Fan shaft casting design: simulation verification and trial casting evaluation

Journal of Materials and Manufacturing 4(2): 25-40 (2025) * Corresponding authors E-mail addresses: [email protected], [email protected] DOI: 10.5281/zenodo.18062212 Received: 10 November 2025, Revised: 15 December 2025, Accepted: 24 December 2025 ISSN: 2822-6054 All rights reserved. Fan shaft casting design: simulation verification and trial casting evaluation Lemuel N. Apusaga*, Earl John T. Geraldo, Karen C. Santos, Joey G. Pangilinan, Alistaire Kerwin A. Acma, and Key T. Simfroso* Materials and Process Research Division, Department of Science and Technology – Metals Industry Research and Development Center (DOST-MIRDC), General Santos Ave., Bicutan, 1631 Taguig City, Metro Manila, Philippines (ORCID: 0000-0002-2491-8161), [email protected] (ORCID: 0000-0002-3078-8201), [email protected] [email protected] (ORCID: 0000-0003-2678-9989), [email protected] (ORCID: 0009-0007-2826-1885), [email protected] (ORCID: 0000-0002-3066-0179), [email protected] Abstract A fan shaft casting design intended for high-temperature service was developed and evaluated through simulation and trial casting. Initial casting simulations using NovaFlow&Solid were conducted to identify an optimal gating and riser design for sand-casting process. A rectangular tapered sprue (52 mm × 26 mm entrance area, taper angle 1°) was employed for manufacturability and to stabilize gravity filling. Melt entry was controlled using a Weir-type pour basin (depth 100mm, Weir radius 2.5mnm) coupled with an offset dross-trap to reduce turbulence and surface oxidation. Continuous tangential gating (length 667.5 mm, thickness 12 mm, width 150 mm, 5° taper) induced a controlled vortex fill, promoting nonturbulent filling behavior while maintaining thermal segregation of hotter metal toward the top of the shaft along the gate length. Directional solidification was supported by a ~1° bottom-shaft taper and adding an exothermic sleeve feeder above the blade-shaft junction. Among four exothermic feeder sleeves evaluated simulated, the optimal riser was an exothermic sleeve (internal volume: 577,267.65 mm3) with a 70 mm neck diameter, providing 17% utilization while eliminating shrinkage porosity at the blade root. The resulting design parameters were then applied to a no-bake molding system, where the final fan shaft was cast using SCH 15 stainless steel. The cast fan shaft was assessed through inspections, followed by sectioning to evaluate internal soundness. The casting trial results showed with most sections free from shrinkage cavities and exhibiting only minimal surface depressions. The observed deviation between the simulated and experimental casting results arises from the software’s limited material database, which lacked the specific alloy employed in the actual casting trials, and also by several additional factors inherent to real casting conditions which were assumed constant during simulation. However, the final casting output demonstrated fair agreement with the simulation results which confirms the effectiveness of the molding system and the adequacy of the developed casting design for producing a sound fan shaft component. Keywords: casting design; casting simulation; fan shaft; no-bake molding system; sand casting. 1. Introduction Casting is one of the most practical and widely used metal-forming processes in the industry. It involves melting a metallic charge, pouring the molten metal into a mold, and letting it solidify into the required shape. This process makes it possible to produce complex components efficiently and at lower cost, without the need for excessive machines or fabrication. Because of its versatility—especially in sand casting—it remains essential for manufacturing large and intricate parts like fan shafts, impellers, housings, and pump bodies. In many industries, casting continues to be a key process not just for productivity but also for sustainability, as it allows for the reuse of scrap metal and supports more resource-efficient production [1]. Journal of Materials and Manufacturing 4(2): 25-40 (2025) 26 In any casting process, the way molten metal flows, cools, and solidifies directly affects the final quality of the product. Common issues such as shrinkage porosity, misruns, cold shuts, and gas inclusions usually come from poor gating or riser design, or improper feeding during solidification [2]. These defects can impact the mechanical strength of components, especially rotating parts like fan shafts that are exposed to continuous stress and high temperatures. Because of this, it is important to control and optimize casting parameters to achieve good metallurgical quality and dimensional accuracy. With the help of