Optimization of High-Precision Titanium Alloy Impeller Casting through Simulation and Process Refinement

In the field of advanced manufacturing for corrosive and high-stress environments, the demand for components combining exceptional strength-to-weight ratio with superior corrosion resistance is paramount. Titanium alloys, particularly commercially pure grades like TA2, offer an ideal solution due to their low density, high specific strength, excellent plasticity, and outstanding resistance to corrosion and erosion. For complex, thin-walled components such as impellers used in fluid machinery, traditional machining from billet can be prohibitively expensive and wasteful. Therefore, precision investment casting emerges as the most viable net-shape forming technology. This process, also known as the lost-wax process, is capable of producing components with intricate geometries, excellent surface finish, and dimensional accuracy that often requires minimal subsequent machining. The core of the process involves creating a wax pattern, building a ceramic shell around it, melting out the wax, and pouring molten metal into the resulting cavity. For reactive metals like titanium, this must be conducted under vacuum or inert atmosphere to prevent contamination.

However, achieving high-quality titanium castings, especially for critical rotating parts like impellers, presents significant challenges. The high melting point of pure titanium (~1668°C for TA2) compared to many alloys leads to lower superheat under conventional melting regimes, increasing the melt viscosity and the risk of filling-related defects. Furthermore, the requirement for stringent dimensional tolerances (e.g., CT7 per ISO 8062) and precise angular alignment of features like blades, often directly from the as-cast state, pushes the precision investment casting process to its limits. Defects such as shrinkage porosity, cold shuts, and microstructural inhomogeneities can severely compromise the performance and service life of the component. This article details a first-person engineering journey in developing a robust precision investment casting process for a high-purity titanium impeller. The approach integrates initial empirical process design with advanced numerical simulation using ProCAST finite element analysis (FEA) software to diagnose, replicate, and ultimately eliminate casting defects, culminating in the successful production of a high-integrity component.

Component Analysis and Initial Casting Challenges

The subject component is a centrifugal impeller fabricated from TA2 commercially pure titanium. Its function involves operating at high rotational speeds within a corrosive atmosphere, placing extreme demands on its dynamic balance, structural integrity, and corrosion resistance. The impeller geometry consists of a central hub with five three-dimensionally curved blades uniformly distributed on both sides of a main disc. The key functional surfaces are the pressure and suction sides of these blades. Critical technical specifications included:

1. Dimensional Accuracy: Maximum envelope dimensions of Φ300 mm × 80 mm with a minimum wall thickness of 6 mm. The casting dimensional tolerance standard was required to be CT7. The angular deviation between the five blades must not exceed 10 arcminutes, and the profile tolerance was specified as ±0.5 mm.

2. Internal Quality: The internal soundness was mandated to comply with Grade C of GB/T 6614-2007 (equivalent to high-integrity aerospace standards), requiring a dense, homogeneous microstructure free from cracks, shrinkage, and significant gas porosity.

The primary challenges identified for the precision investment casting of this part were twofold:

First, the Metallurgical Challenge: The high melting point and associated low superheat of pure titanium increase its viscosity during pouring. This, combined with the intricate, thin-section geometry of the blades, creates a high propensity for mistrun and cold shut defects. Cold shuts, where two fluid fronts meet but fail to fuse completely, can be particularly detrimental as they act as stress concentrators and initiation sites for fatigue cracks under dynamic loading.

Second, the Precision Challenge: Achieving the specified CT7 tolerance and the ultra-tight angular control on the blades directly from the casting process is exceptionally demanding. The process involves two major shrinkage events: the solidification contraction of the wax pattern and the solidification contraction of the titanium metal. Accurately compensating for both in the tooling design is critical for final dimensional accuracy.

Initial Process Design and Prototype Outcome

The initial precision investment casting process was designed based on empirical knowledge and standard practices for titanium. A key step was determining the double shrinkage allowance. For the wax pattern (using a medium-temperature wax), a linear contraction factor of 1.2% was applied uniformly. For the titanium metal contraction from liquidus to room temperature, an established empirical factor of 1.0% was used. The total compensation built into the mold tooling was therefore a combination of these factors, though not simply additive due to the sequence of events.

The initial gating system was designed as a top-pour, center-gated scheme to leverage centrifugal force for feeding. The layout consisted of a central down sprue with a diameter of 60 mm, connected to five separate in-gates (15 mm diameter) entering the impeller hub at a 25° angle. The wax patterns were assembled into a cluster, shelled with a zirconia-based ceramic slurry and stucco to build a shell approximately 10 mm thick, dewaxed, and fired. Melting and casting were conducted in a vacuum arc skull furnace with centrifugal pouring. The parameters were: melting current 15 kA, voltage 40 V, centrifugal speed of 250 rpm.

