Our company is one of the major manufacturers of coal mine hydraulic support systems in China. Each year, we produce approximately 3500 to 4000 tonnes of cast components for hydraulic supports, including column sockets, connectors, guide rails, and racks. Traditionally, these castings were produced using sand casting processes, which often resulted in rough surfaces, poor dimensional accuracy, and excessive machining allowances. To meet increasing demands for surface finish and dimensional precision, we began exploring the application of lost foam castings in the production of hydraulic support components. This paper describes our systematic trials on lost foam castings for steel column sockets and the results achieved.
Introduction to Lost Foam Castings
Lost foam castings, also known as evaporative pattern casting, is a process in which a foam pattern (made from EPS, STMMA, or EPMMA) that replicates the final casting geometry is coated with refractory slurry, dried, embedded in dry sand within a flask, compacted by vibration, and then poured with molten metal under vacuum. The foam pattern vaporizes upon contact with the high-temperature metal, leaving a cavity that exactly matches the pattern. This technique was first successfully tested by H. F. Shroyer in 1956, patented in 1958, and began industrial application in 1961. In recent decades, lost foam castings have matured significantly, and many foundries worldwide have adopted the method. The share of lost foam castings in total casting output continues to grow.
Compared with conventional sand casting, lost foam castings offer numerous advantages: lower production cost, higher quality, design flexibility, and minimal sensitivity to casting complexity. Each pattern is used only once, eliminating parting lines, cores, flash, and draft angles. The process can be automated for mass production. Dry sand molding requires no binders, thus avoiding defects caused by moisture in the mold. Gating and riser systems are easy to remove, and cleaning effort is greatly reduced, improving labor conditions and productivity. Dimensional accuracy of lost foam castings can reach CT5–CT7, and surface roughness can be as low as 6.3–12.5 µm. The machining allowance can be controlled within 1.5–2 mm, significantly improving subsequent machining efficiency. Lost foam castings are environmentally friendly: they produce less dust, waste sand, and emissions. The dry sand reuse rate exceeds 95%, and the generation of CO and noise is substantially reduced. This process eliminates the occupational health hazards associated with silica dust. Many foundry experts call lost foam castings “the casting technology of the 21st century” and “the green revolution of the foundry industry.”
Technical Process Preparation
The column socket is a critical load‑bearing component of hydraulic supports, typically made of ZG27SiMn or ZG25MnTiB steel. It requires heat treatment (quenching and tempering) and machining before service. Areas that are difficult to machine must be cast to near‑net shape. To improve surface quality and dimensional accuracy, we conducted lost foam castings experiments. Prior to the trials, extensive research was performed, and a detailed experimental plan was developed.
Selection of Experimental Lost Foam Castings Component
We selected the column socket as the test piece, material ZG27SiMn, with high requirements for surface finish and dimensional precision. The casting has complex features, including a spherical socket and ears, which are difficult to machine.
Equipment Preparation
| Equipment | Specification |
|---|---|
| Vacuum stabilization system (energy‑saving) | 2BE–303, surge tank Φ1.5 m × 3 m, suction rate 45 m³/min, vacuum range 30–100 kPa |
| Variable‑frequency air‑suspension vibrating table | TQSW, PLC frequency controller |
| Vacuum sand flasks (4 units) | 1.5 m × 1.2 m × 1.2 m, welded from 6 mm steel plate, reinforced with channels. Bottom and sides feature four‑corner and perimeter vacuum chambers separated by 200‑mesh stainless steel mesh to prevent sand ingestion. |
| Roller coating mill | Drum type for grinding refractory coatings |
| Foam cutting tables | One horizontal, one vertical, with adjustable resistance wires |
| Two voltage regulators | For hot‑wire cutting power control |
| Drying room | Constant temperature 45 °C |
Raw Materials
- EPS foam boards (expandable polystyrene)
- Ceramic sand (20–40 mesh spherical particles)
- Refractory coating (self‑formulated from purchased raw materials: quartz powder, BY binder, white latex, water, CMC, sodium bentonite, sodium carbonate)
- Foam adhesive (cold glue)
- Resistance wires for cutting
Process Preparation Requirements
We analyzed the casting structure, designed the gating/riser system, and established strict process operation methods and technical measures to meet lost foam castings requirements. The goal was to ensure dimensions conformed to drawing specifications and to achieve a smooth surface finish.
