Commercial-grade heavy tow carbon fiber has been improved over the years to the point that its properties rival that of aerospace-grade 12K fiber, in terms of tensile strength and tensile modulus. Low-cost heavy tow is a natural choice for commercial, high-production applications, especially those involving filament winding technology. Filament winding equipment
Source: Entec
Simple changes in the fiber handling components on filament winding equipment have vastly improved product performance and aesthetics when using commercial-grade 50K carbon fiber tow
But, filament winding companies have historically been reluctant to use heavy tow for many applications, citing a variety of performance shortfalls, including inconsistencies in both fiber spread and fiber wet out, as well as excessive fuzzing, poor fiber-to-part strength translation and poor part aesthetics. Zsolt Rumy, president of Zoltek Corp. (St. Louis, Mo.), a major manufacturer of commercial grade fiber, decided to address the issue head on. He tasked a team at Entec Composite Machines (Salt Lake City, Utah), a Zoltek-owned composites equipment and machine software producer, to come up with changes to filament winding equipment to address customer concerns.
"Our success criteria were to achieve good translation of the modulus and strength of large tow to the finished filament wound parts, together with easy fiber handling and visual cosmetics similar to 12K or 24K fiber," says Entec's president Robert Murdock. "Our goal was to use the simplest possible approaches requiring the least amount of hardware for easy retrofits onto existing filament winding equipment."
Correcting fiber feed
In a nutshell, wet filament winding works by unspooling multiple tows of dry fiber from a creel, pulling the fiber through a resin bath and then applying it to a rotating mandrel, usually in a simple tubular shape. Traditional filament winding with 12K tow involves guiding the fiber tows through a series of rollers, rings, eyelets and, ultimately, a multiple-pin "comb" just behind the delivery head, which keeps the tows separate until they are brought together in a single combined band on the part, as the mandrel rotates. If the part being produced is a fairly small pressure tank, for example, perhaps four 12K tows might be needed to make up the correct fiber "bandwidth" of about 9 mm/0.375 inch that is laid down with each pass. Larger parts need a correspondingly wider bandwidth and, thus, more fiber tows.
The biggest obstacle to straight substitution of 50K heavy tow for 12K, says Entec project manager Dave Rasmussen, is that one heavy tow is equivalent to four 12K tows bunched together, with a strong "lateral tenacity" that tends to inhibit spreading of the filaments into a flat band. While a slight twist or misalignment of one 12K tow might be barely detectable in a part, failure to break down and flatten rope-like 50K tow is definitely noticeable, and can have a much greater effect on part performance.
As the first step in addressing large tow winding, an Entec team fabricated sample tubular parts and pressure vessels to create a baseline, using a standard 4-axis winding machine. The machine, set up to handle 12K carbon fiber, was loaded with Zoltek Panex 35 50K fiber spools. The team found that, just as customers had reported, fiber handling was often a problem and resulting parts had poor cosmetics and less-than-desirable strength translation.
Four principal machine areas were identified where improvements might be possible: 1) the fiber creel; 2) resin bath; 3) creel-to-head transfer; and 4) the delivery head itself. Specific equipment add-ons or new parts were designed and tested for each area, says Murdock, in a series of experiments in which spool tension, payout speed and fiber sizing also were varied.
The creel, usually situated behind and several feet away from the filament winder, contains the multiple spools of dry fiber that are fed into the machine. In a typical 12K winding machine, the creel holds the fiber spools horizontally, at right angles to the winding axis. The tow moves through sets of small vertical and horizontal rollers that change the fiber's direction several times as it moves out of the creel and toward the resin bath.
However, when large tow was loaded into the 12K creels, filament breakage and subsequent fuzz generation occurred as fibers were pulled out of plane through the rollers. "We had to find a way to minimize the out-of-plane fiber unspooling," says Murdock. Entec addressed the fuzz problem by aligning the fiber spool axes parallel to the winding machine's rotational axis (Fig. 1). This kept the fiber correctly in plane as it unwound, thus minimizing breakage. In addition, the creel was fitted with a fan and filter to contain any generated fuzz, minimizing the chances of it reaching the work area and resin bath. Pressure tank winding
Source: Entec
Pressure tank winding with two Panex 35 tows on a 4-axis machine, with all fiber handling improvements in place.
