I developed the manufacturing route for a large second-stage turbine moving blade by treating every operation as part of one connected thermal, dimensional, and metallurgical system. The part envelope is approximately 351 mm by 168 mm by 89 mm, and the airfoil length is about 203 mm. The internal cavity contains a complex grid, the exhaust tail slot must be held to 1.2 mm ± 0.2 mm, and the casting must pass a water-flow check. Because the blade tip is thin, the tip region is vulnerable to distortion during solidification. The tip waist slot is close to the airfoil, with a local separation of about 2.6 mm, so shell preparation is difficult. The tenon pocket on the back side is deep, which can cause poor drying, local shell weakness, and hard-to-clean excess metal. I kept the logic of evaporative pattern casting in mind throughout the program, because evaporative pattern casting teaches that pattern removal, shell permeability, and thermal uniformity cannot be separated. Even though the production route here is investment casting with a wax pattern and a ceramic core, the evaporative pattern casting mindset helped me organize the process window, the drying strategy, and the defect-prevention logic.
I first summarized the part features that controlled the process choice. I used these data to define the wax injection strategy, the shell-building sequence, and the melting parameters. I did not treat the blade as a simple investment casting, because the combination of a fragile ceramic core, a narrow tip gap, a deep tenon pocket, and a water-flow requirement makes the process closer to a coupled system. In evaporative pattern casting, the pattern disappears while the mold is filled, and in this wax-based route, the pattern and core are removed before pouring, but the same principle applies: every hidden channel must remain open, every support must be stable, and every thermal path must be controlled.
| Feature | Nominal value or condition | Process risk | Control direction |
|---|---|---|---|
| Overall envelope | Approximately 351 mm × 168 mm × 89 mm | Large thermal mass and long feeding path | Top pouring with a split riser |
| Airfoil length | Approximately 203 mm | Tip distortion during solidification | Uniform shell strength and controlled cooling |
| Tip waist slot | Local gap about 2.6 mm | Shell bridging | Fine mullite sand in early layers |
| Tenon pocket | Deep pocket on the back side | Poor drying and excess metal | Compressed-air drying and sand removal |
| Internal cavity | Complex grid | Core cracking and core shift | Wax support at weak positions |
| Exhaust tail slot | 1.2 mm ± 0.2 mm | Dimensional deviation and flow restriction | Core stability and shell rigidity |
| Water-flow requirement | Must pass flow check | Internal blockage or rough passage | Core integrity and clean removal |
The alloy selected for the blade was K444, a nickel-base precipitation-hardening equiaxed cast superalloy. I used its reported melting range of about 1262 °C to 1319 °C as the basis for superheat calculation. K444 provides good high-temperature strength and hot-corrosion resistance below about 900 °C. I did not rely only on the alloy data sheet, because the blade geometry changes the local cooling rate, the shell modulus, and the feeding distance. I therefore treated the alloy as one part of a larger process window that also included the wax pattern, the ceramic core, the shell layer sequence, the preheat schedule, and the vacuum level. I repeated the evaporative pattern casting idea that the mold must be viewed as a temporary but critical machine: it must support the part, allow removal of the pattern and core, and transfer heat in a controlled manner.
I defined the wax injection parameters after considering both pattern fill and core survival. The target wax tank temperature was 80 °C ± 5 °C, the nozzle temperature was 80 °C ± 5 °C, the pressure was 5 bar ± 2 bar, and the flow rate was 40 cc/s. I first ran trials without a ceramic core so I could see the natural filling behavior of the wax. The surfaces showed only slight flow lines. When I added the ceramic core and a cold wax block, the flow lines disappeared, but local incomplete filling appeared near the exhaust side. I then used a short-shot method to reveal the filling sequence. I injected for 5 s, 10 s, 12 s, 13 s, 14 s, and 15 s, and I examined the partially filled patterns. During the first 10 s, the basin side and the back side advanced at nearly the same speed. In the final stage, the back side encapsulated the core faster than the basin side, which left a narrow space at the basin-side exhaust edge. The wax could not enter that space, so the basin-side exhaust edge was not completely filled.
