plates and this may also be the case
in fatigue. Thus the fatigue performance characteristics of various geometries of
corrugated sandwich also warrants further investigation to establish design protocols.

6. Conclusions

This study has shown that adhesively bonded steel corrugated core sandwich
structures are feasible for typical marine applications and compare favourably with
other fabrication methods available, especially in fatigue loading situations. They
have been shown to have structural benefits over conventional stiffened panelling but
are highly orthotropic and have a complex behaviour pattern transverse to the
corrugations. To maximise structural capabilities the corrugated sandwich panel
should be oriented such that the lines of corrugations are parallel to the maximum
load. Further evaluation is required to produce comprehensive design tables trans-
verse to the corrugations. The following specific conclusions are drawn:
(i) Heavy-grade steel sandwich structures using adhesives have been shown to be
possible using a single-part hot cure epoxy adhesive. This laboratory fabrication
 E.M. Knox et al./Marine Structures 11 (1998) 185–204 203

method could easily be scaled up for use in the offshore fabrication and
shipbuilding industries.
(ii) The required weight of material, i.e. the thickness of material, for most offshore
applications makes steel sandwich construction only feasible using through-
thickness laser welding, spot welding or adhesive bonding technology. Adhesive
bonding has distinct practical advantages over laser welding.
(iii) The fatigue performance of adhesively bonded construction in the low stress,
high-cycle fatigue regime is superior to equivalent laser welded construction.
(iv) Failure mechanisms and theory to predict corrugated core sandwich perfor-
mance transverse to the corrugations are complicated due to (a) the faces are
thick and no allowance is made for their local bending stiffness about their own
centroidal axes and (b) the core is stiff enough to contribute to the overall
stiffness of the sandwich construction.
(v) Detailed stress distributions in the corrugated construction may be predicted
using appropriate finite element analysis.
(vi) Many variables influence the performance of a corrugated sandwich panel. In
the direction parallel to the corrugation, the web angle and core depth are the
most influential.
(vii) Corrugated core sandwich construction is a structurally and volume efficient
form of stiffened panelling.
(viii) Failure mechanisms in the direction transverse to the corrugations are complex
and depend upon the relative dimensions of the core and face plate material.

Acknowledgements

This work was funded by the Science and Engineering Research Council (new
EPSRC) through the Marine Technology Directorate Ltd. and the Ministry of
Defence. This support is gratefully acknowledged.

References

[1] Knox EM. Marine applications for structural adhesives. PhD Thesis, Glasgow University, Glasgow,
UK, 1996.
[2] Hashim SA. Assessment of adhesive bonding for structural design with thick adherends. PhD Thesis,
Glasgow University, Glasgow, UK, 1992.
[3] Cowling MJ, Smith EM, Hashim SA, Winkle IE. Performance of adhesively bonded steel connections
for marine structures. Proc. Evalmat’89, JISI, Kobe, Japan, 1989;827—834.
[4] Turner AF. Composites and Laminates. British Steel Technical Information Sheet, Swindon, Moor-
gate, Rotheram, 1993.
[5] Tollouka JA. Potential sandwich structure applications offshore. Sandwich Structure Newsletter
1990;1(1):7.
[6] Hoare J. The application of sandwich structures to ship design. Phase 4 Summary Report, University
of Newcastle Upon Tyne, UK, 1988.
[7] Barnes TAL. Redesign of ships structures to facilitate auto methods of production. MSc Thesis,
University of Newcastle Upon Tyne, UK, 1985.
204 E.M. Knox et al./Marine Structures 11 (1998) 185–204

[8] Bird J. Private communication. DRA, Dumferline, Fife, UK, 1990.

[9] Norris C. Spot welded corrugated core steel sandwich panels subjected to lateral loading. PhD Thesis,
University of Manchester, Manchester, UK, 1987.
[10] Montague P, Norris C. Spot welded corrugated core steel sandwich panel subjected to lateral loading.
Steel and Aluminium Structures Int Conf Papers, Amsterdam: Elsevier, 1987;564—574.
[11] Norris C, Montague P, Tan KH. All steel structural panels to carry lateral load: experimental and
theoretical behaviour. Struct Eng 1989;67(9/2):167—189.
[12] Tan KH, Montague P, Norris C. Steel sandwich panels: finite element, closed solution and experi-
mental comparison on a 6 m;2.1 m panel. Struct Eng 1989;67(9/2):159—165.
[13] Chandel RS. Resistance spot welding of mild steel sheets. A literature review. Ind Welding J 1985;
17(2):27—32.
[14] Allen HG. Analysis and design of structural sandwich panels. London, Permagon Press, London,
1969.
[15] Clark JD. Predicting the properties of adhesively bonded corrugated core sandwich panels. Proc.
Adhesion ’87, York, UK. The Plastics and Rubber Inst., 1987, ppw1-w6.
[16] Libove C, Batdorf SB. A general small deflection theory for flat sandwich plates. NACA TN1526,
1948.
[17] Libove C, Hubka RE. Elastic constants for corrugated core sandwich plates. NACA TN2289, 1951.
[18] Winkle IE, Cowling MJ, Hashim, SA, Smith EM. What can adhesives offer shipbuilding? J Ship Prod
1991;7(3):137—152.
[19] Sowerby JG, Mechlad Mechtool Engineering Ltd, Mechtool House, Mains Works, Whessoe Road,
Darlington, Co. Durham, UK, 1989.
[20] Sen P, Shi BW, Caldwell JB. Efficient design of panel structures by general multiple criteria utility.
Engng Opt 1989;14:287—310.
[21] Wiernicki CJ, Franz Liem PE, Gregory PE, Woods D, Furio AJ. Structural analysis methods for
lightweight metallic corrugated core sandwich panels subjected to blast loading. Naval Engng J 1991;
192—303.
[22] ESDU, Data Sheets, Structures Subseries Vol 3, London: ESDU International Ltd, 1980.
[23] Rollins, MA and Xia YR. Dynamic failure of sandwich structures. Structures Under Shock and
Impact. Report, Cambridge University, Cambridge, UK, 1987.
[24] Johnson AF, Sims GD. Mechanical properties and design of sandwich materials. Composites 1986;
17(4):321—328.
[25] Lotsberg I, Andersson H. Fatigue handbook offshore steel structures. A. Almar-Naess (editor). Tapir,
Trondheim, ISBN 82 519 0662 8, 1985, chap 11, p 464 Table 11.1.
[26] Luckyram J, Vardy AE. Fatigue performance of two structural adhesives. J. Adhesion 1988;
26:273—291.
[27] Liechti K, Johnson WS, Dillard DA. Experimentally determined strength of adhesively bonded joints.
In: Mathews FL, editor. Joining of fibre reinforced plastics, Chap 4, Elsevier, London, ISBN 1 85166
019 4, 1987, p. 144.