August 1999
Technical Paper for presentation at the Conference on
Pattern and Die Manufacturing Technology, Pune, October 7-8, 1999
Computer-Aided Design of Tooling
for Casting Process
B. Ravi, Associate Professor
Department of Mechanical Engineering
Indian Institute of Technology, Bombay 400076, India
M. M. Akarte, Lecturer
Production Department
SGGS college of Engineering & Technology, Nanded
Abstract
Tooling development is an important activity bridging product design and manufacturing
activities, and is often a bottleneck in new product development. Ever-shrinking product
life cycles have made it necessary to explore, adopt and adapt new technologies to
produce quality tooling in a shorter time while remaining competitive. This is especially
true of metal casting sector, where long lead times (weeks to months) for producing the
first article of approval are no longer acceptable.
This paper shows how to compress the tooling development time for metal casting using
an intelligent integrated approach. This is based on combining three ‘young’ but proven
technologies: (1) solid modeling of part design, (2) knowledge-based tooling design and
(3) rapid prototyping and tooling. With this approach, it has been possible to produce
pattern and coreboxes (for small complex parts required in low quantity), within a
working week, and costs slightly higher than high-speed NC milling. As the technologies
become more mature and user-friendly, it is foreseen that well within a decade, they will
become more economically viable and might largely replace the current approach to
produce the tooling for casting. Tooling engineers need to be proactive and start
exploring the new technologies: it is better to be prepared in advance!
Keywords: casting, tooling design, pattern, mold, die, rapid prototyping, CAD/CAM.
* Author for Correspondence. Prof. B. Ravi is In-Charge of Casting Simulation Lab, Co-Founder of Rapid Prototyping
Cell and a guest faculty in the School of Management. He can be contacted at bravi@[Link]
“You see, it takes all the running you can do, to keep in the same place.
If you want to get somewhere else, you must run at least twice as fast.”
- From ‘Alice in Wonderland’ by Lewis Carroll.
1. Survival of the Fastest
In an age when innovation or change is the only constant, time has emerged as the single
most important factor for competitive success. Rising customer demands and fierce
global competition are forcing engineering and consumer product companies to
continuously seek ways to compress the lead-time from concept to market – in the case of
new products, this could mean the difference between assured growth and doubtful
survival. The automobile sector – where the above scenario has unfolded in a global
context – has witnessed slower players being replaced by those who have been proactive
in exploring, adopting and adapting new technologies for slashing the development time.
Indeed, the lead-time to introduce a new automobile (from concept to market), which was
more than 60 months a decade back, has reduced to less than 18 months today.
However, product designers generally delay the freezing of the design to the extent
possible to accommodate the feedback from all departments, who in turn have to take
into account the latest technologies and trends. This includes industrial designers, who
have to visualize which designs will become ‘fashionable’ a few months later and remain
‘acceptable’ for another few years. Since component designs can not be completely
finalized until manufacturing and testing, this naturally puts the pressure on all
downstream activities (after product design) to be compressed to the logical minimum.
Thus tooling development, which is a prerequisite to manufacturing, becomes the most
critical activity since it is not only consumes expensive resources, but also can become a
bottleneck and potentially throw the entire project out of gear.
The above is especially true in the case of cast components, tooling for which takes
weeks to months to develop. Indeed, in a recent benchmarking survey of American
foundries, in which the first author was involved, it was found that the average lead time
for the first article of approval is 10-14 weeks, of which tooling development accounted
for nearly 70% of the total lead time [Creese,1996].
Over the last ten years, researchers have developed and improved several technologies
which are gradually being accepted by the industry, and are of direct interest to tooling
engineers. This paper describes three of those technologies in the context of the metal
casting sector: (1) solid modeling of cast components, (2) knowledge-based casting
design and (3) rapid prototyping and tooling of patterns and dies.
2. Solid Modeling of Cast Components
The casting geometry conceived by the product designer is usually communicated to the
tooling engineer through engineering drawings in terms of orthographic and sectional
views. These drawings have to be interpreted for mental reconstruction of the three
dimensional product before proceeding with the design of pattern and core boxes. It is a
challenging task, particularly in case of intricate castings, and may take several hours for
even experienced engineers. The tooling design is communicated to pattern makers and
foundry engineers through another set of drawings. Trial runs are made to validate the
design, and any changes are recorded. If the tool design is modified, (in some difficult-to-
cast cases, even product design has to be modified), a fresh set of drawings are prepared.
