Full Mold Casting Simulation for K418 Turbine

I treated the K418 superalloy turbocharger turbine as a coupled full mold casting, precision casting, heat-transfer, and solidification problem. My purpose was not merely to describe a single casting trial, but to build a predictive full mold casting framework that could connect pouring temperature, mold filling, thermal gradients, feeding behavior, shrinkage porosity, and final microstructure. I used a casting simulation environment to track the fluid temperature field during filling and the liquid-phase temperature field during solidification. I then compared the simulation results with a trial turbine produced under the same nominal conditions and examined the resulting metallographic structure. The central conclusion of my work is that a bottom-gated full mold casting arrangement with a pouring temperature of 1550 °C provides strong filling capability, a favorable solidification sequence, and a defect concentration zone that remains in the riser rather than in the turbine blade. This makes the full mold casting route practical for K418 turbocharger turbine production.

In my full mold casting analysis, I used a three-dimensional turbine geometry with a nominal diameter of 134 mm, an overall height of 65.38 mm, and a blade height of 45.9 mm. The turbine contains 11 thin blades and a central shaft. The blade edges are thin, the blade bodies are curved, and the shaft is comparatively massive. These geometric features create competing thermal conditions: the thin blade edges cool rapidly, while the shaft and riser remain hot for a longer time. I therefore designed the full mold casting system as a bottom-gated arrangement. Bottom gating is important in this full mold casting process because it promotes smoother metal delivery, reduces turbulence, and protects the thin blade edges from premature freezing. The casting shrinkage was taken as 2.5%, the pouring temperature was set to 1550 °C, and the mesh size was 1 mm. The shell material was zircon sand, preheated to 900 °C. I assigned a casting-to-shell heat-transfer coefficient of 650 W/(m²·K), gravity of 9.8 m/s², a weight factor of 0.8, a relaxation factor of 1.6, an initial time step of 0.0001 s, and a shell-to-environment heat-transfer coefficient of 0.023 W/(m²·K).

I began the full mold casting model by defining the alloy composition. The material is a gamma-prime precipitation-strengthened nickel-based cast superalloy. Its high nickel content provides a stable austenitic matrix, chromium supports oxidation and corrosion resistance, aluminum and titanium participate in gamma-prime formation, and niobium and molybdenum contribute solid-solution strengthening and carbide stability. I summarized the nominal composition in Table 1. The composition is not merely a list of elements; it controls liquidus temperature, solidus temperature, latent heat, viscosity, thermal conductivity, and the final gamma-prime population. In my full mold casting simulation, these variables influence both filling and solidification.

Element Content, wt% Element Content, wt%
C 0.110 Nb 2.07
Cr 13.05 Mo 4.33
Ni 73.02 Si 0.0547
Zr 0.0939 S <0.001
Al 6.1 P <0.001
Ti 0.796 Fe 0.0739
B 0.0133 Mn 0.036

For the full mold casting model, I treated the alloy melting range as 1295 °C to 1345 °C. This means the superheat at a pouring temperature of 1550 °C is approximately 205 °C above the liquidus. In a thin-blade full mold casting geometry, superheat is a resource. It raises fluidity, extends the time before local blockage, and improves replication of fine surface details. However, excessive superheat can increase gas entrapment, shell reaction, grain coarsening, and shrinkage defects in the feeder region. My full mold casting design therefore balances a high enough pouring temperature for filling with a solidification sequence that still directs final feeding to the riser.

I expressed the basic mass conservation constraint of the full mold casting filling stage as

$$ \frac{\partial \rho}{\partial t}+\nabla\cdot(\rho\mathbf{u})=0 $$

where ρ is density, t is time, and u is the velocity vector. Because the alloy liquid is treated as a nearly incompressible Newtonian fluid during the early filling stage, the mass conservation equation simplifies to a divergence-free velocity field in most of the liquid domain:

$$ \nabla\cdot\mathbf{u}=0 $$

I used the momentum equation to capture pressure-driven flow, gravity, and viscous resistance in the full mold casting system:

$$ \rho\left(\frac{\partial \mathbf{u}}{\partial t}+\mathbf{u}\cdot\nabla\mathbf{u}\right)=-\nabla p+\mu\nabla^2\mathbf{u}+\rho\mathbf{g} $$

Here p is pressure, μ is dynamic viscosity, and g is the gravity vector. In thin sections, the balance between pressure drop and viscous resistance controls whether the blade edge fills completely. In the shaft and riser, inertial terms and gravity become more significant. This is why the full mold casting simulation must solve the entire mold rather than only the blade cavity.

