Sie sind auf Seite 1von 5

Finite Element Analyses of

Composite-to-Steel Adhesive Joints


Yu-Ping Yang
George W. Ritter
David R. Speth
Edision Welding
Institute
Columbus, Ohio

Finite element
modeling can
assist in the
design of an
adhesivebonded joint
system for a
compositesteel interface
to meet the
functional
requirements
of structures
and joint
longevity.

Composite part

Core

omposite structures are used extensively in aircraft, space vehicles, marine, and automotive structures due to their light weight, high stiffness, and high ultimate strength. These
structures often are joined to metal structures and other composite structures using mechanical fasteners, which are expensive and use labor-intensive installation procedures, and the
joints require long-term maintenance. Therefore, new cost-effective processes must be developed
to meet the functional requirements of structures and joint longevity.
Edison Welding Institute (EWI) and several partners (Boeing, Applied Research Lab at Penn
State University, Northrop-Grumman Ship Systems, Bath Iron Works, and the Composites Materials Technology Center) demonstrated an adhesive bonding method to join composite-to-steel
for large ship structures. Adhesive bonding can provide load continuity and can be used on both
internal and external bulkhead type structures[1].
Logistics International, Abu Dhabi, Dubai, UAE, obtained two large frigate hulls (141M and
135M), originally belonging to the Dutch government, to convert the hulls into giga-yachts at the
Abu Dhabi MAR shipyard by (among other modifications) removing the steel deckhouse entirely
and replace it with a composite structure. EWI was engaged to help design and verify the composite-to-metal joint and to provide supporting data pertinent to obtaining the manufacturing process
certification of the ships from Det Norske Veritas (DNV), Baerum, Norway.
Experimental testing and finite element analyses (FEA) were conducted to qualify the adhesive bonding process for joining composite to steel. Experimental and FEA data developed at EWI
were submitted to and subsequently approved by DNV. Experimental results for the adhesive were
published in Ref. 2 together with FEA results. This paper reports the finite element model development for assisting the design of a bonded joint system for a composite-steel interface. The modeling method including the approach, input, validation, and application are discussed.
Composite-to-steel joint
An effective composite-to-steel adhesive joint normally incorporates a double lap shear joint
design, adhesive layers, composite skins, and cores. Many types of steel section were evaluated
for carrying the structural load. The clevis double-leg design (Fig. 1) is capable of transmitting
structural loads and maintaining the integrity of the
composite-steel joint[1].
Figure 1 represents a manufacturing approach to
make a composite-to-steel joint[1]. A paste adhesive can
be applied to the steel shoe, into which the composite
part is fitted. After making the composite-to-steel adhesive joint, the bonded steel receiver can be welded to the
deck. Weld points should match the steel legs to existing below-deck stiffeners. As an example, Fig. 2 shows
an adhesive bonded composite-to-steel structure.
Composite

Paste adhesive

Steel H-section
Fig. 1 Adhesive joining of composite to steel[1].
24

ADVANCED MATERIALS & PROCESSES JUNE 2011

Modeling method
The modeling method was developed based on
commercial finite element software, ABAQUS. A threedimensional (3-D) model was used in which the metal,
adhesive, and core were meshed with solid brick elements and the composite was meshed with both solid
brick elements and cohesive elements. The number of
layers and material property orientations in the composite can be considered using this modeling method.
Figure 3 shows the modeling approach used in the
finite element analysis of composite-to-steel adhesive
joint, which includes modeling input, validation, and application. The model input includes the joint geometry

Geometry
Model input

Tensile test
Material
properties

Composite

Model
validation

Steel

Double-lap shear test


Design

Fig. 2 Adhesive-bonded composite-to-steel structure.

Temperature, C
23

60

0.38

0.38

1.13E-06

1.13E-06

Tensile modulus, GPa

1.3

0.177

Failure plastic strain, %

18.0

56.5

Poissons ratio
3

Density, kg/mm

Fig. 3 Modeling approach.


