I have spent a significant part of my recent engineering work on the intersection of additive manufacturing, investment casting, and gas turbine blade production. The central problem I wanted to solve was straightforward in statement but difficult in practice: traditional investment casting requires a long chain of tooling, wax pattern injection, gating assembly, shell building, dewaxing, burn-out, and pouring. When a casting dimension or process parameter must be changed, the entire chain often has to be repeated. That repetition consumes weeks or months and creates a large economic penalty. In my work, I therefore investigated a rapid investment casting route that uses an acrylonitrile butadiene styrene pattern produced by fused deposition modeling, a wax-based gating system, a ceramic shell, and a controlled burn-out and pouring cycle. I also studied the related family of expendable pattern processes, including evaporative pattern casting, because the underlying philosophy is similar: the pattern is sacrificial, and the mold or shell is built around a shape that later disappears.
The main technical difficulty I encountered was shell cracking. The shell sometimes cracked during burn-out or during metal pouring, and the resulting blade castings showed flash, surface roughness, and dimensional drift. I did not want to solve this problem only by trial and error. I therefore combined a deformation-coordination model, finite element analysis, and practical process changes. The result was a more robust route that produced sound gas turbine blade castings with less iteration. In this article I present the reasoning, equations, tables, and optimization steps that I used. I also connect the work to evaporative pattern casting, because many of the same thermal-expansion and pattern-removal issues appear there as well.
1. Motivation and Process Route
In conventional investment casting, a metal die is used to produce wax patterns. The wax patterns are assembled onto a gating tree, dipped into ceramic slurry, stuccoed with sand, dried, dewaxed, fired, and filled with molten metal. This route is excellent for repeatability and high volume, but it is not agile. A gas turbine blade has a complex airfoil, a root, a shroud, internal cooling passages in many designs, and tight tolerances. If the first shell or die is wrong, the cost of correction is high.
Rapid investment casting replaces the wax injection tool with a printed pattern. In my case, I used fused deposition modeling to print an ABS blade pattern. The ABS pattern was bonded to a conventional wax gating system. A ceramic shell was then built around the assembly. After drying, the pattern and gating wax were removed, the shell was fired, and molten stainless steel was poured. This approach removes the die-making step and allows geometry changes to be made directly in the digital model. It is especially useful for prototypes, repair parts, and low-volume gas turbine blades.
I also reviewed evaporative pattern casting as a related expendable-pattern method. In evaporative pattern casting, a foam pattern is coated with refractory and buried in sand. When molten metal enters the mold, the foam pattern vaporizes and the metal takes its place. The mold is not removed before pouring, unlike the burn-out step in investment casting. This difference is important. In evaporative pattern casting the pattern disappears during pouring, while in rapid investment casting the pattern disappears before pouring. Both routes, however, depend on controlled thermal decomposition, gas venting, dimensional stability, and shell or mold strength. I kept the lessons of evaporative pattern casting in mind when I designed the burn-out cycle and the shell support system.
Table 1 summarizes the comparison I used during route selection. I did not treat evaporative pattern casting as a direct replacement for rapid investment casting. Instead, I treated it as a benchmark for pattern removal, gas evolution, and mold support. The comparison helped me identify which variables were likely to matter most in my own process.
| Feature | Conventional Investment Casting | Rapid Investment Casting with Printed Pattern | Evaporative Pattern Casting |
|---|---|---|---|
| Pattern material | Wax or filled wax | ABS, PLA, or other polymer | Expanded polystyrene or copolymer foam |
| Tooling requirement | Metal die for wax injection | No die; direct digital print | No die; foam mold or CNC foam |
| Pattern removal | Autoclave or furnace dewaxing | Burn-out, solvent, or thermal decomposition | Vaporization by molten metal |
| Mold or shell | Ceramic shell | Ceramic shell | Refractory coating plus unbonded sand |
| Typical lead time | Weeks to months | Days | Days to weeks |
| Best application | High-volume precision parts | Prototypes, repair parts, complex blades | Large castings, automotive, some industrial parts |
| Main defect risk | Wax shrinkage, shell cracking, inclusions | Polymer expansion, shell cracking, ash residue | Gas porosity, carbon pick-up, mold collapse |
The process route I finally selected had seven main steps:
- Build the gas turbine blade model in a three-dimensional computer-aided design environment.
