Engineering topics
Publications and source records attributed to Cox, B. N..
This viewgraph presentation provides an overview of research being conducted on the use of C-SiC composite structures for actively cooling rocket nozzles. Potential payoffs and design constraints are discussed. Other topics covered include: testing parameters, material selection, thermal analysis of joined tube structure, pressure containment, H2O2 combustion testing, and cooled re-entry.
The goals of this research program were to: (1) determine how microstructural factors, especially the architecture of reinforcing fibers, control stiffness, strength, and fatigue life in 3D woven composites; (2) identify mechanisms of failure; (3) model composite stiffness; (4) model notched and unnotched strength; and (5) model fatigue life. We have examined a total of eleven different angle and orthogonal interlock woven composites. Extensive testing has revealed that these 3D woven composites possess an extraordinary combination of strength, damage tolerance, and notch insensitivity in compression and tension and in monotonic and cyclic loading. In many important regards, 3D woven composites far outstrip conventional 2D laminates or stitched laminates. Detailed microscopic analysis of damage has led to a comprehensive picture of the essential mechanisms of failure and how they are related to the reinforcement geometry. The critical characteristics of the weave architecture that promote favorable properties have been identified. Key parameters are tow size and the distributions in space and strength of geometrical flaws. The geometrical flaws should be regarded as controllable characteristics of the weave in design and manufacture. In addressing our goals, the simplest possible models of properties were always sought, in a blend of old and new modeling concepts. Nevertheless, certain properties, especially regarding damage tolerance, ultimate failure, and the detailed effects of weave architecture, require computationally intensive stochastic modeling. We have developed a new model, the 'binary model,' to carry out such tasks in the most efficient manner and with faithful representation of crucial mechanisms. This is the final report for contract NAS1-18840. It covers all work from April 1989 up to the conclusion of the program in January 1993.
Composites with three-dimensional woven architectures exhibit large strains to failure when compared to composites made up of the same materials but not with three-dimensional interlocking tows. The fracture mechanics of such three-dimensional architectures is a subject requiring substantial investigation and experimental testing. Classical fracture mechanics concepts (for instance, an isolated defect in a homogeneous body) will not be applicable to the woven fracture test specimen. The use of an isolated singularity to characterize an entire specimen is inadequate when the density of defects is considerable and the material is heterogeneous. Modelling of such a complex system requires a great deal of insight and consideration as well as prudent choices of model sizes to make numerical schemes feasible. The purpose of this manuscript is to review our recently acquired knowledge of damage accumulation in woven composites and to describe a practicable model of the macroscopic behavior in these and other complex composite architectures based on such knowledge. In this manuscript, discussion will be limited to uniaxial compressive loading; considerations of general loading (monotonic and cyclic) will appear in a subsequent manuscript. Our modelling efforts may be briefly described as follows: the composite is subdivided into microstructural elements (microelements) in which the micromechanical modelling is either understood rigorously or can be represented adequately by statistical parameters. There can be microstructural elements for many different types of composite components, such as the various types of warp and weft and matrix for three-dimensional woven composites. The physical dimensions of microelements are made as large as possible while the response within the element can still be represented by a single micromechatlical calculation. The various elements are linked together(sometimes by associating distinct corners and edges, sometimes by superposition) in a pattern which resembles a particular weave architecture. The model can then be loaded in any manner and the linear and nonlinear elastic responses of representative weaves can be calculated. After the elastic regime, the fracture response is determined by monitoring the damage accumulation.
The present study of angle-interlock woven polymer matrix composite behavior under uniaxial monotonic compression notes these materials to be macroscopically ductile, with compressive strains in failure that can exceed 15 percent. Some of the tests conducted on stitched laminates indicated brittle behavior. Woven composite failure mechanisms have been studied by a combination of optical microscopy, moire interferometry, stereoscopy, and digital image comparisons; the critical failure event was in all cases a kink band formation in the primary load-bearing axial rows.