35
30
Stress. MPa

Material properties
Typically, steel, composite, adhesive, and core are involved in a composite-to-steel joint. Isotropic elastic and
plastic material behavior was assumed for the steel, the adhesive, and the core. Orthotropic elastic material behavior
was assumed for the composite. Steel material properties
were obtained from public literature. Adhesive material
properties were obtained from testing, because they are
determined by the bonding process. The composite and
the core material properties were provided by material
suppliers.
Type 316 stainless steel was selected for hull structures
based on the corrosion requirements. Material properties
can be obtained from material handbooks such as ASM
specialty handbook[3]. In the analysis, elastic-plastic material properties were input into the finite element model.
The adhesive selected for the exterior joint is 3M 2216
translucent epoxy adhesive (3M Co., St. Paul, Minn.) to
which an accelerant was added to boost its cure rate and
temperature resistance. Tensile tests of adhesive were conducted from cast specimens configured as ASTM D638
Type I dogbones. Tensile properties were measured at
room temperature (23C) and at 60C. The higher temperature was selected based on the possible service conditions
of composite to steel structures at sea. The strain rate was
kept constant at 12.5 mm/min.
Poissons ratio, tensile elastic modulus, and plastic failure strain at room temperature and elevated temperature
for the epoxy adhesive are shown in the table below.

Modeling
application

25
20
Average

15
10

Rt-5
Rt-4

Rt-6

0
0

0.04

(a)

0.08
0.12
Strain

0.16

0.2

12
Average

10
Stress. MPa

and materials properties. Material properties were obtained from public literature and supporting tensile tests
within this program. Double-lap shear tests were used to
validate the finite element model, and the model was used
to predict the strength of complex composite-to-steel
structures.

8
6
4
T4

T2

T1

0
0
(b)

0.01

0.2

0.3
Strain

0.4

0.5

0.6

Fig. 4 Tensile material properties of 3M 2216 translucent


epoxy adhesive at room temperature (a) and 60C (b).

Figures 4a and 4b show the stress-strain curves for the


adhesive at room temperature and (60C), respectively.
Tensile tests show that the failures at the higher temperature were caused by stretching the adhesive to the material
limit (most deformation is plastic). Three duplicate tests
were conducted. The average material properties were
input to the models.
The core material was Diab Divinycell H200 (Diab Inc.,
DeSoto, Tex.). Elastic and plastic material properties taken
from the product literature were used for the analyses:
Poissons ratio is 0.32, tensile modulus is 0.23 GPa, yield
stress is 1.6 MPa, and tensile strength is 6.4 MPa.
Composite material properties were provided by the
material supplier. Table 1 shows tensile and shear elastic
ADVANCED MATERIALS & PROCESSES JUNE 2011

25

moduli and Poisson ratio. The composite is strong in plane


(x-y plane) and weak in the thickness direction (z-direction). The composite failure was modeled using the progressive damage and failure method. Failure parameters
were calibrated with experimental testing results and published in Ref. 2.
Modeling validation
FE model was validated by analyzing a double lap shear
(DLS) test sample, as shown in Fig. 5. The DLS specimen
configuration is in accordance with ASTM 3528, Type A.
A single large bonded plate about 300 mm wide was produced and individual 25-mm test specimens were cut from
Fix this end

Adhesive
15
2.
4
mm 25
.4
Steel (2.5 mm)
mm

15
2.4
mm

the plate. The adhesive cured for at least one week at room
temperature prior to testing.
Tensile tests were conducted by fixing the steel end and
applying the load at the composite end (Fig. 5). Load-displacement curves were used to validate the finite element
model.
A finite element model was built based on the DLS
configuration (Fig. 6). The steel and adhesive were meshed
with an 8-node brick element. There were 2040 nodes and
1200 elements for the steel and 640 nodes and 360 elements for the adhesive. The composite was first modeled
as skin only with cohesive interlayers (no core). The composite skin was meshed with 8-node brick elements, and
the composite cohesive interlayers were meshed with cohesive elements. There were 2550 nodes and 1000 elements
for the skin and 2520 nodes and 976 elements for the cohesive strength.
Finite element analysis (FEA) was performed to predict
the strength and failure mode for DLS testing at room temperature and 60C. The load capacity was predicted and
compared with the experimental results as shown in Table
3(2). Three replicates (E3, E4, and E10) were tested at room
Broken
Failure
initiation

Composite
(4.25 mm)

Apply load at this end


Fig. 5 Dimensions and loading methods of double lap shear test.

