resin just prior to winding; (2) wet rolled prepreg winding, in which the dry reinforcement is impregnated with the liquid resin and then rewound prior to filament winding; and (3) towpreg winding, in which a commercially impregnated tow is purchased from a material supplier.

Filament winding equipment costs can be low, moderate, or high, depending on part size, the type of winder, and the sophistication of the control system (mechanical or NC control). For high rate applications, some winders have been designed with multiple pay-out systems and multiple mandrels so that several parts can be fabricated simultaneously. Filament wound parts are usually cured in ovens rather than autoclaves, the compaction pressure being provided by mandrel expansion and fiber tension on the lay-up. Circular windings, separated from the part by caul plates or separator sheets, can also be used to provide compaction during cure. Forced air convection ovens are the most common curing equipment. Others, such as microwave curing are faster but result in higher equipment costs. Mandrel costs can be moderate to high, depending on part size and complexity. The mandrel must be able to be removed from the part. This is often accomplished by shrinkage of the mandrel during cool-down, incorporation of a slight draft or taper, wash-out mandrels, plaster break-out mandrels, inflatable mandrels, or for complex parts, segmented mandrels that can be taken out of the inside of the part in sections. While the inner surface (mandrel side) of the part is usually smooth, the outer surface can be quite rough. If this presents a problem, it is possible to wind sacrificial layers on the outer surface and then grind, or machine, the outer surface smooth after cure.

Fiber orientation can be a problem for some filament wound designs, i.e. the minimum fiber angle that can usually be wound is 10-15° due to slippage of the fiber bands at the mandrel ends. However, schemes such as temporary pins inserted in the mandrel ends during winding can sometimes be used to overcome this limitation.

Helical, polar, and hoop are the three dominant winding patterns used in filament winding. Helical winding (Fig. 7.22) is a very versatile process that can produce almost any combination of length and diameter. In helical winding the mandrel rotates, while the fiber carriage traverses back and forth at the speed necessary to generate the desired helical angle (0).12 As the band is wound, the circuits are not adjacent and additional circuits must be applied before the surface begins to be covered with the first layer. This winding pattern produces band cross-overs at periodic locations along the part, which can be somewhat controlled by the newer NC winding machines. Due to this cross-over winding pattern, a layer is made up of a two-ply balanced laminate. If the end dome openings are the same size, a geodesic wind pattern may be used. This pattern produces the shortest band path possible and results in uniform tension in the filaments throughout their length. An additional advantage of the geodesic pattern is that it produces a no-slip condition, i.e. there is no tendency for the bands to slip or shift on the mandrel surface.

Hoop or Circumferential Winding

Fig. 7.22. Filament Winding Patterns12

Hoop or Circumferential Winding

Fig. 7.22. Filament Winding Patterns12

Polar winding (Fig. 7.22) is somewhat simpler than helical winding in that: (1) a constant winding speed can be used; (2) it is not necessary to reverse the carriage during winding; and (3) the bandwidths are laid adjacent to each other as the part is wound. This is an excellent method for fabricating spherical shapes. In this process, the bands pass tangentially to the opening at one end of the part, reverse direction, and then pass tangentially to the opening at the opposite end of the part. The lay-down is planar, with the bandwidths adjacent to each other due to the winding arm, generating a great circle during each pass. Simple polar winders have only two axes of motion, the mandrel and the winding arm. Polar winding machines are generally much simpler than helical winding machines, but they are also somewhat limited in their capabilities. The length-to-diameter ratio must be less than 2.0. They are frequently used to wind spherical shapes by utilizing a continuous step-out pattern. A variation of the polar winder is the tumble winder, in which the mandrel is mounted at an inclined axis and tumbles in a polar path, while the roving strands remain stationary. While tumble winders are very efficient for spherical shapes, they are usually limited to diameters of 20 in. or less.

Hoop winding, also known a circumferential or circ winding, is the simplest winding process. This winding action, shown in Fig. 7.22, is similar to a lathe, where the mandrel speed is much greater than the carriage travel. Each full rotation of the mandrel advances the carriage one full bandwidth, so that the bands are wound adjacent to each other. During part fabrication, hoop winding is often combined with longitudinal (helical or polar) winding to provide adequate part strength and stiffness. The hoop windings can be applied to the cylindrical portion of the part, while the longitudinal windings are applied to both the cylindrical and domed portions of the part. It again should be pointed out that the minimum wind angle for longitudinal winding is generally about 10-15°, to preclude slippage of the bands at the ends of the mandrel.

