The following is a review of some of the most important composites manufacturing technologies, including a discussion in each case of the current state-of-the art in the United States and Japan and the relevant research issues.
Spray layup has very little application in aerospace. This technology produces low specific strength structures which usually do not belong on the end product. Spray layup is being used to join back-up structures to composite face sheets on composite tools. Spray layup is also in limited use for obtaining fiberglass splash from transfer tools.
The JTEC team did not observe spray layup equipment in Japan, nor was evidence of its use seen in any of the tooling. The large bond jigs used for Boeing and Douglas parts are usually dictated by customer tooling handbook designs. This may stifle some innovation. Handbooks help avoid "reinvention of the wheel," but can promote "not invented here" problems.
Hand layup is the most common method of producing composites parts in the U.S. aircraft industry. At McDonnell Aircraft, 100% of the aircraft composites are produced by hand layup. An area which relates to this and which is covered in a later section is ply cutting and stacking.
The primary methods of automation in hand layup relate to computer software. Each ply has a separately identified part number.
Software is used to generate flat patterns from the layer surface and the ply boundary. Software is also used to find the most efficient nest of cut plies to minimize the scrap. Gerber-type textile nesting programs are used. Many different types of parts are laid up, and plies are often spoiled during handling. This usually means a different mix of parts are nested each day. Unidirectional parts may be broken up into two or three subparts to make a better fitting nest. The cut parts are usually hand sorted into ply packs and packaged.
The material to be laid up is delivered to the layup cell at the start of the shift. Most large layup facilities use electronic work orders and work instruction. Bar code is widely used for worker identification and part identification.
Plies are located on the tool in various ways. The most common is within the instruction set, in the form of a sketch or a verbal description. Mylar templates are used for large parts and with multiple, closely spaced ply boundaries. Mylars are generated with the same software that generates the digital data for ply cutting.
As in the U.S., hand layup is used extensively in Japan. The Japanese are not as automated on the shop floor. The U.S. is working toward a paperless system. The resolve to eliminate paper was not evident in Japan. There are not large numbers of CRTs in the manufacturing areas. Textile type pattern cutters do not, extensively, use bar code labelers. Plies are located by template or layout. The work force is so well trained that automation is not needed to ensure quality. If it does not pay off well, the Japanese will not pursue it.
Research in this area is directly improving the methods of locating the ply on the tool. The laser projector system is in limited use. It traces ply boundary, ply identification, and ply orientation information on the tool or composite surface. The system has contour limitations and requires high projection heights for large parts. Larger parts require multiple systems. A second system under development uses a video camera and a computer image on a CRT. Software provides the ply boundary information to the CRT and the video provides the actual position. The operator looks at the CRT while aligning the ply to the CRT image. It is believed that none of these systems are used for even limited production.
This technology is in widespread use and is probably the most frequently cited example of composites automation. These machines are primarily located in factories that make relatively large, mildly contoured parts. The machines are generally furnished by Milicron and Ingersoll, with head designs which originate at Grumman and Vought. Some of these tape layers have also been equipped with water jet cutters and stitching heads. One has been converted to fiber placement. This conversion consisted of a head change for running 12 individual tows, major software additions, and added heat control function hardware.
The Japanese use tape layers for some of their large parts. The tape layers are U.S. made and run the delivered software. The gap and overlap criteria are similar to those applied to U.S. made parts. The tape layers appear to run with less human intervention and with fewer stops than they do in the U.S.
Current tape layers put heavy demands on raw material suppliers in terms of tack, tape width, release paper adhesion, and material splice locations (or lack of splice). A cassette tape layer has been introduced, but is not in widespread use. It was designed to alleviate some of the material demands by separating the cutting and laying into two operations.