In an effort to start the tow breakdown and begin the flattening process, Entec tried two approaches. In the first, each spool was fitted with an individual tow roller that collected the tow from the creel and turned it 90° from horizontal to vertical, as it was pulled toward the resin bath. The polished roller was fairly effective at aligning the fibers, says Murdock. A second method involved collecting the tow in 100-mm/4-inch wide steel D-rings, made from 12-mm/0.5-inch diameter bar stock. The D-rings work well at accommodating the side-to-side motion of the tow as it unspools, and as the spool gets smaller and changes in diameter. According to Murdock, they also "trap" fiber twists and help to flatten the tow as it unwinds.
A final adjustment was aimed at ensuring adequate and adjustable fiber tension on each spool. "In order for the D-rings to work and spread the fiber," says Rasmussen, "about 2 lb to 3 lb of counter-tension or drag is required on each spool to keep the tow taut." Spool tension can be applied with a closed-loop servo control mechanism, in which a load cell senses the tension, or by using a simpler, spool-diameter-compensated, constant voltage brake, both of which are manufactured by Entec, he explains. But the amount of filament breakage and subsequent fuzz is directly related to fiber tension as the fiber unspools, he adds, so fiber tension at the creel can't be too high. Hoop winding
Source: Entec
Photo: Hoop winding with two tows, giving a bandwidth of just over 25 mm/1 inch. The next challenge was to optimize fiber spreading upstream of the resin bath. During the baseline sample part trials, says Murdock, tow width varied widely from 6.4 mm to 19 mm (0.25 inch to 0.75 inch) prior to wetout, which was causing highly variable resin pickup. "We experimented with both stationary and more dynamic, intrusive means to achieve a consistent 0.75-inch spread," he says.
Static means included pulling the fiber over and under a pair of "bumpy bars," essentially bars with variable diameter. A pair of angle bars, made from 76.2-mm/3-inch square tubing with a 9.5-mm/0.375-inch radius, also were tried. Both bar types were chrome-plated and highly polished, to minimize fiber damage and resulting fuzz. The team determined that the square angle bars were better than the bumpy bars at working the fiber.
More aggressive means included a grooved expanding bar and a kneading roller. The radially expanding grooves of the expanding bar and kneading roller bite into the tow as it passes between and help overcome the tendency of the tow to bunch together due to criss-crossing individual filaments (Fig. 3). A vibratory bar also was tried to see if a low-amplitude shaking motion would assist in spreading the filaments apart. Held in place with spring mounts, the 12-lb force, adjustable electromagnetic vibrator's effect was time-dependent — that is, the extent of filament separation depended on the tow's dwell time on the bar. Since increasing dwell time would necessarily decrease the winding rate, setting quality and efficiency goals at odds, the device was eliminated from consideration.
In the end, the simplest means worked the best: static bars (Fig. 2) created the best fiber spread, with minimal hardware changes. "Standard Panex 35 with normal sizing doesn't require a lot of special hardware," says Rasmussen. Spools of 50k tow
Source: Entec
Figure 1: Spools of 50K tow in the creel are aligned with spool axes parallel to the winding machine rotational axis, to keep the fibers in plane and avoid out-of-plane turns. D-rings that collect the tow and start to flatten it are visible at left center. Pre-resign bath
Source: Entec
Figure 2: An overview of the pre-resin bath shows the spreading hardware, including a combination of bumpy bars and angle bars and a roller that leads into the bath. Arrangement of two kneading rollers
Source: Entec
Figure 3: One aggressive option to the bumpy bar/angle bar configuration was this arrangement of two kneading rollers. The fiber band passed between the two rollers, which bit into the tow to disentangle filaments. An adjustable spring on the lower roller maintained pressure. Standard resin bath pickup roller and doctor blade
Source: Entec
Figure 4: The standard resin bath pickup roller and doctor blade (fiber is moving into the foreground). The tow picks up resin as it passes over the pickup roller, then turns from horizontal to vertical as it exits the bath toward the head. A ceramic eyelet guides the tow
Source: Entec
Figure 5: A ceramic eyelet guides the tow and provides a point location on the way to the head. Series of vertical rollers
Source: Entec
Figure 6: An optional alternative to the eyelets is a series of vertical rollers, which worked well to maintain spread in the tow band. A rotating head machine configuration
Source: Entec
Figure 7: A rotating head machine configuration. Vertical rollers, straight static bars and the curved band forming bar just upstream of the part are visible. The curved forming bar and application roller
Source: Entec
Figure 8: The curved forming bar and application roller are clearly visible as the band is laid onto the part mandrel. Concentric D-ring spreaders
Source: Entec
Figure 9: A nonrotating or static head works best with concentric D-ring spreaders. The rings can be interchanged and moved relative to each other to change tension and provide a degree of control over spreading.
