Machine Design

Here Comes Rapid Tooling

New materials and better processes let rapid-prototyping machines fabricate tooling that is good enough to produce parts.

Even in the early days of rapid prototyping, the long term goal of practitioners was to quickly generate real parts rather than just concept molds. Today that goal is being realized. The same techniques that generate fragile prototypes are increasingly being extended to provide tough production tooling. In many cases this tooling produces parts having tolerances and surface finishes acceptable for finished products.

The first examples of rapid tooling became available in 1993 when Soligen Technologies Inc. commercialized a process invented at M.I.T. for creating casting molds more quickly without patterns. Three years ago, DTM Corp. in Austin developed a method called RapidTool for creating powder- metal green parts that are sintered and infiltrated with copper to produce production molds. More recently, a process called Keltool has also been used to quickly make tool-steel molds. Now owned by 3D Systems, Valencia, Calif., this process results in injection molds capable of withstanding a million shots when used with nonabrasive resins.

Recent developments aim at reducing the amount of postprocessing required for rapidly produced molds. Researchers are also experimenting with new materials such as thermoplastics, composites, and maybe even cement. Bolstering this work are improvements in underlying RP hardware, particularly lasers.

It has long been a goal of rapid-prototyping companies to use powder metal as an RP material. The newly devised Laser Engineered Net Shaping or Lens process now uses powder metal to build parts with excellent material properties, say its developers. The process comes out of Sandia National Laboratories and works like most additive RP methods: A computer slices a solid model into many layers, and a machine builds parts layer-by-layer from metal powder or liquid resin. The machine uses a 700-W laser focused onto a substrate to create a molten puddle into which a mechanism injects metal powder. A computer moves the substrate relative to the laser beam depositing thin metallic lines until the part finishes.

However, engineers with developer Optomec Design Co., Albuquerque, stress that Lens can build a limited overhang, to about 30°. And because it works slowly compared to other systems, at rates of 0.3 to 1.0 in.3/hr, quick production of tools rather than parts will be its forte. The firm hopes to eventually build molds with internal sensors to more closely monitor pressure and temperatures. Molds devised in the Lens process could also feature built-in cooling channels that conform to the shape of the part. This optimizes cooling and will allow the shortest possible process times.

The half-dozen materials tested so far include 304 and 316 stainless steel, ironnickel alloys, and titanium. Surface finishes are also encouraging. A laser-remelt process improves the standard finish of about 400 μin. to about 10 μin.

Other developments in rapid tooling focus on processing new materials in existing machines. For example, one experimental rapid-production method showing good results so far marries powder injection-metal feed stocks, from injection-metal molding methods, with an RP machine from Sanders Prototype Inc., Merrimack, N.H. Powder-injection metals contain a polymer that lubricates and suspends the powder as it’s deposited in the RP machine.

“A lot of what’s happening now is just the confluence of existing technology,” says Rand German, professor of powder metals at Pennsylvania State University, University Park, Pa. German is working with the Sanders machine partly because it boasts one of the best accuracies available. This makes it a good candidate for producing short-run complex parts such as gears.

“A group member recently benchmarked available powder metals for strength, wear, and dimensional change,” says German. “Then we set goals to top their characteristics and limit shrinkage to near zero,” he says. With a proprietary combination of offthe- shelf powders, German’s crew came up with a material having an Rc hardness range of 30 to 35, a wear resistance that is four times better than tool steel, and a surface finish of two micrometers as sintered.

Other efforts in rapid tooling concern the making of production molds out of ordinary SLA resins. Though SLA epoxies would melt at typical injection-molding temperatures, judicious cooling can let molds last long enough to make a few dozen parts. The process works like this: An RP machine constructs a core and cavity with hollow backs. Copper tubing for cooling is looped about the protrusion of the part shape inside the mold. The shell is filled with an epoxy mixed with aluminum powder to improve heat transfer, and 24 hr later the mold can be mounted to a base ready to make parts. “We can get 20 to 100 parts out of a mold depending on the material,” says James Mishek, developer and president of Vista Technologies LLC, White Bear Lake, Minn. “Polycarbonate must be shot at high temperatures. The technique gives about 20 parts before the mold breaks down. But rubber or polypropylene is less abrasive and tends to give closer to 100 parts,” he adds. Accuracy has been to about ±0.002 in. The advantage of the system is that it produces parts in the production material in less than two weeks. Mishek also says he expects SLA materials soon that tolerate temperatures beyond the 158°F limit of the most rugged existing resins. Then the part count and surface quality from a mold should improve.

