32. Rapid Prototyping
• The key concept is RAPID: generally this is an all-in-one-step production of a part geometry.
prototypes to allow fast review of part shape, simple assembly, aesthetics, manufacturability, etc.
low volume production: very small numbers of parts can be made using this technology.
reduce prototype/production times from months to weeks or days
a physical model is easier to “sell” to customers and management
physical models are easier to check for errors. Graphical methods often result in cluttered views
avoids the high cost of prototype tooling, and allows (more) design iterations
prototypes costs can be lower than production types
tolerances are generally >.005”
primary materials are specialized, and other steps are required to produce metal parts
• Originally developed for use with stereolithography, but now used by many other processes.
• The standard for Rapid prototyping systems
• Basically connected 3D triangles
• 3D smooth surfaces are tessellated into triangles. the higher the degree of tessellation the closer the surface approximates the smooth surface.
1. with a direction of nodes defined clockwise for the out direction
2. with similar nodes at the corners of triangles. If the triangles don’t overlap, the model will have gaps and be invalid.
• A general approach to determining a rapid prototype slice is to use a ray projection through the collection of polygons. When the ray strikes a triangle it is in/out of the solid. (This is a simple geometrical problem.) A set of lines constitute a slice.
3. Fire a “ray” through the triangles, and find intersections
For the stereolithography machine we need to develop a “scan line” for the laser. When the laser scan-line (the ray) is “in” the part, the laser is on, and thus developing the light hardening polymer.
• If polygon vertices don’t match up, then there will be gaps between the polygons. This can result in non contact, that leads to material added/subtracted unexpectedly.
• Another problem can arise from mobius strip representations. Because the outside is defined by the order of node definitions, a mobius strip will lead to a back touching a front.
• Invented by Charles Hull 1984
• Now developed by 3D systems Inc (90% of market in 1991)
• The physics of the process is based on a photo-sensitive polymer that will harden when exposed to high intensity laser light
• The process uses a vat of photopolymer with an elevator for the part to lower on. The elevator starts at the top of the bath, and drops down a layer at a time as the laser develops each layer.
• supports can be created in the CAD model of afterwards with programs such as bridgeworks (by Solid Concepts, Los Angeles, California)
to prevent the recoating blade from striking the elevator platform
to correct for variations in elevator platform surface.
to allow easy removal of the part from the platform
• There are a few basic supports used,
• CAD files are converted to .STL files
• .STL files converted to slices using the parameters, such as,
various parts on a platform (different .STL files can be mixed)
blade sweeps per layer and sweep period, and “z-wait” to wait after recoating for material mixing
• The part is then immersed (approx. 0.5”) with a waiting period to recoat the surface and the wiper blade is used to clear the excess fluid from the top of the surface. (Note: the sweep is optional, but it is used to get consistent thickness)
• The part is then moved, and the laser has a focal point near the surface that hardens the polymer.
• As the polymer hardens, it shrinks. This shrinkage causes a change in the volume of the fluid. To correct for this the tank has a fluid level detector that will control an adjustable reservoir that will add enough fluid to compensate for volume change.
• Layers vary in thickness from 0.002” to 0.020”. This is controlled by the amount the platform is lowered into the photopolymer. Thinner layers give smoother, higher tolerance parts (with polymer more cured) but these take longer.
• The laser is stationary, but optics and mirrors are used to guide the beam to x-y coordinates on the surface of the fluid.
• After the part is done, the part raises above the fluid, and resin drains out. The elevator can be tipped to drain trapped volumes.
• After removal from the bath the part is cleaned off with towels and Q-tips, and hardening of the resin is completed in a curing oven.
• The laser: is often about 10-200 mW (more power is required for faster operation)
often He-Cd or Argon-Ion to produce UV radiation about a 320-370 nm wavelength
• The optics: the user can set the focal distance of the laser to the range of the slice thickness.
• x-y positioning: uses 2 computer controlled mirrors to reflect beam.
• polymer vat: generally holds 20-200 liters of resin. Can be interchanged to speed up change of resin (lower downtime).
