# Computational structural analysis of composites with spectral-based stochastic multi-scale method

- 383 Downloads

## Abstract

Composite materials and structures may be characterized at different length scales, ranging from the micro-scale at the fiber–matrix level, meso-scale at the lamina–laminate level, to structural macro-scale. However, uncertainties in material properties and geometric parameters due to manufacturing, defects and assembly processes may occur at various length scales. This paper presents a computational framework for stochastic analysis of composites with consideration of stochastic parameters at micro- and meso-scales. The novelty of the proposed framework is the integration of the spectral stochastic finite element method and asymptotic homogenization method within a finite element technique, which was implemented through \(\hbox {ABAQUS}^{\circledR }\). This spectral stochastic homogenization method efficiently predicts the propagation of uncertainties from the constituent to ply levels. The derived probability distributions of effective properties were verified by Monte Carlo simulation. Another novelty is the study of the influence of stochastic parameters at both micro-scale and meso-scale on the failure prediction of composite structures, without assumptions of probabilistic characteristic of ply properties commonly used in a single-scale stochastic analysis. The up-scaled uncertainties combined with other randomness at meso-scale (strength properties and ply orientations) are provided as the input of meso-scale stochastic strength analysis of a quasi-isotropic laminate based on classical lamination theory (CLT) and ply discount. The probability distribution of first-ply failure and ultimate failure loads are obtained and their sensitivity factors with respect to input variations are presented.

## Keywords

Stochastic homogenization Probabilistic strength Polynomial chaos Spectral stochastic method Monte Carlo simulation## 1 Introduction

As shown in Fig. 1, typical stochastic responses of composite structures at macro-scale include displacement, stress, strength and natural frequency, which can be obtained based on ply properties directly or effective properties derived from micro- scales (Sriramula and Chryssanthopoulos 2009). Strength analysis is of special interest due to the requirement of structural integrity and the complicated damage patterns of composites at both micro- and meso-scales, as indicated in Fig. 1. The study of failure mechanism in composites could be enhanced with consideration of stochastic factors, such as stochastic material properties (e.g., strength) and geometric randomness (e.g., fiber distribution) in the simulation of intralaminar damage, including fiber breakage (Swolfs et al. 2016; St-Pierre et al. 2017; Pimenta and Pinho 2013; Pimenta 2017) and transverse matrix cracking (Vinogradov and Hashin 2005; Andersons et al. 2008). Although it is ideal to consider each failure patterns faithfully to achieve high-fidelity simulations, nonlinear computational models involving damage processes are too computationally intensive in design applications, where multiple design variables are involved. When reliability or risk of failure of composite structures (Chiachio et al. 2012) is required to be assessed, efficiency becomes particularly important and simple analytical models or linear finite element analysis are preferred.

Furthermore, stochastic finite element methods (SFEM) have been employed to solve static and dynamic problems (Stefanou 2009; Arregui-Mena et al. 2016). The Monte Carlo simulation (MCS) is a general and direct approach for generating parametric and geometric randomness. Since the solution of large number of MC simulations via sample problems is not computationally viable, various metamodel or surrogate models, such as response surface methodology, artificial neural network and kriging (Dey et al. 2017), are alternative approaches. If only parametric uncertainties are considered, the perturbation method and the spectral stochastic method (Stefanou 2009) may be employed. The perturbation method characterizes the random feature by the Taylor series expansion of stochastic FE matrix and response vectors. It is relatively efficient compared to MCS and useful in estimating the statistical moments of stochastic response. However, the nature of Taylor series expansion means it is only applicable to problems with small random variations. The spectral stochastic method introduced by Ghanem and Spanos (2003) involves not only the discretization with FE in spatial dimension but also the approximation with polynomial chaos (PC) expansion in stochastic space, the error of which is minimized through the Galerkin approach. The advantages of this method are that only one deterministic equation needs to be solved and the whole probabilistic structure of quantities of interest can then be obtained.

