Professional Documents
Culture Documents
Aoki_et_al_fatigue_cfrp_2020
Aoki_et_al_fatigue_cfrp_2020
Aoki_et_al_fatigue_cfrp_2020
A R T I C L E I N F O A B S T R A C T
Keywords: This study proposes a simulation method to evaluate fatigue progressive damage in composite laminates using
Polymer matrix composites intra-laminar and inter-laminar damage models. The intra-laminar and inter-laminar damages are modeled using
Fatigue modelling continuum damage mechanics and cohesive zone model, respectively, and the evolution laws of damages are
Damage mechanics
formulated. A simulation method combining finite element method and fatigue models are described in this
Life prediction
paper. Fatigue simulations are conducted on a curved specimen and the results of that are compared with the
experimental results. The comparison indicates that the proposed method can predict the stiffness degradation
and the number of final failure cycles in good accuracy.
1. Introduction used to model transverse crack and delamination. On the other hand,
continuum damage mechanics (CDM) is a suitable way to model the
In recent years, fiber-reinforced composite materials such as CFRP intra-laminar damage by homogenizing it for evaluating the stiffness
(Carbon Fiber Reinforced Plastic) have been increasingly applied to loss.
primary structures of aircraft thanks to their excellent mechanical CDM was first proposed by Kachanov [8,9] to describe a creep
properties. Fatigue damage accumulates inside aircraft structures damage growth of metallic materials. Damage models for not only creep
because aircraft are subjected to repeated loading during long-term damage but also damage due to static and fatigue loading have been
operation. Since fatigue damage leads to a decrease in the strength proposed using CDM [10–24]. Ladeveze and Allix [10–12] applied a
and stiffness of composite structures, it is indispensable to appropriately CDM model for composite materials and formulated a damage evolution
evaluate the fatigue damage progress for the structural integrity of model. In the context of CDM, a damage state of materials can be
aircraft. Establishing of a numerical simulation method to evaluate the expressed using an internal state variable called damage variable, which
fatigue progressive damage has become an important issue for assessing describes the mechanical behavior and damage growth of materials by
the residual stiffness and strength and fatigue life. formulating its evolution law. It is an attractive approach for predicting
Many previous studies have investigated the damage mechanisms of the residual stiffness due to damage growth.
composites in fatigue and have shown that composite laminates exhibit As a method to predict the stiffness degradation of composite lami
complex damage growth processes under fatigue [1–6]. Reifsnider et al. nates under fatigue loading, many previous studies have proposed intra-
[1] proposed a three-stage stiffness degradation model to describe the laminar fatigue damage models based on CDM [13–24]. Hwang and Han
stiffness reduction process in composites, and Jamison et al. [2] clearly [13,14] introduced a concept of fatigue stiffness, which defined the
described the fatigue damage mechanisms in each stiffness degradation stiffness change in loading cycles as a function of applied stress and
process. In CFRP laminates, intra-laminar damage (including micro strain. Besides, Sidoroff and Subagio [15] proposed a d(D)/dN model in
cracking, interfacial debonding between fiber and resin, and transverse which the damage growth rate was formulated using a damage param
crack) and inter-laminar damage represented by delamination occur and eter D and an applied strain. Van Paepegem et al. proposed a phenom
progress interactively due to fatigue loading. In order to evaluate the enological fatigue model with a coupled residual stiffness and strength
fatigue damage process by numerical simulation, it is necessary to apply for glass-fiber woven composites [16–19]. In addition, Carrella-Payan
an appropriate numerical model for damage modes in each stage. As a et al. [20] extended the model proposed by Van Paepegem et al. for
numerical model, the cohesive zone model (CZM) approach [7] is widely multidirectional CFRP laminates and conducted fatigue simulation
https://doi.org/10.1016/j.ijfatigue.2020.106015
Received 4 August 2020; Received in revised form 17 October 2020; Accepted 22 October 2020
Available online 26 October 2020
0142-1123/© 2020 Elsevier Ltd. All rights reserved.
