Post-processing of SLA parts and molds remains a time-consuming and somewhat problematic necessity. Typically, SLA parts and molds must be post-cured in a UV “oven” for 1 - 1.5 hours. After that, sanding, filing, and polishing are often necessary to produce nice surfaces. Research performed by Bryan Blair, led by Dr. Jon Colton, studied both the post-cure and polishing of SLA parts and molds. Bryan graduated with his MSME degree in Spring 1998.

Current SLA technology is limited in surface roughness and finish by the thickness of each layer, which is approximately 127 microns (0.005 inches). Smooth surfaces of SLA parts and molds are critical to producing parts in injection molding. An average roughness of 1.27 microns (50 micro-inches) is required. The best surfaces that Bryan achieved in his work had roughnesses consistently under 20-microinches. For polishing with an abrasive, results indicate the use of 500 -1200 grit. For polishing by pumping abrasive pastes, a viscoelastic media such as borax/glue mixtures mixed with 30-micron abrasive. A pressure of 600 psi should be used. This work indicates that an automated polishing device could be made that would quickly finish SLA parts and molds.

The development of a plastic part frequently involves several prototype iterations. Production of these prototypes with conventional metal tooling often results in high costs and long lead-times. A group of materials and processes known as rapid tooling can produce a limited number of prototypes faster and more economically than conventional tooling. However, the differences in the material properties of conventional and rapid tooling result in mechanical property differences in the final plastic parts.

Tooling Life

Predicting the number of parts that can be molded in a SLA tool is very difficult due to the complexity of the molding process and the nature of SLA resins. The goal of this project is to reliably mold 50 parts in SLA tools. To do so, we must understand the failure mechanisms of SLA tools and relate these failures to molding process variables, mold material properties, part geometries, and the polymer being molded. We hypothesize that four failure mechanisms predominate: part shrinkage onto the core, thermal cycling of the mold, mechanical interlocking of the part and mold due to stair-stepping, and adhesion between part and mold.

Thomas is conducting experiments to relate surface finish to mold and process factors. More specifically, mold layer thickness and build style will be related to draft angle and number of parts. Ejection force, pre- and post-molding surface roughness, and mold hardness will be monitored. Does ejection force increase or decrease as surface roughness decreases, or as layer thickness decreases? These are some of the questions that will be answered by this research.

Given that an estimation of ejection forces on different areas and features of a part, it will be possible to develop ejection strategies for the part. We plan to develop a synthesis package that will determine the number of ejector pins, their sizes, and locations. Sunji should finish his Masters thesis early in 1999.

Electrodes for electrodischarge machining (EDM) applications are often difficult and expensive to machine. Also, traditional carbon material used to make electrodes wears too quickly. This project addresses both of these issues by using the Powder Injection Molding (PIM) process we developed for molding ceramic parts to form a zirconium diboride (ZrB2) material.

Many have made claims about the merits of new RP and RT related developments, but few can back up those claims with comprehensive dimensional data. We have a significant effort underway to develop better and faster ways to measure what we produce.

This project, supervised by Tom Kurfess, is to develop algorithms and procedures for extracting artifact quality information from the combination of a set of three-dimensional coordinates with the design CAD model. Significant developments to date include:

The ACIS solid modeling kernel is also used for analysis of general solid models.

We began by investigating the effect that eight SLA variables had on accuracy: hatch spacing, blade gap, z level wait time, layer thickness, hatch overcure, fill overcure, and sweep period. Planar, cylindrical, and conical surfaces were studied. As a result of a variable screening experiment, four of these variables were identified as most influential on accuracy: z level wait time, hatch overcure, fill overcure, and sweep period. Somewhat surprisingly, layer thickness was not one of the most influential variables.

The result of not accounting for anisotropic shrinkage is the inaccurate prediction of final dimensions. In ceramic injection molding this problem is magnified when compared to plastic injection molding because the shrinkage is an order of magnitude higher. Results to date include:

As use of RPM technologies is becoming more widespread, the issue of how to effectively use the tools has become more important. Specifically, we are interested in helping users to better understand when and how to use these tools - and when it is better not to use them.

