Rapid prototyping: a tool for casting design and verification.

The adoption of CAD as the primary form of design communication has allowed rapid prototyping to emerge as a means of component verification and short-run casting production.

As the foundry industry continues to embrace 3-D CAD systems as the tool for design communication, rapid prototyping (RP) increasingly has become a quick and reliable option for casting design verification. RP is the generation of a physical model from a 3-D computer design. RP systems use the computer model as the blueprint and its dimensions to fabricate physical models layer-by-layer using liquid, powder and sheet materials.

OEMs often make their decision on where to source a metal component based on time. Conventional methods of casting production, including the construction of tooling and pouring of a casting, are often too time-exhaustive to provide a cast metal model to an OEM for verification of the component's shape and its fit in the overall end-product design. RP allows for design verification before a casting is poured and can make the process of "print-less" tooling and casting quotation easier and more accurate.

RP processes also have evolved into an option for cast metal prototypes and short-run casting production. By using rapid prototypes that serve as patterns in investment casting and reusable masters in sand and plaster molds, RP can replace expensive tooling or serve as a stand-in for short-runs while the tooling is being made.

For the foundry and OEM, the verification of a cast component before full production is vital to reducing lead times and total costs. RP, combined with concurrent engineering, facilitates early-stage changes by allowing OEM "fit checks," and promoting pattern shop, foundry and machine shop feedback early in the design cycle. The cost of changing the basic design of a product increases rapidly as the design advances through the development cycle (Table 1). If a change is made during the design of a casting it will cost $100 on average, while waiting until production could cost $100,000 or more.

Table 1. The Cost of Design Changes From Casting Concept to
Production

Conceptual modeling                     $10
Detail Design                          $100
Prototype/test                        $1000
Manufacturing                       $10,000
Product release                    $100,000

Source: Wohlers Assoc.'s "Rapid Guide to Rapid Prototyping"

[TABULAR DATA FOR TABLE 2 OMITTED]

RP Processes

RP begins with a 3-D CAD model that has been translated into the RP industry-standard .STL format file. The .STL file consists of the X, Y and Z coordinates necessary to represent the surfaced or solid casting model as groups of small polygons. The RP system software then slices the .STL file model into horizontal layers, generating any necessary support structures (for the prototype model to rest on once it is built) and controlling all RP system functions according to user-supplied parameter settings. Following (and also in Table 2) is a look at the various RP processes used to generate physical models and tooling for castings.

Fused Deposition Modeling (FDM) - Stratasys' FDM process heats thermoplastic filament (0.070-in. diameter plastic wire) to a semiliquid state and then deposits it - via an extrusion process - layer-by-layer onto a foam base until the part is completely formed [ILLUSTRATION FOR FIGURE 1 OMITTED]. One extrusion tip deposits model material while another tip deposits "release" material that forms support structures designed to easily release from the prototype during post processing. The FDM plastic prototypes can be used as a waxlike pattern for investment casting, or prototypes can be built from investment waxes. FDM is best suited to parts with a thin-wall thickness that do not contain extensive undercut geometry.

Laminated Object Manufacturing (LOM) - Developed by Helisys, LOM builds wood-like prototypes using paper that is coated with a heat-activated adhesive [ILLUSTRATION FOR FIGURE 2 OMITTED]. During LOM, the paper automatically is pulled and laminated over the build envelope and then is cut into the 2-D casting cross-section by a laser beam. The excess material remaining after the LOM prototype is built is then removed by picking the diced cubes away from the prototype during postprocessing. As a subtractive process, LOM is best suited to geometry with heavy walls. The wood-like nature of LOM prototypes make it a good option for producing patterns and coreboxes for sand casting. The large working envelope and relatively inexpensive raw material have maintained the viability of this technology.

Selective Laser Sintering (SLS) - As one of the newest RP processes, SLS, developed by DTM Corp., uses a heat-fusible powder that is layered across the build chamber. A C[O.sub.2] laser beam traces over the surface area of the 2-D cross-section being built, melting and fusing the powder in its path. This process is repeated, layer-by-layer, until the build is complete. Post processing consists of gently brushing and blowing the loose un-sintered powder from the prototype. SLS is a good method when the design contains extensive undercut geometry because the un-sintered powder acts as a support structure without having to build one. SLS also can be performed on resin-coated sand (both DTM and the German company Electro Optical Systems GmbH have this technology) for the tool-less production of sand molds and cores, and, testing is underway on an SLS polystyrene material for building waxlike patterns for investment casting.