• The photopolymer is light sensitive and toxic. Therefore the operation vat is often out of sight and the unit uses a ventilation unit to evacuate fumes.
• Post Curing Apparatus (PCA): uses high power ultraviolet light to complete the curing of mostly solidified polymer.
times are typically 1 hour and up
after curing parts become non-toxic
sharp-edges tend to get “filled” by resin, thus reducing the “stepped” effect between slices
popularity makes this process well supported
extra time required for postcuring (up to 16 hrs)
polymer shrinks as it hardens: the result is stress that warps the part
toxic chemicals (resin and cleaners)
limited selection of chemicals (general cost $100-200 a liter)
experts needed for process setup
work required after to remove supports
used to reduce curing time and part stresses that cause warping
most of the part is cured before, because of multiple exposures
staggered hatching uses exposure in-between lines of previous exposure
• SL units from 3D systems (also see attached specs)
• Units are sensitive to vibration
basic polymer is slightly brittle and therefore is best suited to conceptual models
“Exactomer” is well suited to trial assemblies, and has been used to make secondary rubber, and spray metal tooling (being less brittle it won’t break when being removed from molds).
Investment casting molds can be made using hollow cores (that minimize polymer expansion when melting) that won’t crack the mold.
Dupont is creating an investment casting resin that won’t crack the mold.
• In some research fibers have been added to a stereolithography process to obtain higher strengths. [Hyer, 1991]
• An inexpensive stereolithography unit can be made using UV light guided by a fiber optic cable.
• Large parts can be created in pieces and glued together. For example, an impeller can be created in sections. The sections are glued with normal resin, and hardened with a UV lamp. Metal inserts can be added by press fitting, and the part can be machined for precision. This process might cost 1/3 of normal prototype costs.
• There are a wide variety of techniques for creating cast metal parts and molds from STL resin parts [Ashly, 1994]. These include parts cast from SLA tooling directly. For example, SLA wax parts can be used to do investment casting.
32.1 Hyer, M.W., and Charette, R.F., “Use of Curvilinear Fiber Format in Composite Structure Design”, AIAA Journal, 1991, pg. 1011-15.
• Basically a loose powder is spread in a layer, and an bonding adhesive is selectively applied to harden a slice. Layers are continually added until one or more parts are completed.
• A trademarked name for this process is 3DP (3 Dimensional Printing)
• The general sequence is pictured below,
• The powders used by this process are starch based/cellulose powders.
• Parts can be colored using dyes
• The water based bonding adhesive is ???unknown???.
• The unbonded powder acts to support the part, and eliminates the need for other supports. This also allows multiple parts in a single build.
• After the part is complete the loose powder is removed. Powder can be easily removed from hollow and recessed cavities.
• Because the part is made of a bonded powder, the final part is porous. Higher part density can be obtained by impregnating parts with materials such as wax or epoxy. Parts may also be sanded for better surface finish.
• As expected there will be some dimensional shrinkage, this will be less than 1% for height and less than 1/2% on the build planes.
• A machine (Z402) is produced by Z-corporation, and the details are given below,
Build speed approx. 1.85 in/hr height for a 4” by 8” area
Maximum build volume 8” by 10” by 8”
Layer thickness 0.005” to 0.009”
Printing head is 0.36” wide and has 128 jets
Equipment size 29” by 36” by 42”
Consumable materials approx $0.65 per cubic inch of finished part
No special environmental requirements
An IBM compatible PC is required to run the machine
investment casting can be done from models
part material limited and not engineering materials
• Powdered material is fused together in layers using a laser
• The powders need fine grains and thermo-plastic properties so that it becomes viscous, flows, then solidifies quickly.
• invented in 1986 by Carl Deckard
• marketed by DTM corp. (Sinterstation 2000)
• The process uses a heated chamber (near the powder melting temperature)
• The product is split into slices from the .STL file and created one layer at a time by spreading layers of powder, sintering the powder with a CO2 laser, then adding new layers of powder and sintering until done.
• When done the part is inside a cake of powder, and putty knives and spatulas are used to remove the loose powder
• Supports not needed as the unsintered powder supports overhangs/etc.
