The JTEC/WTEC panel visited 14 sites in Japan that have RP tooling applications:
Professor Kimura of the University of Tokyo's Department of Precision Machinery Engineering noted in discussions with panelists that Japan has a long tradition of process improvement, or "kaisan." The Japanese are experts at incrementally improving manufacturing methods and processes to ultimate perfection. Professor Kimura told panelists that given time, an evolution of kaisan similar to what has taken place in manufacturing and production will certainly take place in Japan in rapid prototyping.
When panelists spoke with Professor Nakagawa of the Institute of Industrial Science at the University of Tokyo, he indicated his belief that the future of rapid tooling lies in the area of high-speed machining. There is extensive work in high-speed milling in Japan, and the Japanese are experts at its use and application. Professor Nakagawa's appraisal of the primary benefit of using rapid prototyping is that it can produce rapid design changes, especially for the electronics industry, which has a lot of different kinds of parts. Professor Nakagawa indicated that he is working on a new rapid prototyping process, which he hoped to be able to present in 1996. Professor Nakagawa provided a photo of a plastic injection mold tool (Fig. 11.9), that was fabricated by a DTM selective laser sintering machine for a project of the Ministry of International Trade and Industry (MITI).
Professor Nakagawa also provided a diagram showing how RP models are used in Japan to make molds and forming tools. See Fig. 11.10.
Professor Fukuda of the Tokyo Metropolitan Institute of Technology is developing a CAD system that is more intuitive to users -- one that is able to integrate changes and modifications so that the system can be worked with rapid prototyping and then integrated into virtual reality (VR). This could be very positive in the area of tooling.
Although this intuitive CAD approach is not an application that is available for rapid prototyping today, it has exciting possibilities. In the present system of producing a finished part, product engineers design the part, but they do not necessarily understand the process of producing the finished part. So when they finish a design and give it to the manufacturer, the manufacturing engineers must determine how to manufacture the part with considerations for expansion factors, holding targets, sequence of operations, and a number of other factors that the part designers are not concerned with. Traditionally, the product design goes from the design sector to the manufacturing sector, where the product is then processed and delivered.
With rapid prototyping and rapid tooling, the process variables noted above must still be accommodated, but can now be accommodated much earlier in the product development cycle. The research that Professor Fukuda is performing can be instrumental in capturing all of the process planning and bringing it up front into the CAD model in an intelligent manner, so that it is not necessary to consult with various process experts to determine how best to make the tool or part. Professor Fukuda's project is highly relevant to tooling, because intricate planning for process variables is a reality that is not always recognized and appreciated, but nevertheless must be accommodated.
D-MEC is manufacturing glass-filled acrylate resins for plastic injection molds. The mold used was able to process about 180 ABS parts before heat distortion of the mold became unacceptable. D-MEC representatives showed the JTEC/WTEC team a lot of examples of using silicon rubber molds, spin-casting, and vulcanization, as well as a different process using a stereolithography master pattern and making a mold around it to make multiple copies. Several Japanese companies are already doing this.
Current applications at Denken are directed to the modeling environment. Engineering and tooling applications are a future goal. It appears that in engineering applications, rapid prototyping is primarily being focused on design verification. The Denken machine is a relatively low-cost, small-build-volume machine. The Ministry of Education in Japan subsidizes academic purchases for this equipment by up to 50%. That can be a terrific incentive for universities to get students to have a greater appreciation for all of the up-front RP processing variables that must be considered for manufacturing.
Teijin Seiki is developing plastic injection molds for prototype applications using a 50-70% inorganic filled resin. The resin requires thermal post-curing.
Teijin Seiki is also using a vacuum casting application for a portable phone housing (Fig. 11.11).
This tool required 40 hours of CAD design time, 14 hours on a Teijin Seiki Soliform machine, and 1 hour cleaning and post-curing, leading to a urethane casting in a lot size of more than 20 parts. In other RP work that is being accomplished in Japan, just preparing a CAD model requires half the time of making it, as in this example. Well over 50% of the production time was dedicated to making a "squeaky-clean" computer model, because without it, the machine cannot produce the desired finished parts.
Kira machine model KSC-50 was initially designed for design verification, not to be used as a tooling machine. However, Kira is finding that some of its customers are discovering creative and innovative ways to use the paper models as tools. Although accuracy is critical to tooling, several customers have successfully used the paper patterns as masters for silicone and epoxy molds, some of which are shown in Fig. 11.12. The Kira KSC-50 machine and its associated dimensions and specifications are provided in Fig. 11.13.
Tokuda Industries, a prototype shop that fabricates models, forming tools, jigs, models for wind tunnel tests, and prototype products, was the first company to acquire the Kira KSC-50 machine. (It also has a CMET stereolithography machine.) Company representatives indicated to the JTEC/WTEC panel that for Tokuda, among the most appealing issues associated with rapid prototyping are its relatively low cost and low risk. The company provides early RP models of parts and or tools, gives them to customers with a quote for cost for the "machined part," and gets good value for that work. Tokuda's managers are very happy with this capability.
Tokuda representatives cited an example, also presented as a paper at a recent Japanese conference, of a sheet-metal die that measures 250 mm x 150 mm x 93 mm (Fig. 11.14). They compared the Kira RP machine capability to conventional numerical control (NC) machining in terms of (1) cost, (2) production time, (3) accuracy, and (4) cutting characteristics. The Tokuda representatives were favorably impressed with the lower cost and low risk of RP but disappointed in the time and dimensional accuracy of RP in comparison to CNC. They pointed to the need to correct for the staircasing effect of the stacked paper layers by coating the outside of the tool with a filler material, followed by manually sanding it smooth. Fig. 11.15 shows these steps.
