Using Artificial Neural Networks to Predict the Fatigue Life of Different Composite Materials Including the Stress Ratio Effect

Artificial Neural Networks (ANN) have been successfully used in predicting the fatigue behavior of fiber-reinforced composite materials. In most cases, the predictions were obtained for the same material used in training subjected to different loading conditions. The method would be of greater value if one could predict the failure of materials other than those used for training the network. In a recent paper, ANN trained using the experimental fatigue data obtained for composites subjected to a constant stress ratio ( R = s _min/s _max ) \left( {{\hbox{R}} = {\sigma_{{ \min }}}/{\sigma_{{ \max }}}} \right) was successfully used to predict the cyclic behavior of a composite made of a different material. In this work, this method is extended to include the stress ratio effect. The results show that ANN can provide accurate fatigue life prediction for different materials under different values of the stress ratio. These results can allow for the development of a materials smart database that can be used for various engineering applications.     

Multiscale finite element modeling of failure process of composite laminates

A multiscale nonlinear finite element modeling technique is developed in this paper to predict the progressive failure process for composite laminates. A micromechanical elastic–plastic bridging constitutive model, which considers the nonlinear material properties of the constituent fiber and matrix materials and their interaction and the damage and failure in fibrous composites at the fiber and matrix level, is proposed to represent the material behavior of fiber-reinforced composite laminates. The micromechanics constitutive model is employed in the macroscale finite element analysis of structural behavior especially progressive failure process of the fiber-reinforced composites based on a 4-node 24-DOF shear-locking free rectangular composite plate element.     

基于频率变化预测玻璃纤维增强树脂复合材料层合板的剩余疲劳寿命

复合材料结构在疲劳过程中的累积损伤将导致结构刚度下降,并进一步引起结构的动态参数如频率发生衰减。因此,可以将结构疲劳状态与结构频率联系起来,基于频率预测结构的剩余疲劳寿命。本文首先基于复合材料在纵向、横向和面内剪切三个方向的疲劳特性,结合ABAQUS与Umat子程序开发了三维有限元模型模拟复合材料层合板中的疲劳损伤演变,并构建了不同疲劳状态下对应的模态分析模型,由此获得了疲劳过程中的频率衰减曲线。之后,基于疲劳过程的频率变化量训练了人工神经网络,用于预测玻璃纤维增强复合材料层合板的剩余疲劳寿命。特别地,在当前的数值模型中为每个单元分配了符合高斯正态分布的材料属性,以模拟实际情况下复合材料性能的离散性。结果表明,疲劳模型数值模拟结果与已有文献的疲劳实验数据吻合,基于频率变化量训练的人工神经网络可以成功预测玻璃纤维增强复合材料试件的剩余疲劳寿命。      

Energy-based fatigue life prediction of fiberglass/epoxy composites using modular neural networks

The strain energy has successfully been used in the past as a fatigue failure criterion for unidirectional fiber reinforced laminae. This approach has the ability to unify the macro- and microscopic behavior and can allow for extending the failure criterion to incorporate the multiaxial case. In this work, the strain energy will be used as an input to the artificial neural network (ANN) to predict fatigue failure. The results obtained will be compared to those obtained using the maximum applied stress, the fiber orientation angle and the stress ratio as inputs to the ANN.     

Progressive damage modeling in fiber-reinforced materials

This paper presents an anisotropic damage model suitable for predicting failure and post-failure behavior in fiber-reinforced materials. In the model the plane stress formulation is used and the response of the undamaged material is assumed to be linearly elastic. The model is intended to predict behavior of elastic-brittle materials that show no significant plastic deformation before failure. Four different failure modes – fiber tension, fiber compression, matrix tension, and matrix compression – are considered and modeled separately. The onset of damage is predicted using Hashin’s initiation criteria [Hashin Z, Rotem A. A fatigue failure criterion for fiber-reinforced materials. J Compos Mater 1973;7:448; Hashin Z. Failure criteria for unidirectional fiber composites. J Appl Mech 1980;47:329–34] and the progression of damage is controlled by a new damage evolution law, which is easy to implement in a finite element code. The evolution law is based on fracture energy dissipation during the damage process and the increase in damage is controlled by equivalent displacements. The issues related to numerical implementation, such as mesh sensitivity and convergence in the softening regime, are also addressed.     