• slow cooling of the parts can prevent distortion due to internal stresses.
• The laser is about 50W infrared (about 10000nm) This power level is much higher than stereolithography
• Optics and x-y scanner are similar to SL
• the process chamber runs hot to decrease the power required from the laser, and reduce thermal shrinkage that would be caused by a difference in operation and cooling temperatures.
• The hot chamber is filled with nitrogen (98% approx.) to reduce oxidation of the powder.
• rate of production is about 0.5-1” per hour
wide varieties of materials: wax for investment casting; polymers/nylon for assembly prototypes
reduced distortion from stresses
rough surface finish (“stair step effect”)
the first layers may require a base anchor to reduce thermal effects (e.g. curl)
material changes require cleaning of machine
• DTM markets the Sinterstation 2000 for $250,000(US) to $497,000(US) depending upon the selection of 1, 2, or 3 materials (investment casting wax, nylon, or polycarbonate). The Sinterstation 2500 starts at $400,000
• Development is being done on,
high power lasers for metal powders/etc.
• Selected specifications for a Sinterstation 2000 are given below,
1. A computer program preprocesses a part so that it is in sliced layers.
2. A plate (glass?) is charged selectively and coated with a back powder. This process is much like photocopying.The result is a photographic mask of clear and opaque areas for a single slice of the part.
3. A thin layer of photopolymer is spread in a part vat.
4. The mask is placed over the photopolymer and a UV lamp is used to expose the layer and selectively harden the polymer.
5. The photographic plate with the mask is cleaned.
6. The unhardened polymer is removed from the surface.
7. A find layer of wax is deposited and hardened.
8. The surface is milled flat for uniform thickness.
9. The process begins again at step 2 and continues until all of the layers have been added. Note: Some steps can be done concurrently for the mask and the vat (i.e., 2,5 AND 3,6,7,8) to decrease build times.
• Developed by Cubital Inc. in Israel, started in 1987.
• two commercial machines: Solider 4600 and 5600
• Uses photosensitive polymers, but these are developed using a UV light and a photopolymer
1. photopolymer is developed and hardened by a UV mask that has the pattern for one slice of the part.
2. Unhardened polymer is cleaned away and replaced with wax, that is solidified with a cooling plate.
3. the polymer/wax layer is machined to exact thickness, and coated with a new layer of polymer. (a vacuum is used to remove cut chips)
4. The process continues until done
• The masks are made using a glass plate with electrostatic powder distribution (similar to photocopiers). A slice is used to electrostatically charge a glass plate, electrostatic sensitive powder coats the charged areas, and the mask is complete. After use the glass is cleaned and reused.
• After completion the wax is melted, and the complete part remains. (the wax was used to support work and eliminate supports.)
************ Include SGC figures from pg 60 and 64
• the UV lamp is 4KW and is exposed to the polymer for a few seconds
• a resin applicator spreads the photopolymer across in thin layers
• an aerodynamic wiper is used to remove excess material to a storage reservoir. This material may be reused if not overexposed (?) thus causing a change in viscosity.
• because the toxic resins are used, exhaust fans and dark work cabinets are required.
no need for time consuming post-curing
part complexity does not effect speed, however volume does.
elimination of postcuring reduces internal stresses, and warping.
jobs can be stopped, other jobs run, then the first job restarted at a later time.
weights may be inserted at any time to alter the center of gravity
models with moving parts can be produced because of the firm holding of work in the process.
layers can be milled off if they are found to be in error
many parts can be run at the same time
overexposure of the polymer may increase the viscosity, and make it unusable, thus greatly increasing the volume of expensive polymers used.
the resins require that light sealed chambers and toxic material handling procedures be used.
maintenance is high, requires supervision
removal of wax after production is required
65 (5600) or 120 (4600) seconds per layer
14” by 14” by 14” (4600) or 20” by 14” by 20” (5600) work vol.
$275,000US (4600), $400,000US (5600)
has been used to produce investment casting
• A selected set of specifications for the Solider 4600 are given below,
• A selected set of specifications for the Solider 5600 are given below,
• Developed by Scott Crump, and Stratasys has been selling the machine since 1991.
