Investment Casting Using Stereolithography: Case of Complex
In this paper we describe procedures and lessons learned in
the use of stereolithography for manufacturing objects using investment casting.
Procedures described are the results of manufacturing complex components used in
a unique structural framework. The complexity of the object is due to varying
radii and angles, hollowness, sockets of constant wall thickness where tubular
structural members are designed to mate, as well as sizes exceeding work
envelope. Complexity of the object required deviation from conventional
techniques used in investment casting. Rapid prototyping proved superior to
traditional methods both from an economic and technical standpoint, especially
where limited numbers are required.
Keywords: Investment casting, Rapid prototyping,
Time to market is an important business factor in todays
competitive environment. Management now monitors its consumption with the same
emphasis as sales and costs. In today's flexible manufacturing and rapid
response environments, there is an increasing requirement for variety and
innovation. There is a growing acceptance of rapid prototyping (RP) technology
in industry today in response to such requirement. Many firms have incorporated
it into the mainstream of their product development process. From the inception
of rapid prototyping, its very function was directed at design verification. RP
technology has moved quickly toward using RP models in support of
The investment casting (IC) industry, a multi-million dollar
industry, is no exception to this new environment. IC as a process involves
creating a wax model for the object to be manufactured, dipping the model in a
ceramic slurry, firing the ceramic in an autoclave, which melts the wax and
leaves behind a cavity. The cavity is then used for casting the object in a
casting material. While the process yields high precision products, the ceramic
shell has to be destroyed to obtain the part. This in turn brings to focus the
question of repeatability when it comes to creating the wax pattern. The larger
issue concerns the generation of wax patterns for complex models, where the
system dynamics include manufacturability and the economics of an acceptable
time frame, especially when it concerns small production runs.
The objective of this research effort is to find a solution
to this problem in the IC industry by extending RP techniques towards the rapid
wax patterns generation. The manufacturing requirement that prompted this
research is the fabrication of a complex swing frame structure. The structure
had unique but complex geometry with varying radii and angles, hollowness,
sockets of constant wall thickness where tubular structural members are designed
to mate, as well as sizes exceeding work envelope.
Complexity of the objects made it impossible to use milling
in its fabrication. IC was selected as the method of fabrication, however cost
and time constraints in addition to the accuracy, repeatability, and the limited
number of units required deviation from conventional techniques used in IC.
Keeping this in mind, it was decided to prototype these parts using a
stereolithography apparatus (SLA-250). The SLA parts could then be used to
create the wax patterns for investment casting. The advantage of this approach
is the rapid pace at which the parts could be created using the CAD/CAM
technology and its support to the freeform designs. The process used here is an
innovative approach because the SL patterns are used to create wax patterns
instead of being used as a substitute for wax and subsequently burnt out, which
is the current trend in the technology. This process ensures the reusability of
the SL pattern for future requirements, and working off the same master pattern
improves the repeatability apart from the overall accuracy the process provides.
In this paper we describe the steps and procedure used in the
manufacturing process and we report on lessons learned in its conduct. First we
introduce a brief overview of investment casting and why some of the
conventional techniques presently available may not be the most viable option in
certain manufacturing applications. We then provide in some details current
trend in the use of rapid prototyping in the investment casting industry. We
followed by a description of the object(s) and the degree of its complexity and
we concentrate on the process used in the development of wax patterns using the
2. Overview of Investment Casting (IC)
IC is a process also known as the "lost-wax"
process, or "precision" casting. In this process a wax pattern must be
made for every casting and gating system; i.e., the pattern is expendable. This
process offers the end user good value for money where good surface finish,
complex geometry, and cast features are desirable without the necessity of
extensive machining or other fabrication/ finishing work required to provide a
usable end item. A number of processes exist, but they have the following points
in common: (1) Disposable or expendable patterns are used, (2) Molding is done
with a fluid aggregate or slurry, (3) The aggregate is hardened in contact with
the pattern, providing precise reproduction of the pattern., (4) The aggregate
is bonded with an inorganic ceramic binder, (5) The mold is heated to drive off
all gases, (6) Pouring is performed with the mold preheated to a controlled
temperature in order to pour thin sections which would not otherwise fill out.