This is a time-consuming exercise and prone to errors.
Replacing traditional drafting practice with computer-aided drafting has greatly reduced
the time taken to produce drawings of modified designs. Advances such as automatic
dimensioning, bill of material generation and customized title block generation have
further increased the productivity of designers.
However, 3D CAD or solid modeling represents a giant leap forward. A solid model is
not only useful for visualizing the part features, but also is the main input for an array of
CAD/CAM/CAE packages available today. These programs help in automatic generation
of orthographic and sectional views, stress analysis, weight calculation, design of pattern,
mold, cores, feeding and gating, simulation of flow and solidification of molten metal in
mold, NC cutter path generation for tool manufacture, materials planning, cost estimation
and computer-controlled inspection.
A number of solid modeling program are available today, including AutoCAD R14,
CATIA, DUCT, Euklid, I-DEAS, Pro-Engineer, SolidWorks, Unigraphics, etc. In these
systems, solid models are created using Boolean operations: union, intersection and
difference on pairs of simpler models to create the desired shape. A library of solid
primitives such as cube, cylinder, sphere, cone and torus are provided to initiate the
modeling. The model created by combining primitive solids is then combined with other
primitives or other solids to eventually obtain the designed shape. The initial model can
also be created by sweeping a 2-dimensional section through a straight or curved path to
produce solids of revolution and extruded shapes. Advanced sweep-based modelers
handle contoured paths along B-spline curves and change in section shape along the path.
Complex contoured shapes are defined using Coons, Bezier, B-spline or non-uniform
rational B-spline (NURBS) surface patches.
The majority of castings, which are quite complex when compared to machined
components, require a combination of all the above techniques. The designer requires
training and experience in deciding the strategy for modeling, in particular, combinations
of primitive solids, which will lead to the final shape with minimal number of steps.
Sometimes it becomes necessary to retrace the steps during modeling and take a different
approach to complete the shape.
In summary, the ease of visualization, time-saving in subsequent modifications and the
benefits of using the model for other applications far outweigh the initial effort in
creating the solid model. Thus, while the first model may take 10-100 hours to create
(depending on complexity), any minor subsequent changes can be done in minutes.
Automatic and accurate weight calculation takes just a few seconds!
The next section describes how the solid model of the part can be used for designing the
casting and creating the pattern model.
2. Knowledge-Based Casting Design
In the context of this paper, we will refer to casting design as the activity (as well as the
result) of converting the part model into the cavity shape which is filled by molten metal
during the casting process. This includes the following decisions and design activities:
• selecting the best orientation of the part in the mold,
• determining the parting line,
• identifying features that have to be produced by cores,
• design of cores (including their supports or core-prints),
• design of core boxes to produce the cores,
• design of risers to provide feed metal (number, location, shape, dimensions),
• design of gating system to lead molten metal into the mold,
• layout of cavities in the mold,
• design of match-plate pattern or die (including applying allowances),
• other systems (mainly for die casting: guidance, cooling, ejection, etc.).
These decisions require an in-depth knowledge of the metal-specific casting process and
considerable experience in casting design for a range of applications. During the last two
decades, many researchers and some enthusiastic foundry engineers have attempted to
capture the relevant knowledge in rules and empirical equations, which can be coded into
a compuer program. Typically, the user enters the type of cast metal, part weight, section
modulus, average section thickness, type of risers and gating system, their location, etc.,
and the program calculates the dimensions of the riser and gating channels based on.
These values are useful for creating a casting model (with risers and gating) in a solid
modeling system.
A few advanced programs for simulating the casting process are also available today
[AFS,1997]. These programs help in predicting the location and intensity of major
casting defects such as shrinkage cavities, porosity, air entrapment, erosion and cold
shuts. Thus foundry engineers can check several different alternatives for feeding and
gating design and finalize the optimal layout without pouring a single casting. However,
the engineer still requires experience for taking the key decisions, setting up the
simulation run, interpreting the results of simulation and finalizing the design. The
programs for designing various elements of a casting as well as programs for modeling
and simulation programs are all different (developed and offered by different companies).
