Advanced Technology Composite Fuselage—Structural Performance

Walker, T.H., Minguet, P.J., Flynn, B.W., Carbery, D.J., Swanson, G.D., and Ilcewicz, L.B.

NASA CR-4732, 1997.

Boeing is studying the technologies associated with the application of composite materials to commercial transport fuselage structure under the NASA-sponsored contracts for Advanced Technology Composite Aircraft Structures (ATCAS) and Materials Development Omnibus Contract (MDOC). Innovative designs, advanced material forms, and automated processes with the desired cost-savings potential are being pursued. This report addresses the program activities related to structural performance of the selected concepts, including both the design development and subsequent detailed evaluation.

Design criteria were developed to ensure compliance with regulatory requirements and typical company objectives. During design sizing, attempts were made to account for the most significant aspects of each design. Accurate analysis methods were selected and/or developed where practical, and conservative approaches were used where significant approximations were necessary. Design sizing activities supported subsequent development by providing representative design configurations for structural evaluation and by identifying the critical performance issues.

Significant program efforts were directed towards assessing structural performance predictive capability. The structural database collected to perform this assessment was intimately linked to the manufacturing scale-up activities to ensure inclusion of manufacturing-induced performance traits. Mechanical tests were conducted to support the development and critical evaluation of analysis methods addressing internal loads, stability, ultimate strength, attachment and splice strength, and damage tolerance. Unresolved aspects of these performance issues were identified as part of the assessments, providing direction for future development.

Activities performed within the ATCAS program have identified and, to varying degrees, addressed critical structural performance issues for composite fuselage concepts. The program’s central design process forced the development and implementation of an acceptable design methodology, including criteria and sizing methods. Structural tests of the developed designs, and of attractive alternatives, have provided a large database for validating the design methodology, and assessing the accuracy of detailed predictive methods over a representative design space. the structural evaluations were closely coupled with the manufacturing assessments to ensure consideration of process-induced performance characteristics.

Design Criteria. In general, acceptable design criteria for composite fuselage structure exist, and are extensions of those developed for previous composite applications. However, improvements in several areas are needed. Specifically, major developments are needed to determine appropriate criteria for (a) stiffness and weight effects on vehicle handling and flutter, (b) ultimate strength and stability with partially disbonded frames and sandwich facesheets, (c) shimming and pull-up requirements, and (d) damage-size requirements for damage tolerance of sandwich configurations. Relatively minor developments are necessary to address criteria for (a) skin postbuckling and stringer column buckling of skin/stringer configurations, and (b) critical damage conditions associated with ultimate strength.

Internal Loads. Analysis methods for predicting general internal load distributions in fuselage structure are well established for metallic materials, and no major obstacles are expected in extending these methods to composite fuselage structure. Large-scale verification of these methods is necessary. Simplified methods and modeling guidelines for simulating the effects of nonlinear phenomenon on internal load distributions (e.g., reduced effective skin stiffness due to postbuckling, distance over which added plies become fully effective) that are compatible with the computationally-intensive preliminary design environment have not been developed. Similarly, the effect of non-local material response on load distributions in the vicinity of cutouts must be experimentally assessed, and appropriate methods developed as necessary for including these effects in loads modeling activities.

Stability. Detailed analysis methods for addressing compression stability of composite structures are well established and verified. These methods accurately predicted the experimental load levels at the onset of instability and the associated mode shapes for a range of scales and configurations. Similar approaches for addressing shear and combined compression/shear stability must be developed and verified. Methods for including the effect of impact damage on stability response must also be developed and verified. Modeling of the damage as a soft inclusion, either explicitly or through a strain-softening approach, appears to be attractive in this regard. Closed-form methods addressing idealized boundary conditions also exist, and are compatible with the needs of design sizing activities. However, calibration of these methods to reflect actual response of fuselage structural arrangements is needed. This calibration can likely be accomplished primarily through detailed structural analyses, augmented by limited experiments.

Ultimate Strength. Detailed predictive methods for ultimate strength are not as well developed. General criteria for predicting damage onset and growth are needed for in-plane and through-thickness concentrations, and combined loadings. Prediction of load redistribution caused by the damage is an essential component to accurate predictions of structural failure. Finite element models incorporating strain-softening material laws have demonstrated the ability to predict structural failures, although only with initial damage located remote to complex design details. Similar capability is needed for the more difficult locations.

In general, structural tests have demonstrated adequate ultimate load capability for the ATCAS design concepts, within the bounds of the conditions tested. The premature failure of the pressure-box panel subjected to high axial loads with internal pressure was strongly influenced by load introduction effects. This emphasizes the importance of extending these test responses to the behavior within the true fuselage shell. Within these limitations, however, the successful demonstration of ultimate strength has validated the methodologies used in developing the designs. Conditions requiring more rigorous validation include impact damage, environment, and the statistical variation of structural failure.

Attachment and Splice Strength. Structural tests have demonstrated the integrity of the bonded attachment of the circumferential frames and stringers to the skin, although the effects of impact damage, environment, and cyclic loading have not been adequately addressed. Specifically, failures only occurred along the bondline when the skin contained a fabric surface ply; otherwise, element separation was controlled by the transverse strength of the skin laminate. Strength-of-materials criteria provide a first approximation of damage initiation. Detailed finite element models in conjunction with energy-based failure criteria can be used to predict bonded element separation, but are costly and/or of limited scope. Approaches are needed for addressing (a) damage propagation through the skin laminate thickness and (b) loading/geometry changes along the length of the element. Failure criteria must also be validated, and integrated into a generalized methodology addressing interactions between bondline stresses and skin concentrations. Methods for sizing mechanical connections within quadrants appear adequate.

