Search

A series of tests was conducted to measure the strength of stitched carbon/epoxy composites containing through-thickness damage in the form of a crack-like notch. The specimens were subjected to three types of loading: pure bending, pure tension, and combined bending and tension loads. Measurements of applied loads, strains near crack tips, and crack opening displacements (COD) were monitored in all tests. The transverse displacement at the center of the specimen was measured using a Linear Variable Differential Transformer (LVDT). The experimental data showed that the outer surface of the pure tension specimen failed at approximately 6,000 microstrain, while in combined bending and tension loads the measured tensile strains reached 10,000 microstrain.

The effect of stitches on the failure of a single lap joint configuration was determined in a combined experimental and analytical study. The experimental study was conducted to determine debond growth under static monotonic loading. The stitches were shown to delay the initiation of the debond and provide load transfer beyond the load necessary to completely debond the stitched lap joint. The strain energy release rates at the debond front were calculated using a finite element-based technique. Models of the unstitched configuration showed significant values of modes I and II across the width of the joint and showed that mode III is zero at the centerline but increases near the free edge. Models of the stitched configuration showed that the stitches effectively reduced mode I to zero, but had less of an effect on modes II and III.

The effect of stitches on the failure of a single lap joint configuration was determined in a combined experimental and analytical study. The experimental study was conducted to determine debond growth under static monotonic loading. The stitches were shown to delay the initiation ofthe debond and provide load transfer beyond the load necessary to completely debond the stitched lap joint. The strain energy release rates at the debond front were calculated using a finite element-based technique. Models of the unstitched configuration showed significant values of modes I and II across the width of the joint and showed that mode III is zero at the centerline but increases near the free edge. Models of the stitched configuration showed that the stitches effectively reduced mode I to zero, but had less of an effect on modes II and III.

The effect of stitches on the failure of a single lap joint configuration was determined in a combined experimental and finite element study. The experimental program was conducted to determine debond growth under static monotonic loading. The stitches were shown to delay the initiation of the debond and provide load transfer beyond the load necessary to completely debond the stitched lap joint. The experimentally determined debond length vs. applied load was used as an input parameter in the finite element analysis of both configurations. The strain energy release rates at the debond from were calculated using plate finite elements. Nonlinear fastener elements were used to model the stitches and multipoint constraints were used to model the contact problem. Models of the unstitched configuration showed significant values of modes I and II across the width of the joint and showed that mode III is zero at the centerline but increases near the free edge. Models of the stitched configuration showed that the stitches were effective in reducing mode I to zero, but had less of an effect on modes II and III.

A series of tests was conducted to support development of an analytical model for predicting the failure strains of stitched warp-knit carbon/epoxy composite materials with through-thicknesss damage in the form of a crack-like notch. Measurements of strain near notch tips, crack opening displacement (COD), and applied load were monitored in all tests. The out-of-plane displacement at the center of the notch was also measured when the specimen was subjected to bending. Three types of loading were applied: pure bending, pure tension, and combined bending and tension.

The effect of stitching on mode I and mode II strain energy release rates for debond configurations is studied using an analysis based on plate finite elements and the virtual crack closure technique. The stitches were modeled as discrete nonlinear fastener elements with a compliance determined by experiment. The axial and shear behavior of the stitches was considered with both the compliances and failure loads assumed to be independent. The mode I strain energy release rate, G(sub I), was shown to decrease once the debond had grown beyond the first row of stitches and was reduced to zero for long debonds, however, the mode II strain energy release rate, G(sub II), continued to be of significant magnitude over the range of debond lengths considered.