I solved this by increasing the wax tank temperature so that the wax still had enough fluidity at the end of filling. I did not increase the temperature without limit, because excessive wax temperature can increase solidification shrinkage and impose compressive stress on the ceramic core. The balance between fillability and core stress is central to this route, and it is also a central lesson from evaporative pattern casting: the pattern or pattern precursor must remain dimensionally stable while the surrounding system is being formed. I used the adjusted window to press ten blades. After the wax patterns cooled completely, I removed the wax and inspected the ceramic cores for cracks. No cracks were found in the cores after dewaxing.
| Wax injection parameter | Initial target | Observed issue | Adjustment |
|---|---|---|---|
| Wax tank temperature | 80 °C ± 5 °C | Late-stage fill difficulty at basin-side exhaust edge | Increase within the upper part of the window |
| Nozzle temperature | 80 °C ± 5 °C | Flow-line formation when too low | Keep near 80 °C |
| Injection pressure | 5 bar ± 2 bar | Incomplete fill when pressure was low | Maintain stable pressure |
| Flow rate | 40 cc/s | Back side filled faster than basin side late in the cycle | Support with higher wax temperature |
| Core support | Filled weak positions with R839 runner wax | Risk of virtual connection | Ensure intimate support without gaps |
| Cold wax block | Used with ceramic core | Improved surface but changed late filling | Balance with temperature and short-shot data |
I also calculated the nominal fill time to connect the short-shot observations with the machine settings. For a wax volume \(V_p\) and a volumetric flow rate \(Q\), the ideal fill time is
$$ t_f = \frac{V_p}{Q} $$
For a representative pattern volume of 120 \(cm^3\) at 40 \(cm^3/s\), the ideal fill time is about 3.0 s. The actual injection includes runner and gate volumes, pressure rise, and cooling, so the complete shot is longer than this ideal value. I used the short-shot trials to identify the point at which the basin-side exhaust edge stopped filling. That point occurred late in the shot, which indicated that the problem was not the initial pressure but the loss of wax fluidity and the local flow resistance. Increasing the wax temperature reduced the viscosity and extended the time available for the narrow section to fill. The Reynolds number for the wax flow provides a useful check:
$$ Re = \frac{\rho v D_h}{\mu} $$
where \(\rho\) is density, \(v\) is velocity, \(D_h\) is hydraulic diameter, and \(\mu\) is dynamic viscosity. Because the wax is viscous and the channels are small, the flow remains laminar. The wall shear stress can be estimated as
$$ \tau_w = \mu \dot{\gamma} $$
with the apparent shear rate
$$ \dot{\gamma} = \frac{8v}{D_h} $$
These relationships helped me understand why a small increase in temperature could improve filling without requiring an unsafe pressure increase. In evaporative pattern casting, the same type of reasoning is applied to foam degradation and gas transport, but here it applies to wax viscosity and core loading. I recorded the short-shot results in a structured table so that the operator could see the filling sequence rather than only the final result.
| Short-shot time | Basin-side fill | Back-side fill | Exhaust-edge condition | Interpretation |
|---|---|---|---|---|
| 5 s | Partial | Partial | Open | Early fill stage |
| 10 s | Advanced | Advanced | Open | Basin and back sides nearly balanced |
| 12 s | Advanced | More advanced | Narrowing | Back side begins to dominate |
| 13 s | Slower | Faster | Restricted | Late-stage imbalance appears |
| 14 s | Nearly complete | Nearly complete | Partially unfilled | Basin-side exhaust edge starved |
| 15 s | Complete except local edge | Complete | Unfilled | Confirms need for higher wax fluidity |
For the gating and feeding design, I used a top-pouring arrangement because the blade length in the Z direction is about 351 mm. That length is sufficient for top pouring to feed the casting when the riser is correctly sized. I estimated the riser from the blade contour and the tenon cross-section. The tenon section length in the Z direction is about 80 mm. I set the riser width to about 75 mm and the riser thickness to about 40 mm. I adjusted these values according to the metallurgical quality required at the tenon. I chose a split riser design. First, I assembled the wax pattern with the riser connected to the tenon. Then I joined that subassembly to the remaining pattern using bonding wax. After assembly, I smoothed the transition between the part and the riser. The smooth transition is important because a sharp transition can create a hot spot, a stress concentration, or a turbulent flow path during pouring. I reviewed this assembly with the same care that I would use in evaporative pattern casting when placing gates and vents, because in both routes the gating system controls filling, feeding, and defect location.
