Sie sind auf Seite 1von 13

Composite Structures 232 (2020) 111553

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Investigation on in-plane shear behavior of large-size composite plates with T


multi-bolt joints

Fa Zhanga,b, Zhendong Hub, Limin Gaob, Yumin Wanc, , Limin Jind, Xiwen Jiab, Ke Wanga,

Qian Maa,
a
Jiangsu R&D Center of the Ecological Textile Engineering & Technology, Yancheng Polytechnic College, Jiangsu 224005, China
b
Beijing Key Laboratory of Civil Aircraft Structures and Composite Materials, Beijing Aeronautical Science & Technology Research Institute of COMAC, Beijing 102211,
China
c
Association for Science and Technology, Beihang University, Beijing 100191, China
d
Shanghai Institute of Applied Physics, Chinese Academy of Sciences, Zhangjiang Campus, Shanghai 201204, China

A R T I C LE I N FO A B S T R A C T

Keywords: This paper aims to investigate the in-plane shear behavior of large-size composite plates with multi-bolt joints. A
Composite plate mechanical joint structure and an universal fixture were designed and in-plane shear tests were carried out. The
Large size distributed strain gauges were used to monitor the mechanical response and ultimate bearing capacity. The
Multi-bolt joints failure areas and failure modes of mechanical connections were studied by ultras-scanning. A non-linear finite
Shear
element model (FEM) based on 2D shell element mesh was developed to predict the load distribution and failure
In-plane
modes of bolt joints between composite and titanium alloys. The results show that the shear failure load reaches
365.95 kN. The shear failure occurs on the outside edges of composites plate, and the delamination extends to
the vicinity of the fixture bolts. The new FEM that requires very small computational cost can evaluate the
structure strength and predict destruction area. The deviation between the predicted shear failure responses and
the testing results is less than 10%. The failure mode and location are consistent with the testing results, which
verifies the validity of the finite element model. It suggests that this model is applicable on large scale structures
and suitable to use in conjunction with iterative schemes.

1. Introduction structures, such as wind turbine blades [4], building [5] and railway
[6]. The bolted joints, on the other hand, are not only good installation
In modern aircraft design, ultra-light and high strength structure has performance and load transfer efficiency, but also high safety and re-
always been the direction pursued by aircraft structural design en- liability because of easily to be maintained and repaired, which is ty-
gineers. Composite structures have become an important part of the pically favored by aviation designers [7].
primary structure of aircraft [1,2]. With the increasing utilization of Efficient bolted joints design is one of the key approaches for weight
these materials, aero-structures could be manufactured with more reduction of aeronautical composite structures. The mechanical con-
complicated and integrated structure with less fastener. While the joints nection can be divided into lap joint and butt joint according to the
are weak spots that limit the overall efficiency of the structure [3]. form of connection, or single shear and double shear according to the
Successful design of such joints requires high efficient, safe and reliable form of bearing. Initial investigations relating to structural bolted joints
connection mode. There are two kinds of connections normally used to of composites have been conducted in-depth [3,8–12]. The relatively
joint composite components together: adhesive joints and bolted joints. well-known support plan is the European Union’s BOJCAS (Bolted
The main advantages of the adhesive joints are simple bonding process, Joints in Composite Aircraft Structures) project in the early 00 s [8]. Its
less weight, low cost, and good shear load transfer. However, due to the overall objectives are developing reliable and user-friendly analysis-
difficult quality control and poor consistency of the bonding process, it based design methods, and understanding fundamentally the composite
could cause cracking and/or debonding of the adhesive joints, even bolted joint behavior. The need for such joint characterization is driven
structural suddenly collapse without being detected in advance over the by the impetus towards the development of predictive numerical tools
life span. So the adhesive joints are generally used for ground to reduce the extent of physical testing. To date, a large number of


Corresponding authors.
E-mail addresses: zhangfa@comac.cc (F. Zhang), wanyumin@buaa.edu.cn (Y. Wan).

https://doi.org/10.1016/j.compstruct.2019.111553
Received 18 August 2019; Received in revised form 7 October 2019; Accepted 9 October 2019
Available online 12 October 2019
0263-8223/ © 2019 Elsevier Ltd. All rights reserved.
F. Zhang, et al. Composite Structures 232 (2020) 111553

