Documente Academic
Documente Profesional
Documente Cultură
Ceramics International
journal homepage: www.elsevier.com/locate/ceramint
A R T I C L E I N F O A B S T R A C T
Keywords: In this study, a thermal shock resistance model over a wide range of cooling environment temperatures for fiber
Thermal shock resistance reinforced brittle matrix composites is developed. The model takes into account the combined effects of cooling
Cooling environment temperature environment temperature and temperature-dependent material parameters. The critical temperature difference
Critical temperature difference causing matrix cracking of SiC fiber reinforced reaction bonded Si3N4 composite and Nicalon fiber reinforced
Fiber reinforced brittle matrix composites
borosilicate composite are predicted over a wide range of cooling environment temperatures. The results show
Matrix cracking
that the thermal shock resistance of composites is strongly dependent on the cooling environment temperature.
Moreover, the quantitative influences of matrix Young's modulus, thermal expansion coefficient of matrix, and
interfacial shear stress on thermal shock resistance with increase of cooling environment temperature are studied
in detail. And some insights into improving the thermal shock resistance of fiber reinforced brittle matrix
composites are obtained.
⁎
Correspondence to: College of Aerospace Engineering, Chongqing University, Chongqing 400030, China.
E-mail address: wgli@cqu.edu.cn (W. Li).
https://doi.org/10.1016/j.ceramint.2017.12.194
Received 16 December 2017; Received in revised form 26 December 2017; Accepted 26 December 2017
0272-8842/ © 2017 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Please cite this article as: Deng, Y., Ceramics International (2017), https://doi.org/10.1016/j.ceramint.2017.12.194
Y. Deng et al. Ceramics International xxx (xxxx) xxx–xxx
models provide good approach for researching the TSR of composites where h is the heat transfer coefficient between quenching medium and
from the aspect of theory; however, they cannot predict the critical specimen; the coefficient a represents the characteristic dimension of
temperature difference of FRBMCs over a wide range of cooling en- specimen; k is the thermal conductivity of the specimen. Then, ac-
vironment temperatures. cording to the study by Kastritseas et al. [15], the thermal shock-in-
Based on our previous work [16–18], environmental temperature duced stress in matrix surface for FRBMCs, which considers the influ-
has significant influence on the thermal shock behavior of ceramic ence of the anisotropic thermal stress field produced during thermal
materials. Accordingly, the TSR is highly affected by cooling environ- shock, has been given by
ment temperature and material parameters which are temperature-de-
pendent. In particular, the failure criterion of thermal shock is strongly α1 (Ec1 + νc1 Ec22 α c 2/ Ec1 α c1 ) Em
σTS = C ΔT
dependent on temperature. Therefore, it is of great importance to (1 − νc21 Ec 2/ Ec1 ) Ec1 (4)
consider the effect of cooling environment temperature in character-
where Ec1 = Em Vm + Ef Vf , Ec 2 = Em Ef /(Em Vf + Ef Vm) , νc1 =νm Vm + νf Vf ,
ization of the TSR over a wide temperature range. Hence, the main
α c1 = αm Vm + αf Vf , α c 2 = (1 + νm) αm Vm + (1 +νf ) αf Vf − α c1 νc1
objective of this paper is to develop a thermal shock resistance model,
the subscripts, 1 and 2, refer to the directions parallel to fiber and
which can be used to predict the critical temperature difference of
perpendicular to fiber, respectively; the subscripts, c, m and f refer to
FRBMCs over a wide range of cooling environment temperatures.
the composites, matrix and fiber, respectively; V is the volume content.
In this study, a thermal shock resistance model over a wide range of
It is noted that the cooling environment temperature is usually
cooling environment temperatures considering the cooling environ-
much higher than room temperature [16,17], during service of high-
ment temperature and its effect on material parameters for FRBMCs is
temperature structure materials. Moreover, in high temperature en-
established. When room temperature is set as the cooling environment
vironment, the thermo-mechanical properties of materials are highly
temperature, the model predictions are compared with experimental
sensitive to temperature [10,16–18]. Considering the influence of
data available measured by water quenching test for some SiC fiber
temperature on material parameters, the thermal shock-induced stress
reinforced silicate composites, and reasonable agreement is obtained.