The Stages of IC involves :
(1) the development of a die for casting the wax patterns. The metal die must
take allowance for shrinkage of both wax and later the metal casting. The wax is
injected under pressure into the mold at wax melting temperature. Polystyrene
plastics are also used, but require higher pressures and temperature than iron
or steel dies. Patterns and gating system must be assembled if cast separately.
(2) Pre-coating, where the wax assembly is dipped into a
slurry of a refractory coating material. A typical slurry consists of silica
flour suspended in ethyl silicate solution of suitable viscosity to produce a
uniform coating after drying. After dipping, the assembly is coated by
sprinkling it with silica sand and allow it to dry. Sometimes pre-coating is not
used, and the wax pattern is directly invested in the molding material. In this
case, the molding mixture must be vacuumed to remove air bubbles which may lodge
next to the pattern. The coated-wax assembly is next invested in the mold. This
is done by inverting the wax assembly on a table, surrounding it with a
paper-lined steel flask, and pouring the investment-molding mixture around the
pattern. The mold material settles by gravity and completely surrounds the
pattern as the work table is vibrated. The molds are then allowed to air-set.
(3) De-waxing and preheating, in which the wax is melted out
of the hardened mold by heating it in an inverted position. The wax may be
reclaimed and reused. Mold is pre-heated to a temperature desirable for pouring
the particular alloy and casting design. The burnout and preheating cycle
completely eliminate wax and gas-forming material from the mold.
(4) Pouring: when the mold is at a the desirable temperature,
the metal is gravity-poured into the sprue. Air pressure may then be applied to
the sprue to force-fill the mold cavity. Pouring is also done in a centrifuge to
fill out thin sections.
(5) Cleaning operations follow cooling of the casting.
3. Trend in the Use of Rapid Prototyping in IC
Historically, the reluctance of the designer or buyer of a
new part design to consider the IC process has been due to the lengthy lead
times required to build the initial tooling. Depending on part complexity, the
tool building process alone can take from 8 to 20 weeks. As a result many
potential investment casting users were driven away to other traditional metal
The advent of rapid prototyping has filled in this niche for
reducing the total turn around in IC. By using SL solid resin models as
expendable patterns for ICs without the need for tooling, the total turn around
time can be reduced to 3 to 4 weeks. The SLA part is used as a substitute for
the wax pattern part in the investment casting process. This eliminates the need
for low-production-run wax pattern tooling. Low quantity makes the IC process
less effective because of the high costs and long lead times for wax pattern
tooling. Thus, castings are typically not integrated into the system until the
full-scale production phase is implemented.
Recently 3D-Systems released their latest update relative to
the Investment casting world, called the QuickcastTM. It is a
software coupled with the use of new epoxy resin. This process directly uses the
resin model as an expendable pattern. The model is build from a
"honey-comp" structure preserving the outer shell of the model. Using
this structure provided savings in both material and time to build. The process
dictates that for every prototype casting a resin model has to be build. Despite
its advantages, QuickcastTM does not lend itself to every design or
quantity to be produced. Following are some of the problems associated with this
Problems with QuickcastTM and IC
The major problems encountered with QuickcastTM
patterns when used in investment casting were as follows:
1. The existence of numerous 'pinholes' resulting from
support removal on down-facing surfaces. If these pinholes are not detected and
filled with investment casting wax or cured resin, the slurry will penetrate
into the interior of the pattern causing an inclusion, and probably an inferior
casting that will later be rejected.
2. A wavy upfacing surface, with regular periodic
undulations corresponding to the 'sag' of the single upfacing skin layer as it
spans the equilateral triangle hatch structure of the pattern.
3. The need to manually drill or punch 'Vent Holes' and
'Drain Holes' into the pattern surface in order to drain the uncured liquid
resin from the pattern
4. Problems with incomplete pattern drainage. This is
especially problematic for the small equilateral triangles and rhombuses
resulting from the offset procedure used in building patterns. Also, simple
gravity drainage is often insufficient to overcome the flow retarding effects of
liquid resin capillary and viscous forces. Consequently, the drainage process
for Quickcast patterns is considered less than optimal.
