A K424 nickel-base superalloy plate for a new aero-engine exhaust nozzle measures 357 mm × 102 mm × 33 mm. Its base plate carries 33 transverse ribs and 8 longitudinal ribs, with 17 through-windows and five pairs of 5 mm thick bosses. The thinnest wall is 0.7 mm, the ribs are about 1 mm thick, and the bosses are locally more than five times thicker than the surrounding web. Only the boss holes are machined; most surfaces are required to be net-shape. This combination of a large projected area, thin webs, abrupt section changes, and tight dimensional requirements creates a difficult casting problem: thin sections are prone to cold shuts and underfill, while rib intersections and bosses become hot spots that feed poorly and are prone to shrinkage porosity. K424 also has limited weldability, so shrinkage or cracking cannot be recovered by weld repair. Acceptance therefore depends on preventing these defects during solidification rather than correcting them afterward.
Alloy and Component Constraints
K424 is a high-Al, high-Ti, low-density cast nickel-base superalloy, roughly equivalent to the Russian grade VZhL-12U. Its strengthening relies on a high volume fraction of ordered γ′ precipitates, which gives good high-temperature strength, plasticity, and casting processability. The nominal composition is given in Table 1. The alloy density is 7.87 g/cm³, the liquidus temperature is 1344°C, the solidus temperature is 1234°C, and the coefficient of thermal expansion is about 1.4 × 10−5/°C. The freezing range is therefore about 110°C, which is wide enough for a mushy zone to persist during solidification. In a thin-wall casting, the interdendritic liquid in this mushy zone must be fed through narrow channels; once those channels freeze, isolated liquid pockets contract and form shrinkage porosity.
| Element | C | Cr | Al | Ti | Mo | W | Co | Nb | V | Ni |
|---|---|---|---|---|---|---|---|---|---|---|
| wt% | 0.17 | 9.3 | 5.3 | 4.5 | 3.0 | 1.4 | 13.5 | 0.7 | 0.75 | Balance |
The alloy is used for parts operating around 950–1000°C, including turbine blades, guide vanes, and exhaust-nozzle components. At around 850°C, long-term exposure can promote σ-phase precipitation, which is detrimental to toughness and stability. Composition control, melting practice, and thin-wall section design are used to reduce that risk. For the plate component, the more immediate issue is castability: the wide freezing range, the high γ′-forming element content, and the 0.7–5 mm section contrast all work against uniform feeding. The process must therefore establish a thermal gradient that keeps the gating liquid long enough to feed the last solidifying regions.
Process Route and Thermal Design
The manufacturing route follows the standard investment-casting sequence: wax injection, pattern assembly, ceramic shell build, dewaxing, shell firing, shell preheating, vacuum induction melting and gravity pouring, controlled vacuum holding, shell breakout, gate removal, cleaning, inspection, and dimensional verification. The plate is poured vertically with a side-gating system and two parts per cluster. A 30 mm × 30 mm square sprue is used, and the pouring cup has a flanged lip to prevent sand or shell fragments from entering the cavity during dewaxing and firing. Internal gates are placed at the boss and rib hot spots rather than on the thin web, because the gate must feed the heavier sections without creating a new thermal concentration on the 0.7–1 mm walls.
The thermal model included the plate, gating, a 12 mm ceramic shell, and a 30 mm insulation blanket. Local mesh refinement of 1.0–1.5 mm was used on the plate, with coarser mesh on the gating and pouring cup. The shell preheat in the initial simulation was 1050°C, and gravity pouring was assumed. These inputs were used to compare filling behavior, solidification sequence, and shrinkage-porosity tendency before physical trials were run.
Gating and Solidification Simulation
Two gating variants were compared. Variant I used a local runner section P of 26 mm × 23 mm × 7 mm and a gate section M 58 mm long, 12 mm wide, and 4–12 mm thick. Variant II enlarged P to 28 mm wide and 10 mm thick and lengthened M from 58 mm to 68 mm. The purpose was not to increase the total poured weight indiscriminately, but to keep a liquid feed channel open to the rib junctions near the sprue, where the thin ribs otherwise freeze before the thicker junction. The simulated response is summarized in Table 2.
| Variant | Local change | Fill time (s) | Solidification time (s) | Predicted shrinkage |
|---|---|---|---|---|
| I | P: 26 mm × 23 mm × 7 mm; M: 58 mm long, 12 mm wide, 4–12 mm thick | 2.55 | 79 | Multiple sites at rib junctions near the sprue |
| II | P width 23→28 mm and thickness 7→10 mm; M length 58→68 mm | 2.50 | 98 | Markedly reduced |
Both variants filled stably. Metal entered the cavity from the sprue, rose through the side gates, and filled the plate from the lower region toward the upper rib network. No significant splashing or gas entrapment was predicted, and the vent path remained open until the pouring cup filled. The difference appeared during solidification. In variant I, the thin ribs connecting the central feed path to the side junctions solidified early, isolating the thicker junction. The isolated pool then contracted without liquid replenishment, producing scattered shrinkage porosity at the intersection of longitudinal and transverse ribs. The affected locations were symmetric about the central axis and concentrated near the pouring-cup end.