- Convert the model to a stereolithography file and print the blade in ABS by fused deposition modeling.
- Bond the printed blade to a wax gating system, choosing the root as the metal entry region for this blade family.
- Build a ceramic shell by repeated slurry dipping and sand stuccoing.
- Remove the wax gating and the ABS blade through thermal treatment.
- Fire the shell to develop strength and remove residual organics.
- Pour molten stainless steel and finish the blade casting.
Although this route is called rapid investment casting, it shares several thermal and mechanical features with evaporative pattern casting. In both routes, the pattern has a much higher thermal expansion coefficient than the ceramic or refractory system. In both routes, gas and vapor must escape without damaging the mold. In both routes, the pattern or its residue can influence surface quality. I therefore treated the shell as a pressure vessel and a thermal expansion constraint, not merely as a container for liquid metal.
2. Experimental Design and Process Parameters
I used a gas turbine compressor blade as the target geometry. The blade had a curved airfoil, a root section, and a relatively thin trailing edge. The printed ABS pattern was produced with a layer thickness that balanced surface finish and build time. I used a sparse internal structure in some trials and a solid internal structure in others. The sparse structure reduced print time and, more importantly, reduced the amount of polymer that had to decompose during burn-out. This is also a known concern in evaporative pattern casting, where foam density strongly affects gas evolution and defect formation.
The gating system was made from conventional wax. The printed blade was bonded to the wax runner and sprue. I selected root feeding because the root is the thickest section and because it provides a more uniform fill path for the airfoil. I avoided feeding directly into the thin trailing edge, since that can create jetting, cold shuts, and shell erosion.
The ceramic shell was built in multiple layers. Each layer consisted of a slurry dip followed by sand stuccoing and drying under controlled temperature and humidity. The first layers used finer sand to reproduce surface detail. The later layers used coarser sand to build thickness and strength. Table 2 shows the shell recipe in a generalized form. The exact commercial identities are not important for the present discussion; what matters is the progression from fine to coarse and the controlled drying between layers.
| Layer | Binder | Powder | Sand Grit | Function |
|---|---|---|---|---|
| 1 | Silica sol | Fine alumina | Fine | Surface reproduction and primary coat |
| 2 | Silica sol | Fine alumina | Medium-fine | Primary backup and edge support |
| 3 | Silica sol | Refractory powder | Medium | Build thickness and thermal shock resistance |
| 4 | Silica sol | Refractory powder | Medium-coarse | Mechanical support |
| 5 | Silica sol | Refractory powder | Coarse | Final strength and handling durability |
| 6 | Silica sol | Refractory powder | Coarse | Added strength after initial cracking trials |
In the first trials, I used five layers. The shell cracked during burn-out and pouring. I then changed to six layers, and in some later trials I used a sixth full layer plus a partial seventh layer. This change increased shell thickness and bending stiffness. The effect was not simply “more is better,” because an excessively thick shell can retain thermal gradients and crack under thermal shock. I therefore combined the thickness increase with a slower burn-out schedule and an improved blade edge radius.
The burn-out and firing cycle was critical. The ABS pattern must decompose without generating excessive internal pressure. If the heating rate is too fast, the polymer produces gas faster than it can escape through the shell. The shell then behaves like a sealed pressure vessel and cracks. If the heating rate is too slow, the process becomes uneconomical. I used a staged schedule with low-temperature soaking, gradual ramping, and a final high-temperature hold. This is similar in principle to the venting strategy used in evaporative pattern casting, although the pattern removal occurs before pouring in my process.
| Stage | Temperature Range | Purpose | Risk if Incorrect |
|---|---|---|---|
| Initial soak | Near 100–150 °C | Melt and soften binder and polymer | Rapid gas generation and shell pressure |
| Slow ramp | 150–350 °C | Decompose ABS and wax residue | Carbon residue, cracking, incomplete removal |
| Intermediate hold | 350–600 °C | Burn out organic residue | Ash, surface defects, incomplete oxidation |
| High-temperature firing | 900–1,050 °C | Sinter and strengthen shell | Low strength, distortion, poor dimensional control |
| Cooling | Controlled furnace cooling | Avoid thermal shock | Cracks before pouring |
I measured shell cracking after each trial. The cracks were most common near the leading and trailing edges of the airfoil. This observation pointed to curvature and constraint, not just to material strength. A flat plate expands and contracts more freely than a curved shell. A blade edge with a small radius creates a stress concentration. I therefore developed a thermomechanical model to explain the observations.