The majority of filament winding fabricators formulate their own resin systems for both wet winding and wet rolled prepreg.11 If a prepreg product form is specified, then they will normally purchase the preimpregnated tow from one of the major prepreg suppliers. Epoxies are the most prevalent matrix resins used for high performance filament wound parts, while polyesters and vinyl esters are often used for less demanding commercial applications. However, many different types of resins have been successfully used for filament winding, including cyanate esters, phenolics, bismaleimides, polyimides, and others.

Viscosity and pot life are two of the main factors in selecting a resin for wet winding. Low viscosity, generally around 2000 Centipoise (cP), is desirable to help wet the fibers, spread the band, and lower the friction over the guides during the winding process. Pot life is primarily a function of the time it will take to wind the part, i.e. large and thicker parts will require a longer pot life than smaller thinner parts. A number of premixed wet winding resin systems are also available from material suppliers. While preimpregnated tow (towpreg) is more expensive than wet winding resin systems, it does offer several important advantages: (1) a qualified fiber and resin system can often be prepregged onto a tow; (2) it allows the best control of resin content; (3) it allows the highest winding speeds because there is no wet resin that will be thrown-off during winding; and (4) the tack can be adjusted to allow less slippage when winding shallow angles.

Wet winding is accomplished by either pulling the dry tows through a resin bath or directly over a roller that contains a metered volume of resin controlled by a doctor blade. The resin content of wet wound parts is difficult to control, being affected by the resin reactivity, the resin viscosity, the winding tension, the pressure at the mandrel interface, and the mandrel diameter. For example, too low a viscosity resin will impregnate the strands thoroughly but will tend to squeeze out during the pressure of the winding operation, resulting in an excessively high fiber content. At the other extreme, too high a viscosity will not sufficiently impregnate the strands and there will be a tendency for the cured part to contain excessive porosity. Due to the generally low viscosity of wet winding resins, it is not uncommon to have parts with higher fiber volume percentages (70% and sometimes higher) than are normally found in composite parts fabricated with higher viscosity prepreg resins (60 volume percent).

To circumvent some of the problems with controlling a direct wet winding process, wet rolled prepreg is sometimes manufactured by wet impregnating the strands in the normal manner, and then respooling them prior to winding.

There are two main advantages to this process: (1) the fabricator can conduct off-line quality assurance on the wet wound prepreg prior to use, and (2) they can somewhat control the viscosity and tack by room temperature staging. Staging at room or slightly elevated temperature is commonly called B-staging. The objective is to advance the resin to increase the viscosity and tack. On the negative side, wet wound prepreg has to be packaged and refrigerated for storage, unless it is immediately used for winding.

Commercially supplied prepregs offer the best control of resin content, uniformity, and band width control, but are also the most expensive of the product forms, usually 1.5-2 times the cost of wet winding materials. While prepreg tows are the predominant prepreg form using in filament winding, some aerospace manufacturers specify slit prepreg tape to insure extremely tight control of the bandwidth, and the resultant gaps, during the fiber placement of flight critical hardware. Prepreg tows for filament winding generally (1) have the longest pot lives; (2) allow higher winding speeds because there is less chance of "resin throw" during the winding process; and (3) allow winding angles closer to longitudinal (0°), because they contain higher tack than most wet winding systems and will not tend to slip as much at the ends.

The choice of a mandrel material and design is to a great extent a function of the design and size of the part to be built. A large number of materials have been used for filament winding mandrels. Dissolvable mandrels are often used for parts with only small openings. This type of mandrel includes water soluble sand, water soluble or breakout plaster, low temperature eutectic salts, and occasionally low melting point metals. After cure, the disposable mandrel is dissolved out with hot water, melted, or broken into small pieces for removal. An alternate to these approaches would be to use an inflatable mandrel that can be either left inside the part as a liner or extracted through an opening. Reusable mandrels can either be segmented or non-segmented. Segmented mandrels are required when the part geometry does not allow the part to be removed by simply sliding the part off the mandrel after cure. Segmented mandrels are generally more expensive to fabricate and use than non-segmented mandrels. Non-segmented mandrels usually have a slight draft or taper to ease part removal after cure.

After the winding operation is complete, wet wound parts are often B-staged prior to final cure to remove excess resin by heating the part to a slightly elevated temperature but below the resin gel temperature. Frequently, the part is heated with heat lamps and the excess resin is removed as the part rotates. The great majority of filament wound parts are cured in an oven (electric, gas fired or microwave) without a vacuum bag or any other supplemental method of applying pressure. As the part heats-up to the cure temperature, the mandrel expands but is constrained by the fibers in the wound part. This creates pressure that helps to compact the laminate and reduce the amount of voids and porosity. Since the majority of filament wound parts are cured in ovens rather than autoclaves, filament winding is capable of making very large structures, limited only by the size of the winder and the curing oven available.