I used a Chvorinov-type relationship to compare solidification times of different sections:
$$ t_s = B \left(\frac{V}{A}\right)^n $$
where \(t_s\) is solidification time, \(V/A\) is the section modulus, \(B\) is a mold constant, and \(n\) is an exponent usually close to 2. The riser must remain liquid longer than the casting section it feeds. Therefore, I required
$$ \left(\frac{V}{A}\right)_{riser} > \left(\frac{V}{A}\right)_{casting} $$
For the tenon region, I increased the riser modulus by selecting the 75 mm by 40 mm section and by using the split-riser assembly to reduce local heat loss. I also checked the feeding distance against the blade length. A first-order estimate of the feeding distance is
$$ L_f \approx k \sqrt{M} $$
where \(L_f\) is feeding distance, \(M\) is the modulus, and \(k\) is an alloy- and process-dependent coefficient. I did not use this estimate as a final design tool, but it helped me decide that a top-pouring route with a robust riser was more suitable than a bottom-gated route for this geometry. The final decision was confirmed by the absence of shrinkage defects after radiography.
The shell-making plan was driven by two difficult locations: the deep tenon pocket on the back side and the narrow tip gap. The tenon pocket dries poorly because air cannot easily reach the inner surfaces. If the shell remains wet or partially dried, the subsequent layers can delaminate, and the final mold can have low strength. I handled this by using compressed air after the transition layer had dried for 2 h. I directed the air into the pocket to remove moisture and to improve local drying. Before each slurry dip, I cleaned floating sand from the basin-side and back-side tenon pockets with compressed air. This step prevented sand accumulation and reduced the risk of shell layering. I also kept the pocket open so that the shell could build uniformly around the deep feature.
The tip gap is about 2.6 mm. If the sand grain size is too large, the grains can bridge across the gap instead of filling it uniformly. A bridge creates a weak, porous, or incomplete shell region. I therefore selected a finer sand for the early shell layers. Layers 2 and 3 used mullite sand of 30–60 mesh. Before each of those layers, I cleaned floating sand from the narrow space. I repeated this cleaning step until the narrow region was clear. I then continued shell building with coarser sand for the remaining layers. The complete shell was built to 10 layers and then sealed. I selected the layer count and materials to provide enough strength for a casting with a pouring mass of about 15 kg. I also kept the evaporative pattern casting principle of permeability in view: the shell must be strong, but it must also allow the core and pattern removal media to escape and must not trap gas or debris.
| Shell layer | Slurry powder | Slurry viscosity | Stucco sand | Purpose |
|---|---|---|---|---|
| 1 | Zircon, 325 mesh | 36 s | Zircon, 100 mesh | Primary face coat |
| 2–3 | Mullite, 200 mesh | 15 s | Mullite, 30–60 mesh | Narrow-gap filling and bridging prevention |
| 4–10 | Mullite, 200 mesh | 12 s | Mullite, 16–30 mesh | Structural strength and thermal stability |
| 11 | Mullite, 200 mesh | 10 s | None | Seal coat |
I quantified the bridging risk with a simple index:
$$ B = \frac{d_p}{g} $$
where \(d_p\) is the characteristic sand particle diameter and \(g\) is the local gap. When \(B\) is too large, bridging is likely. For the tip gap of about 2.6 mm, a 30–60 mesh mullite sand has a particle size range that is small enough to enter the gap but large enough to build a coherent shell. I also considered the capillary number and the slurry penetration behavior, because the slurry must wet the narrow surfaces before the sand is applied. A simplified drying time scaling is
$$ t_d \propto \frac{L^2}{D} $$
where \(L\) is the diffusion path length and \(D\) is the effective diffusivity. The deep tenon pocket has a long diffusion path, so its drying time is much longer than that of an open surface. The compressed-air step reduced the effective path length and improved moisture removal. I used this relationship to justify the 2 h transition-layer dry before air was applied. I also checked the shell modulus for the overall mold:
$$ M_s = \frac{V_s}{A_s} $$
where \(V_s\) is the shell volume and \(A_s\) is the shell surface area. The 10-layer shell with a seal coat provided a modulus and strength sufficient for the 15 kg pour. The shell did not show local collapse or run-out during pouring. I inspected the shell after knockout and found no evidence of major bridging at the tip or delamination at the tenon pocket.