research works have been done experimentally on bearing, protruding- of test or simulation at low level based on one unit cell does not fully
head, countersunk, preloads of bolts and size effect morphologically consider the stability of the overall composite, thus fails to represent its
[13–18]. Determining the local stress field in a multi-fastener, compo- macroscopic mechanical behavior.
site joint is a complex task. The corresponding numerical models can Herein, the in-plane shear behavior of large-size composite plates
yield further insight into complex damage mechanisms of the joint, fastened with multi-bolt joints was investigated using a modified Two-
enabling better exploitation of these composites in aerospace struc- Rail shear test method. A new kind of large scale multi-bolted joints
tures. In many instance, previous modeling studies have mainly sought structure was designed. The composite plates were lapped with the ti-
to predict the effect of the load distribution [19,20], bolt-hole clearance tanium alloy as the strip plate. Six rows of 28 high-lock bolts are me-
[21,22], bearing stress [23,24], and friction [3,23] on the joint struc- chanically connected. The in-plane shear response of the structure was
tures. Meanwhile, these works have also attempted to predict the tested in a relatively uniform shear field. Distributed strain gauge
failure of the connected composite components using various criteria monitoring system was used to study the strain field distribution and
related to either the Yamada-Sun criterion [25] or the Tsai-Wu criterion ultimate bearing capacity of the large-scale joint structures. The failure
in conjunction with cohesive element [26] and XFEM [27]. The results modes were studied by combining ultrasonic A-scan and C-scan. A high-
showed a favorable accuracy that these models could be applied to efficiency 2D shell mesh finite element model incorporated with a user
guide bolt joint design, and a helpfully insight understood the load material subroutine (UMAT) in ABAQUS was developed to determine
transfer mechanism between the connected elements. The relevant re- the bolts load distribution and predict the failure mode of the multi-
search works on coupon-level mechanical connection have been rela- bolted connections.
tively mature in aircraft structural design [28], and some countries
have developed corresponding design manuals [29]. Whereas most of 2. Experimental program
the above studies focus on a small number of bolted connections of
composite plates, classified as the low level in the aircraft multi-scale 2.1. Composite panel
analysis process [30]. The arrangement of bolts is relatively regular.
The connection form is limited. Therefore, these methods are too The carbon fiber reinforced plastic (CFRP) panels were manu-
computationally expensive to be used in the optimization of large scale factured with T800/Epoxy prepreg with density of 1600 kg/m3, fiber
structures. volume fraction of 61%, in clean room environment using hand layup
In the context of virtual testing of aircraft structures, the term multi- and autoclave molding technique. The nominal thickness of one layer
scale analysis describes the process of sequentially coupling different after cured is 0.191 mm. The optimized laminate ratio of 0/ ± 45/90 is
analysis models at different scales and levels of fidelity [30]. Besides, 4/5/1, stacking sequence of [ ± 45/0/90/ ± 45/0/−45/0/0/ ± 45/0/
the composite has obvious size effect as well [31,32]. Because the 90/ ± 45/0/ ± 45/45/90/−45/0/45/0]s, totally 50 layers. The
present manufacturing technologies have been able to precise control of thickness of the composite plate is 9.55 ± 0.1 mm. In order to generate
hierarchical architectures down to the nano- and micro-scale [33] even the relatively uniform shear field and match the modified V-Notched
adding carbon nanotube (CNT) [34], the mechanical performance of rail shear test method, the size of plate is cut to 453*295*9.55 mm3, as
raw materials can meet the theoretical expectations. At basic level of shown in Fig. 1.
virtual testing of aircraft structures, the detailed knowledge of how the
material responds to stress, sufficient detail geometry parameters and 2.2. Bolted joints structure
very confidence modelling approach are required, which have been
proven in work [3,4,19–27,35]. Finally, going to higher level, the large The mechanical properties of lamina were tested in the room tem-
scale, the scholars have conducted further research on how to extend perature and dry environment according to ASTM D3039, D6641, and
the single bolted joint to the group bolted joint on the large-scale D7078 under different loads, respectively, as shown in Table 1. The
structure with thousands of bolts [7,30,36]. Although the test method maximum tensile and shear failure strain of the laminate is about
can accurately obtain the strain field distribution, failure strength and 14,000 με and 10,000 με, respectively. The bolted joints structure was
failure mode of large structural components under external load, the composed of two equal and symmetrical composite plates, titanium
cost of obtaining the key influencing parameters is too high. Usually, a alloy butt plate and 28 high-lock bolts in the form of single shear butt,
relatively coarse analysis is conducted to calculate how the load is as shown in Fig. 2. The titanium alloy bolts with diameter of 7.94 mm of
distributed among the fasteners, followed by a more detailed analysis of were assembled to six rows. The high-lock bolt and a single bolt con-
critical fasteners [3]. The bolts load distribution analysis method for nection diagram are presented in Fig. 3. The size of the butt plate is
theoretical analysis of large-scale structural connections was originally 217*215*5 mm3. The total length of the structure is 0.91 m. The stiff-
based on the stiffness analysis method [36]. In recent years, the re- ness and physical properties of lamina and titanium alloy are shown in
searchers have gradually focused on reasonably simplified numerical
simulation combined with representative experimental research, which
can effectively save costs, and get ideal results [30,37].
Generally, the shear strength of composite is usually much lower
than the tensile strength, which leads to the shear failure is prone to
occur at the stress concentration regions such as holes. It is inevitable
that the multiple loading holes are needed in the test fixture system for
the large scale load conditions. Another challenge in measuring the
shear properties of the large scale bolted joint structure is generating a
pure shear stress state without eliminating edge crushing, although
several publications have reported shear strength at material level
based on the V-Notched test method (ASTM standard D7078 or D5379)
[1,38] or the in-plane shear test method (ISO 14129) [39] that require
small and easily fabricated specimens. The size effect cannot be ig-
nored. There are few literature reports on the shear behavior of com-
posite with bolted joints at large scale. Only the studies [5,7,40–42]
reported the mechanical response of full-scale structures under bending
or cyclic fatigue load. These studies suggest that the multi-scale method Fig. 1. Schematic of connected composite panel.

2
F. Zhang, et al. Composite Structures 232 (2020) 111553

Table 1
Stiffness and physical properties of lamina and titanium alloy.
Material E11 (GPa) E22 (GPa) G12 (GPa) G23 (GPa) ν12 T (mm) ρ(kg/m3)

Lamina 163.50 9.00 4.14 3.08 0.319 0.191 1600


Ti 116.00 44.28 0.31 – 4500

Note: t—One layer thickness.

directly in the automatic data acquisition system.


The engineering shear strainγxy is obtained by the rosette strain
gauge, whose number is the combination of j ~ j + 2. No. j measured
the 0°direction strain along the length of specimen. No. j + 1 measured
the 45°direction strain. No. j + 2 measured the 90°direction strain. The
local shear strain measured by the rosette strain gauges was calculated
as follows:
γxy = 2ε45o − (εx + εy ) (1)
where γxy is shear strain. εx is 0° direction strain. εy is 90° direction
Fig. 2. Bolted joints structure. strain. ε45o is 45° direction strain.