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
∂ed σ 2ij
Yij = = (1)
∂dij 2Eij0 (1 − dij )2
2
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
component when the damage variable dij reaches 1 under static loading. where Dmax is a maximum value of the damage variable stored at an
In the intra-laminar fatigue damage model, ck,ij (k = 1, 2, 3, ij = 11,
integration point of cohesive element and |Δ̃| is a norm of the normalized
12) are parameters that govern the rate of damage evolution. The pa
rameters c1,ij, c2,ij, determine the slope of stiffness reduction in the low relative displacement vector Δ̃ = {ΔI /ΔIc , ΔII /ΔIIc , ΔIII /ΔIIIc }T .
cycle range (Stage I) and parameters c3,ij determine the slope of the The damage variable D needs to increase monotonically to prevent
gradually increasing damage region (Stage II). The intra-laminar fatigue the recovery of inter-laminar damage. The maximum value of the
damage model intends to reproduce the stiffness reduction behavior in normalized relative displacement norm in loading history is stored as
Stage I and II. the damage variable at each integration point of cohesive elements.
Taking Eq. (3) as an example, the functionality of the parameters is Once the inter-laminar damage initiates, the inter-laminar strengths
described as follows. When damage variable d11 takes a value close monotonically decrease as the damage variable increases. When the
enough to 0, the exponential part in Eq. (3) is close to 1, and the second damage variable finally reaches 1 (D = 1), the cohesive element absorbs
term takes a value close enough to 0. Thus, the first parameter c1,11 energy equal to the fracture toughness and at the same time delamina
determines the initial rate of stiffness degradation because the coeffi tion is generated. Additionally, it should be noted that the initial value of
cient of the first term c1,11 becomes dominant. As the damage variable the inter-laminar damage variable D needs to be larger than 0 to avoid
increases to some extent, the exponential part becomes dominant, so the an infinite stiffness of cohesive element.
second parameter c2,11 determines the end region of Stage I. As the
damage variable further increases, the exponential part approaches zero 3.2. Definition of inter-laminar fatigue damage model
and the second term becomes dominant. It means that the third
parameter c3,11 controls the slop of Stage II. To model the fatigue propagation of inter-laminar damage, the d(D)/
dN model is also applied to the cohesive element. We utilized the fatigue
2.2. Stress-Strain constitutive law with intra-laminar damage evolution model for inter-laminar damage, proposed by Yashiro et al.
[31] and the evolution law is defined as follows:
Taking into account the decrease in stiffness, the stress-strain
d(D) F βm
constitutive law when the intra-laminar damage occurs is defined as = αm (12)
dN (1 − D)γm
follows:
( ) √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2 ( )2 ( )2̅
σ = M − 1 : C0 : M − T : ε (8) τI τII τIII
F= + + (13)
τIc τIIc τIIIc
where σ is a stress tensor, ε is a strain tensor, and C0 is an undamaged
orthotropic elastic modulus tensor. M is a damage effect tensor, which is where F is a dimensionless parameter that represents the stress state on
a 4th order symmetric tensor. The damage effect tensor is defined as the cohesive element. αm, βm, γ m are parameters that govern the crack
follows using the damage variables dij (written in matrix form by Voigt growth rate under mixed-mode conditions. These values are assumed to
notation) [20]. be linearly interpolated using the values in mode I and II.
⎡ ⎤
1 αII − αI
√ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅
⎢ 1− d ̅ 0 0 0 0 0 ⎥ αm = (Δm − ΔI ) + αI (14)
⎢ 11 ⎥ ΔII − ΔI
⎢ ⎥
⎢ 1 ⎥
⎢
⎢ 0 √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 0 0 0 0 ⎥
⎥ where Δm is defined as an equivalent relative displacement Δm =
⎢ 1 − d22 ⎥ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
〈ΔI 〉2 + Δ2II + Δ2III . αI, αII are the values in modes I and II, respectively.
⎢ ⎥
⎢ ⎥
M=⎢ 0 0 1 0 0 0 ⎥ (9)
⎢
⎢
⎥
⎥ βm and γm also take the same form as above.