The energy equation is the core of the full mold casting solidification model:

$$ \rho c_p\left(\frac{\partial T}{\partial t}+\mathbf{u}\cdot\nabla T\right)=\nabla\cdot(k\nabla T)+L\frac{\partial f_s}{\partial t} $$

In this expression, c_p is specific heat, T is temperature, k is thermal conductivity, L is latent heat, and f_s is solid fraction. The last term represents latent heat release during solidification. In my full mold casting analysis, this term is essential because the blade edges, blade bodies, shaft, and riser release latent heat at different times. The temperature field therefore cannot be interpreted as a simple cooling curve; it is a coupled field in which latent heat modifies the local cooling rate and the local cooling rate modifies the solidification path.

I also used a mushy-zone permeability relation to describe feeding resistance in the full mold casting process:

$$ \mathbf{u}_l=-\frac{K}{\mu}\left(\nabla p-\rho_l\mathbf{g}\right) $$

where u_l is the liquid velocity in the mushy zone, K is permeability, and ρ_l is liquid density. As the solid fraction increases, K decreases rapidly, and liquid feeding becomes difficult. In the full mold casting of a thin-blade turbine, this relation explains why late-stage shrinkage defects tend to appear where the mushy zone is isolated from the riser. The design objective is to keep the mushy-zone pressure high enough in the blade and to place the final low-pressure region inside the riser.

The Niyama criterion was used as a practical indicator of shrinkage porosity in the full mold casting simulation:

$$ N_y=\frac{G}{\sqrt{R}} $$

where G is the thermal gradient and R is the cooling rate. In my full mold casting interpretation, regions with low N_y are more likely to form shrinkage porosity. For the K418 turbine, the blade edges solidify first and therefore do not remain in a low-N_y condition for a long time. The last regions to solidify are the riser and riser center. As a result, the full mold casting defect risk is concentrated in the riser rather than in the blade. This is precisely the condition I wanted.

I used the Chvorinov rule as a first estimate of local solidification time in the full mold casting body:

$$ t_s=B\left(\frac{V}{A}\right)^n $$

where t_s is solidification time, B is a mold constant, V is volume, A is surface area, and n is an exponent commonly close to 2. The modulus V/A of the thin blade is small, while the modulus of the shaft and riser is larger. This relationship predicts the observed full mold casting behavior: blade edges freeze first, blade bodies follow, the shaft solidifies later, and the riser remains liquid long enough to feed shrinkage.

I also tracked the casting yield and feeding ratio in the full mold casting design. For a first-order mass balance,

$$ Y=\frac{W_c}{W_c+W_r+W_g}\times 100\% $$

where Y is casting yield, W_c is casting mass, W_r is riser mass, and W_g is gating mass. A full mold casting design with a large riser may improve feeding but reduce yield. My goal was not to maximize yield alone; it was to achieve a robust full mold casting window in which the blade is sound, the riser captures shrinkage, and the process remains repeatable.

I summarized the main full mold casting process parameters in Table 2. These values define the simulation and the trial production condition. The table is written as a process control summary rather than as a descriptive list, because each parameter has a direct physical role in the full mold casting outcome.

Parameter Value Role in full mold casting
Turbine diameter 134 mm Controls overall heat capacity and shell volume
Turbine height 65.38 mm Sets vertical thermal gradient and riser position
Blade height 45.9 mm Determines thin-section filling distance
Number of blades 11 Creates multiple thin cavities and local hot spots
Melting range 1295-1345 °C Defines liquidus, solidus, and freezing interval
Casting shrinkage 2.5% Compensates solid contraction in pattern and die
Pouring temperature 1550 °C Provides superheat for complete filling
Mesh size 1 mm Resolves thin blade edges and shaft details
Shell material Zircon sand Controls heat extraction and shell stability
Shell preheat 900 °C Reduces thermal shock and controls cooling rate
Casting-shell heat transfer 650 W/(m²·K) Sets interfacial heat extraction
Gravity 9.8 m/s² Drives feeding and vertical pressure head
Weight factor 0.8 Controls equation weighting in the numerical scheme
Relaxation factor 1.6 Stabilizes iterative solution behavior
Initial time step 0.0001 s Resolves early filling transients
Environment heat transfer 0.023 W/(m²·K) Represents weak external heat loss

I then calculated the thermo-physical properties required by the full mold casting model. These properties are not constant over the full temperature range. Density decreases as temperature rises, especially near melting. Specific heat increases with temperature and absorbs latent heat effects in the mushy zone. The thermal expansion coefficient increases with temperature and influences stress and distortion after solidification. I summarized these properties in Table 3. The missing values in the table indicate temperatures at which the property is not used directly in the current full mold casting model because the alloy is already liquid or because the property is replaced by latent-heat treatment.

Temperature, °C Density, g/cm³ Specific heat, J/(g·K) Thermal expansion, 10^-6/°C
20 8.0 — 12.6
200 7.95 0.45 12.8
400 7.85 0.46 13.2
600 7.80 0.47 13.5
800 7.74 0.49 13.5
1000 7.67 0.50 14.6
1200 7.6 0.53 —
1400 6.7 — —
1500 6.5 — —

I considered the shell preheat temperature of 900 °C to be one of the most influential full mold casting variables. A cold shell would extract heat too quickly at the blade edges, raising the risk of mistun and cold shut. A very hot shell would slow solidification, coarsen the microstructure, and increase the risk of shrinkage outside the riser. The 900 °C preheat is an intermediate condition: it protects the thin blade edges during the first seconds of filling, but it still permits a directional solidification sequence from the blade edges toward the riser.

