Johnson, W. S. — Page 5 | Engineering Papers

NASA NTRS1984

Fatigue behavior of continuous fiber silicon-carbide-aluminum composites

Four lay-ups of continuous fiber silicon carbide (SCS2) fiber/aluminum matrix composites were tested to assess fatigue mechanisms including stiffness loss when cycled below their respective fatigue limits. The lay-ups were 0 (sub 8), 0(sub 2)/ + or - 45 (sub 2s), 0/90 (sub 2s),and 0/ + or 45/90 (subs). The data were compared with predictions from the author's previously published shakedown model which predicts fatigue-induced stiffness loss in metal matrix composites. A fifth lay-up, + or - 45 (sub 2s), was tested to compare shakedown and fatigue limits. The particular batch of silicon-carbide fibers tested in this program had a somewhat lower modulus (340 GPa) than expected and displayed poor bonding to the aluminum matrix. Good agreement was obtained between the stiffness loss model and the test data. The fatigue damage below the fatigue limit was primarily in the form of matrix cracking. The fatigue limit corresponded to the laminate shakedown for the + or - 45 (sub 2s) laminate.

Johnson, W. S.↗

NASA NTRS1984

Repeatability of mixed-mode adhesive debonding

The repeatability of debond growth rates in adhesively bonded subjected to constant-amplitude cyclic loading was studied. Debond growth rates were compared from two sets of cracked-lap-shear specimens that were fabricated by two different manufacturers and tested in different laboratories. The fabrication method and testing procedures were identical for both sets of specimens. The specimens consisted of aluminum adherends bonded with FM-73 adhesive. Critical values of strain-energy-release rate were also determined from specimens that were monotonically loaded to failure. The test results showed that the debond growth rates for the two sets of specimens were within a scatter band which is similar to that observed in fatigue crack growth in metals. Cyclic debonding occurred at strain-energy-release rates that were more than an order of magnitude less than the critical strain-energy-release rate in static tests.

Everett, R. A., Jr.↗

NASA NTRS1984

Geometrically nonlinear analysis of adhesively bonded joints

A geometrically nonlinear finite element analysis of cohesive failure in typical joints is presented. Cracked-lap-shear joints were chosen for analysis. Results obtained from linear and nonlinear analysis show that nonlinear effects, due to large rotations, significantly affect the calculated mode 1, crack opening, and mode 2, inplane shear, strain-energy-release rates. The ratio of the mode 1 to mode 2 strain-energy-release rates (G1/G2) was found to be strongly affected by the adhesive modulus and the adherend thickness. The ratios between 0.2 and 0.8 can be obtained by varying adherend thickness and using either a single or double cracked-lap-shear specimen configuration. Debond growth rate data, together with the analysis, indicate that mode 1 strain-energy-release rate governs debond growth. Results from the present analysis agree well with experimentally measured joint opening displacements. Previously announced in STAR as N83-13497

Dattaguru, B.↗

NASA NTRS1984

Cyclic debonding of adhesively bonded composites

To analyze the fatigue behavior of a simple composite-to-composite bonded joint, a combined experimental and analytical study of the cracked-lap-shear specimen subjected to constant-amplitude cyclic loading was undertaken. Two bonded systems were studied: T300/5208 graphite/epoxy adherends bonded with adhesives EC 3445 and with FM-300. For each bonded system, two specimen geometries were tested: (1) a strap adherend of 16 plies bonded to a lap adherend of 8 plies, and (2) a strap adherend of 8 plies bonded to a lap adherend of 16 plies. In all specimens tested, the fatigue failure was in the form of cyclic debonding with some 0 deg fiber pull-off from the strap adherend. The debond always grew in the region of adhesive that had the highest mode I (peel) loading and that region was close to the adhesive-strap interface. Furthermore, the measured cyclic debond growth rates correlated well with total strain energy release rates G(T) as well as with its components G(I) (peel) and G(II) (shear) for the mixed-mode loading in the present study.

Mall, S.↗

NASA NTRS1984

Bonded joint strength - Static versus fatigue

Adhesives are commonly characterized only by their static strength even though they are used in structural joints that are subjected to fatigue loads. This paper reviews the relationship between static and fatigue strength for four different specimen types: single-lap-shear, edge-delamination, double cantilever beam, and cracked-lap-shear. It was found that the ratio of static strength to fatigue strength varied from 2.3 to 4.7, depending on the adhesive and specimen configuration.