⎢ 0 0 0 1 0 0 ⎥
⎢ ⎥
⎢ ⎥
⎢
⎢
0 0 0 0 1 0 ⎥
⎥ 4. Simulation method
⎣ 1 ⎦
0 0 0 0 0 √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
1 − d12 4.1. Procedure of fatigue simulation
Applying the above stress-strain constitutive law to solid elements,
In this study, it is assumed that the damage progress only when the
the stiffness reduction due to the intra-laminar damage can be simulated
maximum stress is applied in one cycle of the fatigue loading and the
in a finite element analysis (FEA).
minimum load of the fatigue loading has no effect on the damage
progress. We adopted a method of repeating the analysis at the
3. Inter-laminar fatigue damage model
maximum load of fatigue loading as illustrated in Fig. 2, which is based
on a concept proposed by Van Paepegem et al. [35] for reducing the
3.1. Constitutive law of cohesive element
computational cost.
Starting from the cycle number N = 1, a finite element analysis is
A cohesive element is used to simulate fatigue propagation of
delamination by using FEA. In this study, a newly defined damage
variable D representing the inter-laminar damage is used for the
following traction τi - relative displacement Δi relationship. It is based on
the cohesive element model proposed by Geubelle et al. [34].
1 − D τic
τi = Δi (i = I, II, III) (10)
D Δic
where τic, Δic are inter-laminar strengths and critical relative displace
ments in each fracture mode, respectively. The damage variable D is
defined using relative displacements.
3
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
4
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
Fig. 4. The results of parameter identification of intra-laminar fatigue damage model: (a) [0/90]8S, (b) [45/− 45]8S.
5
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
Fig. 6. The results of parameter identification of inter-laminar fatigue damage model: (a) DCB, (b) ENF.
6
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
5.1. Validation of intra- and inter-laminar fatigue damage model Secondly, no stiffness loss due to initial damage was observed in the
simulation using only the inter-laminar fatigue model, and the predicted
For validation, fatigue simulations were carried out using only the stiffness hardly decreased up to 400,000 cycles. It is thought that
intra-laminar fatigue model, using only the inter-laminar fatigue model, delamination did not progress because the simulation could not capture
and using both fatigue models. It should be noted that the same values the initial degradation of stiffness.
were used for material properties and fatigue parameters. The applied Finally, the results from the simulation using both the intra- and
maximum load of fatigue loading was 1,000 N, which was about 50% of inter-laminar fatigue models presented in this study confirmed that the
the static strength of the curved specimen. behavior of stiffness reduction due to initial damage, subsequent gradual
Fig. 8 shows the stiffness normalized by initial stiffness obtained stiffness reduction, and abrupt stiffness reduction was in good agree
from the simulations and experiments. The normalized stiffness is ment with the experimental results. The predicted stiffness value at
determined as the stiffness divided by the initial stiffness with no 300,000 cycles was 96.4% of the initial stiffness while the experimental
damage. In this study, the stiffness of the curved specimen is defined as values were 97.7% and 96.9%. Furthermore, regarding the number of
the applied load Pmax divided by the corresponding displacement δmax: cycles of final failure, the analysis results showed that the final failure
Stiffness = Pmax/δmax. Furthermore, in this study, the final failure of the occurred at 340,000 cycles, while the experimental results showed that
curved specimen is defined as when the stiffness decreases to 50% or the failure occurred at 344,000 cycles and 380,000 cycles. These results
more of the initial stiffness. The black solid line, black dashed line, and demonstrated that the proposed method can predict the stiffness value
red solid line in the figure indicate the results obtained from simulations and the number of final failure cycles with good accuracy.
