RAPID PROTOTYPING AND MANUFACTURING
Rapid Prototyping
Key to Speedy Manufacturing
By Lea McLees
SPEED AND ACCURACY are the hallmarks of successful, profitable manufacturing in the 1990s. Products that miss target market dates or bear unanticipated design flaws can cost manufacturers dearly in lost sales and development investments, giving competitors an advantage.
And the window for meeting target market dates is shrinking, notes principal research scientist Tom Starr.
photo by Gary Meek
An ultraviolet laser cures a syrupy eposy liquid
into a solid prototype inside a stereolithography machine.
|
"Today, products can miss their markets if they are even six months late," says Starr, a researcher in Georgia Tech's School of Materials Science and Engineering.
Leon McGinnis, professor of industrial and systems engineering, agrees.
"Companies today can't afford to make mistakes when they are bringing products to market," he says.
Rapid prototyping and manufacturing (RPM) technologies hold vast potential for ensuring quickly designed, precise products. These methods allow prototypes -- and perhaps one day, the products themselves -- to be built quickly from computer-aided design (CAD) files using photochemically sensitive resins or other materials. Rapid prototypes are ready within hours or days of design; conventional prototypes can require weeks or months to build or mold.
RPM offers monetary savings, as well. Pratt and Whitney's Rapid Prototyping and Casting Laboratory in Connecticut reported in 1995 having made 2,000 RP castings at a cost of $600,000 to $700,000 -- a savings of $6.4 million, when compared to the $7 million that would have been spent using conventional prototyping methods.
But the availability of RPM technology doesn't guarantee its productive use. Education and experience are necessary, and, as with any technology, RPM could be enhanced with further research and development.
A Project for Students and Industry
To that end, Georgia Tech has formed the Rapid Prototyping and Manufacturing Institute (RPMI) to educate students, address industry RPM needs and shed light on future research directions.
photo by Gary Meek
Georgia Tech mechanical engineering graduate Marcial
Machado displays molds used to produce plastic, ceramic or metal parts.
|
Working together to make RPM potentials reality are a dozen educators and researchers in mechanical engineering, materials science and engineering, chemical engineering, industrial design, management and aerospace industrial and systems engineering; representatives of eight companies interested in RPM; and 14 graduate students, says Tom Graver, RPMI director of operations.
"Industry has a clear need to learn how to apply these technologies and how to develop more challenging applications," Graver says. "By primarily focusing on education, we can address the needs of industry while creating outstanding opportunities for our students and faculty."
Member companies include Coca-Cola, Durden Enterprises, Eastman Kodak, Lucent Technologies, Motorola, Siemens, Procter & Gamble and 3D Systems.
"The idea of having a relationship with the RPMI allows us to take advantage of and participate in research and projects to press the edge of the envelope that we normally would not be able to do ourselves," says Allen Brand of Motorola Energy Products.
RPMI began almost three years ago as part of a $1 million Technology Reinvestment Program grant from the Defense Advanced Research Projects Agency. Since then, member companies have each contributed $25,000 a year to support the educational mission of the RPMI, while researchers have brought in $500,000 in National Science Foundation monies, says McGinnis, one of the writers of the initial grant.
"Industry's immediate need is to be focused on understanding the technology and how to deploy it," McGinnis explains. "The interesting thing is that what comes out of our solving industry's technical problems is a research agenda. Educational activities that involve solving problems often lead to research."
RPMI researchers, member companies and students are working with three technologies, says Reginald Ponder, RPMI lab manager.
Read on to learn more about RPMI's achievements and goals for these technologies.
Rapid Prototyping for Ceramics
Combining rapid prototyping and powder injection molding technologies could enhance ceramics use in manufacturing by reducing the amount of time and money needed to use them.
