This is the third of three articles published by InCast magazine on the results of a study I did comparing AM methods to make investment casting patterns.

1. ®
8 ❘ June 2017
Feature Article
A Comparison of 3D Printing Technologies
Used to Make Investment Casting Patterns
by Tom Mueller, Mueller Additive Manufacturing Solutions
I
n this series, the four leading
methods of creating printed patterns
for investment casting are compared.
Part 1 of the series compared the
performance of the printers themselves.
Part 2 compared operating costs. In
this, the third and final segment, the
casting performance of the patterns is
compared.
Pattern Performance
The performance of the printer and
operating costs are only part of the
equation. Having a printer that performs
well and is inexpensive to operate is
of little value if the patterns it creates
cannot be reliably or economically
converted to good castings.
Of the four technologies, only the
Projet wax printer builds patterns in wax.
For the most part, those patterns can
be processed exactly like molded wax
patterns. The other three technologies,
however, create patterns in materials that
require some modification to the typical
casting process used by investment
foundries.
In this section, several factors
related to the ability to successfully
cast printed patterns are discussed and
evaluated with respect to each of the
four technologies.
The first three performance
considerations are all related to thermal
expansion of the pattern. In general,
patterns, even wax patterns, expand
with heat and can result in cracking of
the shell, usually in the autoclave. The
next three considerations examine three
different aspects of thermal expansion
issues.
Shell Cracking - By far the most
common cause of failure in casting
printed patterns is shell cracking in
the autoclave as a result of thermal
expansion of the pattern. The Voxeljet
and CastForm patterns have an advantage
in this respect in that they both shrink
slightly as the pattern heats up. This is
most likely due to the fact that they are
powder based and not fully dense when
built. In my experience, there has not
been a single case of shell cracking in
the autoclave when Voxeljet patterns are
used.
ProJet wax patterns have a
coefficient of thermal expansion close to
that of casting wax. Consequently, they
should be no more likely to crack a shell
than a molded wax pattern.
QuickCast has the highest
probability of cracking the shell in the
autoclave. Even though the QuickCast
hexagonal internal support structure is
hollow and was designed to allow the
pattern to collapse inward as it heats
up, there is a solid post at each corner
of the hex structure that runs from skin
to skin and is well supported along its
length. Consequently, it is virtually
impossible for the structure to collapse.
This explains the high rate of failure
of QuickCast patterns in the autoclave
step. There is, however, a work around.
If the pattern is vented and the skin of the
pattern punctured under the vent prior
to the autoclave step, steam can enter
the interior of the pattern quickly. The
heat of the steam softens the material
significantly, allowing the structure to
collapse as intended.
Required Shell Thickness – Given
the tendency of some printed patterns
to crack shells in the autoclave,
foundries often add dips to the shells
they use for printed patterns to create a
stronger shell that can better resist the
expansion. Until recently no study was
undertaken to determine how many
dips were necessary to avoid cracking.
In the past two years, research showed
that QuickCast patterns required one
additional dip compared to molded wax
patterns, assuming they were adequately
vented. As might be expected, Projet
patterns required the same number of
Figure 1: Variations to the normal investment casting process required for printed
patterns.

2. ®
June 2017 ❘ 9
dips as molded wax patterns. Voxeljet
patterns required one less dip than would
be required for molded wax patterns.
CastForm patterns were not included in
the study. More detail on the previous
reseach mentioned above is available in
the referenced Incast article.
Reducing dips on the shell can save
some money in materials and labor.
However, the major benefit is in the
reduction of process time. For most
foundries, saving one layer on the shell
means the casting can be shipped one
day earlier.
Thin wall patterns – Patterns with
thin wall sections have always presented
difficulties in the casting process. Not
only are they more difficult to fill, the
stress raisers created by the sharper
corners make shells more likely to crack.
Recent research documented the thin
wall performance of printed patterns.
For QuickCast patterns, the situation
is complicated by the inability to build
hollow thin walls. 1.5mm (0.060 in) is
a practical lower limit for hollow walls.
Thinner walls will be solid. If the walls
are not hollow, then there is no way to
counteract the thermal expansion of the
pattern and the chances of shell cracking
are much higher.
Projet wax patterns expand less and
are less likely to break in the autoclave.
However, the patterns are so fragile that
walls less than 1.5mm require extreme
care during assembly and may break
during dipping.
Voxeljet patterns, because they
don’t expand with heat, are able to
survive autoclaving and are relatively
durable. Walls as thin as 0.030” were
successfully autoclaved.
Although the thin wall performance
of CastForm patterns was not tested,
there is reason to believe that it would
be acceptable. More detail on this study
is available in the referenced Incast
article.
Pattern Durability – Patterns are
subjected to force in both assembly
and in dipping. A pattern that is too
fragile may break as a result of the
force. QuickCast patterns are by far
the most durable of the printed patterns
and are much stronger than molded
wax patterns. An offsetting issue is that
because they are hollow, there can be
significant buoyancy forces acting on
the pattern when it is dipped.