An analysis based on plate finite elements and the virtual crack closure technique has been implemented to study the effect of stitching on mode I and mode II strain energy release rates for debond configurations. The stitches were modeled as discrete nonlinear fastener elements with a compliance determined by experiment. The axial and shear behavior of the stitches was considered, however, the two compliances and failure loads were assumed to be independent. Both a double cantilever beam (mode I) and a mixed mode skin-stiffener debond configuration were studied. In the double cantilever beam configurations, G(sub I) began to decrease once the debond had grown beyond the first row of stitches and was reduced to zero for long debonds. In the mixed-mode skin-stiffener configurations, G(sub I) showed a similar behavior as in the double cantilever beam configurations, however, G(sub u), continued to increase with increasing debond length.

During the past 15 years NASA has taken the lead role in exploiting the benefits of textile reinforced composite materials for application to aircraft structures. The NASA Advanced Composites Technology (ACT) program was started in 1989 to develop composite primary structures for commercial transport airplanes with costs that are competitive with metal structures. As part of this program, several contractors investigated the cost, weight, and performance attributes of textile reinforced composites. Textile composites made using resin transfer molding type processes were evaluated for numerous applications. Methods were also developed to predict resin infiltration and flow in textile preforms and to predict and measure mechanical properties of the textile composites. This paper describes the salient results of that program.

Tests were conducted on over 200 center-crack specimens to evaluate: (a) the tension-fracture performance of candidate materials and laminates for commercial fuselage applications; and (b) the accuracy of several failure criteria in predicting response. Crack lengths of up to 12 inches were considered. Other variables included fiber/matrix combination, layup, lamination manufacturing process, and intraply hybridization. Laminates fabricated using the automated tow-placement process provided significantly higher tension-fracture strengths than nominally identical tape laminates. This confirmed earlier findings for other layups, and possibly relates to a reduced stress concentration resulting from a larger scale of repeatable material inhomogeneity in the tow-placed laminates. Changes in material and layup result in a trade-off between small-notch and large-notch strengths. Toughened resins and 0 deg-dominate layups result in higher small-notch strengths but lower large-notch strengths than brittle resins, 90 deg and 45 deg dominated layups, and intraply S2-glass hybrid material forms. Test results indicate that strength-prediction methods that allow for a reduced order singularity of the crack-tip stress field are more successful at predicting failure over a range of notch sizes than those relying on the classical square-root singularity. The order of singularity required to accurately predict large-notch strength from small-notch data was affected by both material and layup. Measured crack-tip strain distributions were generally higher than those predicted using classical methods. Traditional methods of correcting for finite specimen width were found to be lacking, confirming earlier findings with other specimen geometries. Fracture tests of two stiffened panels, identical except for differing materials, with severed central stiffeners resulted in nearly identical damage progression and failure sequences. Strain-softening laws implemented within finite element models appear attractive to account for load redistribution in configured structure due to damage-induced crack tip softening

Transport fuselage structures are designed to contain pressure following a large penetrating damage event. Applications of composites to fuselage structures require a database and supporting analysis on tension damage tolerance. Tests with 430 fracture specimens were used to accomplish the following: (1) identify critical material and laminate variables affecting notch sensitivity; (2) evaluate composite failure criteria; and (3) recommend a screening test method. Variables studied included fiber type, matrix toughness, lamination manufacturing process, and intraply hybridization. The laminates found to have the lowest notch sensitivity were manufactured using automated tow placement. This suggests a possible relationship between the stress distribution and repeatable levels of material inhomogeneity that are larger than found in traditional tape laminates. Laminates with the highest notch sensitivity consisted of toughened matrix materials that were resistant to a splitting phenomena that reduces stress concentrations in major load bearing plies. Parameters for conventional fracture criteria were found to increase with crack length for the smallest notch sizes studied. Most material and laminate combinations followed less than a square root singularity for the largest crack sizes studied. Specimen geometry, notch type, and notch size were evaluated in developing a screening test procedure. Traitional methods of correcting for specimen finite width were found to be lacking. Results indicate that a range of notch sizes must be tested to determine notch sensitivity. Data for a single small notch size (0.25 in. diameter) was found to give no indication of the sensitivity of a particular material and laminate layup to larger notch sizes.

