20 Years of Rapid Prototyping

Last December marked a significant anniversary; 20 years ago, 3D Systems shipped the first rapid prototyping (RP) system, the SLA 1. Systems were sent to six beta sites including Baxter Healthcare (Deerfield, IL), General Motors (Detroit, MI), Kodak (Rochester, NY), two sites of Pratt & Whitney (East Hartford, CT) and AMP Electronics (now part of Tyco) (Harrisburg, PA). For a few months prior, employees of each of the beta sites had spent several weeks training at 3D Systems, then located in Simi Valley, CA. The systems were hardly the plug-and-play type we see today, and there were major problems to overcome before the system could be used effectively:

• It was still early in the development of the SLA and changes to the software and hardware were being made daily
• A new data format, the STL file, was created to drive the system and no translators existed to create the format from CAD data. Even more of an issue, most CAD systems of the day were either 2D or wireframe systems, both lacking the information necessary to create a complete STL file. Aries, an early solid modeling system was the first to offer a commercial translator. Some of the beta users, including Baxter, found it necessary to write their own STL translators
• The system was crude by today’s standards; no re-coater blade, no restart capability and it was very slow
• The one resin available had relatively high levels of shrink and shrink-related distortion was a continual and frustrating problem
• The resin was so brittle to be of little or no use for functional tests

However, the system did something that was previously impossible—create a model of a design in hours rather than days or weeks.

Baxter was the first system to be installed. In early January 1988, application engineers for 3D Systems arrived in 20 below weather to install the system. After a few days of assembly, calibration and tweaking, the system was up and running and Baxter began supplying stereolithography models to internal customers.

3D Systems’ PR firm did a good job of getting the word out and created a flood of interest. CNN taped a news segment at Baxter, which aired several times, both on CNN and on local Chicago stations. Not only did technical magazines run articles on the new technology, but the mainstream magazine Popular Science also ran an article on the new device with several pictures. As a result, Baxter received requests for tours of the RP lab from individuals interested in starting service businesses, and within six months, at least six service providers had been formed. By the end of 1988, 3D had released production versions of the system and shortly thereafter replaced the SLA 1 with the SLA 250.

A new industry had begun.

Now, a short 20 years later, we have come a long way. Consider:

• More than 22,000 RP systems have been installed worldwide, with more than 9,000 in North Americai
• More than 40 companies manufacture such systems, four of whom have sold more than 2,000 systemsii
• Most medium to large manufacturers own at least one system and many own several
• More than 400 companies provide RP services, with more than 200 located in the U.S.iii

Has There Been an Impact?

While these numbers are very small compared to the sales of popular consumer products such as the iPod or Wii, in the world of relatively expensive industrial capital equipment, they are huge. It would appear that RP has been tremendously successful. However, has the development of RP really had an impact on product development? Do we create new products any differently than we did before these technologies were available? Is there any evidence to show that RP is anything more than an interesting engineering toy?

I understand such questions may seem silly in light of the large numbers of systems sold. However, consider another sales success, the George Foreman Grill. Many, if not most of us, have one in our own homes. It is the most successful appliance in history, selling more than 100 million units and earning Salton, its manufacturer, hundreds of millions of dollars. George Foreman himself has earned more than $150 million from sales of the grill, more than he earned through his entire boxing career. Without question, this has been a successful product and it is in fact a useful cooking device. But has it really had an impact on the way we cook food? Of the hundreds of millions of meals cooked every day in the United States alone, I suspect that the percentage cooked on a George Foreman grill is at best in the thousandths of a percent. We would have to conclude that in spite of its success as a product, it has not had a significant impact on cooking in general.

Are RP systems more than a very expensive George Foreman grill? To find out, I conducted a very unscientific survey. I called several friends and acquaintances, all of whom are intimately involved in RP at major corporations and asked them: “In your opinion, what percentage of your company’s new component designs is built using a rapid prototyping process before they are released for manufacture?”

I didn’t ask for data to support their answers nor did I ask how many parts were designed by their companies each year (although I knew that in each case it was hundreds, if not thousands of parts). Furthermore, I knew the person interviewed was in as good a position as anyone in their company to estimate the answer. The results surprised me (see Figure 1, page 16).

The approximate average (approximate because we don’t know the relative number of parts designed by each company) is about 91 percent. This is an extraordinarily high level of acceptance by industry.

To put this in perspective, if the George Foreman grill had the same level of acceptance, 91 percent of your meals would be cooked on one. I would be hard pressed to find any technology other than CAD that has anything close to that level of acceptance in product development. Consequently, we have to conclude that RP has had a significant impact on the product development process.