A new fatigue failure theory for multidirectional fiber-reinforced composite laminates with arbitrary stacking sequence

Fatigue failure is one of the most important failure types of fiber-reinforced composites. In this paper, a new fatigue failure theory for multidirectional fiber-reinforced composite laminates with an arbitrary stacking sequence is developed, by combining nonlinear residual strength and residual stiffness models with the recently improved Puck’s failure theory which includes the in situ strength effect. This fatigue theory can predict the fatigue life, residual strength and residual failure envelope of fiber-reinforced composite laminates under multidirectional loadings. For these predictions it is necessary to recalculate the fatigue lives of laminae after each cycle since the stresses in the laminae change due to stiffness degradation. It is also necessary to account for the nonlinear accumulation of damage at the new stress level in the laminae resulting from stiffness degradation. This is achieved by using the concept of equivalent cycle. The theoretical predictions are in good agreement with available experimental results.     

Prediction and control of surface roughness for the milling of Al/SiC metal matrix composites based on neural networks

In recent years, there has been a significant increase in the utilization of Al/SiC particulate composite materials in engineering fields, and the demand for accurate machining of such composite materials has grown accordingly. In this paper, a feed-forward multi-layered artificial neural network (ANN) roughness prediction model, using the Levenberg-Marquardt backpropagation training algorithm, is proposed to investigate the mathematical relationship between cutting parameters and average surface roughness during milling Al/SiC particulate composite materials. Milling experiments were conducted on a computer numerical control (CNC) milling machine with polycrystalline diamond (PCD) tools to acquire data for training the ANN roughness prediction model. Four cutting parameters were considered in these experiments: cutting speed, depth of cut, feed rate, and volume fraction of SiC. These parameters were also used as inputs for the ANN roughness prediction model. The output of the model was the average surface roughness of the machined workpiece. A successfully trained ANN roughness prediction model could predict the corresponding average surface roughness based on given cutting parameters, with a 2.08% mean relative error. Moreover, a roughness control model that could accurately determine the corresponding cutting parameters for a specific desired roughness with a 2.91% mean relative error was developed based on the ANN roughness prediction model. Finally, a more reliable and readable analysis of the influence of each parameter on roughness or the interaction between different parameters was conducted with the help of the ANN prediction model.The full text can be downloaded at https://link.springer.com/article/10.1007/s40436-020-00326-x     

Artificial neural networks for predicting the macromechanical behaviour of ceramic-matrix composites

In ceramic-matrix composites (CMCs) a weak fibre/matrix interface is required to achieve satisfactory toughening so that the composite exhibits damage tolerant characteristics. Due to the presence of such a weak interface, debonding and sliding occur at the interface making the mechanics of the material very complex. As a result, developing analytical models for simulating the macromechanical behaviour of these composites is extremely difficult and necessitates simplifying assumptions compromising accuracy. In the present paper, a novel approach to modelling the macromechanical behaviour of CMCs, using the artificial neural network (ANN) approach has been presented. The ability of neural networks in learning the complex multi-parametric interaction among the various microstructural parameters has been demonstrated with an example of SiC/SiC ceramic composite. An artificial neural network has been used to postulate the macromechanical behaviour of SiC (matrix)/SiC (fibre) composite. The training examples for the network have been generated through an accurate micromechanical finite element analysis that models the interfacial debonding and sliding realistically. The network learning is demonstrated and the network is validated by asking it to predict the behaviour of the composite for new specimens. Various stages in the development of ANN such as the preparation of training set, selection of a network configuration, training of the net and a testing scheme, etc, have been addressed at length in this paper.     