I used the Biot number to evaluate the relative importance of interfacial heat transfer and internal conduction in the full mold casting blade:

$$ Bi=\frac{hL_c}{k} $$

where h is the heat-transfer coefficient, L_c is a characteristic length, and k is thermal conductivity. For a thin blade, L_c is small, so Bi is moderate to low. This means the blade temperature tends to be relatively uniform across its thickness, while the shell interface controls heat extraction. For the shaft and riser, L_c is larger, so internal conduction becomes more important and the thermal gradient is more persistent. This contrast is central to the full mold casting solidification behavior.

I also examined the Reynolds number and Peclet number in the full mold casting filling stage:

$$ Re=\frac{\rho uL}{\mu} $$

$$ Pe=\frac{uL}{\alpha} $$

In the gating system, Re is high enough that inertial effects matter, but the flow remains sufficiently controlled because the gating design reduces jetting. In the thin blade channels, the small hydraulic diameter increases viscous resistance. Pe indicates that heat advection is important during filling; the liquid metal carries thermal energy into the blade before the shell can remove it. This is why my full mold casting filling simulation showed fluid temperatures above 1400 °C throughout the blade cavity at the selected pouring temperature.

The filling simulation results are summarized in Table 4. I recorded the fluid temperature field at several times and confirmed that the alloy remained well above the melting range during cavity filling. The lowest fluid temperatures still exceeded 1400 °C, which is above the liquidus of 1345 °C. This margin is important in a full mold casting of a thin-blade turbine because local premature solidification can block a blade passage before the cavity is completely filled.

Filling time, s Minimum fluid temperature, °C Comparison with melting range Full mold casting implication
0.35 >1400 Above liquidus Initial metal front remains fluid
0.94 >1400 Above liquidus Blade channels continue to fill
1.67 >1400 Above liquidus Thin edges are not blocked
4.07 >1400 Above liquidus Late filling remains stable

I observed that the filling pattern progressed from the bottom gate into the runner, then into the turbine shaft, and finally into the blades and blade edges. The bottom-gated full mold casting design reduced free fall and limited surface turbulence. In a top-gated full mold casting arrangement, the liquid metal would fall through the cavity, entrain gas, and create oxide films. Those defects would be difficult to remove and would reduce fatigue resistance in the blade roots. My full mold casting approach instead filled the cavity from below, allowing the metal front to rise in a more orderly manner.

After filling was complete, I analyzed the temperature field at the end of pouring. At approximately 6.60 s, the blade edges had cooled below the melting range and began to solidify. This is consistent with the thin blade edge geometry. The edge has a small modulus and a high surface-to-volume ratio, so it loses heat rapidly. The blade body remained hotter because it has greater mass and because it receives heat from the still-liquid shaft region. This created a favorable full mold casting gradient from the blade edge toward the blade center and then toward the shaft.

At approximately 148.26 s, the riser center still contained liquid metal above the melting range. This is a critical full mold casting condition. If the riser remains liquid longer than the casting, it can continue to feed volumetric shrinkage through the gating path. If the riser freezes too early, isolated liquid regions in the blade or shaft may form shrinkage porosity. In my full mold casting simulation, the riser center remained liquid long enough to act as the final feeding reservoir.

I summarized the solidification sequence in Table 5. The sequence is not arbitrary; it follows the thermal modulus and heat-extraction path of the full mold casting geometry. The blade edge freezes first, the blade body follows, the blade axis solidifies next, then the gating riser, and finally the riser center. The last region to solidify is the riser center, which is exactly where I want residual shrinkage to occur.

Stage Location Solidification behavior Full mold casting consequence
1 Blade edge Fast cooling, thin section Preserves edge shape and surface detail
2 Blade body Directional cooling from edge inward Promotes dense blade structure
3 Blade axis Later freezing due to greater mass Maintains feeding path toward shaft
4 Pouring riser Hot reservoir with liquid metal Feeds shrinkage in shaft and root
5 Riser center Last liquid region Contains final shrinkage defects

I used the liquid-phase temperature field to track the progressive disappearance of liquid metal in the full mold casting. The liquid fraction is a useful scalar because it condenses the thermal history into a single field:

$$ f_l=1-f_s $$

When f_l approaches zero, the region becomes solid and can no longer feed remote shrinkage. In my full mold casting analysis, the blade edge reached f_l = 0 first, while the riser center retained f_l > 0 for much longer. The solidification path therefore followed the intended sequence and avoided isolated liquid pockets in the blade.