3. Thermomechanical Model and Deformation Coordination
To make the problem tractable, I modeled the printed pattern and the surrounding shell as two concentric cylinders. This is a simplification, but it captures the essential constraint: the polymer expands more than the ceramic shell, so the shell is forced to expand with the pattern. The pattern is compressed inward, and the shell is stretched outward. The two bodies must remain in contact, so their radial deformations at the interface must be compatible.
Let the inner radius of the pattern be \(a\), the outer radius be \(b\), and the shell thickness be \(x\). Let the initial temperature be \(T_1\), the final temperature be \(T_2\), and the temperature change be \(\Delta T = T_2 – T_1\). Let the thermal expansion coefficients be \(\alpha_1\) for the pattern and \(\alpha_2\) for the shell. Let the elastic moduli be \(E_1\) and \(E_2\), and let the interfacial stresses be \(\sigma_1\) and \(\sigma_2\).
The free thermal expansion of the pattern would be:
$$ \Delta r_{1,\text{free}} = b \alpha_1 \Delta T $$
The elastic compression of the pattern caused by the shell is:
$$ \Delta r_{1,\text{elastic}} = -\frac{b \sigma_1}{E_1} $$
Therefore, the net outward deformation of the pattern at the interface is:
$$ X_1 = 2\pi b \alpha_1 \Delta T – \frac{2\pi b \sigma_1}{E_1} $$
Similarly, the free thermal expansion of the shell is:
$$ \Delta r_{2,\text{free}} = b \alpha_2 \Delta T $$
The elastic expansion of the shell caused by the pattern is:
$$ \Delta r_{2,\text{elastic}} = \frac{b \sigma_2}{E_2} $$
The net outward deformation of the shell at the interface is:
$$ X_2 = 2\pi b \alpha_2 \Delta T + \frac{2\pi b \sigma_2}{E_2} $$
Because the pattern and shell remain in contact, their deformations must be equal:
$$ X_1 = X_2 $$
Substituting the expressions gives the deformation coordination equation:
$$ 2\pi b \alpha_1 \Delta T – \frac{2\pi b \sigma_1}{E_1} = 2\pi b \alpha_2 \Delta T + \frac{2\pi b \sigma_2}{E_2} $$
Dividing by \(2\pi b\) gives a simpler form:
$$ \alpha_1 \Delta T – \frac{\sigma_1}{E_1} = \alpha_2 \Delta T + \frac{\sigma_2}{E_2} $$
At the interface, the force exerted by the pattern on the shell must equal the force exerted by the shell on the pattern:
$$ \sigma_1 t_1 = \sigma_2 t_2 $$
where \(t_1\) is the effective pattern wall thickness and \(t_2 = x\) is the shell thickness. I then solved for the shell stress. The interfacial pressure \(p\) can be written as:
$$ p = \frac{(\alpha_1 – \alpha_2)\Delta T}{\frac{1}{E_2} + \frac{t_2}{E_1 t_1}} $$
For a thin shell, the hoop stress is approximately:
$$ \sigma_\theta = \frac{p b}{t_2} $$
Combining these expressions gives:
$$ \sigma_\theta = \frac{b}{t_2} \frac{(\alpha_1 – \alpha_2)\Delta T}{\frac{1}{E_2} + \frac{t_2}{E_1 t_1}} $$
This equation shows the key dependencies. The shell stress increases with the difference in thermal expansion coefficients, with the temperature change, and with the radius of curvature. It decreases as the shell thickness and shell modulus increase. That is why a thicker shell helped in my trials, but only up to a point. If the shell becomes too thick, thermal gradients through the thickness become more severe, and the assumption of uniform temperature becomes less valid.
I also used the von Mises equivalent stress to compare the finite element results with the analytical trend:
$$ \sigma_v = \sqrt{\frac{1}{2}\left[(\sigma_1-\sigma_2)^2+(\sigma_2-\sigma_3)^2+(\sigma_3-\sigma_1)^2\right]} $$
For the thermal transient, I used a simple linear ramp in the finite element model:
$$ T(t) = 22 + at $$
where \(T(t)\) is the temperature at time \(t\), \(22\) is the initial room temperature in degrees Celsius, and \(a\) is the heating rate. In the actual furnace, the ramp was not perfectly linear, but the linear model was sufficient to identify the stress concentration regions.
