Search

Various types of internal defects occur during manufacturing and handling of composite materials. It is practically impossible to manufacture composite structures without defects, making it crucial to understand the effect of defects on their failure behavior to maintain structural safety. In this work, the effect of pre-defects on the failure behavior of plain weave textile composites was studied. Unit cell configurations with symmetric, in-phase, and shifted fiber tow arrangements were considered. Inter-laced warp and fill tows and matrix pockets of plain weave unit cells were modeled in three-dimensional finite elements, and cohesive elements were inserted between all bulk elements to account for the fracture modes of the fiber and matrix direction failure of warp and fill tows, matrix pocket failure, and interface failure. Unit cell models containing pre-defects of voids, tow-matrix pocket separation, warp-fill tow separation, and cracks in the warp and fill tows were analyzed, and their effects on progressive failure behavior were investigated in terms of the interaction between fiber tow arrangements and defects. Results indicated that initial failure occurred in matrix-direction failure mode in fill tows, whereas fiber tow-matrix pocket separation was the major failure mode under uniaxial tensile load. Furthermore, failure behavior was found to be highly dependent on the fiber tow arrangement pattern and the location of pre-defects.

In this work, the effective electrical conductivity (EEC) and percolation threshold (PT) of a nanocomposite containing spherical particle fillers are predicted by computational homogenization schemes (CHS) with three-dimensional random representative volume elements (RVEs) by random filler removal (RFR). For this prediction, we prepare the RVE having the maximum filler volume fraction (Vf) of 52% with random particle fillers, also termed the master model, by the discrete element method (DEM), and the corresponding finite-element (FE) model is created. Then, 100 RVE samples for each Vf are derived by randomly replacing the material properties of several fillers by those of the matrix from the master model with diverse Vf from 5% to 50%. In addition, the interphase layer is employed by replacing some matrix elements with interphase elements according to the neighboring distance of the fillers. To demonstrate the performance of the proposed scheme, its randomness of RVEs is verified by spatial and physical metrics in terms of autocorrelation analysis, near-neighbor analysis, and directional conductivity ratio. The EEC prediction results with diverse Vf values are compared with those of an analytical model and test results. As a result, the PT at which EEC of the nanocomposites suddenly increases is successfully evaluated, and the effect of the void, interphase thickness, and conductivity, as well as the size of fillers on the EEC and PT is investigated through a sensitivity analysis.

In this paper, we propose a simple and efficient generator of random fiber distribution with diverse fiber volume fractions ( V f ) for unidirectional composites by a random fiber removal technique. From the representative volume element (RVE) consisting of 100 fibers that have a maximum V f of about 65% in this work, also termed the master model, we randomly eliminate fibers to match the predefined V f ranging from 60%, 55%, 45%, 35%, 25%, 15%, and 5%, which are lower than that of the master model. Accordingly, 100 RVE samples for each V f can be constructed in a straightforward manner. To demonstrate the performance of the proposed algorithm, its fiber locations are verified in terms of statistical spatial metrics, such as the nearest neighbor orientation, Ripley's K function, and pair distribution function. Furthermore, the elastic properties and the anisotropic ratios of the generated RVEs are investigated and compared to those of other random fiber generation algorithms.

In this work, we propose a prediction model of the transverse mechanical behavior of unidirectional (UD) composites containing complex microstructure with the help of a convolutional neural network (CNN). For this prediction, a total of 900 representative volume elements (RVE) samples were generated by constructing 300 RVEs for each V f of 40%, 50%, and 60% with the random sequential expansion (RSE) algorithm. The stress-strain (S–S) curves in terms of transverse elastic modulus, transverse tensile strength, and toughness considering interphase debonding were obtained by a finite element (FE) simulation with the RVE samples. After converting FE models with 900 RVE samples to corresponding microstructural binary images, CNN modeling was employed to construct a prediction model on the microstructural images. To demonstrate the performance of the proposed CNN model, we predicted the transverse mechanical behavior in terms of the S–S curves on various test datasets. Prediction accuracy was verified in terms of the loss functions and the error of the S–S curve. The prediction results were in excellent agreement with the test datasets, and the transverse mechanical behavior was quickly predicted for other microstructures. This confirmed that the proposed CNN model is simple and powerful and can efficiently clarify the relationship between the microstructure and transverse mechanical behavior of UD composites.

An improved finite-element (FE) modeling scheme of misaligned noncircular fibers of a unidirectional composite is proposed with the aid of micro-computed tomography (μCT). The proposed scheme selects several key images among many raw CT images containing fiber misalignment and noncircular fibers. Then, it constructs two-dimensional (2D) drawings based on the signed distance obtained by solving the level-set equation with the trimmed mesh technique for phase segmentation meshing. Subsequently, these 2D drawings are bridged to construct a complex three-dimensional (3D) solid model for 3D FE model generation. To demonstrate the performance of the proposed scheme, we made an intensive comparison with other microstructure FE models in terms of the fiber volume fraction and equivalent elastic properties. Furthermore, the transverse strength was predicted through simulation and compared with the test results, showing excellent agreement with the test result that cannot be achieved with other simple modeling methods.