Skin
(a)
Cohesive
Failure Initiation

Fig. 6 Finite element model of a double-lap shear test sample.

Table 1 COMPOSITE MATERIAL PROPERTIES

26

Tensile modulus,
GPa

Shear modulus,
GPa

Poissons
ratio

17.50 (Ex)

6.90 (Gxy)

0.30 (Vxy)

17.50 (Ey)

6.90 (Gxz)

0.30 (Vxz)

3.0 (Ez)

6.90 (Gyz)

0.30 (Vyz)

ADVANCED MATERIALS & PROCESSES JUNE 2011

(b)

Fig. 7 Failure initiation during loading at room temperature


(a) and 60C (b).

Table 2 COMPARISON OF PEAK LOAD AND DISPLACEMENT BETWEEN EXPERIMENT


AND PREDICTION
Peak load, kN
Temp., C
23

Sample

Test

Avg.

E3

24.1

23.2

E4

21.3

3.5

E10

24.1

3.8

Prediction
60

Displacement at peak
load, mm

E13

Test

Avg.

Failure mode

Failure location

3.0

3.4

Interlaminar shear

Exterior composite skin

2.5

Tearing

Adhesive

23.9
8.7

3.3
9.8

2.4

E14

8.8

2.5

E15

11.8

2.5

Prediction

9.6

temperature and three replicates (E13, E14, and E15) at


high temperature, all at strain rate of 1.25 mm/min. The
average load capacity is 23.2 kN with a displacement 3.4
mm at room temperature and 9.8 kN with a displacement
2.5 mm at high temperature. The predicted load capacity is
23.9 kN with a displacement 3.3 mm at room temperature
and 9.6 kN with a displacement 2.6 mm at high temperature. The model predicted load capacity that correlated
well with the testing results.
In addition to the load capacity, the model can predict the joint failure modes and failure locations. As
shown in the Table 2, interlaminar-shear fracture was observed at room temperature[2]. The fracture location is the
exterior composite skin. Finite element analysis results
(Fig. 7a) shows that the crack started in the cohesive area
between plies and then induced skin failure. This corroborates the observed failure mechanism. The failure mode
at high temperature is cohesive tearing in the adhesive,
rather than failure in the composite skin. Finite element
analysis (Fig. 7b) shows that the crack started and propagated in the adhesive.
The comparison of the peak load, the displacement at

2.6

the peak load, the failure mode, and the failure location between the experiment and prediction indicates that the
model accurately predicts the load response. Therefore, the
model is ready to be used to predict the joint strength of
large composite-to-steel joint.

Modeling application
in optimizing adhesive thickness
The validated FE model was used to assist the composite-to-steel joint design and optimize the adhesive thickness. To understand the effect of adhesive thickness on the
joint strength, the adhesive thickness shown in Fig. 5 was
changed. The following cases were analyzed at high temperature:
2 mm thick at one side and 2 mm thick at another side
(2 mm-2 mm)
2 mm thick at one side and 4 mm thick at another side
(2 mm-4 mm)
4 mm thick at one side and 4 mm thick at another side
(4 mm-4 mm)
Figure 8 shows the plastic strain distributions and failure locations for the three analyzed cases after applying 3.5
mm displacement at the composite end. For the 2 mm-2
2 mm-2 mm
2 mm-4 mm
4 mm-4 mm
mm case, a crack started in
both sides of adhesive. For the
Two-side
One-side
2 mm-4 mm case, a crack
failure
failure
No failure
started on the 2 mm side and
there was no crack in the 4 mm
side. For the 4 mm-4 mm case,
no crack was observed in the
analysis. This shows that thick
adhesive bondline allows more
displacement.
Figure 9 shows the resulting force (load) after applying
a 3.5 mm displacement. The
highest peak load was obtained for the 2 mm-2 mm
case. The lowest peak load
Fig. 8 Effect of adhesive thickness on joint failures at 60C.
ADVANCED MATERIALS & PROCESSES JUNE 2011

