AIAA 98-2062

Analytical and Finite Element Modeling of

Riveted Lap Joints in Aircraft Structure
Y. Xiong and O. K. Bedair
Structures, Materials and Propulsion Laboratory
Institute for Aerospace Research
National Research Council Canada
Ottawa, CANADA
39
th
AIAA/ASME/ASCE/AHS/ASC Structures,
Structural Dynamics and Materials Conference
April 20-23, 1998 / Long Beach, CA
1
AIAA-98-2062
ANALYTICAL AND FINITE ELEMENT MODELING
OF RIVETED LAP JOINTS IN AIRCRAFT STRUCTURE
Y. Xiong* and O. K. Bedair**
Structures, Materials and Propulsion Laboratory
Institute for Aerospace Research
National Research Council Canada
Ottawa, CANADA K1A 0R6
ABSTRACT

Modeling procedures for the stress analysis of
riveted lap joints in aircraft structure are proposed in
this paper using both analytical and numerical methods.
In the analytical method, a complex variational
approach is employed to determine the stresses in
joined plates containing single or multiple loaded holes.
The effects of finite geometry are taken into account by
the variational formulations. An iterative scheme is
carried out to handle the deformation compatibility
between all joined members. In the numerical method,
finite element analyses are conducted using the
commercial packages MSC/Patran and MSC/Nastran.
Gap elements are used to simulate the rivet-hole
interactions. Both linear and nonlinear deformations
are considered. The two modeling procedures are
complementary to each other. While the analytical
method is efficient for parametric studies, the numerical
method is capable of dealing with relatively more
complicated geometry and loading conditions. Results
of several example problems are presented and
reasonably good agreement between the two modeling
procedures is demonstrated.
INTRODUCTION
Riveted lap joints, made of metallic and composite
materials, are structural components commonly used
for assembly of parts in aircraft structures. The current
economical climate requires that both military and
civilian aircraft be operated beyond their design lives.
Therefore maintenance and repair are important issues
to be investigated for aging aircraft structure [1].
Research programs have been underway at the Institute
for Aerospace Research of the National Research