Rapid prototyping is quickly becoming a key factor in reducing the cost and time to market associated with new product development. It allows the designer to evaluate aspects of a proposed design’s form, fit and function using a prototype made in days instead of weeks. This means that design flaws can be detected, corrected and the new design tested again much faster than was previously possible. Unfortunately a lot of the time and cost savings associated with rapid prototyping are often lost due to the use of processes that are incompatible with the type of information needed by the designer at any given point in the product development process. A common example is using a relatively expensive Stereolithography (SLA) model to check the overall form of a proposed design when a concept model costing much less and built in a fraction of the time would have provided the necessary information.

The goal of this research is to improve the selection of prototyping techniques in order to reduce both the costs and cycle times associated with product development processes. In order to accomplish this we must understand and quantify the value associated with each prototyping technique at any given stage in the design process. For the purposes of this research value will be defined as the benefit (the amount of useful information generated) divided by the cost (resources used) associated with creating the prototype. Of particular interest is the development of a benefits metric, which requires two sets of attributes: one that describes the capabilities of the individual prototyping technologies (prototype attributes) and another that describes the designs informational needs (design attributes). The final step is to correlate the two sets of attributes in an attempt to model the informational content of a given prototype.

The major result of this research to date is the completion of a value model for prototyping activities. The first task accomplished was identifying a set of attributes that can be used to describe a wide array of prototyping technologies. The attributes were selected based on our experiences as well as the examination of industry case studies conducted with both Siemens and NCR. The second task completed was identifying the major attributes that drive prototyping needs/decisions. The next step in creating this model was the construction of mathematical relationships that correlate individual prototyping technologies to the informational needs of a given design process. Finally these relationships were combined with an existing metric for measuring informational content during the design process to construct a quantitative measure of rapid prototyping technology benefits and cost.

Today, many firms are faced with a high rate of technological change, shrinking product life cycles, and intense competition in global, dynamic, and fragmented markets comprised of discerning customers. There is overwhelming evidence in the business world to show that a majority of technology based initiatives, in spite of scoring high marks on technical performance metrics, fall short of achieving their intended business objectives. A lack of understanding of the fundamental drivers of successful implementation results in their failure to accomplish the established business goals.

Under the guidance of Nagesh Murthy, we are identifying best practices in the development and implementation of RP technology. Bill Griffin and Atul Mandal are researching different methods through literature search. At the same time we are scheduling site visits in order to develop a more thorough understanding of the use of RP and its relation to the overall design process.

The product realization process, driven by market factors, is changing dramatically. Increased competition is forcing product realization to become faster, enabling shorter time to market. At the same time, globalization, core-competencies, out-sourcing, etc. are changing the structure of the product realization process--it is becoming distributed, both organizationally and geographically. Rapid prototyping has the potential to dramatically reduce time to market by shortening the time required to produce tooling. Realizing this potential, however, requires creating a technological infrastructure for both rapid tooling and distributed product realization. In response, a Rapid Tooling TestBed (RTTB) is proposed in order to focus on injection-molded products and processes. A team of eight Georgia Tech faculty from three units on campus have been funded by a three-year NSF Distributed Design and Fabrication Initiative grant to develop the RTTB.

The key question to be investigated in this proposal is: How early in the product realization process, and under what conditions, can design be separated safely from manufacture? This question will be investigated from the manufacturing viewpoint by (1) developing a distributed fabrication testbed for rapid tooling (injection molds) and (2) experimenting with different product realization processes for injection molds and molded components. Several product decomposition strategies will be developed and tested to ensure that these molds and components are manufacturable. As a result of this proposed project, “customers” of the rapid tooling testbed will receive nearly production-representative components in a variety of polymer, ceramic, and metal materials within three days of submitting a product model.

Janet Allen, Farrokh Mistree, and David Rosen are leading this thrust area. The goal is to translate a product design description into fabrication process plans, including process plans for polymer or powder injection mold tooling. A series of activities are required to perform this translation. Given a preliminary part design as input, our testbed will select the appropriate component material and fabrication process, tailor the design to that material and process, design molding tools for the parts, design the tool fabrication process, fabricate those tools, design the molding process, and mold the part.

