Lost foam casting has been widely adopted in the production of iron castings due to its excellent surface finish, dimensional accuracy, and lower production cost. However, for steel castings, the application of the lost foam process is relatively limited because of the inevitable carbon pickup from the decomposition of the foam pattern. In our foundry, we decided to explore the feasibility of using the lost foam process for steel castings by selecting the tractor traction frame, a critical component for small and medium horsepower tractors. The traction frame demands high quality and strict control of casting defects such as shrinkage, slag inclusion, and cracks. If the lost foam process could successfully produce this part, it would not only enable mass production but also provide a practical reference for other steel castings. This article summarizes our entire development work, including pattern material selection, coating, molding, pouring, composition adjustment, heat treatment, and cost analysis, all focused on realizing reliable lost foam castings for steel parts.
The traction frame has a “U” shape with a single casting weight of 12 kg. The material specification is ZG310-570 with the main wall thickness of 15 mm and local thickness of 20 mm. Reinforcing ribs are present at the corners, which create hot spots and may lead to distortion. The technical requirements do not allow any shrinkage cavities, slag entrapment, or cracks. To meet these requirements, we designed a casting process as shown in the schematic below. The pattern was designed with one pattern producing two castings, using two hot risers with dimensions of φ90 mm × 120 mm for feeding. The gating system and the two patterns were assembled into a single unit so that the whole white mold could be removed from the molding tool as one piece. The sprue was placed in the center and connected to a runner. The upper end of the sprue had a circular recess and the lower end had a circular protrusion to facilitate layer-by-layer assembly of multiple patterns.
The runner inlet was lower than the outlet, which ensures that during vertical multi-layer pouring, the molten steel fills the lower mold first and then enters the upper mold after the lower cavity is full. This integrated pattern assembly also helps reduce casting deformation after pouring. For the first trial, we used a copolymer STMMA foam material to make the pattern with a foaming density of 23 g/L. The pattern assembly consisted of two patterns per group and two layers in height direction, giving four castings per cluster. A carbon steel No. 2 dry powder coating was used with a mixing ratio of dry powder to water of 1:0.8. After mixing for 3 hours, the Baume degree of the coating was measured as 63. The white patterns were dipped and then dried in a drying room at 40–50°C for 8 to 24 hours. After the first dip and dry, the patterns were dipped twice more and dried each time. The yellow dried patterns were then touched up and repaired according to process requirements.
We conducted the first pouring experiment on our 3,000 t lost foam casting test line. The molten steel was melted in a 3-ton medium frequency induction furnace and transferred to a 200 kg tilting pouring ladle. The pouring temperature was set to 1,570–1,620 °C. The molding sand box vacuum was -0.05 MPa. We used manual tilting ladle pouring. In the first attempt, severe back-spraying of molten steel occurred, and the steel could not fill the cavity. The reasons for the steel back-spraying were analyzed as follows: (1) The gas evolution of the copolymer STMMA is high, about 1.5 times that of EPS; (2) during manual tilting pouring, the flow rate was too high, blocking the pouring cup, and a large amount of gas generated by thermal decomposition of the copolymer could not escape quickly through the coating layer, so some gas rushed back to the pouring cup; (3) after molding, we did not burn out the gating system of the yellow pattern, which led to excessive total gas generation.
Based on the first trial results and the failure analysis, we changed to EPS foam material for the second trial. The foaming density was reduced to 21 g/L. The pattern assembly, coating, and drying of the yellow patterns were the same as the first trial. After drying, we poked several vent holes of φ0.5 mm on the risers and casting parts of the yellow mold to facilitate the escape of gas generated during pouring. After molding and before pouring, the vacuum was applied to the sand flask at -0.05 MPa, and then we used an oxy-acetylene cutting torch to burn out the gating system. When the molten steel composition and temperature in the medium frequency furnace met the process requirements, we used a 1-ton shaking ladle to receive the molten steel. The ladle was preheated twice before receiving the steel to avoid rapid temperature drop. The pouring temperature was still 1,570–1,620 °C. The entire pouring process was smooth without any back-spraying. After pouring, the pressure was maintained for 15 minutes, and the flask was opened after 2 hours. The castings were removed and inspected. The overall shape was well formed, and the steel in the pouring cup had sunk significantly.
