Engine flywheels are fully machined and dynamically balanced, so a casting must combine dimensional stability with internal soundness. The trial component is a blank of Ø274 mm × 32 mm, with section thickness from 14 to 32 mm and a mass of 8.3 kg, cast in HT250 low-alloy gray iron. The specification requires graphite length grade 4–7, A+B graphite with A ≥ 70% and no C-type graphite, ferrite < 10%, and a hardness of 187–241 HBS measured at least 1.5 mm below the as-cast surface, with a hardness spread no greater than 25 HBS. These requirements make the flywheel a useful test case for lost foam castings: the shape is not complex, but the part is fully machined and must be free of surface and internal defects.
Lost foam casting uses an expanded polystyrene (EPS) pattern bonded into a cluster, coated with a refractory layer, buried in compacted dry sand, and poured under atmospheric or vacuum pressure. The metal vaporizes the pattern and replaces its shape. Because pattern gas evolution, coating permeability, sand compaction, vacuum level, and pouring temperature interact, the process is a chain rather than a single operation. A defect can originate in pattern density, be amplified by coating damage or loose sand, and become visible only after machining. The production trial therefore treated pattern quality, cluster design, coating, molding, melting, and pouring as one controlled system, with the aim of reducing flywheel deformation, slag inclusion, and iron penetration with sand burn-on.
Casting and alloy requirements
| Item | Requirement |
|---|---|
| Blank size | Ø274 mm × 32 mm |
| Section thickness | 14–32 mm |
| Blank mass | 8.3 kg |
| Material | HT250 low-alloy gray iron |
| Chemical composition | C 3.10–3.30%; Si 1.85–2.40%; Mn 0.60–0.95%; P 0.070–0.10%; S 0.070–0.10% |
| Tensile strength | ≥ 250 MPa |
| Graphite | Length grade 4–7; form A+B, with A ≥ 70%; C-type graphite not permitted |
| Ferrite | < 10% |
| Hardness | 187–241 HBS after at least 1.5 mm of surface removal; hardness spread ≤ 25 HBS |
| Additional treatment | Stress relief; full machining; dynamic balancing of the finished part |
Pattern density and pattern conditioning
The EPS pattern is both the shape carrier and the main internal gas source during pouring. In the trial, an EPS grade B106 was selected after preliminary material trials. The pre-expansion bulk density was controlled to 18–20 g/L so that the molded pattern density could be held at 20–22 g/L. This range is a compromise: a pattern that is too dense generates more decomposition gas and residue, while a pattern that is too light may lack handling strength and dimensional stability.
After pre-expansion, the EPS beads were aged for 4–8 h before molding. The flywheel pattern was produced as a single piece on a hydraulic semi-automatic molding machine, using a tool designed to avoid separate hub and rim sections that would require additional bonding. The molded pattern was then held at ambient temperature for 10–15 days to allow residual water and blowing agent to diffuse out. Before coating, the patterns and gating system were dried at 40 ± 5 °C and a relative humidity below 30% until fully dry.
Finishing and assembly were kept simple but controlled. Flash and burrs were removed, damaged surfaces were repaired, and critical dimensions were checked. The pattern and gating system were joined with hot-melt adhesive. The adhesive was applied evenly and in the smallest amount consistent with a sound joint, because excess adhesive increases local gas evolution and can leave carbonaceous residue or voids.
Coating, cluster assembly, and gating
The refractory coating must resist metal penetration while allowing pattern decomposition products to escape. A commercial lost foam coating was applied in two coats, with drying between coats, to a total thickness of 0.7–1.0 mm. A coating that is too thin is easily breached by metal, especially at high pouring temperature or high vacuum; a coating that is too thick lowers permeability and can cause gas entrapment, cold shuts, or coating spalling.
The trial used an open gating system with \(F_{\text{sprue}} \lt F_{\text{runner}} \lt F_{\text{gate}}\). One cluster contained 16 flywheel patterns and was poured in 30–35 s. Ingate position was treated as a critical variable. The ingates were placed to distribute metal without direct impingement on fragile pattern walls, to avoid local overheating of the coating and sand, and to promote progressive filling of the hub and rim. The cluster also required adequate support and connection, because an unsupported thin section can deform during handling, coating, drying, and sand compaction.
Sand compaction and flask preparation

The flask was filled with 30–50 mesh dry sand. A 120 mm bottom sand layer was added and compacted before the coated pattern cluster was placed. The cluster was positioned with the pouring cup as close as practical to the flask edge to give the ladle a stable approach. The flask was then filled in two stages. The first fill covered the flywheel patterns and was compacted for 20–25 s at a controlled frequency. The second fill provided the cover sand. Sufficient cover sand was required to prevent mold dilation and “flask rise” during pouring.