using only the intra-laminar fatigue damage model, the only inter- In the case of the simulation using both fatigue models, the behavior
laminar fatigue damage model, and both fatigue models, respectively. of the stiffness degradation of CFRP laminates up to the final failure was
Firstly, the simulation result using only the intra-laminar fatigue successfully predicted while the drastic stiffness loss due to delamina
model shows good agreement with the experimental results regarding tion could not be reproduced in the case of the analysis using only the
the stiffness reduction behavior up to 50,000 cycles. However, the inter-laminar damage model. This suggests that the relative displace
predicted stiffness values from 50,000 to 300,000 cycles were over ments between the interlayers were expanded caused by the in-plane
estimated in comparison with the test results. Even though the simula stiffness loss and consequently the inter-laminar damage progressed.
tion was performed up to 400,000 cycles, no rapid decrease in stiffness In addition, Fig. 9 shows a comparison of the damage distribution d22 at
was observed because the inter-laminar fatigue damage was not 340,000 cycles in cases of using only intra-laminar damage model (a)
implemented. and using both fatigue models (b), and microscopic image at the final
Fig. 9. Comparison of damage distribution d22 between (a) in case of using only the intra-laminar damage model and (b) in case of using both fatigue models at
340,000 cycles and (c) failure mode in the experiments at 50% load level.
7
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
simulation using both fatigue models showed the best agreement with
the experimental results regarding the stiffness degradation and can
predict the number of final failure cycles at several load levels.
Furthermore, the comparison of damage distribution suggested that the
proposed method was able to reproduce the fatigue damage growth
where intra-laminar and inter-laminar damages interact on the numer
ical simulation.
Acknowledgement
Fig. 10. Prediction of the number of final failure cycles. This work was supported by JST-Mirai Program Grant Number
JPMJMI19C6, Japan, and Grant-in-Aid for JSPS Fellows Grant No.
failure with 50% load level condition (c). As shown in Fig. 9 (c), the final 20J13151 from Japan Society for the Promotion of Science (JSPS) .
fracture in the experiment is caused due to large-scale delamination and
the occurrence of intra-laminar crackings in 30, − 30, and 90 deg. layers
References
were observed. The simulated damage in Fig. 9 shows intra-laminar
matrix damage and corresponds to the intra-laminar cracking in 30 [1] Reifsnider KL, Jamison R. Fracture of fatigue-loaded composite laminates. Int J
deg. plies, which was observed on the edges of the 30 deg. outermost Fatigue 1982;4(4):187–97.
[2] Jamison R, Schulte K, Reifsnider K, Stinchcomb W. Characterization and analysis of
plies in the experiment. As can be seen from the figure, little damage
damage mechanisms in tension-tension fatigue of graphite/epoxy laminates. In:
distribution in the thickness direction was observed in case (a), while it Effects of Defects in Composite Materials, ASTM STP 836; 1984. p. 21–55.
can be confirmed that intra-laminar damage has progressed significantly [3] Talreja R. Fatigue of composite materials: damage mechanisms and fatigue-life
in the vicinity of the inter-laminar damage in case (b). Once delamina diagram. Proc R Soc Lond A Math Phys Sci 1981;378:461–75.
[4] Spearing SM, Beaumont PW. Fatigue damage mechanics of composite materials. I:
tion has occurred, stress relaxation occurs at the location where the Experimental measurement of damage and post-fatigue properties. Compos Sci
inter-laminar damage occurred, while higher in-plane stresses occur in Technol 1992;44:159–68.
the layers above and below it, resulting in higher intra-laminar damage [5] Fujii T, Amijima S, Okubo K. Microscopic fatigue processes in a plain-weave glass-
fibre composite. Compos Sci Technol 1993;49:327–33.
growth. It can be said that the damage progress in which the intra- and [6] Carraro PA, Maragoni L, Quaresimin M. Characterisation and analysis of transverse
inter-laminar damages interfere with each other was numerically crack-induced delamination in cross-ply composite laminates under fatigue
reproduced. loadings. Int J Fatigue 2019;129.
[7] Elices M, Guinea GV, Planas J. The cohesive zone model: advantages, limitations
and challenges. Eng Fract Mech 2002;69:137–63.