of FRBMCs at cooling environment temperature, T ′, should be:
Moreover, the critical temperature difference over a wide range of
cooling environment temperature for a Nicalon/Duran composite and a σTS (T ′) = Cα c1 (T ′) φ (T ′) ΔTc (T ′) (5)
SiC fiber reinforced reaction bonded Si3N4 composite (SiC/RBSN) are
Em (T ′)[Ec1 (T ′) + νc1 (T ′) Ec22 (T ′) α c 2 (T ′) / Ec1 (T ′) α c1 (T ′)]
predicted by the model. The dependency of TSR on some material where φ (T ′) = Ec1 (T ′)[1 − νc21 (T ′) Ec 2 (T ′) / Ec1 (T ′)]
parameters is also systematically analyzed using the model. And some T ′ is the cooling environment temperature, ΔTc (T ′) is the critical
approaches to improve the TSR over a wide temperature range of temperature difference corresponding to matrix cracking of composite
FRBMCs have been put forward. This study provides a quantitative tool at T ′; so TI = T ′ + ΔTc (T ′) is the initial thermal shock temperature,
for the design and prediction of TSR for FRBMCs at high temperatures. σTS (T ′) is the thermal shock-induced stress at T ′; α (T ′) , E (T ′) and ν (T ′)
are the thermal expansion coefficient, Young's modulus, and Poisson's
2. Theoretical model ratio at T ′, respectively.
Moreover, the temperature-dependent critical matrix cracking stress
When a fiber reinforced brittle matrix composite is subjected to a in composites, σcy (T ) , for unidirectional fiber reinforced ceramic com-
thermal gradient, the early damage due to thermal shock is in the form posites is given in our previous work [20]:
of matrix micro-cracks [14]. In general, the thermal shock-induced 1
T 3
stress corresponding to initial damage should be equal to the critical ⎡ 12τ (T ) γmo Ec2 (T ) Ef (T ) Vf2 ⎡ ∫T0 CP (T ) dT ⎤ ⎤
matrix cracking stress during thermal shock process [6,12]. To establish σcy (T ) = ⎢ 2 ⎢1 − Tm ⎥⎥
⎢ Em (T ) Vm R ⎢ ∫T0 CP (T ) dT ⎥ ⎥
the theoretical thermal shock resistance model over a wide range of ⎣ ⎣ ⎦⎦ (6)
cooling environment temperatures, a simple approach can be elabo-
rated as follows. where Ec (T ) and τ (T ) are the Young's modulus and interfacial shear
The matrix cracking occurs when the total of thermal shock-induced stress of composite at temperature T , respectively; γmo is the matrix
stress and residual thermal stress is equal to the critical matrix cracking fracture surface energy at an arbitrary reference temperature T0 ; R is the
stress, based on the approach of Blissett et al. [6] and Boccaccini [14]. fiber radius; Cp (T ) is the specific heat capacity of matrix for constant p
As the thermal shock-induced stress, residual thermal stress, and critical pressure and temperature T ; Tm is the melting point of matrix.
matrix cracking stress are very sensitive to temperature, the tempera- So let the temperature T in Eq. (6) equals to the cooling environ-
ture-dependent thermal shock failure criterion for FRBMCs should be ment temperature, T ′, during thermal shock, the critical matrix
cracking stress in matrix for FRBMCs at T ′ should be
σCM (T ′) = σTS (T ′) + σRTS (T ′) (1)
Em (T ′)
where T ′ is the cooling environment temperature, σCM (T ′) , σTS (T ′) and σCM (T ′) = σcy (T ′)
Ec1 (T ′)
σRTS (T ′) are the critical matrix cracking stress, thermal shock-induced 1
T′ 3
stress and residual thermal stress at T ′, respectively. ⎡ 6τ (T ′) Γmo Em (T ′) Ef (T ′) Vf2 ⎡ ∫T0 CP (T ) dT ⎤ ⎤
When a ceramic material is subjected to an abrupt temperature =⎢ ⎢1 − Tm ⎥⎥
⎢ Ec1 (T ′) Vm R ⎢ ∫T0 CP (T ) dT ⎥ ⎥
decrement ( ΔT ), the surface thermal shock-induced stress under the ⎣ ⎣ ⎦⎦ (7)
assumption of plane stress is [19]:
where σCM (T ′) is the critical matrix cracking stress in matrix at T ′;
Eα Γmo = 2γmo , is the fracture energy of matrix at reference temperature T0 ;
σTS = C ΔT
1−ν (2) Em (T ′) , Ef (T ′) and Ec1 (T ′) are the Young's modulus of matrix, fiber and
composites at T ′, respectively; τ (T ′) is the interfacial shear stress at T ′.