The building block approach is currently used to design composite structures. With this approach, the data from coupon tests are scaled up to determine the design of a structure. Current standard impact tests and methods of relating test data to other structures are not generally understood and are often used improperly. A methodology is outlined for using impact force as a scale parameter for delamination damage for impacts of simple plates. Dynamic analyses were used to define ranges of plate parameters and impact parameters where quasi-static analyses are valid. These ranges include most low-velocity impacts where the mass of the impacter is large, and the size of the specimen is small. For large-mass impacts of moderately thick (0.35-0.70 cm) laminates, the maximum extent of delamination damage increased with increasing impact force and decreasing specimen thickness. For large-mass impact tests at a given kinetic energy, impact force and hence delamination size depends on specimen size, specimen thickness, boundary conditions, and indenter size and shape. If damage is reported in terms of impact force instead of kinetic energy, large-mass test results can be applied directly to other plates of the same thickness.

Impacters of various masses were dropped from various heights onto thick graphite/epoxy filament-wound cylinders. The cylinders represented filament-wound cases made for the booster motors of the Space Shuttle. Tups of various shapes were affixed to the impacters. Some of the cylinders were filled with inert propellant, and some were empty. The cylinders were impacted numerous times around the circumference and then cut into tension coupons, each containing an impact site. The size of the damage and the residual tension strength were measured. For hemispherical tups, strength was reduced as much as 30 percent by nonvisible damage. The damage consisted of matrix cracking and broken fibers. Analytical methods were used to predict the damage and residual tension strength. A factor of safety to account for nonvisible damage was determined. For corner and rod shaped tups, any damage that resulted in strength loss was readily visible.

Base programs and system programs are discussed. The base programs include fundamental research of composites and metals for airframes leading to characterization of advanced materials, models of behavior, and methods for predicting damage tolerance. Results from the base programs support the systems programs, which change as NASA's missions change. The National Aerospace Plane (NASP), Advanced Composites Technology (ACT), Airframe Structural Integrity Program (Aging Aircraft), and High Speed Research (HSR) programs are currently being supported. Airframe durability is one of the key issues in each of these system programs. The base program has four major thrusts, which will be reviewed subsequently. Additionally, several technical highlights will be reviewed for each thrust.

A compilation of slides presented at the NASA Workshop on Impact Damage to Composites held on March 19 and 20, 1991, at the Langley Research Center, Hampton, Virginia is given. The objective of the workshop was to review technology for evaluating impact damage tolerance of composite structures and identify deficiencies. Research, development, design methods, and design criteria were addressed. Actions to eliminate technology deficiencies were developed. A list of those actions and a list of attendees are also included.

Static indentation, falling weight, and ballistic impact tests were conducted in clamped plates made of AS4/3501-6 and IM7/8551-7 prepreg tape. The transversely isotropic plates were nominally 7-mm thick. Pendulum and ballistic tests were also conducted on simply supported plates braided with Celion 12000 fibers and 3501-6 epoxy. The 20 degree braided plates were about 5-mm thick. The impactors had spherical or hemispherical shapes with a 12.7 mm diameter. Residual compression strength and damage size were measured. For a given kinetic energy, damage size was least for IM7/8551-7 and greatest for the braided material. Strengths varied inversely with damage size. For a given damage size, strength loss as a fraction of original strength was least for the braided material and greatest for AS4/3501-6 and IM7/8551-7. Strength loss for IM7/8551-7 and AS4/3501-6 was nearly equal. No significant differences were noticed between damage sizes and residual compression strengths for the static indentation, falling weight, and ballistic tests of AS4/3501-6 and IM7/8551-7. For the braided material, sizes of damage were significantly less and compression strengths were significantly more for the falling weight tests than for the ballistic tests.

