Stanford University

News Service



David F. Salisbury, News Service (650) 725-1944; e-mail:

Industry reps, academics swap theories on rapid design, assembly of prototypes

Suppose, as sometimes happens in real life, your 6-year-old daughter has just ruined her favorite Barbie doll's face with nail-polish remover. Though you've gamely tried repainting the lips, eyes and eyebrows, somehow they don't look quite right. Ah, if only you could build your own Barbie from scratch, right here at home.

But building a model is tedious, time-consuming, error prone and always harder than it looks. In the world of manufacturing, where competition is stiff and time is money, people would like to make model-building a lot easier, too. The models they build are typically prototypes of a component or a tool, to be copied a million times in a mass-manufacturing operation.

At a two-day workshop held on Stanford's campus in early May, industry representatives met with Stanford engineers and their graduate students to talk about new, rapid ways of designing and assembling physical prototypes. The underlying question: Can we now produce one-of-a-kind models that do more than serve as simple, show-and-tell displays, but that actually harbor some of the final product's physical characteristics ­ smooth surfaces, resistance to heat or wear, mechanical strength ­ so that they can actually perform like the component or tool in question?

Participants discussed several methods of rapid prototype generation, examined models produced by various technologies and speculated on their possible applications well beyond the research lab or factory floor.

Speakers even suggested that within five years a consumer version of rapid prototyping might help you solve the Barbie problem. It may be possible to download electronic directions from Mattel over the Internet and, using a special "3-D printer" attached to your home computer, construct a new plastic replica of the damaged Barbie's head.

The workshop was organized by Stanford mechanical engineering Professors Friedrich B. "Fritz" Prinz and Mark R. Cutkosky and sponsored by the Alliance for Innovative Manufacturing at Stanford (AIMS). Formerly known as the Stanford Integrated Manufacturing Association, AIMS is a campus-based joint venture initiated by Stanford's Graduate School of Business and School of Engineering and several large corporate partners to promote the exchange of technical ideas between academia and industry.

Layered manufacturing

The workshop's primary focus was on the potential of a technique called layered manufacturing. In principle, any three-dimensional shape, no matter how complex, can be produced by decomposing its design into thin cross-sections, then assembling the piece by producing those cross- sectional layers and piling them one on the other.

In practice, Prinz said, this can be achieved by any of a number of methods, each with its advantages and drawbacks. For instance, a very fine wire of metal or plastic can be extruded, heated to the melting point, and deposited according to a preset pattern over several iterations. An alternative method, which employs the same principle behind the ink-jet printers attached to many home computers, builds objects by targeted spraying of molten plastic layer by layer. Voids are filled with water-soluble wax to provide temporary support. After the full three-dimensional shape is produced, the supporting wax is removed by dissolving it with hot water.

The practice of composing 3-D models in layers took root about a decade ago and since then has grown tenfold to more than $500 million in annual sales, Prinz told participants, who represented manufacturing sectors ranging from automobiles to primary metals, semiconductors and aircraft manufacturing. Recently, however, sales have been leveling off, he added, because these prototypes typically lack the physical properties of the true product, so they can't be used for engineering tests. As a result, the prototypes' use is limited to show and tell.

What manufacturers need, Prinz said, is a methodology that is fast and extremely precise, yet yields strong, smooth-surfaced, accurately shaped prototypes that can actually be tested and put to work as parts or tools. "Our research at Stanford," he told the audience, "is geared at rapid prototyping techniques that allow for tremendous shape complexity but do not sacrifice engineering quality."

Shape deposition manufacturing favored

The approach favored by Prinz's group is called shape deposition manufacturing, or SDM. It alternates steps that add material and those that cut away the excesses, said Prinz, the Rodney H. Adams Professor in the School of Engineering.

The first step is to figure out where to cut cross­sections through a part's or tool's design. In simple thin-sectioning, all the layers are the same thickness. In SDM, however, the thickness of each layer varies, and is carefully selected in order to reproduce the shape with the fewest number of layers possible. After the thickness and shape of all the layers are determined, the object is built up by the successive deposition of layers of molten prototype material ­ usually a wax that is not water soluble ­ supported by a matrix of space-filling, water-soluble wax. After each layer is added, it is allowed to cool and then it is machined to achieve the precise geometry required. The process is repeated for each layer. When all the layers have been deposited and machined, the supporting matrix is dissolved away by immersion in water.

Using SDM, Prinz said, "you can make any shape you want" in principle, no matter how complex. So it can reproduce extremely intricate engineering parts. And because it uses thicker slices than existing alternatives ­ and therefore requires fewer of them ­ SDM is faster than other methods, he noted.

One of the current limitations of SDM, he said, is that it cannot yet produce quality engineering artifacts. Another problem, shared with other layered manufacturing techniques, is that the deposition of each new layer can change the properties of layers already deposited; this is especially true if the materials deposited are metals, whose crystal structure is altered by heat or exposure to other substances.

For layered manufacturing to work well with metals, suitable alloys must be found. Several Stanford graduate students working under Prinz and Cutkosky discussed their research on alloys and the fine-tuning of deposition techniques to solve problems of shrinkage, deformation, lack of strength and surface roughness. Ceramics are another promising material type.