Early work on computational stochastic analysis of composite structures was mostly conducted at ply level. Vinckenroy and Wilde (1995) adopted MCS combined with FE modeling and evaluated the variation of maximum stress in an open-hole composite plate due to randomness in material properties (\(E_1\), \(E_2\), \(v_{12}\)) and geometric parameters (hole size and position). Probabilistic strength analysis was developed by Jeong and Shenoi (2000) based on MCS and first-ply failure analysis with applied load, elastic moduli, geometric and strength values assumed as random variables following normal and Weibull probability distributions. Spatial variation of strength properties was considered by Wu et al. (2000) who used a random field model with MCS and FPF based on Tsai–Hill and Tsai–Wu criteria. The importance of including the variability of material elastic properties in reliability analysis of laminates was demonstrated by Lekou and Philippidis (2008) and an overestimation of reliability could be caused without considering these variabilities. Sánchez-Heres et al. (2014) further investigated the influence of mechanical models for laminate failure and probabilistic models for ply properties on reliability estimation of cross-ply laminates. Among all the factors, the definition and modeling of matrix cracking in mechanical models (failure criteria and ply discounts) and the choice of probability distribution (normal, Weibull and log-normal distributions) have comparatively large influence. Other applications in composites with MCS or related metamodels include strength analysis with spatial variation of loading position (Karsh et al. 2018), buckling analysis (Kepple et al. 2015), free vibration analysis (Dey et al. 2015; Chakraborty et al. 2016) and strength analysis of joints (Vijaya Kumar and Bhat 2015; Zhao et al. 2017).

First-order perturbation technique was adopted by Onkar et al. (2007) to improve the efficacy in predicting the statistics of the first-ply failure load of composite plates with random material properties subject to uniform distributed random load. Lal et al. (2009, 2011) studied the effect of random system properties such as material properties and laminate thickness on thermal buckling load and post buckling load of composite plates using higher order shear deformation plate theory and first-order perturbation technique. While stiffnesses are usually taken as random variables, Noh and Park (2011) focused on the spatial randomness of Poisson’s ratio and statistical moments of displacement were calculated with first-order Taylor-series expansion. However, the computed variation in response is not significant (coefficient of variation is 1.0–6.0%, the input variation is 0.1), which suggests the variation in Poisson’s ratio may not be critical. The correlation among random variables for composite structures (e.g., \(E_{11}\) and \(E_{22}\)) was considered in a recent work based on the combination of the Perturbation method and so-called Copula function (a tie function linking the marginal cumulative distribution with joint cumulative distribution), which mitigates the defects of traditional transformation method in terms of efficiency, accuracy and application (Cui et al. 2017). The spectral stochastic method has also attracted many applications in modeling composite structures. Chung et al. (2005) implemented a spectral stochastic version of solid-like shell element within an object-oriented computational framework and applied it to fiber metal laminates with the material properties of glass fiber epoxy layers treated as independent random field models. Ngah and Young (2007) compared the accuracy of the perturbation method and the spectral stochastic method in the stochastic analysis of a rectangular unidirectional composite panel with structural stiffness assumed as a Gaussian random field. It shows that the former method is only efficient for low material variability (coefficient of variation up to 10%) while the later method can achieve accurate results with a comparatively large material variability (24%). Chen and Soares (2008) extended the application of spectral method to multi-layer laminates and various moduli (\(E_{1}\), \(G_{12}\), \(G_{23}\), \(G_{13}\)) were modeled as independent random fields. The accuracy and efficiency of this developed formulation were benchmarked through MCS. Other recent applications include free vibration analysis (Sepahvand 2016), aeroelastic response analysis (Scarth et al. 2014) and stochastic analysis of composite structures with both normal random and interval variables (Chen and Qiu 2018).

In meso-scale stochastic analysis, assumptions regarding the probabilistic characteristics of material properties are often difficult to justify (Charmpis et al. 2007). Charmpis et al. (2007) proposes that macro-properties be derived from micro-mechanical stochastic information. Therefore, only the stochastic constituent properties need to be quantified. By casting simple analytical micro-mechanical model in a probabilistic framework, Shaw et al. (2010) derived macro-level statistics for reliability analysis of composites with MCS. Stochastic fiber and matrix properties as well as volume fraction were also considered by Li et al. (2016) in the stochastic thermal buckling analysis of laminated plates using perturbation method. Other analytical homogenization methods include self-consistent model and differential model explored by Ma et al. (2011). Although stochastic analytical homogenization method is simple and fast due to the closed formulations, the simplified assumptions limit their applications. Hence, stochastic computational homogenization has received more attention recently. Perturbation-based stochastic finite element method was first developed by Kamiński and Kleiber (2000). Sakata et al. (2008) extended the perturbation-based stochastic homogenization to detailed three-dimensional analysis and considered randomness in Young’s modulus and Poisson’s ratio of constituent materials. They conclude that the perturbation-based method does not perform well for nonlinear stochastic problems with large variations. A recently developed computational homogenization method Perić et al. (2011) was implemented with a second-order perturbation technique by Zhou et al. (2016a, b) for the stochastic homogenization of unidirectional composites and woven textile composites. The multi-scale uncertainty issue was later addressed in reliability analyses for composite structures (Zhou et al. 2016c, 2017). However, besides being inaccurate in applications with large variation, the perturbation-based analysis is unable to predict probability distribution. The spectral stochastic method is promising in solving this problem but related research work still remains few (Tootkaboni and Graham-Brady 2010). Other recent work includes the application of artificial neural networks (Balokas et al. 2017) and interval method (Chen et al. 2017).

