Abu Talib CDS
Abu Talib CDS
Abu Talib CDS
Technical Report
Department of Aerospace Engineering, University Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia
Department of Mechanical and Manufacturing Engineering, University Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia
a r t i c l e
i n f o
Article history:
Received 14 April 2009
Accepted 10 June 2009
Available online 14 June 2009
a b s t r a c t
In this study, a nite element analysis was used to design composite drive shafts incorporating carbon
and glass bers within an epoxy matrix. A conguration of one layer of carbonepoxy and three layers
of glassepoxy with 0, 45 and 90 was used. The developed layers of structure consists of four layers
stacked as 45glass =45glass =0carbon =90glass . The results show that, in changing carbon bers winding angle
from 0 to 90, the loss in the natural frequency of the shaft is 44.5%, while, shifting from the best to the
worst stacking sequence, the drive shaft causes a loss of 46.07% in its buckling strength, which represents
the major concern over shear strength in drive shaft design.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Drive shafts for power transmission are used in many applications, including cooling towers, pumping sets, aerospace, structures, and automobiles. In metallic shaft design, knowing the
torque and the allowable service shear stress for the material allows the size of the shafts cross-section to be determined. In order
to satisfy the design parameter of torque divided by the allowable
shear stress [1], there is unique value for the shafts inner radius
because the outer radius is constrained by the space under the
car cabin. Metallic drive shafts have limitations of weight, low critical speed and vibration characteristics.
Composite drive shafts have proven that they can solve many
automotive and industrial problems that accompany the usage of
the conventional metal ones. Numerous solutions, such as ywheels, harmonic dampers, vibration shock absorbers, multiple
shafts with bearings and couplings, and heavy associated hardware, have shown limited success in overcoming the problems [2].
When the length of a steel drive shaft is beyond 1500 mm [3], it
is manufactured in two pieces to increase the fundamental natural
frequency, which is inversely proportional to the square of the
length and proportional to the square root of the specic modulus.
The nature of composites, with their higher specic elastic modulus (modulus to density ratio), which in carbon/epoxy exceeds four
times that of aluminium, enables the replacement of the two-piece
metal shaft with a single component composite shaft which resonates at a higher rotational speed, and ultimately maintains a higher margin of safety. A composite drive shaft offers excellent
vibration damping, cabin comfort, reduction of wear on drive train
* Corresponding author. Tel.: +60 17 249 6293; fax: +60 38 656 7122.
E-mail address: [email protected] (A. Ali).
0261-3069/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2009.06.015
components and increased tire traction. In addition, the use of single torque tubes reduces assembly time, inventory cost, maintenance, and part complexity. The rst application of composite
drive shafts to automobiles was developed by Spicer U-joint divisions of the Dana Corporation for the Ford Econoline van models
in 1985 [3].
Polymer matrix composites such as carbon/epoxy or glass/
epoxy offer better fatigue characteristics because micro cracks in
the resin do not freely propagate as in metals, but terminate at
the bers. Generally, composites are less susceptible to the effects
of stress concentration, such as are caused by notches and holes,
compared with metals [4]. The lament winding process is used
in the fabrication of composite drive shafts. In this process, ber
tows wetted with liquid resin are wound over a rotating male
cylindrical mandrel. In this technique, the winding angle, ber tension, and resin content can be varied. Filament winding is relatively
inexpensive, repetitive and accurate in ber placement [5].
An efcient design of composite drive shaft could be achieved
by selecting the proper variables, which are specied to minimize
the chance of failure and to meet the performance requirements.
As the length and outer radius of drive shafts in automotive applications are limited due to spacing, the design variables include the
inside radius, layers thickness, number of layers, ber orientation
angle and layer stacking sequence. In the optimal design of the
drive shaft, these variables are constrained by the lateral natural
frequency, torsional vibration, torsional strength and torsional
buckling of the shaft. In this study, another constraint is added that
relates to torsional fatigue and selection of the stacking sequence.