Table 3 lists the material properties I used in the finite element model. The ABS properties were taken at a temperature above its glass transition, where the modulus is low and the expansion is high. The shell properties represent the ceramic body after partial sintering. These values are approximate, but they reproduce the qualitative behavior observed in the trials.
| Property | ABS Pattern | Ceramic Shell |
|---|---|---|
| Elastic modulus | Approximately 1.0 MPa at elevated temperature | Approximately 630 MPa |
| Poisson ratio | 0.43 | 0.26 |
| Thermal expansion coefficient | Approximately \(92 \times 10^{-6} /^\circ\text{C}\) | Approximately \(4 \times 10^{-6} /^\circ\text{C}\) |
| Thermal conductivity | Low | Moderate |
| Role in process | Expands, decomposes, generates gas | Constrains, supports, resists thermal shock |
I created two-dimensional finite element models of the blade cross-section. I used a bonded contact between the pattern and the shell. The mesh used four-node thermal-solid elements. I analyzed solid and hollow blade patterns, and I compared shell thicknesses of 6 mm and 8 mm. The purpose was not to predict the exact crack location in every trial, but to rank the stress states and guide the optimization.
| Model Case | Pattern Type | Shell Thickness | Temperature Rise | Output |
|---|---|---|---|---|
| A | Solid ABS | 6 mm | 22 to 150 °C | Von Mises stress, interface pressure |
| B | Solid ABS | 8 mm | 22 to 150 °C | Von Mises stress, interface pressure |
| C | Hollow ABS | 6 mm | 22 to 150 °C | Von Mises stress, interface pressure |
| D | Hollow ABS | 8 mm | 22 to 150 °C | Von Mises stress, interface pressure |
| E | Hollow ABS with grid | 8 mm | 22 to 150 °C | Von Mises stress, gas path |
The finite element results showed that the maximum equivalent stress was concentrated at the leading and trailing edges. These are the regions of smallest curvature radius. The stress decreased toward the central portion of the airfoil, where the curvature is gentler. This matched the physical cracks I observed. The shell did not crack randomly; it cracked where the geometry forced the largest mismatch strain.
I also found that a hollow or sparse pattern reduced the maximum shell stress. The reason is that a hollow pattern has less material to expand, and it can deform inward more easily. The gas path is also shorter, so decomposition products can escape more readily. This observation is directly relevant to evaporative pattern casting, where foam density and pattern permeability are known to control gas defects and mold pressure.
4. Finite Element Results and Failure Interpretation
The stress cloud from the finite element analysis showed a clear pattern. The maximum stress appeared at the edges, and the minimum stress appeared in the middle of the convex and concave surfaces. I used the results to define a risk index:
$$ R = \frac{\sigma_{\max}}{\sigma_{\text{allow}}} $$
where \(\sigma_{\max}\) is the maximum equivalent stress and \(\sigma_{\text{allow}}\) is the allowable shell strength. When \(R \gt 1\), cracking is likely. When \(R \lt 1\), the shell has a margin. Table 4 summarizes the relative results for the cases I analyzed. The values are normalized because the exact shell strength depends on firing and testing conditions.
| Case | Pattern Type | Shell Thickness | Relative Maximum Stress | Risk Index | Observed or Predicted Outcome |
|---|---|---|---|---|---|
| A | Solid ABS | 6 mm | 1.00 | 1.18 | High cracking risk at edges |
| B | Solid ABS | 8 mm | 0.82 | 0.97 | Marginal, still cracked in some trials |
| C | Hollow ABS | 6 mm | 0.78 | 0.92 | Improved but not fully robust |
| D | Hollow ABS | 8 mm | 0.61 | 0.72 | Good margin, fewer cracks |
| E | Hollow ABS with grid | 8 mm plus partial layer | 0.54 | 0.64 | Best result before edge radius change |
The results confirmed three main points. First, shell thickness has a strong effect on stress. Increasing the shell from 6 mm to 8 mm reduced the relative maximum stress by about 18 percent in the solid-pattern case. Second, pattern architecture matters. A hollow or sparse pattern reduced the stress by about 22 percent compared with a solid pattern at the same shell thickness. Third, the edge radius matters. The stress concentration at the leading and trailing edges was the limiting factor. I therefore treated the edge radius as a design variable.
