If developing packaging for Hewlett-Packard's top-selling DeskJet printers is a high-pressure job, packaging engineer Kevin Howard doesn't show it. With more than a half-million DeskJet printers shipping out of three manufacturing sites worldwide each month, the stakes for the packaging team are incredibly high. If the rush to meet a launch date results in a package that causes damage, losses can top $1 million in a matter of weeks. Conversely, spending too much time on cushion design could jeopardize the launch date, leaving room for fast-acting competitors to grab marketshare. It's a tightrope that Howard seems comfortable walking. This is an engineer who dismisses cushion curves as erroneous: he routinely designs packaging foam pieces with half the recommended volume. Yet he attains shock pulses that aren't possible according to such curves. How does he get away with it? In a word, testing. By substituting real-world testing of actual molded prototype parts instead of designing from textbook cushion curves, Howard is able to carve away unneeded foam from the design. All without compromising product protection or missing a launch date. Ordinarily, creating prototype molds would be considered a luxury in time and expense. But Howard, along with a group of technicians at HP's Vancouver, WA, facility where DeskJet printers are designed and manufactured, developed a prototyping process that produces high-quality molded parts for final package testing in one to two weeks. In some cases, as quickly as four days. The most impressive piece of the prototyping puzzle, put in place a year and a half ago, consists of the ability to produce a prototype mold in-house using nothing other than a computer numerically controlled (CNC) milling machine and CAD software found in the product R&D lab of any large manufacturer. Here's what the prototype mold offers: * It eliminates the drawbacks of the traditional prototyping process, including hand-carved samples from foam billets. Usually carved by vendors at no charge, such samples don't test reliably, Howard charges, since consistent part geometry is difficult to replicate from part to part. * By using a prototype mold, HP can test a molded part before actually committing to a final production mold. Compared to hand-carved or machine-milled parts, molded parts have higher consistency in geometry and density, so test results will virtually match the characteristics of production parts. * Molded prototypes are cheaper to produce. HP requires at least 40 sets of end caps-too many for hand-carving-for its extensive drop testing. A hundred or more additional parts are needed to send prototype printers to various business partners. * Modifications to the prototype mold can be done in-house in a day or two, versus weeks with conventional molds. * Final package samples are ready before the first prototype build of the actual product they're designed to protect, months before a new product is released. Previously, molded samples of packaging weren't available until just weeks prior to launch, raising the spectre of a missed launch due to packaging snafus. "I can't claim that it actually speeds up our product introduction to the marketplace, but I know that we won't ever hang it up anymore," says Howard, who described HP's rapid prototyping process at the Transpack '96 conference in February, sponsored by the Institute of Packaging Professionals. Virtual package design The design process starts out on a computer. Once R&D completes the design of the printer on a CAD system, it turns the electronic CAD model over to the packaging department. "We take the computer image of the printer and 'push' it directly into what looks like a block of foam on the computer screen. That forms all the details of the cavity that we want," explains Howard. The computer mathematically reverses out the geometry of the printer to automatically create a solid-body model of the foam end cap. A physical prototype of the printer isn't even seen at this stage. "You don't even need it," Howard says. When the CAD model is completed, typically in just a few days, packaging personnel can compute the exact volume and weight of the part at a given density of foam. Such a computation not only allows HP to later verify the correct weight of molded parts, but it allows the company to get a fast-and more accurate-quote from a foam molder. That's because the molder can calculate the quote off the CAD drawing, and doesn't have to hand-carve a sample in order to weigh it. "We've gotten quotes back in as little as four hours," says Howard. The next step involves milling a preliminary prototype out of a block of foam using a CNC milling machine in the company's model shop. (HP had been using industry-standard foam billets, but has switched to molded foam blocks that eliminate inconsistencies inherent in billets, says Howard.) Technicians in the model shop first convert the CAD model into detailed milling instructions for the CNC machine. This step requires considerable programming expertise, emphasizes Howard. Once the programming is completed (usually less than a day), the actual piece can be cut in 30 minutes. Also, any part revisions requiring changes to the program can be done quickly, often in a matter of minutes. After a half-dozen test-and-revise sessions, the refined design is ready for the next stage, producing a prototype tool. Previously produced by HP's toolmaker via the traditional sandcasting method, the tool is now produced by a radical new method that HP developed itself. The method involves cutting different layers out of standard aluminum sheet and bolting the layers together to form the final tool (see sidebar). Why are molded parts necessary? Howard answers with two reasons: first, molded parts provide more realistic final test results than milled parts. That's attributable to the "skin" that forms on a molded part, which contributes to the resiliency of the cushion. The second reason is money. Howard explains: "When we CNC-mill parts, it costs us between $50 and $100 per set. When we use parts from a single-cavity prototype tool, it's around $3 a set. If I have to make up 200 sets and I save $97 per set, it more than pays for the tool." Tough questions Certainly the concept is clever. And the cost of prototype tooling seems justified when spread out over a hundred or more sets of end caps. But why go to the expense of developing in-house tooling, especially when it's double or triple the cost of traditional sandcasted tooling? "Our model shop is geared to turn things around quickly and concisely," Howard admits. "Cost is secondary." Plus, foam vendors often hand-carve prototypes for free. The expense is justified, Howard claims, on two counts: first, it allows HP to save on material costs. If HP were forced to rely only on hand samples for testing, it would have to design in an extra margin of foam to account for performance differences between hand-carved foam billets and final production parts. Multiply that extra foam by half a million printers a month, and it adds up. Second, rapid prototyping virtually guarantees that the packaging department won't miss launch dates. With new DeskJet models introduced twice a year or more, the timeliness of rapid prototyping justifies the cost. But how transferrable is this technology to other packaging applications? Will it be restricted to deep-pocketed high-tech firms like HP? Not necessarily, argues Howard. Although he concedes that HP's model shop is stocked with millions of dollars of high-tech gear, none of it was purchased with packaging in mind. It's equipment that any large manufacturer would need for product development; the only difference is that he's using the same resources for package development, which he says is not common. "There are an awful lot of companies using large quantities of molded foam that would benefit from this kind of technique," he says. He also notes that there could be applications beyond EPS foam molding. "In speaking with other suppliers in other industries, they're thinking they can use this kind of technique for tooling on thermoforms." Finally, isn't developing prototype tooling the job of a mold maker, not a packaging engineer? Not so, says Howard. "It does require a different mentality, the difference between asking are you hired to design a package, or are you being hired to get a product from point A to point B at the lowest cost possible with no damage? With that much larger definition of your job comes all sorts of things such as developing this type of rapid prototyping."