The result of this initial trial was unsatisfactory. Visual inspection revealed significant issues:

1. Cold Shuts: Severe cold shuts were present at the center of the hub and at the fillet regions where the blades meet the disc. These manifested as both “re-melt” type (closed, linear features) and “mis-run” type (open, linear features).

2. Shrinkage Porosity: Substantial micro-porosity and localized shrinkage were evident along the outer edges and thicker sections of the casting.

Radiographic inspection confirmed the presence of scattered shrinkage cavities internally, validating the visual assessment. The prototype failed to meet the internal quality specification.

Diagnosis via ProCAST Finite Element Simulation

To understand the root cause of these defects without costly and time-consuming physical trials, the process was replicated using ProCAST simulation software. The 3D CAD models of the impeller and the initial gating system were imported. A critical step was the meshing strategy:

1. Surface Meshing: All solid parts (wax pattern, gating) were surface meshed to create an accurate representation.

2. Shell Generation: Using the “shelling” command, a 10 mm thick shell mesh was generated around the assembly to represent the ceramic mold.

3. Volume Meshing: The entire system, including the furnace enclosure volume, was volume meshed, keeping the total element count below 10 million for computational efficiency.

The following initial and boundary conditions were defined to model the vacuum centrifugal casting process realistically:

Component / Condition Material / Setting Parameters / Values
Mold Zirconia Preheat Temperature: 300°C; Stress: Rigid
Casting & Gates Pure Ti (TA2) Liquidus/Solidus Temp.: Defined; Fluid Model: Newtonian; Initial Temp.: 300°C (shell filling start)
Heat Exchange Radiation (Vacuum) Emissivity Constant: 0.9
Furnace Wall Radiation Boundary Radiation Direction: Inwards; Emissivity: 0.8
Pouring Velocity Boundary Velocity defined as a time-function to match furnace tilting
Rotation Revolution Boundary Rotational speed defined as a time-function (250 rpm)

The simulation outputs for the initial process were revealing. The temperature distribution plot at the end of filling clearly showed a large thermal gradient. The central hub region, fed last by the top-gate, was significantly cooler than the metal arriving at the blade tips via centrifugal force. This temperature differential is the primary driver for cold shut formation, as the cooler metal fronts lose the fluidity required to fuse. The solidification simulation accurately predicted shrinkage porosity in the locations where it was found physically. The Niyama criterion (often used to predict shrinkage) can be expressed as:
$$ G / \sqrt{\dot{T}} $$
where \( G \) is the temperature gradient and \( \dot{T} \) is the cooling rate. Low values of this criterion indicate a high risk of shrinkage porosity. The simulation showed low-criterion zones correlating perfectly with the observed defective areas. The simulation thus successfully diagnosed the problems: inadequate thermal management leading to cold shuts and insufficient feeding pressure and direction leading to shrinkage.

Process Optimization Strategy

Based on the simulation diagnosis, the precision investment casting process was comprehensively redesigned with the following key modifications:

1. Gating System Redesign: From Top-Pour to Bottom-Pour.
The initial top-pour design was abandoned in favor of a bottom-gated system. The sprue was attached to the bottom center of the impeller hub. This ensures that the mold cavity fills from the bottom upwards, promoting stable, progressive filling. The hottest metal remains at the top of the casting (the blade tips and outer rim) throughout the process, while the gate area, which solidifies last, acts as an effective feed source for the central hub.

2. Cluster Layout and Feeding Enhancement.
To improve yield and thermal mass, a cluster of four impellers was designed around a central downsprue. The horizontal runners were lengthened and strategically placed to ensure even metal distribution. A large-diameter, reusable spherical-bottomed centrifugal pouring cup was employed. This cup provides a substantial reservoir of hot metal for effective “hot topping” or feed metal supply during the critical late stages of solidification, compensating for shrinkage.

3. Centrifugal Parameter Optimization.
The simulation was used to optimize the centrifugal radius. Multiple simulations with varying effective radii showed that a radius of 450 mm provided the most stable, controlled rise of the metal front within the mold cavity, minimizing turbulence and promoting directional solidification towards the feeder. The centrifugal force pressure \( P_c \) acting on the metal is given by:
$$ P_c = \frac{1}{2} \rho \omega^2 (r_2^2 – r_1^2) $$
where \( \rho \) is the melt density, \( \omega \) is the angular velocity, and \( r_2 \) and \( r_1 \) are the outer and inner radii of the liquid metal ring. Optimizing the radius \( r_2 \) (through cluster layout) increases this feeding pressure, helping to push metal into shrinking regions.