The ability to tailor the elastic constants in composites provides
numerous alternatives for the variables to meet the desired
stability and strength of the structure. At rst, the bers are
selected to provide the best stiffness and strength, together with
515
Nomenclature
[A]
A66
E
Eh
Ex
G
g
I
Im
J
K
L
[N]
Q1j
r
stiffness matrix
shear stiffness component
modulus of elasticity
average modulus in the hoop direction
average modulus in the axial direction
modulus of rigidity
acceleration due to gravity
second moment of inertia
mass moment of inertia
polar moment of inertia
torsional spring rate
length
force matrix
reduced stiffness coefcient
radius
ey
/
cxy
g
h
sl
t
nTe
their cost. Indeed, it is the best choice to use carbon bers in all
layers to achieve desired stability. However, due to the cost constraint, a hybrid of layers of carbonepoxy and E-glassepoxy
was introduced. It is evident that the ber orientation angle dictates the maximum bending stiffness, in turn leading to the maximum natural frequency in bending. In this design, the bers were
arranged longitudinally at the zero angle with respect to the shaft
axis. On the other hand, the angle of 45 was used to obtain the
maximum shear strength, while 90 was the best for buckling
strength. The main goal was to achieve the minimum weight while
adjusting the parameters in order to meet a sufcient margin of
safety. The safety criteria specify that a critical speed (natural frequency) must be higher than the operating speed, a critical torque
must be higher than the ultimate transmitted torque and a nominal stress (the maximum at ber direction) must be less than the
allowable stress after applying the failure criteria.
In the shaft design, the shear strength could be increased by
increasing the diameter of the shaft. However, the crucial parameter to consider was the buckling strength. The variable of the laminate thickness has a big effect on the buckling strength and slight
effect on the natural frequency in bending. A discrete variable optimization algorithm could be employed for optimization of layer
thickness and orientation. Vijayarangan et al. [6] used a Genetic
Algorithm, and Rastogi [3] used GENESIS/I-DEAS optimizers for
the optimization of variables in the design of a drive shaft for automotive applications. In other work, Darlow and Creonte [7] employed the general-purpose package, OPT (version 3.2), in
optimizing the lay-up of a graphiteepoxy composite drive shaft
for a helicopter tail rotor.
2. Design procedure
Q 12
E1
1 t12 t21
t12 E2
;
1 t12 t21
Q 22
A
N
X
Q K zK zK1 :
K1
The matrix, [A], is in (Pa m), and the thickness of each ply is calculated in reference to their coordinate location in the laminate. The A
matrix is used to calculate Ex and Eh, which are the average moduli
in the axial and hoop directions, respectively:
"
#
1
A212
;
Ex
A11
t
A22
"
#
1
A212
Eh
:
A22
t
A11
3=2
t
:
r
Q 11
outer radius
inner radius
mean radius
torque
failure torque
thickness.
weight per unit length
strain in y direction
angle of twist
shear stress
coefcient of mutual inuence
carbon bers orientation angle
in-plane shear strength of the laminate
Poissons ratio
torque coupling coefcient
r0
ri
rm
T
Ts
t
W
E2
fn
;
1 t12 t21
E2
and t21 t12 ;
E1
Q 66 G12 ;
1
p gEx I
2
WL4
516
Ix
p
4
r40 r4i pr 3 t:
T s 2pr 2m t sl :
Z
X
Table 1
Material properties [12].
Material
E11 (GPa)
E22 (GPa)
E12 (GPa)
E-glassepoxy
Carbonepoxy
40.3
126.9
6.21
11.0
3.07
6.6
827
1170
1910
1610
Fig. 2. Set of the rst six modes as a function of natural frequency along with their corresponding frequency values.