Johnson, W. S.↗

NASA NTRS1983

Experimental and analytical investigation of the fracture processes of boron/aluminum laminates containing notches

Experimental results for five laminate orientations of boron/aluminum composites containing either circular holes or crack-like slits are presented. Specimen stress-strain behavior, stress at first fiber failure, and ultimate strength were determined. Radiographs were used to monitor the fracture process. The specimens were analyzed with a three-dimensional elastic-elastic finite-element model. The first fiber failures in notched specimens with laminate orientation occurred at or very near the specimen ultimate strength. For notched unidirectional specimens, the first fiber failure occurred at approximately one-half of the specimen ultimate strength. Acoustic emission events correlated with fiber breaks in unidirectional composites, but did not for other laminates. Circular holes and crack-like slits of the same characteristic length were found to produce approximately the same strength reduction. The predicted stress-strain responses and stress at first fiber failure compared very well with test data for laminates containing 0 deg fibers.

Johnson, W. S.↗

NASA NTRS1983

A fracture mechanics approach for designing adhesively bonded joints

An analytical and experimental investigation was undertaken to determine if the adhesive debond initiation stress could be predicted for arbitrary joint geometries. The analysis was based upon a threshold total strain-energy-release rate (Gth) concept. Two bonded systems were tested: T300/5208 graphite/epoxy adherends bonded with either EC-3445 or FM-300 adhesive. The Gth for each adhesive was determined from cracked-lap-shear (CLS) specimens by initiation tests. Finite-element analyses of various tapered CLS specimen geometries predicted the specimen stress at which the total strain-energy-release rate (GT) equaled Gth at the joint tip. Experiments verified the predictions. The approach described herein predicts the maximum stress at which an adhesive joint can be cycled yet not debond. Furthermore, total strain-energy-release rate appeared to be the driving parameter for cyclic debonding and debond initiation in structural adhesives. In addition, debond initiation and growth were found to occur with virtually no peel stress present.

Johnson, W. S.↗

NASA NTRS1983

Damage tolerance evaluation of adhesively laminated titanium

Basic damage tolerance properties of Ti-6A1-4V titanium plate can be improved by laminating thin sheets of titanium with adhesives. Compact tension and center cracked tension specimens made from thick plate, thin sheet, and laminated plate (six plies of thin sheet) were tested. The fracture toughness of the laminated plate was 39 percent higher than the monolithic plate. The laminated plate's through the thickness crack growth rate was about 20 percent less than that of the monolithic plate. The damage tolerance life of the surface cracked laminate was 6 to over 15 times the life of a monolithic specimen. A simple method of predicting crack growth in a crack ply of a laminate is presented. Previously announced in STAR as N82-33737

Johnson, W. S.↗

NASA NTRS1983

Modeling stiffness loss in boron/aluminum laminates below the fatigue limit

Boron/aluminum can develop significant internal matrix cracking when fatigued. These matrix cracks can result in a 40 percent secant modulus loss in some laminates, even when fatigued below the fatigue limit. It is shown that the same amount of fatigue damage will develop during stress or strain-controlled tests. Stacking sequence has little influence on secant modulus loss. The secant modulus loss in unidirectional composites is small, whereas the losses are substantial in laminates containing off-axis plies. A simple analysis is presented that predicts unnotched laminate secant modulus loss due to fatigue. The analysis is based upon the elastic modulus and Poisson's ratio of the fiber and matrix, fiber volume fraction, fiber orientations, and the cyclic-hardened yield stress of the matrix material. Excellent agreement was achieved between model predictions and experimental results. With this model, designers can project the material stiffness loss for design load or strain levels and assess the feasibility of its use in stiffness-critical parts. Previously announced in STAR as N82-24298

Johnson, W. S.↗

NASA NTRS1982

Cyclic debonding of adhesively bonded composites

The fatigue behavior of a simple composite to composite bonded joint was analyzed. The cracked lap shear specimen subjected to constant amplitude cyclic loading was studied. Two specimen geometries were tested for each bonded system: (1) a strap adherend of 16 plies bonded to a lap adherend of 8 plies; and (2) a strap adherend of 8 plies bonded to a lap adherend of 16 plies. In all specimens the fatigue failure was in the form of cyclic debonding with some 0 deg fiber pull off from the strap adherend. The debond always grew in the region of adhesive that had the highest mode (peel) loading and that region was close to the adhesive strap interface.

Mall, S.↗

NASA NTRS1982

Geometrically nonlinear analysis of adhesively bonded joints

A geometrically nonlinear finite element analysis of cohesive failure in typical joints is presented. Cracked-lap-shear joints were chosen for analysis. Results obtained from linear and nonlinear analysis show that nonlinear effects, due to large rotations, significantly affect the calculated mode 1, crack opening, and mode 2, inplane shear, strain-energy-release rates. The ratio of the mode 1 to mode 2 strain-energy-relase rates (G1/G2) was found to be strongly affected by he adhesive modulus and the adherend thickness. The ratios between 0.2 and 0.8 can be obtained by varying adherend thickness and using either a single or double cracked-lap-shear specimen configuration. Debond growth rate data, together with the analysis, indicate that mode 1 strain-energy-release rate governs debond growth. Results from the present analysis agree well with experimentally measured joint opening displacements.