Table 1, and the strength properties are given in Table 2. 2.4. Ultrasonic scanning

The inspection principle of ultrasonic scanning for composite in-


2.3. Shear test set-up ternal damage, A-scan and C-scan, are obtained with respect to the
pulses of ultrasound transmitted into and received from the specimen.
Base on Two-Rail shear test method [1,38], a special picture frame The ultra-A scan equipment used was USM GO with 1 MHz and dia-
type fixture was designed and fabricated in house for applying shear meter of 6 mm of contact vertical wave detector, where the transducer
loading on the bolted joint structure as shown in Fig. 4. Typically, it is a was fixed in position and one looks at the strength of the returning
loading frame with four symmetrical borders and two parallel tracks echoes as equivalently of penetration depth. The automatic immersion
bolted together using high strength steel bolts. The test specimen is C scanning imaging, UST 200, was employed with a frequency of
sandwich between the borders of thickness 15 mm with 22 bolts of 0.2–30 MHz and a detecting length range of 1*2 m, where the trans-
diameter 16 mm (M16). The performance parameters of the bolts are ducer was scanned in two dimensions above the specimen being in-
given in Table 3. The sliding tracks provide a single sliding degree of spected.
freedom, which ensures that the relatively uniform shear field is gen-
erated. A vertical displacement-controlled loading was applied to the 3. Finite element model
bottom end of the loading frame at a constant cross-head speed of
1 mm/min. The other end was fixed. The experiment was conducted A high-efficiency 2D shell mesh finite element model incorporated
using an MTS servo-hydraulic testing machine with a load capacity of with a user material subroutine (UMAT) in ABAQUS/Standard was
500kN. developed, based on the spring-shell element model, to determine the
In order to monitor the neutral, strain field distribution and critical bolts load distribution and predict the failure mode of the multi-bolted
failure load of the test pieces, a set of strain gauges of 120 Ω capacity connections. Both material and geometric nonlinearities were taken
were used to measure the development of strain in the test area on into account in the calculation.
composite and butt-cover plate. All of the strain gauges positions were
symmetrically pasted front and opposite sides. The number of strain 3.1. Mesh
gauge on opposite side is the number at the same position on the front
side plus 100. The position and serial number of strain gauges pasted The composite plates were defined as a kind of transverse isotropic
are shown in Fig. 5. The main unidirectional strain gauges were located shell. The butter-cover plate was treated as an isotropic shell. Both were
on the top/bottom sides of tested area to measure the tension and meshed using reduced integration shell elements (S4R in ABAQUS). The
compression strains, while the strain rosettes were placed at the center sizes of these element were set to be approximately uniform, about
to measure the shear strains. The total number of the strain gauges was 5 mm on the specimen and 15 mm on the fixture, as shown in Fig. 6.
54. Strain data were collected by JM5938 system provided by YANG- The bolts were simplified to be a combination of a linear spring element
JING corporation. The load and displacement data were recorded (CBUSH) with six elasticity modulus and several 1D rigid elements. The

Fig. 3. Titanium alloy bolt.

3
F. Zhang, et al. Composite Structures 232 (2020) 111553

Table 2
Strength of lamina and titanium alloy.
Material Xt (MPa) Xc (MPa) Yt (MPa) Yc (MPa) S12 (MPa) S23 (MPa) δ (%) Interlaminar strength (MPa)

Lamina 2920 1138 70 278 74 102 ≈1 62


Ti 931 848 931 848 545 545 15 –

combination of RBE2-BEAM-CBUSH elements was used to simulate the ⎡ (1 − df ) E1 (1 − df )(1 − dm) ν21 E1 0 ⎤
slide track mechanism, where the RBE2 was a rigid element to fix the 1⎢ ⎥
Cd = (1 − df )(1 − dm) ν12 E2 (1 − dm) E2 0
slide track, the BEAM element was the slide track, the CBUSH with only D⎢ ⎥
⎣ 0 0 (1 − ds ) GD ⎦
one zero stiffness simulated the slide. In addition, the interactions be-
tween structural parts were also created in ABAQUS using general (3)
contact algorithm choosing the friction coefficient of 0.28. A total of where G is shear modulus and D is macroscopic damage value, which
13,410 elements constitute this model. can be expressed as:

D = 1 − (1 − df )(1 − dm) ν12 ν21 (4)


3.2. Material constitutive
where, df ,dm and ds correspond to fiber, resin and shear damage state
respectively, and ν12 ν21 are Poisson’s ratio.
A suitable material constitutive used is fundamental importance to
Based on Hashin theory [43], which included fiber tensile fracture,
the validity of finite element prediction results. This section briefly
fiber compression buckling, matrix fracture under tensile and com-
introduces the constitutive and continuum damage models that are
pression, and shear failure, the criterion of lamina is defined.
used to describe the deformation and damage response of bolted joint
structure, consisting of the lamina and the titanium alloy.
3.2.2. Butter cover plate
The butter cover plate, titanium alloy, is assumed to undergo
3.2.1. Composite plate elastic-plastic deformation obeying the J2-isotropic hardening plasticity
A large number of researches have developed the constitutive theory. Its typical stress-strain curves tested by ASTM E8 and ASTM E9
models for laminated composites which include coupled damage and are given in Fig. 7. Damage initiation in the titanium alloy was based on
failure capabilities at global model scale. However, material model a ductile damage criterion in conjunction with shear damage criterion
available today that are based on physical composite failure modes and that allowed for the removal of element from the mesh [35].
fracture mechanics principles are too expensive for use at most model The ductile criterion is specified by providing the equivalent plastic
scales [30]. For the large scale structure, explicit modelling of fiber strain at the onset of damage, which is met when the following con-
fracture, matrix fracture, and shear damage are considered in this dition is satisfied:
model. The response formula of composite in UMAT is as follows: pl
ωD = ∫ ε¯d¯
ε
pl
(η)
=1
(5)
σ = Cd ε (2) D