The above review has shown that although numerous work has been conducted for stochastic analysis of composites based on randomness at meso-scale with various methods, there has been more recent work exploring randomness at the micro-scale level. However, work on uncertainty propagation from micro-properties to macro-response still remains few, with the exception of recent work by Zhou et al. (2016c, 2017) based on the perturbation method. Considering the aforementioned advantages of the spectral stochastic method, we propose a computational framework for spectral stochastic multi-scale analysis of composites structures. The structure of this paper is organized as follows. In Sect. 2, the formulation of spectral-based stochastic multi-scale methods is presented for both homogenization and macro-analysis. The numerical implementation is detailed in Sect. 3. Section 4 shows the predicted stochastic outputs including stochastic effective properties and stochastic macro-behavior. Conclusions are drawn in Sect. 5.

## 2 Spectral-based stochastic multi-scale methods

### 2.1 Stochastic formulation of asymptotic homogenization

*h*denotes homogenization.

*Y*. Equation (8) is usually solved by FEM, the matrix form of which is given by:where \(n_{ele}\) is the number of elements used for discretizing the RVE and \(\varvec{B}^e\) is strain displacement matrix.

### 2.2 Spectral stochastic homogenization method

*p*. Equation (12) may be truncated after the

*p*th order polynomial chaos terms and expressed in a compact form

### 2.3 Stochastic failure analysis of laminates

*n*layers of fiber orientation \(\theta _i(\varvec{\eta }), i=1,2,\ldots ,n\), where \(\varvec{\eta }\) denotes the randomness due to fiber misalignment at ply level. In Fig. 3, \(\varvec{N} = \{ N_x,N_y,N_{xy} \}\), \(\varvec{M} = \{ M_x,M_y,M_{xy} \}\) are resultant forces and moments per unit length applied along the edge. Based on Kirchhoff’s hypothesis, the displacements can be parametrized by a translation and a rotation. The laminate strain in the global coordinate

*x*–

*y*–

*z*is expressed as \(\varvec{\epsilon } = \varvec{\epsilon }^0 + z \varvec{\kappa }\), where \(\varvec{\epsilon }^0\) and \(\varvec{\kappa }\) are midplane strains and curvatures (Kaw 2005). For the

*i*th layer, the strain in the local material coordinate 1–2–

*z*is \(\varvec{\epsilon }_{loc} = \varvec{T}_1 (\varvec{\eta }) \varvec{\epsilon }\),

*i*th layer, the stress–strain relationship in global coordinates is given as:

*maximum stress*failure theory

*Tsai*–

*Wu*failure theory

*S*is shear strength in the 1–2 plane. As the laminate is composed of multiple layers, the failure process is progressive and involves stress redistribution. Although the failure mechanisms are generally complex, a conservative approach of ply discounts of failed plies are often used in design from first-ply failure (FPF) to ultimate failure (UF). The FPF strength is the load at which the initiation of failure in the first ply, determined by the failure criterion. In the following stress analysis, the stiffness matrix of a partially failed layer is partially discounted as:

## 3 Numerical implementation

## 4 Results and discussion

In this section, a quasi-isotropic AS4/3501-6 laminate with lay-up of \([90^\mathrm{o}/\pm 45^\mathrm{o}/0^\mathrm{o}]_s\) has been analyzed with uncertainties at both micro- and meso-scales. Based on the aforementioned methods, the probabilistic characteristics of effective properties of the lamina and strength values of the laminate are calculated.