A micromechanical approach for the fatigue failure prediction of unidirectional polymer matrix composites in off-axis loading including the effect of viscoelasticity

A novel micromechanical approach is used to study the fatigue failure of unidirectional polymer matrix composites subject to off-axis loading. The main advantage of the present micromechanical model lies in its ability to give closed form solutions for the effective nonlinear response of unidirectional composites and to predict the material response to any combination of shear and normal loading. The fatigue failure criterion is expressed in terms of the fatigue failure functions of the constituent materials. The micromechanical model is also used to calculate these fatigue failure functions from the knowledge of the S–N diagrams of the composite material in longitudinal, transverse, and shear loadings; thus, eliminating the need for any further experimentation. Unlike previous works, the present study accounts for the viscoelasticity of the matrix material rendering it the capability of modeling creep damage accumulation in high-temperature composite materials. The results are found to be in good agreement with the literature. In particular, for higher off-axis angles, the results are seen to be in better concurrence with the experimental data compared to when the effect of viscoelasticity is overlooked. The present approach is also capable of accounting for the strain evolution due to viscoelasticity of the matrix material.     

Composite concrete/GFRP slabs for footbridge deck systems

In recent years the use of advanced composite materials has gained wider space in the civil engineering sector, due to some favorable characteristics such as lightweight, high specific strength, resistance to corrosion and fatigue. Innovative systems that combine concrete with advanced composite materials have proved to be a viable and efficient solution as compared to conventional systems. In this work, a new slab system composed of a fiber-reinforced concrete top laid on glass fiber reinforced polymeric (GFRP) wide-flange-section pultruded profiles, filled in with foam blocks, is presented. The material properties of the GFRP profiles were obtained both theoretically and experimentally. Experimental tests to choose the appropriate resin to bond the concrete to the GFRP profiles and to select the appropriate short fiber and volume fraction to be used in the concrete top have also been conducted. The slab was designed to sustain constructive loads and live pedestrian loads for footbridge deck applications. To investigate the slab flexural behavior up to failure, three specimens were tested under four-point bending, and theoretical and finite element analyses were also performed. Comparisons of theoretical, numerical and experimental results show good agreement. Studies under way to complete the development of the proposed slab are briefly described at the end of the work.     

Thermoreversible and remendable glass-polymer interface for fiber-reinforced composites

Adhesion of the reinforcement to the polymer matrix is essential for load transfer from the polymer matrix to the reinforcement material in fiber-reinforced composites. The reversible Diels-Alder reaction between a furan-functionalized epoxy-amine thermosetting matrix with a maleimide-functionalized glass fiber was used to impart remendability at the polymer-glass interface for potential application in glass fiber-reinforced composites. At room temperature the Diels-Alder adduct is formed spontaneously and above 90 °C the adduct breaks apart to reform the original furan and maleimide moieties. Healing of the interface was investigated with single fiber microdroplet pull-out testing. Following complete failure of this interface, significant healing was observed, with some specimens recovering over 100% of the initial properties. Healing efficiency was not affected by the distance of displacement, with an overall average of 41% healing efficiency. Up to five healing cycles were successfully achieved. It is expected that a glass fiber-reinforced composite of maleimide-sized glass within a furan-functionalized network will demonstrate extension of fatigue life.     

Relationship Between Hysteresis Dissipated Energy and Temperature Rising in Fiber-Reinforced Ceramic-Matrix Composites Under Cyclic Loading

In this paper, the relationship between hysteresis dissipated energy and temperature rising of the external surface in fiber-reinforced ceramic-matrix composites (CMCs) during the application of cyclic loading has been analyzed. The temperature rise, which is caused by frictional slip of fibers within the composite, is related to the hysteresis dissipated energy. Based on the fatigue hysteresis theories considering fibers failure, the hysteresis dissipated energy and a hysteresis dissipated energy-based damage parameter changing with the increase of cycle number have been investigated. The relationship between the hysteresis dissipated energy, a hysteresis dissipated energy-based damage parameter and a temperature rise-based damage parameter have been established. The experimental temperature rise-based damage parameter of unidirectional, cross-ply and 2D woven CMCs corresponding to different fatigue peak stresses and cycle numbers have been predicted. It was found that the temperature rise-based parameter can be used to monitor the fatigue damage evolution and predict the fatigue life of fiber-reinforced CMCs.     

