thermal shock resistance of laminated ceramic matrix composites

1、JOURNAL OF MATERIALS SCIENCE 26 (1991) 2961--2966 Thermal shock resistance of laminated ceramic matrix composites Y. R. WANG, T.-W. CHOU Center for Composite Materials and Department of Mechanical Engineering, Universit

2、y of Delaware, Newark, DE 19716, USA The thermal shock resistance capability of laminated ceramic matrix composites is investigated through the study of three-dimensional transient thermal stresses and laminate failure

3、 mechanisms. A ( - 45~176 SiC/borosilicate glass laminate is utilized as a reference composite system to demonstrate the analytical results. The maximum allowable temperature change, A Tmax, has been taken as a measur

4、e of the thermal shock resistance capability of composites. The effects of fibre orientation, volume fraction, thermal expansion coefficient, Young's modulus, and thermal conductivity on the thermal shock resistanc

5、e capability, expressed in terms of the maximum allowable temperature change, A Tmax, have been assessed. Numerical computations are also performed for six composite systems. 1. Introduction Ceramics and ceramic matri

6、x composites have dem- onstrated the desirable characteristics of high-temper- ature strength, and resistance to creep and corrosion. But they also display an unfavourable property, namely, brittleness or notch-sensi

7、tivity, which can render such materials highly susceptible to cata- strophic failure under thermal shock. The thermal shock resistance capability of mono- lithic ceramics has been studied since the 1950s. Cheng [1], i

8、n 1951, first demonstrated that the thermal shock resistance of ceramics can be quantified by analysing the nonsteady state thermal stresses in the material. The thermal shock resistance parameter, R ,-~ (1 - v)~K/~E

9、, for the ceramics was proposed; where is the tensile strength, v Poisson's ratio, K the thermal conductivity, = the thermal expansion coefficient, and E Young's modulus. Kingery [2] reported in 1955, that th

10、e thermal shock resistance capability is not an intrinsic material property, and it depends on the manner in which heat is applied and the geometrical shape of the specimen. Buessem [3] concluded from the experimenta

11、l investigations that thermal shock tests usually do not lead to useful experimental data. This is because the synergistic effects of all the material properties on the thermal shock resistance, and the influence of

12、a single property cannot be readily quan- tified from the test data. There is the need of standard tests for identifying the individual material property effects. More recent interests in thermal shock resistance prob

13、lems of ceramics and ceramic matrix composites have been developed since 1980. Singh et al. [4] first employed heat transfer theory to analyse thermal stress fracture of ceramics subjected to cooling by quenching int

14、o fluid media. In addition to specimen size and geometry, and thermal convectivity, the effect 0022-2461/91 $03.00 + .12 ?9 1991 Chapman and Hall Ltd. of specimen density on thermal shock resistance was also introduce

15、d. Lewis [5] demonstrated the appreci- able disagreement between the water-quench thermal shock test data, A To, and the calculated thermal shock resistance parameter, R. Results showed that the quenching experiments

16、 do not appear to be very suitable for a quantitative assessment of thermal shock resistance. Becher et al. [6] studied two general thermal shock resistance parameters; one relates the material properties to its resi

17、stance to crack initiation under thermal stress and the other relates the proper- ties to the ability of the material to retain its strength in a thermal stress situation where crack initiation is unavoidable. Faber

18、et al. [7], in 1981, proposed a novel thermal shock resistance testing and evaluation procedure intended to alleviate the difficulties associ- ated with the variable and very high heat transfer rates and edge effects

19、 in testing, and to provide quantitative predictions of thermally induced failure. Thomas et al. and Singh et al. [8, 9-1, in 1981, performed theoretical investigations of thermal shock resistance capability of ceramic

20、s induced by convect- ive and radiative heat transfer. Oguma et al. [10-1, in 1986, compared the predicted and experimental beha- viour of thermal stress resistance of circular rod speci- mens of a sodaqime-silica gl

21、ass and polycrystalline alumina. The comparison indicates that the thermal stress fracture of the specimens is caused by conduct- ive heat transfer. Tiegs and Becher [11-1 performed the thermal shock testing of an al

22、umina-20 vol % SiC whisker composite. There is no decrease in the com- posite flexural strength with AT 400 ~ The improvement in per- formance is due to the increase in fracture toughness due to fibre reinforcement.

