Characterizing the failure of composite structures.
There are two main types of composites: particulate and fiber. In particulate composites, small particles are uniformly distributed and embedded in a matrix system. In fiber composites, the reinforcing fibers--either continuous or chopped--are embedded within a matrix.
At Lawrence Livermore National Laboratory (LLNL), continuous-fiber composite materials are being studied for a number of new applications. For example, composites are attractive for weapons applications due to their light weight and high strength. Such composites are currently being considered as candidates for outer weapon cases because they could provide structural strength and yet save weight. They are also being considered for certain projects for the Strategic Defense Initiative, where very stiff, very lightweight structural components that undergo only minimal dimensional changes with changes in temperature are needed.
Even though composites are not new, much information on their mechanical characteristics has yet to be obtained. Extensive data bases on material properties and accurate models of composites' behavior must be developed if structural designers are to avoid the costs and pitfalls of a cut-and-try approach and efficiently tailor composites to specific applications.
Fiber-composite materials offer two major advantages over metals: high specific strength and high specific stiffness. In applications where weight is a limiting factor, fiber composites are very attractive because they provide good strength and stiffness with very little weight. At LLNL, we have examined a number of composites that have greater tensile strength by weight than metals, including T300, IM6, Toray 1000, and Kevlar.
Fiber composites are designed by selecting a reinforcing fiber and an appropriate matrix and then orienting the fibers within the laminate to create a material with the desired strength, stiffness, weight, wear resistance, and thermal expansion. Our goal is to understand the behavior and properties of the fibers, matrices, composite material, and various laminates so that we can tailor composites to specific applications.
A major disadvantage of these materials is that the improved properties are exhibited only in the plane of the fibers, that is, only in two dimensions. This generally restricts the use of composites to two-dimensional structures since these materials tend to be very weak in the third (out-of-plane) direction. The structural designer must be assured that the out-of-plane loading conditions for the intended applications are acceptable i.e., within the strength limits of the desired material). Another problem with fiber composites is that it is difficult to fabricate a composite part without flaws, such as voids or microscopic cracks, that can lead to the premature failure of the composite part.
Laminating and filament winding are the two methods used to fabricate fiber-composite parts. To fabricate a laminated part, a pre-impregnated tape of continuous fibers is embedded in a thin film of resin that is layered onto a nonstick backing paper. The layers, or plies, of prepreg tape are stacked in the desired orientation and shaped over a mold or mandrel. The polymeric resin is usually dormant at room temperature and is cured at high temperature, usually in a laminating press or autoclave and usually under vacuum.
In filament winding, a continuous fiber that is coated in resin is wound around a mandrel. A numerically controlled winding machine is used to carefully control the winding pattern and tension at which the fiber is laid on. After the desired thickness is obtained, the part is cured.
it is possible to create very complex shapes using both fabrication methods. Filament winding is more efficient for spherical or cylindrical parts such as pressure vessels. Lamination is more efficient for parts that are basically flat, such as turbine blades, wing tips, and structural body parts.
Composite materials are difficult to analyze and model because they are orthotropic--they have three mutually perpendicular planes of symmetry and thus their material properties are markedly different in these different directions. Nine individual material constants are needed to model the elastic constitutive behavior of orthotropic materials, compared to only two constants needed for isotropic materials such as most metals. The experimental effort required to characterize the material constants is very complex and represents a significant portion of LLNL's composite research.
One of the most difficult analytical problems encountered with composite materials is that of predicting failure response. Composite materials, especially those with multiple plies at different orientations or multiple layers of filament-wound fibers, often contain myriad internal failure surfaces such as voids, bubbles, and fiber flaws. With so many possible sites at which a flaw or crack can begin to grow and lead to failure of the composite part, it is very difficult to establish an accurate or appropriate failure load. Many failure theories have been developed, but they have generally proven inadequate. Work is underway at LLNL to improve our understanding of failure and to develop a more adequate set of failure criteria.