### 4.1 Stochastic effective properties

^{TM}i5-2400 CPU @ 3.10 GHz, 8 GB memory) while that for the spectral-based method with up to third-order terms is 6.74 s, which illustrates the efficiency of spectral stochastic homogenization method.

Polynomial chaos from order 1 to 3 (\({\varPsi }_0 = 1\))

| | \({\varPsi }_j (\xi _1, \xi _2)\) |
---|---|---|

1 | 1–2 | \(\xi _1\), \(\xi _2\) |

2 | 3–5 | \(\xi _1^2-1\), \(\xi _1 \xi _2\), \(\xi _2^2-1\) |

3 | 6–9 | \(\xi _1^3-3\xi _1\), \(\xi _2(\xi _1^2-1)\), \(\xi _1(\xi _2^2-1)\), \(\xi _2^3-3\xi _2\) |

*X*,

*Y*,

*Z*) and shear in each of the three planes \((X-Y, Y-Z, X-Z)\). Having obtained \(\varvec{X}_j\), the coefficients in the expansion of stochastic homogenized stiffness matrix can be calculated through Eq. (18). Through post-processing, the probability distribution of effective properties is plotted in Fig. 7 and compared with the results by MCS. It can be seen that high accuracy is achieved with the new method for both cases (\(C_v=0.1\), 0.2) when second-order and third-order PC expansions are adopted. Only very small differences from the MCS results are found for the Poisson’s ratios \(v_{12}\) and \(v_{23}\) with \(p=1\). The reason is that the first-order PC expansion in Table 1 is only capable of reproducing a linear combination of two Gaussian random variables. The statistics including mean value and standard deviation for the case with large variation are listed in Table 2 and comparisons are shown in Fig. 8. The coefficient of variation in \(E_1\), \(E_2\), \(G_{12}\) and \(G_{23}\) are all smaller than the input variation (0.2) and the variations in \(v_{12}\) and \(v_{23}\) are extremely small, which illustrates that the randomness considered in stiffness matrix (due to moduli) hardly propagates to effective Poisson’s ratio. Similarly, only the standard deviation in Poisson’s ratio with \(p=1\) deviates from MCS results.

Statistics of effective properties by MCS (\(C_v=0.2\), moduli in GPa)

Properties | Mean | STD | COV |
---|---|---|---|

\(E_1\) | 136.409 | 26.911 | 0.197 |

\(E_2\) | 9.358 | 1.380 | 0.148 |

\(G_{12}\) | 4.598 | 0.724 | 0.157 |

\(G_{23}\) | 3.004 | 0.473 | 0.157 |

\(v_{12}\) | 0.253 | 0.002 | 0.008 |

\(v_{23}\) | 0.258 | 0.001 | 0.005 |

Correlation of effective properties from results (\(C_v=0.2\)) by MCS and spectral-based method with \(p=2\) (MCS-upper right triangular, spectral-lower left triangular)

Properties | \(E_1\) | \(E_2\) | \(G_{12}\) | \(G_{23}\) | \(v_{12}\) | \(v_{23}\) |
---|---|---|---|---|---|---|

\(E_1\) | 0.736 | 0.407 | 0.408 | −0.706 | 0.417 | |

\(E_2\) | 0.738 | 0.915 | 0.916 | −0.109 | 0.344 | |

\(G_{12}\) | 0.404 | 0.914 | 1.000 | 0.287 | 0.157 | |

\(G_{23}\) | 0.405 | 0.914 | 1.000 | 0.286 | 0.159 | |

\(v_{12}\) | −0.708 | −0.112 | 0.291 | 0.290 | −0.566 | |

\(v_{23}\) | 0.443 | 0.330 | 0.120 | 0.121 | −0.608 |

Random variable | Mean value (\(\mu \)) | Standard deviation (\(\sigma \)) | Distribution | |
---|---|---|---|---|

Strength | \(X_T\) | \(1969 \, \mathrm{MPa}\) | \(0.1\mu \) | Log-normal |

\(X_C\) | 1480 MPa | \(0.12\mu \) | Log-normal | |

\(Y_T\) | 48 MPa | \(0.06\mu \) | Log-normal | |

\(Y_C\) | 200 MPa | \(0.08\mu \) | Log-normal | |

| 79 MPa | \(0.11\mu \) | Log-normal | |

Ply misalignment | \(\hat{\theta }\) | \(0^\mathrm{o}\) | \(0.9^\mathrm{o}\) | Normal |

