Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
net/publication/281736670
CITATIONS READS
6 50
3 authors:
Mojtaba Yazdani
Sahand University of Technology
28 PUBLICATIONS 118 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Saeed Shiri on 16 February 2019.
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems MARCH 2016, Vol. 2 / 011001-1
Part B: Mechanical Engineering Copyright © 2015 by ASME
3 Proposed Model
2 Fatigue Damage
Based on the characteristics of damage evolution in the
Due to the complexity of the composite materials, their trend of composite materials, the rates of damage development are approx-
damage accumulation process is quite different with homogenous imately same in the first and final stage of fatigue life. Hence,
materials. As shown in Fig. 1, damage is accumulated nonlinearly in the relation between A and B is approximated as linear (more detail
composite materials. This process is divided into three main stages. is available in Ref. [7]). Herein, the parameters are modified as
The damage accumulates rapidly during the first few cycles. During follows:
this stage, microcracks initiate in multiple locations in the matrix.
Also, some fibers with low strength may break during this stage. log N
The next stage shows a slow and steady damage growth rate. This B¼C ð2Þ
ð1 − RÞðσmax =σr Þ
occurs when the crack density is saturated in the matrix. Finally, the
damage again grows rapidly. In brief, it is commonly known that
matrix cracking, delamination, and fiber breakage are the main A ¼ 0.85B þ 0.42 ð3Þ
dominant damage mechanisms in the explained stages [16,17].
Stiffness-based models are an appropriate approach for damage where σmax is the maximum stress, σr is the residual strength, C is a
development simulation of composites. The damage accumulation proportional constant, and R is the stress ratio (σmin =σmax ).
process can be quantified by measuring the Young’s modulus of the As seen, the residual strength is coupled to B and is approxi-
material. The proposed stiffness-based model by Wu and Yao [7] is mated by Eq. (4), which was originally proposed by Broutman
capable of describing the nonlinear damage evolution of composites and Sahu [20]
in all explained stages. The present paper is based on the following
stiffness-based model: n
σr ¼ σu − ðσu − σmax Þ ð4Þ
B A N
E − EðnÞ n
DðnÞ ¼ 0 ¼1− 1− ð1Þ
E0 − Ef N
where σu is the ultimate strength. Therefore, a combination model
where E0 is the static Young’s modulus; Ef is the failure Young’s of residual stiffness and strength is developed. It is a modified form
modulus; EðnÞ is the Young’s modulus of the material subjected to of the Wu and Yao model presented in Ref. [7]. The proposed model
has the advantage that it requires no estimation of additional param-
eters beyond those required for the choice model and considers the
degradation of two material properties simultaneously.
Based on the aforementioned model, the following equations are
proposed for life prediction of composites under two-stress level
loading
ni þ nki;i−1 Bavg Aavg
Dðni Þ ¼ 1 − 1 − ð5aÞ
Ni
ni−1 Bavg Aavg
ni;i−1 ¼ N i 1 − 1 − ð5bÞ
N i−1
DðnF Þ ¼ 1 ð6Þ
where ni and ni−1 are the number of applied cycles under ith and
(i − 1)th cyclic loadings; N i and N i−1 are the corresponding fatigue
Fig. 1 Sketch of fatigue damage development [1] lives, DðnF Þ is the critical damage, which causes fatigue failure
under Fth cyclic loading; and k is a constant to consider the effect of Table 2 Predicted remaining fatigue cycles by the modified
load sequence. It is set to 0.95 for the high-low sequence and 1.02 model for E-glass/epoxy [20]
for the low-high sequence; Aavg and Bavg are the average values of
parameters, which are defined as follows: Current
Experimental study
Ai þ Ai−1 σ1 σ2
Aavg ¼ ð7Þ
2 (MPa) (MPa) n1 n2 Wu and Yao n2
241 289 49,950 3730 11,619 (211.5) 5302 (42.14)
241 289 19,975 9490 13,877 (46.23) 8325 (12.27)
Bi þ Bi−1 241 337 49,950 391 2120 (442.2) 924 (136.33)
Bavg ¼ ð8Þ
2 241 337 19,975 804 2414 (200.25) 1405 (74.73)
241 386 19,975 124 490 (295.16) 281 (126.93)
289 337 10,000 293 846 (188.74) 222 (24.19)
where Ai , Ai−1 , Bi , and Bi−1 are the parameters under ith and (i − 1) 289 337 2000 1290 2276 (76.43) 1204 (6.69)
th fatigue loadings, respectively. 289 386 2000 355 476 (34.08) 241 (31.98)
337 386 1000 297 328 (10.44) 126 (57.68)
337 386 250 503 468 (6.96) 242 (51.93)
386 241 250 192,000 64,267 (66.53) 105,047 (45.29)
4 Numerical Implementation 386 241 100 193,000 100,320 (48.02) 121,218 (37.19)
386 289 250 5940 5944 (1.78) 8144 (39.44)
In order to validate the modified model and compare with 386 289 100 11,970 9605 (19.76) 9617 (19.66)
the original model, published data in Refs. [20,21] are used. 386 337 250 1250 1090 (12.8) 1254 (0.34)
Residual fatigue lives are calculated based on Eqs. (5–8). A pro- 386 337 100 1635 1784 (9.11) 1512 (7.49)
portional constant, C, is set to a value of 0.05 to achieve a good 337 241 1000 86,000 83,293 (3.15) 110,727 (28.75)
compatibility between the experimental values of model parame- 337 241 250 162,500 130,098 (19.94) 132,166 (18.67)
ters and those for two-stage test data. The results are presented in 337 289 1000 8670 7905 (8.82) 8678 (0.09)
337 289 250 8000 12,298 (53.73) 10,654 (33.17)
Tables 2 and 3. 289 241 10,000 96,500 52,292 (45.81) 97,968 (1.52)
The numerical results demonstrate that by applying the modi- 289 241 2000 110,800 135,150 (21.98) 131,181 (18.39)
fied model, the difference is decreased between the predicted
and experimental values for most cases. The percentage of relative Note: The numbers in parentheses indicate the percentage of relative
error clarifies the extent of improvement. Therefore, a combination error for each prediction.
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems MARCH 2016, Vol. 2 / 011001-3
Part B: Mechanical Engineering
Fig. 3 Weibull distribution to the histogram of predicted residual fatigue lives for E-glass/epoxy laminates [20]:
(a) Low-high loading sequence and (b) high-low loading sequence
Fig. 5 Reliability variation of predicted residual fatigue lives for E-glass/epoxy laminates [20]: (a) σ 1 241, σ 2 289,
n 1 49;950 and (b) σ 1 386, σ 2 337, n 1 100
Fig. 6 Reliability variation of predicted residual fatigue lives for E-glass/epoxy laminates [21]: (a) σ 1 315, σ 2 340,
n 1 86;300 and (b) σ 1 340, σ 2 315, n 1 1350
ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems MARCH 2016, Vol. 2 / 011001-5
Part B: Mechanical Engineering
DownloadedViewFrom:
publicationhttp://risk.asmedigitalcollection.asme.org/
stats on 07/21/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use