STRAIN AND STRESS DISTRIBUTION IN A ROTATING

DISK MADE BY 2D C/C LAMINATED COMPOSITES

Fenghua Zhou, Akinori Ogawa and Ryosaku Hashimoto

Aeroengine Division, National Aerospace Laboratory

7-44-1 Jindaijihigashi, Chofu, Tokyo 182-8522, JAPAN

SUMMARY: Spin tests of 2D laminated C/C disks have been performed to study the
materials’ deformation and bursting failure properties under centrifugal loading. The strain
distributions on disk-shape specimens were measured in the tests. In this paper, we use the
composite mechanics method to analyze the stress and strain states within the C/C specimen.
Firstly, classical laminate theory is used to deduce the in-plane elastic constants of the
composites. Then, to analyze the complicated deformation behavior of an orthotropic rotating
disk, a Ritz-type elastic analysis based on variation theory was proposed. By adequately
assuming the deformation functions, the distributions of displacement, strain and stress
components within the disk are accurately calculated. The calculated results are compared
with the experimental measurement, from which the elastic constants of the unidirectional
C/C lamina are determined. The failure phenomena of different specimen are discussed. From
measured bursting velocity, the shearing strength of the UD-lamina is estimated.

KEYWORDS: C/C Composites, 2D Laminates; Rotating Disk, Ritz Method Elastic Analysis

INTRODUCTION

Carbon fiber reinforced carbon matrix composites (C/C composites) have excellent
properties of retaining strength at high temperature up to 2000oC, good stiffness, light
weightiness, low thermal expansion coefficient, etc. These properties make the composite a
good candidate material in high-temperature aerospace structures. A potential use of C/C
composite is manufacturing high temperature rotating components [1]. There are numerous
reports on the developments and applications of these materials (for example, [2, 3]).
However, because of the following reasons: 1. The mechanical properties of C/C composites
are strongly dependent on their manufacturing process; 2. These materials are expensive so
that generally there are not many specimens for testing; 3. The experimental methods for
testing these materials/components are not mature; etc., the fundamental data of C/C
composites, especially those important for designing a rotating component, is not sufficient.
In the authors’ research group, systematic investigation on C/C composites is performed,
aiming to develop a turbine blisk (bladed disk) rotor with 3D woven C/C composite. In the
fundamental phase of this project, tensile, bending, and spin test experiments of different 2D
laminated C/C composites were performed to investigate the materials mechanical properties.
A summary of these researches can be found in Ref. [4]. From the tensile tests, it is found that
[0/90] laminated C/C composites have orthotropic elastic properties, with very high elastic
modules in fiber directions (Exx and Eyy), but very low in-plane shearing modulus (Gxy) and
very small Poisson’s ratio (νxy). Spin tests of C/C composites, laminated in [0/90] and
[0/45/-45/90] sequences, were performed to study the deformation and failure (bursting)
properties of these materials under centrifugal loading. It was found that the failures of the
specimen were due to the in-plane shear stress of each layer. The bursting speed of [0/90]
composites was about 50-60% that of [0/45/-45/90] composites.
As one kind of fiber reinforced composites, the C/C laminate plate can be analyzed by
the structural analytical method frequently used to fiber-reinforced-plastics (FRP). As a trial,
we measured the elastic constants of unidirectional (UD) C/C composites by off-axis tensile
experiments, and used the resulting data to predict the mechanical properties of multilayer
C/C laminates. We found that classical laminate plate theory can be applied to predict 2D
laminated C/C composites quite well [5].
In the present paper, we will analyze the deformation and failure properties of 2D
laminated C/C composite disk under centrifugal loading. Firstly, we give a brief account of
experimental results. Then, we apply the classical laminate theory to deduce the expressions
of the in-plane elastic constants for [0/90] and [0/45/-45/90] laminate plates. We propose a
Ritz-type elastic analyzing approach based on variation theory, to calculate the stress of a
rotating orthotropic disk. By adequately assuming the deformation functions of the rotating
disk, the mechanical state variables: displacement, strain and stress on the disk can be
evaluated quickly and accurately. By comparing the results of analysis and experiments, the
basic mechanical parameters of the materials: elastic constants, shearing strength, etc., are
determined. The bursting phenomenon of different specimens will be explained finally.