where σTS is the thermal shock-induced stress; E , α and ν are the
Then, it is known that the thermo-mechanical responses of FRBMCs
Young's modulus, thermal expansion coefficient and Poisson's ratio,
are strongly affected by the existence of residual thermal stress devel-
respectively; C is the thermal stress reduction factor depending on the
oped during composites processing, due to the thermal expansion
thermal shock conditions, which can be quantified by the Biot modulus,
β, given by [4] mismatch between fibers and matrix. Thus, the effect of residual
thermal stress on TSR must be considered. The residual thermal stress in
ah the composite as a function of temperature, σRTS (T ) , can be given as
β=
k (3) follow [20]
2
Y. Deng et al. Ceramics International xxx (xxxx) xxx–xxx
Ef (T ) Vf ϕt (T )
σRTS (T ) = A [αf (T )−αm (T )](T −Tproc )
1−νm (8)
0.5(1 + Ec1 (T ) / Ef (T ))
where ϕt (T ) = ⎡ 1 − 0.5(1 − 2ν )[1 − E (T ) / E (T )] / (1 − ν ) ⎤
⎣ c1 c1 f c1 ⎦
Tproc is the processing temperature of material; αf (T ) and αm (T ) are
the mean coefficients of thermal expansion for fiber and matrix from
temperature T to Tproc, respectively; νm and νc1 are the Poisson's ratio of
matrix and composite, respectively; A is the residual thermal stress
release factor, which characterizes the release of residual thermal stress
due to weak fiber/matrix interface and micro-flaws in composites [21].
And the value of A is less than one.
In Eq. (8), let the temperature T equals to the T ′, the residual
thermal stress in matrix as a function of cooling environment tem-
perature can be expressed as:
Ef (T ′) Em (T ′) Vf ϕt (T ′) Fig. 1. The predicted and experimental values of ΔTc by water quench test.
σRTS (T ′) = A [αf (T ′)−αm (T ′)](T ′−Tproc )
Ec1 (T ′)[1−νm (T ′)] (9)
0.5[1 + E (T ′) / E (T ′)]
c1 f necessary to do one experiment for determining the value of C when
where ϕt (T ′) = ⎡ 1 − 0.5[1 − 2ν (T ′)][1 − E (T ′) / E (T ′)] / [1 − ν (T ′)] ⎤
⎣ c1 c1 f c1 ⎦ predicting the ΔTc of FRBMCs over a wide temperature range using our
Finally, submitting Eqs. (5), (7) and (9) into Eq. (1), the thermal model. In addition, based on the residual thermal stress reported by
shock resistance model for FRBMCs over a wide range of cooling en- Beyerle et al. [22] and Eq. (8), the residual thermal stress release factor,
vironment temperatures is obtained A, is set as 0.232 for Nicalon fiber reinforced lithium aluminosilicate
(Nicalon/LAS). For Nicalon/Duran composites and Nicalon/Pyrex
ΔTc (T ′)
composites, the residual thermal stress is ignored due to the same
1
⎧ 6τ (T ′) Γmo Em (T ′) Ef (T ′) V 2 ∫TT ′ CP (T ) dT ⎤ ⎤3 ⎫ thermal expansion coefficient between fibers and matrix.
1 ⎪⎡ f ⎡
1 − T0m + ⎪ As can been seen from Fig. 1, our model predictions of ΔTc for Ni-
⎢ Ec1 (T ′) Vm R ⎢ ∫T0 CP (T ) dT ⎥ ⎥
= ⎣ ⎣ ⎦ ⎦
Cα1 (T ′) φ (T ′) ⎨ ⎬ calon/Duran, Nicalon/Pyrex, and Nicalon/LAS composites are in good
⎪ A Ef (T ′) Em (T ′) Vf ϕt (T ′) [α (T ′) − α (T ′)](T ′ − T ) ⎪ agreement with the experimental results by water quench test. For
E (T ′)[1 − ν (T ′)] f m proc
⎩ c1 m ⎭ Nicalon/LAS composites, the predicted data is larger than the experi-
(10) mental data, which is likely due to the effect of C. Meanwhile, the
where ΔTc (T ′) is the critical temperature difference corresponding to predictions by Kastritseas et al. [15] model are also shown in Fig. 1,
matrix cracking at cooling environment temperature T ′. their model predictions overestimated the ΔTc of all FRBMCs. Then,
The model not only provides an approach for predicting the TSR of Kastritseas et al. developed a modified Coulomb-type model with con-
FRBMCs over a wide temperature range, but also provides a technical sideration of thermal shock stress to estimate the interfacial shear stress
means for quantitatively estimating the effects of cooling environment [15], i.e
temperature and temperature-dependent material parameters on the
critical temperature difference corresponding to matrix cracking of
FRBMCs.