I introduced a simple curvature stress concentration factor:
$$ K_t = 1 + \frac{2t_2}{R} $$
where \(R\) is the local radius of curvature and \(t_2\) is the shell thickness. This is not a rigorous fracture mechanics solution, but it captures the trend: as \(R\) decreases, \(K_t\) increases. For a thin trailing edge, \(R\) is small, so the local stress is amplified. By increasing the edge radius in the digital model, I reduced \(K_t\) and moved the maximum stress away from the most fragile part of the shell.
I also considered the thermal transient. The shell does not heat uniformly. The outer surface may heat faster or slower than the inner surface depending on the furnace and the shell thickness. The transient heat conduction equation is:
$$ \rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + Q $$
where \(\rho\) is density, \(c_p\) is specific heat, \(k\) is thermal conductivity, and \(Q\) is any internal heat source. In my process, the decomposition of ABS and wax provides a transient internal gas source, but the dominant source is the furnace. The important practical consequence is that a very fast ramp creates a large temperature difference through the shell thickness. That temperature difference adds to the expansion mismatch. A slower ramp reduces the transient thermal stress but increases process time. The optimized schedule therefore used a moderate ramp with a hold in the decomposition range.
I also examined the gas pressure generated by polymer decomposition. If the shell is treated as a closed vessel, the pressure rise depends on the gas generation rate, the permeability of the shell, and the venting path. A simple mass balance is:
$$ \frac{dm_g}{dt} = \dot{m}_{\text{gen}} – \dot{m}_{\text{vent}} $$
where \(m_g\) is the mass of gas inside the shell, \(\dot{m}_{\text{gen}}\) is the generation rate from the decomposing pattern, and \(\dot{m}_{\text{vent}}\) is the venting rate through the shell and gating openings. When \(\dot{m}_{\text{gen}} \gt \dot{m}_{\text{vent}}\), pressure builds. This can crack the shell even if the thermal expansion mismatch is moderate. This is a central issue in evaporative pattern casting as well, where foam decomposition gases must escape through the refractory coating and sand. In my process, the gating opening and the shell porosity provided the vent path. A hollow or sparse pattern reduced \(\dot{m}_{\text{gen}}\), which helped avoid pressure-driven cracking.
5. Optimization Measures
Based on the analytical model and the finite element results, I implemented four main optimization measures. I did not apply them one at a time in every trial, because the interactions were important. Instead, I used a structured optimization matrix and then verified the best combination.
The first measure was to increase the leading-edge and trailing-edge radii in the digital blade model. This reduced the stress concentration factor. The change was small enough to preserve the aerodynamic function of the blade for the trial, but large enough to reduce the local shell stress. In a production blade, such a change would require aerodynamic review, but for the rapid casting trial it was an acceptable geometry modification.
The second measure was to print the ABS pattern with a sparse or grid-like internal structure. This reduced the mass of polymer that had to decompose. It also allowed the pattern to compress inward more easily under shell constraint. The external surface remained solid and smooth, so the casting surface quality was not compromised. This approach is conceptually similar to density control in evaporative pattern casting, where a lower-density foam can reduce gas defects but may also reduce pattern stiffness.
The third measure was to increase the shell thickness. I changed the shell from five layers to six layers and added a partial seventh layer at the most highly stressed edges. The additional thickness increased the bending stiffness of the shell and lowered the hoop stress. I also ensured that the additional layers were well dried before the next dip. Moisture trapped between layers can cause steam-driven cracks during burn-out.
The fourth measure was to revise the burn-out and firing schedule. I used a slower ramp in the temperature range where ABS decomposes, and I added a hold to allow gases to escape. I also increased the venting area near the gating system. This reduced the internal pressure peak.