5.2. Prediction of final failure on several load levels [8] Kachanov M. Time of the rupture process under creep conditions. Izestiia Akademii
Nauk SSSR, Otedelenie Teckhnicheskikh Nauk 1958;8:26–31.
[9] Kachanov M. Elastic solids with many cracks: A simple method of analysis. Int J
As a validation for several load levels, fatigue simulations were Solids Struct 1987;23:23–43.
conducted at the load levels of 60% (1,150 N) and 65% (1,250 N) of the [10] Ladevèze P. A damage computational method for composite structures. Comput
static strength. The numbers of failure cycles were 75,000 cycles and Struct 1992;44:79–87.
[11] Ladeveze P, LeDantec E. Damage modelling of the elementary ply for laminated
21,000 cycles, respectively, and the prediction of the number of cycles of
composites. Compos Sci Technol 1992;43:257–67.
final failure was summarized with experimental results in Fig. 10. The [12] Allix O, Bahlouli N, Cluzel C, Perret L. Modelling and identification of temperature-
dotted lines are the trend lines for the final failure cycle of curved dependent mechanical behaviour of the elementary ply in carbon/epoxy laminates.
Compos Sci Technol 1996;56:883–8.
specimens. The red one shows the experimental results and the black one
[13] Hwang W, Han KS. Cumulative damage models and multi-stress fatigue life
shows the predicted results in Fig. 10. As shown in the figure, it can be prediction. J Compos Mater 1986;20:125–53.
said that the fatigue simulation by the present method can predict the [14] Hwang W, Han KS. Fatigue of composites—fatigue modulus concept and life
number of failure cycles at several load levels It is because the simula prediction. J Compos Mater 1986;20(2):154–65.
[15] Sidoroff F, Subagio B. Fatigue damage modelling of composite materials from
tion was able to capture the initiation of delamination by determining bending tests. Proceedings of 6th International Conference on Composite Materials,
the parameters in the inter-laminar fatigue damage model as to predict London, UK, July. 1987.
the onset numbers of crack growth at several energy conditions. [16] Van Paepegem W, Degrieck J, De Baets P. Finite element approach for modelling
fatigue damage in fibre-reinforced composite materials. Compos B Eng 2001;32:
575–88.
6. Conclusions [17] Van Paepegem W, Degrieck J. Coupled residual stiffness and strength model for
fatigue of fibre-reinforced composite materials. Compos Sci Technol 2002;62:
687–96.
Focusing on progressive damage in CFRP laminates under fatigue [18] Van Paepegem W, Degrieck J. Experimental set-up for and numerical modelling of
loading, we modeled intra- and inter-laminar damage based on CDM and bending fatigue experiments on plain woven glass/epoxy composites. Compos
CZM approaches, respectively. The fatigue evolution laws of the intra- Struct 2001;51:1–8.
[19] Van Paepegem W, Degrieck J. A new coupled approach of residual stiffness and
laminar damage variables and the inter-laminar damage variable were strength for fatigue of fibre-reinforced composites. Int J Fatigue 2002;24:747–62.
formulated as the damage growth rates defined by damage and stress [20] Carrella-Payan D, Magneville B, Hack M, Lequesne C, Naito T, Urushiyama Y,
states. A fatigue simulation method combining both fatigue models and Yamazaki W, Yokozeki T, Van Paepegem W. Implementation of fatigue model for
unidirectional laminate based on finite element analysis: theory and practice.
FEA was proposed adopting the cycle jump calculation. It was verified
Frattura ed Integrità Strutturale 2016;10:184–90.
that the intra-laminar fatigue damage model predicted the stiffness [21] Samareh-Mousavi SS, Mandegarian S, Taheri-Behrooz F. A nonlinear FE analysis to
reduction and the number of crack initiation cycles could be predicted at model progressive fatigue damage of cross-ply laminates under pin-loaded
several energy levels using the inter-laminar fatigue damage model. To conditions. Int J Fatigue 2019;119:290–301.