* Member of ASC, CSME, and CASI
** Member of ASME and ASCE
Copyright 1998 by National Research Council Canada. Published
by the American Institute of Aeronautics and Astronautics, Inc. with
permission.
Council Canada on bolted patch repair of damaged
composite skin panels [2] and multiple site damage
(MSD) of lap splice joints [3]. A main task in the
research programs is to develop efficient modeling
procedures for mechanically fastened composite joints
and riveted lap joints in aircraft fuselage structure.
Both analytical and numerical methods are to be
examined. This is challenging work because of the
complex geometry and load transfer in the joints. An
important consideration about the modeling procedures
is to maintain a good balance between accuracy and the
computational efficiency. In the discussion below, the
geometry configuration of the two types of joints are
idealized as two plates joined together by frictionless
rivets without distinguishing the differences in
geometry.
Various analytical approaches following
Lekhnitskii [4] have been developed for composite
joints [5-6]. These approaches are based on an infinite
plate assumption. Finite width correction factors are
required when applying the analytical approaches to
finite sized plates [7]. These correction factors, in most
cases, are developed based on finite element analyses.
This requirement of using a numerical technique to
generate correction factors makes the Lekhnitskii's
complex potential based analytical approaches less
suitable for designing complex joint configurations. In
addition, the closed-form solutions are valid for pure
bearing loading only.
To eliminate the requirement for finite width
correction factors, the analytical solutions involving
complex stress potentials have been employed in
different fashions. The boundary collocation method
was used for crack problems [8] and contact problems
[9]. This method involves a conformal mapping and a
least-squares scheme. A hybrid method in conjunction
with complex stress functions was formulated for
aircraft panels with multiple site damage near the rivet
holes [10]. More recently, a mixed form of complex
variational approach was developed for composite
2
joints containing multiple rivet holes under a uni-
directional load [11].
In addition to the analytical approaches, numerical
methods have been widely used for both composite
joints and metallic lap joints. Two-dimensional finite
element analyses with assumed distribution of rivet
loads were conducted for strength prediction of
composite joints [12-14]. Finite element analyses were
also carried out for metallic lap joints in different ways
such as 2D model with merged nodes [15], 2D model
with assumed displacement boundary conditions [16],
and 3D model with interface or gap elements [17, 18].
While the three-dimensional analyses can provide more
accurate stress predictions, particularly for locations
around rivet holes, extensive modeling efforts and
computer time are required and thus only a small
number of rivet holes can be modeled. In the present
work, two-dimensional finite element models have
been developed for rivet lap joints using gap elements
to simulate the rivet-hole interactions. The effects of
the number of gap elements are examined from the
accuracy and computational efficiency point of view.
In this paper, the development of the analytical and
numerical modeling procedures for lap joints involving
single and multiple rivet holes is described. Detailed
results for several example problems of lap joints under
uni-directional loading condition are presented and
discussed. The two modeling procedures are
complementary to each other and reasonably good
agreement between the results from the two modeling
procedures is demonstrated.
STATEMENT OF PROBLEM
The problem under study in this paper is a single
shear lap joint, which involves two composite or
metallic plates joined by single or multiple rivets, as
demonstrated in Figure 1 for a joint with two rivets. In
analytical modeling, the two plates are treated, in
general, as anisotropic plates since the isotropic
metallic plate can be considered as a special anisotropic
case. The two plates are of the same finite width and
they have, in general, m circular loaded holes of various
sizes that are arbitrarily located. The tensile load, P,
applied at the edge of the lower plate (Plate 2), is in
equilibrium with the m rivet loads, P
i
( i = 1, 2, ..., m).
This load is transferred to the upper plate (Plate 1).
It is noted that the actual stress state at the rivet
location is complex and three dimensional in nature.
To develop relatively simple modeling methods with
reasonable accuracy, some assumptions are made in
this work:
(i) the load on one plate is transferred by flexible
frictionless rivets over half of the hole edges, and
(ii) the secondary bending of the plates due to load
eccentricity is negligible compared to the in-plane
deformations.
Therefore the two plates are assumed to be in a plane
stress state and the rivets are modeled as elastic beams
with appropriate end conditions.
P
Plate 1
Fateners
P
1 2
P
Plate 2

1

2
A B
C D
A B
C D
P P
u
1
u
2
1
P
2
P
Figure 1. Configuration of two-rivet lap joint
ANALYTICAL MODELING METHOD
Iterative scheme
As seen in Figures 1, the external load, P, applied
on Plate 2 is transferred by the rivets to Plate 1. This
load transfer results in the deflection of rivets and the
elongation of the holes in the upper plate. To consider
the deformation compatibility between the jointed
members, an iterative scheme that involves the
following steps is established for joint stress analysis.
Step 1. Assign initial values for rivet loads that are
in equilibrium with the externally applied load, P.
Step 2. Calculate the hole elongation in Plate 1
under the applied load and the rivet loads. The rivet
loads are simulated by a cosine function.
Step 3. Calculate the deflections of the rivets under
the shear loads transferred by the rivets.
Step 4. Calculate the summation of the hole
elongations and rivet deflections in Plate 1. The
3
combined deformation is taken as the rigid body
movement of the contact region of the corresponding
rivet holes in Plate 2. Calculate the reactions on holes
of Plate 2 under the rigid body motions and the applied
load, P.
Step 5. Take the reactions on the holes of Plate 2 as
new rivet loads and repeat steps 2~4 until a convergent
solution is obtained.
Variational Formulations for Plate 1 and Plate 2
Since the two plates are in a plane stress state, the
theories of two-dimensional anisotropic elasticity are
applied to the stress analysis of the two plates. That is
two complex stress potentials are used to derive the
stress and displacement components. For the sake of
simplicity an identical set of notations are employed for
the two plates and the basic equations are written as:
N
d z
dz
d z
dz
N
d z
dz
d z
dz
N
d z
dz
d z
dz
x
y
xy
= +
= +
= +
2
2
2
1
2 1 1
1
2
2 2 2
2
1 1
1
2 2
2
1
1 1
1
2
2 2
2
Re[
( ) ( )
]
Re[
( ) ( )
]
Re[
( ) ( )
]