### 4.2 Stochastic strength predictions

As shown by the distribution of the FPF load and UF load in Fig. 11, it can be observed that the scatter of UF load is larger than that of FPF load since the FPF load only depends on the strength of the weakest ply and progressive failure of all the plies influences the final failure. In addition, the Tsai–Wu criterion is more conservative in evaluating the UF load than maximum stress criterion but the difference in prediction of FPF load is small. With the stochastic strength and given load variation, reliability factor of the composite plates can be evaluated for a limit state design. Furthermore, sensitivity analysis is implemented for understanding the effect of elastic properties (\(E_1\), \(E_2\), \(G_{12}\) and \(v_{12}\)), strength properties (\(X_t\), \(X_c\), \(Y_t\), \(Y_c\) and *S*) and ply angles (\(\theta _i\), \(i=1,2,\ldots ,8\)) on the failure load. The results for two loading cases based on two criteria are given in Fig. 12. The FPF load is strongly correlated with \(E_1\) and \(v_{12}\) among elastic properties and \(Y_t\), which indicates transverse matrix crack occurs at first-ply failure. As for the ultimate failure, longitudinal tensile strength \(X_t\) plays an important role. Moreover, the effect of ply misalignment on failure load is not significant in the current study involving tensile loading, which indicates this type of geometric uncertainties can be ignored in future development of non-sampling-based stochastic methods at meso-scale. However, fiber misalignment and waviness may affect the compressive failure strength (Liu et al. 2004), but a more sophisticated model should be established.

## 5 Conclusion

In this paper, a spectral-based stochastic multi-scale method is proposed for computational structural analysis of composites with multi-uncertainties. Classical asymptotic homogenization method is employed to connect material properties at micro-scale and meso-scale. Uncertainty propagation between these two scales are achieved by combing asymptotic homogenization with the spectral stochastic finite element method. With two random variables standing for uncertainties in fiber and matrix taken into account, the probability structure of effective elastic properties is fully predicted with polynomial chaos expansion. The statistical results including probability distribution, statistical moments and correlation are compared with converged results by Monte Carlo simulations, which shows good agreements. By incorporating stochastic effective properties and other uncertainties at meso-scale, stochastic strength analysis of a quasi-isotropic composite laminate has been conducted with analytical approach and Monte Carlo simulations. Probability distribution of first-ply failure and ultimate failure load values are predicted, which can provide a basis for reliability analysis of composite structures. Furthermore, sensitivity analysis shows that parameters such as \(E_1\), \(v_{12}\), \(X_t\) and \(Y_t\) are critical while the parameters \(\theta _i\) hardly affect stochastic outputs in current analysis of tensile strength, which indicates the dimension of this stochastic problem can be reduced for the uniaxial and biaxial tensile load cases studied here. Besides, a further analysis with uncorrelated effective properties shows larger scatter in the stochastic failure load prediction. It implies that a meso-scopic stochastic analysis cannot achieve reasonable prediction unless the correlations among material properties are quantified in advance.

In conclusion, this work proposes a computational framework combining different mechanics models and probability methods and illustrates stochastic multi-scale analysis of composite structures through numerical examples. As regards the uncertainty propagation from micro-scale to meso-scale, the spectral stochastic finite element method is efficient, however, the application of which is usually limited in engineering due to its intrusive formulation. To this end, a specific type of element with variable nodes is developed and implemented in a commercial FE package. Although only Gaussian random distribution is considered in this work, other types of distribution can be included in this framework by employing generalized polynomial chaos (Xiu 2010). Future development of this framework with nonlinear mechanics may consider the collocation method (Panayirci and Schuëller 2011), which is promising in stochastic analysis of complicated problems. Finally, to build an integral framework of stochastic analysis of composite structures, frequency statistics of large amount of test data or stochastic inverse analysis based on limited available experimental data is necessary for the quantification of uncertainties in the input of current analysis.

## Notes

### Acknowledgements

The support of the research scholarship for the first author and the research Grant (No. R265000523646) from NUS are gratefully acknowledged.