| Shell issue | Cause | Countermeasure | Verification |
|---|---|---|---|
| Tenon pocket poor drying | Long diffusion path and trapped moisture | Compressed air after 2 h of transition-layer drying | No shell delamination after pour |
| Floating sand in pocket | Sand accumulation during dipping | Compressed-air cleaning before each slurry | Clean pocket surfaces |
| Tip bridging | Coarse sand in 2.6 mm gap | 30–60 mesh mullite in layers 2 and 3 | Open gap and uniform shell |
| Insufficient shell strength | Thin shell or incomplete layer build | 10 layers plus seal coat | No run-out or mold deformation |
| Gas trapping | Low permeability or blocked channels | Controlled layer sequence and clean narrow spaces | Sound casting after radiography |
I applied a shell preheat schedule with the sprue cup facing downward. The preheat temperature was 1100 °C ± 10 °C, and the holding time was at least 1 h. This schedule removed residual moisture and volatiles, stabilized the shell, and reduced the thermal shock during pouring. I maintained a vacuum level below 10 Pa during melting and pouring. I refined K444 at 1520 °C ± 10 °C for at least 3 min. The pouring temperature was set to 1440–1470 °C. The pouring speed was controlled to complete the fill in less than 2 s. After pouring, I held the mold for at least 10 min before taking it out for cooling. The casting cooled for more than 6 h before shell removal. These conditions were selected to balance superheat, fluidity, feeding, and grain structure. I compared the superheat with
$$ \Delta T_s = T_p – T_l $$
where \(T_p\) is the pouring temperature and \(T_l\) is the liquidus temperature. Using a liquidus near 1319 °C and a pouring temperature of 1440–1470 °C gives a superheat of about 121–151 °C. That range is high enough to fill the thin exhaust tail slot and the complex internal passages, but not so high that the shell or core is damaged. I also estimated the cooling rate as
$$ \dot{T} = \frac{T_p – T_m}{t_f} $$
where \(T_m\) is the mold temperature and \(t_f\) is the local solidification time. The shell preheat of 1100 °C reduces the initial temperature difference and helps avoid steep thermal gradients at the tip. In evaporative pattern casting, thermal gradients are also critical because the foam pattern decomposes and the gas must escape while the metal advances. Here the pattern is gone before pouring, but the same need for controlled thermal gradients remains.
| Melting parameter | Set point | Reason |
|---|---|---|
| Shell preheat orientation | Sprue cup downward | Uniform heating and stable positioning |
| Shell preheat temperature | 1100 °C ± 10 °C | Remove moisture and reduce thermal shock |
| Shell preheat holding time | At least 1 h | Through-thickness temperature uniformity |
| Alloy | K444 nickel-base cast superalloy | High-temperature strength and corrosion resistance |
| Melting range | 1262–1319 °C | Basis for superheat calculation |
| Refining temperature | 1520 °C ± 10 °C | Melt cleanliness and homogeneity |
| Refining time | At least 3 min | Inclusion flotation and temperature equalization |
| Pouring temperature | 1440–1470 °C | Filling thin sections and internal passages |
| Vacuum level | Below 10 Pa | Prevent gas defects and oxidation |
| Pouring speed | Less than 2 s | Avoid premature freezing and cold shuts |
| Post-pour hold | At least 10 min | Complete filling and initial solidification |
| Cooling before knockout | More than 6 h | Reduce residual stress and cracking risk |
I estimated the pouring time from the pour weight and the effective pour rate:
$$ t_p = \frac{W}{Q_p} $$
For a pour mass of about 15 kg and a target pour time below 2 s, the effective pour rate must be high enough to fill the mold before the thin exhaust tail slot freezes. I used a controlled tilt-pour practice to maintain a smooth metal front. A slow pour increases the risk of cold shuts, while a violent pour increases turbulence, mold erosion, and gas entrapment. I kept the pouring speed within a window that produced complete filling without visible turbulence defects. I also monitored the vacuum level because poor vacuum can cause gas porosity, and excessive vacuum can increase metal vaporization and shell damage. The vacuum conductance relationship
$$ Q = C(P_1 – P_2) $$
where \(Q\) is gas flow, \(C\) is conductance, and \(P_1-P_2\) is pressure difference, helped me understand why the shell must remain permeable. A dense but impermeable shell can block gas removal, which is a familiar concern in evaporative pattern casting as well. Although the actual gas source differs, the need for controlled evacuation and venting is common to both processes.