| Optimization Measure | Problem Addressed | Expected Effect | Implementation Detail |
|---|---|---|---|
| Increase edge radii | Stress concentration at leading and trailing edges | Lower \(K_t\), lower maximum stress | Modify digital blade model before printing |
| Sparse or grid ABS pattern | High expansion and gas generation | Lower pattern mass, easier inward deformation | Use internal grid instead of solid infill |
| Increase shell layers | Insufficient shell strength and stiffness | Lower hoop stress, better handling strength | Six full layers plus partial seventh at edges |
| Revised burn-out cycle | Rapid gas generation and thermal shock | Lower internal pressure, lower transient stress | Slow ramp, intermediate holds, controlled cooling |
| Improved venting | Gas trapped inside shell | Lower pressure, less cracking | Enlarge gating opening and avoid sealed pockets |
I also checked the gating design. The original gating entered at the root, which is generally correct for a blade. However, a narrow gate can accelerate the metal and create a hot spot. I slightly enlarged the gate and used a rounded transition. This reduced turbulence and improved feeding. It also provided a larger vent path during burn-out. The modulus of the gating system can be estimated by:
$$ M = \frac{V}{A} $$
where \(V\) is the volume and \(A\) is the cooling surface area. A larger modulus means slower cooling. I used this concept to ensure that the gate remained liquid longer than the blade root, so that feeding could occur toward the shrinkage zone.
After these changes, I ran a new set of casting trials. Table 6 shows the final process window that produced the best results. The window is expressed as ranges because the furnace and shell conditions vary slightly from run to run.
| Parameter | Initial Trial | Optimized Trial |
|---|---|---|
| Pattern material | Solid ABS | Sparse ABS with grid |
| Shell layers | 5 | 6 full plus partial 7th |
| Edge radius | As-designed sharp | Increased leading and trailing radii |
| Burn-out ramp | Fast single ramp | Staged ramp with holds |
| Venting | Small gate opening | Enlarged gate and rounded transition |
| Pouring temperature | Standard stainless steel range | Same range with tighter control |
| Casting result | Shell cracks, flash, dimensional drift | Sound shell, clean blade, stable dimensions |
6. Results and Discussion
The optimized process produced complete ceramic shells without visible cracks after burn-out. The poured blades had clean surfaces and no significant flash at the leading or trailing edges. Dimensional inspection showed that the airfoil profile was within the trial tolerance, and the root section was sound. I did not observe the large dimensional drift that had appeared in the early trials. The improvement came from the combined effect of lower shell stress, lower internal gas pressure, and better feeding.
Table 7 compares the before-and-after results in a semi-quantitative way. The values are based on visual inspection, dimensional measurement, and sectioning of trial castings. They are not intended to be a full statistical process capability study, but they show the direction and magnitude of improvement.
| Response | Before Optimization | After Optimization | Improvement |
|---|---|---|---|
| Shell cracking frequency | High | Low | Major reduction |
| Surface flash | Moderate to severe | Minimal | Major reduction |
| Dimensional deviation | Large | Small | Improved stability |
| Edge integrity | Poor | Good | Improved |
| Shell removal | Difficult due to cracks and metal penetration | Easier and cleaner | Improved |
| Process iteration count | Multiple | Reduced | Time and cost savings |
The deformation coordination equation was especially useful because it explained why the shell cracked at the edges. It also explained why a thicker shell helped, and why a hollow pattern helped. When I increased the shell thickness, the term \(t_2\) in the denominator of the pressure equation increased. When I used a sparse pattern, the effective \(t_1\) decreased and the pattern could deform more easily. Both changes reduced the shell stress. The finite element analysis confirmed that the maximum stress was not uniformly distributed around the blade section.
I also compared my results with the general behavior of evaporative pattern casting. In evaporative pattern casting, the foam pattern is not removed before pouring. The molten metal must vaporize the foam, and the decomposition products must escape through the coating and sand. If the coating is too impermeable, gas defects and mold pressure increase. If the foam density is too high, the gas load is too large. My rapid investment casting process removes the pattern before pouring, so the gas load occurs during burn-out rather than during pouring. Nevertheless, the same principles apply: lower pattern mass, higher permeability, and controlled heating reduce defects. This cross-process insight helped me optimize the burn-out cycle and the pattern infill.
One important difference is that in evaporative pattern casting the mold is usually supported by unbonded sand. The sand can accommodate some deformation and can vent gas. In investment casting, the ceramic shell is a bonded structure. It cannot deform as easily. Therefore, the shell is more sensitive to expansion mismatch and internal pressure. This is why the deformation coordination model was so relevant to my work. The shell must be treated as a structural component, not just a refractory container.