9
9
8
8
>
>
=
=
< Nx >
< 0 >
0 ;
A1
feg A1 Ny
>
>
>
>
;
;
:
:
Nxy
Nxy
N xy
T
:
2pr 2
8
>
<
38
9 8
e1 >
m2
n2
mn
>
> r1 >
=
= <
=
< ex >
6
7
2
e2 4 n
r2
m2
mn 5 eh
>
> >
>
>
>
:
:
c12 ;
cxh ; : s12 ;
2mn 2mn m2 n2
9
2
38
Q 11 Q 12
0 >
=
< e1 >
6
7
4 Q 21 Q 22
0 5 e2
>
>
:
c12 ;
0
0 Q 33
517
The torsional frequency is another parameter and is directly related to the torsional stiffness (T/u), where u is the angle of twist
and T is the applied torque. The frequency of torsional vibration
can be presented as:
1
ft
2p
s
K
;
Im
10
T Gxy J
;
/
L
11
where J is the polar moment of inertia and L is the length. The shear
modulus can be tailored to its maximum value by orienting the bers at an angle equal to 45. In some applications, like racing cars,
less torsional stiffness is required [2]. The shear modulus can be directly obtained from the extensional stiffness matrix [A], by dividing the shear stiffness component, A66, by the total thickness of
the drive shaft as follows:
Gxy
A66
:
t
Fig. 3. The effect of ber orientation angle on the natural frequency as determined
by changing the rst two 45 layers angles.
12
Fig. 4. The effect on natural frequency of changing the carbon ber orientation
angle in a hybrid drive shaft of stacking 45glass = 45glass =0carbon =90glass .
518
Woven roving Fabric bers used in both [0/90] and [45] lay-up.
The thicknesses of the composites were measured to be: Carbon/
epoxy layer thickness = 0.35 mm and glass/epoxy layer thickness =
0.37 mm.
A linear eigenvalue buckling analysis was conducted to estimate the maximum torque that can be supported prior to losing
stability. In this analysis, the specied load must be closer to the
collapse load in order to obtain accurate results. The output from
the analysis is a factor that can be multiplied by the actual magnitude of the applied load in order to obtain an estimate of the critical torque. Fig. 6 presents the contour of maximum shear stress,
and the deformed shape after linear eigenvalue analysis.
From the contour of maximum shear stress, it can be observed
that higher shear stress accumulated at the two bands that heli-
7000.00
glass ( 0.762)
carbon (0.3022)
carbon (0.532)
carbon (0.762)
6000.00
5000.00
4000.00
3000.00
2000.00
1000.00
0.00
0
10
21
30
38
50
60
70
80
90
519
Fig. 6. First buckling mode shape, and the corresponding contour of maximum shear stress.
1 [ , 0, 90 ]
3 [ 45,45,0, ]
2 [ 45, 45, ]
3000
2500
cally wrapped the cylindrical tube at 45. The tubes buckled when
they lost their stability, and the circular cross-sections became
ovoid.
The best ber orientation angle for maximum buckling strength
is 90. At this angle, the bers are oriented in the hoop direction,
thereby increasing the hoop modulus (Eh). Fig. 7 presents the effect
on buckling torque of changing the ber orientation angles with
only two glass/epoxy layers.
As shown in Fig. 8, there is little dependence of the buckling torque on the ber orientation. It can be observed that, by changing
the angles of the 3rd or the 4th layer, the critical buckling torque
of the drive shaft is not substantially affected by the ber orientation angles. This is attributed to the fact that the modulus, Ex, has
its maximum value at the zero degree ber orientation angle,
2000
1500
1000
500
3200
2700
10
20 30 40 50 60 70
Fiber Orientation Angle (Deg.)
80
90
Fig. 8. Effect of ber orientation angle on the buckling torque of a drive shaft having
stacking of 45glass =0carbon =90glass .
2200
1700
Table 2
Five selected laminates with different stacking sequences and their corresponding
stiffness component D22 (the bulking strength descending from top to bottom).
1200
700
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Fiber Orientation Angle (Deg.)
Fig. 7. Effect of ber orientation angle on buckling torque by changing the rst two
layers of a hglass =0carbon =90glass stack.