I controlled the Niyama criterion as a qualitative check for shrinkage porosity:
$$ N_y = \frac{G}{\sqrt{R}} $$
where \(G\) is the temperature gradient and \(R\) is the cooling rate. A higher \(N_y\) generally indicates a lower risk of shrinkage porosity. The top-pouring split-riser design, the shell preheat, and the controlled pouring temperature all helped maintain a favorable gradient from the riser toward the blade. I did not rely on a single criterion, because the blade has thin walls, a complex core, and a large length-to-thickness ratio. Instead, I combined the Niyama concept with the Chvorinov modulus, the superheat calculation, and the radiography results. This combined approach is also consistent with evaporative pattern casting practice, where filling, gas evolution, and solidification interact strongly.
After pouring and cooling, I removed the shell and cut off the riser. The castings showed no missing metal and no incomplete filling. I then used blue-light scanning to check the profile dimensions. The dimensional requirement was +0.75 mm / −0.5 mm. The measured dimensions were within tolerance on both the basin side and the back side. The shell layer count and shell strength were therefore sufficient to maintain the blade shape. I also verified that the tip region did not show unacceptable distortion. The dimensional capability can be expressed with
$$ C_{pk} = \min\left(\frac{USL-\mu}{3\sigma}, \frac{\mu-LSL}{3\sigma}\right) $$
where \(USL\) is the upper specification limit, \(LSL\) is the lower specification limit, \(\mu\) is the process mean, and \(\sigma\) is the process standard deviation. I used this index as a summary of the dimensional results. The measured data indicated that the process was centered well enough for the profile tolerance. I did not treat the dimensional result as separate from the shell and core behavior, because the tip gap, core support, and shell rigidity all influence the final profile. This is another point where the evaporative pattern casting viewpoint is useful: dimensional accuracy is not only a machining or inspection issue, but a result of the entire pattern-mold-metal system.
| Inspection item | Method | Acceptance condition | Result |
|---|---|---|---|
| Missing metal or incomplete fill | Visual inspection after knockout | No short runs or unfilled features | Pass |
| Profile dimensions | Blue-light scanning | +0.75 mm / −0.5 mm | Within tolerance on basin and back sides |
| Surface defects | Fluorescence inspection | No over-limit indications | Pass |
| Internal defects | Radiographic inspection | No over-limit indications | Pass |
| Water-flow performance | Flow check | Meets flow requirement | Pass |
| Exhaust tail slot | Dimensional check | 1.2 mm ± 0.2 mm | Within tolerance |
| Tenon tooth form | Wire-cut after inspection | No surface or sub-surface defects | Pass |
I performed radiographic inspection according to HB/Z60 and found no over-limit indications. The fluorescence inspection also showed a good surface condition. The castings did not show the typical defects that I had identified at the start: wax pattern core cracking, shell bridging at the tip, poor drying at the tenon pocket, incomplete filling at the exhaust edge, and shrinkage in the tenon. The success came from adjusting the wax temperature, supporting the weak core positions with R839 runner wax, cleaning the narrow tip space before each early shell layer, using 30–60 mesh mullite sand in layers 2 and 3, drying the deep pocket with compressed air, building 10 shell layers plus a seal coat, and controlling the melting and pouring parameters. I also maintained a process log so that each change could be traced to a specific defect or risk. I reviewed the process log with the same discipline that evaporative pattern casting requires, because in both routes small changes in pattern, mold, or thermal conditions can produce large changes in the final casting.
The ceramic core was a major source of risk. Its exhaust-side grid is complex, and the cylindrical core print on the tenon side is fragile. The thinnest tip region is about 3 mm. Before wax injection, I filled the weak positions with R839 runner wax. This support had to be intimate; a virtual connection would allow relative movement, which could crack the core or create a dimensional shift. I checked the support by visual inspection and by handling the assembled core. I also used the short-shot trials to confirm that the wax front did not push the core unevenly. The core survived dewaxing in all ten trial blades. The absence of core cracks after wax removal was a strong indication that the wax injection pressure and temperature were within a safe window. I used the following core stress estimate as a qualitative guide:
$$ \sigma_c = \frac{E_c \alpha_c \Delta T}{1-\nu_c} $$
where \(\sigma_c\) is the thermal stress in the ceramic core, \(E_c\) is the elastic modulus, \(\alpha_c\) is the coefficient of thermal expansion, \(\Delta T\) is the temperature change, and \(\nu_c\) is Poisson’s ratio. A high wax temperature increases \(\Delta T\) during cooling and can increase stress on the core. Therefore, I raised the wax temperature only enough to solve the late-stage filling problem. I did not use the highest possible temperature. This balance is central to the process, and it is also a theme that appears in evaporative pattern casting, where pattern degradation and mold stability must be balanced.