modern casting simulation software, foundries today can predict how molten metal behaves even before production starts. Programs like NovaFlow & Solid (NovaCast), MAGMASoft, and ProCAST allow engineers to visualize metal flow, heat transfer, and solidification patterns to identify where defects might form [3]. By simulating the process, it is possible to adjust casting parameters such as gating and riser design, pouring temperature, and cooling rate to minimize errors. This reduces material waste, shortens trial runs, and improves overall yield and process consistency. According to Jin, 2025, optimizing sand-casting parameters through simulation can help eliminate unfilled regions and reduce shrinkage by improving feeding and solidification of uniformity [4]. His simulation-based design for a geometrically complex backward-curved fan produced a defect-free casting, with mechanical performance showing negligible deformation and stress that validated the approach. Likewise, Sultana et al., 2019 showed that using casting simulation together with actual validation tests improves defect prediction accuracy and aligns well with real microstructural results [1]. In their work, experimental trials confirmed that a proposed gating and feeding design for an aluminum flywheel improved casting yield by approximately 15% compared with conventional practice. Anggono et al., 2020 also reported a casting design, simulation and manufacturing validation of air compressor fan blade, establishing that defects are predictable and that process and design improvements can be finalized ahead of actual casting [5]. More recently, Patwari et al., 2024 also found that combining manual and digital optimization techniques in casting improve reliability and product quality [6]. These studies clearly show how simulation-based design connects computational modeling with actual foundry performance. Most studies that integrate manual and digital approaches in casting primarily focus on symmetric metal components such as fans, impellers, and similar geometries [5, 7-9]. However, to date, there is a notable lack of literature addressing the casting of fan-like components featuring very thin blades attached to a long, slender rod which poses a complex type of component. Casting, whether conducted manually or through simulation, remains highly dependent on practitioner expertise, and producing this type of metal component is quite challenging. This study focuses on the design and development of a fan shaft casting intended for high-temperature service. The selected material, SCH 15 stainless steel, is a heat-resistant alloy suitable for components exposed to oxidizing environments and mechanical stress. The work combines simulation verification and trial casting evaluation to develop a sound and defect-free component. Initial simulations using NovaFlow&Solid were carried out to refine the gating and riser design for sand-casting process. The design was applied in a no-bake, Furan in particular, molding system. Despite minor casting defects observed in the initial trials, the design serves as a promising foundation for further improvement through advanced simulation tools and optimized pouring parameters. 2. Methodology 2.1 Model Generation Parametric CAD modeling was used to generate the 3D models and technical drawings, which served as the basis for casting, machining, and assembly. Notably, the part design features multiple diameters in a stepped fashion reaching nearly one meter in length and tapered blades going down to a thickness of 9 mm, as can be seen in Fig. 1. The combination of the long arm and relatively thin features poses a significant challenge in ensuring sound cast to meet its service requirements. The casting cavity model was designed with a 1.5% shrinkage factor and included machining allowances. The model was then adjusted according to Fig. 2 and 3 to account for machining allowance. Noticeably, the machining allowance is larger for the end section, as this would be the determined placement for the feeder. The casting cavity model is an upscaled model of the model with machining allowance, by a 1.5% factor, to Journal of Materials and Manufacturing 4(2): 25-40 (2025) 27 account for the shrinkage of the material as common practice by the foundry for steel alloy casts. This casting allowance was also adopted by the researchers for this product development and was applied when making the pattern. Fig. 1. CAD model and dimensional drawing of the fan shaft, showing stepped diameter transitions, along the shaft body, and blade sections with a minimum thickness of 9 mm. Fig. 2. Adjusted casting model incorporating machining allowance. Fig. 3. Overlay of actual part (light area) to design with machining