Thus the user has the additional burden of managing these programs and switching
between them; also there is a high possibility of errors when transferring the data
(manually or electronically) between the programs.
We propose an intelligent integrated approach to enable better and faster decision making
during casting design. Intelligent CAD software is distinguished from conventional CAD
by its capability to perform tasks which require domain knowledge and geometric
reasoning. (Expert Systems, which also contain domain knowledge and an inference
engine, can not handle geometry). Geometric reasoning is required for automatic
recognition of casting features (such as cored holes from a solid model of the part). The
domain knowledge is required for assessing the features recognized (for example,
determining whether the feature has to be produced by a core, designing the core print
and evaluating the entire core in terms of strength, venting, cooling and other criteria).
Intelligent CAD software will thus act like a manufacturing expert, on call anytime, to
assist product engineers in design for manufacture; tooling and casting engineers can
verify their decisions and ensure that a feasible design alternative has not been
overlooked. Such an intelligent assistant to casting designers – called AutoCAST – has
been developed [Ravi,1999]. IT simulates the way expert engineers design various
elements of a casting such as parting line, core, mold, feeders and gating (Fig.1). All
programs of this system are linked through an in-built casting project database
management system. These are outlined below.
Product: This program helps in importing a solid model of the part from an external solid
modeler through a standard exchange format, followed by computation of geometric
properties. Wall thickness, section variation, complexity, holes and other criteria are used
for castability analysis, followed by suggestions for design improvement.
Parting: This program suggests the best orientation of the casting in the mold, generates
several parting lines and determines the best alternative, aligns the parting line with the
mold parting plane, and finally analyzes the current parting in terms of flatness, draw
distance, draft volume, dimensional stability, etc. to suggest improvements.
Core: This program first identifies cored features in the part model: through holes, deep
pockets and undercuts. For each cored feature, it designs the core print and creates a solid
model of the entire core. Finally, it analyzes the cored feature for failure, venting and
other criteria, based on which guidelines for design improvement are presented.
Mold: This program selects the most appropriate mold box to enclose the casting,
determines the optimal number of cavities and displays the cavity layout. The mold
design is analyzed in terms of metal to sand ratio, cavity shape and other criteria, based
on which guidelines for design improvement are displayed.
Feeding: This program first simulates the solidification of the casting to determine the
location of hot spots and suggests an appropriate location for the feeder. It also calculates
the feeder dimensions, creates its solid model and attaches it to the casting. The feeding
design is further verified by progressive solidification plots on a section and directional
solidification vectors (feed metal flow paths) inside the casting. Yield, feeding efficiency
and ease of fettling are computed to compare different layouts.
Gating: This program suggests the connection points for ingates and the location of sprue,
followed by the layout of runners. It determines the optimal pouring time, designs the
entire gating system and creates its solid model. The actual filling time is estimated to
correct the gating design if necessary. The gating design is analyzed in terms of yield,
ease of fettling and other criteria based on which suggestions for design improvement are
presented.
Process: This program suggests an appropriate casting process for producing the part
(given its metal, weight, lot size, quality specifications, etc.). It plans the requirement of
cast metal, mold sand, core sand and other materials. An activity-based approach is used
for analyzing the lead time and costs for producing the casting.
Quality: This program helps in casting inspection in two ways. It assists in setting up
reference dimensions between specified locations on the casting surface and determines
the ideal distance between them. It also simulates radiography to produce the radiograph
of a defect-free casting, which can be compared with actual radiographs to identify
internal defects in the casting.
The knowledge-based approach to casting design dramatically reduces the total time for
optimizing a casting. For example, 6-10 iterations may be required for optimizing the
feeder design alone for a small-to-medium complex part, and the total time for feeder
design-model-simulate using separate programs may take 10-15 days (1.5 to 2 days per
iteration). For optimizing the entire casting design (including orientation, gating, etc.),
more iterations, and hence more time will be needed. An intelligent program like
AutoCAST not only reduces the iteration time, but also cuts down the number of
iterations required (by giving good first solutions), with the result that a casting design
can be optimized within a single day!
After the casting design is optimized, the pattern model can be generated using a solid
modeling program. This involves splitting across the parting line, removing the holes,
adding core prints, applying draft, fillets and various allowances. Some solid modeling
programs have semi-automatic facilities to speed up the above steps. The pattern model is
needed for the next step: actual fabrication of the tooling.