I also considered the shrinkage of the wax pattern. The linear contraction can be estimated by
$$ \Delta L = \alpha L \Delta T $$
where \(\alpha\) is the coefficient of thermal expansion, \(L\) is the original length, and \(\Delta T\) is the temperature change. For a long airfoil of about 203 mm, even a small value of \(\alpha\) can produce a measurable length change. I compensated for this by controlling the wax temperature, the cooling time, and the handling of the pattern after injection. I did not allow the pattern to cool unevenly on a cold fixture. I also maintained a consistent time between injection and assembly. The dimensional scan results confirmed that these controls were adequate. In evaporative pattern casting, pattern shrinkage is also a major dimensional variable, so the same principle of controlling thermal history applies.
The shell drying process was also linked to dimensional stability. If the shell dries unevenly, it can warp or develop local weakness. I controlled the drying time for each layer. For the transition layer, I waited 2 h before applying compressed air to the tenon pocket. For the narrow tip region, I cleaned floating sand before each layer. I used the same slurry viscosity for the layers that required fine sand. I measured viscosity with a standard flow cup and kept it within the specified range. A viscosity that is too high produces a thick slurry that cannot penetrate the narrow gap. A viscosity that is too low produces a thin slurry that does not build enough green strength. I found that the 15 s viscosity for layers 2 and 3 and the 12 s viscosity for layers 4–10 gave a good balance. The face coat used zircon with a 36 s viscosity. The seal coat used a 10 s viscosity. These values are summarized in the shell table.
I also used a simple mass balance for the slurry pickup:
$$ m_{pickup} = \rho_s V_{retained} $$
where \(m_{pickup}\) is the retained slurry mass, \(\rho_s\) is the slurry density, and \(V_{retained}\) is the retained volume. I did not need a precise value for every layer, but the concept helped me understand why narrow gaps can become blocked if the slurry is too viscous or if the draining time is too short. In evaporative pattern casting, coating weight and permeability are also critical, so the same principle of controlling coating thickness and drainage applies.
The melting operation required tight control because K444 is sensitive to temperature and vacuum. I refined at 1520 °C ± 10 °C for at least 3 min. This step improved melt homogeneity and allowed inclusions to float. I poured at 1440–1470 °C. If the temperature was too low, the thin exhaust tail slot and the internal passages could freeze before filling. If the temperature was too high, the ceramic core could be eroded, the shell could be damaged, and the grain structure could become too coarse. I used a vacuum below 10 Pa to reduce gas porosity and oxidation. I held the mold for at least 10 min after pouring to allow the gate and riser to feed the casting. I then cooled the mold for more than 6 h before knockout. This slow cooling reduced residual stress in the thin tip and in the tenon. I also verified that the shell preheat was uniform before pouring. A cold spot in the shell can cause local premature freezing, which is especially harmful in the narrow tip region.
I evaluated the process capability with the following general relation:
$$ C_p = \frac{USL – LSL}{6\sigma} $$
where \(C_p\) is the potential capability and \(\sigma\) is the process standard deviation. When the process is centered, \(C_p\) and \(C_{pk}\) are close. The blue-light scanning data showed that the process was centered well enough for the profile tolerance. I did not rely on a single measurement, because the blade has many features. I checked the tip, the airfoil, the platform, and the tenon transition. The most critical dimensional feature was the exhaust tail slot, which must remain within 1.2 mm ± 0.2 mm. The core support and shell rigidity were the main controls for that feature. The water-flow check also confirmed that the internal passages were clear. A blocked or rough passage would reduce flow and would indicate a core-related defect. The flow check is therefore both a functional test and a process audit. I kept the evaporative pattern casting idea that internal channels must be treated as first-class features, because in both processes a hidden channel can determine whether the part works.