(1)
u p z p z
v q z q z
= +
= +
2
2
1 1 1 2 2 2
1 1 1 2 2 2
Re[ ( ) ( ) ]
Re[ ( ) ( ) ]


(2)
where N
x
, N
y
, and N
xy
are the stress components, u and v
are the displacement components,
1
and
2
are the two
complex stress potentials, and other variables in Eqs.
(1) and (2) are complex constants associated with the
material properties.
Plate 1 is under the applied load, P, and rivet loads,
P
i
. The rivet loads cause a pressurized stress against
the right half of the hole edge and this stress is assumed
to be in a cosine distribution as:
p
P
r
r
i i
i
( ) cos

=
2
(3)
where r
i
is the radius of the i-th rivet hole and the
angle , with respect to the loading direction, denotes
the location on the hole edge.
Since Plate 1 does not have any displacement
boundary conditions, the minimum potential energy
theorem is applied. The total potential energy of Plate
1 is written as:
~
{ ( ) ( )

1 11
2
12 22
2
1
2
2 = + +

A
u
x
A
u
x
v
y
A
v
y

+ + +
+ +
=
2
16 26
66
2
2
( ) ( )
( ) } |
/
A
u
x
A
u
x
u
y
v
x
A
u
y
v
x
dx dy
P
W
u dy
x L
y
y
B
C

/
/
( ) ( )
2
2
1
p u r d
r
i
i
m
r i
(4)
where A
ij
(i, j = 1, 2, 6) are the in-plane stiffness
coefficients, denotes the domain, L the length and W
the width of the plate.
Using the integration by parts, the first order
variation of the above energy functional with respect to
u and v, respectively, is derived and written as:


~
( ) |
[ ( ) ]|
/
/

1
2
= +
+ +
=
=

N u N v dx
N
P
W
u N v dy
xy y y W
x
x
x xy x L
y
y
A
B
B
C

+
+ +
+
=
=

( ) |
( ) |
{ [ ( ) ]
/
/
/
/
N u N v dx
N u N v dy
N p u N u r d
xy y y W
x
x
x xy x L
y
y
r
i
r
i
i
m
r
i
r
i i
i
C
D
D
A

2
2
2
2
1
+ +

/
/
( ) }
2
3 2
N u N u r d
r
i
r
i
r
i i
i
(5)
where the stress components are related to the
displacement components by the constitutive equations.
In addition, the area integration in Eq. (5), which are
associated with the two equilibrium equations, are
eliminated because of the automatic satisfaction by
using the stress potentials. Therefore the variational
formulation involves only boundary integrations which
are associated with the boundary conditions.
For Plate 2 which has both force and displacement
boundary conditions, a mixed variational formulation is
established. The energy functional in this case is
written as:
~
{ ( ) ( )

2 11
2
12 22
2
1
2
2 = + +

A
u
x
A
u
x
v
y
A
v
y

+ + +
+ +
=
2
16 26
66
2
2
( ) ( )
( ) } |
/
A
u
x
A
u
x
u
y
v
x
A
u
y
v
x
dx dy
P
W
u dy
x L
y
y
D
A

4
+

/
/
( )[ ( ) ]
2
3 2
1
i
i
m
r r
i
i
u u r d (6)
where ) (
i
is the Lagrangian Multiplier for the i-th
hole which represents the compressive stress against
the left half of the hole edge caused by the rivet load.
In a similar way, the first order variation of the above
energy functional with respect to u, v, and
i
,
respectively, is derived and written as:


~
( ) |
( ]|
/
/

2 2
2
= +
+ +
=
=

N u N v dx
N u N v dy
xy y y W
x
x
x xy x L
y
y
A
B
B
C
+
+ +
+
=
=

( ) |
[ ( ) ]|
{ ( )
/
/
/
/
N u N v dx
N
P
W
u N v dy
N u N u r d
xy y y W
x
x
x xy x L
y
y
r
i
r
i
r
i i
i
i
m
C
D
D
A

2
2
2
2
1
+

/
/
[ ( ) ( )
2
3 2
N u u u
r
i i
r
i
r
i
r
i i
+ N u r d
r
i i
i
] } (7)
Similarly, only boundary integrations are involved
in Eq. (7). To seek the solutions using the formulations
in (5) and (7), the trial functions are selected, for the
stress potentials, as:

1 10 1 1
1
1 10 1 1 1 1
2 20 2 2
1
1
20 2 2 2
1
= +
+ +
= +
+ +
=
=

=
=
=

C C z
D D
C C z
D D
n
n
n
N
i
m i
i n
i
i
n
n
N
n
n
n
N
i
m i
i n
i
i
n
n
N
( ln )
( ln )
(8)
where ) 2 , 1 ( , = + = k y x z
k k
are complex coordinate
variables, C D C D
n n
i
n n
i
1 1 2 2
, , , are complex constants to
be determined, and the mapping functions are:

ki
k k i k
i k
z z r
r I
k i m
=
+

= =
2 2 2
1
1
1 2 1 2
( )
( )
, , , ,...,
(9)
For the Lagrangian multipliers, the following
trigonometrical functions are selected as:
m i
j j
J
j
i
j
i
j
i i
..., , 2 , 1
, ] ) ( cos ) ( sin [ ) (
1
1 2 2 1
=
+ + =

=
+

(10)
For accurate and convergent solutions from the
variational formulations, some additional conditions
must be imposed. The logarithm terms in the stress
potentials must result in single-valued displacements
around each rivet hole. The applied load must be in
equilibrium with the compressive stresses at rivet holes
represented by the Lagrangian multipliers.
There are 2(m + 1)(N + 1) complex constants in
Eq. (8) and m (J + 1) real constants in Eq. (10) which
are determined by the variational formulations. Once
these constants are determined, the stress and
displacement components are calculated using Eqs. (1)
and (2).
FINITE ELEMENT MODELING METHOD
Finite element modeling techniques have been
proposed for the purpose of analyzing problems with
relatively complicated geometry and loading
conditions. A major difficulty in modeling riveted
joints is the idealization of the load transfer between the
rivet and the plate. The resulting stress distribution
around the rivet hole is largely influenced by the
procedures followed in the idealization. Therefore
several alternative strategies for FE modeling have been
investigated in the present program on multiple site
damage in lap joints. For comparison with the
analytical method outlined in the previous section, the
two-dimensional modeling method using gap/spring
elements is discussed in this section. The commercial
finite element packages MSC/Patran and MSC/Nastran
were used for the analyses.
A salient idea in the present modeling method is to
simulate the rivet by an assembly of a circular beam
and two discs that are rigidly connected at the end of
the beam. The thickness of the two discs is equal to
that of the two plates and the geometry of the beam is
identical to that of the rivet (countersink is not
considered). The two plates with rivet holes and the
two end discs of the rivet are modeled by 2-D plane
stress elements. The rivet itself is modeled by a single
beam element. Gap/spring elements are then generated
to connect the nodes in the contact region along the
rivet/plate boundary. Figure 2 demonstrates a finite
element model and the rivet hole representation for a
four-rivet single lap joint using 80 gap elements around
each rivet hole.
Gap/spring elements are one-dimensional elements
that stimulate unidirectional point-to-point contact
between two media (in the present models the plate and
the rivet). Gap elements are appropriate in modeling
the load transfer between the rivet and the plate under
general loading conditions. This is because gap
elements require two separate values of axial stiffness
5
constants: the opening (or tensile) and the closed (or
compressive). Two types of nonlinear analyses are
usually conducted when using gap elements. One is
associated with both the update of the contact region
and the large deformation of the whole structure,
referred to as Nonlinear 1 in this paper. The other,
referred to as Nonlinear 2, is associated with the update
of the contact region only. Spring elements are used
only if the contact region is known to be fixed during
the loading history as only one stiffness constant is
required for spring elements. Linear analyses are
usually conducted when using spring elements, as there
is no update of the contact region.
A major deficiency when using gap elements is
that it requires non-linear analysis. This may not be
favorable if the structure contains a large number of
rivets since the computational time required is much
more than that in the traditional linear analysis.
Therefore a good compromise needs to be achieved
between the accuracy and the computational efficiency.
A convergence study was conducted to determine
the appropriate number of gap/spring elements that
would be sufficiently adequate to model the load
transfer between the rivet and the plates. For
simplicity, the analysis was performed for a single rivet
joint under a uni-directional load. Linear analysis was
done using 30 to 200 spring elements along one half of
the rivet hole because the load is uni-directional.
Nonlinear analysis with large deformation of the whole
structure (Nonlinear 1) was done in the corresponding
models by changing the springs to gap elements. When
200 gap/spring elements are used, the model contains a
total of 28,788 degrees of freedom. The degrees of
freedom decrease to 6584 for a model with 50
gap/spring elements. Note that as the number of
gap/spring elements increases along the rivet hole the
mesh becomes finer in the vicinity of the rivet area.
This obviously influences the calculated stress values
since the size of the elements around the rivet hole is
different for each case.
Figure 3 shows the convergence of the peak stress
obtained from linear and non-linear analyses with
varying number of gap/spring elements. Apparently,
the values obtained by the linear analysis are upper
bounds for those by the non-linear analysis. The
change in the peak stress becomes insignificant by
increasing the number of gap/springs from about 100 to
200. The numerical values of the peak stress and the
CPU time required to run each model are listed in Table
1. The values of the peak stress obtained from the
model with 40 gap/spring elements are relatively close
to the convergent values but the CPU time for this
model is significantly less than that for the model with
200 gap/spring elements. This indicates that a model
with 40-100 gap/spring elements would be a good
compromise between the accuracy and the
computational efficiency. Therefore 80 gap/spring
elements are used in the present finite element models
for the example problems discussed below.
0
1
2
3
4
5
6
0 20 40 60 80 100 120 140 160 180 200
Number of gaps/springs
P
e
a
k

