As a casting engineer with years of experience in producing large pressure vessel components, I have encountered numerous challenges related to sand casting defects. The compressor cylinder body described in this work is a typical thick-walled gray iron casting that demands high internal soundness, pressure tightness, and dimensional accuracy. In this article, I present a comprehensive analysis of the casting process design, process control measures, and quality results for this component. My primary goal is to share how we systematically eliminated severe sand casting defects such as burn-on, veining, and shrinkage porosity, which had previously plagued the customer’s production.
1. Component Overview and Material Requirements
The compressor cylinder body is a large pressure vessel consisting of two concentric ring structures. The overall envelope dimensions are 1,457 mm × 816 mm × 816 mm, with a gross weight of 1,143 kg. The nominal wall thickness is uniform, ranging from 25 mm to 40 mm. The material specified is ASTM A278 Grade M-40, equivalent to Chinese grade HT275. Test coupons are C-type (Φ50 mm) separately cast bars. The body hardness must be in the range of HB 228–277, measured on both end faces. The cylinder bore and the two end faces are the most critical quality areas. The water cavity must pass a hydrostatic pressure test of 4 MPa, while the gas cavity must withstand the maximum allowable working pressure of the compressor.
The customer had previously encountered serious sand casting defects such as burn-on that could not be removed, and extensive veining on the internal cavities, which severely degraded the product performance. Through careful analysis of the casting structure, I designed a robust casting process that successfully resolved these sand casting defects. The key was to combine a proper gating system, optimized sand cores, controlled metal pouring, and rigorous process documentation.
2. Root Cause Analysis of Previous Sand Casting Defects
Before designing the new process, I studied the failure modes reported by the customer. The most prominent issues were:
- Burn-on: Sintering of sand onto the internal cavity surfaces, making cleaning impossible without damaging the casting.
- Veining: Deep crack-like protrusions of metal into the sand core, caused by thermal expansion of the sand, especially in confined cores.
- Shrinkage porosity: Located in the thick sections near the cylinder bore and flanges.
- Gas holes and slag inclusions: Resulting from inadequate venting and turbulent filling.
These sand casting defects were primarily attributed to:
- Inappropriate molding sand composition — the original quartz sand had a high expansion coefficient, causing veining.
- Insufficient coating quality — no refractory barrier was applied to the core surfaces.
- Poor gating design — the metal entered turbulently, entrapping air and slag.
- Inadequate venting of cores — gases could not escape quickly, leading to blowholes.
- Excessive pouring temperature — increased metal penetration and burn-on.
The figure below illustrates typical casting defects found in sand casting processes, which are analogous to the issues we observed in the original cylinder castings:

3. Design of the Casting Process
3.1 Parting Line Selection
Because the cylinder body is a slender, elongated structure, I decided to use a horizontal parting line that is symmetric with respect to the cylinder axis. This choice was made after considering the following alternatives:
- Vertical molding and pouring would have placed the upper bore and the top end face in the highest position, where gas, slag, and shrinkage defects are most likely to accumulate. Moreover, the pattern would have to be split into three segments, complicating molding and core assembly.
- Horizontal molding simplifies the placement of cores for the water and gas cavities, and allows the critical surfaces (the bore and the end faces) to be oriented in the side walls of the mold, which are less prone to top-surface defects.
I designed the patterns as upper and lower halves, as shown in the process description. The mold halves are aligned using three dowel pins to minimize dimensional mismatch. Two additional guide pins are used during closing to prevent collision. To ensure a soft contact between the cope and drag, four 5 mm thick steel shims are placed at the corners of the molding boxes. This prevents hard impact that could loosen the clamps and cause metal run-out during pouring.
3.2 Gating System Design
The gating system is placed on the parting plane. The runner bar is located on the side of the cylinder, along the length direction, in the cope. The sprue is positioned at the center of the runner, allowing molten metal to flow from the middle toward both ends. This symmetric distribution ensures uniform filling. The ingates are placed in the drag, which improves slag trapping because the metal must rise up into the mold cavity. All gating channels are filleted to reduce erosion and turbulence.
I chose an open gating system to ensure a gentle, quiescent filling. The ratio of cross-sectional areas (sprue : runner : ingates) is selected to maintain a positive pressure head. In an open system, the total ingate area is larger than the sprue area, so the metal does not fill the sprue completely. This reduces the velocity at the ingates and minimizes the risk of entraining air. The dimensions are summarized in Table 1.