• The concept is that material is heated and then in controlled quantities deposited directly on previous layers. Eventually layers are built up to complete the entire part.
• The materials are available on spools of 1/2 mile in length, at costs from $175(US) to $260(US). The filaments are 0.05”
• As usual the .STL file is sliced into layers, and the slices are used to drive the machine.
• The key to this method is an extrusion head,
the material is fed into the head
the material is heated until melting
the material is then extruded from the tip in controlled quantities
the material is wiped on the previous layer
• The extrusion head is moved about the table with an x-y positioning system to deposit material on each layer
• The platform the part is on drops when a layer is complete to allow the addition of a new layer.
• slice thickness is 0.002” to 0.03”
• material changeover requires a few minutes of “flushing-out’
a good variety of materials available
tolerance of +/- 0.005” overall
the extrusion head must continue moving, or else material bumps up
part strength is weak perpendicular to build axis.
more area in slices requires longer build times
temperature fluctuations during production could lead to delamination
• selected specifications for the FDM1000 are,
• Invented by Michael Feygin 1985, marketed by Helisys as LOM 1015 and LOM 2030. Also manufactured by Paradigm and Sparx AB as HotPlot
• uses thin sheets of material (most notably paper and polystyrene) that has a heat activated adhesive on one side. Sheets are piled up one at a time, and heat is used to melt sheets together. A laser then cuts the sheet into thin sections that form the slice.
• Slice thickness depends on material and ranges from 0.002” to 0.02”. Materials in use are,
• The laser uses the typical x-y and optics systems for the laser
• More than one layer can be cut at once, but the accuracy decreases as the number of layers increases.
• As material is cut, it is not removed. Material that is to be discarded is cut into “tiles”. There are chunks of material that will support the part, and are easily removed to recover the part.
• When complete the part is in the middle of a block. Outside there is a “wall” to support the tiles, and in turn the tiles support the parts.
• A heated roller compresses the laminate to the other layers. The thickness is harder to control, so the height of the material is measured each time to ensure accuracy.
• The final part requires careful removal from the tiles, and is finally sealed to keep moisture out, and prevent layer separation.
***************** Include LOM process photos
• The system uses a CO2 laser, and the cuts are done at varied powers and speeds. Note that these cuts are not done in raster lines such as other techniques
• The other laser positioning systems operate differently, in this case the mirrors are rotated, which is better suited to drawing vectors, as opposed to rasters.
no chemical changes, and minimal heating, so the shrinkage is trivial, and stress induced deformation is very small.
no “developing/heating time” is required
the laser only has to cut the part outline and hatching, not all the internal area.
a large variety of materials can be used: butcher paper is $2/lb for 0.004” thickness
the system is inexpensive to maintain
these machines are well suited to desktop operation
removal of the tiles can be difficult because the laser cuts through the layers, not between them. This requires schemes to weaken material layers that are at a solid/air interface. Cross hatching is used to “burn-out” and weaken materials
delicate parts can be damaged when removing tiles.
enclosed volumes will trap the support material
the material properties change with the direction of the laminate
a great percentage of the material is wasted
machinability is limited because of delamination
ventilation is required for fumes when burning
• The machine costs are itemized below,
• Accuracy is +/- 0.002” in x-y and +/- 0.001” in z
• time required is hours to days
• wax can be used to add fillets
1. build polyurethane molds from the LOM model
2. build polyurethane, or epoxy pattern equipment
4. cast many metal parts (tolerances were in the range of 1/5 to 3/100”)
1. build rubber, epoxy, or polyurethane plastic molds from LOM model
2. pour rubber pattern from mold
4. cast up to 100 parts (tolerances in the range of +/- .01” to .02”)
• Investment Casting (one shot only)
2. Develop a mold using ceramic slurry
3. place in autoclave to cure ceramic shell
4. Burn out LOM paper in oven and remove ash
1. make RTV silicone or epoxy mold of LOM part
3. make ceramic mold out of part
4. use autoclave to melt wax and cure shell
5. put in oven and burn out wax.