where, df, dm is the stress, ds is the strain, and Cd is the stiffness matrix, εT+ sinh[k 0 (η− − η)] + εT− sinh[k 0 (η − η+)]
ε¯Dpl (η) =
which can be expressed as: sinh[k 0 (η− − η+)] (6)

Fig. 4. Schematic diagram of test fixture and test setup.

4
F. Zhang, et al. Composite Structures 232 (2020) 111553

Table 3
Performance parameters of bolt.
Brand Material Diameter (mm) Strength of single shear (N) Strength of tension (N)

CFBL1001-6 Titanium alloy 4.76 23,914 7112


CFBL1001-8 Titanium alloy 6.35 41,339 13,335
CFBL1001-10 Titanium alloy 7.94 64,897 22,225
CFBL1001-12 Titanium alloy 9.53 93,345 31,115
M16 Steel 16.00 129,645 74,238

where ωD is a state variable that increases monotonically with plastic εS+ sinh[f (θS − θS−)] + εS− sinh[f (θS+ − θS )]
ε¯Spl (θS ) =
deformation. ε̄Dpl is the equivalent plastic strain, which is a function of sinh[f (θS+ − θS−)] (8)
stress triaxiality η , where η = −p q, p is the pressure stress, andq is the
Mises equivalent stress. εT+ and εT− correspond to the equivalent plastic where ωS is a state variable that also increases monotonically with
strain at ductile damage initiation for uniaxial tensile and uniaxial plastic deformation proportional to the incremental change in equiva-
compressive deformation, respectively. η+ = 1 3 and η− = −1 3 are the lent plastic strain. ε̄Spl is a function of shear stress ratio,
stress triaxiality in uniaxial tensile and compressive deformation state. θS = (q + kS p) τmax , τmax is the maximum shear stress. εS+ and εS− corre-
k 0 is material parameter. Here, k 0 = 3.765 is recommended in this spond to the equivalent plastic strain at shear damage initiation for
paper. uniaxial tensile and compressive deformation, respectively. The para-
Similarly, the damage initiation of shear criterion is met when the meters θS+andθS−correspond to the values of θS at η = η+ and η = η−,
following condition is satisfied: respectively. kS and f are material parameters. For Titanium alloy,
kS = 0.12 and f = 1.5 were used in this paper.
pl
ωS = ∫ ε¯ pld¯ε(θ ) = 1 3.2.3. Bolt
S S (7)
The spring element, Cbush, was assigned corresponding Ti/Steel
material properties, whose elastic shear stiffness (K2, K3) K (K = 1 C )

Fig. 5. Position and number of strain gauge.

5
F. Zhang, et al. Composite Structures 232 (2020) 111553

Fig. 6. Schematic of the finite element model of shear test set-up.

where Aij represents the coefficient of in-plane stiffness matrix of la-


minate, ij = 11, 22, 12. t is the thickness of lamination.
The axial stiffness (K1) of the spring element is given as:

Ef A
K1 =
L (12)

where A is the cross-sectional area of bolt. L = t1 + t2, is the effective


length of bolt.
Based on experience, it recommends that the torsional stiffness (K4)
of bolt is given to be 100 N/mm. The bending stiffness (K5, K6) of bolt is
defined to be 108N/mm. The stiffness of CBUSH elements are sum-
marized in Table 4.

4. Results and discussions

4.1. Load vs. displacement


Fig. 7. Typical stress-strain curves of titanium alloy.
Take cost and other performance tests into consideration, only one
were calculated by Huth expression: of three specimens was loaded to catastrophic failure. The other two
were only loaded to 155 kN due to most composite structures are brittle
t + t2 a b ⎛ 1 1 1 1 ⎞ with a little bit of plasticity. Fig. 8. shows the load vs. displacement
C=⎛1 ⎞ · + ⎜ + + ⎟

⎝ 2d ⎠ n ⎝ t1 E1 nt2 E2 2t1 Ef 2nt2 Ef ⎠ (9) behavior of the multi-bolted joints structure under in-plane shear
loading. It can be seen from the figure that the three curves have good
where t1 and t2 denote the thickness of jointed plates, respectively, t1 is
consistency with a low discrete coefficient of 7.6%, and the load re-
the middle plate thickness when double-shear joints. d is the diameter
sponse increases linearly with the increase of displacement under
of the bolt. a and b represent the type of bolt joint, here, a is 2/3, b is
155 kN. The linearity of all three curves is greater than 0.99, as shown
4.2. n represents the form of shear, single shear n = 1, double shear
in Fig. 9. The discrete coefficient CV can be calculated by the following
n = 2, respectively. E1 and E2 are the equivalent elastic modulus of
formula:
jointed plates, respectively. Ef is the elastic modulus of bolts material.
Equivalent elastic modulus of jointed composite plate can be cal-
culated:
Table 4
2 Stiffness of CBUSH elements.
(A11 − A12 A22 )
E1 =
t (10) Brand Fastened_1 Fastened_2 K1 (N/mm) K2 (N/mm) K3 (N/mm)