After compaction, the sand surface was leveled and kept 25 mm below the top of the pouring cup, with the cup fully exposed. A plastic film was placed over the sand surface and covered with a 20 mm protective sand layer, also leveled. The film helps maintain vacuum during pouring, while the protective sand prevents the film from being damaged by spilled metal or ladle contact. Over-vibration can distort the pattern or loosen the sand around thin sections; under-vibration leaves voids and sand bridges that promote metal penetration and dimensional error.
Melting, inoculation, and pouring
The iron was melted in a 2.0 t medium-frequency induction furnace. The tap temperature was controlled to 1560–1580 °C. Stream inoculation was used with a ferrosilicon inoculant at 0.2–0.3%. A teapot ladle was preheated to a dark red heat before use. The pouring cup was filled with a small initial stream to establish a controlled start; after black smoke appeared and the suction sound of the vacuum system was heard, the flow was increased. The flow was reduced again before the cup became full, so that the cup remained filled without overflow. This sequence limits turbulence and slag carryover while maintaining enough metal flow to replace the vaporizing pattern.
| Process step | Parameter |
|---|---|
| Pattern material | EPS grade B106 |
| Pre-expansion bulk density | 18–20 g/L |
| Molded pattern density | 20–22 g/L |
| Bead aging before molding | 4–8 h |
| Pattern ambient aging | 10–15 days |
| Pattern drying | 40 ± 5 °C; relative humidity < 30% |
| Coating | Two coats; total thickness 0.7–1.0 mm; dry between coats |
| Gating | Open system, \(F_{\text{sprue}} \lt F_{\text{runner}} \lt F_{\text{gate}}\); 16 flywheels per cluster |
| Molding sand | 30–50 mesh dry sand; 120 mm bottom sand |
| Compaction | First fill covers patterns; vibration 20–25 s; second fill; 20 mm protective sand over film |
| Pouring vacuum | −0.020 to −0.030 MPa gauge |
| Tap temperature | 1560–1580 °C |
| Final pouring temperature | > 1480 °C |
| Pouring time | 30–35 s per cluster |
| Cooling before shakeout | 1.0 h in the flask |
Defect mechanisms and control priorities
Deformation. The flywheel has a relatively thin rim and a thicker hub, so differential cooling and uneven sand support can produce out-of-flatness or runout. The trial addressed this through a one-piece pattern, controlled pattern density and aging, minimal adhesive, adequate cluster support, uniform sand compaction, and a 1.0 h in-flask cooling period before shakeout. Early shakeout removes support while the casting is still hot and is therefore a major risk for distortion.
Slag inclusion. Slag and pattern residue can be trapped when filling is turbulent or when metal arrives too cold. The open gating system, teapot ladle, stream inoculation, controlled vacuum, and two-stage pouring sequence were used to reduce turbulence and slag carryover. The final pouring temperature above 1480 °C provides margin against cold shuts and viscous slag, but excessively high temperature increases gas evolution and metal penetration.
Iron penetration and sand burn-on. Metal can enter the sand if the coating is too thin, damaged, or poorly dried, or if sand compaction is insufficient. The trial controlled coating thickness at 0.7–1.0 mm, applied two coats with intermediate drying, used 30–50 mesh dry sand, compacted in two stages, and kept the vacuum within −0.020 to −0.030 MPa. Vacuum that is too low leaves decomposition gas in the mold; vacuum that is too high can draw metal into the sand and worsen penetration.
Gas-related defects. Pattern density, pattern aging, adhesive quantity, coating permeability, and vacuum all affect the volume and removal of decomposition products. The 20–22 g/L pattern density and 10–15 day ambient aging were selected to reduce residual volatiles, while the coating and vacuum system were balanced to remove gas without causing metal penetration.
Trial outcome and production implications
The process trial showed that lost foam casting is feasible for the HT250 flywheel blank. The recent trial reached a blank qualification rate above 85%, and the production target is above 90% with tighter management and operating discipline. The result supports the use of lost foam casting for a fully machined flywheel blank, provided that pattern density, coating integrity, cluster support, sand compaction, vacuum, and pouring temperature are maintained as a coupled process window.
The reported 85% is an overall blank qualification rate, not a defect-by-defect breakdown, and the process changes were implemented as a package rather than as a factorial experiment. For a rotating part, blank qualification is necessary but not sufficient: machining allowance, section soundness, hardness distribution, and dynamic balance of the finished flywheel remain separate acceptance requirements. The parameter window is also specific to this 8.3 kg, Ø274 mm geometry and HT250 alloy; transfer to other flywheel sizes or section thicknesses requires revalidation of pattern density, gating, coating, and pouring conditions.
