21st Century Manufacturing Episode 1 Part 9 potx

4.4 CASTING METHODS FOR RAPID PROTOTVPING 4.4.1 Introduction The classic manufacturing textsbyDeGarmo and associates 1997, Kalpakjian 1997, Schey 1999, and Groover 1999 are remarkably co

ComparisoD of Approxlnule AeeurllCY

of Rapid PwlutypinK PIon:_

Sinterstation

2000 (SLS)

Sinterstation

2500 (SLS)

SOC 4600 (SGC)

0.005

0.005 0.006 13~' SOC 5600 (SOC)

~ l5 LOM-2030H (LOM)

~~

~ FDM 2000 (FOM)

0.006

0.01

0.005

0.003

o 0.001 0.002 0.003 0.004 0.(lO5 0.006 0.007 0.008 OJXl9 0.01

Accuracy (inches per inch) Note that the machine suppliers quote in "thou' and that one "thou' '" 25 microns

Figure 4.15 Comparison of accuracy (as of March 2000).

4.4 CASTING METHODS FOR RAPID PROTOTVPING

4.4.1 Introduction

The classic manufacturing textsbyDeGarmo and associates (1997), Kalpakjian (1997), Schey (1999), and Groover (1999) are remarkably comprehensive in their coverage of the casting process The several methods of casting include:

• Lost-wax investment casting

• Ceramic-mold investment casting

• Shell molding

• Conventional sand molding

• Die casting

Rather than duplicate the material found in other books, this section focuses

on casting as it is done by rapid prototyping companies Batch sizes from 50 to 500 are typical The key market strategy is that casting is cheap and fast However, it may not be the choice for the final product because of its tolerances Depending on the SLA·250(SLA)

wax processes to +/- 375 microns (0.015 inch) for standard sand castings (also see Chapter 2)

4.4.2 Lost-Wax Investment Casting

As mentioned in Chapter 1, the fundamentals of casting were invented by Korean

and Egyptian artists many centuries ago The following steps are known as the

lost-neering or art object is first carved from wax; (d-f) it is surrounded by a ceramic slurry that soon sets into solid around the wax; (g) the wax is melted out through a hole in the bottom, leaving a hollow cavity; (h) this hole is plugged, and liquid metal the ceramic shell can be broken away to get the part;(j)some cleaning, deburring, and polishing are needed before the object is finished

The process was greatly improved and made more accurate during World War

II for aeroengine components Today it is used for products such as jet engine turbine from injection molds, assembled on treelike forms, and then treated with the slurry Alternate layers of fine refractory slip (zircon flour at 250 sieve or mesh size) are applied, followed by a thicker stucco layer (sillimanite at 30 sieve or mesh size) and liquid acid hardener Drying takes place in ammonia gas.The next step is to elim-inate the wax in a steam autoclave at 150°C,fire the mold for 2 hours at 950 "C, then pour in the liquid steel or aluminum

In summary, the modern lost-wax method has one of the best tolerances in the casting family because the original wax patterns are made in nicely machined molds Today,tolerances of+/-75 microns (0.003inch) are readily obtainable Also the as-cast surface is relatively smooth and usable for the same reason Other advantages include:

• No parting lines if the wax original is hand finished

• Waxes with surface texture can give direct features such as the dimples on a golf club

• Automation of the slurry dipping is possible using robots, thereby reducing costs

• Products such as turbine blades can be unidirectionally solidified, giving good mechanical properties in the growing direction

4.4.3Ceramic-Mold Investment Casting Procedures

The snag about the previous method is that the wax pattern is destroyed The

ceramic-mold investment casting technique therefore employs reusable submaster

patterns in place of the expendable wax patterns This version of investment casting fine care and expense that go into creating the original master positive in Step 1 The steps are as follows:

• Step L Positive: make an original master pattern with stereolithography or

Mold to make pattern

CIW~.

or plastic pattern Ejecting pattern

(c) ~~

Pattern assembly (tree)

Slurry coating Stucco coating Completed mold

Autoclaved

"~Jt <, -/

Molten

-s-c-wax or

plastic

W~.'.CW7

.[( J"

.:• ::.,

~

Casting

Figure 4.1f The lost-wax investment casting process Upper diagrams (a) through (c) lead to the tree of wax master patterns Middle diagrams show the slurry and stucco being applied Lower diagram shows the casting (adapted from literature of the Steel Founders' Society of America)

• Step 2 Negative: create a shell around the master with highly stable resin A negative space is created around the original positive master pattern This shell can be pulled apart to give a parting line