## References

- Andersons J, Joffe R, Spārniņš E (2008) Statistical model of the transverse ply cracking in cross-ply laminates by strength and fracture toughness based failure criteria. Eng Fract Mech 75(9):2651–2665CrossRefGoogle Scholar
- Arregui-Mena JD, Margetts L, Mummery PM (2016) Practical application of the stochastic finite element method. Arch Comput Methods Eng 23(1):171–190MathSciNetCrossRefzbMATHGoogle Scholar
- Balokas G, Czichon S, Rolfes R (2017) Neural network assisted multiscale analysis for the elastic properties prediction of 3d braided composites under uncertainty. Compos Struct 183:550–562CrossRefGoogle Scholar
- Chakraborty S, Mandal B, Chowdhury R, Chakrabarti A (2016) Stochastic free vibration analysis of laminated composite plates using polynomial correlated function expansion. Compos Struct 135:236–249CrossRefGoogle Scholar
- Chamis CC (2004) Probabilistic simulation of multi-scale composite behavior. Theor Appl Fract Mech 41(1):51–61CrossRefGoogle Scholar
- Charmpis DC, Schuëller GI, Pellissetti MF (2007) The need for linking micromechanics of materials with stochastic finite elements: a challenge for materials science. Comput Mater Sci 41(1):27–37CrossRefGoogle Scholar
- Chen X, Qiu ZP (2018) A novel uncertainty analysis method for composite structures with mixed uncertainties including random and interval variables. Compos Struct 184:400–410CrossRefGoogle Scholar
- Chen NZ, Soares CG (2008) Spectral stochastic finite element analysis for laminated composite plates. Comput Methods Appl Mech Eng 197(51):4830–4839CrossRefzbMATHGoogle Scholar
- Chen N, Yu DJ, Xia BZ, Liu J, Ma ZD (2017) Interval and subinterval homogenization-based method for determining the effective elastic properties of periodic microstructure with interval parameters. Int J Solids Struct 106:174–182CrossRefGoogle Scholar
- Chiachio M, Chiachio J, Rus G (2012) Reliability in composites—a selective review and survey of current development. Compos Part B Eng 43(3):902–913CrossRefGoogle Scholar
- Chung DB, Gutiérrez MA, de Borst R (2005) Object-oriented stochastic finite element analysis of fibre metal laminates. Comput Methods Appl Mech Eng 194(12):1427–1446CrossRefzbMATHGoogle Scholar
- Cui XY, Hu XB, Zeng Y (2017) A copula-based perturbation stochastic method for fiber-reinforced composite structures with correlations. Comput Methods Appl Mech Eng 322:351–372MathSciNetCrossRefGoogle Scholar
- Dey S, Mukhopadhyay T, Adhikari S (2015) Stochastic free vibration analyses of composite shallow doubly curved shells—a Kriging model approach. Compos B Eng 70:99–112CrossRefGoogle Scholar
- Dey S, Mukhopadhyay T, Adhikari S (2017) Metamodel based high-fidelity stochastic analysis of composite laminates: a concise review with critical comparative assessment. Compos Struct 171:227–250CrossRefGoogle Scholar
- Ghanem RG, Spanos PD (2003) Stochastic finite elements: a spectral approach. Courier Corporation, North ChelmsfordzbMATHGoogle Scholar
- Jeong HK, Shenoi RA (2000) Probabilistic strength analysis of rectangular FRP plates using Monte Carlo simulation. Comput Struct 76(1):219–235CrossRefGoogle Scholar
- Kamiński M, Kleiber M (2000) Perturbation based stochastic finite element method for homogenization of two-phase elastic composites. Comput Struct 78(6):811–826CrossRefGoogle Scholar
- Karsh PK, Mukhopadhyay T, Dey S (2018) Spatial vulnerability analysis for the first ply failure strength of composite laminates including effect of delamination. Compos Struct 184:554–567CrossRefGoogle Scholar
- Kaw Autar K (2005) Mechanics of composite materials. CRC Press, Boca RatonzbMATHGoogle Scholar
- Kepple J, Herath MT, Pearce G, Prusty BG, Thomson R, Degenhardt R (2015) Stochastic analysis of imperfection sensitive unstiffened composite cylinders using realistic imperfection models. Compos Struct 126:159–173CrossRefGoogle Scholar