allowance (shaded area). Journal of Materials and Manufacturing 4(2): 25-40 (2025) 28 2.2 Casting Design and Simulation The supposed fabrication of the fan shaft is to cast SCH 15 stainless steel, with its chemical composition presented in Table 1, along with the report of the metal composition used during the cast. The casting simulation was conducted using NovaFlow&Solid 6.5 Release 3 software. However, as the version employed was an educational license with a limited material database, the specific grade SCH 15 was not available. SCH 15 is a Japanese Industrial Standards (JIS) G5122 specification and an equivalent alloy under the DIN (German Institute of Standardization or “Deutsches Institut für Normung”) standard is the GX40CrNiSiNb38-18 [10,11]. To approximate its behavior in the simulation, the alloy GX40CrNiSiNb38-19—chosen for its closely comparable chemical, physical, and mechanical characteristics [12,13]—was selected as an alternative material. Its corresponding chemical composition and relevant properties are presented in Table 2. Table 1. Chemical composition, physical and mechanical properties of SCH 15. Chemical Composition Physical and Mechanical Properties Element Range Actual Remarks Properties Physical Mechanical C 0.35 – 0.70 0.048 Pass Tensile strength 115-234 231-231 σb/MPa Si 2.5 max 0.416 Pass Yield strength 23 15 σ 0.2 ≥/MPa Mn 2.0 max 0.527 Pass Elongation 65 56 δ5 ≥ (%) P 0.04 max 0.0023 Pass S 0.04 max <0.0010 Pass Cr 15.0 – 19.0 15.86 Pass Ni 33.0 – 37.0 33.25 Pass Mo 0.126 Al 0.0129 Co 0.112 Cu 0.0858 Ti 0.0100 V 0.0431 Table 2. Chemical composition, physical and mechanical properties of GX40CrNiSiNb38-19. Chemical Composition Physical and Mechanical Properties Element Composition Properties Physical Mechanical Fe 38.09 Tensile strength 115-234 231-231 σb/MPa Ni 37.50 Yield strength 23 15 σ 0.2 ≥/MPa Cr 19.50 Elongation 65 56 δ5 ≥ (%) Si 1.70 Mn 1.00 C 0.40 Mo 0.25 Nb 1.50 P 0.035 S 0.025 Journal of Materials and Manufacturing 4(2): 25-40 (2025) 29 The simulation setup began with importing the 3D model of the casting design, which included all essential appendages such as the sprue, runners, ingates, feeders, and vents. Additional components, including exothermic sleeves, chills, and filters, were incorporated as required. After importing the necessary geometries, the gravity direction was defined, and a mold box representing the cavity boundaries was generated as shown in Fig. 4. Fig. 4. Simulation setup displaying the complete casting geometry, added feeding and gating elements, and the defined mold box and gravity direction, as defined by orientation. Thermal boundary conditions were specified for different material interfaces, namely casting-to-mold, casting-to-atmosphere, and mold-to-atmosphere interactions, illustrated in Fig 5. These boundary conditions are used by the software to facilitate interactions between the different materials. Additionally, a casting-toatmosphere zone was manually added to the riser metal since it is not on the same height plane as the basin. The mesh size was selected to achieve an optimal balance between computational accuracy and simulation efficiency. A minimum of three mesh cells at the smallest feature is maintained throughout the iterations. This is done as a general rule of thumb for meshing, as recommended by NovaCast. After the setup of materials, the resulting mesh model was then used to define the casting process simulation parameters, which are summarized in Table 3. Fig. 5. Cross-section of mold (orange), fan shaft (yellow), gate system (light blue), riser (green) and feeder metal (red) showing the mesh cells efficiency. Journal of Materials and Manufacturing 4(2): 25-40 (2025) 30 Table 3. Material simulation parameters. Solid Names Materials Temperature (ºC) Fan shaft GX40CrNiSINb38-19 1650.00 Mold material Core – Furane Sand 24.00 Cavity medium Air – In mold 20.00 Throughout the simulation, a real-time graphical interface enabled monitoring of key variables such as temperature distribution, solidification behavior, shrinkage formation, and defect prediction which facilitates iterative improvements to the casting design. Once an acceptable design was determined, the corresponding tooling was fabricated, and the process proceeded to the trial casting stage. 