2 CFBL1001-10 Butt-cover Composite 396,529 54,768 45,226


(A22 − A12 A11 )
E2 = M16 Fixture Composite 2,857,733 295,801 406,977
t (11)

6
F. Zhang, et al. Composite Structures 232 (2020) 111553

Fig. 10. Discreteness of strain gauge data monitored under the load of 155KN.
Fig. 8. Comparison of load vs. displacement curves of three specimens.
4.2. Strain analysis

4.2.1. Strain discreteness


Based on the three tests of 155 kN, the discreteness of strain gauge
data at corresponding location is shown in Fig. 10. It can be found that
the most of discrete coefficient is approximately within 10%, which
confirms that the good repeatability of strain gauge data is obtained.
Among them, the large discrete coefficient, which exceed 10%, mainly
located at the edge of the butt cover plate. This is because the area is
very close to the free edge of the bolted joints area and the strain value
is small, where is sensitive to installation errors.

4.2.2. Distribution of strain extremes


Fig. 11 shows the maximum tensile, compression and shear curves
of strain on the composite and butt cover plate, respectively. It ob-
viously presents that the maximum tensile and compression strains on
the composite plate are bigger greater than those of the butt cover plate
Fig. 9. Linearity of load vs. displacement curves.
due to the positions of strain gauge of 1 and 64 are further away from
the shear center. While the shear strain is almost equal under initial
1
N 175kN, which indicates the shear strain field is relatively harmonious
∑ (x i − μ)2
N
i=1
until nonlinearity appears. The other phenomenon is that the composite
CV = responses nearly linearly, and the butt cover plate of titanium alloy
μ (13)
shows a nonlinear response. This could attribute to the characteristic of
where x i is sample value, N is sample size, μ is mean. multi-bolted joints structure, especially, the contact nonlinear of bolt
And, the linear correlation coefficient R2 can be calculated by the hole extrusion, rather than material nonlinear. The strain extremes at
following equation: failure load are listed in Table 5.

n 2
⎛ ∑ (x − x¯)(y − y¯) ⎟⎞
⎜ i i
R2 = ⎝i = 1 ⎠
n n
∑ (x i − x¯)2 · ∑ (yi − y¯)2
i=1 i=1 (14)

where R2 islinear correlation coefficient, xi is x-value of the ith array, x̄


is abscissa mean of the entire array, yi is y-value of the ith array, ȳ is
ordinate mean of the entire array.
Go over 155 kN, there is a gradual appear of non-linear response
before the final failure occurring due to either the initiation of the ex-
truding damage around bolt hole or large geometrically nonlinear de-
formation. When the displacement meets 11.20 mm, the bearing ca-
pacity of the specimen reach 365.95 kN, a catastrophic brittle failure
occurs after a very little period of holding load developing. The holding
phenomenon should be a expansion process from microscopic damage
to macroscopic failure.

Fig. 11. Distribution of strain extremes on composite and butt cover plate.

7
F. Zhang, et al. Composite Structures 232 (2020) 111553

Table 5 model and experimental results. Firstly, there is a 0.23% error between
Strain extremes on composite and butt cover plate at failure load. the experimental fixture average displacement and numerical predic-
Component Tension (με) Compression (με) Shear (με) tion under the load state of 155 kN, see Fig. 14. Secondly, the strains at
the key points in different area predicted by the FEM are mostly close to
Composite 6536 −7882 9886 the experimental values within 10%, see Fig. 15. Nevertheless, the
Butt cover 3863 −3194 8093
different regions show different contrast varies. Similar to discreteness,
the strain comparison result between experimental and numerical is
also that the composite plate is better than the butt cover plate. And the
4.2.3. Strain field analysis
more close to center point on the butt cover plate, the higher contrast is
Pick up the strain curves at the four corners of the composite plate
obtained. Finally, Fig. 16 shows the comparison of displacement vs.
and the center line of the butt cover plate to illustrate the symmetry in
force curves between FEM and EXP-1 results. There is a good match
large size strain field, as shown in Fig. 12. It can be seen from Fig. 12(a)
observed until a displacement of approximately 5.5 mm as there is a
that a good diagonal symmetry is able to be get on either side of the
linear responds. Beyond that point, due to the FEM model does not
connected composite, and bolt hole extrusion leads a good linearity
consider the nonlinearity of bolt hole connection, it induces there is a
under tension while nonlinear under compression. However, the sym-
small deviation between the FEM and the experiment results in the the
metry on the front and back sides don't match well because local
nonlinear responds. However, in case of failure load, the predicted
bending happens. That could be verified by the linear eigenvalue ana-
value is 385 kN, approximately 5.8% bigger than the experimental
lysis, as shown in Fig. 13. Besides, the up-down asymmetry on butt
value (364 kN), which meets the requirements of engineering applica-
cover plate is good as well, see Fig. 12(b), and the more closer to the
tion.
center, the bigger the value of shear strain.

4.4. Failure mode


4.3. Verification of FEM
Fig. 17 shows shear failure mode of the multi-bolted joints structure,
From the discreteness and linearity analysis above, it can be con- when the shear load reaches 365.95 kN, the maximum shear strain on
cluded that the test data obtained from the three specimens has high the composite plate monitored is up to 9886 με. The outside edge of
reliability and good validity. Therefore, it is used to verify the feasi- composites plate occurs macroscopic shear failure while local com-
bility of the developed FEM by comparing the displacement, strain and pression failure with a bolt collapse. Its damage area is measured by the
failure load between FEM and experimental result, as shown in ultrasonic A-scan firstly, as shown in the white circles in Fig. 17. This
Figs. 14–16, respectively. It can be concluded that a generally accep- could directly prove there is a nonlinearity of bolt hole connection, and
table accuracy is shown for the results predicted by the constitutive progressive damage happens by comparing with the nondestructive

Fig. 12. Comparison of strain curves on composite plate and butt cover plate.