Stitched laminates and angle interlock woven composites have been studied in uniaxial, in-plane, monotonic compression. Failure mechanisms have been found to depend strongly on both the reinforcement architecture and the degree of constraint imposed by the loading grips. Stitched laminates show higher compressive strength, but are brittle, possessing no load bearing capacity beyond the strain for peak load. Post-mortem inspection shows a localized shear band of buckled and broken fibers, which is evidently the product of an unstably propagating kink band. Similar shear bands are found in the woven composites if the constraint of lateral displacements is weak; but, under strong constraint, damage is not localized but distributed throughout the gauge section. While the woven composites tested are weaker than the stitched laminates, they continue to bear significant loads to compressive strains of approx. 15 percent, even when most damage is confined to a shear band.
Elementary results are presented for the buckling of stitched, laminated composites containing delamination cracks. The stitching fibers are assumed to provide continuous, linear restoring tractions opposing the deflection of the delaminated layer adjacent to the crack. It is shown that there exists a characteristic length a(0) for buckling: if the length, 2a, of the delamination crack exceeds 2a(0), then, when buckling occurs, it will consist of waves of period 2a(0) and will usually not span the whole delamination. Simple expressions are derived for the critical buckling load and the minimum stitching density required to suppress buckling of the delaminated layer.
The far-reaching effects of such extrinsic factors as specimen shape and load distribution on bridged crack propagation, in cases where bridging-zone length is comparable to any of the crack and/or specimen dimensions, are presently demonstrated in view of calculation results for single-edge-notch specimens under uniform remote tension. The inherent risk of nonconservative predictions or reliability and strength may be reduced by considering the relationship between the bridging tractions and the crack-opening displacement as a fundamental material property. The fundamentality of the 'bridging length scale', or initial crack extension over which the bridging zone matures, is demonstrated.
The goal is to determine how microstructural factors, especially the architecture of microstructural factors, control fatigue damage in 3D reinforced polymer composites. Test materials were fabricated from various preforms, including stitched quasi-isotropic laminates, and through-the-thickness angle interlock, layer-to-layer angle interlock, and through-the-thickness stitching effect weaves. Preforms were impregnated with a tough resin by a special vacuum infiltration method. Most tests are being performed in uniaxial compression/compression loading. In all cases to date, failure has occurred not by delamination, but by shear failure, which occurs suddenly rather than by gradual macroscopic crack growth. Some theoretical aspects of bridging are also examined.
This paper presents a simple shear lag model of interfacial sliding at a free surface in a layered or continuous fiber composite. The interface is characterized by a critical interfacial shear stress, tau0, which might represent the critical stress for frictional sliding at a weakly bonded interface, or the shear flow stress of a thin ductile interface layer at a well bonded interface. The history, during heating and cooling, of the relative normal displacement of the reinforcing inclusions and the matrix on a free surface cut normal to the inclusions is calculated and shown to depend on both the absolute value and the temperature dependence of tau0, as well as on the magnitudes of the bulk residual stresses. The variety of possible displacement histories suggests that they are a rich source of information about tau0 and the residual stresses.
Both in-plane and out-of-plane displacements were determined in patches of the toughened 1962-ERLX epoxy bounded by three closely packed high-modulus P75S graphite fibers. The P75S/1962-ERLX composite was unidirectional, with an average fiber volume of 60 percent. On heating to 44 C from room temperature, the epoxy confined between clusters of three fibers sank in a trough beneath the plane of the fiber ends by 50-120 nm. When the maximum temperature achieved in the prior thermal cycles exceeded 100 C the depth of the trough increased. However, when cycles exceeding 100 C were followed by thermal cycles of decreasing amplitude, chosen to reduce interfacial stress relaxation, the depth of the trough decreased. The results illustrate the feasibility of deducing quantitative data on local deformation from high-resolution strain measurements on cut surfaces.