In variant II, the enlarged runner and longer gate kept the junction connected to the gating for a longer period. The freezing front advanced from the far edges and thin web toward the gate, so the casting solidified before the gating and the last liquid remained in the feed system. The predicted shrinkage was substantially reduced. The longer solidification time of 98 s, compared with 79 s in variant I, also lowered the cooling rate and thermal gradients in the thin plate, which is favorable for reducing cracking. Variant II was therefore selected for the casting trials.
Pouring temperature was examined separately at 1520°C, 1500°C, and 1480°C with the shell preheat held at 1050°C. At 28.32 s after pouring, the critical rib-junction region was coolest in the 1480°C case and hottest in the 1520°C case. The predicted shrinkage followed the same order: the 1520°C casting showed the largest number of shrinkage sites, the 1500°C casting showed fewer, and the 1480°C casting was essentially free of predicted shrinkage at the rib junctions. Higher superheat increases the total liquid contraction that must be fed and prolongs the mushy zone; in a thin-wall plate with a limited feed path, that extra contraction is difficult to compensate and appears as porosity.
Wax Pattern Formation
The wax pattern is the first geometric replication of the plate, so its surface quality, dimensional stability, and freedom from shrinkage depressions directly affect the final casting. The component was injected with F28-44B, a filled medium-melting wax with good flow, low ash, and low shrinkage. A 100 t hydraulic wax-injection machine was used because the die is large and requires independent upper and lower die clamping and automatic ejection. The die was heated above 35°C before injection. Four parameter sets were evaluated, as shown in Table 3.
| Scheme | Wax temperature (°C) | Injection pressure (MPa) | Hold time (s) | Observation |
|---|---|---|---|---|
| 1 | 65–70 | 1.0–1.5 | 25–35 | Incomplete fill at thin edges remote from the injection point |
| 2 | 68–73 | 1.2–1.8 | 25–35 | Rib underfill and depression on the boss face |
| 3 | 68–73 | 1.5–2.0 | 25–35 | Complete fill and good surface, but visible boss depression |
| 4 | 68–75 | 1.5–2.0 | 30–40 | Less boss depression, but cracking during ejection |
At the lowest wax temperature and pressure, the melt reached the thin edges after those regions had already cooled, leaving underfill. Raising the temperature and pressure improved the rib fill, but the 5 mm boss remained a local hot spot: the wax there stayed soft after the hold stage and contracted during subsequent cooling, producing a depression on the boss face. Extending the hold time reduced the depression but made the thin ejected edges brittle and prone to cracking. Parameter adjustments alone therefore could not eliminate the boss depression.
The solution was a cold wax core placed in the boss cavity. The core was shaped to the boss outline and left a 2 mm wax layer between the core and the die wall. Small locating spurs on the core prevented it from being displaced by the injection pressure. The core reduced the effective wax section at the boss, removed the local thermal mass, and allowed the surrounding wax to cool and solidify more uniformly. With the core in place, the boss depression disappeared. The final wax-injection window was 68–73°C, 1.5–2.0 MPa, and 25–35 s, with the cold core used at the thick boss locations.
Ceramic Shell Build
The shell must provide enough green strength for handling, enough hot strength to resist the molten alloy, sufficient permeability for gas escape, and enough compliance to avoid restraining the casting during contraction. Three binders were considered. Water glass is inexpensive and fast-drying but produces a rougher surface and lower dimensional accuracy. Ethyl silicate hydrolysate gives good slurry stability and high shell strength, but requires hydrolysis, has a higher cost, and needs controlled drying conditions. Silica sol requires no chemical hardening and gives excellent high-temperature strength and stability, but wets the wax poorly and dries slowly. A hybrid silica-sol/ethyl-silicate shell was therefore used: silica sol for the face and selected backing layers, ethyl silicate for alternating backing layers and the final seal.