3. Rapid Prototyping and Tooling
Rapid prototyping (RP) techniques enable complex shaped parts to be produced directly
from a computer model within a few hours: no part-specific tooling or machining is
required. These models can be directly used as patterns for casting or can be used for
producing the tooling through rapid tooling (RT) processes. While RP techniques were
initially intended for creating prototypes of complex shaped products (to verify their
form, fit and to some extent, their function), in recent years these techniques have been
combined with RT techniques by several foundries, tool rooms and service bureaus. This
has enabled significant reductions in the lead-time to manufacture cast products. The two
technologies (RP and RT) are outlined below, and the next section will describe how
these can be applied to the metal casting sector.
3.1 Rapid Prototyping Techniques
The Rapid Prototyping (RP) technology is based on the philosophy of fabricating the
cross-sectional layers on top of each other to produce a part. The sections are generated
from a solid model of the part created on a computer and then fabricated using one of the
several RP techniques available. These include selective liquid solidification, liquid
deposition, powder binding and sheet laminating [Xue,1996]. The main RP techniques
currently available in the industry include stereolithography (SLA), solid ground curing
(SGC), fused deposition modeling (FDM), laminated object manufacturing (LOM) and
selective laser sintering (SLS). These are briefly described here (also see Fig.2).
Stereolithography uses an ultraviolet laser beam moving in a criss-cross fashion to cure
photo-curable polymer resin contained in a vat. The polymer layer is lowered by a
platform attached to it to enable generating the next layer on top.
Solid Ground Curing is similar to stereolithography; the difference is that the entire layer
of polymer within the specified boundary is cured by a flood of ultraviolet light passing
through a glass mask containing a negative image of the cross-section.
Fused Deposition Modeling technique relies on melting and depositing a thin filament of
thermoplastic polymer to form each layer. A separate head deposits the support material
in each layer, which can be broken off later.
Laminated Object Manufacturing involves laser beam cutting of cross-section contours
out of sheets of heat sensitive or polymer coated paper; the adjacent layers are joined by
heating and compression by a roller.
Selective Laser Sintering process uses a high power laser beam to melt thermoplastic
powder spread on a layer. A roller spreads the next layer of powder on the previous layer.
The unsintered powder serves the function of supports for undercuts.
Table 1 provides a summary of the characteristics of the above RP techniques.
3.2 Rapid Tooling Techniques
Rapid tooling can be considered as secondary operations, which produce a negative
replica or mold from a master (the master can be produced by Rapid Prototyping
processes). The mold can be used to obtain a replica of the original model. Different
processes used in secondary operations can thus produce the desired tooling through
several routes (Fig.3). Some of the routes are useful for creating one-off parts, others are
useful for small, medium or large batch products. The most widely used materials for
producing rapid tooling include thermosetting polymers: epoxy resins, polyurethanes
(elastomers and foams) and silicones. Metals, especially those with low melting point are
also used in some processes. Some of the important operations employed for rapid
tooling are outlined here: epoxy resin casting, laminated shell molding, silicone rubber
molding, polyurethane casting, metal arc spraying and investment casting [Reg,1999].
Epoxy Resin Casting process essentially involves mixing a suitable epoxy resin with a
hardener and pouring it into a mold or onto the face of the model placed in a box. A
parting agent applied to the face of the mold facilitates easy release of the fabricated
model.
Laminated Shell Molding involves creating a laminated shell around a master model
using alternate layers of gel, epoxy resin and glass cloth. While it is labor intensive and
time consuming, it couples light weight with good strength, suitable for large flat parts.
Silicone Rubber Molding involves pouring RTV silicone rubber around a master model
and cutting open the mold after curing. These molds have excellent chemical resistance,
low shrinkage and high dimensional stability, suitable for producing parts in polyester,
epoxy and polyurethane foam by injection molding.
Polyurethane Casting involves pouring the resin around a master placed in a box, similar
to epoxy casting. The material is more expensive than epoxy, but sets faster and can
produce better detail.
Metal Arc Spraying process uses a high velocity electric arc metal spray generating
system to deposit finely atomized particles of molten metal (usually kirksite) onto a
model surface to create a metal shell mold. The metal shell can be reinforced by epoxy.