I also examined the economics of the rapid route. The main savings come from eliminating the metal die and reducing the number of iterations. Table 8 presents a simplified cost comparison. The numbers are normalized because actual costs depend on labor, furnace, materials, and machine time. The key point is that the rapid route shifts cost from tooling to digital design and printing, and it reduces the cost of design changes.
| Cost Category | Conventional Investment Casting | Rapid Investment Casting | Evaporative Pattern Casting |
|---|---|---|---|
| Pattern tooling | High | None | None or low |
| Pattern production | Low per part at high volume | Moderate, depends on print time | Low to moderate |
| Design change cost | High | Low | Low |
| Lead time for first article | Long | Short | Short |
| Best volume range | High | Low to medium | Low to high for suitable parts |
I found that the rapid investment casting route is most valuable when the part is complex, the production volume is low, or the design is still evolving. For gas turbine blades, these conditions are common in repair, overhaul, and prototype development. The route is also useful for testing new cooling schemes or new alloys before committing to hard tooling. In that sense, it complements rather than replaces conventional investment casting. It also shares a common technical foundation with evaporative pattern casting, especially in the areas of pattern decomposition, gas venting, and dimensional control.
7. Practical Guidelines from the Work
From my trials, I developed several practical guidelines. I list them here because they may be useful to others working with rapid investment casting, expendable patterns, or evaporative pattern casting.
- Treat the ceramic shell as a pressure vessel and a thermal expansion constraint. Do not design it only for handling strength.
- Use a lower-mass pattern when possible. A sparse or hollow ABS pattern reduces gas generation and expansion load.
- Increase edge radii in the digital model before printing. Small radii create large stress concentrations in the shell.
- Use a staged burn-out cycle. Fast heating increases gas pressure and transient thermal stress.
- Provide a clear vent path through the gating system. Trapped gas can crack the shell even when the thermal expansion mismatch is moderate.
- Control drying between shell layers. Residual moisture can cause steam-driven cracks during firing.
- Use finite element analysis to rank designs, not to replace physical trials. The model is most useful for comparing edge radii, shell thickness, and pattern architecture.
- Remember the lessons of evaporative pattern casting. Pattern density, permeability, and gas evolution are central to defect control in any expendable-pattern process.
I also recommend recording the thermal cycle in detail. The shell stress depends on the temperature difference \(\Delta T\), but the transient temperature distribution depends on the heating rate. Two furnaces with the same final temperature can produce different crack behavior if their ramp rates differ. A simple thermocouple near the shell can reveal whether the actual cycle matches the intended cycle.
In addition, I suggest measuring the shell thickness at several locations after firing. The nominal number of layers does not guarantee uniform thickness. Slurry drainage and sand coverage can vary, especially near sharp edges and internal corners. If the shell is thinner at the leading edge, that is exactly where the stress is highest. A few extra layers at the edges can be more effective than adding layers everywhere.
8. Conclusion
I optimized a rapid investment casting process for gas turbine blades by combining 3D-printed ABS patterns, wax gating, ceramic shell building, controlled burn-out, and stainless steel pouring. The initial trials produced shell cracking and casting defects. I explained the cracking with a deformation coordination model and verified the stress distribution with finite element analysis. The maximum stress occurred at the leading and trailing edges, where the curvature radius is smallest. I then applied four main optimizations: increased edge radii, sparse ABS pattern architecture, increased shell thickness, and a revised burn-out cycle with improved venting. The optimized process produced sound blade castings with minimal flash and improved dimensional stability.
The work also clarified the relationship between rapid investment casting and evaporative pattern casting. Both are expendable pattern processes. Both require careful control of pattern decomposition, gas venting, and thermal expansion. In rapid investment casting, the pattern is removed before pouring; in evaporative pattern casting, it is vaporized during pouring. Despite this difference, the underlying physics of gas generation, pressure, and mold constraint is closely related. I found that insights from evaporative pattern casting were directly useful in my investment casting trials, especially regarding pattern density and permeability.
For future work, I would extend the finite element model to three dimensions, include the temperature-dependent properties of the decomposing polymer, and measure the internal pressure during burn-out. I would also apply a design-of-experiments approach to separate the effects of edge radius, shell thickness, pattern infill, and heating rate. A statistical model could then predict the risk of shell cracking before a physical trial. Finally, I would evaluate the fatigue and creep properties of the cast blades to confirm that the optimized geometry and process meet service requirements. The rapid route is not a substitute for qualification, but it is a powerful way to reduce development time and cost while maintaining the essential metallurgical and dimensional quality of investment cast gas turbine blades.