Cutkosky, who is the Charles M. Pigott Professor of Mechanical Engineering, walked participants through technical details of the SDM design process, whereby the part or tool to be built is efficiently "decomposed" via various mathematical methods into the smallest number of layers that can accommodate its particular geometry. The ability to make extremely complex 3-D structures as well as to vary material composition gives a huge range of possible design characteristics, Cutkosky said. Moreover, the process is flexible enough so that subcomponents such as sensors ­ or even moving pieces such as pistons ­ can be embedded in the prototype at various stages of the deposition cycle.

As you gain the capacity to embed intelligence in parts, you begin to produce parts that keep track of their own fatigue history (what stresses they've undergone, whether they're about to fail, etc.), said Prinz.

Layered manufacturing is ideal for making "mesoscale" items whose dimensions are measured in the tens to hundreds of microns. Electronic devices in this size range bestow portability on such gadgets as computers, telephones, radios, watches, medical devices, and various sensors. Beyond that, small motors generate more power per unit of weight than large ones do. In theory, large arrays of tiny jet engines, made with shape-deposition techniques, could replace the giant solo engines currently mounted on airplane wings. Designers could build in enough redundancy ­ say, by using 500 of these small engines when only 450 are actually required ­ so that the plane still will fly even if 10 percent of the microengines fail, an outcome far preferable to the catastrophe arising from the failure of a single large engine.

Using shape-deposition techniques, the Prinz lab has created a "mesicopter," a tiny flying machine made of ceramic components and weighing a mere 1.7 grams. In tests, the mesicopter has proven capable of lifting itself into the air, although it is not yet aerodynamically stable enough to fly without external support. The mesicopter is actually an array of four propellers -- each sporting fully shaped, 80-micron-thick blades -- occupying the four corners of a square plastic connecting frame and powered by a tiny commercial motor. The motor was acquired from a European company, said Rudy Leitgib, a Stanford graduate student working under Prinz's direction; but in the next two or three years Leitgib and his colleagues hope to have produced a superior, smaller micromotor of their own design.


Rapid prototyping in automotive industry

Participant Dawn White, a technical staff specialist in the Manufacturing Systems Department of the Ford Research Laboratory (an arm of Ford Motor Co.), said the era of fast prototyping already has arrived in the automotive industry, where, she said, even "experimental" processes must be able to support volumes of 50,000 units per year or more. In this environment, she said, "pennies matter." Piggybacking on this, she said, is unrelenting pressure to reduce the product-cycle time.

"Traditionally in the automotive industry, when a new model is designed, it's first built as a hand-built prototype for show and tell," White said. Conventional means take three or four months. But the new methods reduce this to five or six weeks. One that Ford is now employing, metal spray forming, operates more or less according to the ink-jet principle, she said.

"We've been working with this technology for about five years. We can get up to 200,000 parts out of a tool produced this way. This may be enough for an entire product run of a niche product, like a new Thunderbird," White said. "Spray forming appears to be cheaper than conventional techniques."

Another locale with an obvious affinity for rapid prototyping would be a military aircraft carrier, which might spend six months at a stretch out on the high seas. Spare arts could simply be produced aboard as needed.

But the real high-impact payoff of layered manufacturing may be much closer to home ­ in fact, right in the home, in the form of a 3-D printer that could quickly produce anything from a Star Wars toy to a replacement bracket for the one you just broke at 2 a.m. As far-fetched as this scenario may sound, White suggested that this capability might not be so far off.

"Someday everybody will be able to print 3-D models at their desktop," she said.

Here's how it might work. Imagine the printer connected to your home computer used not only four colors of ink but, say, four colors of polyurethane (or some other kind of plastic) and a water-soluble wax to serve as a temporary space-filling support. You could go online, connect with the website for Hasbro or your friendly neighborhood hardware store, and for a small fee download a set of instructions for "printing" the necessity du jour. You turn on your printer, press the "3-D print" option and get out your nose plugs to fend off the olfactory assault as the "Desktop Factory" fires up. The plastic and wax are extruded in a skinny, fast­drying liquid jet, layer by thin layer, until a complex shape is built up. Stick it in a tub of water for a while to remove the wax, and there's your geegaw, punctual and pristine.

White's musings were echoed by Paul Fussel, a senior technical specialist in product design and development at Alcoa Technology. Fussel suggested that the prospect of having your own 3­D printer, capable of creating complex artifacts, is perhaps as little as five years off. "You'd have to get the price of one of these down below $500 to ignite a mass market," he said.

Quite likely these 3-D printers won't appear initially in the home, but will first show up in a neighborhood copy shop or in the back room of the retail supplier itself. Said White: "Maybe this is where a Kinko's would come in. Or maybe it'll just be a good way for Toys 'R' Us to keep their inventory down."


Other relevant sites:

Stanford Rapid Prototyping Laboratory


Stanford Dextrous Manipulation Laboratory



By David F. Salisbury

© Stanford University. All Rights Reserved. Stanford, CA 94305. (650) 723-2300. Terms of Use  |  Copyright Complaints