• Step 3 Positive: create reusablesubmaster rubbery molds from the shells

• Step 4 Negative: create the destroyable slurry/ceramic molds

• Step 5 Positive: pour metal into the ceramic molds, which are then broken apart to get the components, which must then be degated and deburred SLA can be used to make the original master pattern, or a CNC machine can

be used to mill the master from brass, bronze, or steel Of course, the process can start

at Step 3, but this might damage the original master, especially if it is SLA Also, to get high productivity in the factory, it is preferable to have many molds at Step 3, all

of which can be made from the stable resin negative in Step 2

Prototyping companies like to use the hard resin to fabricate the negative in Step 2, because the resin has good dimensional stability Note that it is typical to have two resin molds, one for each side of the casting, separable by a parting line Once the hard resin shells have set, they can be filled with a slurry gel that solid-ifies to a hard "rubbery positive" for Step 3 This intermediate submaster mold can

be stripped away from the resin shells while it is still "rubbery." The material is ideal for the rather rough handling environments of a foundry, and the rubbery properties mean that no draft angles are needed for stripping these submasters off the resin shells

The Step 4 negative mold is made from a graded aluminosilicate with a liquid binder (ethyl silicate) and isopropyl alcohol This is poured around the subrnasters inner cavity, the slurry is fired at 950 ~C to give it strength, and the casting process, say with molten aluminum, can begin

After solidification, the component is broken out of the ceramic, cleaned up, and deburred The parting line can cause problems, but in general, good accuracy is

obtained: +f- 125 to 375 microns (+1-0.005 to 0.015 inch).

4.4.4Shell Molding

An alternative form of high-accuracy casting isshell molding. Metal pattern plates are first heated to 200°Cto 240°C A thin wall of sand,S to 15 millimeters (0.25 to 0.75 inch) thick, is then sprayed over the plates The sand is resin-coated to ensure adhesion to the metal plate Phenolic resins, with hexamethylene-tetramine addi-tives, are combined with the silica to ensure rigid thermosetting of the sprayed sand accurate for casting Once the excess sand is removed and casting is finished, accu-racies can be as low as +/- 75 microns (0.003 inch)

4.4.5 Conventional Sand Molding

The cruder, cheaper version of casting starting with wooden or plaster patterns is calledsand casting. A sand impression is made around the pattern with gates and

risers for the poured metal This gives tolerances of +1- 375 microns (0.015 inch).

Newer developments include:

1 A high-pressure jolt-and-squeeze method: Here mechanical plungers push the sand against the mold at a jolt of 400 psi This gives a tighterfitof the sand against the pattern and hence better tolerances after casting

2 Carbon dioxide block molding: Here the interfacing between the sand and tbe pattern is made up of a special material about 12 millimeters (0.5 inch) thick

It is a refractory mix of zircon or very fine silica, bonded with 6% sodium sili-cate, which is then hardenedbythe passage of carbon dioxide

4.4.6 Die Casting

Die casting is predominantly donebythe high-pressure injection of bot zinc into a permanent steel die Today, the die or mold for this type of casting is almost certain

to be milled on a three- or five-axis machine tool

Die costs are relatively high, but smooth components are produced with

accu-racies in the range of +1- 75 microns (0.003 inch) However, these high costs for the

permanent molds mean that die casting does not really fit into the rapid prototyping family It is mostly used for large-batch runs of small parts for automobiles or con-sumer products Since low melting point materials such as zinc alloys are used in the process, component strengths are relatively modest

Today, the injection molding of plastics (Chapter 8) is often preferred over zinc die casting

4.5 MACHINING METHODS FOR RAPID PROTOTYPING

4.5.1 Overview

Chapter 7 deals with the generalized machining operation including the mechanics

of the process This chapter focuses on advances in CAD/CAM software that allow CNC machining to be more of a "turnkey rapid prototyping" process One goal is to the intensely hands-on craft operations (e.g., process planning and fixturing) that demand the services of a skilled machinist

CyberCut™ is an Internet-based experimental fabrication test bed for CNC machining The service allows client designers on the Internet to create mechanical components and submit appropriate files to a remote server for process planning and fabrication on an open-architecture CNemachine tool Rapid tool-path planning, novel fixturing devices, and sensor-based precision machining techniques allow the original designer to quickly obtain a high-strength, good-tolerance component (Smith and Wright, 1996)