TEST MATERIALS AND EXPERIMENTS

The 2D laminated C/C composites being tested were manufactured by following

sequences: 1. The unidirectional carbon fiber cloth plies are stacked to form the plate; 2. The
plate is pitch impregnated and carbonized; 3. The impregnation and carbonization processes
are repeated in multiple cycles to form the densified materials. Two kinds of laminates were
manufactured: the CC-2 laminate with [0/90]ns (cross-ply) stacking sequence; and the CC-3
laminate with the [0/45/-45/90]ns (quasi-isotropic) stacking sequence. Thin, circular disk-type
specimens were cut from the laminates. The dimensions of the specimen were: internal
diameter 2a=100mm, external diameter 2b=160mm. The experiments were performed in a
spin-test apparatus, where the specimen was rotated by an air-driven turbine. Fig. 1a and Fig.
1b are the photographs of the test facility and the specimen before testing.
Two types of the spin-test experiments: non-failure test and failure test were performed.
In the non-failure test, the rotating speed of the specimen was low. During the test, the strain
components on different points of the specimen (A: r=65mm, θ=0o and B: r=65mm, θ=45o)
were measured by a slip-ring measuring system. Fig. 2 illustrates the measuring points on the
specimen surface and the controlling/measuring system.
 Fig.1 (a) Spin test facility; (b) C/C specimen before testing

9 0o
45o
B

0o
A

Fig.2 (a) Strain measuring points on the specimen; (b) Strain measuring system

After non-failure test, the rotating speed was increased until the specimen was broken by
centrifugal force (failure test). The rotating speed at which specimen broke was recorded as
bursting speed. The appearance of specimen at the instant of bursting was photographed by a
special photographic system where the stroboscopic lamb was triggered by the burst-sensing
circuit. Details of the experiments were reported in Refs. [6] and [7]. In Fig. 3 the snapshots
of specimen at bursting were shown. The broken specimens were shown in Fig. 4. The
bursting speeds of different materials were listed in Table 1.

Fig.3 Bursting instants of C/C specimen: (a) [0/90] specimen; (b) [0/45/-45/90] specimen
 Fig.4 Specimen after burst: (a) [0/90] specimen; (b) [0/45/-45/90] specimen

TABLE 1 BURSTING SPEED OF DIFFERENT C/C LAMINATES

Lamination Bursting Speed
CC-2 [0/90]ns 27600 rpm
CC-3 [0/45/-45/90]ns 49500 rpm

IN-PLANE ELASTIC CONSTANTS OF 2D C/C LAMINATES

For a unidirectional lamina of C/C material, there are four in-plane elastic constants: the
elastic modulus in fiber direction EL, elastic modulus in transverse direction ET, the shearing
modulus GLT, and the Poisson’s ratio νLT. When the UD lamina were stacked to form a 2D
laminate plate (Fig. 5), the elastic parameters of the laminate plate can be calculated from the
UD lamina elastic constants, by using classical laminate theory. Detailed deducting
procedures are omitted here, which can be found in a composite mechanics textbook (e.g., [8]).
In case of the [0/90] cross-ply laminate, the engineering elastic parameters in x- and
y-directions are identical. Orthotropic property like this is called cubic anisotropy, there are
three independent in-plane elastic constants, which are expressed as:
E + 2 E L ET + ET (1 − 4ν LT )
2 2 2
Ex = E y = L
2( E L + ET )(1 − ν LT ET E L )
2
(1)
2ν E
ν xy = ν yx = LT T , G xy = GLT
E L + ET
The [0/90] cross-ply laminate has the property of orthotropic elasticity. We define the
orthotropic parameter α, by the following formula:
2(1 − ν xy )G xy 4( E L + ET − 2ν LT ET )(1 − ν LT ET E L )G LT
2
α= = (2)
E L + 2 E L ET + ET (1 − 4ν LT )
2 2 2
E xx
Ordinarily, ET<<EL, GLT<<EL, it can be estimated that for a cross-ply laminate, νxy<<1,
α<<1, the material has strong orthotropic property.
 L

zz yy
0°
°
°
°
xx 9
0°
°
°
°

Fig.5 [0/90] Laminate: fiber oriented in x-, y-directions

On the other hand, For the [0/45/-45/90] laminate, the elastic constants are as following:

[E L + ET (1 + 2ν LT )][E L + ET (1 − 2ν LT ) + 4G LT (1 − ν LT ET / E L )]
2

Ex = Ey =
[
(1 − ν LT ET E L ) 3E L + ET (3 + 2ν LT ) + 4G LT (1 − ν LT ET / E L )
2 2
]
E L + ET (1 + 6ν LT ) − 4G LT (1 − ν LT ET / E L )
2

ν xy = ν yx = (3)
3E L + ET (3 + 2ν LT ) + 4G LT (1 − ν LT ET / E L )
2

E L + ET (1 − 2ν LT ) + 4G LT (1 − ν LT ET / E L )
2

G xy =
8(1 − ν LT ET E L )
2

It is seen that for this laminate, the relationship G = E 2(1 − ν ) is satisfied. In other

words, the orthotropic parameter α=1. The material’s elastic constants are identical in all
directions, i.e., material is quasi-isotropic.

ORTHOTROPIC DISK UNDER CENTRIFUGAL LOADING: ELASTIC ANALYSIS

The problem to be analyzed is shown in Fig. 6. For an (quasi-) isotropic rotating disk
whose elastic constants are expressed by (3), the closed-form elastic solution exists that can
be found in most textbooks of elasticity. However, for a rotating disk with cubic anisotropy as
described by (1), no exact solution exists. Ordinarily, FEM calculation is used to analyze this
problem. Although FEM analysis can be easily performed, it outputs huge amount of discrete
data, which is difficult to sort out for further analysis. Moreover, the calculation accuracy for
the stress is not high in FEM analysis unless fine meshes are used.
Y

θ r
R

θ X

r
D e n s i ty : ρ
R o ta ti o n : ω

Ex=Ey=E
Gxy=G=α α E/2(1+ν
ν)
ν x y=ν
ν

Fig.6 Analyzing problem: a rotating disk as the spin test specimen

 We have developed a Ritz-type elastic analyzing approach based on variation theory to
analyze this problem. The elastic analysis procedures are as following:
In polar coordinate system, the disk’s displacement components in radial direction ur,
and in circumferential direction uθ, are functions of the geometric coordinate (r, θ):

u r = u r (r ,θ ) , uθ = uθ (r ,θ ) (4)

The strain components are then expressed as:

∂ur u ∂u ∂u ∂u u
εr = , ε θ = r + θ , γ rθ = r + θ − θ (5)
∂r r r∂θ r∂θ ∂r r
The total potential energy of the rotating disk system are calculated by integration:
b 2π b 2π
W =ò òθ rΨ (r ,θ ) drdθ = ò òθ r (Ψe + Ψw )drdθ (6)
r =a =0 r =a =0

where: Ψe =
1
2
[ ]
{ε r , ε θ , γ rθ } E~ (θ ) {ε r , ε θ , γ rθ }T , Ψω = − ρω 2 u r (7)

are the densities of potential energy originated from deformation and centrifugal loading,
respectively.
If the displacement functions ur and uθ are expressed by certain algebraic equations
containing unknown coefficients, the total system potential W can be calculated by integration
throughout the whole circular area. With the condition that W should take the minimum value,
unknown coefficients in the deformation functions are calculated.
We found, with experience, that if displacement functions (4) take the form:
u r = (a1r −3 + a2 r −1 + a3 r + a 4 r 3 + a5 r 5 ) + (b1r −3 + b2 r −1 + b3 r + b4 r 3 + b5 r 5 )Cos (4θ )
(8)
uθ = (c1r −3 + c2 r −1 + c3 r + c4 r 3 + c5 r 5 ) Sin(4θ )
where ai, bi and ci (totally 15) are unknown coefficients, then the present problem can be
solved with satisfactory accuracy [9].