I also evaluated the cooling rate and its effect on grain structure. The cooling rate can be approximated from the thermal history as

$$ \dot{T}=\frac{\partial T}{\partial t} $$

and the local solidification time can be approximated as

$$ t_f=\frac{T_L-T_S}{\dot{T}} $$

where T_L is liquidus temperature and T_S is solidus temperature. In the full mold casting blade, t_f is short at the edge, moderate in the blade body, and longer in the shaft and riser. This variation produces different grain morphologies in different regions. The blade edge experiences a high cooling rate and forms fine equiaxed grains. The blade body experiences a directional thermal gradient and forms columnar dendrites together with fine equiaxed grains. The root and axial center cool more slowly and form more equiaxed grains.

I related the primary dendrite arm spacing to cooling rate using a power-law form:

$$ \lambda_1=A\dot{T}^{-n} $$

where λ_1 is primary dendrite arm spacing, A is a material constant, and n is an exponent. A higher cooling rate gives a smaller spacing. In the full mold casting blade, the thin edge has the highest cooling rate and the finest structure. The shaft has a lower cooling rate and a coarser structure. This is consistent with my metallographic observations.

I also estimated the grain size using a similar relation:

$$ d=d_0+B\dot{T}^{-m} $$

where d is grain size, d_0 is an initial size, B is a coefficient, and m is an exponent. The measured average grain size in the turbine blade was 24.561 micrometers. This fine grain size is beneficial for fatigue resistance and tensile properties, although high-temperature creep resistance can sometimes favor directionally solidified or single-crystal structures. For a polycrystalline turbocharger turbine, the combination of columnar and fine equiaxed grains in the blade is a practical and robust outcome.

I evaluated the thermal modulus of several full mold casting regions using

$$ M=\frac{V}{A} $$

where M is the modulus, V is volume, and A is active cooling surface area. The blade edge has the smallest modulus and therefore the shortest solidification time. The blade body has a slightly larger modulus. The shaft has a much larger modulus. The riser has the largest effective modulus among the feeding regions. This ordering is consistent with the observed full mold casting solidification sequence and explains why the riser remains liquid while the blade is already solid.

I summarized the predicted defect risk in Table 6. The table compares different regions of the full mold casting and indicates whether shrinkage porosity, cold shut, gas entrapment, or misrun is likely. In my full mold casting simulation, the blade region showed low risk, the shaft showed moderate risk near the thermal center, and the riser showed the highest risk. This is a favorable distribution because the riser can be removed after casting.

Region Filling risk Solidification risk Predicted defect location Full mold casting assessment
Blade edge Low at 1550 °C Fast freezing Low Good shape replication
Blade body Low Moderate gradient Low Dense structure expected
Blade root Low Moderate feeding demand Low to moderate Riser feeding required
Shaft Low Hot spot possible Moderate Feeding path must remain open
Riser Not applicable Last to freeze High Accepts shrinkage
Riser center Not applicable Final liquid Highest Sacrificial defect zone

I used the simulation to refine the full mold casting window rather than to simply accept the first design. The most important variables were pouring temperature, shell preheat, gating location, riser size, and heat-transfer coefficient. I summarized a sensitivity assessment in Table 7. This table reflects my interpretation of how changes in each variable would influence the full mold casting process.

Variable Increase Decrease Effect on full mold casting
Pouring temperature Better filling, coarser grains, more shrinkage Risk of cold shut and misrun 1550 °C is the selected balance
Shell preheat Slower cooling, better filling, coarser structure Faster cooling, risk of incomplete edge filling 900 °C supports thin edges
Gating position Top gating increases turbulence Bottom gating improves stability Bottom gating is preferred
Riser volume Better feeding, lower yield Higher yield, higher shrinkage risk Riser must remain last
Heat-transfer coefficient Faster cooling, finer grains Slower cooling, coarser grains 650 W/(m²·K) is consistent

I validated the full mold casting simulation by producing a trial turbine under the selected conditions. The trial turbine had a smooth surface and no visible external defects. The blade passages were complete, the blade edges were well formed, and the shaft showed no obvious surface shrinkage. This external inspection is not sufficient to prove internal soundness, so I also performed metallographic sampling. I selected samples from the blade, blade root, and axial center, and I compared the observed microstructure with the full mold casting prediction.

I summarized the validation results in Table 8. The table combines visual inspection, metallographic observation, and simulation prediction. The agreement between the predicted and observed results supports the use of the full mold casting model as a process design tool.