The primary activity over the first four months of the project was to design the RTTB, to map the information flows through the various activities required to design and product rapid tools in a distributed environment. Present work is focused on three primary decisions, Resource Selection, Mold Design, and Fabrication Process Design, and on information modeling to support those decisions. A new selection decision formulation has been developed to match target values of attributes, which is necessary when trying to select materials and RP/RT processes that will yield production representative parts. This new formulation will be used for Resource Selection. SLA rapid tools act differently than conventional steel tools and must be designed somewhat differently. Based on Kent Dawson’s and Anne Palmer’s work in rapid tooling, a set of mold design rules are being developed to enable tailoring SLA mold designs. Additionally, Sunji Jangha is developing an ejection system design tool for use with SLA rapid tools and our standard mold bases. Under Fabrication Process Design, Aaron West is developing a method for selecting favorable values of SLA process variables to achieve build goals of accuracy, surface finish, and build time. This work extends that of Charity Lynn-Charney and Joel McClurkin.

Jon Colton is leading this research thrust. The goal is to characterize polymer injection molding in support of the molding process design activity. This enables testing and verification of DFM rules and process designs. With SLA and other RP technologies, resulting molds suffer from two main problems: low strength and poor thermal conductivity.

A prerequisite for good experimental studies is state-of-the-art experimental equipment. In the past year, we have a new insert mold with flexible cavity size and ejector pin pattern. Plus, a data acquisition system has been added with transducers for indirect cavity pressure measurement, in-cavity melt temperature, screw position and velocity, and hydraulic pressure. Several sets of experiments have been run, studying the effect of draft angle on ribs and slots on mold failure. Also, the resin flow direction relative to ribs and slots is critical for mold life. One set of experiments demonstrated that fan gates are far superior to pin gates. Additional experiments are underway to study feature height, width, depth, and draft angle interactions. Design rules for SLA molds will be developed that are analogous to those for steel tools. Results will be integrated with the Mold Design Methods RTTB thrust.

Tom Starr leads this research area. As described in the Rapid Prototyping of Ceramics project, the powder molding and sintering process differs from polymer molding. A ceramic or metal powder, mixed with suitable thermoplastic binder, is molded under relatively low temperature and pressure. After cooling and removal from the mold, the body is heated to high temperature, eliminating the binder and the inter-particle porosity to form a fully dense ceramic or metal part. The sintering step produces additional and significant dimensional changes in the as-formed shape. The primary research issue of this project is to develop a detailed understanding of these changes and the computational methods for predicting as-formed shapes. Our near-term focus will be on processing of stainless steel materials using a methodology similar to that outlined in the RP of Ceramics project.

Tom Kurfess is leading this thrust from the perspective of three-dimensional metrology. The objective is to characterize rapid prototyping processes and encode their characteristics for use in the SLA process design. We will also enable on-line monitoring and control of SLA builds. Research issues include: (1) quantification of systematic errors of SLA systems, (2) development of methods by which SLA systems can be validated and calibrated, and (3) direct feedback to SLA controller from metrology systems to “invert” anticipated deviations from target geometry.

Richard Fujimoto and Karsten Schwan from the College of Computing are leading this thrust. Their goal is to develop the distributed computing environment that enables the RTTB to function across the web. As required by the NSF Initiative, the RTTB must support distributed design and fabrication. It should be possible to search for materials and manufacturing processes on the web. Designers in one geometric location should be able to collaborate with manufacturers and tool makers in other locations. Mold-filling simulations and mold design optimization runs should be observable and controllable from remote locations. These challenges call for a new approach to developing distributed computing environments.

Overall, this project area is concerned with extending the suite of applications of SLA machines, particularly in the fabrication of functional assemblies and mechanisms. Generally, it is necessary to build each part in an assembly separately, then assemble them after the build outside of the SLA vat. It would be of significant benefit to eliminate the secondary assembly steps. Furthermore, functional assemblies and mechanisms may require stronger materials in high-stress areas, such as joints, in order to operate effectively. In the context of SLA, one solution is to incorporate inserts (of a different material) into SLA parts or assemblies that are placed into the build vat during or prior to the start of a build. Another is to fabricate mechanisms entirely out of SLA resin, taking care not to fuse the parts together. A second application is the production of smooth surfaces on SLA parts directly out of the vat. A meniscus smoothing approach will be taken.