Speed of Adoption

While this level of acceptance is surprising, it is even more surprising that it was accomplished in a relatively short period of time. While 20 years may not seem short in the sense of technology adoption, it is, in fact, extraordinarily fast. Many of the most significant technological innovations, including the light bulb, the automobile and the xerography process, took more than 25 years to find widespread use and even then were nowhere near the level of acceptance of RP.iv

Consider other technologies which have had a significant impact on manufacturing and my rough estimate of their penetration after 20 years of exposure (see Figure 2, page 16).

Clearly, RP has been adopted much more rapidly than other technologies. Why would this be so? The inescapable conclusion is that RP provides substantial value in the product development process, more than was provided by these other technologies.

What is the Value?

What value does RP provide in product development to drive adoption so quickly? While we don’t even have an unscientific survey to give us more insight on this, we can make some educated speculations based on sales evidence.

Perhaps the greatest value is the simplest—the ability to provide a reasonably accurate physical model of a design quickly and at low cost. Such models are typically referred to as concept models and a number of low cost systems, often referred to as 3D printers, have been developed to build them quickly and at very low cost. Two manufacturers announced last year that they would be introducing 3D printers that would sell for less than $10,000. That part of the market is by far the fastest growing segment with a unit sales growth rate exceeding 21 percent in 2005. In fact, 3D printers accounted for more than 72 percent of total RP unit sales in 2006.v

It is not hard to understand the attraction of a concept model. The traditional method of documenting a design is an engineering drawing. Most non-technical people have a great deal of difficulty visualizing a part from looking at the drawing. Even engineers and toolmakers that look at drawings every day may require several minutes of studying an engineering drawing before they can create that geometry in their mind and really understand it. A model part, however, instantly and unambiguously communicates the design to everyone regardless of their ability to read drawings.

The availability of quick and inexpensive models allows us to use more efficient design processes. For example, a design engineer can go into a design review meeting with models of five alternative designs for a component. The models enable everyone involved in the selection process to quickly grasp the details of each alternative and discuss the relative advantages and disadvantages of each, quickly coming to a conclusion about the best design. Prior to the advent of RP, it was generally too expensive to create a model of a single design, let alone one for each of several alternative designs. Design decisions were often made without a complete understanding of design alternatives.

Concept models also allow preliminary fit checks, enable tooling vendors to quickly identify design features, which can make tooling more expensive, and provide physical models for package design and manufacturing automation. Without a doubt, the availability of fast, inexpensive concept models has provided tremendous value in the development of new component designs.

RP systems have also tremendously lowered the cost and time required to obtain functional prototypes for testing. For example, RP materials are now available which simulate a wide range of injection molded plastics. As a result, it is possible to build functional prototypes that can reasonably simulate a production part for testing purposes. While RP materials are not the equivalent of injection-molded plastics, in many cases they are close enough that some verification testing can be done and conclusions about the production design can be drawn without the need for molding prototype parts in the production material.

RP systems can also be used to create functional metal components. Using RP systems to create patterns for investment, sand or plaster mold casting is now commonplace. Newer systems are allowing metal parts to be directly created on the RP system. Prior to the development of RP systems, it was impossible to quickly and cost effectively create functional parts for testing. The ability to do so has allowed us to identify and correct design deficiencies long before the product is released, improving the quality of products we develop.

Is the Market Saturated?

If more than 90 percent of new component designs are already prototyped using an RP process, are we approaching a market saturation situation? Is there really much of a need for additional systems?

It is tempting to conclude that the market should be slowing considerably. However, sales data does not support such a conclusion. If anything, the market has grown considerably in the last couple years. Sales of 3D printers continue to accelerate and even the sales of mainstream RP systems such as SL and LS systems is growing.

While it may be that the number of new components designed every year is increasing, it is more likely that the average number of additive fabrication parts per design is increasing. We strongly suspect that concept models are being built for several alternative designs before the final design is selected, and that multiple prototypes are being built in functional materials for testing prior to building tooling. Furthermore, I expect that trend to continue for several reasons:

• The price of concept modeler systems continues to fall, reducing the price of concept models. The cost of making additional concept models will be insignificant compared to the total cost of product development, making it easier to evaluate a greater number of alternative designs.
• As materials continue to improve and more closely simulate injection molded plastics, the number of functional prototypes built will increase significantly, allowing more and more design issues to be resolved prior to the build of tooling.
• Also as materials improve, additive fabrication systems will be used to build initial production parts, allowing companies to get their product to market while waiting for production tooling to be built.
• As materials reach production quality, the market for very low volume production will grow rapidly, greatly increasing the average number of parts per design built.

I think the next 20 years could be every bit as interesting as the last. 

Tom Mueller is partner and co-founder of Express Pattern (Vernon Hills, IL), a supplier of RP services and a provider of RP patterns for investment casting. Tom’s involvement with RP began in the late 1980s at Baxter Healthcare where he headed the first beta test for 3D Systems. After two years there, he and a partner formed Prototype Express, one of the earliest service bureaus.