T. Kevin O'Brien

Comparison of 2D Finite Element Modeling Assumptions with Results from 3D Analysis for Composite Skin-Stiffener Debonding

The influence of two-dimensional finite element modeling assumptions on the debonding prediction for skin-stiffener specimens was investigated. Geometrically nonlinear finite element analyses using two-dimensional plane-stress and plane-strain elements as well as three different generalized plane-strain type approaches were performed. The computed skin and flange strains, transverse tensile stresses and energy release rates were compared to results obtained from three-dimensional simulations. The study showed that for strains and energy release rate computations the generalized plane-strain assumptions yielded results closest to the full three-dimensional analysis. For computed transverse tensile stresses the plane-stress assumption gave the best agreement. Based on this study it is recommended that results from plane-stress and plane-strain models be used as upper and lower bounds. The results from generalized plane-strain models fall between the results obtained from plane-stress and plane-strain models. Two-dimensional models may also be used to qualitatively evaluate the stress distribution in a ply and the variation of energy release rates and mixed mode ratios with delamination length. For more accurate predictions, however, a three-dimensional analysis is required.

Exploratory Investigation of Failure Mechanisms in Transition Regions Between Solid Laminates and X-cor Truss Sandwich

Small sub-component specimens consisting of solid laminates at the ends that transition to X-cor® truss sandwich in the center, were tested in a combination of three point bending, uni-axial tension, and combined tension and bending. The failure process in the transition region was documented for each loading using digital video and high-resolution cameras. For the three-point bending tests, most of the deformation occurred in the solid laminate regions on either end of the specimen. Some pin debonding from the skin of the X-cor® truss sandwich was observed in the transition region and was accompanied by audible “pings” throughout the loading. Tension loaded specimens failed in the sandwich skin in the middle of the gage length, accompanied by separation of the sandwich core from the back skin and by delamination between the top skin and bottom skin at the transition region. The pinging associated with pin debonding occurred as the load was increased. However, the frequency of the pinging exceeded any visual observations of pin debonding in the video of the transition region. For specimens tested in combined tension and bending, the greatest amount of pinging occurred during initial application of the axial load. High-resolution images in the transition region indicated that the pinging corresponded to pins debonding and buckling due to the through-thickness Poisson contraction of the specimen. This buckling continued to a much smaller extent as the transverse load was applied.

Fatigue Delamination Behavior of PEEK Thermoplastic Composite Laminates

Edge delamination tension fatigue tests were conducted on AS4/PEEK composite lami nates to determine the fatigue delamination behavior of these graphite reinforced, semi crystalline thermoplastic matrix composites. The strain at onset of edge delamination, ∈ c, observed during fatigue tests of (35n/ — 35 n/0n/90n)s laminates, where n = 1,2, was plotted as a function of fatigue cycles, N. The delamination onset strain decreased dramatically with fatigue cycles and then began to level off to an endurance limit at 106 cycles. The delamination onset strains were used to calculate critical strain energy release rate, Gc, values as a function of fatigue cycles to provide a more generic representation of the fa tigue delamination behavior of the material, i.e., one that is independent of layup and ply thickness. Although the static interlaminar fracture toughness of the AS4/PEEK compos ite is much greater than the toughness of graphite epoxy composites, the delamination fa tigue Gc threshold, calculated from the cyclic strain endurance limit at 106 cycles, was only slightly greater than the delamination fatigue threshold for graphite epoxy composites. This comparison, however, was based only on the contribution of mechanical strain to the strain energy released. An asymmetric (0 2/906)T AS4/PEEK curved panel was heated in an oven to determine the stress free temperature of the composite, and quantify the contri bution of residual thermal stresses to the strain energy release rate at delamination onset. The contribution of residual thermal stresses to delamination in AS4/PEEK was substan tial due to the large temperature range between manufacture and room temperature. More work is needed to fully characterize the influence of residual thermal and moisture stresses, and the degree of crystallinity of the PEEK matrix, on the strain energy release rate for delamination growth, and the interlaminar fracture toughness of the composite.