We are currently evaluating the strength of composite materials and predicting their failure under thermal-mechanical loading. in particular, we are attempting to develop a new, more appropriate failure criterion. The key to developing such a criterion is the ability to experimentally determine the necessary failure data. This is not an easy task. In order to evaluate the failure criterion in a multiaxial stress field, we must first be able to generate controlled multiaxial failure surfaces. The most common way to do this for metals (isotropic materials) is combined axial tension and torsion of tube specimens. Although a similar procedure can be used for composites, it is not problem-free. For instance, it is almost impossible to grip the composite tube to apply the loads without inducing failure in the grip region.
To eliminate this problem, we designed a composite tube that flared at both ends and then designed grips to hold the flared tube stationary. This unique combination of tubes and grips worked; it minimized the stress that was previously concentrated in the grip region and transferred it uniformly along the length of the composite. We can now place these tubes in a numerically controlled, servo-hydraulic, multiaxial test instrument and subject them to combinations of tension, compression, torsion, and internal pressurization to generate the failure surfaces. Moreover, we can be confident that the resulting data reflect material failure as opposed to grip failure. The development of these tools has been a critical step toward our goal of successfully predicting the strength and failure modes of fiber composites.
Using the test instrument, we have successfully applied axial load, torsion, and internal pressure simultaneously to a T300/F263 prepreg tape. When we pressurized the tube, it leaked. It appears that once a small crack forms, the internal pressure is lost, even though the composite may be structurally sound. We are currently investigating different types of nonstructural internal bladders to help increase the internal pressure.
Our success in loading composites to failure has led us to a point where we can establish realistic criteria for strength and failure in composites. This, in turn, has allowed us to begin altering fabrication parameters so as to increase the strength of a given composite. Currently, we are using the test apparatus in an experiment to increase the compressive strength of Toray 1000/DER 332-T403 composite, which is a material of interest to LLNL's Earth Penetrator project. Our goal is to obtain the highest specific compressive strength possible.
We have found several ways to increase compressive strength, one of which is to increase the percentage of fiber in the composite. (Theoretically, 60 percent is the optimum fiber volume, but we have not yet verified that figure.) One way to achieve the highest possible fiber volume is to make the composite as compact as possible. We have found that by manipulating the parameters of two of the fabrication steps, filament winding and curing, we can increase the compaction of the composite. In filament winding, we wound fiber around a mandrel under 8 pounds of tension, a much higher tension than previously tried (any greater tension could damage the fiber). We achieved a fiber volume of 56 percent, close to the optimum 60 percent volume.
In the second method, we cured the composite in an autoclave, which allowed us to use pressure to squeeze resin out of the composite, thereby making the tube as compact as possible. The second technique resulted in a fiber volume of 70 percent. Next we tested both tubes to failure and found that the first technique was more successful: the tube withstood greater stress before failing, even though it had a lower fiber volume. The reason that the tube with the higher fiber volume failed earlier was that the pressure applied during the curing process severely wrinkled the fibers, reducing the tube's compressive strength. Defects such as wrinkling, fiber misalignment, fiber waviness, and voids can all be easily introduced during the fabrication process and need to be guarded against. They can result in fiber microbuckling local failure of a fiber), which in turn can lead to the premature failure of the composite, rather than to compressive fiber failure where the composite sustains the predicted load. In our experiment, we abandoned using an autoclave to increase fiber volume, but have continued to wind fiber under 8 pounds of tension.
Another way to increase compressive strength is to vary the ply layup parameters, that is, the angle at which the fiber is wound around the mandrel, the number of layers wound at a given angle, and the order of the various layers. We are currently investigating a composite whose fibers are wound at O-, 45-, and 90-degree angles (the actual angles are 1.5, 45, and 89 degrees because the winder cannot wind exactly 0- and 90-degree angles). The fibers are wound at a variety of angles because if they were wound at one angle, the tube would collapse when subjected to axial compression--it would split down its length as though it were a barrel whose staves had caved in. To prevent such splitting, we wrap a layer of 90-degree hoop fibers around the 0-degree fibers. Now when we compress the structure, we find that the failure mode is one in which the axial fibers buckle inwardly, because the hoop fibers stop the axial fibers from bowing outwardly. That is, when we compress the composite, the compressive force on the axial fibers O degrees) is transferred to the hoop fibers (90 degrees) as tensile stress. Because the tensile stress of the hoop fibers is much greater than the compressive strength of the axial fibers, the latter will buckle inwardly.