Tape laying research is an evolutionary, not a revolutionary, process. The major areas of interest are in contact show material, cutting improvements, and more user- friendly software. Some efforts are being made to do real-time quality assessment using line scan cameras and oblique angle strobe lighting. There was at one time a major effort to tape-lay thermoplastics. However, the overall emphasis on thermoplastics has dwindled. Most tape laying of thermoplastics involved hot shoes or laser welding. Some PEEK work was done using an amorphous film carrier, which became molten material when heated. This severely limited the temperature capability of the system, but did not provide good laminates without subsequent autoclave compaction. Both Milicron and Ingersoll are now emphasizing fiber placement. Milicron has developed one machine and has orders from Boeing. Ingersoll will be entering the market in 1993.
These technologies support the hand layup process. Ply cutting uses reciprocating knives, ultrasonics, lasers, reciprocating chisels, and most recently, water. All of these methods and also textile methods, such as steel rule dies, are in use in the U.S. The various methods have the advantages of speed, sharpness of inside and outside corners, multiple ply cutting, and the ability to cut very close to the fiber direction. Labeling the ply is also different on the various machines. The labeling usually is dictated by the next operation, which is clearing the cutter bed and sorting the plies. If a person is clearing the bed, the labels are alphanumeric and the nesting program has been adjusted so that all the plies in a stack are in close proximity. Some cutter beds are 60 inches wide and 60 feet long. Clearing and sorting can be a labor-intensive operation unless plies in a stack are within a few feet of each other, and remakes or add kits are at the head or tail of a table. If the plant has automated sorting, the label is usually bar coded. A few use alphanumeric character readers, but they are generally unreliable. Full automation of ply kitting has had many unsuccessful and expensive attempts. The purpose of automating the sorting and stacking job was to have no human intervention after the bar code was read. The combination of numbers of plies, ply kits, ply sizes, and speed demands has doomed most of these processes. The most successful attempt used bar codes and software to display the bin number and sequence in the bin. The person must then place it in the proper position. In most high volume operations, kits are packaged in heat-sealed poly with bar code identification. Most operations are just- in-time; therefore kits are not frozen prior to use. Some automated storage and retrieval is used to deliver kits to the workcells. This is prevalent in operations that are highly paperless.
The Japanese use ply cutting machines to support their hand layup process. These machines are all U.S. made and are running the delivered software. Nests looked to be more efficient for scrap reduction. This could be due to a more stable nest, possible because fewer plies were scrapped; another reason might be that the operators are willing to search the full bed to complete a kit.
In the area of research, major emphasis is on improved software and links to various CAD systems. Potential exists for some improvement from automated sorting, with IRADs looking into systems similar to mail sorting systems. Automated flat ply collation is being done by Boeing and by others on a limited basis. Two types of machines have been built to pick up plies and roll them into place. One system cuts the plies without release paper on either side, removes the ply with vacuum handlers, sorts the ply into hundreds of trays and then places the plies directly on a flat plate tool. The flat plate tool rotates with the pick-and-place arm taking care of X and Y translation. The other machine is built for Boeing and only takes the cut ply, orients it, places it, and does a small amount of debulking. A limited amount of work is continuing on these systems.
In Japan, all three of the "heavy industries" companies (MHI, KHI, FHI) showed interest in ply sorting and kitting. However, it seemed that no work was ongoing in this area. This is an area where cooperative efforts between the U.S. and Japan could take place to vigorously seek low cost methods.
This technology is used to apply more composite material than all other techniques combined. It is well suited for pressure vessels, therefore it has primary applicability in the missile business. Both intercontinental and tactical missile cases are built using this technology. Glass, aramid, carbon and boron fibers have been or are being wound. Wet winding is the most prevalent technique, utilizing resins with room temperature viscosity in the range of 2000 cps or less. These structures have void contents of 3 to 10%, with resin contents of 40% by volume. Winding with prepreg tow is used for structures that require higher temperature performance limits, and that also demand lower void contents to sustain loads other than tension.
Filament winding has been combined with other fiber application methods such as hand layup, pultrusion, and braiding. Classic filament winding involves a spindle with a carriage or carriages to apply hoop and helical fibers. Compaction is through fiber tension. Resin content is now primarily metered. The machines are generally all computer controlled with up to six axes independently monitored. The additional axis comes into play at the fiber turn-arounds. The extra head axis allows for better placement of the band, and more uniform band width.