The concept of pressure proof testing of fuselage structures with fatigue cracks to insure structural integrity was evaluated from a fracture mechanics viewpoint. A generic analytical and experimental investigation was conducted on uniaxially loaded flat panels with crack configurations and stress levels typical of longitudinal lap splice joints in commercial transport aircraft fuselages. The results revealed that the remaining fatigue life after a proof cycle was longer than that without the proof cycle because of crack growth retardation due to increased crack closure. However, based on a crack length that is slightly less than the critical value at the maximum proof stress, the minimum assured life or proof test interval must be no more than 550 pressure cycles for a 1.33 proof factor and 1530 pressure cycles for a 1.5 proof factor to prevent in-flight failures.

The concept of pressure proof testing of fuselage structures with fatigue cracks to insure structural integrity was evaluated from a fracture mechanics viewpoint. A generic analytical and experimental investigation was conducted on uniaxially loaded flat panels with crack configurations and stress levels typical of longitudinal lap-splice joints in commercial transport aircraft fuselage. The results revealed that the remaining fatigue life after a proof test was longer than that without the proof test because of crack growth retardation due to increased crack closure. However, based on a crack length that is slightly less than the critical value at the maximum proof test stress, the minimum assured life or proof test interval must be no more than 550 pressure cycles for a 1.33 proof factor and 1530 pressure cycles for a 1.5 proof factor to prevent in-flight failures.

The steady state solution topology for absorption line-driven flows is investigated for the condition that the Sobolev approximation is not used to compute the line force. The solution topology near the sonic point is of the nodal type with two positive slope solutions. The shallower of these slopes applies to reasonable lower boundary conditions and realistic ion thermal speed v(th) and to the Sobolev limit of zero of the usual Castor, Abbott, and Klein model. At finite v(th), this solution consists of a family of very similar solutions converging on the sonic point. It is concluded that a non-Sobolev, absorption line-driven flow with a realistic values of v(th) has no uniquely defined steady state. To the extent that a pure absorption model of the outflow of stellar winds is applicable, radiatively driven winds should be intrinsically variable.

Impacters of various shapes and masses were dropped from various heights onto 36 mm (1.4 in.) thick graphite/epoxy cylinders, which represented filament wound cases (FWC) for the booster motors of the Space Shuttle. Insert solid propellant was cast into some of the cylinders. The cylinders were impacted numerous times around the circumference and then cut into 51 mm (2.0 in.) wide tension specimens, each containing an impact site. Four indenters were used: a sharp corner, two hemispheres, and a bolt-like rod. The diameters of the hemispheres were 12.7 mm mm (0.5 in.) and 25.4 mm (1.0 in.), and the diameter of the rod was 6.3 mm (0.25 in.). Impacts with the rod were simulated by pressing the rod against the face of specimens. For the hemispheres, the damage initiated beneath the surface at a critical contact pressure and was not visible on the surface until an even larger pressure was exceeded. The damage consisted of matrix cracking and broken fiber. The rod an corner made visible surface damage in all tests. For the hemispheres, the tension strength was reduced considerably before the damage was visible on the surface, 30 percent for the 25.4 mm (1.0 in.) diameter hemisphere and 10 percent for the 12.7 mm (0.5 in.) diameter hemisphere. Analytical methods were used to predict the damage and residual tension strength. A factor of safety to account for nonvisible damage was determined.

The performance is studied of a stitched uniweave fabric composite and that of a toughened tape composite. The effects of stitching on compression fatigue life are addressed. Post impact compression fatigue and open hole fatigue tests were run on an AS4/3501-6 uniweave with stitching and a toughened IM7/8551-7 tape without stitching. Stitching was found to increase the thickness and consequently the weight of the composite material. The two materials were compared on an equal carbon content basis as well as on an equal weight basis. The excess thickness in the stitched uniweave composite was responsible for the lower fatigue life, on an equal carbon basis, compared to the toughened resin tape composite. Comparison of fatigue lives on an equal carbon content basis indicated that puncture or crimp type damage from stitching has very little effect on compression failure. Post impact fatigue test showed that although the damage area in the stitched uniweave composite was twice that of the toughened tape composite, the fatigue lives of the stitched composite were significantly longer than those of the toughened composite. Thus, it appears that the increase in thickness from stitching is much more of a penalty than crimped fibers or puncture type damage from stitching.