For lost foam castings, one of the key concerns is the carbon pickup from the decomposition of the foam pattern. The traction frame material ZG310-570 has the chemical composition requirements shown in Table 1. In the first trial melt, the molten steel composition was: 0.46% C, 0.86% Mn, 0.38% Si, 0.033% S, 0.032% P. After pouring, we cut samples from three different positions of the casting: position 1 near the ingate, position 2 at the middle section, and position 3 at the far end. The spectral analysis results are presented in Table 2. The carbon content increased in all positions, with the range of carbon pickup being 0.04%–0.09%. The other alloying elements showed negligible changes. In the second trial melt, the initial steel composition was 0.44% C, 0.88% Mn, 0.46% Si, 0.026% S, 0.049% P. The post-pouring compositions are listed in Table 3. Again, the carbon content increased by similar amounts. The measured carbon pickup across different positions suggests a non-uniform distribution due to the flow pattern and solidification sequence.
| Element | C | Si | Mn | S | P |
|---|---|---|---|---|---|
| Requirement | 0.40–0.50 | 0.30–0.60 | 0.70–1.10 | ≤0.04 | ≤0.04 |
| Element | Position 1 | Position 2 | Position 3 |
|---|---|---|---|
| C | 0.48 | 0.52 | 0.50 |
| Si | 0.43 | 0.42 | 0.42 |
| Mn | 0.82 | 0.80 | 0.81 |
| S | 0.026 | 0.026 | 0.025 |
| P | 0.032 | 0.033 | 0.030 |
| Element | Position 1 | Position 2 | Position 3 |
|---|---|---|---|
| C | 0.49 | 0.53 | 0.48 |
| Si | 0.44 | 0.42 | 0.43 |
| Mn | 0.78 | 0.80 | 0.82 |
| S | 0.024 | 0.026 | 0.025 |
| P | 0.039 | 0.038 | 0.038 |
The results clearly indicate that carbon pickup is the dominant chemical change in lost foam castings of steel. The carbon pickup can be expressed as:
$$ \Delta C = C_{final} – C_{initial} $$
where \( C_{final} \) is the carbon content measured in the solidified casting and \( C_{initial} \) is the carbon content in the molten steel before pouring. From our tests, the carbon pickup ranged from 0.04% to 0.09% depending on the location and process parameters. To compensate for this effect, we adjusted the target carbon content in the furnace. The adjusted molten steel composition is shown in Table 4. The carbon content was reduced to 0.36%–0.40% so that after the carbon pickup from the lost foam pattern, the final chemistry would fall within the required range. The other elements were maintained in the same ranges as the final specification because they did not significantly change during pouring.
| Element | C | Si | Mn | S | P |
|---|---|---|---|---|---|
| Target range | 0.36–0.40 | 0.30–0.60 | 0.70–1.10 | ≤0.04 | ≤0.04 |
In addition to composition control, we also implemented other measures to minimize carbon pickup and ensure sound lost foam castings. First, selecting EPS foam with a lower density of 21 g/L reduced the total amount of foam to be decomposed, thereby lowering the carbon source. Second, burning out the gating system before pouring eliminated the foam in the sprue and runner, so the molten steel would not decompose that portion of the pattern. Third, vent holes of φ0.5 mm were poked in the risers and remote sections of the pattern to facilitate the escape of pyrolysis gases, which reduced the gas pressure and the residence time of carbon-rich gases in contact with the molten steel. Fourth, a controlled pouring temperature of 1,570–1,620 °C was maintained to ensure complete decomposition of the foam and proper filling without excessive gas generation.
After the casting trials, the traction frame castings were subjected to normalizing heat treatment. The normalizing temperature was 880 °C ± 10 °C. The castings were held at that temperature for 2 hours and then cooled slowly in air. After heat treatment, we sampled the hardness from several castings and obtained HB 195, HB 190, and HB 193, all within the required hardness range of HB 156–217. The microstructure was examined and found to be uniform ferrite-pearlite, typical of normalized ZG310-570 steel. The castings also showed good surface quality with no visible shrinkage, slag, or cracks. Dimensional inspection confirmed that the lost foam castings met the drawing requirements, and the elimination of the parting line and core parting flash greatly reduced the need for post-casting grinding.
The success of the lost foam process for the tractor traction frame is not only a technical achievement but also a significant economic benefit. Compared with the conventional green sand casting process, the lost foam process eliminates many pieces of equipment and several production steps, such as sand mixing and core making. It also reduces the cleaning workload because there are no fins, parting lines, or core prints. The sand can be reused, and the number of production workers is reduced. Table 5 shows the cost comparison per ton of castings between the conventional green sand process and the lost foam process. The melting cost is slightly lower in the lost foam process because the adjusted carbon content slightly reduces the need for carburizer and the overall energy consumption. The molding cost is drastically reduced from 480 yuan per ton in green sand to 160 yuan per ton in lost foam. The core making cost in green sand is replaced by pattern making in lost foam, with a slight increase from 245 to 260 yuan per ton. Cleaning cost (including heat treatment) is reduced from 955 to 875 yuan per ton because of the cleaner casting surface and less fettling. The labor cost is reduced from 1,067 to 660 yuan per ton because fewer operators are needed. The total cost per ton of castings is reduced from 6,084 yuan to 5,228 yuan, saving 856 yuan per ton. This is a substantial saving for mass production of tractor components.