Dattaguru, B.↗

NASA NTRS1982

Improved damage tolerance of titanium by adhesive lamination

Basic damage tolerance properties of Ti-6A1-4V titanium plate can be improved by laminating thin sheets of titanium with adhesives. Compact tension and center cracked tension specimens made from thick plate, thin sheet, and laminated plate (six plies of thin sheet) were tested. The fracture toughness of the laminated plate was 39 percent higher than the monolithic plate. The laminated plate's through the thickness crack growth rate was about 20 percent less than that of the monolithic plate. The damage tolerance life of the surface cracked laminate was 6 to over 15 times the life of a monolithic specimen. A simple method of predicting crack growth in a crack ply of a laminate is presented.

Johnson, W. S.↗

NASA NTRS1982

Modeling stiffness loss in boron/aluminum below the fatigue limit

Boron/aluminum can develop significant internal matrix cracking when fatigued. These matrix cracks can result in a 40 percent secant modulus loss in some laminates, even when fatigued below the fatigue limit. It is shown that the same amount of fatigue damage will develop during stress or strain-controlled tests. Stacking sequence has little influence on secant modulus loss. The secant modulus loss in unidirectional composites is small, whereas the losses are substantial in laminates containing off-axis plies. A simple analysis is presented that predicts unnotched laminate secant modulus loss due to fatigue. The analysis is based upon the elastic modulus and Poisson's ratio of the fiber and matrix, fiber volume fraction, fiber orientations, and the cyclic-hardened yield stress of the matrix material. Excellent agreement was achieved between model predictions and experimental results. With this model, designers can project the material stiffness loss for design load or strain levels and assess the feasibility of its use in stiffness-critical parts.

Johnson, W. S.↗

NASA NTRS1982

Mechanisms of fatigue damage in boron/aluminum composites

(Previously announced in STAR as N81-16135)

Johnson, W. S.↗

NASA NTRS1981

Multi-parameter yield zone model for predicting spectrum crack growth

A phenomenological load interaction model referred to as the Multi-Parameter Yield Zone model is presented which accounts for crack growth retardation caused by previous overloads, acceleration due to current overloads, and underload effects. In the present model, load interaction effects are calculated utilizing a residual stress intensity concept. Crack growth retardation and acceleration are accounted for by decreasing or increasing, respectively, the effective stress ratio used in a modified Forman's (1967) crack growth equation. Results of spectrum crack growth predictions for 2219-T851 aluminum are presented.

Johnson, W. S.↗

NASA NTRS1981

Fatigue mechanisms in metal matrix composite laminates

Certain aspects of the damage process are described. The shakedown phenomenon in laminates is first explained. This is followed by a presentation of matrix test data used in evaluating the shakedown limits of the laminates used in the subsequent experiments. Test results suggesting that there is no loss of elastic stiffness in laminates loaded at constant amplitudes within their shakedown limits are presented. Damage accumulation and the development of the saturation damage state are illustrated by experimental results on specimens loaded outside the shakedown range. Attention is also given to the behavior of laminates subjected to variable cyclic loading.

Dvorak, G. J.↗

NASA NTRS1980

Mechanisms of fatigue damage in boron/aluminum composites

Tensile fatigue tests were conducted on several laminates of boron/aluminum (6061-0). In laminates with 0 deg fibers on the outside, an analysis that identifies "shakedown" conditions predicted the stress amplitude below which no fatigue damage accumulated. A fatigue damage accumulation model which relates matrix fatigue cracking and the overall laminate properties is described. A model for the saturation damage stage development is presented, that identical laminates, tested in directions 90 deg apart (such that one layup has 90 deg outer plies and the other 0 deg), have different fatigue behaviors due to the stacking sequence. The 90 deg plies on the surface develop cracks earlier than predicted by shakedown. An attempt was made to explain this stacking sequence effect. Variable load history effects on the fatigue damage response were investigated. Tests reveal that for a given stress ratio the specimen seeks the saturation damage state for the largest stress range to which it is subjected. It was also found that little damage is generated by shifting a given stress range down, whereas significant damage may be created by shifting it upward. The laminate stresses were always tensile.

Johnson, W. S.↗

NASA NTRS1980

Multi-Parameter Yield Zone Model for Predicting Spectrum Crack Growth

A systematic technique is presented for modeling crack growth load interaction effects due to spectrum loading. The Multi-Parameter Yield Zone (MPYZ) model accounts for crack growth retardation, acceleration, and underload effects. The load interactions are attributed to the residual stress intensity due to the plastic deformation at the crack tip. Fatigue crack growth was predicted and compared with test data for a variety of spectrum loadings.

Johnson, W. S.↗