Pressure proof testing of aircraft fuselage structures has been suggested as a means of screening critical crack sizes and of extending their useful life. The objective of this paper is to study the proof-test concept and to model the crack-growth process on a ductile material. Simulated proof and operational fatigue life tests have been conducted on cracked panels made of 2024-T3 aluminum alloy sheet material. A fatigue crack-closure model was modified to simulate the proof test and operational fatigue cycling. Using crack-growth rate and resistance-curve data, the model was able to predict crack growth during and after the proof load. These tests and analyses indicate that the proof test increases fatigue life; but the beneficial life, after a 1.33 or 1.5 proof, was less than a few hundred cycles.

A study was made to determine the relevance of impacter shape to nonvisible damage and tensile residual strength of a 36 mm (1.4 in.) thick graphite/epoxy motor case. The shapes of the impacters were as follows: 12.7 mm (0.5 in.) and 25.4 mm (1.0 in.) diameter hemispheres, a sharp corner, and a 6.3 mm (0.25 in.) diameter bolt-like rod. The investigation revealed that damage initiated when the contact pressure exceeded a critical level. However, the damage was not visible on the surface until an even higher pressure was exceeded. The damage on the surface consisted of a crater shaped like the impacter, and the damage below the surface consisted of broken fibers. The impact energy to initiate damage or cause visible damage on the surface increased approximately with impacter diameter to the third power. The reduction in strength for nonvisible damage increased with increasing diameter, 9 and 30 percent for the 12.7 mm (0.5 in.) and 25.4 mm (1.0 in.) diameter hemispheres, respectively. The corner impacter made visible damage on the surface for even the smallest impact energy. The rod impacter acted like a punch and sliced through the composite. Even so, the critical level of pressure to initiate damage was the same for the rod and hemispherical impacters. Factors of safety for nonvisible damage increased with increasing kinetic energy of impact. The effects of impacter shape on impact force, damage size, damage visibility, and residual tensile strength were predicted quite well assuming Hertzian contact and using maximum stress criteria and a surface crack analysis.

It was suggested by Zickgraf et al. (1986) that the outer atmospheres of some B(e) stars have a two-component structure: a fast, radiation-driven wind from the pole, and a dense, slow outflow from the equator. Poe et al. (1989) developed this theory to explain the momentum problem associated with WR stars. This paper uses the multiforce wind theory of Poe et al. to model the B(e) outflow phenomenon. Two general questions are investigated: (1) whether B(e) stars can be rotating near critical speed, and if so, (2) what constraints can be placed on the parameters that determine the two-component flow structure.

The residual tensile strength of a thick graphite/epoxy composite with impact damage was predicted using surface crack analysis. The damage was localized to a region directly beneath the impact site and extended only part way through the laminate. The damaged region contained broken fibers, and the locus of breaks in each layer resembled a crack perpendicular to the direction of the fibers. In some cases, the impacts broke fibers without making a visible crater. The impact damage was represented as a semi-elliptical surface crack with length and depth equal to that of the impact damage. The maximum length and depth of the damage were predicted with a stress analysis and a maximum shear stress criterion. The predictions and measurements of strength were in good agreement.