s
t
r
e
s
s

Non-Linear Analysis
Linear Analysis
Figure 3. Convergence of peak stress
(a)
(b)
(c)
Figure 2. Finite element model of a four-
rivet joint: (a) front view, (b) top view,
and (c) rivet representation
6
RESULTS AND DISCUSSION
Using the analytical and numerical modeling
methods outlined above, several example problems
have been analyzed. Detailed results from the two
modeling methods are presented and discussed below.
In the first case, for the purpose of demonstrating
the effectiveness of the complex variational approach, a
rectangular composite plate AS4/3501-6 [45/0/-
45/0/90/0/45/0/-45/0]
2s
with three loaded holes was
analyzed. The dimensions of the plate, as shown in
Figure 4, are: W = 101.6 mm, L = 152.4 mm, e = 41.2
mm, s = 35 mm, d
1
= d
3
= 20 mm, and d
2
= 30 mm. A
linear finite element analysis was also conducted where
the rivets were modeled by a set of stiff spring
elements. A comparison of the stress contours in the
loading direction from the complex variational
approach and the finite element analysis are presented
in the upper and lower parts of Figure 5, respectively,
and an excellent agreement is shown.
e s s
W
d
1
d
2
d
3
p
L
In the following examples of lap joint geometry
configurations, the two plates are aluminum with a
thickness of 2.54 mm. All rivet holes have a diameter
of 7.9375 mm. Three types of analyses (Linear,
Nonlinear 1 and Nonlinear 2) were conducted using the
finite element method. The finite element results are
compared with those from the analytical method. In the
analytical method for multi-rivet joints, 3 complex and
6 real terms were taken in each stress potential and each
Lagrangian multiplier, respectively.
The stress values around rivet holes are of
importance in strength prediction of lap joints. A single
rivet joint, as shown in Figure 6, was analyzed in the
second case by the analytical method, with 6 terms in
each stress potential, and by the finite element method.
In this joint, L1 = 63.5 mm, L2 = 79.375 mm, W = 38.1
mm, and e = 31.75 mm. The distributions of the hoop
and bearing stresses around the rivet hole in Plate 2 are
shown in Figures 7 and 8, respectively. It is seen that
the overall distributions from the analytical method and
the three types of FE analyses agree very well. The
unreasonably high peak values from the linear analysis
are due to the discontinuity of stiffness at the junctures
of the contact and non-contact regions. It is therefore
concluded that linear finite element analysis using
spring elements is not appropriate for determining peak
stresses.
-6
-5
-4
-3
-2
-1
0
1
2
3
0 30 60 90 120 150 180 210 240 270 300 330 360
Angular position (deg.)
N
o
r
m
a
l
i
z
e
d