5. Difficulty achieving a high 'void ratio,' (Rv) defined as
the fraction of a Quickcast pattern's volume is ultimately filled with air
subsequent to draining the uncured liquid resin and post-curing the pattern. For
Quickcast patterns the maximum measured void ratio was Rv = 0.66, which may not
be good for complex designs.
6. Occasional shell cracking during pattern/gating
Despite problems cited, Quickcast set a trend in IC
technology. Research efforts are concentrated in resolving the cited problems
and pushing forward with new material and quicker methods. Problems cited above
may not make it economically feasible to use Quickcast technology in the
manufacturing of complex objects, especially if the number of units required is
small. This was the problem that we faced in manufacturing 2-3 units of
relatively complex objects. The next section describes the object in question
and the procedure followed in its fabrication.
4. Experiment Setup and The Product development
The objects in question are multiple angle "Junctions" and "end-caps" for a number of tubular carbon
composite structures. The end-caps are required to be manufactured of stainless
steel with high precision and internal surface finish. Figure 1.0 is sample
structure and the end-caps, targeted for casting.
Given the complexity of the object s geometry, it was not
possible to build the objects through milling. IC was selected as the technique
to be used in the fabrication of the intended objects. The complexity of the
objects is due to varying radii and angles, sockets of constant wall thickness
where tabular structural members are designed to mate, hollow geometry, as well
as sizes exceeding the work envelop for the existing SLA. These features coupled
with the limited number of units to be fabricated, make it difficult to
manufacture the objects using conventional pattern making or the Quickcast
technology. Following are the procedures that were developed to obtain wax
patterns which in turn were used to cast the objects to near net shape. There
are two phases to this effort. The initial phase includes converting the
design to a solid-model file using CAD/CAM software, developing the STL file
format, manipulating the data to provide for casting tolerances, orienting and
supporting the object using support structure software, before building the
model on the SL apparatus. Due to the limited size of this paper we will not
describe the regular SLA steps. Instead the reader may consult the references
[3,6] for the details of the process. Figures 2.0 shows the outcome of a typical
CAD process for an object and its core. An SLA model was developed for each
end-cape and its core. Shrinkage factors were added/subtracted to account for
casting of the objects in stainless steel. The outcome of the SLA process is an
actual part and core made-up of solidified resin. The second phase
involves developing wax patterns using the stereolithography model, and casting
in stainless steel. The SL model of the object and its core were used to
generate a wax pattern of the object/ cavity using Green wax, and to create a
wax pattern of the core using Yellow wax. The process of making the wax patterns
are as follows :
Figure 1.0 Sample structure and the end-caps targeted for
Figures 2.0 CAD process outcome (Object and Core)
Materials and setup
1. The mold box : Wooden mold boxes were prepared to house
the objects. Large complex objects with multiple angles required the mold to be
designed as several parts. Up to four part molds in certain cases.
2. Mold Material : The mold material that was chosen is plaster of Paris -
Ultracal 30, manufactured by US-Gypsum. Ultracal 30 ( white powder ) was added
to water in the ratio of 3 to 1. The mixture has to be continuously mixed for
7-11 minutes to obtain the correct consistency. This is an exothermic reaction
that sets within 5 minutes of pouring. When the mixture was ready for pouring a
1" base was poured into the mold box.
3. Wax Material : To obtain the "hollowness" in the
objects the core wax would have to be dissolved and
drained, leaving the wax pattern of the object hollow. The
two types of wax used in this experiment is the Yellow wax for the core, which
dissolves in Muriatic acid, and foundry green wax for the object.
1. Wax pattern - core: After the 1" base is hardened in
the mold box, the SL core was positioned and sheet metal inserts were placed
along previously strategically marked parting lines. Larger cores require
several sheet metal inserts. After the first metal insert was positioned, the
plaster of Paris was poured into the segmented mold, and upon hardened, the
metal sheet was removed. The hardened surface of the mold was then coated with
Johnson wax. Coating the surface of the mold prevents it from fusing to the
other segments of the mold when they are poured and aids in easy separation of
the final multi piece mold. Subsequently, the remaining segments of the mold
were similarly created. On completion of the multi-piece mold, the mold was
separated and the SL core removed. The mold was re-assembled and injected with
water soluble Yellow wax at a pressure of 50 lbs and 140 F. The Yellow wax takes
about 1 hour to completely solidify and can then be removed from the mold.