When Berok Khoshnevis was troweling plaster onto the wall of his home not long ago, it occurred to him that a small trowel on the nozzle tip of a rapid-prototyping machine might remove the layered appearance that characterizes these parts. He tested the idea and it worked. With funding from the National Science Foundation, the professor at the University of Southern California came up with a system that could work with a wide variety of materials. Theoretically, it can work with any substance that can be extruded. The system can use standard thermosets, thermoplastics, and photopolymers as well as those not associated with RP such as plaster, cement, clay, or concrete.

So far Khoshnevis has built a limited-capacity machine and fabricated smoothwalled cones and cylinders as test shapes. The build method resembles fused deposition modeling, extruding UV curable or airdry materials one layer at a time, which Khoshnevis terms as Contour Crafting. “Surface quality is improved by removing the obvious stair steps,” he says. One cylinder built by the machine has a surface of 24 microns and was made with 2-mm layers. The machine works faster than existing systems that use fused deposition. Preliminary studies show that even for layers 0.25-mm thick, the process would be about eight times faster and give a much smoother surface than current FDM. Use of thicker layers would speed the process even more without affecting surface quality.

The mechanics involved in Contour Crafting are scalable. This opens the door to the possibility of RP machines that incorporate large gantry robots. These might produce not only large prototypes but also final products that could include boats and internal components for aircraft and automobiles. What’s more, structural properties of large parts could be improved through the introduction of resins with fillers such as fibers or glass.

So what next? Pennsylvania State University’s Rand German half jokes that since most developments have come from shoestring budgets, “with a little serious funding, we could be dangerous.” He’s suggesting there are many more ideas looking for funding and the rapid tooling we’ve seen so far is just the tip of the iceberg.

To test the design of a new water-pump impeller, engineers with Airtex Product, Fairfield, Ill., used a recently introduced SLA material that was fitted around a steel rod. Mounted in an aluminum housing, the device ran at speeds of up to 5,000 rpm until it broke. What’s notable is that the impeller was moving a relatively hot 225°F-water-glycol mixture at 48 gpm and about 17 psi. The material, SLA 5210, comes from 3D Systems, Valencia, Calif.

“Because there were almost no step lines in the SLA prototype, we were able to more accurately gauge the performance of the impeller under working conditions,” says Kerry Austin, product designer for Airtex. “We might even be able to hit 7,000 rpm on future applications by adding strengthening features such as improved fillets or increasing the draft on the impeller.” Engineers with 3D Systems credit an advanced vinyl ether/acrylate formulation for the durability of the resin. They say parts exhibit almost no swelling or softening when exposed to extreme levels of moisture, and have good side-wall surface quality.

An RP system produced intricate sewing machine parts that range from 12.7 to 152 mm high with an accuracy of ±50 microns. Union Special Corp., Huntley, Ill., used a ModelMaker RP machine from Sanders Prototype Inc, Merrimack, N.H., to produce patterns for loopers. These are sewing-machine parts that work with needles to form an interlocking stitch. In doing so, the firm cut nine days from a developed cycle that once spanned 27.

Loopers have complex blends and compound angles that are easily defined in CAD software but difficult to show in a drawing or to a pattern maker. Such small parts are usually investment cast. Thus concept models fabricated on RP machines generally are scaled up. The RP machine chosen has a dimensional accuracy of ±24 microns with 2-micron rms surface finish.

“We were able to produce benchmarks for hard tooling while eliminating nine days, and that saved $3,750 in labor costs per master pattern over the previous process,” says Walt Taylor, project engineer with Union Special. The accompanying chart compares the time and cost of the old wire- EMD-and-machining process versus RP and investment casting.

© 2010 Penton Media, Inc.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.