• making RTV Silicone Rubber Molds
1. Gating is added to the LOM model
2. the model is placed in a box and RTV silicone is poured in to cover the model
3. the silicone is degassed and cured
4. the silicone mold is split, and the part is remove.
5. 10 to 30 polyurethane parts can be cast in the mold
• Selected specifications are given for a LOM 1015,
• Selected specifications are given for a LOM 2030,
• Invented by Emanuel Sachs 1989 at MIT
1. layer of powder is deposited, spread, and compressed on a pallet.
2. The material for the slice is fused using a print head that moves in a raster and sprays adhesive in required spots.
• the unfused powder is not removed, and thus supports the rest of the part
• when complete the powder is removed and reused
• the result is a shell that can be used in casting. Therefore these parts often include the gating required for the metal flow.
produces good castings directly
the variety of usable common powders is large (using about 320 grit)
allows tests using metal parts for strength and fit
eliminates costly time consuming intermediate stages to casting
can produce very complicated molds
the mold can be removed from cavities after molding by using a caustic bath. (the rest is simply smashed off)
many parts can be made at once
it is faster to spray adhesive than fuse/cut with laser
final materials only limited by casting
rough surface finish: details down to 0.175mm; tolerance +/- 0.05mm
unbound powder can clog in hidden cavities
the printing jet tends to clog.
not commercially available yet
• work volume is 8” by 12” by 8”
• resolution of print head 0.007”
• cost for alpha machine $200,000US
• Expected machine in 1994 is,
• Sprays material (wax) in 0.002” drops at rates of 12,500 drops per sec to build up slices
• The elevator drops as slices are formed
• Variable slice thickness is set by changing the flow rate
• Part material supports are made from water soluble wax (polyethelene glycol) and are removed after completion by placing the model in water
• The BPM personal modeler is $35,000
• Incremental fabrication is a ballistic particle method developed by Incre Inc. but molten metal is used instead.
32.9.1 Sanders Prototype
• This methods uses two thermoplastic materials the positive having a higher melting temperature. The materials are distributed by a head that will melt and deposit either material (much like an ink jet printer head doing multiple colors). A raster scan is used to build up layers until the final composite part is done. The lower temperature material is melted away to leave the inner part.
• This method is very good for small parts, and produces parts in engineering materials.
• There are two commercially available units,
32.9.2 Design Controlled Automated Fabrication (DESCAF)
• Invented in 1986 by Efrem Fudim
• Marketed by Light Sculpting Inc.
• Uses photomasks of layers to develop sections.
• Exposes the parts to UV light, and develops the photomask
• An early case study [Rapid Prototyping Report, 1992b] revealed part costs were generally $88 to $344 for a 1.5 by 1.5 by 3 in. speedometer adaptor. The figures are given below, and dollar figures are based on Chrysler labor (Starts with an STL file). The SGC method assumes 35 parts are made at one time, but all figures quoted are for a single part.
• Costs for machines from other vendors are listed below. Most of these are Stereolithographt units not available in the US because of patents.
• My first exposure to rapid prototyping was from an early report written by Leo Matteo when he was an undergraduate student at Ryerson Polytechnic University
32.2 Aronson, R.B., “So Rapid Prototyping Works, Now What?”, Manufacturing Engineering, Nov., 1993, pp. 37-42.
32.3 Ashley, S., “Special Report: Rapid Prototyping Systems”, Mechanical Engineering, April, 1991, pp. 34-43.
32.4 Ashley, S., “New Material Introduced for Rapid Prototyping Systems”, Mechanical Engineering, Feb., 1992, pp. 16.
32.5 Ashley, S., “Prototyping With advanced Tools”, Mechanical Engineering, June 1994, pp. 48-55.
32.6 Brown, A.S., “Rapid Prototyping: Parts Without Tools”, Aerospace America, Aug., 1991, pp. 18-23.
32.7 Burns, M., “automated Fabrication;”, Prentice Hall, 1993.
32.8 Cabriele, M.C., “More Rapid Prototyping Systems Reach Commercialization”, Plastics Technology, June 1991, pp. 45-48.