Detailed methods for assessing performance of splices exist, though no splice tests of significant size have been conducted to verify the ATCAS structural arrangements. Design details of specific concern include the splicing of closed-section hat stringers and full-depth sandwich close-outs. Methods for predicting load distribution exist, but must be verified for ATCAS configurations. Moreover, methods to address the redistribution associated with local damage formation are needed. Splice development efforts must remain closely tied to manufacturing progress to ensure consideration of expected part tolerances.

Damage Tolerance. Extensive experimental data has augmented the development of damage tolerance predictive methods. This data suggests that strain-softening and non-local material responses are active in damage tolerance scenarios. The success of strain-softening analytical models in accurately scaling coupon results to structural configurations, and the dependency of these laws on element size, provide strong support for this conclusion. Additional development, however, is needed to address efficient determination of the material laws and the degradation of load transfer to undamaged structural elements. Development of analytical capability to address non-local material response is also needed to improve predictions of damage-induced load gradients.

Axial damage tolerance was experimentally demonstrated for both the crown and the aft keel. Hoop damage tolerance was demonstrated for the perceived critical loading condition (i.e., internal pressure only) for the skin-stiffened crown with a bolted frame/skin attachment; however, the ability of bonded frames to arrest damage under these conditions was not clearly demonstrated for the crown designs. The effectivity of bonded elements in arresting damage growth, therefore, requires additional analytical and experimental evaluation. Graphite-glass intraply hybrid skin concepts were also demonstrated to have vastly superior damage tolerance performance, which can potentially eliminate damage tolerance as a design driver and increase inspection intervals. Major issues requiring additional experimental data include the effects of damage variables (e.g., shape, location, and orientation), combined loads, dynamic events, cyclic loading of large damage, and environment.the delaminated plies must be considered in addition to those for damage with dents. Existing analysis methods form the basis for predicting (a) sublaminate buckling, (b) facesheet fracture due to the buckled sublaminate, and (c) static delamination growth, all as a function of damage size and delamination depth. These predictions are compared to the applied loading to preclude static failure and fatigue delamination growth. Conservative techniques are employed to address limitations of in-service inspection methods relative to identifying delamination depth and the existence of multiple delaminations.

An implementation strategy to support in-service use of these strength predictions was developed. A number of damage size and spacing measurements are necessary to support the above methods. SRM allowable damage data for one zone of a specific aircraft part include one plot addressing single-site damage, while a second plot provides a correction factor for multi-site damage scenarios.

The single-site ADL curve results from the hole/impact residual strength capability, as well as consideration of delaminations without dents. The left-hand portion of the curve is based on the impact and hole residual strength predictions. The curve is generated by determining the damage size that results in failure at the applied Ultimate strain from the residual strength curves for the specific layup. The right-hand portion of the curve reflects the results for delaminations with no dent. These constant values are a function only of delamination depth. Each line represent the smallest of the damage sizes associated with the failure modes related to sublaminate buckling (facesheet fracture, static and fatigue delamination growth) for the applied strains and the delamination depth indicated.

The central portion of the single-site ADL curve results from consideration of facesheet springback, which results when a relatively undamaged facesheet either partially or fully returns to its initial position, despite the damaged core. Evaluations of impact damage surveys indicate that springback occurs only in less-severe damage scenarios, and that the severity level below which springback might occur is a function of facesheet thickness. The residual strength at this severity threshold is used as an upper limit for less severe damages. However, damage depths associated with facesheet springback might also relax over time when subjected to environmental and mechanical cycling associated with service. Damages where this is considered to be a possibility are therefore conservatively treated as a delamination with no dent, except in cases where intermediate inspections can and will be performed to monitor the damage for growth.

Multi-site corrections are based on single-site strength reductions due to interactions between pairs of damages. Specifically, isotropic stress concentration factors were found to reasonably predict the interactions observed in test data, and are therefore used to define strength knock-down factors based on damage spacing, relative damage size, and loading orientation. The correction factors for damage size are determined from these strength-based knock-downs, and are dependent on layup due to the dependence of the notch sensitivity on layup. Severity differences between damages within a pair are treated conservatively by assuming both damages are of the greater severity of the two damages.

In-service usage is necessarily simple. The operator obtains the required damage size and spacing measurements. For single-site damage, the damage severity (d2/y) is calculated, then the allowable damage limit is obtained from the upper curve. If the maximum size of the actual damage is less than the ADL, and all spacing requirements are met, it is acceptable. Otherwise, the damage must be repaired.

For multi-site damage, each unique damage pair are treated separately. The severity of both damages in a pair are calculated, and the smaller of the two is used to determined a single-site ADL from the upper curve. Spacing and diameter ratios (S/D3 and d3/D3, respectively) are calculated, and used to obtain the multi-site diameter correction factor (MDCF) for each of the smaller and larger damages from the lower plot. The multi-site ADLs are determined for each damage as the product of the single-site ADL and the respective MDCF. If the maximum size of the actual damages are both less than their respective ADLs, and all spacing requirements are met, they are acceptable. Otherwise, both damages must be repaired.