Validation item Simulation prediction Experimental observation Agreement
Overall fill Complete at 1550 °C Complete turbine shape High
Surface quality Good Smooth surface, no visible defects High
Shrinkage location Riser region No blade shrinkage observed High
Blade microstructure Columnar plus fine equiaxed Columnar and fine equiaxed grains High
Root and axial center Equiaxed grains Ordered equiaxed grains High
Average grain size Fine 24.561 micrometers Consistent

I examined the blade microstructure in detail. The blade region was dense and showed no obvious defects. The structure consisted of columnar crystals and fine equiaxed crystals. This mixed morphology is common in thin-section castings where the thermal gradient is high at the surface and the cooling rate is rapid. The surface layer solidifies quickly and forms fine equiaxed grains, while the interior can develop columnar dendrites aligned with the heat-flow direction. In the full mold casting blade, this combination is desirable because it provides both strength and damage tolerance.

The root and axial center of the turbine showed more equiaxed grains. These regions have larger thermal mass and lower cooling rates. The grains are more equiaxed because the thermal gradient is lower and constitutional undercooling can develop in multiple directions. The equiaxed structure in the root and axial center is also beneficial because it reduces anisotropy and distributes strain more uniformly under thermal cycling. The full mold casting process therefore produces a functionally graded microstructure across the turbine: finer and more directional at the blade edges, more equiaxed in the root and center.

I also considered the role of latent heat in the full mold casting simulation. During solidification, the release of latent heat slows the local cooling rate and delays the temperature drop. This effect is strongest in the shaft and riser, where the solidifying volume is large. In the blade edge, the volume is small and the latent heat release is quickly removed by the shell. This difference amplifies the natural thermal gradient and helps the full mold casting sequence progress from thin to thick sections. If latent heat were neglected, the simulation would overpredict cooling rates and underpredict feeding distances.

I used the following relation to estimate the volumetric shrinkage that must be compensated by the riser:

$$ \epsilon_v=\frac{\rho_s-\rho_l}{\rho_s} $$

where ε_v is volumetric shrinkage, ρ_s is solid density, and ρ_l is liquid density. The riser must supply enough liquid metal to compensate this shrinkage in the casting before the feeding path freezes. In my full mold casting design, the riser center remains liquid until after the blade and shaft have mostly solidified. This ensures that the shrinkage demand is transferred to the riser instead of creating isolated porosity in the blade.

I also estimated the feeding distance using a simplified thermal criterion:

$$ L_f=C\left(\frac{G}{\dot{T}}\right)^m $$

where L_f is feeding distance, G is thermal gradient, \dot{T} is cooling rate, and C and m are empirical constants. A higher thermal gradient and a lower cooling rate increase feeding distance. In the full mold casting turbine, the bottom gate and shell preheat create a favorable gradient from the blade to the riser. This allows the riser to feed the shaft and root even though the blade edges freeze quickly.

I considered the influence of the full mold casting shell on heat extraction. The shell has a lower thermal conductivity than the metal, but it still controls the interfacial heat transfer. The heat flux across the casting-shell interface can be written as

$$ q=h(T_c-T_s) $$

where q is heat flux, h is the heat-transfer coefficient, T_c is casting surface temperature, and T_s is shell surface temperature. A higher h increases heat extraction and promotes faster solidification. In my full mold casting model, h = 650 W/(m²·K) gave a realistic balance between blade edge filling and directional solidification. If h is too high, the blade edge may freeze before the cavity is full. If h is too low, the blade may remain liquid too long and the riser may not be able to feed the entire casting.

I examined the effect of the bottom-gated full mold casting design on turbulence. A simple estimate of the gate velocity is

$$ v_g=\frac{Q}{A_g} $$

where v_g is gate velocity, Q is volumetric flow rate, and A_g is gate area. A high gate velocity increases Reynolds number and can cause jetting, gas entrainment, and mold erosion. A low gate velocity reduces turbulence but may increase filling time and allow premature freezing. The bottom-gated full mold casting arrangement allows a moderate gate velocity while keeping the metal front stable. This is a key reason for the good surface quality in the trial casting.

I summarized the relationship between filling time and flow rate in Table 9. The table is based on the continuity principle and illustrates how gate area and flow rate influence the full mold casting filling stage. The actual simulation includes pressure losses, but the table provides a useful process design check.

Volumetric flow rate, cm³/s Gate area, cm² Approximate gate velocity, cm/s Full mold casting tendency
500 5 100 Moderate, stable filling
500 2.5 200 Turbulence risk increases
500 10 50 Low turbulence, longer fill
800 5 160 Faster fill, higher inertia
300 5 60 Slow fill, premature freezing risk

I also used the heat balance of the shell to check the full mold casting preheat condition. The shell absorbs heat from the casting and releases heat to the environment. A simplified lumped balance is

$$ m_s c_{p,s}\frac{dT_s}{dt}=hA(T_c-T_s)-h_eA(T_s-T_e) $$

where m_s is shell mass, c_{p,s} is shell specific heat, h_e is environment heat-transfer coefficient, and T_e is environment temperature. With a shell preheat of 900 °C and a low external heat-transfer coefficient, the shell remains hot during the early filling stage and then gradually cools. This delays edge freezing just long enough to complete the blade cavity.