s
t
r
e
s
s
Linear
Non-Lin.1
Non-Lin.2
Analytical
Figure 7. Bearing stress around rivet
hole
P
P
e
W
P
l
1
L1
L2
Figure 6. A single rivet lap joint
Figure 4. A composite plate with three loaded
Figure 5. Stress distributions around three loaded
holes (upper: analytical; lower: finite element)
7
The second lap joint analyzed was one with two
rivets as shown in Figure 9. In this joint, L1 = 95.25
mm, L2 = 111.125 mm, W = 38.1 mm, and e = 31.75
mm. The rivet loads, peak bearing stress and peak hoop
stress calculated from the analytical and finite element
methods are presented in Table 2. It is seen that the
analytical results are in close agreement with the finite
element data, particularly with those from the nonlinear
analysis in which the nonlinearity is only associated
with the update of the gaps opening and closing
(Nonlinear2).
A third lap joint, containing four rivets as shown in
Figure 10, was analyzed. In this joint, L1 = 95.25 mm,
L2 = 111.125 mm, W = 69.85 mm, e = 31.75 mm, and s
= 31.75 mm. The rivet loads, peak bearing stress and
peak hoop stress calculated from the analytical and
finite element methods are presented in Table 3.
Similar agreement between the analytical results and
the finite element data is noted as in the two-rivet joint.
The last lap joint analyzed was one containing five
rivets as shown in Figure 11. In this joint, L1 = 190.5
mm, L2 = 206.375 mm, W = 38.1 mm, e = 31.75 mm,
and s = 31.75 mm. The rivet loads, peak bearing stress
and peak hoop stress calculated from the analytical and
finite element methods are presented in Table 4. Similar
agreement between the analytical results and the finite
element data is also noted as in the two-rivet and four-
rivet joints. To show the comparisons in a graphical
form, the rivet loads and peak stresses in this joint are
displayed in Figures 12 and 13, respectively.
-2
-1
0
1
2
3
4
5
6
7
0 30 60 90 120 150 180 210 240 270 300 330 360
Angular position (deg.)
N
o
r
m
a
l
i
z
e
d

s
t
r
e
s
s


Linear
Non-Lin.1
Non-Lin.2
Analytical
Figure 8. Hoop stress around rivet
hole
P
e s
P
s s s
W
l
5
P P
4
P
3
P
1
P
2
L1
L2
Figure 11. A four-rivet lap joint
Figure 12. Rivet loads in five-rivet joint
0
0.1
0.2
0.3
0.4
1 2 3 4 5
Hole No.
R
i
v
e
t