The properties of fiber reinforced composites are mainly determined by the fraction of reinforcement and matrix. Thus, to design a system based on composite materials, it is vital to carefully measure the volume fractions of the composites with a proper method. Digestion by strong acid or ignition at high temperature in an oxidizing environment are conventional for measuring content fractions. In essence, these methods assume that the reinforcement does not lose weight by digestion or ignition. However, by neglecting reported vulnerable oxidization characteristics of carbon fiber, these conventional methods result in inaccurate fiber volume fractions of carbon fiber/epoxy composites. In this study, an effective and accurate method, having only 2 steps in measuring process and 1.5%p maximum error, for determining the fiber volume fractions of two different PAN (Polyacylonitrile)-based carbon fiber reinforced composites via thermo-gravimetric analysis was developed and subsequently verified using the results from 80 microscopic images.

A simple and accurate scheme for modeling microstructures is proposed with the help of element trimming combined with signed distance function based boundary smoothing. To accommodate randomly distributed fibers in unidirectional composites, digital image processing is used. The interfaces of multi-materials are identified by introducing a signed distance function, and then, square background elements crossing the interfaces are simply trimmed and divided to represent a single material behavior by a single element. After element trimming, the elements that are polygon-shaped in the two-dimensional domain are split into conventional three-node triangle elements (six-node prism elements in the three-dimensional domain) available in many commercial software packages. The present modeling scheme was verified through benchmark examples in terms of the accuracy and efficiency and then applied to the modeling of unidirectional composites based on real microscopic images to evaluate the equivalent elastic properties.

This paper presents the development of artificial neural network (ANN) modeling for predicting the transverse elastic modulus of a unidirectional composite (E-glass/MY750) with fiber positions and volume fraction. For this prediction, random representative volume elements were generated according to the fiber volume fraction ( V f ). A training dataset consisting of input data ( V f and fiber locations) and output data (the effective elastic modulus), which were computed by the computational homogenization scheme (CHS), was used to train the ANN model to have proper weight and bias by back propagation. To demonstrate the performance of the proposed ANN model, prediction of the transverse elastic modulus of various test datasets whose transverse elastic moduli are known by CHS was conducted. The prediction accuracy was verified in terms of the mean squared error, correlation coefficient (R), and relative error. The prediction results showed excellent agreement with the test dataset and quickly predicted the transverse elastic modulus having random microstructures.

Measuring fiber volume fraction properly is very important in designing composite materials because the fiber volume fraction mainly determines mechanical and thermal properties. Conventional Ignition methods are effective for ceramic fiber reinforcing composite materials. However, these methods are not proper for applying to carbon fiber reinforcing composites because of the venerable characteristic against oxidation of carbon fiber. In the research, fiber volume fraction of carbon fiber composites was obtained by a thermogravimetric analysis considering oxidation characteristic of the carbon fiber and the method was compared and verified with the results from microscopic cross section images. 초 록: 섬유체적비에 따라 복합재료의 기계적 열적 특성이 크게 달라지기 때문에, 복합재료 설계시 섬유체적비 를 올바르게 측정하는 것이 매우 중요하다. 일반적으로 섬유체적비를 측정하는 여러 방법은 산화되지 않는 유리 섬유나 세라믹섬유를 사용한 복합재료의 경우에는 적합하고 효율적이다. 하지만 산화현상이 있는 탄소섬유의 경 우에는 산화 방법과 조건에 따라서 다른 결과를 가져오게 되며 그러므로 올바른 섬유체적비를 측정이 어렵다. 본 연구에서는 Thermogravimetric analysis를 수행하여 산화되는 탄소섬유의 질량 감소량을 보정하여 탄소섬유 복합 재료의 섬유체적비를 측정하였고 현미경 단면 이미지를 이용하여 그 결과를 검증하였다.

In this work, the coefficients of thermal expansion (CTEs) of fibers in composite materials that contain microstructures are numerically evaluated using a lamina-scale cost function with quasi-analytical gradients. To consider the effects of fiber arrangements and local defects, such as interface debonding and voids, a variety of representative volume elements are modeled with a number of finite element meshes. Then, the CTEs of fibers are evaluated by minimizing a lamina-scale cost function that represents the difference between the measured CTEs and the computed CTEs by means of a computational homogenization scheme for the composite lamina. The descent direction of the cost function is obtained using quasi-analytical gradients that take partial derivatives from prediction models, such as the Schapery model and Hashin model defined in an explicit manner, which accelerates the minimization procedure. To verify the performance of the proposed scheme in terms of accuracy and efficiency, the CTEs of constituents calculated using the proposed scheme in a unidirectional composite lamina are compared with experimental values reported in the literature. Furthermore, the convergence behavior of the proposed scheme with quasi-analytical gradients is also investigated and compared with other minimization methods.