8
F. Zhang, et al. Composite Structures 232 (2020) 111553

Fig. 13. Model analysis of in-plane shear test.

Fig. 14. Comparison of global displacement between FEM and experimental results.

Fig. 15. Comparison of strain between FEM and experimental results.

9
F. Zhang, et al. Composite Structures 232 (2020) 111553

Fig. 16. Comparison of displacement vs. force curves between FEM and EXP-1 Fig. 18. Nondestructive results of EXP-3 after 155KN test.
results.
failure mode of composite plates is delamination, and the catastrophic
results of EXP-3 around bolt holes after the 155 kN test, as shown in failure is caused by the fiber fracture from the corner edge to the middle
Fig. 18. Additionally, Fig. 17 shows a good agreement is achieved be- bolt hole of fixture. The damage area is nearly 10,000 mm2.
tween the macroscopic failure morphology from the experimental and
numerical method.
4.6. Bolts load distribution

4.5. Damage analysis Extract the useful results from the FEM model To reveal the bolts
load distribution and verify the load transfer of the multi-bolted con-
In order to further accurately reveal the internal damage of com- nections. Fig. 20(a) shows the vector distribution of the bolts load and
posite plates after the EXP-01 shear test, ultrasonic C-scan inspection the maximum principal strain distribution of specimen at the state of
technology was used to determine the delamination size and image 155 kN load. It can be easily observed that the bolts load in the butt
internal damage, as shown in Fig. 19. Where the details of damage area is a semi-circular and anti-symmetric distribution with nearly
information are parsed into Table 6. It can be found that the main value, while the bolts load in the clamped area is a circular distribution

Fig. 17. Failure mode of multi-bolted joints structure.

10
F. Zhang, et al. Composite Structures 232 (2020) 111553

Fig. 19. Internal damage image of composite plate of 01# specimen.

Table 6 And the bolt shear failure occurs firstly at 6#–8# bolt grade, then it
Damage detection results of ultrasonic C-scan. changes to the composite plate fracture after 10# bolt grade. The in-
Number Damage area (mm2) Damage depth Damage type
crease of bolt grade has limited effect on the overall strength of the
structure when the composite plate fracture is the main failure mode.
D1 54.34 3/4H Delamination
D2 128.31 1/2H Delamination
D3 73.97 1/4H Delamination
5. Conclusions
D4 311.71 1/8–3/4H Delamination
D5 9967.91 1/8–5/8H Delamination + Fiber fracture
D6 221.14 5/8H Delamination In the present study, an experimental study was conducted to
D7 130.95 3/8H Delamination characterize the in-plane shear behavior of the large-size composite
D8 266.8 5/8H Delamination
plates fastened with multi-bolted joints using a modified Two-Rail shear
D9 47.17 1/2H Delamination
D10 141.14 1/2–5/8H Delamination
testing method. A new kind of multi-bolted joints structure was de-
signed. A shear strain field was generated in accordance with the the-
Note: H is the thickness of composite plate. oretical expectations by the newly designed fixture. Three specimens
were tested. The experimental displacement and strain data shows a
with very small value in the middle. The maximum shear force among good repeatability and symmetry, which confirms the rationality of test
these bolts appears at the first bolt in the first row on the left of the design for the large-scale joints structure. One of them was loaded to
clamped area. Comparing with the delamination direction images catastrophic failure in order to study the ultimate bearing capacity and
around the bolt holes detected by the ultrasonic method, as shown in failure modes. The load reaches 365.95 kN, and the maximum shear
Figs. 17 and 19, it well matches the bolts load vector distribution. On strain monitored on the composite plate is up to 9886 με. The outside
the other hand, Fig. 20(b) shows the maximum principal strain dis- edge of composites plate occurs macroscopic shear failure while local
tribution of specimen, which clearly gives the load transfer in the shear compression failure with a bolt collapse. The internal damage of the
field that a tension-compression cross filed is generated in the test composite plate and induced by the delamination around bolt hole
center area. Moreover, it indicates the bolts are the key channel of force extruded were revealed by the combination of ultrasonic A-scan and C-
flow. The force to the tested specimen comes from the bolts load on the scan technique.
clamped area, then propagates to the cross filed. The fiber fracture area In addition, a high-efficiency 2D shell mesh finite element model
is where the compression strain concentration is, which could be used incorporated with a user material subroutine (UMAT) in ABAQUS was
together with the strain gauge data (see Fig. 12) to explain the me- developed, which could give a relatively accurate prediction results of
chanism of the local compression damage initiation. shear response and failure mode of large size structures within minutes
level. The strain/displacement data, ultimate load and failure mode
4.7. Influence of geometric parameters obtained experimentally were used to validate the FEM model, and a
good agreement is achieved. Further, helpful results are extracted from
Similarly, taking advantage of the validated model to analysis the the model to reveal the bolts load distribution, verify load transfer of
effect of the thickness of butt cover plate and the diameter of bolt on the the multi-bolted connections, and analysis the effect of the different
in-plane shear property of the bolted structure. The comparison of geometry parameters on the inplane shear property of the bolted
displacement vs. force curves with two geometry parameters is pre- structure.
sented in Fig. 21. The bolt diameter and stiffness with different grades From all above, it suggests the experimental method is flexible to
are given in Tables 3 and 7, respectively. It could be found from evaluate the in-plane shear behavior of the large size specimen, the
Fig. 21(a) that the thicker the butt cover plate is, the greater the stiff- multi-bolted joints structure designed is potential to be used as a pri-
ness of the bolted joint structure is, which means the ability of com- mary aircraft structure, the FEM model is robustly applicable on the
posite plate to resist compression instability is improved. However, as mechanical analysis of large scale structures and suitable to use in
the thickness increases, the strength increment and plastic deformation conjunction with iterative schemes. Further work is focused on scaling
of bolted structure gradually decreases. The failure mode remains un- up to a subcomponent design of aircraft scale and verifying larger
changed in the range of thickness variation of butt cover plate while it multi-bolted joints areas through more complicated structural char-
changes from bolts failure to composite plate fracture in the range of acterization.
bolt grades. It could be found from Fig. 21(b) that the smaller the bolt
cross-sectional area is, the lower the in-plane stiffness and strength are.