27

was obtained for the 4 mm-4 mm case, for which no failure was predicted. This study shows that the thicker the
adhesive layer, the more flexible it becomes. As a result,
for the same load, the elongation goes up as the bondline
thickness increases.

12
2 mm-2 mm
2 mm-4 mm

10

4 mm-4 mm

Load, kN

8
6
4
2
0

0.5

1.0

1.5
2.0
2.5
Displacement, mm

3.0

3.5

Fig. 9 Predicted load-displacement curves at 60C.

Adhesive (3.5 mm)


Fix

Modeling application in predicting joint strength


The validated model was expanded to predict the joint
strength of a composite-to-steel joint shown in Fig. 10. The
design includes a steel section, adhesive, composite, and
now the core. The adhesive has the same thickness on both
sides. The end of the steel was welded to the hull, which
was simulated by fixing the end numerically during the
analysis. To save computation time, only one half of the design was analyzed. Symmetric boundary conditions were
applied on the plane of top surface.
Finite element analyses were conducted to predict the
load capacity of the structure by applying load at the composite end and fixing the steel end. Figure 11 shows the
predicted load capacity at room temperature and high temperature; the joint can carry a higher load at room temperature than at high temperature.

156 mm
Symmetric plane

76.2
Metal

Core

Composite (5 mm)

Load
Fig. 10 Complex design of composite-to-steel joint.

Reference
1. J. Simler and L. Brown, 21st Century Surface Combatants
Require Improved Composite-to-Steel Adhesive Bonds, AMPTIAC Quarterly, Vol 7, No. 3, p 21-25.
2. G.W. Ritter, D.R. Speth, and Y.P. Yang, Qualifications of Adhesive for Marine Composite-to-Steel Bonded Applications,
J. of Ship Production, Vol 25, No. 4, p 198-205, Nov. 2009.
3. J.R. Davis, Stainless Steels, ASM Specialty Handbook, 1994.

300

250

RT

Load, kN

200
60C
150

Acknowledgement: Johan Valentijn, CEO, Logistics International; Brian Climenhaga, BJC Design; Mark Bishop, MB Design; Frank Crane, J. Frank Crane Inc; and Jim Gardner,
Compmillenia. Visual imagery is available at www.abudhabimar.com.

100

50

0
0

0.5 1.0

1.5

2.0 2.5 3.0 3.5 4.0


Displacement, mm

Fig. 11 Load responses during tension.


28

Summary
A finite element analysis method was developed to predict the load carrying capacity of adhesively bonded composite-to-steel joints. In the procedure, the material
properties of steel, adhesive, and core were assumed to be
isotropic and the material properties for composite were
assumed to be orthotropic. Progressive damage and failure were modeled by defining failure criteria (damage initiation and evolution) of the adhesive and composite. The
failure parameters were obtained from experimental testing results. The analysis procedure was validated by analyzing a DLS sample and comparing the calculations with
experimental testing results. The validated model was applied to assist the composite-to steel-joint design and optimizing adhesive thickness.

ADVANCED MATERIALS & PROCESSES JUNE 2011

4.5 5.0

For more information: Dr. Yu-Ping Yang is a senior engineer


in modeling group (614/688-5253; email: yyang@ewi.org), Dr.
George W. Ritter is technology leader (614/688-5253; email:
gritter@ewi.org), and Dr. David R. Speth is senior engineer in
adhesive bonding (614/688-5253; email: dspeth@ewi.org),
Edison Welding Institute, 1250 Arthur E. Adams Dr., Columbus, OH 43221.

Das könnte Ihnen auch gefallen