Voxeljet patterns are the second
most durable. While they can be easily
broken, they are stronger than molded
wax patterns.
CastForm and Projet wax patterns
are more fragile than molded wax
patterns and must be handled with care
to avoid breaking them during assembly
and initial dips.
EaseofCasting–Formostinvestment
foundries, nearly all their production
is assemblies of molded wax patterns.
Assemblies with printed patterns are by
far the exception, typically less than 1%
of any day’s production. The ease of
casting is directly related to how much
the process required for the printed
pattern varies from that used for molded
wax patterns. More variation increases
the probability that one of the variant
steps will be missed, resulting in a
scrapped casting.
Figure 1 illustrates the variations
in the casting process required for
each of the four technologies. Any
step colored in light blue is a variation
from the conventional process. The
first column lays out the major steps
in the conventional process including
assembly, shelling, de-wax, preheat,
pouring, shell removal and cleanup.
• QuickCast patterns require
significant deviations from the
conventional process. It begins
with a pattern inspection prior
to assembly. Most foundries that
process many QuickCast patterns
will do a leak inspection prior to
assembly to make sure there is no
possibility of getting slurry into the
pattern. That typically requires
applying a vacuum to the inside
of the pattern and checking to see
if it will hold a vacuum. A loss of
vacuum would indicate a leak. The
port where the vacuum is pulled
must then be patched.
The next variation is in assembly.
At least one vent is applied to
each pattern on the assembly. The
vents serve two purposes; allowing
steam to enter the pattern during
autoclaving, and allowing air to
flow through the mold during
burnout replenishing oxygen used
in combustion. If the pattern is larger
than 15-20 cm in any direction, it
may be necessary to use multiple
vents to ensure that steam can get
to all parts of the pattern and soften
the support structure, allowing it
to collapse inwardly as the pattern
expands.
Prior to autoclaving, the vents
must be opened and the skin of the
pattern must be punctured so that
steam can enter.
After autoclaving, the patterns need
to be burned out. After burnout,
the shell is cooled, and the ash
cleaned out, either by compressed
air or water rinse. The vents are
then patched. The final variation
is to grind the vents during casting
cleanup.
Table 1: Residual Ash Comparison
Method Material Residual Ash % Apparent Density (lb/in3
) Ash Mass (10-6
lb/in3
) Ratio
QuickCast CastPro 0.00570% 0.01075 0.613 1.0
CastForm Polystyene 0.02000% 0.01660 3.320 5.42
Project Wax M3 HiCast NA 0.02926 NA NA
Voxeljet PMMA PPB .000533% 0.02529 1.35 2.20

3. ®
10 ❘ June 2017
• CastForm patterns require fewer
steps because vents typically are
not required. Most foundries do,
however, burn out the pattern
requiring three additional steps
related to the burnout.
• ProJet wax is the best in terms
of ease of casting. There are no
variations from the conventional
process.
• Voxeljet patterns require a process
similar to casting QuickCast
patterns except that fewer vents are
required. It is recommended that a
single vent be added to each pattern
during the assembly step to provide
for airflow through the shell during
burnout. That vent is opened prior
to burnout. The shell then goes
through a burnout step, followed
by a clean out step for the shell.
The vents are then patched prior
to preheat. In casting cleanup, the
vent stubs must be removed and
ground flat.
Residual Ash – QuickCast, CastForm
and Voxeljet patterns will not melt in
an autoclave de-wax step and must be
burned out, typically in a preheat oven.
Even though the residual ash levels of
these materials is typically low, there
can be enough ash remaining in the
shell to cause unacceptable ash related
surface defects in the casting. To avoid
such defects, many foundries cool the
shell to room temperature after burnout.
The ash is removed either by blowing out
the shell with compressed air or rinsing
it with water. At that time any vents can
be patched as well. The shell is then
pre-heated and poured conventionally.
For fused silica shell systems,
it is important to keep the burn out
temperature below the cristobalite
conversion temperature (approximately
1650ºF) so that cristobalite formation
does not take place. Cracking of the
cristobalite during cool down for clean-
out could seriously weaken the shell at
pouring.
Manufacturers publish residual ash
percentage values for their materials.
QuickCast patterns, however, are
hollow, reducing the mass of the pattern
Feature Article
Table 2: Summary of Results
Technology/Printer
QuickCast
ProX 800
CastForm sPro 60
ProJet Wax
3510 CPX
Voxeljet
Printer
Performance
Accuracy 0.1 - 0.2% ? 0.1 - 0.2% 0.30%
Surface Finish G/E F/VG E/E F/VG
Build Volume (in3
) 16,362 3,353 683 18.307
Multiple Layer Build No Yes No Yes
Build Rate (in3
/hr) 166.7 101.3 14.5 269.7
OperatingCost
Printer Price $550,000 $495,000 $100,000 $806,000
Capacity Cost ($/in3
/hr) $3,294 $4,687 $6,884 $2,948
Material Cost ($/in3
) $1.98 $1.13 $6.70 $0.81
Depreciation Cost ($/in3
) $0.12 $0.17 $0.26
Maintenance Cost ($/in3
) $0.12 $0.19 $0.20
Total Cost ($/in3
) $4.16 $4.47 $10.82
PatternPerformance
Shell Cracking Highest Low Low
Minimum Shell Thickness +1 Layer Same Same -1 Layer
Minimum Wall Thickness >0.060 ? >0.030 in. <0.030 in.
