skip to main content

In this paper, the effect of matrix viscoelasticity on the development of residual stresses in 3D woven composites is investigated using Finite Element Analysis. Based on experimental observations, it is hypothesized, that the stresses develop mainly due to the difference in the coefficients of thermal expansion between the fiber reinforcement and the matrix. The model considered is a “1x1 orthogonal” 3D woven composite unit cell that is generated using x-ray computed microtomography data. In this study, cooling after curing is considered under the assumption of zero stress at the beginning of the cooling. In addition to the full time- and temperature-dependent viscoelastic formulation, the applicability of two simplified constitutive methods, elastic and variable time pseudoviscoelastic, is investigated. It is observed that the pseudo-viscoelastic method predicts similar cumulative stress distribution (Von Mises and Hydrostatic) compared to the full viscoelastic results. The elastic model presented the highest stress values while the full viscoelastic model presented the lowest stress values.
; ; ; ;
Award ID(s):
Publication Date:
Journal Name:
Proceedings of NUMIFORM 2019: The 13th International Conference on Numerical Methods in Industrial Forming Processes
Sponsoring Org:
National Science Foundation
More Like this
  1. Manufacturing-induced residual stresses in carbon/epoxy 3D woven composites arise during cooling after curing due to a large difference in the coefficients of thermal expansion between the carbon fibers and the epoxy matrix. The magnitudes of these stresses appear to be higher in composites with high throughthickness reinforcement and in some cases are sufficient to lead to matrix cracking. This paper presents a numerical approach to simulation of development of manufacturing-induced residual stresses in an orthogonal 3D woven composite unit cell using finite element analysis. The proposed mesoscale modeling combines viscoelastic stress relaxation of the epoxy matrix and realistic reinforcement geometry (based on microtomography and fabric mechanics simulations) and includes imaginginformed interfacial (tow/matrix) cracks. Sensitivity of the numerical predictions to reinforcement geometry and presence of defects is discussed. To validate the predictions, blind hole drilling is simulated, and the predicted resulting surface displacements are compared to the experimentally measured values. The validated model provides an insight into the volumetric distribution of residual stresses in 3D woven composites. The presented approach can be used for studies of residual stress effects on mechanical performance of composites and strategies directed at their mitigation.
  2. Three panels of 3D woven carbon fiber/RTM6 epoxy composites with a ply-to-ply weave with 12x12 (warp/weft) picks per inch (ppi), 10x12 ppi, and 10x8 ppi were fabricated by resin transfer molding. Realistic finite element models of each weave architecture were constructed using Dynamic Fabric Mechanics Analyzer. The resin properties were isotropic and linear elastic and dependent on temperature. The resin-infiltrated fiber tow properties were estimated using homogenization based on Hashin and Shapery formulas. The model was considered to be at zero stress at the 165C curing temperature. The stresses resulting from cooling the composite to 25C were estimated using the resin temperature-dependent properties and the temperature independent properties of the tows. The displacement fields resulting from holes drilled through the middle of the top warp or weft yarn were estimated by virtually drilling a hole in the finite element model and were measured on the specimens using electronic speckle pattern interferometry. In general, the measured displacements transverse to the yarn were lower than the predicted displacements. This suggests the resin in the infiltrated yarns relieves some of the stress by permanently deforming during cooling. The measured displacements along the yarn were approximately the same for the 12x12 ppi,, lower formore »the 10x12 ppi, and significantly higher for the 10x8 ppi.« less
  3. 3D woven carbon/epoxy composites are often produced using resin transfer molding technique which includes epoxy curing at elevated temperatures. The process may lead to accumulation of the intrinsic residual stresses during cooling of the material caused by the mismatch between carbon and epoxy coefficients of thermal expansion. This paper deals with implementation of mesoscale finite element models to evaluate intrinsic residual stresses in 3D woven composites. The stresses are determined by correlation of the surface displacements observed after drilling 1-mm diameter blind holes with the corresponding predictions of the models. We investigated how a numerical representation of the composite plate surface affects the correlation between the experimental measurements and numerical predictions and how it influences the evaluation of the process-induced residual stresses. It has been shown for ply-to-ply woven composites with different pick spacing that the absence of the resin layer leads to more accurate interpretation of the experimental measurements. The prediction of the average residual stress in the matrix phase of the composite was found to be sensitive to the surface representation accuracy, however, the residual stress magnitude and distribution was not affected fundamentally.