COMPARISON OF CALCULATION RESULTS AND EXPERIMENTAL RESULTS

The elastic constants of UD-C/C lamina are listed in Table 2. These data are substituted
into (1) and (3), to calculate the elastic constantans of two C/C laminates: CC-2 and CC-3
(Table 3). In Table 3 is also listed the orthotropic parameter α.
TABLE 2 ELASTIC PROPERTIES OF UD C/C PLY
EL (GPa) ET (GPa) GLT (GPa) νLT ρ (Kg/m3)
251.1 2.454 7.0 0.41 1630
TABLE 3 ELASTIC PROPERTIES OF C/C LAMINATES
Ex (GPa) Gxy (GPa) νxy α
CC-2 127 7.0 0.00794 0.1111
CC-3 90.51 35 0.2930 1

Using the data shown in Table 3, the displacement, strain and stress of the rotating disk
are calculated. Fig. 7 shows the distribution of circumferential strain εr, radial strain εθ and the
shear strain εrθ. The rotating velocity of the disk is 18020rpm. The units of the strain
components are microns (µ).
0.08
22.5937

0.06

0.04

0.02

-1015.78
0
0 0.02 0.04 0.06 0.08
a.
0.08
1206.8

0.06

0.04

0.02

137.676
0
0 0.02 0.04 0.06 0.08
b.
0.08
1065.64

0.06

0.04

0.02

-1065.64
0
0 0.02 0.04 0.06 0.08
c.
Fig.7 Distribution of cylindrical strain components on [0/90] disk: (a) εr; (b) εθ; (c) εrθ

Fig. 8 shows the changing of the strain values on A (130mm, 0o) and B (130mm, 45o)
points with the increase of rotating velocity. The experimental data are also plotted in the
figures for comparison. The calculation curves agree with the experimental results quite well.
 1500
CC-2 (R-Str0)

C/C-2 Disk CC-2 (T-Str0)

45o , εθ CC-2 (R-Str0)

1000
CC-2 (T-Str0)
Theo (R-Str0)
Theo (T-Str0)
500 0o , εθ CC-2 (R-Str45)
CC-2 (T-Str45)
CC-2 (R-Str45)
0
0o , εr CC-2 (T-Str45)
Theo (R-Str45)
Theo (T-Str45)
-500

45o , εr

-1000
0 5000 10000 15000 20000
a. Rotation Speed (rpm)

1500
CC-3 (R-Str0)

C/C-3 Disk CC-3 (T-Str0)

CC-3 (R-Str0)
1000
CC-3 (T-Str0)
Theo (R-Str0)

0 or 45o , εθ
o
Theo (T-Str0)
500
CC-3 (R-Str45)
CC-3 (T-Str45)
CC-3 (R-Str45)
0
CC-3 (T-Str45)

0 or 45o , εr
o
Theo (R-Str45)
Theo (T-Str45)
-500

-1000
0 5000 10000 15000 20000
b. Rotation Speed (rpm)

Fig.8 εr and εθ on (130, 0) and (130, π/4) points: (a) [0/90] disk; (b) [0/45/-45/90] disk

FAILURE CHARACTERISTICS

From Fig. 3a, it can be found that the failure of [0/90] disk originated from four radial
cracks propagating in 45o, 135o, 225o and 315o. As the result, the disk specimen broke into
four parts (Fig. 4a). However, the failure patterns of [0/45/-45/90] specimen is complicated,
the specimen burst to many small pieces under centrifugal loading (Fig. 4b).
The stress distributions on every ply of the two laminates is investigated. It is found that
for [0/90] laminate, the maximum shearing stress (in reinforcing Fiber direction LT) appears
at the same location of both 0o-ply and 90o-ply: the inner circle of 45o, 135o, 225o and 315o
points (Fig. 9a). However, for [0/45/-45/90] laminate, the maximum shearing stress of 0o-ply
and 90o-ply appears at 45o, 135o, 225o and 315o points, but those of 45o-ply and -45o-ply
appears at 0o, 90o, 180o and 270o points (Fig. 9b). This explains the different failure
phenomena observed in CC-2 and CC-3 disks.
The absolute values of the shearing stress for different materials are checked below. Fig.
10 compares the distribution of shearing stress in (LT) fiber-direction, τLT,, on the 0o-ply of
CC-2 disk and CC-3 disk, both under centrifugal loading of 18020rpm. The maximum of τLT
appears at the same location of the disk surfaces. The absolute values of τLT(max) on [0/90]
disk and on [0/45/-45/90] disk are 15.55Mpa and 3.38Mpa, respectively. The maximum τLT
on [0/90] disk is more than 4 times as large as that on [0/45/-45/90] disk. This is the main
reason that the burst speed of CC-2 disk is much less than the burst speed of CC-3 disk.