Solid Ground Curing (SGC) - Developed by Cubital, SGC solidifies an entire horizontal cross-sectional layer of photo-reactive polymer by projecting UV light through a photo-mask that is created by using an electrostatic process similar to laser printing. Any uncured resin is vacuumed from the surface and a layer of wax support material is distributed over the build envelope. The layer is then milled to the correct thickness and the process is repeated until the build is complete. Post processing of the physical model involves melting or rinsing away the wax support structure. Because the entire 2-D cross-section layer is solidified all at once, and the post processing stage is automated, this RP process is most efficient when multiple parts are batched for build.

Stereolithography (SLA) - This process, manufactured by 3D Systems, uses a computer-guided laser beam to harden liquid plastic cross-sections layer-by-layer until the 3-D physical model is completely formed. The support structure necessary to sustain a model's overhangs or undercut geometry is built independent of the physical model. Once the model has been generated, SLA post-processing consists of removing the manual support structure and any uncured resin, and the reduction of "stair-stepping" by using sandpaper to finish the model. Stair-stepping is common to rapid prototypes because the models are built in horizontal layers (as thin as 0.004 in.) that are stacked one on top of the other.

The SLA process using its Quickcast build style also can build a physical model for use in place of a wax pattern in short-ran investment casting. This method of SLA hardens the liquid plastic in a honeycomb internal structure enclosed by an external "skin," which results in an accurate model that minimizes the ceramic shell damage during the flash-fire burnout process.

Three Dimensional Printing (3DP) - Developed by Z Corp., this new RP process builds prototypes by spreading starch-based powder onto the build box and selectively dropping binder that bonds the powder to form a 2-D cross-section of the part. This process is repeated until the part(s) is built. The 3DP system was designed to quickly build conceptual or physical appearance models in an office environment to provide engineers with "instant" models. The manufacturer says that its build time is 10 times faster than other RP processes, however, the process' disadvantages include part inaccuracy and fragility.

Direct Shell Production Casting (DSPC) - Performed by Soligen, this process uses the 3DP technology to produce ceramic foundry molds (that can receive any metal and alloy) directly from a CAD database. This tool-less molding process is fast and does not require the draft angles, parting work or coreprints associated with sand and plaster casting. One disadvantage is the Z-axis stair stepping (which can be removed from accessible areas) that is transferred from the ceramic mold to the metal casting. Also, the one-to-one ratio of ceramic mold to metal casting can be a limiting factor when producing more than a few prototype castings, and small diameter pipe core situations may prevent the use of this process.

CNC Machining - Although it isn't an additive layering process, CNC machining can be the fastest, most accurate process for RP. Patterns can be machined using a CAD database to generate tool paths from a variety of wood, plastic, metal and hybrid materials. Metal bar stock prototypes also can be fabricated if a few prototypes are needed. The accuracy of machined prototypes can be controlled by the cutting tools used as well as by the feed and speed parameters chosen by the machinist. The accuracy of CNC machining is unrivaled by any RP process and is more flexible in regard to raw material usage. Patterns can be cut to any surface finish desired to minimize the inaccuracies and labor involved with hand finishing. CNC machining is best applied to components that: require accuracy not achievable on RP equipment, are very large and solid, and are simple enough that RP isn't warranted. Parts that have numerous undercuts requiring multiple CNC setups may be a better fit for RP.

Choosing an RP Method

Despite the recent advances in computer-based simulations of casting solidification, many OEMs still require prototype metal castings to be tested before approval is given for mass production. The problem is that the production of hard tooling (such as metal dies for diecasting, permanent molding and investment wax injection, or cope/drag tooling to withstand long sand casting production runs) often doesn't meet the OEMs lead time requirements as it wants, for example, a shipment of 100 prototype castings for use in 3 weeks. To combat this, RP is used in conjunction with metalcasting processes to facilitate cast metal prototypes and short-run production.

To determine the best combination of RP and metalcasting processes to manufacture a short-run of metal castings, a number of criteria must be considered including: raw material, number of castings needed, accuracy, surface finish, mechanical characteristics, budget, project timeline and part geometry (size and cross-sectional area/thin walls or thick). Three metalcasting processes - investment, sand and plaster molding - work with RP processes to initiate these short runs.

Investment Casting

The key element that makes RP and investment casting compatible is that investment casting uses an expendable pattern (wax or foam), and RP can produce an expendable pattern directly from a database without the need for tooling [ILLUSTRATION FOR FIGURE 3 OMITTED]. Because the process is capable of casting virtually all metals and producing a surface finish comparable to other processes, complex investment cast prototypes can be produced for die, plaster and permanent mold castings. Cost, however, is usually the limiting factor when considering the direct investment casting method because RP patterns are expensive when compared to traditional wax patterns.