| Cost item | Green sand process | Lost foam process |
|---|---|---|
| Melting | 3,337 | 3,283 |
| Molding | 480 | 160 |
| Core/pattern making | 245 (core making) | 260 (pattern making) |
| Cleaning (including heat treatment) | 955 | 875 |
| Labor | 1,067 | 660 |
| Total | 6,084 | 5,228 |
The cost saving can be calculated by:
$$ \Delta Cost = Cost_{green} – Cost_{lostfoam} = 6,084 – 5,228 = 856 \text{ yuan per ton} $$
If the annual production of traction frames is, for example, 500 tons, the total annual saving would be:
$$ S = 856 \times 500 = 428,000 \text{ yuan} $$
This substantial saving, combined with improved surface quality and reduced environmental burden from dry sand reclamation, makes the lost foam process an attractive alternative for steel castings. In our further trials, we also applied the same principles to other steel castings with different wall thicknesses and weights. We found that the carbon pickup is influenced by several factors, including the foam density, the coating permeability, the vacuum level, the pouring temperature, and the geometry of the casting. For thicker sections, the carbon pickup tends to be higher due to the longer contact time between the molten steel and the pyrolysis products. Therefore, we established a correlation between the section modulus and the carbon pickup, which can be approximated as:
$$ \Delta C = k \cdot \frac{A}{V} \cdot \rho_{foam} \cdot t_{p} $$
where \( A/V \) is the surface area to volume ratio of the casting section, \( \rho_{foam} \) is the foam density, \( t_{p} \) is the filling/pyrolysis time, and \( k \) is an empirical constant that depends on coating, vacuum, and material type. This model helps us predict the amount of carbon pickup for new steel castings and adjust the initial carbon content accordingly.
Another important aspect in producing sound lost foam castings is the coating quality. The coating must be permeable enough to allow the escape of pyrolysis gases while maintaining sufficient strength to withstand the metallostatic pressure of the molten steel. We optimized the coating thickness and permeability by controlling the Baume degree at 60–65 and the drying parameters. We also found that the vacuum level of -0.05 MPa is critical. Too low a vacuum may cause gas defects, while too high a vacuum can cause the sand to collapse or produce rough surfaces. Our experiments showed that -0.05 MPa provides the optimal balance.
The foam pattern quality also plays a key role. We compared STMMA and EPS patterns. Although STMMA generally provides lower carbon residue in aluminum castings, for steel castings its higher gas evolution caused back-spraying. EPS with a lower density of 21 g/L, combined with burning out the gating system, gave the best result. The white mold assembly was designed to integrate the sprue, runner, risers, and castings into one piece, which not only facilitated handling and coating but also ensured dimensional consistency. The pattern density was strictly controlled to avoid excessive carbon source. We measured the density of the EPS boards and patterns regularly to ensure uniformity.
The pouring operation is also critical for lost foam castings. We learned from the first failure that the pouring rate must be controlled to keep the sprue filled but not overflowing. In the second trial, using a 1-ton shaking ladle with preheating provided a steadier stream. We maintained a continuous pouring without interruption to prevent the collapse of the foam pattern. The pouring temperature was monitored with an optical pyrometer. We observed that the metal front advanced smoothly through the foam, and the gas generated was evacuated through the coating and vent holes. The vacuum pump continued to operate during pouring and for a short time after pouring to remove any residual gases.
After the castings were shaken out, we performed a thorough inspection. The surface finish was significantly better than that of green sand castings. The as-cast surface roughness was measured and found to be Ra 6.3 µm, which is acceptable for the traction frame application. The dimensional tolerances were within ±1 mm for the critical mounting holes and ±2 mm for overall dimensions. The absence of parting lines and core fins reduced the cleaning time by about 30%. We also tested the mechanical properties of the castings after heat treatment. The tensile strength was measured as 610 MPa, which is well above the minimum of 570 MPa required for ZG310-570. The yield strength was 320 MPa and the elongation was 15%, both meeting specification. The impact toughness at room temperature was 35 J/cm², which is good for a tractor structural component.
In conclusion, we have successfully produced tractor traction frame steel castings using the lost foam process. The key technical measures include:
(1) Selection of EPS foam with a controlled density of 21 g/L to reduce gas evolution and carbon pickup.
(2) Burning out the gating system before pouring to eliminate the foam in the runner and sprue, thus reducing the total carbon source.
(3) Pocking φ0.5 mm vent holes in the risers and remote parts of the pattern to facilitate gas escape.
(4) Adjusting the molten steel carbon content to 0.36%–0.40% in the furnace, so that after the carbon pickup of 0.04%–0.09% from the lost foam pattern, the final composition meets the ZG310-570 specification.