This study investigated the transverse damage behavior of a unidirectional composite containing a complex microstructure having noncircular fiber cross-sections. For this purpose, a finite-element (FE) model based real microscopic images of the M55J/M18 composite was generated with the signed distance function (SDF) by the level-set method and the trimming mesh technique. In addition, the interphase zone was constructed along the interface between the fiber and the matrix. Subsequently, a virtual experiment simulating a three-point bending test was conducted with the generated FE model. Crack propagation analysis was carried out with the cohesive zone model (CZM) by applying transverse tension under plane strain condition. The crack length and propagation directions were compared to those of the three-point bending test. To make a nice correlation between the virtual experiment and actual experiment, the effects of the fiber shape, thermal residual stress, and various fracture toughnesses were investigated on the length and propagation direction of the cracks. It was found that, the fiber shape is vital to the inter-fiber distance and fiber volume fraction (Vf), which strengthen or weaken the stress concentration, thus making the propagation direction of the cracks warped and the length of cracks varied.

In this paper, a variation of the critical projectile diameter of honeycomb sandwich panel (HC/SP), due to the channeling effect, is investigated by using smoothed particle hydrodynamics (SPH). Before performing numerical study on the channeling effect, the results of 2-D hypervelocity impact simulation were compared with the experimental results in order to validate the simulation along with searching proper simulation parameters such as equation of state, strength model, and failure model. And then, analysis of the channeling effect is performed with different cell sizes, wall thicknesses and depths of HC core in accordance with the specifications of commercial HC cores. As a result of the study, it is revealed that the HC core cell size is the most influential parameter on the channeling at normal incidence impact. Also the critical projectile diameter appears to be varying with the HC core cell size due to the channeling effect. Finally, the modified critical projectile diameters of HC/SP are presented by introducing the decrement of critical projectile diameter.

This study is concerned with the deformation or warping behavior of thin layered semiconductor structure comprising a silicon substrate, a pattern layer and a polyimide coating layer with its thickness varying from 100um to 50 um. In contrast with the conventional thick semiconductor structure, today’s semiconductor structure is increasingly thin and therefore the warping is extremely conspicuous, being among the major concerns in the structural design of a chip. In the view of thermomechanical analysis of an extremely thin layer structure considered in the present paper, a few parameters on the deformation should be taken into consideration such as the pattern layer and intrinsic stress. To account for the effect of the pattern layer, we make a well educated guess for the mechanical properties, employing the test results and the CBA (Composite Beam Analysis) theory. In addition, we take into consideration the effect of the intrinsic stress due to moisture absorption on deformation. We show that the chip warpage is accurately predicted when all these are properly considered. Furthermore, we have found that the local instability or wrinkling, associated with the nonuniformity or the inhomogeneity in material properties and bonding quality between any two neighboring layers, appears as one important mode of energy relaxation in addition to the overall warpage when the chip thickness becomes very small.

As an increasing number of polymer dielectric layers were laminated, the maximum bow values were measured layer by layer using a laser profilometry during thermal cycling. In the lamination process, a polymeric film is overlaid on a silicon substrate using a polymeric adhesive. Since the lamination process uses relatively thick polymer films, the classical stress analyses assuming infinitesimally thin films, are no longer effective. In this letter, a simple model based on the composite beam theory is presented to analyze the experimental results, and compared with the well-known Stoney’s formula. The thermomechanical behavior of the laminated multilayer polymer films on a silicon substrate was better described by the proposed model, while an error as much as 30% was involved using Stoney’s formula. The model can be applied for the design and fabrication of multilayer multichip module substrates.

In this paper, the two-way shape memory effect (TWSME) in a Ti?54.5 Ni(wt%) alloy was investigated experimentally to develop a NiTi linear actuator. The two-way shape memory effect was induced through a compressive shape memory cycle comprising four steps: (1)?loading to maximum deformation; (2)?unloading; (3)?heating; and (4)?cooling. Six types of specimen (one solid cylindrical and five tubular) were used to obtain the two-way shape memory strain and two-way recovery stress and to evaluate the actuating capacity. The two-way actuating strain showed a saturated tendency after several training cycles for the same maximum deformation. A maximum value of the two-way strain was obtained for 7% of maximum deformation, independently of the geometry of the tubular specimens. The two-way strains obtained by the shape memory cycles and two-way recovery stress linearly increase as a function of the maximum deformation and the two-way strain, respectively, and the geometry of specimen affects the two-way recovery stress. Although the results show that sufficient recovery stress can be generated by either the two-way shape memory process or by the one-way shape memory process, the two-way shape memory process can be applied more conveniently to actuating applications.