| Process step | Critical variable | Target or range | Effect on blade |
|---|---|---|---|
| Core preparation | Weak-position support | R839 runner wax, no virtual connection | Prevents core cracking and shift |
| Wax injection | Wax tank temperature | 80 °C ± 5 °C | Controls fill and core stress |
| Wax injection | Nozzle temperature | 80 °C ± 5 °C | Reduces flow lines |
| Wax injection | Pressure | 5 bar ± 2 bar | Fills thin sections without overloading core |
| Wax injection | Flow rate | 40 cc/s | Balances basin-side and back-side fill |
| Assembly | Riser design | Split riser, 75 mm × 40 mm | Improves feeding and metallurgical quality |
| Shell building | Tip layer sand | 30–60 mesh mullite | Prevents bridging in 2.6 mm gap |
| Shell building | Tenon pocket drying | Compressed air after 2 h | Prevents delamination and excess metal |
| Shell building | Layer count | 10 layers plus seal | Provides strength for 15 kg pour |
| Preheat | Temperature and time | 1100 °C ± 10 °C, at least 1 h | Reduces thermal shock |
| Melting | Refining | 1520 °C ± 10 °C, at least 3 min | Improves cleanliness |
| Pouring | Pouring temperature | 1440–1470 °C | Fills thin walls and internal passages |
| Pouring | Pouring speed | Less than 2 s | Avoids cold shuts and turbulence |
| Cooling | Post-pour hold | At least 10 min | Supports feeding |
| Cooling | Knockout delay | More than 6 h | Reduces residual stress |
I compared the final process with the initial risks and found that each major risk had a direct control. The ceramic core cracking risk was controlled by wax support, wax temperature, and injection pressure. The tip shell bridging risk was controlled by fine mullite sand and cleaning. The tenon pocket drying risk was controlled by compressed air and pre-dip cleaning. The exhaust-edge incomplete filling risk was controlled by higher wax fluidity and the short-shot study. The dimensional risk was controlled by shell rigidity and uniform cooling. The internal defect risk was controlled by vacuum, refining, pouring temperature, and feeding. The water-flow risk was controlled by core integrity and clean removal. I also used the evaporative pattern casting concept of a robust process window, meaning that the process should tolerate small variations without producing defects. I did not set the parameters at the edge of the allowable range. Instead, I kept them near the center of the proven window.
I derived an approximate heat balance for the mold and metal:
$$ Q = \dot{m} c_p \Delta T $$
where \(Q\) is heat flow, \(\dot{m}\) is mass flow rate, \(c_p\) is specific heat, and \(\Delta T\) is temperature change. This relation helped me confirm that the pouring temperature and pour rate were consistent with the mold preheat. A higher preheat reduces the heat loss from the metal front and helps fill thin sections. However, too high a preheat can reduce the temperature gradient needed for feeding. I therefore used the 1100 °C preheat as a compromise. I also used the thermal diffusivity
$$ \alpha = \frac{k}{\rho c_p} $$
where \(k\) is thermal conductivity, \(\rho\) is density, and \(c_p\) is specific heat. The shell and core have different diffusivities, which means the heat extraction is not uniform. The ceramic core can act as a heat sink in some regions and as an insulator in others. I considered this when I set the core support and shell layers. In evaporative pattern casting, the mold fill and gas evolution also create non-uniform thermal conditions, so a robust process must account for local heat transfer.
I also estimated the solidification modulus of the tip and the tenon. For the tip, the thin wall freezes quickly, so it needs a high pouring temperature and a fast fill. For the tenon, the section is thicker, so it needs a riser that stays liquid longer. The split riser provided that. The top-pouring arrangement allowed the metal to enter from above and feed downward. The smooth transition between the blade and the riser reduced turbulence and avoided a sharp hot spot. I inspected the radiographs for shrinkage at the tenon and found none over the limit. I also inspected the tip for cold shuts and incomplete fill and found none. The fluorescence inspection showed no surface indications over the limit. The blue-light scan showed the profile within tolerance. The water-flow check passed. These results confirmed that the process window was robust.