Test and Analysis of Composite Hat Stringer Pull-off Test Specimens

Hat stringer pull-off tests were performed to evaluate the delamination failure mechanisms in the flange region for a rod-reinforced hat stringer section. A special test fixture was used to pull the hat off the stringer while reacting the pull-off load through roller supports at both stringer flanges. Microscopic examinations of the failed specimens revealed that failure occurred at the ply termination in the flange area where the flange of the stiffener is built up by adding 45/-45 tape plies on the top surface. Test results indicated that the as-manufactured microstructure in the flange region has a strong influence on the delamination initiation and the associated pull-off loads. Finite element models were created for each specimen with a detailed mesh based on micrographs of the critical location. A fracture mechanics approach and a mixed mode delamination criterion were used to predict the onset of delamination and the pull-off load. By modeling the critical local details of each specimen from micrographs, the model was able to accurately predict the hat stringer pull-off loads and replicate the variability in the test results.

Analysis of Local Delaminations Caused by Angle Ply Matrix Cracks

Two different families of graphite/epoxy laminates with similar layups but different stacking sequences, (0θ/-θ) s and (-θ/θ/0) s , laminates, were analyzed using three-dimensional finite element analysis for 0 = 15 and 30 degrees. Delaminations were modeled in the -θ/θ interface, bounded by a matrix crack and the stress free edge. The total strain energy release rate, G, along the delamination front was computed using three different techniques: the virtual crack closure technique (VCCI), the equivalent domain integral (EDI) technique, and a global energy balance technique. The opening fracture mode component of the strain energy release rate, G1, along the delamination front was also computed for various delamination lengths using VCCT. Although the finite element model did not have an orthogonal mesh, VCCT still yielded accurate results which were in agreement with the global energy balance and yielded similar G distributions across the delamination front as the EDI technique. For both layups analyzed, the matrix crack length influenced the magnitude of G for delamination. Furthermore, the opening mode, G., was greatest near the matrix crack and decreased near the free edge. The laminate stacking sequences with a matrix crack in the surface angle ply had a greater GI value than the laminate stacking sequences with an angle ply matrix crack in the interior of the specimen thickness. This is consistent with test results in the literature that show delamination occurs earlier in the fatigue life of laminates with matrix cracks in the surface plies than in the interior plies.

Combined tension and bending testing of tapered composite laminates

A simple beam element used at Bell Helicopter was incorporated in the Computational Mechanics Testbed (COMET) finite element code at the Langley Research Center (LaRC) to analyze the responce of tappered laminates typical of flexbeams in composite rotor hubs. This beam element incorporated the influence of membrane loads on the flexural response of the tapered laminate configurations modeled and tested in a combined axial tension and bending (ATB) hydraulic load frame designed and built at LaRC. The moments generated from the finite element model were used in a tapered laminated plate theory analysis to estimate axial stresses on the surface of the tapered laminates due to combined bending and tension loads. Surfaces strains were calculated and compared to surface strains measured using strain gages mounted along the laminate length. The strain distributions correlated reasonably well with the analysis. The analysis was then used to examine the surface strain distribution in a non-linear tapered laminate where a similarly good correlation was obtained. Results indicate that simple finite element beam models may be used to identify tapered laminate configurations best suited for simulating the response of a composite flexbeam in a full scale rotor hub.

Delamination of Composite Materials

Abstract Fracture mechanics has been found to be a useful tool for understanding composite delamination. Analyses for calculating strain energy release rates associated with delamination growth have been developed. These analyses successfully characterized delamination onset and growth for particular sources of delamination. Low velocity impact has been found to be the most severe source of composite delamination. A variety of test methods for measuring interlaminar fracture toughness are being developed to identify new composite materials with enhanced delamination resistance.