Autoclave curing may also be used to further reduce the amount of porosity; however, the compaction pressure applied by an autoclave can also induce fiber buckling, and even wrinkles in the part. The use of thin caul plates that are allowed to slip over the surface may help to alleviate some wrinkling on cylindrical surfaces, but these are prone to leaving mark-off on the part surface where they terminate. Caul plates with circumferential windings over the outside of the caul plates have also been used in oven cured parts to improve compaction and provide a smoother surface finish. Occasionally, the part will be wrapped with shrink tape to provide compaction pressure, a common method employed in manufacturing carbon fiber golf club shafts.

7.7 Fiber Placement

In the late 1970s, Hercules Aerospace Co. (now Alliant Techsystems) developed the fiber placement process. Shown conceptually in Fig. 7.23, it is a hybrid between filament winding and tape laying. A fiber placement, or tow placement, machine allows individual tows of prepreg to be placed by the head. The tension on the individual tows normally ranges from 0 up to about 2 lb. Therefore, true 0° (longitudinal) plies pose no problems. In addition, a typical fiber placement machine (Fig. 7.24) contains either 12, 24, or 32 individual tows that may be individually cut and then added back in during the placement process. Since the tow width normally ranges from 0.125 to 0.182 in., bands as wide as 1.50-5.824 in. can be applied depending on whether a 12 or 32 tow head

Fig. 7.23. Fiber Placement Process1

is used. The adjustable tension employed during this process also allows the machine to lay tows into concave contours, limited only by the diameter of the roller mechanism. This allows complicated ply shapes, similar to those that can be obtained by hand lay-up. In addition, the head (Fig. 7.25) contains a compliant compaction roller that applies pressure in the range of 10-400 lb during the process, effectively debulking the laminate during lay-up. Advanced fiber placement heads also contain heating and cooling capability. Cooling is used to decrease the towpreg tack during cutting, clamping, and restarting processes, while heating can be used to increase the tack and compaction during lay-down. For the current generation of fiber placement heads, a minimum convex radius of approximately 0.124 in. and a minimum concave radius of 2 in. are obtainable. One limitation of the fiber placement process is that there is a minimum course (or ply) length, normally about 4 in. This is a result of the cut-and-add process. A ply that is cut or added must then pass under the compliant roller, resulting in a minimum length that is dependent on the roller diameter. Fiber-placed parts are usually autoclave cured on carbon/epoxy, steel, or low-expansion invar tools to provide dimensionally accurate parts. Typical applications for fiber placement are engine cowls, inlet ducts, fuselage sections, pressure tanks, nozzle cones,

Fig. 7.25. Fiber Placement Head Source: The Boeing Company

tapered casings, fan blades, and C-channel spars. The aft section of a V-22, in which the skin is fiber placed over cocured stiffeners, is shown in Fig. 7.26.

Extensive testing has shown that the mechanical properties of fiber-placed parts can be essentially equivalent to hand layed-up parts.14 Like hand layed-up parts, gaps and overlaps are typically controlled to 0.030 in. or less. One difference between fiber placed and hand layed-up plies are the "stair-step" ply terminations obtained with fiber placement, since each tow is cut perpendicular to the fiber direction. Again, this stair-step ply termination has been shown to be equivalent in properties to the smooth transition you obtain with manual lay-up. In fact, some parts have been designed so that either fiber placement or manual hand lay-up may be used for fabrication. Since the tows are added-in and taken-out as they are needed, there is very little wasted material; scrap rates of only 2-5% are common in fiber placement. In addition, since the head can "steer" the fiber tows, there is the potential for the design of highly efficient load-bearing structure.

The software required to program and control a fiber placement machine is even more complex than that required for an automated tape layer or modern filament winder. The software translates CAD part and tooling data into 7-axis commands, developing the paths and tool rotations for applying the composite tows to the part's curved and geometric features, while keeping the compaction roller normal to the surface. A simulator module confirms the part program with 3-D animation, while integrated collision avoidance post-processing of the NC program automatically detects interferences.

Fig. 7.26. V-22 Aft Fuselage Source: The Boeing Company

Modern fiber placement machines are extremely complex and can be very large installations. Most machines contain seven axes of motion (cross-feed, carriage traverse, arm tilt, mandrel rotation, and wrist yaw, pitch, and roll). The larger machines are capable of handling parts up to 20 ft in diameter and 70 ft long, with mandrel weights up to 80 000 lb They typically contain refrigerated creels for the towpreg spools, towpreg delivery systems, redirect mechanisms to minimize twist, and tow sensors to sense the presence or absence of a tow during placement.

Although complex part geometries and lay-ups can be fabricated using fiber placement, the biggest disadvantages are that the current machines are very expensive, complex, and the lay-down rates are slow compared to most conventional filament winding operations.

Was this article helpful?

0 0

Post a comment