Investment Casting requires an expendable pattern (usually of wax) which is coated with
layers of silica slurry to obtain a shell, which is heated to remove the wax. Molten metal
is poured in the shell and allowed to solidify to give a metal replica of the wax model.
4. Rapid Tooling for Casting
The range of RP and RT techniques available today provides several routes for producing
the tooling for metal casting. Some of these are useful for creating one-off parts or master
patterns, others are useful for creating patterns and coreboxes that can be used for short,
medium or long runs. Indeed, RPT technologists are continuously opening up new
materials and routes for producing the tooling (for various casting processes): polymer
patterns, investment patterns, polyurethane foam patterns, metal patterns, pattern plates
and coreboxes. Various rapid prototyping and tooling techniques suitable for fabricating
different types of patterns are briefly described here [Karunakaran,1998].
Patterns: Models fabricated by a few RP systems can be directly used as casting patterns
for low to medium size batches. The LOM process has been the most widely used for this
purpose, since it produces wood-like models. The low cost of the fabrication material and
the fast build rate makes the LOM process economical for large size patterns, where
dimensional accuracy is not very critical. These patterns are however, required to be
initially and periodically coated with lacquer to maintain a smooth surface finish and
have a longer life. The ABS models produced by the FDM process and some polymer
models produced by SLA, SGC and BPM processes can also be used as casting patterns.
These are more suited to small, intricate components requiring good dimensional
accuracy and surface finish. All these patterns are lightweight, have adequate strength,
produce good surface finish, are easy to clean after molding and are economical for small
to medium runs. The plastic patterns can also be hand finished to obtain a smooth surface
and can be coated with resins, both initially and periodically, to enhance their life. The
SLS process can directly produce silica shell molds into which metal can be poured to
produce metal patterns.
Investment Patterns: RP processes can directly produce models in wax or a polymer
which can be used in the investment casting process. Most RP manufacturers have
recognized the importance of addressing this requirement and offer special materials and
build techniques to produce investment patterns. The FDM machine comes with a special
head to produce wax models. The SLA equipment now supports a photocurable resin
suitable for investment casting and uses a proprietary QuickCast build technique to
produce patterns with hollow honeycomb-type walls ensuring better collapsibility during
burnout. Even LOM models can be invested, though they leave a high ash residue. The
SLS process can produce metal molds, which can be used for producing investment
casting wax patterns. Wax and polyurethane foam patterns can also be produced by
injecting wax/resin in a silicone rubber or polyurethane mold produced by the rapid
tooling techniques.
Pattern Plates and Coreboxes: Pattern plates and coreboxes can be produced by making a
negative replica of the corresponding master model using epoxy resin. For producing a
pattern plate, a negative (mold) is first produced by pouring epoxy or polyurethane resin
around a master model placed in a box. After curing both sides of the mold, the pattern is
then cast by pouring the resin in the epoxy mold. The epoxy patterns can be reinforced
with glass cloth during curing or enclosed in a metal frame. No reinforcement is required
if the patterns are to be used on automatic high pressure molding equipment, since there
is no jolting action as in conventional molding machines. For producing a corebox, a
model of the core is required. This is can be easily fabricated by extracting the
corresponding feature (hole) from the computer model of the component, reversing the
feature to obtain the corresponding solid and fabricating the model on an RP machine.
The corebox is then build around the core feature by either epoxy resin casting or
laminated shell molding. While such pattern plates and coreboxes are suitable for small to
medium runs, the ease of manufacture and repair makes them quite attractive, especially
for complex shapes.
Given the number of RP and RT processes, there are a large number of routes to produce
the tooling for casting processes. Each route has its own capabilities, strengths and
limitations, in terms of materials, quality characteristics and costs. Choosing the correct
route for a given set of requirements is therefore an important task. Once the route is
chosen, fabricating the actual tooling involves creating the master pattern or master mold
using RP, and then creating the master mold or regular pattern using RT. If the solid
model of the pattern is ready, the first step (RP) may take 12-60 hours depending on size,
and the second step (RT) may take 24-48 hours depending on the process. The patterns
can thus be fabricated within a working week!