[45,45,0,90]
[0,45,45,90]
[0,45,90,45]
[45,45,90,0]
[0,90,45,45]
58.8
55.4
42.9
36.31
36.23
520
2500
2000
1500
1000
500
[4
5,
[9 4 5,
0, 0,
0, 90
[ 9 45 ]
0, ,
0 [9 ,-4 45]
0, 5
4 ,
[0 5, 45]
,4 0,
5, -4
[0 -4 5]
, -4 5
,9
[9 5 ,4 0]
0, 5,
9
[ 0 4 5 0]
,4 ,0
,
5
[ 9 ,90 45]
0, ,
45 -4
[0 , 5]
,-4 -4
5
[4 5 ,9 ,0]
5, 0
,4
[-4 4 5, 5]
5, 90
[0 45 ,0]
,9 ,9
0 0
[ 0 ,45 ,0]
,9 ,
0, -4
- 4 5]
5,
45
]
and the modulus, Eh, has its maximum value at a 90 angle. Since
the expression of buckling torque is related to both moduli, then
the peak value for this torque is realized when the bers are oriented at 0 and 90.
[]4
Carbon-epoxy
The stacking sequence of the layers has an effect on the buckling strength. Although the [A] matrix is independent of the stacking sequence, both the [B] and [D] matrices are dependent upon it.
The drive shaft buckled when its bending stiffness along the hoop
direction could not support the applied torsion load. This normal
bending stiffness is correspondent to the component, D22, of the
bending stiffness matrix [D].
Therefore, the value of D22 species the buckling strength. Fig. 9
presents the effect of stacking sequence on the buckling strength
and it is concluded that the best case scenario stacking sequence
is [45/45/0/90], and the worst case scenario is [0/90/45/45].
Table 2 shows the correspondent D22 components for ve laminates with different stacking sequences. The best stacking offers a
buckling torque of 2303.1 Nm and the worst stacking offers a torque of 1242 Nm, with a loss in buckling resistance capability equal
to 46.07%.
5.4. Effect of coupling between the twisting moment and normal
curvature
The twisting moment resulted from a pure torque loading coupled with a normal curvature in terms of the components D16 and
D26 in the bending stiffness matrix [D]. The D16 component represents the curvature in the longitudinal direction, and as its value
increases the drive shaft tends to bend, and its natural frequency
in bending decreases. The coupling between the twisting moment
and the normal curvature in the hoop direction can be directly related to the coefcient of mutual inuence (g), which is a normalshear coupling. One form of this coupling is:
e
S
gx;xy y 26 :
cxy S66
13
5000
5.3. Effect of layers stacking sequence on buckling torque
- - - []2
6000
4000
Glass-epoxy
3000
2000
1000
0
0
10
20
30
40 50
60
70
Fiber Orientation Angle
80
90
Fig. 10. The effect on the buckling torque of coupling between twisting moment
and normal curvature in hoop direction.
or the curvature in the hoop direction (D26) tends to increase. Angle-ply laminates that consist of 0 and 90 angles do not experience such coupling. On the other hand, a conguration of [h2n]
has a zero coupling or zero for D16 and D26, because for every +h
there is h at the same distance from the mid-plane. However, a
laminate of the conguration [hn]S has a coupling. Here, the component D26 contributes to the buckling strength of the drive shaft.
Fig. 10 exhibits the effect of coupling on the buckling torque for
two congurations having the same value of the component of normal bending stiffness in the hoop direction, D22.
6. Conclusions
The present nite element analysis of the design variables of ber orientation and stacking sequence provide an insight into their
effects on the drive shafts critical mechanical characteristics and
fatigue resistance. A model of hybridized layers was generated
incorporating both carbonepoxy and glassepoxy. Buckling,
which dominates the failure modes, has a value does not increases
regularly with increasing the winding angle. For the worst stacking
sequence, the shaft loses 46.07% of its buckling strength compared
521