Vous êtes sur la page 1sur 5

Failure Mode and

Strength Predictions Anisotropic Bolt Bearing Specimens

of

J. P. WASZCZAK AND T. A. CRUSE, Department of Mechanical Engineering, Carnegie-Mellon University, Pittsburgh, Pennsylvania 15213
(Received April 6, 1971)

A major consideration in the design of a structure made of composite materials is the bolted joint since a bolted joint in a composite material has a significantly lower efficiency than the same joint in metals. Furthermore, the composite joint may fail in unique modes not found in metal joints. This study, in an attempt to further understand the failure characteristics of such bolted joints, investigates the stress concentrations induced in anisotropic plates loaded by means of a single fastener. The development of a prediction capability for both failure mode and ultimate load is the major goal of the early part of this work. Such a capability would allow synthesis rather than analysis to be used in the future design of fastener joints. An implied goal in this study is a relative evaluation of the three proposed anisotropic failure criterion: maximum stress, maximum strain, and distortional energy. COMPUTER SIMULATION
A constant strain, finite element computer program modified to handle anisomaterials [1]was used for the stress analysis. Only specially orthotropic laminates, i.e. laminates which are mid-plane symmetric and are crossplied, were investigated since the published data on bolt bearing specimens is also limited to this class of laminates. The bolt bearing test specimen contains two lines of specimen symmetry as shown in Figure 1. Thus it was only necessary to include one-fourth of the specimen in the finite element simulation. The grid representation used contains 480 triangular elements and 279 nodes. A cosine distribution of normal stress acting over the upper half of the hole surface was used to simulate the resulting stress distribution caused by the bolt. The interaction was assumed to be frictionless. Bickley [2] shows this to be an excellent approximation for isotropic bolt bearing specimens. Several isotropic bolt bearing test specimens were simulated using the cosine distribution of normal stress and the grid mesh previously described. The computed stress concentration

tropic composite

factors for these

specimens agreed with [3]

cosine distribution (see Figure 2) were used to simulate the bolt-specimen interface stress boundary condition. The net force in the load direction in each case was equal. It was observed that significant variation about the cosine distribution

Finally,

two other

to within six percent. normal distributions of stress significantly different from the

J. COMPOSITE MATERIALS, Vol. 5

(July 1971),

p. 421

Figure 1. Bolt bearing

test

specimen.

resulted in considered.

insignificant

alterations of the calculated

stress

fields for the

specimen

ANALYSIS PROCEDURE AND RESULTS

The selection of specimen geometries for the investigation was made from data which has been published by General Dynamics, North American Rockwell and Grumman Aerospace. Figure 3 illustrates the various failure modes included in
this investigation. A strength analysis of a laminated composite structure is based on the strengths of its individual laminae. Lamina failure may be predicted to occur by the distortional energy failure criterion [4] when the following combination of lamina principal stress ratios add to a number greater than or equal to one:

The maximum stress (or strain) failure criterion, on the other hand, requires that the ratio of principal stresses (or strains) to their respective ultimate stresses (or strains) be greater than or equal to one for failure to occur. The distortional energy failure criterion has been found in this study to be the only reliable means of predicting bolt bearing specimen failure modes. Interpretation of the maximum stress and maximum strain failure criteria, in an attempt to predict specimen failure modes, has been very unsuccessful. The difficulty results from the anisotropy of the composite material. Figure 4 shows distortional energy contour plots for the main load carrying lamina in various experimentally failed specimens. An initial application of the experimentally determined failure load was applied in each case. The contour plots were suflicient to enable the prediction of failure modes in all but the shear-out cases. For these specimens it was necessary to consider the ratios of lamina principal stresses to their respective
422

2. Variations about the cosine distribution of normal stress.

Figure

Figure 3. Bolt bearing


failure modes.

test

specimen

Figure
for bolt

4. Distortional energy contour plots

bearing specimens.

423

ultimate stresses in the regions of high distortional energy to differentiate between shear-out mode and a bending tear-out mode. Prediction of failure load was also based on normalized distortional energies. The values of DIST in the first row of circumferential elements around the hole were considered for each lamina. A successive failure analysis was used to predict ultimate load. As soon as an element in any given lamina achieved a value of DIST equal to 1.0 that lamina was assumed to have failed and was removed from the laminate. The load was then redistributed among the remaining laminae and all values of DIST were recalculated. If all recalculated vaules of DIST were less than 1.0 more load was applied until another lamina reached failure. This process was repeated until total laminate failure occurred. The resulting predictions of failure load, based on distortional energies, were conservative (see Table I). The degree of conservatism varied with failure mode, but more importantly it appeared to be a function of specimen anisotropy. Note that the predicted failure loads for the net tension specimens improve greatly as the percentage of 45laminae decreases. This same type of behavior was reported by Grumman Aerospace in a study they performed on laminate tension data. Both the maximum stress failure criterion and the maximum strain failure criterion underpredicted one specimen ultimate load. However, failure load predictions made using the maximum stress failure criterion are in agreement with those of the distortional energy failure criterion to within about ten percent. The major disadvantage encountered with the maximum stress criterion, as previously stated, is its inability to clearly indicate failure modes. The experimentally failed specimens exhibit excellent agreement with the predicted failure behavior of this study. For example, a specimen which failed in a shear-out mode exhibited a relatively smooth, clean fracture surface. The distortional energy contour plot for this specimen is shown in Figure 4 ( b ) . The region of high distortional energy results from very high principal shear stress ratios which
a

Table 1.

Nomenclature

T C
S B

Tension Combination Shear-Out


Bearlna

G/E

3/E

Graphite-Epoxy Boron-Epoxy

424

rather smooth shear fracture surfaces. On the other hand, a ( 0 / which failed in a bending tear-out failure mode exhibited a very coarse, jagged fracture surface along lines at 45. This behavior is again expected from the computed stress ratios. Along lines at 45, where the distortional energies are high, the largest stress ratios act in the first principal direction. These are the stresses which are trying to break fibers in tension. As a result, as the triangular section is being torn away from the specimen, the fibers along these lines are being broken in tension; resulting in a very coarse, jagged fracture surface. Another interesting feature of most of the experimentally failed specimens was the presence of a highly localized region of laminate destruction at the bolt-specimen interface. This damage occurred in conjunction with almost every type of experimental failure mode and was predicted by the stress analysis, Figure 4.
to

would lead

90/45) specimen

REFERENCES
1.

Technomic
2. W 3. 4.

J.

E.

Ashton, J. C. Halpin, and (1969).

P. H. Petit, Primer

on

Composite Materials: Analysis,

Bickley, "The Distribution of Stress Round a Circular Hole in a Plate", Phil. Trans. Roy. Soc., A (London), Vol. 227 (1928), p. 383. R. E. Peterson, Stress Concentration Design Factors, Wiley (1953). S. W. Tsai, "Mechanics of Composite Materials, Part II—Theoretical Aspects", Air Force Materials Laboratory Technical Report AFML-TR-66-149 (1966).

425