4.5.2WebCAD: Design for Machining "onthe Internet" on

the Client Side

A key idea is to use a "process aware" CAD tool during the design of the part This

prototype system is called WebCAD (Kim et al., 1999) Sun Microsystems' Java™_

used as a framework for serving mini-applications The GUI is a 2.5D feature-based-design system that uses the destructive solid geometry (DSG) idea introduced in the last chapter (Cutkosky and Tenenbaum, 1990;Sarma and Wright, 1996).Recall that the user starts out with a prismatic stock and removes primitives or "chunks" of building up a part incrementally from "nothingness." In the "destructive" paradigm, tain shapes of material, referred to as features These features take the form of pockets, blind holes, and through-holes

WebCAD also contains an expert system capturing rules for machinability.At the top of Figure 4.17, the designer is shown being guided by these rules For from being designed too close to an edge In the event that the designer violates a ther into the block-typically by its radius dimension WebCAD also uses a WYSIWYG ("what you see is what you get") environment, with explicit cutting tool selection and visible comer radii on pockets At the time of this writing, further improvements also include Ireeform surface editing and selection of different cutting tool sets depending on final fabrication location (Kim, 2000)

The rationale for imposing destructive features upon the designer is that each

of these features can readily be mapped to a standard CNC milling process The scheme thus resembles the interaction between a word processor and a printer

to DSG limits the set of parts that can be designed However, the key advantage of this design environment is that the design-to-manufacture process is more deter-ministic than conventional methods, which rely on unconstrained design and on looser links between design, planning, and fabrication Experience shows that designers are somewhat concerned at first that they are constrained; however, the opportunity to be provided with the correct part very quickly proves to be attractive 4.5.3 Planning on the Server Side

When the client's design is finished, the resulting geometry can be sent over the pipeline takes the geometry and determines in which order the features should be tool

Macroplanning orders the individual features and creates the specific machining setups in fixtures CyberCut's current macroplanner is a feature recogni-tion module that can reliably extract the volumetric features from2.5V parts The output of the module is not just a machining feature set but a rich data structure that also gives important connectivity information that relates one feature to another A recent advance in the macroplanner is its ability to recognize and process features containing freefonn surfaces (Sundararajan and Wright, 2000)

2A 2.50 featureis a machinable feature with arbitrary outside peripheral contour and a uniform

(WebCAD)

A novel JAVA-based design

with WYSIWYG part features

A design consultant: 'The hole

is too dose to the edge"

Freeforrn feature based manufacture Custom control algorithms for precision and flexibility Sensor integration

FIgure 4.17 The CyberCul project: integrated design, planning, and f"bricotion.

spe-cific tool motions Colloquially speaking, this is the step that is like lawn mowing: each volume has to be carved out with a specific tool diameter, and the overlap roughness The comers of pockets Gust like lawns) might require special methods

Mill;l'opiaD

• Make hole 1 before

pocket 2

Change setup

• Refixture

Mkro_

• Use tool 1

Cutatl600rprn

• Feedrate Jrnm/s

• Perform one pass

F_brlcatloll

Freeform surfaces must be divided into flat and steep regions The flat regions are machined with a slicing tool-path pattern.This yields good tool-path uniformity and only moderate computational complexity The decomposition aims at minimizing machining time within the constraints of the specified surface roughness, tolerance, which can be sent back to the designer, providing an early estimation of the machining costs

4.5.4 Fabrication by Milling on the Server Side

Finally,a stream of NC commands performs the machining on an open-architecture milling machine (By contrast, if it had been determined along the way that the client would have been better served by SFF technology, CyberCut can connect to a fused deposition modeling [FDM] machine.) The particular milling machine being used is

an open-architecture machine that can execute advanced tool-path trajectories One represented by NURBS This ability brings a richer surface generation capability to the ostensibly traditional machining process By doing so, it continues to compete with the SFF methods from the point of view of geometric complexity (Hillaire et al., 1998) More details of the open-architecture machine tool itself are reserved for Chapter 7 on machining

4.6 MANAGEMENT OF TECHNOLOGY

4.6.1 Summary

Solid freeform fabrication (SFF) techniques and conventional rapid prototyping techniques such as machining and casting are key technologies for improving product realization cycles and reducing time-to-market The availability of lntemet-based software tools (Berners-Lee, 1989; Java, 1995) has accelerated the links between CAD and prototype creation The Internet has also allowed access to man-ufacturing sites in many different countries (Smith and Wright, 1996;DeMeter et al.,

• SLA emerges as the most commercially accepted of the newer SFF proto-typing methods

• SLS emerges as a very useful, commercially accepted alternative to SLA for overhanging structures needing support and for stronger materials that can be sintered rather than photocured