I evaluated the solidification path using the Scheil-like relation for solute redistribution in the full mold casting:

$$ C_l=C_0(1-f_s)^{k-1} $$

where C_l is liquid composition, C_0 is initial composition, k is partition coefficient, and f_s is solid fraction. This relation is approximate for a complex superalloy, but it explains why the last liquid regions become enriched in solute elements and why carbides, borides, and gamma-prime form in the interdendritic regions. In the full mold casting blade, the rapid cooling limits segregation and produces a fine distribution of second phases. In the slower-cooling shaft and riser, segregation can be more pronounced, but these regions are either removed or less critical for fatigue.

I considered the formation of gamma-prime in the full mold casting microstructure. The gamma-prime phase is the principal strengthening precipitate in K418. Its volume fraction, size, and distribution depend on cooling rate and subsequent heat treatment. In the as-cast full mold casting blade, the gamma-prime is fine and distributed in the dendrite cores and interdendritic regions. A high cooling rate suppresses coarsening and retains a fine precipitate structure. This contributes to high-temperature strength and creep resistance, although a full heat treatment may still be required for optimal properties.

I also noted that the full mold casting process can produce carbides and borides at grain boundaries. These phases can improve creep resistance by pinning grain boundaries, but excessive or continuous networks can reduce ductility and toughness. In the full mold casting blade, the rapid solidification helps disperse these phases. In the root and shaft, slower cooling may produce larger carbides, but the equiaxed grain structure and the absence of sharp thermal gradients reduce the risk of cracking.

I summarized the microstructural features in Table 10. The table connects the local cooling conditions to the observed grain morphology and the expected mechanical behavior.

Region Cooling condition Observed structure Expected mechanical effect
Blade edge Highest cooling rate Fine equiaxed grains High fatigue resistance
Blade body High gradient, moderate cooling Columnar plus fine equiaxed Good strength and creep resistance
Blade root Moderate cooling Ordered equiaxed grains Uniform load distribution
Axial center Lower cooling rate Equiaxed grains Reduced anisotropy
Riser Slowest cooling Coarse equiaxed or dendritic Sacrificial feeding zone

I considered the industrial implications of this full mold casting study. The simulation reduced the number of trial casts, shortened the development cycle, and lowered the cost of process optimization. Instead of physically testing many pouring temperatures and gating designs, I could screen them numerically and then validate only the most promising condition. This is especially valuable for K418 because the alloy is expensive and the turbine geometry is complex. The full mold casting model provides a rational basis for selecting the pouring temperature, shell preheat, gating position, and riser size.

I also recognized the limitations of the full mold casting model. The simulation assumed a uniform shell thickness and a constant interfacial heat-transfer coefficient, while in reality the shell may have local variations in thickness, permeability, and thermal properties. The model also simplified the nucleation and grain growth kinetics. A more advanced full mold casting model could include stochastic nucleation, dendrite growth, and solid-state phase transformation. These additions would improve the prediction of grain size, texture, and precipitate distribution. Nevertheless, the current full mold casting model captures the dominant physical mechanisms: filling, heat transfer, solidification sequence, and feeding.

I used the following dimensionless superheat parameter to compare the full mold casting conditions:

$$ S^*=\frac{T_p-T_L}{T_L-T_S} $$

where T_p is pouring temperature, T_L is liquidus temperature, and T_S is solidus temperature. For T_p = 1550 °C, T_L = 1345 °C, and T_S = 1295 °C,

$$ S^*=\frac{1550-1345}{1345-1295}=4.1 $$

This high dimensionless superheat confirms that the alloy has a large thermal margin above the freezing range. It supports complete filling of the thin blade edges. However, the same high superheat increases the total heat that must be removed and can enlarge the shrinkage demand. This is why the riser must be properly sized and why the full mold casting sequence must be directional.

I also calculated the relative freezing range of the alloy:

$$ \Delta T_f=T_L-T_S=50\ ^\circ C $$

A relatively narrow freezing range is favorable for directional solidification because the mushy zone is narrow and feeding resistance is lower. In a wide freezing range alloy, the mushy zone can extend over a large region and isolate liquid pockets. K418 has a moderate freezing range, and the full mold casting design further promotes directionality by placing the riser at the hot end and the thin blades at the cold end.

I evaluated the thermal gradient along the blade using

$$ G=\frac{\Delta T}{\Delta x} $$

where ΔT is the temperature difference and Δx is the distance. In the full mold casting blade, the gradient is steep near the edge and becomes gentler toward the root. A steep gradient favors columnar growth, while a gentle gradient favors equiaxed growth. This explains the mixed columnar-equiaxed structure in the blade body and the more equiaxed structure in the root and axial center.

I summarized the relationship between thermal gradient and grain morphology in Table 11. This table is a simplified solidification map for the full mold casting turbine.