A micromechanical approach for the prediction of the time-dependent failure of high temperature polymer matrix composites

In the present study, a novel micromechanical approach is introduced to study the time-dependent failure of unidirectional polymer matrix composites. The main advantage of the present micromechanical model lies in its ability to give closed-form solutions for the effective nonlinear response of unidirectional composites and to predict the material response to any combination of shear and normal loading. The creep failure criterion is expressed in terms of the creep failure functions of the viscoelastic matrix material. The micromechanical model is also used to calculate these creep failure functions from the knowledge of the creep behavior of the composite material in only transverse and shear loadings, thus eliminating the need for any further experimentation. The composite material used in this study is T300/934, which is suitable for service at high temperatures in aerospace applications. The use of micromechanics can give a more accurate insight into the failure mechanisms of the composite materials in particular at high temperatures where the general behavior of the polymer matrix composite is governed by matrix viscoelasticity and the time-dependent failure of the matrix is a localized phenomenon. The obtained creep failure stresses are found to be in reasonable agreement with the experimental data.     

Fabrication and modelling of the macro-mechanical properties of cross-ply laminated fibre-reinforced polymer composites using artificial neural network

The mechanical behaviour of fibre-reinforced polymer composites (FRPCs) is considered very complex due to many factors such as composition, material type, manufacturing process and end user applications. This article presents the mechanical properties and artificial neural network (ANN) modelling results of cross-ply laminated FRPCs. Twenty composite samples were fabricated by varying the number of layers of carbon fibre and glass fibre as reinforcement and polyphenylene sulphide and high-density polyethylene as matrix. Mechanical properties were measured in terms of flexural modulus, hardness, impact and transverse rupture strength. Multilayer feed-forward backpropagation ANN approach was used to predict the mechanical properties by using material type, composition and number of reinforcement and matrix layers as input variables. From 20 data patterns, 16 were used for network training and remaining 4 were used to test the models. Furthermore, trend analysis was also performed to understand the influence of inputs on developed models. It is evident from the ANN prediction results that there is good correlation between predicted and actual values within acceptable mean absolute error. The outcomes of this research will help to reduce cost and time by eliminating tedious composite property measurements and to fabricate tailored composites meeting application requirements.     

Predicting Failure Strength of Randomly Oriented Short Glass Fiber-Epoxy Composite Specimen by Artificial Neural Network Using Acoustic Emission Parameters

Acoustic emission (AE) peak amplitude and cumulative energy emitted during 50% of failure of composite specimen was collected, analyzed, and utilized to predict the ultimate tensile strength (UTS) using artificial neural network (ANN) and the performance of various training algorithm on prediction was analyzed. AE data have been collected from finite numbers of randomly oriented short glass fiber-epoxy tensile specimens, while loading up to failure in a tensile testing machine. AE response from each of the specimen was classified and segregated by understanding the failure mechanism. A feed forward back-propagation type ANN was designed and the segregated data of amplitude hits and cumulative energy was processed using two separate networks to predict the UTS of corresponding specimens using it with appropriate parameters and the results were analyzed.     

Bridged versus cohesive crack in the flexural behavior of brittle-matrix composites

A nonlinear fracture mechanics model, which explains and reproduces the constitutive flexural behavior of a brittle-matrix composite, is proposed. It embraces in a unified dimensionless formulation two peculiar models, i.e., the cohesive-crack and the bridged-crack, which are used to analyze the composite failure process. Dimensionless parameters, which depend on the mechanical and geometrical properties, characterize the structure in flexure. It is shown that, based on the assumptions of the bridged-crack model, which simulates the composite as a multiphase material, the flexural response is controlled by two dimensionless parameters, whereas, based on the assumptions of the cohensive-crack model, which simulates the composite as a homogeneous material, the parameters reduce to one. The influence of the dimensionless parameters on the behavior is studied, along with the size-scale effects on the structural ductility. It is also shown how the matrix toughness affects the response. The two theoretical models are compared through the simulation of an experimental test on a fiber-reinforced beam, and it is shown that both the models can predict approximately the same overall behavior.     