A method was previously developed to predict the fracture toughness (stress intensity factor at failure) of composites in terms of the elastic constants and the tensile failing strain of the fibers. The method was applied to boron/aluminum composites made with various proportions of 0 to + or - 45 deg plies. Predicted values of fracture toughness were in gross error because widespread yielding of the aluminum matrix made the compliance very nonlinear. An alternate method was developed to predict the strain intensity factor at failure rather than the stress intensity factor because the singular strain field was not affected by yielding as much as the stress field. Strengths of specimens containing crack-like slits were calculated from predicted failing strains using uniaxial stress-strain curves. Predicted strengths were in good agreement with experimental values, even for the very nonlinear laminates that contained only + or - 45 deg plies. This approach should be valid for other metal matrix composites that have continuous fibers.

A study was made to determine the extent of fiber damage caused by low-velocity impact of spherical impactors to a very thick graphite/epoxy laminate. The laminate was cut from a filament wound case being developed for the Space Shuttle solid rocket motors. The case was wound using a wet process with AS4W graphite fiber and HBRF-55A epoxy. Impacts were simulated under quasi-static conditions by pressing hemispherically shaped indenters against the laminate at different locations. The contact force and indenter diameter were varied from location to location. The forces were chosen for each indenter diameter to produce contact pressures below and above that required to initiate damage. After the forces were applied, the laminate was cut into smaller pieces so that each piece contained a test site. The pieces were then deplied and the individual plies examined to determine the extent of fiber damage. Broken fibers were found in the outer layers directly beneath the contact site. The locus of broken fibers in each layer resembled a crack normal to the direction of the fibers. The maximum length and depth of the cracks increased with increasing contact pressure and indenter diameter. The internal stresses in the laminate were calculated using Hertz's law and Love's solution for pressure applied on part of the boundary of a semi-infinite body. The maximum length and depth of the cracks were predicted using a maximum shear stress criterion. Predictions and measurements were in good agreement.

The National Aeronautics and Space Administration is developing graphite/epoxy filament-wound cases (FWC) for the solid rocket motors of the Space Shuttle. The membrane region is about 36 mm thick. A study was made to determine the reduction in strength of the FWC due to accidental damage caused by low-velocity impacts. Two 76.2 cm diameter by 30.5 cm long cylinders were impacted every 5 cm of circumference with 1.27 cm radius impacters of various mass. The impacters represented tools and equipment dropped from various heights. One cylinder was empty and the other was filled with inert propellant. Five cm wide test specimens were cut from the cylinder. Each was centered on an impact sight. The specimens were X-rayed and loaded to failure in uniaxial tension. The strengths and depths of impact damage were analyzed in terms of maximum impact force. Rigid body mechanics and the Hertz law were used to derive an equation for impact force in terms of kinetic energy and the masses of the impacter and target. The depth of damage was predicted in terms of impact force using Love's solution of pressure applied on part of the boundary of a semi-infinite body.

A model for predicting the fracture strength of homogeneous materials is proposed. Impacted FWC samples were evaluated using ultrasonic testing and an X-ray dye penetration method. The ability of the model to measure fracture strength was also examined. The relation between attenuation and velocity measurements is studied. It is observed that the X-ray method is not useful for predicting fracture strength because the dye could not penetrate the matrix. It is noted that fracture strength predictions derived from the fracture mechanical model and the ultrasonic measurements correlate well with actual measured fracture strengths.

Impacts to fibrous composite laminates by objects with low velocities can break fibers giving crack-like damage. The damage may not extend completely through a thick laminate. The tension strength of these damage laminates is reduced much like that of cracked metals. The fracture toughness depends on fiber and matrix properties, fiber orientations, and stacking sequence. Accordingly, a parametric study was made to determine how fiber and matrix properties and fiber orientations affect fracture toughness and notch sensitivity. The values of fracture toughness were predicted from the elastic constants of the laminate and the failing strain of the fibers using a general fracture toughness parameter developed previously. For a variety of laminates, values of fracture toughness from tests of center-cracked specimens and values of residual strength from tests of thick laminates with surface cracks were compared to the predictions to give credibility to the study. In contrast to the usual behavior of metals, it is shown that both ultimate tensile strength and fracture toughness of composites can be increased without increasing notch sensitivity.