What Engineers Should Know about Direct Digital Manufacturing

Given the significant technical barriers to direct digital manufacturing, additive fabrication systems still have a way to go before they become a common means for making production metal and plastic parts. Still, these additive processes are in use right now in a limited number of applications. Here's a list of factors that can help you determine whether you have a job that falls in direct digital's current sweet spot:

Small Parts Are Better. Rapid manufacturing machines currently have relatively small build envelopes, notes Greg Morris, cofounder of Morris Technologies. For example, the EOS direct metal laser sintering machines his company uses have a maximum working envelope of 10 x 10 x 7.5 inches. And while you can join rapid manufactured parts, secondary processes tend to offset some of the advantages of additive fabrication. So the larger number of whole parts you can squeeze on the build platform, the better.

Expect Finish Machining. The prototyping machines best suited for manufacturing tasks don't, at least on paper, have any difficulties getting hold of the dimensional tolerances found on workaday injection molding and casting jobs. The hardware suppliers all claim they typically can hold  +/- 0.002 inch/inch, so that smaller parts may tend to have somewhat better dimensional accuracy. In reality, though, you will likely have to plan for finish machining to achieve the critical tolerances found in many molded and cast parts. Morris' experience has been that parts coming off an additive machine still need finish machining, sometimes even on features whose tolerances fall within machines’ theoretical accuracy limits. "I have to make parts in the real world," says Morris. "And in the real world, additive parts require finish machining."

Machining will be even more likely when you consider that the surface finish on additive plastic parts doesn't stack up to what you can get from conventional plastic manufacturing methods. The plastics processes can't achieve anywhere near the best finishes offered by molded parts. There's no Class A finish off an additive machine. "At this point, we consider the layered manufacturing methods only when the part can't be seen," says Ron Hollis, president of Quickparts.

In the case of additive metal parts, the situation is a bit like what you would encounter with a cast part. Morris says his EOS parts typically have a surface finish of 125 RMS after they get a bit of shot peening. With some bench work he can take that down to about 63 RMS. With polishing, "we can get to any required finish," he says. In this regard, additive parts are like any traditional metal part that requires polishing.

How Many Parts Do You Need? With the exception of some high-volume hearing aid and dental implant parts made on SLA machines, most rapid manufacturing jobs start with low production volumes. "Pinning down an ideal range of volumes can be tricky," says Hollis. It's true that rapid manufacturing machines don't have the throughput of high-volume processes like injection molding or die-casting. But Hollis notes that rapid manufacturing machines eliminate the need to create, maintain and transport expensive tooling. Rapid manufacturing machines may trigger other cost savings too – including inventory reductions in the short term and over the product's lifecycle via on-demand spare parts. So at what point do rapid manufacturing advantages offset the productivity disadvantage? That's the tricky part, says Hollis. He and others who have used rapid manufacturing systems, though, tend to give a range that goes from a handful of parts up to about 1,000 parts/year. Larger jobs may be on the horizon, though. Morris is currently in the planning stages for a complex aerospace component that will have volumes in the 20,000-30,000 parts/year range.

Favor Complex Geometries and Parts Consolidation. The best jobs for rapid manufacturing systems are those with complex geometries that cannot be made at all or made cost effectively using traditional manufacturing methods. Morris, for instance, doesn't even turn his DMLS machine on unless the part has the right amount of complexity. "There's really very little advantage to putting a part that can be conventionally milled or turned on a DMLS machine," he says. Once you start talking about intricate machining operations with multiple set-ups, five-axis machining or lots of sink EDM work, then Morris may fire up one of his EOS machines. "DMLS is a good fit when you have something you just can't mill or cast, like very complex internal cavities," he says. With all that ability to produce complex geometry comes design freedom and the potential to do big parts consolidations. In this sense, complex geometry actually becomes a cost justification all its own.

Additive Parts Are Different. Just how different are additive parts? Part of the problem is that no one really knows, since these technologies are only now entering the manufacturing arena. According to Tim Gornet, director of operations at the University of Louisville's rapid prototyping center, there's very little data about long-term performance of additive parts – for example, how the plastic ones experience creep. And less common mechanical and physical properties are harder to come by for additive materials.

One thing that engineers should be aware of is that additive parts exhibit anisotropy. As Gornet explains, they have good, predictable tensile and flexural properties in the x-y axis, but tend to have some loss of those properties in the z axis. That's because the z axis is the one containing the interfaces between the individual layers of the part. Gornet says this z-direction property loss doesn't have to be a big deal, as long as engineers take it into account. "Design engineers will have to learn to specify a build orientation when working with additive fabrication," he says. And they'll have to design around any strength differences. "If the x-y axis is 20 times as strong as you need, the z-axis loss of properties may not be an issue in the first place," says Morris.

He also believes that the strength differences in different orientations are already starting to diminish, at least for metal-based processes that fully melt the build material. When he first started with the EOS machines, Morris noticed that they produced parts with a tensile property differential of roughly 15-20 percent between the x-y and z axes. "Today it's more like 5 percent and getting better all the time," he says.