To prevent inward buckling, we next put together a fiber sandwich; we wound a set of 0-degree axial fibers between two sets of 90-degree hoop fibers. This sandwich approach, using a 90-, O-, 90-degree pattern, exhibited greater compressive strength than the O-, 90-degree pattern. We are currently investigating whether fibers at a 45-degree angle will also increase compressive strength.
We plan to continue investigating several other questions concerning compressive performance, such as how sensitive the compression performance is to the angle of the axial fiber and whether increasing the stiffness of the hoop fibers increases compression strength. Cylindrical Shear
In addition to using the multiaxial test apparatus to determine the compression strength of fiber composites, we have used it to evaluate the shear strength of four different cylindrical joints made of graphite and epoxy. The four designs consist of a bonded wedge cone, pinned wedge cone, bonded and pinned wedge cone, and an integrally wound wedge cone. All four designs were similar to our multiaxial test part: they all had an axial length of I inch and a 15-degree tapered cone. Our goal was to achieve an average shear strength of 8000 psi along the joint. This value was derived by calculating the maximum shear strength of the epoxy system used in the bonding. Results obtained for the four joints were approximately 14,000, 10,200, 11,700, and 14,500 psi, respectively. We believe the increase in available shear strength can be attributed to the hydrostatic compression effects from the gripping mechanism.
To help us further understand the complex failure performance of composite materials, we have developed out-of-plane shear loading fixtures, a three-dimensional compression fixture, and an external ring compression fixture. To graph a complete failure surface, we must merge the data on out-of-plane stress fields obtained with these tools with the data on in-plane stress fields generated with the multiaxial test apparatus. We do not yet have the ability to load a part in three dimensions.
We have also developed techniques for loading composites at high stress/strain rates because some of our composites may, for example, be used to protect the innards of a missile from a nearby explosion. Our purpose is to develop a strain-rate-sensitive material model for inclusion in LLNL's DYNA code (see January 1991 ME, pp. 56-60).
A Hopkinson bar--a standard device for measuring stress/strain behavior at high rates--was used to test a composite at a strain rate of 1000 inches/inch/second. The toughness, or area under the curve that plots axial compression stress versus compression strain, is approximately eight times greater for a composite subjected to a 1000-inch/inch/second stress/strain rate than for one subjected to a 0.001-inch/inch/second stress/strain rate. Although we do not yet understand the mechanisms behind this result, a working hypothesis is that the increased toughness can be attributed to several factors, including the viscoelastic behavior of the epoxy and the formation of non-coalescing internal damage mechanisms such as microcracking. At these higher strain rates, it is possible that the loading rate is fast enough to prevent the usual delamination of the composite before failure.
All of these tests have been conducted in compression; however, we have also developed a 0.5-inch-diameter version of the composite tube and have measured its stress/strain rate in tension at rates up to 100 units of displacement per second. We found that, as in compression, the greater the stress/strain rate, the higher the tensile strength.
Future work will include generating multiaxial failure data for the composite Toray 1000/DER 332-T403. Our plan is to vary the winding patterns of this composite and see what changes in behavior occur. We will also evaluate several published failure criteria and begin to formulate our own more general criterion. As part of that formulation, we will continue to generate high strain rate data using the drop tower and Hopkinson bar.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Lawrence Livermore National Laboratory|
|Author:||Groves, Scott E.|
|Date:||Feb 1, 1991|
|Previous Article:||Connecting rods that crack by design.|
|Next Article:||The state of education.|