l
o
a
d
s

(
p
i
/
P
)
Analytical
FE Non-Lin.2
P
P
e s
W
P P
l
1 2
L1
L2
Figure 9. A two-rivet lap joint
P
P
e
s
s
1
4
3
2
W
P P
P P
l
L1
L2
Figure 10. A four-rivet lap joint
8
CONCLUSIONS
Analytical methods have been developed based on
the complex variational approach for lap joints with
single or multiple rivet holes. The joined plates can be
either metallic or composite materials. The stresses in
the two joined plates and the rivet loads can be
determined. There is no need for finite width
correction factors when the rivet holes are not small
compared to the width of the joined plates. An iterative
procedure is implemented to take into account the
flexibility of all the joined members.
Finite element simulations have also been carried
out with the objectives: (i) to verify the analytical
methods developed and (ii) to explore effective
modeling procedures to simulate the rivet-hole
interactions. Both gap and spring elements were used
and the effects of these elements on the load transfer
were examined. Particularly, the results using different
numbers of gap/spring elements around rivet holes
were evaluated from the points of view of
computational efficiency and accuracy. Comparisons
between the results from the analytical and the finite
element modeling procedures were presented and
discussed.
Based on the results discussed, the following
conclusions can be drawn:
(i) The analytical method developed using the
complex variational approach is efficient for
determining the stresses in multi-rivet lap joints
under a uni-directional load;
(ii) Preliminary design work can be carried out using
the analytical method in a PC environment;
(iii) Further development work can be conducted to
extend the capability of the analytical method to
deal with multi-rivet joints under general loading
conditions;
(iv) The two-dimensional finite element modeling
technique using gap/spring elements is an
effective way to analyze riveted lap joints with
relatively complicated geometry and loading
conditions;
(v) The rivet representation by an elastic beam with
two end discs is appropriate and approximately
40-100 gap/spring elements along the contact
region are sufficient for adequate accuracy and
efficiency;
(vi) The linear analysis using spring elements predicts
accurate overall stress distributions in the joint but
it is not appropriate for determining the peak
stresses at the junctures between the contact and
non-contact regions;
(vii) The nonlinear analysis associated with both the
update of gap elements and the large deformation
of the whole structure provides the lower bounds
for the peak stresses. This analysis requires the
most extensive computing time;
(viii) The nonlinear analysis associated with the update
of gap elements only provides the peak stresses,
which are between the values calculated from the
linear analysis and the complete nonlinear
analysis (Nonlinear 1). Also the computing time
required by this type of nonlinear analysis is
significantly less than that required by Nonlinear
1. Therefore it is recommended to use this type of
nonlinear analysis with 40-100 gap elements in
future finite element modeling of riveted lap
joints.
ACKNOWLEDGEMENT
This work has been carried out under IAR Program
303, Aerospace Structures, Projects SAA-00 and SAB-
00, Computational Structures Technology
Development. The financial assistance received from
DND CRAD Funding (FE#1450VA 3GB12 101A
7298) is gratefully acknowledged.
REFERENCES
[1] Jones, R and Miller, N. J., eds., Proceedings of
International Conference on Aircraft Damage
Assessment and Repair, The Institute of
Engineering, Australia, 1991.
[2] Poon, C. and Xiong, Y., Bolted joint technology
in composite structures --- analytical tools
Figure 13. Peak stresses in five-rivet joint
-2
-1
0
1
2
3
4
5
1 2 3 4 5
Hole No.
Analytical
FE Non-Lin.2
Hoop
Bearing
9
development, AGARD Conference Proceedings
590, Florence, Italy, 1996.
[3] Eastaugh, G., Simpson, D. L., Straznicky, P. and
Wakeman, R. B., A special uniaxial coupon test
specimen for simulation of multiple site fatigue
crack growth and link-up in fuselage skin splices,
AGARD Conference Proceedings 568, Rotterdam,
The Netherlands, 1995.
[4] Lekhnitskii, S. G., Anisotropic Plates. Gordon and
Breach Science Publishers, New York, 1968.
[5] De Jong, Th., Stresses around pin-loaded holes in
elastically orthotropic or isotropic plates, Journal
of Composite Materials 11, 313-331, 1977
[6] Zhang, K. D. and Ueng, C. E. S., Stresses around
a pin-loaded hole in orthotropic plates with
arbitrary loading direction. Composite Structures
3 119-143, 1985.
[7] Garbo, S. P. and Ogonowski, J. M., Effect of
variances and manufacturing tolerances on the
design strength and life of mechanically fastened
composite joints, I. Methodology development
and data evaluation. AFWAL-TR-81-3041, 1982.
[8] Bowie, O. L. and Neal, D. M., A modified
mapping-collocation technique for accurate
calculation of stress intensity factors. Int. J.
Fracture Mechanics 6, 199-206, 1970.
[9] Madenci, E. and Ileri, L., Analytical
determination of contact stresses in mechanically
fastened composite laminates with finite
boundaries, Int. J. Solids Structures 30, 2469-
2484, 1993.
[10] Tong, P., Greif, R. and Chen, L., Residual strength
of aircraft panels with multiple site damage,
Computational Mechanics 13, 285-294, 1994.
[11] Xiong, Y., An analytical method for failure
prediction of multi-rivet composite joints.
International Journal of Solids and Structures 33,
pp. 4395-4409, 1996.
[12] Waszczak, J. P. and Cruse, T. A., Failure mode
and strength predictions of anisotropic bolt
bearing specimens. Journal of Composite
Materials 5, 421-425, 1971.
[13] Crews, Jr., J. H., Hong, C. S., and Raju, I. S.,
Stress-concentration factors for finite orthotropic
laminates with a pin-loaded hole. NASA
Technical Paper 1862, 1981
[14] Chang, Fu-Kuo, Scott, R. A., and Springer, G. S.,
Strength of bolted joints in laminated composites.
AFWAL-TR-84-4029, 1984.
[15] Roach, D. P. and Jeong, D. Y., Experimental and
analytical program to determine strains in 737 lap
splice joints subjected to normal fuselage
pressurization loads, AIAA-96-1558-CP, 1996.
[16] Beuth, J. L. and Hutchinson, J. W., Fracture
analysis of multi-site cracking in fuselage lap
joints, Computational Mechanics 13, 315-331,
1994.
[17] Fung, C.-P. and Smart, J., Riveted single lap
joints Part 1: a numerical parametric study, Proc.
Instn. Mech. Engrs. 211, 13-27, 1997.
[18] Bellinger, N. C., Krishnakumar, S. and
Komorowski, J. P., Modeling of pillowing due to
corrosion in fuselage lap joints, Canadian
Aeronautics and Space Journal 40, 1994.
Table 1. Stress concentration and CPU time (sec.)
for various models
No. of Analysis type
Gaps or Spring (Linear) Gap (Nonlinear 1)
Springs SCF CPU SCF CPU
30 5.16 18 4.01 72
40 5.20 22 4.18 86
50 5.35 26 4.30 93
100 5.60 32 4.74 151
200 5.73 110 4.97 853
Table 2. Results of a two-rivet lap joint
Hole No.
Method Quantity 1 2
p
i
/p 0.49 0.51
Analytical Peak
r
-2.51 -2.55
Peak