Experimental Procedure
Foam Pattern Fabrication
The quality of the foam pattern directly affects the gasification rate and pyrolysis products during pouring. High‑quality foam has low carbon content, high molecular weight, and low density while maintaining adequate strength. Since the column socket is produced in relatively small batches, we chose manual pattern making instead of mold‑based foam shaping. The pattern was cut into several blocks and assembled using cold glue. For complex surfaces like the spherical socket, we designed a dedicated cutting fixture. A hot resistance wire centered at the socket’s pivot point was rotated without stopping to form the spherical curvature. The pattern shrinkage allowance was set to 2.5%. Machining allowance on the riser face for inspection was 2 mm, and on other inspection surfaces it was 1 mm. Cardboard templates were used as cutting guides. For vertical cutting of arcs, the foam height did not exceed 300 mm to avoid wire deflection. After cutting, surfaces were sanded and assembled. Cold glue was applied sparingly to minimize gas generation during pouring.
Coating Preparation, Application, and Drying
| Component | Function | Remarks |
|---|---|---|
| Quartz powder (refractory aggregate) | Provides high‑temperature resistance and coating strength | — |
| BY binder + white latex (organic binders) | Ensures good coating adhesion, permeability, and strength after drying | Mixed before milling |
| Water (solvent) | — | — |
| Carboxymethyl cellulose (FM6 grade) | Thickener and suspending agent | Pre‑soaked for 4 hours to form a paste |
| Sodium bentonite + sodium carbonate | Suspension aid and binder modifier | Diluted with water before adding |
The coating was milled in a drum mill for 8–10 hours and then allowed to rest for 30 minutes. The final coating must exhibit good permeability, applicability, anti‑stick properties, and adequate mechanical strength. It should remain suspended without sedimentation, adhere firmly to the foam, and have leveling characteristics. After drying, the coating should not crack or rehydrate easily. The coating was applied using a combination of flow coating and brushing to achieve uniform thickness. Three coats were applied, with a total average thickness between 1.5 mm and 2.5 mm. Between coats, the patterns were dried at 45 °C for a total of 28 hours. To prevent deformation from the added weight, patterns were turned over after the second coat.
Molding and Embedding
We designed a sand flask measuring 1.5 m × 1.2 m × 1.2 m, capable of accommodating two castings per flask. The flask was fabricated from 6 mm steel plate with reinforcement at lifting points. The bottom and sides were reinforced with U‑channels. The flask had four corner vacuum chambers connected to peripheral chambers; two suction pipes were placed at the bottom on one side. A 200‑mesh stainless steel screen was placed between the guard plate and side plates to prevent sand from entering the vacuum system.
Determining the pouring position was the first critical step in lost foam castings molding. The pouring position during pouring must exactly match the orientation of the pattern in the flask. To ensure smooth metal filling and avoid mold collapse, we avoided placing large flat surfaces horizontally. Large planes were tilted 15° from vertical to facilitate sand filling. For areas where sand could not easily flow, manual assistance and vibration were applied.
We used 20–40 mesh spherical ceramic sand because of its high permeability and excellent flowability, which is ideal for complex steel castings. The dry sand molding process is much simpler than traditional green sand molding. The sequence was: place bottom sand, level it, vibrate in three dimensions (x, y, and z). The vibrating table (TQSW) with PLC control provided frequency‑modulated vibration. The pattern was positioned, sand added, vibration applied, more sand added, vibration repeated — alternating directions (30 seconds forward, 30 seconds reverse) each cycle of 1 minute. This “frequent and short” vibration strategy ensured dense packing. After the flask was full, the top surface was covered with a plastic film, leaving the pouring cup and riser exposed. A thin layer of dry sand was then placed on the film to seal the system and maintain vacuum during pouring.
Vacuum System and Pouring
Control of the vacuum inside the flask is crucial for lost foam castings. We used the energy‑saving vacuum unit 2BE–303 with a large surge tank (1.5 m diameter, 3 m length). The high pumping capacity (45 m³/min) and large tank volume helped stabilize negative pressure and prevented sand collapse during pouring. The pouring system was a “sprue‑riser combined” open‑top solution. Since the steel was melted in an electric arc furnace and poured via a bottom‑pouring ladle, we designed a two‑chamber pouring cup with a larger capacity to reduce metal pressure and avoid splashing. The cup was made of a water‑glass‑bonded silica sand and was sealed with refractory clay at the junction to the pattern sprue. During pouring, the cup was kept full of metal to prevent air aspiration. The pouring temperature was controlled between 1590 °C and 1610 °C. The initial vacuum was set at 0.07–0.075 MPa, and after stable pouring it dropped to 0.06–0.065 MPa. No mold collapse occurred, and the castings exhibited excellent filling.
Experimental Results
Surface Quality of Lost Foam Castings
The as‑cast surface appearance was good. After shakeout, the coating layer peeled off in large flakes. Only a small amount of coating remained attached to irregular surfaces and required manual cleaning. Local sand adhesion occurred between the ears of the column socket. After riser removal, the riser on the pouring side (combined sprue‑riser) showed no shrinkage cavity at the root, while the riser on the opposite side had a small concentrated shrinkage cavity. Flame cutting revealed a cavity depth of 5–10 mm. After shot blasting, the castings were clean, with no deformation or mold expansion defects.