| Risk | Primary control | Secondary control | Final verification |
|---|---|---|---|
| Core cracking during wax injection | R839 runner wax support | Controlled wax temperature and pressure | No cracks after dewaxing |
| Wax pattern incomplete fill | Short-shot study | Higher wax tank temperature | Complete pattern fill |
| Shell bridging at tip | 30–60 mesh mullite layers 2–3 | Cleaning before each layer | Open tip gap and sound shell |
| Tenon pocket poor drying | Compressed air after 2 h | Pre-dip floating sand removal | No delamination or excess metal |
| Cold shuts or incomplete fill | Pouring temperature 1440–1470 °C | Pouring speed less than 2 s | No missing metal |
| Gas porosity | Vacuum below 10 Pa | Shell permeability | Radiography pass |
| Shrinkage porosity | Split riser with adequate modulus | Shell preheat and mold hold | Radiography pass |
| Dimensional deviation | Stable shell and core support | Controlled cooling | Blue-light scan within tolerance |
| Blocked internal passage | Core integrity and clean removal | Water-flow check | Flow test pass |
I recorded the final process window as a set of controlled parameters. The wax tank temperature and nozzle temperature were both 80 °C ± 5 °C. The injection pressure was 5 bar ± 2 bar, and the flow rate was 40 cc/s. The shell used a zircon face coat, mullite layers 2–3 with 30–60 mesh sand, mullite layers 4–10 with 16–30 mesh sand, and a seal coat. The shell was preheated to 1100 °C ± 10 °C for at least 1 h. The alloy was K444. The refining temperature was 1520 °C ± 10 °C for at least 3 min. The pouring temperature was 1440–1470 °C. The vacuum was below 10 Pa. The pouring speed was less than 2 s. The post-pour hold was at least 10 min. The cooling time before knockout was more than 6 h. These parameters produced castings that passed fluorescence inspection, radiographic inspection, dimensional inspection, and water-flow testing. I also confirmed that the exhaust tail slot remained within 1.2 mm ± 0.2 mm.
I concluded that the successful route depended on integration rather than on any single parameter. The wax injection had to fill the pattern without cracking the core. The shell had to support the narrow tip without bridging and dry the deep pocket without delamination. The melt had to fill the thin sections without overheating the core or the shell. The cooling had to avoid residual stress and distortion. The inspection had to confirm that the internal passages were clear and that the dimensions were within tolerance. I used the evaporative pattern casting principle of treating pattern, mold, and metal as one system. In evaporative pattern casting, the foam pattern decomposes and the gas must escape. In this investment casting route, the wax and core are removed before pouring, but the same requirement for a stable, permeable, and thermally balanced mold applies. I therefore used the phrase evaporative pattern casting as a reminder throughout the program: every hidden path must remain open, every support must be stable, and every thermal path must be controlled. The final result was a qualified second-stage moving blade with no over-limit indications in fluorescence or radiography and with dimensions and flow performance that met the requirements.
| Final process window | Value |
|---|---|
| Wax tank temperature | 80 °C ± 5 °C |
| Nozzle temperature | 80 °C ± 5 °C |
| Injection pressure | 5 bar ± 2 bar |
| Injection flow rate | 40 cc/s |
| Core support wax | R839 runner wax at weak positions |
| Shell layers | 10 layers plus seal coat |
| Tip fine sand | 30–60 mesh mullite in layers 2–3 |
| Tenon pocket drying | Compressed air after 2 h of transition drying |
| Shell preheat | 1100 °C ± 10 °C for at least 1 h |
| Alloy | K444 |
| Refining | 1520 °C ± 10 °C for at least 3 min |
| Pouring temperature | 1440–1470 °C |
| Vacuum | Below 10 Pa |
| Pouring speed | Less than 2 s |
| Post-pour hold | At least 10 min |
| Cooling before knockout | More than 6 h |
For future work, I would keep the same evaporative pattern casting-inspired systems view. I would continue to monitor wax viscosity, core support integrity, shell drying, and thermal gradients. I would expand the short-shot study to more time steps and use pressure sensors to record the injection profile. I would also use additional thermal couples in the shell to measure the preheat uniformity and the cooling rate. I would maintain the radiographic and fluorescence inspections as the primary acceptance tests and continue to use blue-light scanning for dimensional control. I would also keep the water-flow test as a functional check. The process is now stable, but the blade remains a complex casting, so continuous monitoring is necessary. The key lesson is that a successful second-stage moving blade is not produced by a single optimized parameter but by a controlled network of parameters. That network includes the wax, the ceramic core, the shell, the alloy, the vacuum, the pouring, and the cooling. I used the logic of evaporative pattern casting to keep that network visible, and the resulting castings met the required quality level.