Interchangeable parts
Once consistent fiber spread was accomplished, attention was focused on effectively wetting the dry fibers in the resin bath. Entec tried three different approaches: 1) an immersion bath; 2) a traditional doctor blade roller bath; and 3) a single-tow injection system. The immersion bath simply submerged the fiber tow into the resin, then pulled it through a set of wipers or rollers to "squeegee" off the excess resin. For some types of heavily sized fiber, the immersion bath worked fairly well and controlled the resin/fiber ratio with a higher tolerance than anticipated, says Murdock. The doctor bar system is essentially a polished rotating cylinder that sits in the resin bath and picks up resin as it turns. The doctor bar presses against the cylinder to create a precise resin film thickness and pushes excess resin back into the bath. As the fiber tow comes over the top of the cylinder, it contacts the resin film and wets out, and resin content is accurately controlled. The third approach, the inline tow injection system, required a lot of additional equipment, including circulating metering pumps and a resin reservoir, and it required that the tows be pulled through a set of machined orifices. While this type of system generated some interest, there was concern that any fiber breakage or fuzz would eventually clog the orifices. It was ultimately deemed unnecessary, as the tried-and-true doctor blade and resin pickup roller system (Fig. 4) worked well with only minor modifications, including elimination of "combs" normally seen on 12K machines. "The pins in traditional combs tend to rope the fiber and prevent good wetout," explains Rasmussen.
To transfer the fiber to the winding head, two basic options were investigated. One was to use ceramic "eyelets," essentially doughnut-shaped, low-friction satin-finished ceramic guides with a 19-mm/0.75-inch diameter center hole and an approximate 6.4-mm/0.25-inch radius (Fig. 5). "The eyelets were a simple way to accommodate the axes of head motion while directing the fiber to the proper location on the carriage," explains Murdock.
The second option was a vertical roller system (Fig. 6) that maintained the flat tow spread from the resin pickup roller and accommodated side-to-side motion of the delivery head along the carriage. While the eyelets simply provided a point location to direct the tows, the vertical rollers (one set for each tow) better maintained the flat tow band from the resin bath. Both options were selected as suitable for customer machines, but where space allows, the vertical roller option is preferred, says Murdock. "Once we have the fiber spread, we want to maintain that spread through the system and not neck it back down through eyelets, if we can help it," he says.
According to Murdock, three classes of head fiber control and spreading hardware were manufactured: hardware for a rotating eye (Figs. 7 & 8), hardware for a static eye (Fig. 9) and hardware that can be used on either system. Rotating vs. static eye refers to the roll motion of the head as it places fiber on the mandrel. A static head can only move back and forth along the carriage as the mandrel spins parallel to it. A rotating head can be programmed to orient the fiber band to the wind-angle, to maintain spread as it is guided around the end domes of a pressure vessel, for example.
Interchangeable equipment for both head types included straight and curved static bars; bumpy bars; chrome-plated tow redirect rollers; concave redirect rollers; and convex spreading rollers. The rotating eye required a separate head roller while the static eye needed additional D-rings.
Rasmussen points out that, for 50K tow, the standard movable comb used for 12K fiber bandwidth control had to be discarded in favor of a curved band forming bar. "The comb tended to bunch the fibers up again after we'd spread them," he explains. "Control of bandwidth is achieved by bringing the tows over and under adjustable curved bars near the final delivery point. The final delivery element at the part may be a D-ring in the case of a nonrotating eye or a cylindrical roller with a rotating eye."
Testing shows improvements
With machine modifications in place, the team began to wind and test a new set of parts. Results for sample pressure vessels showed that the structural translation factor averaged a surprisingly high 93 percent, meaning only 7 percent of the tow's structural properties were lost during manufacturing — a better outcome than can be achieved with most manufacturing processes.
Burst pressure tests on samples improved significantly. For example, a baseline 152.4-mm/6-inch diameter tube burst at 245.25 bar/3,557 psi, while a similar tube wound after the modifications had a burst result of 427.2 bar/6,196 psi. Observable changes included improved part aesthetics, decreased void percentage and increased fiber-to-resin ratio, says Murdock. In addition, fiber speeds were the same as those used for 12K. "No speed changes are required with the heavy tow," Rasmussen maintains.
More than two dozen customers in the U.S., Europe, Asia and South America have purchased the machine improvement package, which has been easily retrofitted onto their existing machines. "The perception of large tow is the problem, not the fiber itself," Murdock concludes. "We've developed a practical, low-cost solution for filament winders that allows them the flexibility of using a lower-cost carbon fiber."