Pattern Durability Excellent Very Brittle Very Brittle Good
Ease of Casting Difficult Relatively Easy Easy Moderate
Residual Ash (10-6lb/in3) 0.61 5.42 ? 1.35
Pattern Leak Potential Yes No No No
Heavy Metal Content Yes No No No

4. ®
June 2017 ❘ 11
and the amount of ash remaining in the
shell. The effective density will vary
depending on the parameters chosen
for the hexagonal support structure, the
thickness of the skin and the surface area
to volume ratio of the pattern geometry.
To determine an average density, volume
and weights of 55 QuickCast patterns
were recorded. The density of each was
calculated by dividing the weight by the
volume and the density values averaged.
The result of the study is shown in the
table below. No value of residual ash
was available for Projet Wax.
The ratio in the last column provides
a quick comparison. If QuickCast is the
base, CastForm will have more than 5
times as much ash and Voxeljet a little
over twice as much ash. However, it is
important to note that residual ash levels
can vary widely depending on burn-out
conditions. Ash levels will be higher
if there is insufficient oxygen in the
furnace atmosphere or if airflow through
the mold is limited.
Pattern Leaks – Pattern leaks are
only an issue with QuickCast patterns
because they are the only patterns that
are hollow. If there is a break in the
skin of the pattern, slurry can seep into
the pattern during dipping, resulting in
inclusions in the casting. CastForm,
Projet Wax and Voxeljet patterns are
solid with no chance of slurry leaking
into the pattern.
Heavy Metal Content – Most
stereolithography resins contain trace
amounts of antimony in the photo-
initiator component of the resin. A study
by DSM showed that 80% of the ash
remaining after burnout was antimony or
antimony salts. Antimony can seriously
degrade material properties of castings,
particularly titanium and superalloys.
Unless all the ash is removed from the
shell after burnout, there is a potential
of antimony contamination. CastForm,
Projet Wax, and Voxeljet have no heavy
metal content in their formulation.
In the past few years, at least two
resins have been released which do not
contain antimony or any other heavy
metal in their formulation. The use of
these resins should eliminate the heavy
metal content issue as well as reduce
residual ash levels.
Summary
Table 2 summarizes the results of the
study.
It is interesting that the printer with
the lowest purchase price, the ProJet
3510 CPX, is by far the most expensive
in terms of cost of capacity and operating
cost.
It is important to note that there is no
single “best” pattern printing technology.
For each of the four technologies,
applications exist for which they are
the best alternative. It is a matter of
determining which printer characteristics
and pattern characteristics are best for a
particular application.
For example, if the application is to
make small highly detailed parts such
as jewelry, the ProJet would clearly be
the best choice for the application. It
will provide the best detail resolution,
best surface finish, and is the easiest to
cast. The high value of the casting will
more than offset the increased cost of
the pattern.
If the application is to make
prototype castings for a large turbine
blade where surface finish and accuracy
are critical, QuickCast would be the best
choice. Its accuracy and surface finish
capabilities are most likely to meet the
high demands of the application.
If the application includes very thin
walls, Voxeljet may be the best choice.
Most applications, however, have
much less stringent requirements; any
of the four technologies can meet the
accuracy and surface finish requirements
and variations in the casting process will
not significantly affect the cost of the
casting. In such cases, the important
criteria become the speed and cost of
obtaining patterns. It is interesting to
plot each of the four printers on a speed
vs. cost chart as shown in Figure 2. For
these applications, the most desirable
technologies will have a high print
speed and a low pattern cost. They will
lie in the upper left of the chart. The
least desirable will be in the lower right
portion of the chart with a low print
speed and high pattern cost.
Conclusion
Requirements for investment castings
vary considerably depending on the
application. The information presented
here can be used to select the technology
most appropriate for the application.
Endnotes
1
“An Evaluation of Autoclave Performance of
Thin Wall Printed Investment Casting Patterns
and Minimum Shell Thickness Necessary to
Avoid Shell Cracking in the Autoclave”,
Tom Mueller, Investment Casting Institute
Technical Conference, October 2014.
2
“An Evaluation of Autoclave Performance
of Thin Wall Printed Investment Casting
Patterns and Minimum Shell Thickness
Necessary to Avoid Shell Cracking in the
Autoclave”, Tom Mueller, Investment Casting
Institute Technical Conference, October
2014.
Figure 2: Average Build Speed vs. Pattern Cost