23、 Orenstein and Green [12] measured the thermal shock resistance of cellular 2961 The above approaches demonstrate that two basic types of information are necessary to assess the mater- ial thermal shock resistance capa

24、bility. The first is the thermal stress analysis, such as Equation 3; the second is the material failure criterion with reliable strength values. The introduction of the dimensionless stress parameter, c~*, in Equati

25、on 4 indicates that the max- imum tensile stress failure criterion has been adopted. Thus, the strength as well as thermo-elastic properties of a material are necessary for thermal shock resist- ance predictions. 2.2

26、. Thermal shock resistance of composite materials Just as in the case of homogeneous and isotropic materials, the evaluation of thermal shock resistance of composite materials requires the accurate thermal stress anal

27、ysis. Considerable effort has been made to investigate the steady state and transient thermal stresses of unidirectional and laminated composites [16-19]. It is clear that due to the mathematical complexities, therma

28、l stresses in fibre composites in- duced by temperature gradients can be expressed in explicit forms only in very limited cases. The second requirement for determining the ther- mal shock resistance of composites is th

29、e failure ana- lysis. It has been found that the failure of ceramic matrix composites is much more complicated than that of monolithic ceramics [20]. The failure of lamin- ated composites could be classified as intra

30、laminar and interlaminar. A commonly observed failure mode in composites is delamination initiated at geometric boundaries such as voids, microcracks, free edges, ply drop-off, co-cured joints or bolted joints. Delam

31、ina- tion by itself cannot lead to final failure, in-plane fracture must occur for the specimen to lose its load- carrying capability. In order to simplify the analysis of thermal shock resistance of laminated composi

32、tes, this paper focuses on the failure initiation (first-ply failure) predicted by the failure criteria. The failure development and ulti- mate fracture are not considered. The relationship between thermal shock resis

33、tance capability and the structural and material character- istics of high-temperature composites cannot be ex- pressed explicitly in a simple form as Equations 6 and 8 for isotropic materials. Therefore, for the purp

34、ose of quantitative studies, the maximum allowable temper- ature change, A Tmax, without causing failure, has been taken as a measure of thermal shock resistance of fibre composites. The transient thermal stress anal

35、ysis given previously [19] for laminated composites is utilized. Two initial failure mechanisms for high-temper- ature composites are considered in the present ana- lysis: (1) delamination within the boundary layer re

36、gion due to the concentration of boundary layer stresses, and (2) matrix micro-cracking induced by in-plane tensile stress. The consideration of the first failure mechanism is consistent with findings in [19]. The se

37、cond failure mechanism is motivated by the low-toughness/high-strength characteristics of ceramic matrix composites; matrix micro-cracking has been extensively documented in the literature [21-24]. The mathematical for

38、ms of the criteria for initial failure corresponding to the above mechanisms are = interlaminar tensile strength (9) (l) % and (2) %, = matrix micro-cracking yield stress (10) where o~ is the laminate boundary layer no

39、rmal stress in the thickness direction, and ~m is matrix tensile stress. Thus in the present analysis, the term “failure“ is defined by the state of local stress which attends a strength value of the composites. The

40、stresses close to the laminate boundary display a high concentration, and they may be 5-30 times larger than those away from the boundary (but still within the boundary layer region) as demonstrated in [19]. Thus, for

41、 applying the failure criteria, both the average stresses and peak stresses within the boundary layer region are compared with the strength data. This will be explained further later. The matrix micro- cracking stren

42、gth, O'mcy , is either obtained from the literature or estimated from the following relationship. (3“racy = Egmcy (1 1) where crecy is the matrix micro-cracking yield strain and E is the composite Young's mod

43、ulus. The matrix micro-cracking yield strain of ceramic matrix com- posites usually ranges from 0.1%-0.2% [25]. The other necessary strength data are calculated from the fibre and matrix strength properties by rule-o

44、f- mixtures [261. 3. Numerical examples A four layer ( - 0/0)s SiC/borosilicate glass (BG) lam- inate is used as a baseline composite system for para- metric studies, Fig. 1. The laminate is of thickness 2h (20 ram)

45、and width 2b (400 mm); it is infinite in extent along the x-direction. The laminate is uniformly heated at time t = 0 + along the surfaces y = + b. The numerical computations of thermal shock resistance capability ar

46、e also performed for five other composite systems (Table I). The constituent fibre and matrix thermoelastic properties can be easily found in the literature [19, 25, 26], and the strength data of these fibres and mat

47、rices are given in Table I. The thermally induced inplane stresses (~x and oy) and interlaminar normal stress (%) of a ( - 45~176 SiC/BG laminate with 30% fibre volume fraction are depicted in Fig. 2. These stresses ar

48、e within the lam- inate boundary layer region and calculated from [19]. The shear stresses are much smaller than the normal stresses. Fig. 2 shows the highly localized stress con- centrations of % and %. The peak val

49、ues (at Y = 1.0) of % and oz are 4.5 and 1.5 MPa, respectively, cyy tends to zero when approaching the boundary as the stress boundary condition requires, and the maximum value (at Y = 0.97) is 0.34 MPa. Because the