2. Wax pattern - object/cavity : Similar to the core, a
cavity for the object was created in a mold box. To create the wax pattern for
the object, the core was first positioned in a mold box with the help of
Aluminum end plugs. The Aluminum plugs were created to establish the wall
thickness for the final part, and they also act as a heat sink during the wax
casting. Once the core was positioned, the rest of the mold was assembled. The
mold then was injected with foundry green wax at 45 lbs at 140 F. The wax sets
within 20 minutes, yielding the cavity. However at this stage the Yellow wax
core is still within the Green wax core. 3. Separating core and cavity : In
order to separate the core from the cavity, the Aluminum end plugs were first
removed. The part was then placed in a solution of Muriatic acid (HCl). 2%
Citric acid can also be used for this purpose. The Muriatic acid dissolves the
Yellow wax core and yields the completed Green wax pattern.
Figures 3.0 Mold cavity box, parting lines and plaster mold
that yielded one of the objects wax patterns.
Figure 4.0 The completed wax casting where the cavity still
encloses the soluble wax core.
Figure 5.0 Wax casting after dissolving the core
This pattern can now be used in the next stage of investment
Figures 3.0 shows the mold cavity, box, parting lines, and
plaster mold that yielded one of the objects wax patterns. Figure 4.0 shows
the completed wax casting where the cavity still encloses the soluble wax core.
Figure 5.0 shows the final wax casting after the core was dissolved in Muriatic
acid. The next stage involves using the wax pattern for stainless steel
A number of wax pattern castings were successfully
manufactured using this procedure. The process has the potential for use in
rapid tooling where complexity, accuracy, quantity and time are important
5. Findings and Lessons Learned
Provided here is an summary of some of the issues experienced
during the conduct of this experiment.
1. The experiment proved the technical feasibility of
building accurate wax patterns for complex objects to be used in investment
casting. It would have been very difficult or even impossible to build with the
required accuracy and within the time frame required. In order to ensure
accuracy of the sockets, the SLA models were cut into pieces and all the sockets
were built oriented along the z-axis, using commercially available software.
Following the procedures described above, wax pattern of complex objects were
2. Model Scaling and accountability for shrinkage factors :
Models built on the SLA for use in foundry applications have to be scaled at two
levels. The first scaling is done to compensate for shrinkage's during the build
process. This scaling is compensation that is resin specific and is done on the
SLA. The second scaling is due to the casting process. Foundry applications
require different shrinkage coefficients based on the object size, complexity
and the casting material. Accounting for shrinkage coefficient should be done
through scaling the original CAD model before any subsequent processes.
3. Acceptable surface finish was obtained. The final wax
patterns met specifications in terms of surface finish and dimensional accuracy.
When building surfaces that are oriented at an angle to the vertical, the
stair-step effect is seen on the model. The level of imperfection depends on the
layer thickness and resolution set for the model, this has a direct effect on
the surface finish. The problem of stair-steps in the SLA was avoided by
orienting the parts correctly and sanding it during post processing.
4. Reduced Cycle Time : Once the process was mastered, the
cumulative cycle from obtaining the STL file to the finished wax casting was on
an average around 5 days. This is a tremendous increase in efficiency when
compared to conventional methods of tool and pattern making.
5. In situations where the size of the object exceeded the
work envelope, the object was sectioned into manageable sections using software
and later assemble (glued) together. Humidity and temperature affect the part
accuracy and are important parameters to consider if sections of the component
are built separately.
6. Sharp corners problems : One of the best features of
Stereolithography is that it can build features of almost any shape. One
geometric limitation that stereolithography does have is the ability to make
absolutely sharp corners and very small features. The ability in these areas are
limited by the width of these lines (.008") drawn by their laser beams. In
this experiment sharp corners were avoided in the design whenever possible.
There were two important issues that influenced cost in this
project. The first one was production time and the second was direct cost
involved in production.
The first aspect in time was converting the designs from 2D
drawing to 3D and subsequently to Solid model. Time estimated at 18 hours per
object, was spent in converting the designs to solid model. It was difficult and
sometimes impossible to create 3D fillets and merge surfaces. For complex
objects it is recommended that objects would be designed using software which
would enable solid modeling and producing the STL file necessary for creating
the SL model. A second aspect was the actual build time, which on average was 10
hrs for each component, and a similar time for the production of the core. It
should be recognized that time could be cut in half, if quick-cast and its
limitations is worked out.