[22] Huang J, Pastor ML, Garnier C, Gong XJ. A new model for fatigue life prediction
validate the proposed method, three types of fatigue simulation were based on infrared thermography and degradation process for CFRP composite
carried out on the curved specimen. It was confirmed that the result of laminates. Int J Fatigue 2019;120:87–95.
8
R. Aoki et al. International Journal of Fatigue 143 (2021) 106015
[23] Sally O, Laurin F, Julien C, Desmorat R, Bouillon F. An efficient computational [32] Okabe T, Yashiro S. Numerical simulation for predicting fatigue damage progress
strategy of cycle-jumps dedicated to fatigue of composite structures. Int J Fatigue in notched CFRP laminates by using cohesive elements. J Solid Mech Mater Eng
2020;135. 2009;3:1202–11.
[24] Tamboura S, Laribi MA, Fitoussi J, Shirinbayan M, Bi RT, Tcharkhtchi A, Dali HB. [33] Yamaguchi T, Okabe T, Yashiro S. Fatigue simulation for titanium/CFRP hybrid
Damage and fatigue life prediction of short fiber reinforced composites submitted laminates using cohesive elements. Compos Sci Technol 2009;69:1968–73.
to variable temperature loading: Application to Sheet Molding Compound [34] Geubelle PH, Baylor JS. Impact-induced delamination of composites: a 2D
composites. Int J Fatigue 2020;138. simulation. Compos B Eng 1998;29:589–602.
[25] Turon A, Costa J, Camanho PP, Dávila CG. Simulation of delamination in [35] Van Paepegem W, Degrieck J. Fatigue degradation modelling of plain woven glass/
composites under high-cycle fatigue. Compos A Appl Sci Manuf 2007;38:2270–82. epoxy composites. Compos A Appl Sci Manuf 2001;32:1433–41.
[26] Harper PW, Hallett SR. A fatigue degradation law for cohesive interface elements – [36] MSC Marc Documentation. Volume D: User Subroutines and Special Routines. MSC
Development and application to composite materials. Int J Fatigue 2010;32: Software; 2017.
1774–87. [37] ASTM D6115-97. Standard Test Method for Mode I Fatigue Delamination Growth
[27] Landry B, LaPlante G. Modeling delamination growth in composites under fatigue Onset of Unidirectional Fiber-Reinforced Polymer Matrix Composites. American
loadings of varying amplitudes. Compos B Eng 2012;43:533–41. Society for Testing and Materials, ASTM international; 2011.
[28] Kawashita LF, Hallett SR. A crack tip tracking algorithm for cohesive interface [38] Murri GB. Evaluation of Delamination onset and growth characterization methods
element analysis of fatigue delamination propagation in composite materials. Int J under Mode I fatigue loading. NASA TM-2013-217966 2013.
Solids Struct 2012;49:2898–913. [39] O’Brien TK, Johston WM, Toland GJ. Mode II Interlaminar Fracture Toughness and
[29] Tao C, Qiu J, Yao W, Ji H. A novel method for fatigue delamination simulation in Fatigue Characterization of a Graphite Epoxy Composite Materials. NASA TM-
composite laminates. Compos Sci Technol 2016;128:104–15. 2010-216838 2010.
[30] Kaushik V, Ghosh A. Experimental and numerical investigation of Mode-I & Mode- [40] ASTM D5528-13, Standard Test Method for Mode I Interlaminar Fracture
II fatigue crack growth in unidirectional composites using XIGA-CZM approach. Int Toughness of Unidirectional Fiber-Reinforced Polymer Matrix Composites.
J Fatigue 2020;134:105461. American Society for Testing and Materials, ASTM international; 2011.
[31] Yashiro S, Okabe T. Numerical prediction of fatigue damage progress in holed [41] ASTM D7905-14, Standard Test Method for Determination of the Mode II
CFRP laminates using cohesive elements. J Solid Mech Mater Eng 2009;3:1212–21. Interlaminar Fracutre Toughness of Unidirectional Fiber-Reinforced Polymer
Matrix Composites. American Society for Testing and Materials, ASTM
international; 2011.