One challenge remains: compensating for the shrinkage involved in powder processing.
photo by Gary Meek
Ph.D. student Beth Judson watches her rapidly made
mold fill with aluminum oxide powder.
|
Ceramic parts created with powder molding shrink 10 to 20 percent during firing as the powder particles merge together at high temperatures to form the dense ceramic. This change from the as-molded size can affect the accuracy of the final product. Principal research scientist Tom Starr and Ph.D. student Beth Judson are developing a model that would predict the dimensions of the final, fired part within 0.5 percent of the intended size. That would allow designers to create molds that compensate for shrinkage.
"We're using finite element analysis," Judson says. "Right now we're working with aluminum oxide. It's the workhorse of the ceramic industry -- most technical ceramic parts are made of it."
If shape is a controlling factor, finite element analysis would be done with each shape. If material flow is a controlling factor, equations could be developed for different types of materials and process conditions.
Industries that might benefit from this research include textiles, where ceramic thread guides are used; and areas such as aerospace, automotive manufacturing or soft drink production that use nozzles, air foils, rotors or other such parts for high-temperature or potentially corrosive applications.
The work Judson and Starr are doing might be especially useful in short-term manufacturing, Starr notes.
"For example, if you wanted 100 parts out of stainless steel for a military aircraft that isn't made anymore, do you invest in a metal mold for $30,000, or do you do it this way?" Starr noted. "You can't amortize that cost over just a few parts. And the process we're working with would work for metal, as well as ceramic, parts."
Georgia Tech is pursuing a patent on the shrinkage prediction model.
Estimating Prototype Build Time
Once a CAD design is sent to a stereolithography machine for building, the user knows the part will be completed -- but not how long it will take.
photo by Gary Meek
RPM has potential applications in preparing for
medical procedures and re-creating damaged bone structure, says lab
manager Reginald Ponder.
|
"It is difficult to estimate build time because it is dependent on so many variables, such as part size, layer thickness, laser power, resin and other factors," says Joel McClurkin, a graduate student who completed his master's degree in mechanical engineering in June.
Notes his adviser, Dr. David Rosen: "The [rapid prototyping] machine must perform thousands or millions of small operations to make a part. Until now, no one has added up the times for all these operations."
Enter the build-time estimator. Developed by McClurkin working with Rosen, the program estimates how long a 3D Systems SLA (stereolithography apparatus) will need to build a part by analyzing the CAD file of the prototype in question.
"The build time estimator reads the vector and range files and makes an estimate based on the information in those files," McClurkin explained. "The vector and range files contain the 'low level' information that actually controls the operation of the SLA. The biggest task here was cracking the vector file to determine the location of the information I was looking for."
The program (available at http://rpmi.marc.gatech.edu/BTE.html) needs only three pieces of information: the name of the resin being used, the resin's penetration depth or critical exposure; the power at which the laser will be operating; and the name of the range and vector files of the drawing for which the user wants a build time estimate.
McClurkin made dozens of comparisons of estimates with the true build time for a variety of parts and gathered feedback from industry users, finding an average error of 2 to 3 percent.
The build time estimator is part of a software package McClurkin completed and demonstrated for his master's thesis. The package will help SL users select from numerous possible SL build styles, taking into account build time, surface finish, accuracy, post processing time and other factors.
Measuring What You've Made
No manufactured part is a perfect match to the CAD file from which it was generated. Defects occur randomly or because of problems associated with the manufacturing process, says Tom Kurfess, an associate professor of mechanical engineering.
His specialty is measuring objects with a coordinate measurement machine (CMM) to determine how to replicate them, or how closely they resemble original CAD files, Kurfess says.
"Once the object is in the CMM, a trigger probe touches the surface of the object in a variety of locations in three dimensions, giving XYZ coordinates off its surface," Kurfess explains. "The data points are connected with curves, surfaces and solids to represent or measure the object."
Among the issues Kurfess and his students are studying are the best ways to take data points, connect them and fit them to complex surfaces; ensuring a good data fit; and identifying the types of deformation that result during RPM, and their causes. If characteristic types of deformation are found to be associated with certain types of RPM, they might be eliminated in the future by modifying the CAD file involved.