• FDM is an excellent choice for an in-house machine that can be used by an industrial design team for an iterative series of prototypes

• LOM is excellent for larger components

• 3-D printing and planarization using the inexpensive Sanders and Z-Corporation machines are gaining commercial acceptance at the time of this writing Very inexpensive 3-D digitizers coupled with miniature milling

machines are also entering the market (see URL for Roland Digital Group at the end of the chapter}

• Machining and casting remain central to the rapid prototyping field, especially for high-strength prototypes and longer batch runs of several prototypes 4.6.2Future Trends

The accuracy of processes such as stereolithography and selective laser sintering is improving as time goesby.These processes are being used more and more in the cre-ation ofthe original, first master for casting and for plastic injection molding As con-sumer products such as stereos, cellular phones, personal digital assistants (PDAs), and handheld computers (Richards and Brodersen, 1995) begin to look more aerody-namic, there is a need to create molds that have unusual curves and reentrant shapes; these are easy to create in SLA or S~ especially in comparison with machining

It has nonetheless been emphasized that SFF's accuracy is poor in comparison with machining Overhanging structures may be hard to support during fabrication, and there are problems with component warping during curing While simple shapes might have accuracies of.ct-:25 to 75 microns (0.001 to 0.003 inch), the range for complex shapes might be as high as+1-125 to 375 microns (0.005 to 0.015 inch) While the strength of SFF parts is today less than machined parts, new trends are closing the gap The FDM parts made by the Stratasys machine can be formed in near full strengthABS and similar polymers Cheung and Ogale (1998),for example, have increased the strength of photopolymers by fiber reinforcement Also, research

at Sandia Laboratories on a process called laser engineered net shaping (LENS) is permitting direct fabrication of high-strength metal molds This and similar projects are modified versions of DTM Inc.'s SLS process

At the same time, CA O/CA M techniques for machining are advancing rapidly For example, the CyberCut freeform design tools linked to open-architecture milling machines will continue to expand machining's capability (Greenfeld et al., 1989; Schofield et al., 1998; Hillaire et al., 1998) There is a subtle point to be made here: much of the increased activity in the SFF prototyping methods was originally prompted bythe poor communication between CAD and CNC machine tools During the late 1980s, stereolithography's competitive edge over machining came from the fact that the CAD model could be instantly "sliced" and then turned into laser scanning paths for rapid part production With the CyberCut methodology and open-architecture control, the conventional machining process can be equally com-petitive from an art-to-part speed standpoint, and it continues to give the high-accuracy and product integrity qualities that it always gave

The evidence is thus clear that the capabilities in both the machining and the SFF fields are constantly improving It has also been noted that several of the tion with machining "to get the best of both worlds." Perhaps in a similar way, the 3D

ment casting." In QuickCast, disposable SLA patterns are fabricated with distinctly

hollow internal structures When the ceramic shells for casting are created around

and ready for use This process, described by Jacobs(1996, 183-252), is gaining rapid acceptance commercially

The issues mentioned are predominantly technical As this chapter draws to a close, it is important to recall an earlier point from Chapter 2 that "prototypes structure the design process" (see Kamath and Liker, 1994).Physical prototypes focus the efforts

of a distributed design team, especially if subcontracting is a big part of the process

Perhaps the most important conclusion is this: each manufacturing process will playa vital role at different points in the product development cycle SFF techniques

through to create highly accurate molds, and plastic injection molding will be most evident in the final high-volume production method for the consumer's product Once again it must be emphasized that "manufacturing in the large" is an integration

of many software tools, physical processes, and market strategies

In summary, rapid prototyping dramatically accelerates time-to-market

• Psychologically, it focuses the attention of the members of the design team in

a "learning organization" (see Chapter 2)

• Physically, it reduces the time necessary to make a full production die from hardened steel and to launch into mass production

4.7 GLOSSARY

4.7.1 01 Casting

Low-pressure casting, often of liquid zinc, into a machined mold

4.7.2 Electrodischarge Machining IEDM

The use of an electrode to melt and vaporize the surface of a hard metal Usually restricted to low rates of metal removal of very hard metals

4.7.3 G-Codes

The standard low-end machine tool command set that gives motion, for example, G1

=linear feed

4.7.4 Injection Molding

Viscous polymer is extruded into a hollow mold (or die) to create a product 4.7.5 Ink-Jet Printing in 3-0

Rapid prototyping by rolling down a layer of powder and hardening it in selected regions with a binder phase that is printed onto the powder layer

4.7.6 Investment Casting

The word investment is used when time and money are invested in a ceramic shell