Filament winding has been used to wind large wind-machine blades up to 150 feet long. Rectangular tubes with hand-laid axial fibers have been split and used for back-to-back channels. These have been used for floor beams. Filament winding is still in widespread use for pressure vessels and other missile components. Some helicopter rotor blades are still manufactured utilizing highly modified filament winding processes.
Winding engine nacelles has been done for many years, but is now being done with prepreg tow. The MD-11 large center engine nacelle is made using hand layup; however, attempts are underway to filament wind.
In Japan filament winding is performed using U.S.-built machines. Most applications observed by the team used prepreg tow. Tooling is unique, utilizing mandrel expansion for pressure applications, and a disposable outside diameter rubber sleeve for surface finish.
Some tape winding is performed using modified filament winders or Japanese-built machines. These machines are making contoured broadgoods for subsequent cure in an outside mold line tool. Figures 1.3 and 1.4 depict a typical wrap machine and the resulting panel.
Figure 1.3. Tape Winding Device
Figure 1.4. Full Scale Panel
As far as research is concerned, filament winding is a very mature technology with innovation and technical advancement at very low level. Computer hardware and software continue to change as robotic advances are made. There are few filament winding machine suppliers left in the U.S., therefore most of the modifications and improvements are being made by the users. Wet winders are being improved in resin control and fiber wetout. In the wet wind process utilized in line impregnations, wetting is the speed-limiting factor. The most ambitious in-line system uses an alpha source and detector to measure the mass before and after an impregnation orifice. The orifice area is controlled by an inflatable metal ring. The total system is a closed-loop, computer-controlled network, capable of variable resin contents. Some winders preimpregnated the spool in a vacuum chamber prior to installation on the machine.
This technology is a combination of filament winding and tape laying. It probably has the most promise as a practical automation tool for skins. There are currently four machines in the U.S. industry. This technology utilizes preimpregnated tow. The fiber placement machine tracks along the tool surface, laying up to 36 tows of preimpregnated tow. The fiber is laid with essentially zero tension; therefore tool concavity is possible. All current machines, except one, include spindles. The machines with spindles handle closed-form as well as flat plates. The only company building production or semi-production parts is General Electric machine. That unit is building blades for the GE-90 High-Bypass engine. Hercules has built numerous CRAD parts. The Hercules technology seems cost-effective and produces structures consistent with hand layup.
The technology has been used to produce skins with substructure embedded in the tool and the part co-cured. This concept is inside-mold-line tooled. The technology has been used to produce contoured skins on a mandrel tool which are then removed. The compacted skins are placed in an outside mold line tool and stiffeners are placed on the skin and co-cured in the outside mold.
Fiber placement has also been used to fiber-place box substructures with additional tooling added and the skin fiber placed over the uncured boxes. Still the largest cost is forming the substructure. The most automated process for the substructure is the Boeing channel laminator. It uses tape-laid ply packs, and hot drape-forms the beam.
Tow placement is still a curiosity in Japan and did not seem to be well understood. There were many requests for visits to Hercules to view fiber placement and discuss technology transfer.
The forming of stiffeners was briefly discussed. One project for the forming of closed free edge blades was observed. It is called an omega stiffener and is shown in Figure1.5, along with a sketch of the machine that makes it. These stiffeners exit the process uncured and enclosed in the silicon rubber cure mandrel.
The area of fiber placement is ripe for extensive research. The fiber lay-down head is branching into two areas of focus. The first is a complex head that forms the tow to the proper thickness and width in the head. The second is the type of head that requires tow from the material supplier at the proper width, which in turn simplifies the head function, but increases the material costs.
Other areas of research concern expanding the capabilities of fiber placement to lay thermoplastic and polyimide. Current thermoplastic work involves laser heating at the lay-down point. This concept is also being explored for thermoset. The material would not cure at lay down but would be well advanced through the viscosity curve. Work is being done on conformable rollers and on reducing the minimum course length, which is currently about six inches.