⎡CΔT ν12 E2 α1 + E2 α2 + −
⎢
τ = −μ ⎢
2 E /E
1 − ν12 2 1 ( Em Ef
Ef (1 + νm) + Em (1 − νf ) R)
Ar ⎤
⎥
Em Vm (αm − αf ) ⎥
⎢− 2[(1 − ν12) − 0.5(1 − 2ν12)(1 − Ec1 / E ) / (1 − ν12)] (Tproc − TI) ⎥
⎣ f
⎦ (11)
3. Results and discussion
where μ is the coefficient of friction between fiber and matrix, Ar is the
3.1. Model verification roughness amplitude of the fiber surface, TI is the initial thermal shock
temperature. This model provided an accurate approach to calculate
In order to verify the present thermal shock resistance model, the the interfacial shear stress, however, the TI in Eq. (11) relies on the
critical temperature difference, ΔTc, for some Nicalon fiber reinforced thermal shock experiment, and the value of μ and Ar of material are not
silicate composites (Nicalon/Duran composites, Nicalon/Pyrex com- easy to obtain.
posites, and Nicalon fiber reinforced lithium aluminosilicate) are pre- The modified interfacial shear stress calculated by Eq. (11) was also
dicted and compared with experimental results. Due to the lack of ex- used to predict the ΔTc by Kastritseas et al. In Fig. 1, their model pre-
perimental data at high cooling environment temperatures, the ΔTc for dictions using modified interfacial shear stress (C=0.66) under-
water quench test is used for comparison at room temperature. The estimated the ΔTc of all FRBMCs. Then, C was also set as 0.55 (the
relative material parameters in the calculations at normal temperature average limits to the value of C in Blissett et al. [6] and Boccaccini [14]
are shown in Table 1 [6,14,15,22–26]. In this work, the value of C in work) to calculate the ΔTc by Kastritseas et al., and Fig. 1 shows that
the model is set as 0.66, which was reported by Kastritseas et al. [15]. their modified model predictions (C=0.55) have a good consistency
The value of C may be different for other dissimilar FRBMCs, due to the with the ΔTc of all FRBMCs. But their model cannot be used to predict
influence of the thermal shock conditions (thermal conductivity, char- ΔTc over a wide range of cooling environment temperatures.
acteristic dimension and heat transfer coefficient). In that case, it is In addition, the ΔTc for Nicalon/Duran composites and SiC/RBSN
Table 1
The relative material parameters at room temperature [6,14,15,22–26].
Nicalon/Duran 63 190 0.6 0.4 0.66 7.5 0.2 0.2 5.2 3.3 3.3 1000
Nicalon/Pyrex 63 190 0.5 0.5 0.66 7.5 0.2 0.2 3.1 3.3 3.3 1000
Nicalon/LAS 83 190 0.6 0.4 0.66 30 0.2 0.3 1.4 0.9 3.3 1350
3
Y. Deng et al. Ceramics International xxx (xxxx) xxx–xxx
Table 2
The relative material parameters of SiC/RBSN composites and Nicalon/Duran composites [21,23,27,28].
Table 3
Material parameters for SiC/RBSN composites and Nicalon/Duran composites [20,21,29–34].
R(μm) 70 8
A 0.1878
νm 0.22 0.2
νf 0.22 0.2
Vm 0.741 0.6
Vf 0.259 0.4
o 72 7.5
Γm (J / m2 )
Tproc (°C) 1200 1000
Tm(°C) 1878 800
T0(°C) 24 24
C 0.66 0.66
αf (T ) (×10−6°C−1) 3.98 + 2.23 × 10−3 (T + 273) 3.19 + 3.6 × 10−3T − 1.68 × 10−6T2
αm (T )(×10−6°C−1) 1.5617 + 6 × 10−3T 1.123 + 1.607 × 0.01 × (273 + T )−
3.496 × 10−5 (273 + T )2 + 2.435 × 10−8 (273 + T )3
Cp (T )(cal/ mol) Cp (T ) = 76.337 + 109.039 × 10−3T Cp (T ) = 48.5 − 20 × 10−3T 24°C < T < 450°C
− 6.535 × 105T −2 − 27.083 × 10−6T2 − 10.9 × 105T −2 + 18.9 × 10−6T2
Cp (T ) = 67.1 + 12.1 × 10−3T 450°C < T < 574°C
− 30 × 105T −2 + 18.9 × 10−6T2
Cp (T ) = 79.1−10.8 × 10−3T 574°C < T < 800°C
− 22.3 × 105T −2 + 6.3 × 10−6T2
4
Y. Deng et al. Ceramics International xxx (xxxx) xxx–xxx
Fig. 4. The sensitivity of ΔTc to Young's modulus of matrix at different cooling en- Fig. 6. The sensitivity of ΔTc to interfacial shear stress at different cooling environment
vironment temperatures. temperatures.
5
Y. Deng et al. Ceramics International xxx (xxxx) xxx–xxx