Fig. 11.14. Model of a sheet metal die from Tokuda Industries.
Fig. 11.15. Tokuda's steps to improve staircasing surface finish.
Fig. 11.16 compares the Laminated Object Manufacturing (LOM) with CNC machining in terms of cost and time. The cost of using the Kira process in comparison to CNC machining is less; however, the Kira RP process is slower.
The JTEC/WTEC panel found many examples where the Japanese are focused on dimensional accuracy and consequently rely on CNC machining over RP approaches. The expected accuracy of the Kira process was noted to be within ±0.1 mm, which is not acceptable for tooling applications. Further, the dimensional accuracy in the stacked layers or z dimension is subject to swelling from humidity, leaving RP with much need for improvement.
At INCS, a very successful service bureau that is doing very well with its 3D Systems stereolithography apparatus (SLA), a variety of applications are underway (see also Fig. 11.17): rapid tooling via QuickCast (7%); investment casting (15%); vacuum casting silicone and epoxy molds (22%); and design verification (56%). Company representatives indicated they would like to do more in the area of tooling, but they are required to create the computer modeling behind the scenes for a lot of their customers. About 80% of the automotive and electronics companies in Japan use CAD, but they present the CAD model in a 2D form from drawings, not in a 3D model; therefore, essentially the whole modeling effort has to be redone before it can be used for rapid prototyping.
INCS has demonstrated the ability to create rapid metal tooling, as in the brass electronic connector shown in Fig. 11.18. Fabricating metal tooling for an electronic connector would normally take about 1.5 months by conventional means, because of the time required to fabricate the pinholes of the connector and the individual pins. A process was used to fabricate multiple insert pins as one assembly by plaster casting a stereolithography QuickCast master pattern.
In the method used by INCS (see also the schematic shown in Fig. 11.19), the original 3D model data of an individual insert pin is combined into one assembly of multiple pins using 3D CAD. Then a positive SLA model of the entire assembly is built, which is used as a master pattern to make a negative silicone rubber mold. Another type of silicone rubber is poured into the previously prepared negative rubber mold to form a silicone rubber part of the same positive geometry as the master SLA pattern. A plaster mold is then created around this silicone pattern. The silicone pattern is removed and metal is cast into the cavity. The resulting part has the same geometry as the original SLA pattern. The choice of metal to be ultimately used for fabricating the insert pins depends on the temperature tolerance of the plaster and the required metal tooling strength. For the electrical connector, a high grade of brass was chosen. The desired metal pins are then finally assembled by integrating them into the previously formed connector in one-third the time required by conventional methods.
The JTEC/WTEC panel encountered the RP cultural acceptance issue again at Omron. The Omron representatives expressed their belief that in five years rapid machining will dominate, while rapid prototyping will incrementally advance and continue to be restricted to design use.
Many camera parts that Olympus makes are very complicated. The company is creating RP models and finishing them to make them look "real." Looking at one of Olympus' RP camera housings or video cameras, it is hard to believe, except when you look on the back side, that it is an RP model, because it is all sanded, painted, and looks very authentic. Panelists did not see any rapid prototype tooling applications per se, but certainly found many examples of functional testing -- up to 80% of its applications are tested using stereolithography models. Olympus is eliminating a lot of tooling rework by going this way. That is the flip side of tooling applications: if tooling rework can be avoided, much time, energy, and cost will be saved. Table 11.1 compares resources used to make a camera case by conventional and RP techniques.
RP applications at Hino Motors are predominantly in design verification and interference checking. Hino Motors also uses prototypes in engine research and development on the intake side of its engines for short-run functional testing. The company is also finding that it can check design geometry for items such as an instrument panel air vent, in which example there was a significant time savings: checking the vent in the RP sample took 4 hours compared to the usual 2 weeks that would have been required if testing a manually constructed wooden model of the air vent.
Shonan also uses RP primarily for design verification. Its designers have coined the term, "CAM-less" for its injection molds. That is really synonymous with what is going on in Germany, where designers are making the cavity and core of the mold inserts directly, using a computer model and RP. Shonan uses some of the filled resins from Teijin Seiki to produce mold inserts directly using a stereolithography process; however, because the resins have low thermal conductivity, Shonan representatives indicated that they are realizing a 1-2 minute cycle time for part production that used to take seconds. Consequently, they confirmed that research is underway to add conductive fillers in the resin for the mold inserts. There was no direct metal mold activity at Shonan Design Company at the time of the JTEC/WTEC visit.
Besides doing functional testing, JAE makes use of rapid prototyping to visualize several of its connector parts. By iterating and using special procedures, such as beam width compensation, thin slice layers, and shrinkage compensation, JAE is finding ways to get the best accuracy from its RP equipment. JAE is able to attain a dimensional accuracy of ±0.1 mm. The company is realizing a very significant advantage in time and cost savings. Its next step is to model some of its mold core pins to speed mold development.
The JTEC/WTEC panelists were unable, because of time constraints and a very structured agenda, to discuss tooling applications with Toyota representatives; however, the panel is aware that there is some work ongoing at Toyota. Toyota representatives did confirm a couple of RP issues that are of concern to them, most of which are identified elsewhere in this study. In addition, Toyota is pursuing studies of birefringence for photoelasticity and development of transparent parts, so designers can look through them to observe colored gasses for evaluation purposes.