Thermal gradient Cooling rate Grain morphology Full mold casting region
High High Fine equiaxed Blade edge
High Moderate Columnar Blade body
Moderate Moderate Columnar plus equiaxed Blade root transition
Low Low Equiaxed Axial center and root
Very low Very low Coarse equiaxed or dendritic Riser

I paid special attention to the blade edge because it is the most difficult region to fill. The edge thickness is small, so the heat capacity of the incoming metal is limited. The edge can freeze if the local residence time is too long or if the shell extracts heat too quickly. My full mold casting simulation showed that at 1550 °C the fluid temperature remained above 1400 °C throughout filling. This provides a margin of approximately 55 °C above the liquidus and 105 °C above the solidus. The edge therefore fills before it freezes, and the final edge structure is fine and dense.

I also examined the shaft region. The shaft has a larger cross-section and a lower surface-to-volume ratio, so it cools more slowly. This makes it a natural hot spot. The full mold casting design must ensure that the shaft remains connected to the riser until it has finished feeding. In my simulation, the shaft solidified after the blade axis and before the riser center. This is the correct order: the shaft receives liquid metal from the riser while it is still mushy, and the final shrinkage is transferred to the riser center.

I considered the possibility of gas porosity in the full mold casting. Gas porosity can form from dissolved gas, gas entrapment during filling, or decomposition products from the mold. The bottom-gated design reduces entrapment, and the high shell preheat reduces the risk of premature skin formation that can trap gas. The simulation did not predict significant gas entrapment in the blade. The trial casting showed no visible gas defects. If gas porosity were observed, I would reduce gate velocity, increase venting, or adjust the shell permeability.

I also considered cold shut and misrun defects. A cold shut occurs when two metal fronts meet but do not fuse completely. A misrun occurs when a cavity does not fill before the metal freezes. Both defects are possible in thin-blade full mold casting. The high pouring temperature and bottom gating reduce both risks. The simulation showed that the metal fronts remained hot and merged before local solidification. The trial casting confirmed complete blade edges and no visible cold shuts.

I summarized the defect control strategy in Table 12. The table links each potential defect to its cause, the full mold casting control variable, and the observed result.

Defect Cause Full mold casting control Observed result
Misrun Low superheat or slow filling Pouring at 1550 °C No misrun
Cold shut Premature front freezing Shell preheat at 900 °C No cold shut
Gas porosity Turbulence or mold gas Bottom gating and venting No visible gas pores
Shrinkage porosity Insufficient feeding Riser remains last to freeze Contained in riser
Hot tearing Restrained contraction during solidification Uniform shell preheat and geometry No visible hot tears
Coarse grains Low cooling rate Controlled shell preheat and thin sections Fine structure in blade

I used the full mold casting model to estimate the time required for complete solidification. The blade edge solidified first, followed by the blade body, the blade axis, the shaft, the gating riser, and finally the riser center. The total solidification time is controlled by the riser because it has the largest thermal modulus. This is desirable because the riser acts as a thermal reservoir and feeding reservoir. The full mold casting process is therefore self-consistent: the region that must feed the casting is also the region that remains liquid longest.

I considered the effect of the gating system on the thermal field. The bottom gate delivers metal into the shaft region, which is a thick section. This preheats the shaft and helps maintain a feeding path. If the gate were placed directly on a thin blade, the local thermal mass would be too small, and the gate would freeze before the casting was fully fed. The full mold casting design therefore avoids direct gating into thin blades and instead uses the shaft as the distribution region. This is an important design rule for thin-blade turbine castings.

I also considered the role of the riser shape. A cylindrical riser with a sufficiently large diameter can remain liquid longer than the shaft. The riser must have a modulus larger than the modulus of the region it feeds. I used the modulus ratio

$$ M_r=\frac{M_{riser}}{M_{casting}} $$

where M_r is the riser modulus ratio. For successful full mold casting feeding, M_r should be greater than 1, typically between 1.1 and 1.4 for many castings. In my full mold casting design, the riser modulus was selected to exceed the shaft modulus. This ensured that the riser remained liquid after the shaft began to freeze and could continue to supply liquid metal.

I evaluated the pressure head required for feeding using the hydrostatic relation

$$ \Delta p=\rho g h $$

where Δp is pressure difference, ρ is density, g is gravity, and h is height. The bottom-gated full mold casting system provides a favorable pressure head because the runner and riser are located above the gate. This pressure head helps drive liquid metal into the mushy zone and compensates for viscous resistance. Without sufficient pressure head, feeding would stop before the last shrinkage was compensated.

I also considered the effect of shell permeability on gas evacuation. During full mold casting, gas and vapor must escape through the shell or through vents. If permeability is too low, gas pressure can build up and cause blowholes or incomplete filling. The zircon sand shell with a controlled preheat provides sufficient permeability in the current design. The simulation did not include detailed gas flow, but the trial casting results suggest that gas evacuation was adequate.

I summarized the full mold casting process window in Table 13. The table lists the lower and upper limits for key variables and the reason for each limit. This process window can be used as a practical guide for production.