5. Conclusion
This paper described how three new technologies: solid modeling, knowledge-based
casting design and rapid prototyping & tooling can help in compressing the total lead
time for producing the patterns for metal casting. All three require substantial initial
investment in terms of hardware, software and trained ‘humanware’. Equally important,
they require a change of culture from the one relying on cost-cutting measure to increase
profitability, to the one which uses time (in terms of shorter time as well as assured
delivery) to add value and gain competitive advantage. Already there are hundreds of
new-age tool-rooms in USA and Europe, which use these technologies to deliver tooling
to their customers in days instead of weeks. As the technologies mature, the costs are
likely to reduce to an extent that these routes become economic enough to replace
conventional pattern-making techniques. Pattern and tool-makers need to be aware of
these new developments taking place, start building up their knowledge and experience in
this area, and take the critical step of transition before it is too late.
6. References
AFS, “Sources of Casting Modeling Software,” Modern Casting, Vol.87, No.10, 1997,
pp.27-29.
Creese RC, “Benchmarking and Lead Time Reduction,” AMC Benchmarking Project
Report, West Virginia University, Oct 1996, pp.1-137.
Reg G, "Rapid prototyping: A tool for casting design and verification," Modern Casting,
pp. 44-47 (March 1999).
Kruth J P, "Material incress manufacturing by Rapid Prototyping Techniques," Annals of
CIRP, Vol. 41, No. 2, pp. 603-614 (1991).
Xue Y and Gu P, "A review of rapid prototyping technologies and systems," Computer
Aided Design, Vol. 28, No. 4, pp. 307-318 (1996).
Karunakaran, KP, Bapat VP and Ravi B “Rapid Prototyping and Tooling,” IDC Press,
IIT Bombay, 1998.
Ravi B, "Computer-Aided Casting - Past, Present and Future," Indian Foundry Journal,
45(1), 1999, 65-74.
Table 1: Comparison of Rapid Prototyping processes
[Kruth, 1991; Karunakaran 1998]
SLA SGC FDM LOM SLS
MATERIAL PHOTO- PHOTO- THERMOPLASTICS: PAPER, POWDERS:
POLYMERS POLYMERS INVESTMENT WAX, CELLULOSE THERMOPLAS
(UV): (UV): MACHINABLE WAX, PLASTICS, TICS (ABS,
ACRYLATES ACRYLATES NYLON LIKE METALS, FABRICS NYLON, PVC,
MATERIAL, WAX SYNTHETIC PC), WAX,
FILLED PLASTIC MATERIAL ANY METALS,
ADHESIVE SHEET PRE- POLYMER
COATED WITH COATED
HEAT ACTIVE CERAMICS
ADHESIVE
RAW FORM LIQUID LIQUID 1.25 mm WIRE SHEET, FOIL POWDER
BUILD SLA - 250: SOLIDER - 4600 FDM - 1650 LOM - 1015 SINTER
ENVELOP 250x250x250 350x350x350 254x254x254 380x250x350 STATION
(mm) SLA - 500: SOLIDER - 5600 FDM - 2000 LOM - 2030 300x380
508x508x584 500x350x500 254X254X254 810x550x500
BUILD RATE 100 365 5 500 20
(mm3/Sec)
ACCURACY(mm)
XY - PLANE 0.1 0.1 0.15 0.25 0.4
Z - PLANE 0.1 0.15 0.13 0.1 - 0.3 0.5
LAYER 0.1 - 0.7 0.05 - 0.15 0.025 - 1.25 0.01 - 0.15 0.08 - 0.13
THICKNESS(mm)
NEED FOR YES NO SOMETIMES NO NO
SUPPORT
POST YES, OVEN YES, MELTING NO NO NO, EXCEPT
CURING CURING OF WAX FOR COATED
CERAMICS
INITIAL 70,000 TO 295,000 TO 110, 000 TO 140,000 TO 366,000 TO
COSTS ($) 490,000 475,000 120,000 272,000 450,000
FLOOR SPACE 0.7x1.2x1.6 - 1.8x4.2x2.9 0.6x0.86x0.93 - 1.2x0.99x1.27 - 3.02x1.53x1.93
(meter) 1.8x1.2x2.0 0.6x0.92x1.07 2.08x1.47x1.4
COMPANY 3D SYSTEMS CUBITAL STRATASYS INC. HELISYS CORP. DTM CORP.