(5) Maintaining a pouring temperature of 1,570–1,620 °C and a vacuum of -0.05 MPa to ensure smooth filling and proper venting.
These measures enabled us to produce lost foam castings with sound internal quality, excellent surface finish, and consistent mechanical properties. The cost analysis demonstrates that the lost foam process saves 856 yuan per ton compared to the conventional green sand process, mainly due to reduced molding, cleaning, and labor costs. This technical breakthrough expands the application range of the lost foam process beyond iron castings to steel castings, offering a new way to reduce production costs and improve quality in the foundry industry. We believe that with further optimization, the lost foam process can be applied to a wider variety of steel castings, provided that the carbon pickup issue is carefully managed through composition adjustment and process control. Our future work will focus on developing more accurate models for carbon pickup, improving coating formulations for lower gas resistance, and automating the pouring process to achieve even greater consistency.
Furthermore, the success of the lost foam castings depends heavily on the synergy between pattern material, coating, vacuum, pouring parameters, and metallurgical control. We have compiled the key process parameters in Table 6 for reference.
| Parameter | Value/Range |
|---|---|
| Pattern material | EPS |
| Foam density | 21 g/L |
| Coating type | Carbon steel No. 2 dry powder coating |
| Coating Baume degree | 60–65 |
| Coating drying temperature | 40–50 °C |
| Vent hole diameter | 0.5 mm |
| Sand vacuum | -0.05 MPa |
| Pouring temperature | 1,570–1,620 °C |
| Initial carbon in steel | 0.36%–0.40% |
| Holding time after pouring | 15 min |
| Normalizing temperature | 880±10 °C, hold 2 h |
The application of the lost foam process to steel castings is not without challenges. One of the main concerns is the potential for surface carburization, which can affect the weldability and machinability. In our case, the carbon pickup was within acceptable limits, and no excessive carburization was observed on the surface. We also checked the hardness of the as-cast surface and found no significant increase compared to internal sections. The uniform distribution of carbon pickup across the casting positions suggests that the pyrolysis gases were well evacuated and did not cause localized enrichment. However, for thicker sections or heavier castings, the carbon pickup may be higher, and further optimization of the foam density and coating permeability may be required.
Another challenge is the deformation of thin-walled castings. The traction frame has a thin wall of 15 mm with reinforcing ribs. The integrated pattern assembly and the use of a continuous sprue helped maintain shape stability. We also designed the gating system with the runner inlet lower than the outlet to control the filling sequence and minimize turbulence. The castings were inspected for distortion after shakeout, and the measured flatness was within 1 mm, which is acceptable. The heat treatment was carried out in a stress-relieving fixture to avoid additional distortion. The successful production of this thin-walled steel casting demonstrates that the lost foam process can handle complex geometries with reasonable accuracy.
In terms of productivity, the lost foam process offers a significant advantage in that multiple patterns can be assembled into a cluster and poured simultaneously. We used one group with two layers, giving four castings per pour. The cluster approach reduces the number of pours and increases the throughput. The modular design of the sprue with circular recesses and protrusions allows easy stacking of multiple layers. We have already tested a three-layer cluster with six castings, and the results were equally good. The ability to produce multiple castings in a single pour significantly improves the efficiency of the lost foam process.
The environmental benefits of the lost foam process are also noteworthy. The unbonded sand is reused without the need for binders and additives, reducing the generation of waste sand. The dry sand reclamation system recovers about 95% of the sand, which is much higher than the green sand process. The absence of core binders eliminates the emission of hazardous gases from core making. The only emission is from the decomposition of the foam pattern, which is controlled by the vacuum system and can be treated with an afterburner if necessary. This makes the lost foam process an environmentally friendly alternative to conventional casting processes.
To sum up, our work has established a robust process for producing lost foam castings of steel tractor parts. The process is technically feasible, economically beneficial, and environmentally sustainable. We have achieved a carbon pickup range of 0.04%–0.09% and developed a method to compensate for it by adjusting the initial carbon content. The resulting castings meet all the requirements of ZG310-570, including composition, mechanical properties, dimensional accuracy, and surface quality. The cost saving of 856 yuan per ton is a decisive factor for the foundry to adopt this process for mass production. We are confident that the lost foam process will play an increasingly important role in the production of steel castings, and we encourage other foundries to explore its potential for their products. Our future development will focus on expanding the product range, optimizing the pattern and gating design, and implementing real-time process monitoring to ensure consistent quality in large-scale production.
Finally, we would like to emphasize that the lost foam process is not just a casting method; it is a complete manufacturing system that requires close coordination between pattern making, coating, molding, melting, pouring, and heat treatment. The experience gained from the tractor traction frame has provided us with valuable knowledge that can be transferred to other steel castings. By sharing our findings, we hope to contribute to the wider adoption of lost foam castings in the steel foundry industry.