RIM Molding for Wind Blade Skins
By anyone's measure, wind power is a good thing. It's clean, renewable and helps cut consumption of fossil fuels - and the industry that produces it has grown more than 30 percent per year in the last decade.
While not as cost-efficient as today's big utility turbines, at least 30,000 small turbines with 65 to 105 kilowatt (kW) ratings were installed during the 1980s at U.S. sites, such as California's Altamont Pass, and elsewhere around the world. While they played an important role in the development of the wind power industry, these turbines are approaching the ends of their public service contracts, yet they are still serviceable, says National Renewable Energy Labora-tory (NREL) senior project leader (II) Trudy Forsyth, a key player in the U.S. Department of Energy's (DoE) Low Wind Speed Technology (LWST) for Small Wind Turbine Development program. LWST's goal is to find new uses for these older machines in wind regimes that aren't as attractive to large utilities. One approach is refurbishment and reuse of smaller turbines to supply power to individual farms and ranches as well as small communities at a fraction of the cost of a new turbine. A number of U.S. states offer energy incentives under this program, including power buy-back programs, to foster the increased use of "distributed"wind power throughout the country (that is, a dispersed, widespread power grid with many small generators). LWST is researching ways to produce critical replacement components - primarily, the composite turbine blades - at a competitive cost.
"We asked ourselves: What's an inexpensive way to produce a lot of lightweight blades that could be used to retrofit these older turbines?"says Forsyth. "A low-cost, mass-producible and lightweight 'universal' blade that can work on both older and newer turbines was a key objective.â€
One possible solution is a wind blade with a carbon fiber spar and reaction injection molded (RIM'd) skins. This concept is currently under investigation by Composite Engineering Inc., Blade Group (CEI, Concord, Mass.) in a project that won DoE/NREL funding in 2003. CEI's project team included LJM Consulting (Amherst, Mass.), which designed the airfoil shape and performed finite element modeling; SweetBriar (Springfield, Mass.), responsible for mold development and manufacturing; and Federal Engineering Assoc. (McLean, Va.), which handled market research and potential business models for the blade. SweetBriar was assisted by RIM-molding specialist Paramount Plastics (Abbingdon, Va.) and toolmaker Art Mold and Polishing (Roselle, N.J.). Several prototype blade sections have been produced, and structural and field testing is planned for late this year.
The Design Process
"The project team wanted a manufacturing method capable of producing an aerodynamically superior and lightweight wind turbine blade,"says David Wright, SweetBriar's owner and an expert in plastic molding and tooling development. Targeted blade lengths range from 24.4 ft to 32.5 ft (7.5m to 10m), for 60 kW to 100 kW turbines. In contrast to the labor-intensive hand layup or resin infusion methods now employed by most turbine blade manufacturers, Wright wanted a high-volume, low-cost manufacturing method that could maintain near-optimum blade shape and taper.
"Ours is a radically different design approach,"states CEI's design engineer Forrest "Woody"Stoddard. Traditional wind blades have an I-beam spar; spar caps are formed by buildups in the upper and lower blade skins, with the beam web bonded in separately to connect the caps. The RIM blade has an entirely separate carbon composite spar. The RIM-molded plastic skins are unreinforced and just stiff enough to act as fairings while load is transferred to the spar, explains Stoddard. When the wind gusts at higher speeds, the loads cause the skins to deform elastically to shed excess load without adverse affect on the turbine, he maintains.
CEI is responsible for spar manufacture. Using the same method it employs for composite masts and rigging on yachts, the company inline braids 48K carbon yarn and S-glass over a reusable, fiberglass-carbon/epoxy mandrel to form a triaxial "sock."The mandrel, with a high-density polyurethane foam core, was designed with a circular cross-section at the root end that transitions to a D-shape along the blade length. The curved portion of the D forms the leading edge. Carbon fiber for the braiding is supplied by Grafil Inc. (Sacramento, Calif.) and Toho Tenax America Inc. (Rockwood, Tenn.). The carbon fiber is placed in the 0* or lengthwise spar direction, with the S-glass forming the +/-45* angle fibers, which keep the carbon tows properly aligned, says Stoddard. To complete the process, spars are vacuum-bagged and infused with epoxy resin, then cured in an autoclave. Final fiber-to-resin ratio is at least 65 percent.
Wright, meanwhile, has concentrated on a RIM mold design for the compliant blade skins. RIM molding, first developed by Bayer MaterialScience LLC (Pittsburgh, Pa.) for thermoset polyurethanes in the 1960s, is an economical closed molding process that involves mixing two highly reactive, low-viscosity, fast-curing resin components and injecting them into a mold at high velocity. Processing temperatures are typically less than 100F/38C, and internal mold pressure is less than 150 psi/10.34 bar, permitting the use of relatively low-cost mold materials. The team selected RIM because of its relative simplicity and low energy demand. While composite tooling can be used for RIM, matched metal molds are generally preferred for their good part cosmetics, accuracy, durability and longer useful lives (assuming high production rates). Wright says that in addition to polyurethane (PU), polyamide (nylon) and dicyclopentadiene (DCPD) also are suitable for RIM.