The direct method is economical early in the design cycle when quantities are low and design changes are anticipated. Any rapid prototype model that can be flash-fire incinerated without damaging the ceramic investment shell can be used as a wax pattern substitute. SLA Quickcast patterns are the most commonly used rapid prototypes for direct investment casting. FDM Quickslice software also allows the "fast" build style that produces results similar to SLA. Both build styles reduce pattern build time and encourage pattern implosion during firing to reduce shell damage.

Indirect investment casting is the use of temporary tooling to make wax patterns. Temporary wax tooling can be manufactured by RP (DTM's Rapidsteel), the machining of aluminum or the production of RP master patterns from silicone mold or metal-filled epoxy tooling. When combined with the skills of a mold maker, virtually any RP- or CNC-machined master pattern can be used to make silicone or metal-filled epoxy wax tooling. Temporary tooling makes sense when the price of multiple RP patterns approaches or exceeds the cost of temporary tooling (usually 5-10 RP patterns). Temporary silicone mold or epoxy tooling can be fabricated in 1-5 weeks (depending on complexity) vs. machined tooling, which takes 6-25 weeks to build.

Sand Casting

Any RP pattern that is durable enough to survive the sand molding process can be used directly as a master pattern to make sand molds. RP coreboxes can be used to make cores, especially "dump boxes" that are filled with sand manually rather than with a core blower.

LOM patterns have been used extensively to make sand casting masters due to the large working envelope of the RP equipment (32 x 22 x 20 in.) and the process' subtractive nature, which lends itself to parts with heavy walls. SLA, SLS, FDM and SGC patterns also are capable of being used as patterns for sand casting if treated with care during the molding process.

Industrial material manufacturers have introduced a plethora of new "tooling board" materials for CNC machining that are excellent for sand casting patterns and coreboxes. These tooling boards are made of easily machinable materials that provide an excellent surface finish with minimal hand finishing. The materials also are more stable than traditional wood patterns and, in many cases, offer better abrasion resistance [ILLUSTRATION FOR FIGURE 4 OMITTED].

Sand molds and cores also can be produced directly on RP equipment by laser sintering resin-coated foundry sand with the SLS process. Parts cast using the RP sand method display mechanical and physical properties similar to parts cast in a production sand foundry. This process allows cores to be integrated into the mold as it is built, providing a method to build parts with virtually no limitations on complexity [ILLUSTRATION FOR FIGURE 5 OMITTED]. So far, due to the expense, the use of RP sand molds and cores has been limited to complex geometry such as an automotive cylinder head water jacket and port cores. The RP sand process has the potential to reduce complex casting lead times from the traditional 8-12 weeks to 1-2 weeks. One limitation of this process is the one-to-one ratio of molds/cores to castings yielded. As with the direct investment process, if the RP sand process does not yield a good casting, an additional mold/core(s) must be built, resulting in additional cost and time. In many cases, RP sand cores are effective when used in conjunction with RP or CNC machined patterns, especially for parts that have complex internal coring but relatively simple external geometry.

Plaster Molding

The plaster mold process consists of pouring plaster slurry over a pattern inside a flask. After the plaster hardens, the mold is stripped from the pattern and placed in an oven. Once the mold is dry, the metal can be poured using a vacuum-assist technique. Due to the composition of the plaster, however, only zinc, aluminum and brass alloys can be used in the plaster-mold process. Due to this temperature restriction, its ability to reproduce detail (thin walls to 0.05 in.) and excellent surface finish, and its dimensional tolerances and repeatability, plaster molding is used to prototype diecast parts.

The plaster molding process requires tooling similar to that used in a sand foundry. Tooling used in a plaster foundry must be watertight due to the moisture in plaster molds. For this reason, LOM patterns are not recommended for direct use in the plaster molding process. One advantage of the plaster process is the low abrasion wear on tooling because the plaster mold is not as abrasive as a sand mold. RP materials unaffected by moisture (such as those used in SLS, FDM. SLA and SGC processes) are suited to the plaster process. By using the loose pattern technique and handcut gating, castings can be made with minimal tooling expense.

Flexible rubber patterns also can be used in the plaster-mold process. Rubber patterns are more forgiving than hard plastic patterns when slight undercuts or minimal draft are necessary.

Plaster molding is a more labor intensive process than nobake molding, thus plaster castings are more expensive than sand castings. In addition, the plaster mold material cannot be reused, adding to the casting cost.