The shell was built at 23–27°C and 50–70% relative humidity. The face coat used silica sol with corundum powder and corundum stucco. Backing layers alternated between silica sol and ethyl silicate hydrolysate, with bauxite powder and coal-gangue stucco. The optimized schedule is given in Table 4. Viscosity is expressed in flow-cup seconds.
| Step | Binder | Viscosity (s) | Refractory filler | Stucco |
|---|---|---|---|---|
| Pre-coat | Silica sol | 25–30 | Corundum powder | None |
| Layer 1 (face) | Silica sol | 35–40 | Corundum powder | Corundum sand |
| Layer 2 | Ethyl silicate hydrolysate | 10–20 | Corundum powder | Corundum sand |
| Layer 3 | Silica sol | 10–20 | Bauxite powder | Coal-gangue sand |
| Layer 4 | Ethyl silicate hydrolysate | 10–20 | Bauxite powder | Coal-gangue sand |
| Layer 5 | Silica sol | 10–20 | Bauxite powder | Coal-gangue sand |
| Layer 6 | Ethyl silicate hydrolysate | 10–20 | Bauxite powder | Coal-gangue sand |
| Layers 7–8 | Silica sol | 10–20 | Bauxite powder | Coal-gangue sand |
| Seal | Ethyl silicate hydrolysate | 10–20 | Bauxite powder | None |
The face-coat viscosity required particular attention. At 25–30 s, the slurry drained unevenly on the large flat base: it accumulated at the lower end and remained thin at the upper end. The thin region had a more open stucco packing, allowing metal penetration and burn-on. Raising the face-coat viscosity to 35–40 s improved coverage and eliminated the burn-on, but the more viscous slurry then bridged the narrow rib intersections and trapped air bubbles. Those bubbles produced small spherical metal nodules on the rib side of the casting.
The final solution was a two-stage face coat. A 25–30 s silica-sol pre-coat was applied first at the rib intersections, where the slurry had to penetrate a narrow gap. The main face coat at 35–40 s was then applied over the whole pattern. The pre-coat prevented bubble entrapment at the intersections, while the higher-viscosity main coat maintained uniform coverage on the flat base. The shell was then built through the alternating backing sequence and sealed with a final slurry coat to prevent loose stucco from entering the mold during handling.
Shell Preheat and Pouring Window
Melting and pouring were carried out in a three-chamber vacuum induction furnace. The alloy was melted and refined at 1590 ± 5°C, then allowed to cool to the pouring temperature before gravity casting. The shell was wrapped in a 30 mm insulation blanket, placed in a sand box to keep it vertical, and preheated in a high-temperature furnace. It was transferred to the mold chamber within 20–25 s. After pouring, the shell was held under vacuum for 10 min before the chamber was opened.
Shell preheat controls both fill and thermal stress. A simulation of a 1050°C shell in the mold chamber showed a drop to 940–950°C after 170 s, an average loss of about 100°C. Because the transfer and evacuation steps consume time, the furnace set point must be higher than the intended shell temperature at pouring. Three preheat levels were tested with the same refining temperature of 1590 ± 5°C, pouring temperature of 1480 ± 5°C, and pouring time of 2–3 s. Six castings were poured for each condition.
| Shell preheat (°C) | Castings | Result |
|---|---|---|
| 1050 ± 5 | 6 | Cold shuts at boss edges and ribs in all castings |
| 1100 ± 5 | 6 | No cold shuts or other visible metallurgical defects |
| 1150 ± 5 | 6 | 4 castings cracked |
At 1050°C, the metal filled the cavity but cooled too much before the last thin sections merged, producing cold shuts at the boss edges and rib junctions. At 1100°C, the additional superheat in the shell was sufficient to eliminate those cold shuts without causing cracking. At 1150°C, solidification slowed further; the casting remained hot and weak for longer, and the thermal contraction and restrained shrinkage produced cracks in four of six parts. The shell preheat was therefore set at 1100 ± 5°C.
Pouring temperature was selected from the relation
$$T_{\mathrm{C}} = T_{\mathrm{L}} + \Delta T$$
where \(T_{\mathrm{C}}\) is the pouring temperature, \(T_{\mathrm{L}}\) is the liquidus temperature of 1344°C, and \(\Delta T\) is the superheat. A superheat of 110–140°C places the theoretical range at 1454–1484°C. The plate is large and thin, so a value near the upper part of this range was expected to provide adequate fill while limiting shrinkage. Three pouring temperatures were tested at a shell preheat of 1100 ± 5°C, refining temperature of 1590 ± 5°C, and pouring time of 2–3 s. Each condition was run as two batches of 20 castings.
| Pouring temperature (°C) | Batch 1 accepted | Batch 2 accepted | Total accepted | Acceptance rate | Dominant defect |
|---|---|---|---|---|---|
| 1500 ± 5 | 13/20 | 14/20 | 27/40 | 67.5% | Shrinkage porosity larger than 4 mm |
| 1480 ± 5 | 18/20 | 18/20 | 36/40 | 90% | No shrinkage above 4 mm; remaining rejects were shell-related |
| 1460 ± 5 | 15/20 | 14/20 | 29/40 | 72.5% | Underfill at corners and thin edges |
At 1500°C, the higher superheat increased volumetric contraction and extended the mushy zone. The rib-junction hot spots remained liquid longer, but the available feed path was still limited, so the extra contraction appeared as shrinkage porosity. At 1460°C, shrinkage was suppressed, but the metal lost too much heat before reaching the corners; underfill accounted for about 80% of the rejected castings. The 1480°C condition provided the best balance: the thin sections filled completely, the rib junctions were fed, and no shrinkage above the 4 mm acceptance limit was found. The four rejected 1480°C castings were associated with shell bubbles, uneven slurry coverage, or shell cracks rather than with shrinkage. The production window was therefore fixed at 1100 ± 5°C shell preheat, 1590 ± 5°C refining, 1480 ± 5°C pouring, and 2–3 s pouring time.