Variable Lower limit Upper limit Reason for limit
Pouring temperature 1520 °C 1580 °C Below 1520 °C risks misrun; above 1580 °C coarsens grains
Shell preheat 850 °C 950 °C Lower risks cold shut; higher slows solidification too much
Gate velocity 50 cm/s 180 cm/s Lower risks freezing; higher risks turbulence
Riser modulus ratio 1.1 1.5 Lower risks shrinkage; higher reduces yield
Shell heat-transfer coefficient 500 W/(m²·K) 800 W/(m²·K) Lower risks slow freezing; higher risks edge blockage

I also evaluated the energy balance of the full mold casting process. The total heat to be removed from the alloy includes superheat, latent heat, and solid-state cooling:

$$ Q_{total}=m\left[c_p(T_p-T_L)+L+c_p(T_S-T_e)\right] $$

where m is mass, c_p is specific heat, T_p is pouring temperature, T_L is liquidus, T_S is solidus, L is latent heat, and T_e is final temperature. This equation shows why pouring temperature and riser size are coupled. A higher pouring temperature increases Q_total and requires more feeding and longer solidification time. In my full mold casting design, the riser compensates for the additional shrinkage by remaining liquid longer.

I considered the heat flux through the shell using the overall thermal resistance:

$$ R_{total}=R_{casting}+R_{interface}+R_{shell}+R_{environment} $$

where each R is a thermal resistance. The interface resistance is often dominant in precision casting. In my full mold casting model, the heat-transfer coefficient of 650 W/(m²·K) represents this interface. A lower interface resistance would increase heat extraction and promote earlier freezing. A higher interface resistance would slow cooling and promote coarser grains. The selected value produced a realistic solidification sequence and matched the observed microstructure.

I also considered the effect of the mold on the cooling curve. The shell preheat of 900 °C means that the initial temperature difference between the casting and shell is relatively small at the start of pouring. This reduces the initial heat flux and helps the blade fill. As the shell heats further, the heat flux decreases, and the casting cools more slowly. This dynamic behavior is important for the full mold casting of thin sections. A cold shell would produce a high initial heat flux and could freeze the blade edge before filling is complete.

I summarized the thermal history of selected regions in Table 14. The table is qualitative but reflects the relative cooling behavior predicted by the full mold casting model.

Region Initial temperature after filling Relative cooling rate Solidification start Solidification end
Blade edge Near 1400 °C High Earliest Early
Blade body Above 1400 °C Moderate to high Early Moderate
Blade root Above 1400 °C Moderate Moderate Moderate
Shaft Above 1400 °C Moderate to low Later Late
Riser Above 1400 °C Low Late Later
Riser center Above 1400 °C Lowest Latest Latest

I concluded that the full mold casting process is capable of producing a sound K418 turbocharger turbine. The simulation results and experimental validation agree on the key points: complete filling, favorable solidification sequence, riser-controlled shrinkage, dense blade structure, and fine grain size. The full mold casting design is therefore suitable as a production guide. The next step would be to extend the model to include stress analysis, hot tearing prediction, and heat treatment simulation. These additions would allow a more complete optimization of the full mold casting route.

I also recommend monitoring several production variables in future full mold casting runs. The pouring temperature should be controlled within the process window. The shell preheat should be uniform to avoid local cold spots. The gating system should remain bottom-gated to reduce turbulence. The riser should be inspected after casting to confirm that shrinkage is concentrated there. The blade should be sampled periodically for grain size and defect analysis. These practices will maintain the benefits of the full mold casting design over repeated production.

I summarized the final full mold casting conclusions in Table 15. The table consolidates the main findings and their practical meaning.

Finding Evidence Practical meaning for full mold casting
Bottom gating and 1550 °C ensure filling Fluid temperature above 1400 °C during filling Thin blade edges fill completely
Solidification is directional Blade edge to blade to axis to riser center Shrinkage is transferred to riser
Riser remains liquid longest Riser center liquid at 148.26 s Feeding path remains open
Blade microstructure is dense No obvious defects; fine grains Good fatigue and strength potential
Grain morphology varies by region Columnar plus equiaxed in blade; equiaxed in root and center Balanced mechanical behavior
Simulation matches trial casting Smooth surface and sound blade Model can guide production

I regard the full mold casting simulation as a successful predictive tool for this K418 turbine. It reduced trial-and-error, clarified the relationship between process variables and defects, and provided a quantitative basis for selecting the pouring temperature and gating design. The full mold casting route with bottom gating, 1550 °C pouring, 900 °C shell preheat, and a properly sized riser produced a turbine with good external quality, dense blade structure, and no visible defects. The final microstructure contained columnar and fine equiaxed grains in the blade and ordered equiaxed grains in the root and axial center. I therefore recommend this full mold casting design for practical production and for further optimization through coupled thermal-stress and microstructural modeling.

Scroll to Top