"We have a lot of support from industry and government to address these issues," Kurfess said. "Our work extends beyond the rapid prototyping issues. Our target is extremely complicated geometry -- and we know exactly how to look at it."
Small, Speedy Product Runs
RPM is extremely useful for creating tooling used in short manufacturing runs, says Dr. David Rosen, assistant professor of mechanical engineering.
|
Making tooling rapidly: Harnessing emerging rapid tooling
methods to develop processes that quickly, inexpensively produce five to
500 parts in end-use materials.
"We're looking for good ways to develop tools to produce a small quantity of plastic parts in end-use material, using manufacturing equipment like injection molds," says Rosen, who also serves as RPMI academic director.
Currently, injection molds can take several weeks or months to produce. Rosen's goal is to go from CAD model to molded parts in four days -- he thinks that will be possible within the next three years.
One aspect of this work involves fine-tuning the production of injection molds using SL machines, and speeding up that process. An approach that Rosen and his students are looking at is building shells and filling them with epoxy or metal.
"Rather than building a solid mold on the SLA machine, which takes about 35 hours, we build the shell of the mold in about 15 hours and at one-tenth the cost," Rosen says.
Beyond Rapid Prototyping
Prototype development and use is part of a larger, holistic picture being explored by Rosen and Dr. Farrokh Mistree, professor of mechanical engineering. They are exploring the application of the best type of prototype -- physical or computer -- in different situations.
"Depending on the information we want to gain from the model, we don't necessarily need a product- representative prototype," Rosen says. "In some cases, we may not need physical prototypes -- virtual, computerized prototypes tend to be even more rapid."
They also are looking beyond product manufacturing to other needs that should be considered during design -- service and disassembly.
"At the end of a product's life there are a couple of different strategies," explains Mistree, who also works in Georgia Tech's Systems Realization Lab.
"One is to throw whatever you have into the waste basket. Another approach is to disassemble it and take out valuable components. Or, you could recycle the materials. How do you assess product assembly and disassembly characteristics? One way is to build a physical prototype and study it."
RPM will greatly contribute to development of principles for disassembly, a companion to the principles for assembly developed 15 years ago, he notes.
"Because we have rapid prototyping and virtual prototyping abilities, we should be able to come up with principles for disassembly much more quickly," he says.
Future RPM Research at Tech
What does the future hold for rapid prototyping and manufacturing technology? It could evolve into rapid manufacturing, predicts Steven Danyluk, director of Georgia Tech's Manufacturing Research Center (MARC). RPMI is one of the main focuses in MARC's work toward enhancing manufacturing education and research.
"If you needed to make a specific honeycomb product, you could make it out of a polymer, and further fabricate and use it in a production environment," he says.
Further materials research could make that prediction reality, says Bob Schwerzel, principal research scientist in the Georgia Tech Research Institute's Electro-Optics, Environment and Materials Laboratory.
"We'd like to develop new polymers and resins with new properties that would allow RPM to be used as a general manufacturing tool, and not just as prototyping tool," he notes. "I think controlling the material properties of the resin in the biggest challenge."
And eventually, RPM may contribute to the increased use of flexible systems in manufacturing that adapt quickly to changing market needs, Rosen notes.
"You really need your manufacturing equipment and tools to be modular and adaptable, and that's quite difficult to do -- so it remains to be seen how well rapid prototyping technologies can help us do mass customization and product variety," he says.
Further information on RPMI is available from Tom Graver, Rapid
Prototyping and Manufacturing Institute, Georgia Institute of
Technology, Atlanta, GA 30332-0406. (Telephone: 404/894-5676)
Check out RPMI's web site at: http://rpmi.marc.gatech.edu/.
Send questions and comments regarding these pages to Webmaster@gtri.gatech.edu
Last updated: Aug. 5, 1997