90o 90o

0o
-45o

45o
90o

0o 0o

a. b.

Fig.9 The points where maximum τLT appears: (a) [0/90] disk; (b) [0/45/-45/90] disk

0.08
0

0.06

0.04

0.02

-15.555
0

0 0.02 0.04 0.06 0.08

a.
0.08
0.

0.06

0.04

0.02

-3.37568
0

0 0.02 0.04 0.06 0.08

b.

Fig.10 Distribution of τLT on the 0-ply of: (a) [0/90] disk; (b) [0/45/-45/90] disk
 The shear strength of UD-C/C ply is assumed to be 36Mpa. Based on the maximum
shear stress failure theory, we can predict that the bursting velocity of CC-2 specimen and
CC-3 specimen are 27,4000rpm and 58,800rpm, respectively. This prediction agrees with the
experimental result of CC-2 material. However, the predicted burst speed is larger than the
experimental result of CC-3 materials. We believe that other stress components, the normal
stress in fiber direction σL and in transversal direction σT, also play roles in composite failure.
A more accurate prediction of composite failure may be made by using stress energy failure
criterion.

CONCLUSION

Spin tests of two 2D laminated C/C composite disks, stacked in [0/90]ns and
[0/45/-45/90]ns sequences, were performed. The characteristics of strain distribution and
specimen failure differ very much for two materials. A mechanical analysis was performed to
investigate the deformation and the stress properties of the two disks under centrifugal
loading. The analytical results agree well with the experimental data. The failure phenomena
of two materials are explained.

REFERENCE

1. Rubin, L., “Applications of Carbon-Carbon”, Carbon-Carbon Materials and Composites,

pp267-281, NASA Reference Publication 1254, 1992
2. Hata, H. et.al. Coordinated, “Present status of the development on C/C composites”, Workshop
WS-7 (pp316-327), Proceedings of 72nd JSME Annual Meeting, Tokyo, March, 1995 (in Japanese)
3. Kayama, A., Serizawa, H. and Sato, S., “Recent trend of the research on C/C composites”, Journal
of Japan Society of Composite Materials, 22, No3, pp91-99, 1996 (in Japanese)
4. Ogawa, A., Sofue, Y., Hashimoto, R., Zhou, F. and Yonaiyama, M, "An Applicability
Investigation of C/C composite on turbine rotors”, pp191-198, Proceedings of the 3rd International
Symposium on Hyper/Super-Sonic Transportation Propulsion System, Tokyo, May, 1999 (in
Japanese)
5. Zhou, F. Sofue, Y. and Ogawa, A., "In-Plane elastic parameters of UD C/C composite evaluated
by off-axis tensile tests”, pp49-52, Proceedings of the 27th FRP symposium, Kyoto, March, 1998
6. Ogawa, A., Hashimoto, R. and Matsusue, S. “Spin test on the burst strength of C/C composites”,
pp205-210, Proceedings of 7th Autumn JSGT symposium, Nagasaki, November, 1992 (in Japanese)
7. Ogawa, A., Hashimoto, R., Morimoto, T. and Yonaiyama, M., “Strain distribution on the spin disk
made by (0/90 laminated or 0/45/90 laminated) C/C composites”, 267-270, Proceedings of 11th
Autumn JSGT symposium, Miyazaki, November, 1996 (in Japanese)
8. Daniel, I. and Ishai, O., Engineering Mechanics of Composite Material, Oxford University Press,
1994
9. Zhou, F. and Ogawa, A., “Elastic Analysis of a Rotating Disk with Cubic Anisotropy based on
Ritz Method”, to be published (in Japanese)