Inspection and Mechanical Performance
Dimensional inspection was performed on a coordinate-measuring machine. For 36 castings produced with the optimized process, all checked dimensions—including plate length, width, height, base thickness, rib thickness, boss thickness, and boss spacing—were within the drawing tolerances. Surface inspection combined visual examination with fluorescent penetrant inspection. The acceptance limit for clustered shrinkage on the penetrant inspection was a maximum dimension of 4 mm. Radiographic inspection used a separate limit: shrinkage defects no larger than 4 mm, no more than three such defects, and spacing no less than the sum of the relevant defect diameters.
The 1480°C castings showed no penetrant indications or radiographic shrinkage above 4 mm at the rib locations. The 1500°C castings did show such indications, consistent with the acceptance-rate difference in Table 6. The remaining 1480°C rejects came from shell-related surface defects, not from internal shrinkage.
Mechanical properties were measured on separately cast disc test bars produced from the same master heat and melting practice as the plate castings. The bars were radiographed before machining to ensure that they were free of shrinkage and cracking. Room-temperature tensile tests and 995°C stress-rupture tests were performed. The results are summarized in Table 7.
| Property | Condition | Requirement | Measured mean | Range |
|---|---|---|---|---|
| Yield strength | Room temperature | ≥850 MPa | 981 MPa | 967–996 MPa |
| Elongation | Room temperature | ≥5.0% | 6.5% | 6.2–6.9% |
| Reduction of area | Room temperature | ≥8.0% | 9.5% | 9.3–9.8% |
| Stress-rupture life | 995°C, 210 MPa | ≥50 h | 79 h | 75–83 h |
All seven heats met the specified tensile and stress-rupture requirements. The measured yield strength was well above the 850 MPa minimum, and the rupture life was more than 50% above the 50 h requirement. These results indicate that the optimized casting route did not compromise the alloy’s mechanical capability, at least within the range represented by the seven heats.
Engineering Implications of the Process Window
The K424 plate is a useful example of how a thin-wall superalloy casting is controlled through three coupled thermal decisions rather than through a single parameter. First, the gating must keep a liquid channel open to the last-feeding hot spots. In this component, the critical locations were not the 0.7 mm webs but the rib junctions and boss transitions, where a thin feed path met a thicker thermal mass. Enlarging the local runner and gate changed the solidification sequence from isolated-pool freezing to directed feeding from the casting toward the gating.
Second, the local wax section must be reduced where the geometry creates a hot spot. The cold wax core in the boss acted as an internal chill for the wax pattern, not for the metal casting. By reducing the effective wax thickness, it prevented the boss from remaining soft after injection and eliminated the shrinkage depression without requiring an excessively long hold time that would crack the thin ejected edges.
Third, the shell preheat and pouring temperature must be balanced against both fill and contraction. The acceptable window is narrow: 1050°C shell preheat produced cold shuts, 1150°C produced cracks, and 1500°C pouring produced shrinkage while 1460°C produced underfill. The selected conditions of 1100°C shell preheat and 1480°C pouring lie between these failure modes. The roughly 20°C margins on either side of the pouring temperature indicate that furnace calibration, transfer time, and shell insulation are not secondary details; they are part of the process capability.
The numerical model was useful for comparing gating variants and ranking pouring temperatures, but it required physical calibration. The initial simulation used a 1050°C shell, which was adequate for comparing shrinkage tendency, yet the physical trials showed that 1050°C did not provide enough fill margin in the actual transfer and evacuation sequence. The production preheat was therefore raised to 1100°C. This is a normal limitation of casting simulation: the model can resolve the relative thermal field, but the absolute window depends on shell properties, furnace conditions, and the time spent between preheat and pouring.
The 90% acceptance rate should be read as a production result for this component and process, not as a universal capability for all thin-wall K424 castings. The remaining rejects were associated with shell bubbles, uneven face-coat coverage, and shell cracks, which are process-control issues rather than a fundamental limitation of the gating design. For similar large, thin-wall nickel-base castings, the same logic applies: identify the last-fed hot spots, preserve a feed path to those locations, reduce local pattern mass where possible, and establish a shell-temperature and pouring-temperature window that avoids both cold shuts and shrinkage. The exact temperatures, slurry viscosities, and gating dimensions will vary with shell system, furnace, and component geometry, but the controlling mechanisms remain the same.