2.3 Casting Process The molten metal used for casting was prepared using an induction furnace, which provided precise temperature control and uniform melting. The temperature of the molten metal was monitored using both an infrared thermometer and a dip pyrometer to ensure measurement accuracy and consistency. Once the target tapping temperature was achieved, the molten metal was carefully poured into the prepared molds to minimize turbulence and potential oxidation during filling. The poured metal was measured with the infrared thermometer. For the fan shaft casting, a purely no-bake molding system was employed during the trial casting process. The no-bake mold was prepared in accordance with the standard procedure described in Ref. [14] using sand mold. The minimum mold thickness is 25 mm. The casting was conducted by pouring SCH 15 stainless steel. A total of four casting trials were conducted. In the first three trials, pouring was performed at the target temperature of 1650°C, with controlled variations to evaluate the effect of using an alloy different from that assumed in the simulation. The fourth trial incorporated insights from the initial trials to improve the casting of the fan shaft. For this trial, four molds were prepared, each assigned a specific pouring temperature: 1590°C for Mold 1, 1580°C for Mold 2, 1570°C for Mold 3, and 1560°C for Mold 4. Fig. 6 and 7 show the molding system used during these trials. The resulting fan shaft casting featured thin sections and an overall length of nearly one meter. Fig. 6. Flask and no-bake molding setup: (a) flask containing the fan shaft pattern: a pattern for part of the rectangular sprue and the continuous runner gate pattern positioned tangentially to the shaft; (b) no-bake blade shaft junction mold pattern; and (c) molded blade shaft junction. (a) (b) (c) Journal of Materials and Manufacturing 4(2): 25-40 (2025) 31 Fig. 7. Final mold assemblies for casting trials: (a) closed mold assembly highlighting the structural interfaces between shaft body, blade junction and feeder neck; (b) pouring cup with sprue extension positioned to support gravity filling and exothermic sleeve fitted; and (c) mold fitted with counterweight for stability during pouring. 2.4 Evaluation of Cast Fan Shaft The cast fan shaft was assessed through a combination of visual inspection and physical verification to determine whether the component was successfully produced according to the simulation-driven mold and process design. First, the as-cast surface was examined to identify any apparent defects such as misruns, surface porosity, cold shuts, or shrinkage indications. Following this, the machining allowance along the shaft margin was removed to expose the underlying material and to confirm that the final geometry conformed to the intended design generated by the casting simulation. To further validate internal soundness, the shaft was sectioned at selected locations. These cross-sections were examined to check whether the internal features and solidification patterns were consistent with the predicted simulation results, particularly in regions where higher thermal gradients or defect risks were anticipated. This evaluation focused primarily on qualitative and geometry-based verification. Mechanical characterization, such as hardness testing, surface roughness measurements, tensile testing, or other quantitative assessments, was not conducted as part of the trials. This constitutes one of the limitations of the present work, since the absence of post-casting mechanical measurements prevents a more comprehensive assessment of the material’s performance relative to the simulation predictions. 3. Results and Discussion 3.1 Simulation Results Fig. 8 presents the casting design of the fan shaft, developed through iterative modeling and casting simulation. The simulation was split into two parts for calculation efficiency, a filling simulation and a solidification simulation. For filling simulation, the basin, sprue, and gate design was investigated and iterated. The design and filling simulation goal was for a non-turbulent metal flow into the part body, while ensuring that the hotter metal would fill up the height of the shaft to slow down the solidification along the blades. This was achieved with the uprunner thin gate that is tangentially connected along the length of the shaft. This helped with smoothing the flow of metal since the high entry speed was redirected into a vortex that slowly fills up the shaft. The waterfall effect was also minimal. A Weir type pouring basin was also utilized with a depth of 100mm and Weir radius of 2.5mm to ensure smooth entry down the sprue and minimizes trapped air once the sprue is filled. The sprue was set to be rectangular, for ease of fabrication, with dimensions 52mm by 26mm and tapered by 1° along the length. The solidification model is where the feeding system was iterated until the final design (including the exothermic sleeve) was determined. The results of the casting simulations for various design configurations were evaluated, and the optimal design was identified as the one exhibiting the casting defects. (a) (b) (c) Journal of Materials and Manufacturing 