Walker, T.H., Ilcewicz, L.B., Bodine, J.B., Murphy, D.P., and Dost, E.F.
NASA CR-194969, 1994.
Benchmark Panel Fabrication. Five curved stiffened panels representative of ATCAS fuselage crown design alternatives were fabricated to provide test specimens for the pressure-box test fixture and for frame/skin bondline strength evaluations. These panels augmented ATCAS pressure-box panels by providing an opportunity for tests of alternate damage scenarios (e.g., large circumferentially-oriented notches, barely visible impact damage), attractive alternate design concepts, and other panel locations.
The designs were characterized by a 122 in. radius, tow-placed skins, hot-drape-formed hat-section stringers, and braided/RTM J-section frames. Attractive design alternatives contained in the panel designs included intraply hybrid skins, higher-stiffness stringers, and bolted frames with smaller mouseholes.
Pulse-echo inspection of the panels revealed no anomalies, however a repeatable pattern was observed in the skins of all panels. The pattern was hypothesized to be a result of thickness variations caused by consistent non-uniformities in band thickness being magnified when the bands of similarly-oriented plies were stacked.
Pulse-echo inspection of the braided frames indicated localized defective regions in the inner chord of six frames. Photomicrographs taken in the defect areas of the frame trim revealed microcracks at the laminate midplane. Three of the frames were trimmed such that all defect regions fell in the trim areas. Fasteners were installed in the inner frame chord of the other three to prevent any delamination growth during testing. The microcracks were hypothesized to be caused by increased residual stresses due to a higher percentage of 0° yarns and a somewhat higher fiber volume fraction than in the other twelve frames.
Load-introduction doublers were fabricated for each pressure-box test panel and provided, with appropriate fasteners, to NASA Langley Research Center for installation. Doubler designs varied with specific panel configuration and loading requirements. Slots were included in all between-frame doublers and the longer between-stringer doublers to maximize decoupling of axial and hoop loads.
Benchmark Panel Finite Element Analysis. Geometrically nonlinear finite element analysis was conducted to support the development of test fixturing, and provide predictions of undamaged panel response. The model utilized shell, brick, and beam elements to properly simulate the panel and load introduction fixturing. The undamaged model contained 2,500 elements and 13,900 degrees-of-freedom. Results were used to refine panel edge-doubler designs to minimize their influence on the central panel bays. Successful nonlinear analyses of panels with high axial load requirements could not be obtained, despite significant efforts. Forces normal to the panel at the hoop-edge grips were observed to reverse sign with each iteration.
Comparisons of similar ATCAS models with test results implied that the representation of the test fixturing requires improvements. Higher strains in the central stringers and higher radial deflection than predicted suggest that the stiffness of the hoop restraints may be the cause.
Strain Softening Model Development. General shell elements and an orthotropic strain softening material model were used to analyze laminate test results from the ATCAS fracture database, concentrating on uniaxial load conditions perpendicular to the axis of blunt notches. Parametric studies indicated that the maximum stress level and initial unloading portions of a strain softening curve have the strongest effect on small notch strength for coupon sizes typically used for composites. The energy associated with the rest of the strain softening curve is most important to large-notch strengths, due to larger notch tip damage zones allowed by larger specimen widths. These analyses confirmed the experimental observation that small fracture coupons do not provide material properties.
Comparison of strain softening analyses with models based on linear elastic fracture mechanics indicated the superiority of the strain-softening approach in predicting the ATCAS test database, including specimen size effects and residual strengths over a wide range of notch sizes. Differences in strain-softening response can also explain differences observed between tow-placed and tape laminates.
Frame/Skin Bondline Analysis and Test. Bonded frame pulloff tests and analyses were performed to support the baseline crown design including a mousehole detail at the intersection between frames and hat stiffeners. Based on the finite element analysis of the ATCAS crown panel under an internal pressurization load, two critical areas for the frame-to-skin bond were identified: along the length of the frame near the middle of bay, and the gap between the frame and the stiffener in the mousehole area.
Initial tests for conditions in the middle of the frame (away from the mousehole) were performed with a clamped fixture that has been used in the past to evaluate bonded element pulloff. Results from these tests indicated that the type of ‘noodle’ material used at the end of the frame web, adhesive or braided inserts, had little effect on failure. In most clamped fixture test cases, damage started to propagate unsymmetrically from each of the frame edges but subsequent noodle damage initiated and became the dominant mode leading to final failure. Membrane forces that occur in the clamped specimen as pulloff loads increase may have been the cause for a shift in the dominant damage location. Detailed finite element analysis used to predict damage initiation in clamped specimens compared reasonably well with experiments.
A simply supported test fixture was also designed to determine frame pulloff failure loads and modes for areas away from the mousehole. This improved fixture was statically determinate. In most test cases with the simply supported fixture, damage initiated and propagated unsymmetrically from each of the frame edges leading to final failure at loads less than that observed with the clamped fixture. Some of the specimens tested exhibited manufacturing anomalies with the frame sinking into the skin and causing some waviness in the top plies. This often caused delaminations to propagate deeper into the skin and resulted in lower failure loads. Based on the results, it appears that the frame pull-off load itself is not the best way to report data but that the skin bending moment at the tip of the frame was the most important load resultant. In view of the static data obtained and the predicted skin bending moments from the finite element analysis, it appears that an adequate margin of safety is obtained for bonded frame pulloff away from the mousehole. Preliminary fatigue tests also indicated that adequate margins exist with the current design.
A frame pulloff test fixture was also designed for testing the mouseholed design detail. For these tests, failure originated under the frame noodle near the stiffener and propagated as a delamination under the frame. Based on the results of the finite element analyses, it would appear that the loads obtained in the test of the mousehole region are not sufficient to reach design ultimate load, although more refined local analysis and test instrumentation may be required to confirm this.