11
F. Zhang, et al. Composite Structures 232 (2020) 111553

Fig. 20. Bolts load and maximum principal strain distribution.

Fig. 21. Comparison of displacement vs. force curves with different geometry parameters.

Declaration of Competing Interest Acknowledgements

The authors declare that they have no known competing financial The authors acknowledge the financial supports from the Special
interests or personal relationships that could have appeared to influ- Project of Civil Aircraft of Ministry of Industry and Information
ence the work reported in this paper. Technology of China (Grant Number MJ-2017-F-20), and the Open
Foundation of Jiangsu Provincial Research platform (Grant Numbers
YGKF-201809 and YGKF-201711).

12
F. Zhang, et al. Composite Structures 232 (2020) 111553

Table 7
Stiffness of CBUSH elements of different bolts.
Brand Fastened_1 Fastened_2 K1 (N/mm) K2 (N/mm) K3 (N/mm)

CFBL1001-6# Butt-cover Composite 142,511 38,939 32,154


CFBL1001-8# Butt-cover Composite 253,619 47,188 38,966
CFBL1001-10# Butt-cover Composite 396,529 54,768 45,226
CFBL1001-12# Butt-cover Composite 571,242 61,855 51,678

This work is also supported by the High Education Science [20] McCarthy MA, Gray PJ. An analytical model for the prediction of load distribution
Foundation of Jiangsu Province (Grant Number 18KJB540005). And in highly torqued multi-bolt compsite joints. Comp Struct 2011;93(2):287–98.
[21] Zhou Yinhua, Yazdani-Nezhad Hamed, McCarthy MA, et al. A study of intra-laminar
the financial from Qinglan Project of the Jiangsu Higher Education damage in double-lap, multi-bolt, composite joints with variable clearance using
Institutions of China (Jiangsu Teacher, 2018 NO.12 and 2019 NO. 3). continuum damage mechanics. Composites B 2014;116(1):441–52.
[22] McCarthy MA, McCarthy CT, Padhi GS. A simple method for determining the effects
of bolt-hole clearance on load distribution in single-column multi-bolt composite
Appendix A. Supplementary data joints. Compos Struct 2006;73(1):78–87.
[23] McCarthy MA, McCarthy CT, Lawlor VP, Stanley WF. Three-dimensional finite
Supplementary data to this article can be found online at https:// element analysis of single-bolt, single-lap composite bolted joints: Part I – model
development and validation. Compos Struct 2005;71(2):140–58.
doi.org/10.1016/j.compstruct.2019.111553. [24] Feo L, Marra G, Mosallam AS. Stress analysis of multi-bolted joints for FRP pul-
truded composite structures. Compos Struct 2012;94(12):3769–80.
References [25] Sun CT, Yamada SE. Strength distribution of a unidirectional fiber composite. J
Comp Mater 1978;12(2):169–76.
[26] Frizzell RM, McCarthy CT, McCarthy MA. Simulating damage and delamination in
[1] Tan Wei, Falzon Brian G. Modelling the nonlinear behaviour and fracture process of fibre metal laminate joints using a three-dimensional damage model with cohesive
AS4 PEKK thermoplastic composite under shear loading. Compos Sci Technol elements and damage regularisation. Comp Sci Technol 2011;71(9):1225–35.
2016;126:60–77. [27] Wang Zhenqing, Zhou Song, Zhang Jifeng. Progressive failure analysis of bolted
[2] Yumin Wan, Fa Zhang, Changxi Liu, et al. Overall buckling of typical thin-wall single-lap composite joint based on extended finite element method. Mater Des
sandwich composites applied on the aircraft. Acta Mater Compos Sinica 2012;37:582–8.
2018;35(8):2235–45. [28] Hart-smith LJ. Design methodology for bonded-bolted composite joints ADA117342
[3] Ekh Johan, Schon Joakim, Zenkert Dan. Simple and efficient prediction of bearing Long beach, California: Douglas Aircraft Company, McDonnell Douglas
failure in single shear, composite lap joints. Compos Struct 2013;105:35–44. Corporation; 1982.
[4] Murray Robynne E, Roadman Jason, Beach Ryan. Fusion joining of thermoplastic [29] Elsenmann JR. Bolted joint static strength model for composite materials. NASA
composite wind turbine blades Lap-shear bond characterization. Renewable Energy Report TMK 3377, 1976: 563–602.
2019;140:501–12. [30] Ostergaard MG, Ibbotson AR, Roux OL, et al. Virtual testing of aircraft structures.
[5] Guo-Feng Du, Bie Xue-Meng, Li Zhao, Guan Wen-Qiang. Study on constitutive CEAS Aeron J 2011;1:83.
model of shear performance in panel zone of Connections composed of CFSSTCs and [31] Xu X, Wisnom MR, Li XQ, et al. A Numerical investigation into size effects in centre-
steel-concrete composite beams with external diaphragms. Eng Struct notched quasi-istropic carbon/epoxy laminates. Compos Sci Technol
2018;155:178–91. 2015;111:32–9.