This method is primarily low technology; however, there have been some high- technology enhancements to the pultrusion process. Feeding in plus and minus bias weave fabric in place of the normal mat feeders is being used. The resins used in combination with some changed die geometry are allowing higher Tg matrix systems to be used while still having reasonable pulling power.
Figure 1.5. Folding Process for the ê-Stringer
Research. When considering low cost, one thinks of continuous processes. Pultrusion is one of the few continuous composite fabrication processes. Research is aimed at introducing carbon fiber, epoxy matrix, and fiber orientations other than zero. A definite objective is the pulling of B- or C-staged structural members for subsequent forming and co-curing. The Xerkon process produced C-staged stiffeners, which were subsequently formed and co-cured to the V-22 fiber-placed skin.
Two research projects were observed in Japan which could be called pultrusion. Both seemed destined to produce very slow process speeds, but that slow speed may allow a very reliable machine. One machine is being researched to build carbon epoxy tubes for the Japanese space station. It is a typical MITI research project in which a research company is formed for a period of 10 years. The company employees come from the various companies that have an interest in the project. This particular research company is led by Mitsubishi Electric. Figure 1.6 shows the machine. It is similar to a driveshaft machine patented by Hercules.
The second pultrusion project is being performed by an independent Japanese company (JAMCO) which has developed a machine for pultruding curved and straight "L" and "T" stiffeners. This machine is shown in Figures 1.7 and 1.8 pultruding a "T" shaped frame. This again is a slow process that uses carbon/epoxy prepreg tape and fabric. This same company uses a similar process to produce flat, glass faced, cored panels. Both of these processes would appear undesirable to an American, who would immediately ask: "How many pounds per hour will it lay?"
This technology has a potential for cost reduction equal to that of fiber placement. Currently, we know of no production application of RTM in aerospace. Non- recurring costs for the tool are very high and recurring preform costs are high. The resin content requirement at 40% by volume also slows the process turnaround.
Research. Aerospace applications of this technology are going in various directions. Cheaper preforms is one avenue. Lubricating the preforms to allow the mold to close over the bulk factor is another major thrust. Tooling which has a larger volume during resin transfer and then seats to final position after filling is being pursued. Some research is actually targeting a structure as large as a wing skin with stiffeners.
This is another area of MITI interest. A "10-year" research company was created to thoroughly investigate the weaving of preforms. When a preferred concept is recognized, a machine is to be developed which will economically produce the preform. This company is currently six years into the 10 years, and is presided over by Mitsubishi Electric. See Figures 1.9 and 1.10 for examples of weaves.
Figure 1.6. Continuously Formed Graphite/Epoxy Composite Tubes for Large Space Structure
Figure 1.7. JAMCO's Continuous Curved Pultrusion Machine (side view)
Figure 1.8. JAMCO's Continuous Curved Pultrusion Machine (front view)
Figure 1.9. 3-D Fabric RTM (1)
Figure 1.10. 3-D Fabric RTM (2)
This technology appeared to have applications during the prime of thermoplastics in aerospace. In Japan there is still a major emphasis on thermoplastics, and some 10-year projects still have three to five years to run. Double diaphragm forming using aluminum carrier sheets was all that was observed. All other thermoplastic work was flat.
There are some promising research projects in the U.S. which are modeling and forming thermosets using rubber diaphragm carriers. We saw no comparable work in Japan.
This material has not been readily available to aerospace. It was promoted for use in a thermoplastic matrix, in order to allow hot thermoplastic/LDF to drape without buckling. No production aerospace parts are currently produced, and material qualification data have not been established.
Research. This material is now being offered in a thermoset matrix. The objective of this research is to find a suitable application for this material. It may have an application as inner stiffening panels on doors and covers. It is also being considered for the web of tall beams to form corrugations or for stiffness.
The Japanese are familiar with long discontinuous fiber; however, no thermoplastic or thermoset projects were observed.