Experimental study on the tension fatigue behavior and failure mechanism of 3D multi-axial warp knitted composites

An experimental study is described in this paper dealing with the tension–tension fatigue and failure mechanism of 3D MWK composites with different fiber architectures and material sizes. Macroscopic fracture morphology and SEM micrographs are examined to understand the fatigue damage and failure mechanism. The results show the fatigue properties and failure mechanism of composites can be affected significantly by the fiber architecture and material size. The fatigue life of material A(0°/0°/0°/0°) with small fiber orientation angle is significantly longer than that of material B(+45°/−45°/+45°/−45°). For material A, the fatigue properties of the long composite are better than that of the short one. It is 0° fiber bundles fracture under fatigue stress which cause the material failure and the long composite provides more space for the formation and propagation of local fatigue micro-cracks. However, for material B, the short composites have better fatigue properties. Moreover, the materials show typical ±45° zigzag fatigue fracture and obvious shear behavior. The fatigue cracks for the long composite can be spread more quickly along the fiber/matrix interface due to the fiber bundles realignment.     

Predicting the Crushing Behavior of Axially Loaded Elliptical Composite Tubes Using Artificial Neural Networks

In this research work, the artificial neural networks (ANN) technique is used in predicting the crushing behavior and energy absorption characteristics of axially-loaded glass fiber/epoxy composite elliptical tubes. Predictions are compared to actual experimental results obtained from the literature and are shown to be in good agreement. Effects of parameters such as network architecture, number of hidden layers and number of neurons per hidden layer are also considered. The study shows that ANN techniques can effectively be used to predict the crushing response and the energy absorption characteristics of elliptical composite tubes with various ellipticity ratios subjected to axial loading.     

An investigation into the flexural and drawing behaviors of GFRP-based fiber–metal laminate

Fiber–metal laminates (FMLs) are advanced composite materials that consist of bonded thin metal sheets and fiber-reinforced composite layers. In this article, mechanical behavior of a thermoplastic-based FML is investigated, which is composed of glass-fiber-reinforced polypropylene (GFRP) laminate and aluminum AA1200-O as the core and skin layers, respectively. Engineering constants of the composite laminate were achieved using Timoshenko's beam theory, flexural and tensile test results. Finite element simulations of the GFRP-based FML were performed to predict the behavior of this material in three-point bending and deep drawing tests. Some experimental verification tests were conducted to prove the reliability of results in the FE analysis of the FML. Comparison of the results shows an excellent correlation between the FE analysis and experimental tests.     

Fatigue variability of a single crystal superalloy at elevated temperature

In order to develop more accurate life prediction tools, an improved understanding of the variability within the fatigue behavior of a material is required. Recent work has shown multiple failure mechanisms that drive the variability in fatigue life of polycrystalline titanium and nickel materials. In addition, the bimodal behavior in the fatigue response is not readily apparent when only a very small number of specimens are tested at each loading condition, as is normal practice.The objective of this work was to investigate the fatigue variability of a single crystal nickel-base superalloy at elevated temperature. PWA1484, a second generation single crystal alloy developed for advanced turbine airfoil applications, was the material of choice for this investigation. A large number of fatigue tests were performed at one condition (stress level, stress ratio, frequency and temperature) to determine the variability and identify the sources of uncertainty in life. Scanning electron microscopy was used to investigate the relationship between failure mechanisms and variability. Crack growth analyses were used to predict lowest life estimates and were compared to experimental data. The results show large variability in fatigue life at fairly high stresses. Evaluation of the fracture surfaces indicated that microstructural features such as carbides and eutectics were responsible for the failures. In addition, the size of the feature responsible for fatigue failure could not be directly related to the fatigue life. The lowest expected life based on fatigue crack growth analyses did agree with the shortest life found experimentally. However, more testing and analysis is required.