2.94 4.48
FE p
i
/p 0.50 0.50
Linear Peak
r
-2.58 -2.28
Peak

3.38 4.67
FE p
i
/p 0.50 0.50
Non-Lin.1 Peak
r
-2.62 -2.31
Peak

2.40 3.89
FE p
i
/p 0.50 0.50
Non-Lin.2 Peak
r
-2.59 -2.28
Peak

2.87 4.55
10
Table 3. Results of a four-rivet lap joint
Hole No.
Method Quantity 1 2 3 4
p
i
/p 0.25 0.25 0.25 0.25
Analytical Peak
r
-2.32 -2.32 -2.66 -2.66
Peak

2.91 2.91 4.29 4.29

FE p
i
/p 0.25 0.25 0.25 0.25
Linear Peak
r
-2.38 -2.38 -2.22 -2.22
Peak

3.24 3.24 5.19 5.19

FE p
i
/p 0.25 0.25 0.25 0.25
Non-Lin.1 Peak
r
-2.42 -2.42 -2.27 -2.27
Peak

2.31 2.31 3.79 3.79

FE p
i
/p 0.25 0.25 0.25 0.25
Non-Lin.2 Peak
r
-2.39 -2.39 -2.22 -2.22
Peak

2.73 2.73 4.47 4.47

Table 4. Results of a five-rivet lap joint
Hole No.
Method Quantity 1 2 3 4 5
p
i
/p 0.28 0.16 0.12 0.15 0.30
Analytical Peak
r
-1.46 -0.88 -0.67 -1.04 -1.54
Peak

1.60 1.77 1.78 2.63 3.48

FE p
i
/p 0.31 0.14 0.10 0.14 0.31
Linear Peak
r
-1.58 -0.68 -0.61 -0.79 -1.49
Peak

2.07 1.80 1.98 2.50 4.01

FE p
i
/p 0.31 0.14 0.10 0.14 0.31
Non-Lin.1 Peak
r
-1.59 -0.71 -0.64 -0.83 -1.52
Peak

1.49 1.55 1.73 2.19 3.46

FE p
i
/p 0.31 0.14 0.10 0.14 0.31
Non-Lin.2 Peak
r
-1.57 -0.69 -0.62 -0.80 -1.48
Peak

1.77 1.80 1.99 2.51 4.00