It is sometimes necessary to build prototype assemblies that operate as mechanisms or that have multiple materials in them. In the context of SLA, one solution is to incorporate inserts into SLA parts or assemblies that are placed into the build vat during or prior to the start of a build. Imagine fabricating a working mechanism with metal shafts and bearings directly in a SLA machine. This vision requires both small and large changes in the operation of a SLA machine, and may require hardware changes as well.

The idea here is to lessen the effects of stair stepping by solidifying resin in the crevices between the stairs while the part is in the vat. In order to smooth the resin meniscus, a suitable scan vector type needs to be developed. It is necessary to investigate modifications to the optics system or to the elevator in the SLA machine so that laser shadowing effects are minimized. We will investigate the use of off-line process planning for meniscus smoothing. And we will explore real-time control of meniscus solidification through partial redesign of the optics, elevator, and control systems of SLA machines.

R&D will be completed that will define processes and methods to achieve SLA part smoothness of 50 micro-inches RMS or better without sacrificing part accuracy. The processes and methods will not be conventional sanding and painting, but will be part of the SLA building process or an automated post processing method. Specific objectives include:

In order to achieve functional assembly fabrication and smooth surfaces, SLA machines will require additional functionality. We will investigate the use of additional degrees of freedom in the operation of SLA machines, investigating 5-axes of motion or more. Conventional RP machines have 3 degrees of freedom; for example, the SLA has two DOF in the laser beam (scans XY), plus a third DOF with the elevator translating in Z. Additional DOF’s could include platform swivel and tilt. Two broad approaches are being taken to provide additional DOF’s. The first involves modifying the mechanical subsystem to provide platform motions beyond simple elevation. The second involves adding additional capabilities to the optics system.

In contrast to SLA based rapid tooling approaches, high speed machining is often used to fabricate tools for short runs or for prototype parts. This project will investigate the machinability characteristics of CIBA tooling board materials, specifically the CIBA-Express epoxy tooling board materials. Applications of machined tooling boards include injection and blow molding and sheet metal forming.

The objectives of this project are to

Prototypes produced by stereolithography (SL) have "stair-step" surfaces as a result of the layered SL build process. This pattern leads to a rough surface that reduces the utility of the SLA prototype. In contrast to the meniscus smoothing approach, this proposed solution involves a post-processing step of coating SLA part and processing that coating. Powder coating provides a means for adding polymer similar to the SLA substrate. Electrostatics can temporarily hold the powder coating on the surface until the coating melts and wets the surface. This liquid resin should preferentially fill in the "stair steps", thereby smoothing the surface. In addition, a second material system, liquid UV coatings, will also be investigated.

NASA has sponsored a first phase of work to investigate the feasibility of creating new methods for building large composite structures more quickly than current methods allow. The work is ongoing.

Jon Colton’s research will develop the science underlying the rapid production of composite structures, specifically those of carbon fiber-epoxy materials. The goal is to provide a minimum 100x reduction in the time required to produce arbitrary, laminated products. This scientific understanding will be reduced to practice in a demonstration device that will produce a part on the order of 12" by 12" by 12".

A laser CVD rapid prototyping system is capable of fabricating complex net-shaped metallic and ceramic structures. In contrast to most metal and ceramic RP systems, LCVD bonding occurs at the atomic level, producing a material that is fully dense, ultra-pure, and mechanically sound. Since LCVD can also produce fibers or layers in any given direction, the proposed system will mimic a production part in form, fit, and function. Furthermore, a capacity for multiple materials permits composite structures and functionally-graded materials and alleviates traditional material restrictions imposed by a given prototyping technique.

Theoretically, enormous possibilities exist, and Jack Lackey will ask the RPMI members and faculty for input to guide this work. Thus far, Lackey's team has performed extensive research into current laser CVD technology, leading to a design for the first LCVD rapid prototyping system.