- Lal A, Singh BN, Kumar R (2009) Effects of random system properties on the thermal buckling analysis of laminated composite plates. Comput Struct 87(17):1119–1128CrossRefGoogle Scholar
- Lal A, Singh BN, Kale S (2011) Stochastic post buckling analysis of laminated composite cylindrical shell panel subjected to hygrothermomechanical loading. Compos Struct 93(4):1187–1200CrossRefGoogle Scholar
- Lekou DJ, Philippidis TP (2008) Mechanical property variability in FRP laminates and its effect on failure prediction. Compos B Eng 39(7–8):1247–1256CrossRefGoogle Scholar
- Li JQ, Tian XP, Han ZJ, Narita Y (2016) Stochastic thermal buckling analysis of laminated plates using perturbation technique. Compos Struct 139:1–12CrossRefGoogle Scholar
- Liu D, Fleck NA, Sutcliffe MPF (2004) Compressive strength of fibre composites with random fibre waviness. J Mech Phys Solids 52(7):1481–1505CrossRefzbMATHGoogle Scholar
- Lopez RH, Miguel LFF, Belo IM, Cursi JES (2014) Advantages of employing a full characterization method over form in the reliability analysis of laminated composite plates. Compos Struct 107:635–642CrossRefGoogle Scholar
- Lorca JL, González C, Molina-Aldareguía JM, Segurado J, Seltzer R, Sket F, Rodríguez M, Sádaba S, Muñoz R, Canal LP (2011) Multiscale modeling of composite materials: a roadmap towards virtual testing. Adv Mater 23(44):5130–5147CrossRefGoogle Scholar
- Ma J, Temizer I, Wriggers P (2011) Random homogenization analysis in linear elasticity based on analytical bounds and estimates. Int J Solids Struct 48(2):280–291CrossRefzbMATHGoogle Scholar
- Nakayasu H, Maekawa Z (1997) A comparative study of failure criteria in probabilistic fields and stochastic failure envelopes of composite materials. Reliab Eng Syst Saf 56(3):209–220CrossRefGoogle Scholar
- Ngah MF, Young A (2007) Application of the spectral stochastic finite element method for performance prediction of composite structures. Compos Struct 78(3):447–456CrossRefGoogle Scholar
- Noh HC, Park T (2011) Response variability of laminate composite plates due to spatially random material parameter. Comput Methods Appl Mech Eng 200(29):2397–2406MathSciNetCrossRefzbMATHGoogle Scholar
- Onkar AK, Upadhyay CS, Yadav D (2007) Probabilistic failure of laminated composite plates using the stochastic finite element method. Compos Struct 77(1):79–91CrossRefGoogle Scholar
- Panayirci HM, Schuëller GI (2011) On the capabilities of the polynomial chaos expansion method within SFE analysis—an overview. Arch Comput Methods Eng 18(1):43–55CrossRefzbMATHGoogle Scholar
- Perić D, de Souza Neto EA, Feijóo RA, Partovi M (2011) On micro-to-macro transitions for multi-scale analysis of non-linear heterogeneous materials: unified variational basis and finite element implementation. Int J Numer Meth Eng 87(1–5):149–170zbMATHGoogle Scholar
- Pimenta S (2017) A computationally-efficient hierarchical scaling law to predict damage accumulation in composite fibre-bundles. Compos Sci Technol 146:210–225CrossRefGoogle Scholar
- Pimenta S, Pinho ST (2013) Hierarchical scaling law for the strength of composite fibre bundles. J Mech Phys Solids 61(6):1337–1356MathSciNetCrossRefGoogle Scholar
- Sakata S, Ashida F, Kojima T, Zako M (2008) Three-dimensional stochastic analysis using a perturbation-based homogenization method for elastic properties of composite material considering microscopic uncertainty. Int J Solids Struct 45(3):894–907CrossRefzbMATHGoogle Scholar
- Sánchez-Heres LF, Ringsberg JW, Johnson E (2014) Influence of mechanical and probabilistic models on the reliability estimates of fibre-reinforced cross-ply laminates. Struct Saf 51:35–46CrossRefGoogle Scholar
- Scarth C, Cooper JE, Weaver PM, Silva GHC (2014) Uncertainty quantification of aeroelastic stability of composite plate wings using lamination parameters. Compos Struct 116:84–93CrossRefGoogle Scholar