As for the casting process, the majority of the time was
spend in the design and setup of the wax molding process. Designs has to be
custom-tailored for each object. However once the design is completed the actual
building of the mold, wax injection, ceramic dipping, and casting in metal took
the regular time for investment casting procedure. In practice, these processes
are done for all objects at the same time, thus reducing the cycle time per
unit. For non specialized facility like the one we used it is estimated that the
turn-around time is 5 days per object.
However the cost is a direct function of the object, the
direct cost attributed to the project other than those associated with time was
the material cost, overhead, and the laser. An average direct cost of $300.00
per object was estimated per object. The elements of the computed cost include,
preprocessing, post-processing, material cost, build cost, SLA maintenance cost,
In this paper we presented a set of procedure to
manufacturing complex object(s) using the technologies of investment casting
coupled with that of rapid prototyping. The outlined procedure is a deviation
from the trend of Quickcast generally practiced in investment casting circles.
Complexity of the objects required deviations from conventional techniques used
in investment casting. Rapid prototyping proved superior to traditional methods
both from an economic and technical standpoint, especially where limited numbers
1. Greenbaum , Perry Y. and Khan, S. ; " Direct Investment Casting of
RP Parts", Conference Proceedings, Rapid Prototyping and Manufacturing
, Dearborn, MI, 1994.
2. Hosni Y., Ferreira L., Burjanroppa R., "Rapid Prototyping through
Reverse Engineering", Proceedings of the Second IER Conference , Los
Angeles, California, 1993.
3. Hosni , Yasser A, "Rapid Prototyping and Reverse Enging'',
Proceedings ICC&IE-94, Ashikaga, Japan, 1994.
4. Hull, C. W. and Jacobs, P. F., "Stereolithography and Quickcast:
Moving Towards Rapid Tooling", 3D Systems, 1995
5. Prioleau F. R., "Applications of Stereolithography in Investment
Casting", Conference Proceedings, Second International Conference on
Rapid prototyping, University of Dayton, Dayton, Ohio, June 23-26, 1991.
6. Sundaram, R., "Investment Casting using Stereolithography";
MS Thesis, University of Central Florida, Orlando, FL, 1996.
Dr. Yasser A. Hosni is a professor of engineering in the department of
Industrial Engineering and Management Systems at the university of Central
Florida (UCF). Hosnis research interest is in area of advanced technology in
support of industrial processes. Lately he has been conducting research in the
area of rapid production in general and that of rapid prototyping and tooling in
particular. This paper is an outcome of such effort as it relates to investment
casting. Hosni, has lectured on the subject in Japan, Egypt, and the US. He
authored and published numerous papers and conference presentations over a span
of 20 years of experience in industry and academia. He had 6 papers over the
last 2 years on the subject of rapid production. Hosni is a senior member of IIE,
IEEE, SME, INFORMS, and ASEE.
Dr. Jamal Nayfeh is an associate professor in the department of
Materials, Mechanical, and Aerospace Engineering at the University of Central
Florida. Nayfehs research interest is in the application of mechanics in the
analysis, design, and manufacture of a advanced materials and structures. He had
published and authored over 30 papers and conducted short courses and workshops
dealing with his research interest. Nayfeh has collaborated with Hosni in
establishing the Design an Rapid Prototyping Laboratory at UCF. DRPL is equipped
with state-of-the-art equipment for design and RP. The research presented in
this paper was conducted in the DRPL. Nayfeh is a member of AIAA, ASME, ASCE,
Ravindran Sundaram is a graduate student in industrial engineering at UCF.
Sundaram completed his MS. thesis based on the experiment presented in this
paper. He had a BS in Mechanical engineering from Bangalore university, India,
in 1991. He had industrial experience in India and in the US. His research
interest is rapid prototyping and the CAD/CAM technology. He is a member in IIE,
Yasser A. Hosni, Jamal Nayfeh, Ravindran Sundaram
Department of Industrial Engineering and Management Systems
University of Central Florida, Orlando, Florida, USA