Dimensional Accuracy and Mass Comparison
| Drawing Dimension | Lost Foam Casting | Sand Casting |
|---|---|---|
| 490 (riser face) | 492 | 501 |
| 420 | 419 | 430 |
| 160 | 161 | 161 |
| 54 | 54 | 57 |
| 53 (ear thickness) | 53 | 51/54 (with draft) |
| 35 (side plate) | 36 | (501–428)/2 |
| 110 (riser face) | 110 | 115 |
| 246 (inspection) | 247 | 251 |
| 490 – 2×35 = 420 | 420 | 428 |
| Process | Gross weight (kg) | Finished weight (kg) | Yield (%) |
|---|---|---|---|
| Lost foam castings | 224 | 305 | 73.4 |
| Sand castings | 238 | 330 | 72.1 |
The dimensional accuracy of lost foam castings was clearly superior. For example, the ear thickness of 53 mm was achieved without draft, whereas sand castings required a taper. The machining allowance was significantly reduced. The finished weight of the lost foam casting was lower because less metal was needed for the rough casting, even though the net casting weight was higher due to less machining allowance.
Defect Analysis and Countermeasures
Shrinkage cavity and porosity under the riser on the non‑pouring side:
Root cause: the riser was undersized; the open‑top pouring system caused faster cooling on that side, leading to incomplete feeding. Additionally, excessive foam adhesive generated gas that was trapped. Solution: increase riser size and reduce process yield slightly; apply insulating cover on the riser after pouring.
Slag inclusions in the spherical socket and sand adhesion between ears:
Root cause: the bottom‑pouring ladle created high metal velocity, dislodging coating from the ear area. The dislodged coating mixed with the melt and accumulated at the top of the spherical socket, forming slag. The pouring cup made of silica sand also contributed to sand erosion. Solution: use a tea‑pot ladle better suited for lost foam castings; improve coating formulation to enhance strength and permeability.
Local sand adhesion:
Root cause: coating layer at the ears was damaged during handling or vibration. Solution: apply thicker coating in vulnerable areas and ensure full drying.
Conclusions
The experiments demonstrated that lost foam castings can produce steel column sockets for hydraulic supports with complete filling, smooth surfaces, and high dimensional accuracy. Compared with sand castings, the lost foam castings exhibited superior surface finish, no draft, simpler gating systems, reduced machining allowances, and near‑net shape capabilities. The manual pattern‑making method (cutting and assembling foam blocks) is suitable for low‑volume trials but results in lower productivity and human‑induced dimensional variations. For production, foam patterns should be produced using either die‑foam molding or CNC machining to ensure consistency and efficiency.
The following mathematical relationships were derived or utilized during the trials:
$$ \text{Pattern shrinkage allowance} = 2.5\% $$
$$ \text{Machining allowance on riser face} = 2\ \text{mm} $$
$$ \text{Machining allowance on inspection faces} = 1\ \text{mm} $$
$$ \text{Coating thickness per layer} = \frac{1.5\ \text{mm} \text{ to } 2.5\ \text{mm}}{3\ \text{coats}} $$
$$ \text{Vacuum range during pouring} = 0.06\ \text{MPa} \text{ to } 0.075\ \text{MPa} $$
$$ \text{Pouring temperature range} = 1590^\circ\text{C} \text{ to } 1610^\circ\text{C} $$
$$ \text{Process yield} = \frac{\text{gross weight}}{\text{finished weight}} \times 100\% \quad (\text{lost foam: } 73.4\%,\ \text{sand: } 72.1\%) $$
$$ \text{Shrinkage cavity depth observed} = 5\ \text{mm} \text{ to } 10\ \text{mm} $$
We also considered the relationship between vacuum and sand bed resistance. Assuming Darcy flow in the packed sand, the pressure drop across the mold can be expressed as:
$$ \Delta P = \frac{\mu Q L}{k A} $$
where \( \mu \) is gas viscosity, \( Q \) is gas flow rate, \( L \) is mold thickness, \( k \) is sand permeability, and \( A \) is cross‑sectional area. Maintaining a stable vacuum within 0.06–0.075 MPa ensured adequate flow of decomposition gases from the foam pattern while preventing mold collapse.
In summary, lost foam castings have proven to be a viable and advantageous process for hydraulic support components. Further work should focus on optimizing the coating system, automating pattern production, and fine‑tuning pouring parameters to eliminate remaining defects. With continued development, lost foam castings will play an increasingly important role in our foundry operations.