4(2): 25-40 (2025) 32 The primary casting defects encountered during the iterations considered included shrinkage cavities, hot and cold tears, and cold laps. The most common defect encountered is the isolated shrinkage field in the bladeshaft junction due to early closing off of the feeder shaft interface before the blade-shaft hotspot is fully fed. To resolve this issue, a ~1° taper along the shaft was introduced to induce directional solidification from the bottom thinner cross-section upward and exothermic sleeves were incorporated. Four (4) available sizes of exothermic sleeves, as listed in Table 4, were examined to evaluate their effectiveness in minimizing shrinkage porosity. After several design iterations exploring different taper degrees, fan shaft orientation, feeding system designs and feeder placements, the design shown in Fig. 8 successfully eliminated the centerline shrinkage porosities below the blades. Table 4. Volume calculations conducted comparing sleeve volume and sleeve neck diameter to keep it close to the original riser dimensions. Fig. 8. Final fan shaft casting design derived from iterative simulations, showing the sprue, runner, dross trap, and riser layout used for gravity-fed pouring. The design utilized exothermic sleeves, where Fig. 9 and 10 illustrate the solidification simulation and fullness using various sizes of sleeves. As illustrated, the shrinkage was completely eliminated from the blade junction, with sleeve no. 1 showing approximately 15% utilization of the riser volume. Exothermic sleeve no. 2, similar to sleeve no. 1, showed no shrinkage at the junction. Sleeve no. 2 was smaller by 42% in volume and had a larger neck diameter. Fullness, however, was still at 83.092% or approximately 17% utilization, indicating that a smaller-volume exothermic sleeve was still acceptable. Sleeve no. 3, on the other hand, showed a shrinkage of about 0.1% by volume, which might have been attributed to the smaller neck diameter that caused pinching of the molten metal at the riser–part junction. Exothermic sleeve no. 4 introduced more shrinkage at the junction compared to sleeve no. 3, which was due to a more drastic diameter difference between the feeder Sleeve No. Original riser volume (mm3) Exothermic sleeve volume (mm3) Feeder volume vs original (%) Fan shaft diameter (mm) Neck diameter of sleeve (mm) 1 1,064,100.05 995,492.17 93.55% 67.45 65 2 577,267.65 54.25% 70 3 1,428,714.06 134.27% 50 4 773,460.11 72.69% 40 Journal of Materials and Manufacturing 4(2): 25-40 (2025) 33 neck and the shaft-to-riser junction. Based on these observations, sleeve no. 2 was used during the casting and simulation process. Fig. 9. Solidification simulation result from NovaCast illustrating temperature-driven progressive solidification of the shaft and blade geometry when exothermic feeding sleeve no. 4 is applied. The diameter mismatch between the feeder and contributed to the simulated isolated shrinkage field in the blade-shaft junction. Fig. 10. Designs with exothermic sleeves (a) no. 1, (b) no. 2, (c) no. 3, and (d) no. 4 and the respective amount of metal remaining (by % volume). 3.2 Trial Castings Results Trial casts were conducted using the no-bake molding system, and each cast specimen is then inspected. External defects were considered acceptable if within the casting allowance, while internal defects are all unacceptable. For internal defects, the cast fan shaft specimen is sectioned and checked for internal shrinkages. While other non-destructive scanning methods exist, sectioning was deemed more practical and economical for product development. A total of four trial casts were conducted, as was the resource available for the project. The first three trial casts yielded no successful specimen and is summarized in Table 5. a) b) c) d) Journal of Materials and Manufacturing 4(2): 25-40 (2025) 40 [9] Zhang, Y., Li, Z-Y., Zhang, L-Y., Rong, B-S., Cai, Q-H., Zhu, R-F., & Xu, Y. (2018) Simulation and Optimization for Investment Casting of Impeller Based on 3D Printing. IOP Conf. Series: Earth and Environmental Science, 186, 012016. [10] Kennametal. (2021). Workpiece Materials Overview Listing. https://www.purdue.edu/bidc/wpcontent/uploads/2021/08/ISOGrade.pdf [11] Mitsubishi Materials Corporation. (n.d.). Material Cross Reference List. https://www.mitsubishicarbide.net/contents/mmus/enus/html/product/technical_information/information/ material_5.html [12] Steel Grades. SCH15 Data Sheet, https://www.steel-grades.com/Steel-Grades/MouldSteel/21/9736/SCH15.pdf [13] Steel Grades. GX40NiCrSiNb38-19 Data Sheet, https://www.steel-grades.com/Steel-Grades/MouldSteel/21/2160/_GX40NiCrSiNb38-19_.pdf [14] ASM Handbook (2008). Volume 15 - Casting, Chapter: No-Bake Sand Molding, 567 - 580.