Aspect Initial Process Optimized Process Rationale for Change
Gating Orientation Top-Pour, Center-Gated Bottom-Pour, Center-Gated Promotes stable fill, keeps hottest metal in casting, enables effective feeding from below.
Pouring Cup Standard Cup Spherical-Bottomed Hot-Topping Cup Provides large thermal mass reservoir for prolonged feeding against shrinkage.
Cluster Layout Single Casting Cluster of 4 Castings Improves furnace yield; longer runners aid in temperature distribution and feeding pressure.
Centrifugal Radius ~250 mm (implied) ~450 mm (optimized) Increases feeding pressure (P_c ∝ r²), ensures smoother cavity fill and better feeding.
Simulation Use Post-mortem Analysis Predictive Design & Optimization Used to test variations virtually before tooling commitment, ensuring optimal thermal and flow fields.

Simulation Validation and Final Production Results

The optimized process design was modeled again in ProCAST using the same rigorous boundary conditions but with the new geometry and an effective centrifugal radius of 450 mm. The results were markedly different:

1. Temperature Field: The simulation showed a much more uniform temperature distribution during filling. The thermal gradient between the incoming metal and the mold walls was reduced, and the solidification sequence became more directional, progressing from the blade tips back towards the central hub and the feeder.

2. Defect Prediction: The areas previously flagged for high shrinkage risk were significantly reduced or eliminated. The simulated solidification pattern indicated a clear path for feed metal from the large central sprue and hot-top cup into the last-solidifying sections of the hub.

3. Filling Pattern: The bottom-fill design showed a calm, progressive upward advance of the liquid front, virtually eliminating the conditions that lead to cold shuts (entrapment of air or premature freezing of merging fronts).

Encouraged by the simulation results, the optimized process was put into production. New wax injection tools were fabricated with the corrected double shrinkage allowance. The shells were built, and the castings were produced using the same vacuum skull furnace parameters (15 kA, 40 V, 250 rpm) but with the new bottom-pour cluster setup.

The final castings, after standard post-processing (cut-off, grinding, shot peening), demonstrated a dramatic improvement:

1. Surface Quality: The castings were free from the severe cold shuts observed previously. The surface was smooth with sharp, well-defined edges on the complex blade geometries.

2. Internal Quality: Radiographic inspection (X-ray) confirmed the absence of shrinkage cavities, hot tears, or significant porosity. The internal soundness met the required Grade C specification.

3. Dimensional Accuracy: 3D laser scanning and comparative analysis against the nominal CAD model showed that the vast majority of the casting profile, including the critical blade contours, was within the ±0.5 mm tolerance band. Any minor deviations were uniformly distributed, indicating excellent process control for dynamic balance. The angular relationship between the blades was also confirmed to be within the 10-arcminute specification.

Conclusion and Perspectives

The successful production of this high-precision, high-integrity titanium impeller underscores the transformative power of integrating advanced numerical simulation into the precision investment casting development cycle. The journey from a defective prototype to a qualifying component was accelerated by using ProCAST FEA software not merely as a post-mortem diagnostic tool, but as a proactive design and optimization platform. The software’s ability to accurately replicate complex physics—including free surface flow under rotation, heat transfer in a vacuum via radiation, and solidification shrinkage—allowed for the virtual testing and refinement of multiple process concepts without the cost of physical tooling and metal.

The key lessons and optimized strategies that emerged are fundamental to advancing precision investment casting for challenging titanium alloys:

1. Thermal Management is Paramount: For high-melting-point, low-superheat metals like pure titanium, the gating system must be designed to maintain thermal gradients favorable to complete filling and directional solidification. A bottom-gated system proved far superior to a top-gated one for this geometry.

2. Feeding Must Be Active and Sustained: Combining centrifugal force (with an optimized radius) with a substantial hot-topping feeder provides the necessary pressure and liquid metal reserve to compensate for shrinkage effectively, eliminating porosity.

3. Precision is Built on Compensation: Achieving net-shape dimensional accuracy in precision investment casting requires meticulous accounting for the sequential contractions of the wax pattern and the metal alloy, validated through iterative simulation and prototype measurement.

The final process configuration—featuring a bottom-pour gating system, a multi-cast cluster layout with extended runners, an optimized centrifugal radius, and a robust hot-top—represents a validated solution for this class of component. This methodology demonstrates a clear pathway for the manufacture of other complex, high-performance titanium castings, reducing lead time, improving first-pass yield, and ensuring component reliability in demanding applications. The fusion of foundational casting principles with modern simulation-based engineering is the cornerstone of next-generation precision investment casting capabilities.

Scroll to Top