- Sepahvand K (2016) Spectral stochastic finite element vibration analysis of fiber-reinforced composites with random fiber orientation. Compos Struct 145:119–128CrossRefGoogle Scholar
- Shaw A, Sriramula S, Gosling PD, Chryssanthopoulos MK (2010) A critical reliability evaluation of fibre reinforced composite materials based on probabilistic micro and macro-mechanical analysis. Compos B Eng 41(6):446–453CrossRefGoogle Scholar
- Soden PD, Hinton MJ, Kaddour AS (1998) Lamina properties, lay-up configurations and loading conditions for a range of fibre-reinforced composite laminates. Compos Sci Technol 7(58):1011–1022CrossRefGoogle Scholar
- Sriramula S, Chryssanthopoulos MK (2009) Quantification of uncertainty modelling in stochastic analysis of frp composites. Compos A Appl Sci Manuf 40(11):1673–1684CrossRefGoogle Scholar
- Stefanou G (2009) The stochastic finite element method: past, present and future. Comput Methods Appl Mech Eng 198(9):1031–1051CrossRefzbMATHGoogle Scholar
- St-Pierre L, Martorell NJ, Pinho ST (2017) Stress redistribution around clusters of broken fibres in a composite. Compos Struct 168:226–233CrossRefGoogle Scholar
- Swolfs Y, Verpoest I, Gorbatikh L (2016) A review of input data and modelling assumptions in longitudinal strength models for unidirectional fibre-reinforced composites. Compos Struct 150:153–172CrossRefGoogle Scholar
- Tay TE, Liu G, Tan VBC, Sun XS, Pham DC (2008) Progressive failure analysis of composites. J Compos Mater 42(18):1921–1966CrossRefGoogle Scholar
- Tootkaboni M, Graham-Brady L (2010) A multi-scale spectral stochastic method for homogenization of multi-phase periodic composites with random material properties. Int J Numer Meth Eng 83(1):59–90MathSciNetzbMATHGoogle Scholar
- Van Vinckenroy G, De Wilde WP (1995) The use of Monte Carlo techniques in statistical finite element methods for the determination of the structural behaviour of composite materials structural components. Compos Struct 32(1–4):247–253CrossRefGoogle Scholar
- Vijaya Kumar RL, Bhat MR (2015) Probabilistic stress variation studies on composite single lap joint using Monte Carlo simulation. Compos Struct 121:351–361CrossRefGoogle Scholar
- Vinogradov V, Hashin Z (2005) Probabilistic energy based model for prediction of transverse cracking in cross-ply laminates. Int J Solids Struct 42(2):365–392CrossRefzbMATHGoogle Scholar
- Wu WF, Cheng HC, Kang CK (2000) Random field formulation of composite laminates. Compos Struct 49(1):87–93CrossRefGoogle Scholar
- Xiu DB (2010) Numerical methods for stochastic computations: a spectral method approach. Princeton University Press, PrincetonzbMATHGoogle Scholar
- Zhao LB, Shan MJ, Liu FR, Zhang JY (2017) A probabilistic model for strength analysis of composite double-lap single-bolt joints. Compos Struct 161:419–427CrossRefGoogle Scholar
- Zhou XY, Gosling PD, Pearce CJ, Kaczmarczyk Ł, Ullah Z (2016) Perturbation-based stochastic multi-scale computational homogenization method for the determination of the effective properties of composite materials with random properties. Comput Methods Appl Mech Eng 300:84–105MathSciNetCrossRefGoogle Scholar
- Zhou XY, Gosling PD, Pearce CJ, Ullah Z, Kaczmarczyk Ł (2016) Perturbation-based stochastic multi-scale computational homogenization method for woven textile composites. Int J Solids Struct 80:368–380CrossRefGoogle Scholar
- Zhou XY, Gosling PD, Ullah Z, Kaczmarczyk Ł, Pearce CJ (2016) Exploiting the benefits of multi-scale analysis in reliability analysis for composite structures. Compos Struct 155:197–212CrossRefGoogle Scholar
- Zhou XY, Gosling PD, Ullah Z, Kaczmarczyk Ł, Pearce CJ (2017) Stochastic multi-scale finite element based reliability analysis for laminated composite structures. Appl Math Model 45:457–473MathSciNetCrossRefGoogle Scholar