[6] Awad ZK, Aravinthan T, Zhuge Y, Gonzalez F. A review of optimization techniques [32] Liu T, Sun BZ, Gu BH. Size effects on compressive behaviors of three-dimensional
used in the design of fibre composite structures for civil engineering applications. braided composites under high strain rates. J Compos Mater
Mater Des 2012;33:534–78. 2018;52(28):3895–908.
[7] Fa Zhang, Wei Zhang, Zhendong Hu, et al. Experimental and numerical analysis of [33] Demir Baris, Henderson Luke C, Walsh Tiffany R. Design rules for enhanced in-
the mechanical behaviors of large scale composite c-beams fastened with multi-bolt terfacial shear response in functionalized carbon fiber epoxy composites. Appl
joints under four-point bending load. Compos B 2019;164:168–78. Mater Interfaces 2017;9:11846–57.
[8] McCarthy Michael. BOJCAS: bolted joints in composite aircraft structures. Air Space [34] Gurkan Idris, Cebeci Hulya. An approach to identify complex CNT reinforcement
Europe 2001;3(3–4):139–42. effect on the interlaminar shear strength of prepreg composites by Taguchi method.
[9] Zhao Xu, Jifeng Libin. Analysis method of advanced composite material connection Compos Struct 2016;141:172–8.
structure. Beijing University of Aeronautics and Astronautics Press; 2015. [35] Zhang Fa, Wan Yumin, Bohong Gu, Sun Baozhong. Impact compressive behavior
[10] Gray PJ, Mc Carthy CT. A global bolted joint model for finite element analysis of and failure modes of four-step three-dimensional braided composites based meso-
load distributions in multi-bolt composite joints. Composites B 2010;41(4):317–25. structure model. Int J Damage Mech 2015;24(6):636–60.
[11] Egan B, Mc Carthy CT, Mc Carthy MA, et al. Static and high-rate loading of single [36] William F McCombs, James C McQueen, Jeffrey L Perry. Analytical design methods
and multi-bolt carbon-epoxy aircraft fuselage joints. Compos A Appl Sci Manuf for aircraft structural joints. AD831711; 1971.
2013;53:97–108. [37] Liu Fengrui, Xuheng Lu, Zhao Libin, et al. Investigation of bolt load redistribution
[12] Sharos PA, Egan B, Mc Carthy CT. An analytical model for strength prediction in and its effect on failure prediction in double-lap, multi-bolt composite joints.
multi-bolt composite joints at various loading rates. Compos Struct Compos Struct 2018;202:397–405.
2014;116:300–10. [38] Zhang Yi, Zhang Litong, Yin Xiaowei, et al. Effects of porosity on in-plane and
[13] Warren Kyle C, Roberto A, Lopez-Anido, Goering Jonathan. Behavior of three-di- interlaminar shear strengths of two-dimensional carbon fiber reinforced silicon
mensional woven carbon composites in single-bolt bearing. Compos Struct carbide composites. Mater Des 2016;98:120–7.
2015;127:175–84. [39] Kolanu Naresh Reddy, Raju Gangadharan, Ramji M. Damage assessment studies in
[14] Starikov R, Schon J. Quasi-static behavior of composite joints with protruding-head CFRP composite laminate with cut-out subjected to in-plane shear loading.
bolts. Compos Struct 2001;51:411–25. Composites B 2019;166:257–71.
[15] Starikov R, Schon J. Quasi-static behavior of composite joints with countersunk [40] Jensen FM, Falzon BG, Ankerse J, et al. Structural testing and numerical simulation
composite and metal fasteners. Composites B 2001;32:401–9. of a 34m composite wind turbine blade. Comp Struct 2006;76(1–2):52–61.
[16] Taheri-Behrooz F, Shamaei Kashani AR, Hefzabad RN. Effects of material non lin- [41] Yang J, Peng C, Xiao J, et al. Structural investigation of composite wind turbine
earity on load distribution in multi-bolt composite joints. Compos Struct blade considering structural collapse in full-scale static tests. Comp Struct
2015;125:195–201. 2013;97:15–29.
[17] Xiao Y, Ishikawa T. Bearing strength and failure behavior of bolted composite joints [42] Ataei Abdolreza, Bradford Mark A, Valipour Hamid R, Liu Xinpei. Experimental
(prat I: experimental investigation). Comp Sci Technol 2005;65(7–8):1022–31. study of sustainable high strength steel flush end plate beam-to-column composite
[18] Mandal Bibekananda, Chakrabarti Anupam. Numerical failure assessment of multi- joints with deconstructable bolted shear connectors. Eng Struct 2016;123:124–40.
bolt FRP composite joints with varying sizes and preloads of bolts. Compos Struct [43] Zhang Fa, Mohmmed Ramadan, Sun Baozhong, Bohong Gu. Damage behaviors of
2018;187:169–78. foam sandwiched composite materials under quasi-static three-point bending. Appl
[19] Gray PJ, McCarthy CT. A global bolted joint model for finite element analysis of Compos Mater 2013;20(6):1231–46.
load distributions in multi-bolt composite joints. Composites B 2010;41(4):317–25.

13

Das könnte Ihnen auch gefallen