32.9 Crump, S.S., “Rapid Prototyping Using FDM”, Modern Casting, April 1992, pp. 36-37.
32.10 Doyle, L.E., Keyser, C.A., Leach, J.L., Schrader, G.F., Singer, m.B., Manufacturing Processes and Materials for Engineers, 3rd. ed., Prentice Hall, 1985.
32.11 Jacobs, P.F., Rapid Prototyping and Manufacturing: Fundamentals of Stereolithography, Society of Manufacturing Engineers, 1992.
32.12 Lindsay, K.F., “Rapid Prototyping Shapes Up As Low-Cost Modeling Alternative”, Modern Plastics, Aug., 1990, pp. 40-43.
32.13 Machine Design, “Solid Prototypes Produced Without Postcure”, Machine Design, Jan., 24th, 1991, pp. 30-31.
32.14 Mateo, L. Rapid Prototyping????, A Bachelors thesis submitted to the Department of Mechanical Engineering, Ryerson Polytechnic University, Toronto, Ontario, Canada, 1994.
32.15 Mechanical Engineering, “Rapid Prototyping Includes Moving Parts”, Mechanical Engineering, April 1992, pp. 12.
32.16 Mechanical Engineering, “Rapid Prototyping for Artificial Body Parts”, Mechanical Engineering, May 1993, pp. 50-53.
32.17 Miller, J.F., Rapid Prototyping Overview: An Introduction to Systems and Technology, Society of Manufacturing Engineers, 1993.
32.18 Montague, R.M., “Rapid Prototyping Offers 3D Model Design”, Industrial Engineering, Oct., 1991, pp. 20.
32.19 Muller, T. and Bex, T., “Rapid Prototyping Draws Widening Foundry Interest”, Modern Casting, Nov., 1991, pp. 37-41.
32.20 Rapid Prototyping Report, “Stereolithography Produces Functional Impeller for Testing”, an article appearing in Rapid Prototyping Report, CAD/CAM Publishing Inc., San Diego CA, Vol. 1, No. 6, June 1992a).
32.21 Rapid Prototyping Report, “Chrysler Benchmarks Rapid Prototyping Systems”, an article appearing in Rapid Prototyping Report, CAD/CAM Publishing Inc., San Diego CA, Vol. 1, No. 6, June 1992b).
32.22 Sachs, E., Cima, M., Williams, P., Brancazio, D., and Cornie, J., “3D Printing: Rapid Tooling and Prototypes Directly From a CAD Model”, Journal of Engineering for Industry, No. 114, pp. 481-488.
32.23 Sprow, E.E., “Rapid Prototyping: Beyond The Wet Look”, Manufacturing Engineering, Nov., 1992, pp. 37-42.
32.24 Stovicek, D.R., “Rapid Prototyping Slices Time-to-market”, Automation, Sept., 1991, pp. 20-23.
32.25 Warner, M.C., Rapid Prototyping Applications: Fro Rapid Prototyping to Functional Metal and Plastic Parts, Mack Industries Inc., Troy, MI, 1993.
32.26 Wohlers, T., “Make Fiction Fact Fast”, Manufacturing Engineering, Mar., 1991, pp. 44-49.
Problem 32.1 Indicate why (one good and one bad point) the following rapid prototyping approaches are well/poorly suited to the part specified.
a) Stereolithography for a hollow ball
b) Selective laser sintering for a door hinge
c) Solid ground curing for a light bulb
d) Fused deposition modeling for a hollow box
e) Laminate object modeling for a complete model of a pyramid with the pharaohs tomb inside
Problem 32.2 In general what common problems are faced by all rapid prototyping techniques?
Problem 32.3 a) Why would a full size engine block be difficult to produce with modern rapid prototyping methods? b) How could the problems in #2a) be overcome by changing the RP technologies?
Problem 32.4 Describe the process of a) Selective Laser Sintering, b) Stereolithography.
Problem 32.5 Sketch a prototype that could be made using Selective Laser Sintering.