  4. Intrinsic residual stresses in woven composites result from the coefficient of thermal expansion mismatch between the fibers and the matrix. Extrinsic residual stresses result from large scale thermal gradients during curing and cooling. Intrinsic residual stresses in 3D woven composites are sometimes severe enough to cause micro-cracking in the matrix. They are also expected to impact the fatigue resistance and the impact resistance. To the best of our knowledge, there have been no spatially resolved measurements of the intrinsic residual stress field as a function of position in the repeating weave pattern. We used digital image correlation (DIC) and electronic speckle pattern interferometry (ESPI) to measure the surface displacement field resulting from drilling a 1 mm diameter hole at four selected locations in two different 3D woven composite architectures that represent low and high through-the-thickness constraint. The two methods are used because the displacements sometimes on the lower end of the resolution for the DIC method and the displacement gradients are sometimes too steep to resolve the fringes for the ESPI method. Finite element models constructed with realistic fiber geometry using Dynamic Fabric Mechanic Analyzer software were utilized to estimate the residual stress field from cooling from the curing temperature.more »Holes were manually inserted by deactivating the elements in the hole region and the resultant displacement fields were compared to the measurements. In general, the measured displacement fields were lower in magnitude than the model predictions. In some cases, the sign of the predicted displacement field is opposite to the observed field which could be attributed to differences between the actual hole location and the hole in the model.« less
  5. Introduction: The mechanical stability of an atheroma fibrous cap (FC) is a crucial factor for the risk of heart attack or stroke in asymptomatic vulnerable plaques. Common determinants of plaque vulnerability are the cap thickness and the presence of micro-calcifications (µCalcs). Higher local stresses have been linked to thin caps(<65µm) and, more recently, our lab demonstrated how µCalcs can potentially initiate cap rupture [1-3]. When combined, these two factors can compromise to a greater extent the stability of the plaque. On this basis, we quantitatively analyzed both individual and combined effects of key determinants of plaque rupture using a tissue damage model on idealized atherosclerotic arteries. Our results were then tested against a diseased human coronary sample. Methods: We performed 28 finite element simulations on three-dimensional idealized atherosclerotic arteries and a human coronary sample. The idealized models present 10% lumen narrowing and 1.25 remodeling index (RI)(Fig.1A). The FC thickness values that we considered were of 50, 100, 150 and 200µm. The human coronary presents a RI=1.31, with 31% lumen occlusion and a 140µm-thick cap(Fig.1B). The human model is based on 6.7μm high-resolution microcomputed tomography (HR-μCT) images. The µCalc has a diameter of 15µm and each artery was expanded up tomore »a systolic pressure of 120mmHg. Layer-specific material properties were de-fined by the HGO model coupled with the hyperelastic failure description proposed by Volokh et al. [4] to repli-cate the rupture of the FC. We considered a max. princi-pal stress for rupture of 545kPa[5]. The lipid core and the µCalc were considered as elastic materials (Ecore = 5kPa, νcore = 0.49; EµCalc= 18,000 kPa, νµCalc=0.3). To obtain a detailed analysis of the cap stresses and rupture progres-sion, a sub-modeling approach was implemented using ABAQUS (Dassault Systemes, v.2019) (Fig. 1). Results: We investigated the quantitative effect of cap thickness and µCalc by simulating tissue failure and de-riving a vulnerability index (VI) for each risk factor. The VI coefficient was defined as the peak cap stress (PCS) normalized by the threshold stress for rupture (545kPa). The relationship between the risk factors and VI was de-termined by deriving the Pearson’s correlation coefficient (PCC) followed by one-tailed t-test (SPSS, IBM, v.25). The null hypothesis was rejected if p<0.05. The presence of the µCalc is the factor that manifests the greater impact on cap stability, leading to at least a 2.5-fold increase in VI and tissue rupture regardless of cap thickness (Fig.2A,B). One µCalc in the cap is the first predictor of vulnerability, with PCCµCalc=0.59 and pµCalc=0.001. Our results also confirm the substantial in-fluence of cap thickness, with an exponential increase in stresses as the cap becomes thinner. The 50µm cap is the only phenotype that ruptures without µCalc (Fig2A). The human sample exhibits PCS levels that are close to the idealized case with 150µm cap and it doesn’t rupture in the absence of the µCalc (PCShuman=233kPa, PCSideal= 252kPa). Conversely, the phenotypes with the µCalc showed an increase in VI of about 2.5 and reached rup-ture under the same blood pressure regime. Conclusions: Our results clearly show the multifactorial nature of plaque vulnerability and the significance of micro-calcifications on the cap mechanical stability. The presence of a μCalc strongly amplifies the stresses in the surrounding tissue, and it can provoke tissue failure even in thick caps that would otherwise be classified as stable. Clearly, plaque phenotypes with a thin cap and μCalcs in the tissue represent the most vulnerable condition. Finally, these observations are well validated by the case of the human atherosclerotic segment, which closely compares to its corresponding idealized model. The novel imple-mentation of the tissue damage description and the defi-nition of a vulnerability index allow one to quantitatively analyze the individual and combined contribution of key determinants of cap rupture, which precedes the for-mation of a thrombus and myocardial infarction.« less