To create two-part metallic clamshell molds, Wright began with a blade model created by LJM Consulting with SolidWorks software (supplied by SolidWorks Corp., Concord, Mass.). He selected a parting line along which the virtual blade could be split lengthwise into upper and lower skins, adding details that would be necessary to locate and bond the spar between the two. These included lengthwise raised ribs (longerons) on the inside surfaces of both skins, which hold the cured spar in place, as well as a groove on the inside of the lower skin into which the trailing edge of the upper skin fits. Innovative, interlocking, snap-together "teeth"mate the skins together at the leading edge, a challenging feature for a part of this size.
Matched metal mold sets then were modeled in SolidWorks - one set for the upper blade skin and one set for the lower skin. Wright and the team selected machined aluminum, which has the advantages of high heat transfer and a lower cost than tooling steel. The 3-D mold files were transmitted to Art Mold and Polishing and imported into a three-axis CNC center, which machined four solid aluminum billets to form two 58-inch/1.5m long prototype mold sets for the blade skins at mid-section (minus the root and tip), for prototyping trials. The tip shape is still under discussion: "A series of solutions will be tried, including a 'sword tip,'"says Wright. Because the tip generates a wind blade's acoustic signature, the team is trying to minimize noise while increasing tip speed for better efficiency.
Each mold set has a female cavity and a matching male "core"that fits into the cavity, with a precise gap between them to control blade skin thickness. Steel rail support structures were welded on both the cavity and core molds to facilitate lifting and moving of the prototype molds. To enable manufacture of a "family"of blade sizes, the mold design is modular (see size chart at the top of this page) - permitting additional sections to be bolted on at the root end to increase blade length. Simple flanges and bolted dowels will be used to hold the segments together.
"The matched metallic mold sets produce parts that are accurate, finely detailed and extremely consistent,"says Wright. This means the geometry of the part is tightly controlled, he explains, which will ease and simplify the process when the molded blade skins and the spar are assembled.
Fast & Cost-Effective
To date, the team has produced test parts with the mid-section molds; the remaining mold modules are under construction and should be finished within a few months. To keep costs low, the molds can be operated freestanding, meaning no press is required. This is a good thing, notes Wright, since "there is no off-the-shelf press configured for a long, skinny part."Each mold sits with the female cavity section on the floor (skin appearance side down) and the male core section is lowered into place from above via forklift or gantry. This ensures that any gas bubbles formed during molding will rise and vent at the surface of the upper core mold half, which forms the interior surface of the blade - making pinholes invisible. Toggle clamps hold the top and bottom mold halves together during injection. To control mold temperature, water channels were drilled about 4 inches to 5 inches (100 mm to 127 mm) apart and about 1.5 inches/38 mm from the face. Since mold release is sufficient to allow easy part extraction, says Wright, air-driven ejector pins, normally specified for high-volume apps, were not used to save cost and time.
Key to the RIM process is the "mix head"built into each mold core. As shown in Step 3, it contains a large piston precisely fitted inside a machined bore. In operation, it acts as an impingement mixer: As the plunger is retracted from the bore, the metered amounts of A and B resin components are instantaneously pulled into the chamber at a high flow rate, causing the components to mix thoroughly. When the plunger is activated, the mixed shot is injected into the mold within seconds. "The pressure in the feed lines can reach 2,000 psi [137.9 bar],"says Wright. "Once the resin enters the sprue and fan gate, pressure drops to between 50 and 100 psi [3.45 bar and 6.89 bar], which is sufficient to move the resin through this large mold without a high-tonnage press."While a RIM mold would normally contain hot runners to distribute resin, Wright designed the lengthwise longerons in the part to serve as runners, again reducing mold machining costs.
In a production scenario, the A and B resin components - for PU, polyol and isocyanate - are stored separately in two "day tanks"on the shop floor. Each tank is jacketed for temperature control and equipped with internal mixers to ensure uniform blending of any added fillers, although in this case, no fillers are used. Pumps equipped with heaters keep the tank contents moving slowly through high-pressure hydraulic hoses to the mix head - fixed in "recirculating mode"so that the components remain unmixed - and back to the tank, to maintain consistent viscosity and temperature.
When an operator calls for a shot of material, hydraulically driven "lance pumps"near the bottom of each tank retract and meter the A and B components into the high-pressure hoses that connect the tanks to the mix head. "Our RIM equipment and lance pumps can deliver a range of ratios to accommodate the resin type,"says Wright. In addition, the amount of resin delivered can be varied to allow use of a range of shot sizes for different molds. Although the final blade's length suggests the use of two mix heads, which would move the resin shot into the mold more efficiently, Wright has decided to rely on one mix head per skin mold to avoid formation of knit lines in the part caused by converging resin flow fronts.