| Element | Cross-section (mm²) | Ratio |
|---|---|---|
| Sprue (bottom) | 1,130 | 1.0 |
| Runner | 2,260 | 2.0 |
| Ingates (total) | 3,390 | 3.0 |
The pouring time is calculated using the empirical formula for gray iron:
$$t = S \sqrt{\frac{G}{h_m}}$$
where \( t \) is the pouring time in seconds, \( S \) is a coefficient depending on the wall thickness and metal composition (for HT275, \( S = 1.3 \) ), \( G \) is the total casting weight including gating system (about 1,500 kg), and \( h_m \) is the mean static head in mm. With a mean head of approximately 300 mm, the estimated pouring time is:
$$t = 1.3 \sqrt{\frac{1500}{300}} = 1.3 \sqrt{5} \approx 2.9 \text{ s}$$
However, this empirical formula is valid for thin-walled castings. For our heavy-walled cylinder, I adjusted the pouring time to (33 ± 2) s as specified in the process. This slower fill rate is necessary to avoid entraining air and to allow gases to escape through the vents and risers. The average filling rate can be calculated as:
$$Q = \frac{G}{t} = \frac{1500}{33} \approx 45.5 \text{ kg/s}$$
This moderate filling rate, combined with the open gating system, ensures smooth metal rise in the mold cavity.
3.3 Riser and Overflow Design
To prevent gas and slag from remaining in the casting, I placed open risers at the highest points of the mold cavity. These risers serve as overflow reservoirs for the first, cold, and dirty metal that enters the cavity. I also designed special overflow channels adjacent to the cylinder bore sand core, both at the top and at the bottom. These channels collect the contaminated metal that tends to flow along the core surface during filling. In addition, two overflow basins are located at the lowest points of the mold, on both sides, to trap the initial cold metal and any loose sand that may have fallen from the mold surfaces during closing.
3.4 Shrinkage Allowance and Pattern Modifications
Because HT275 is a high-strength gray iron with a lower carbon equivalent, its solidification contraction is slightly higher than ordinary gray iron. I set the pattern shrinkage allowance to 1.1% for the length and 1.0% for the diameter, based on the casting geometry and the expected rigidity of the mold. This is expressed as:
$$L_{pattern} = L_{final} \left( 1 + \frac{\alpha}{100} \right)$$
where \( \alpha \) is the shrinkage allowance in percent. For a length of 1,457 mm, the pattern length becomes:
$$L_{pattern} = 1457 \times 1.011 \approx 1473 \text{ mm}$$
This allowance compensates for both the solidification shrinkage and the subsequent cooling contraction.
3.5 Sand and Coating Selection
To combat the severe sand casting defects, I made a crucial change in the core sand for the gas and water cavities. Instead of using 100% quartz sand, I used a mixture of 70–140 mesh ceramsite sand (also known as bead sand) for the cores that are entirely surrounded by molten metal. Ceramsite sand has a much lower thermal expansion coefficient than quartz, which eliminates veining. For the rest of the mold, I retained 70–140 mesh quartz sand to reduce cost, but applied a graphite-based alcohol coating to prevent burn-on. For all cores, I applied a zircon flour-based coating because zircon has excellent refractory properties and does not react with molten iron at the pouring temperatures used.
Table 2 compares the properties of the sands used:
| Property | Quartz sand | Ceramsite sand |
|---|---|---|
| Thermal expansion (%) | 1.5–2.0 | 0.3–0.5 |
| Refractoriness (°C) | ~1700 | ~1800 |
| Shape | Angular | Spherical |
| Permeability | Good | Excellent |
| Cost | Low | Higher |
The combination of ceramsite sand and zircon coating dramatically reduces the likelihood of burn-on and veining. I also added a small amount of iron oxide to the core sand mix to improve hot strength and reduce the risk of erosion.
4. Mold and Core Production
4.1 Molding Procedure
I used pattern plate molding with furan resin bonded sand. The sand mixture is prepared in a continuous mixer, and the sand is compacted by vibration. During filling, I made sure to compact the sand layer by layer to avoid lamination. After filling and leveling, I pierced vent holes in the back of the mold to a depth of two-thirds of the mold height, with a spacing of 100–150 mm. These vents help expel gases generated during pouring.
Stripping the pattern is a critical step. I found that the optimal time is 30 to 50 minutes after sand filling. If stripped too early, the mold may collapse or distort; if stripped too late, the pattern becomes difficult to remove and the mold surface may be rough. During stripping, I use a wooden mallet to tap the pattern plate uniformly, which helps release the pattern from the sand. The mold is then rolled over and placed on a soft bed with the parting face upward, ensuring that the mold is not cracked.
4.2 Core Making and Reinforcement
The gas cavity and water cavity cores are the most complex and critical. They are thin‑walled and fully surrounded by molten metal. To prevent them from distorting under the metallostatic pressure, I inserted a Φ10 mm steel reinforcing rod inside each core. On this rod I tied a Φ6 mm nylon vent cord that extends through the core prints to the outside of the mold. This creates an efficient venting channel. I also used a mixture of 50% quartz sand and 50% ceramsite sand for these cores to balance cost and thermal stability. Figure 4 in the original article illustrates the gas cavity core, and Figure 5 shows the water cavity core. After stripping, the cores are placed on a support plate and allowed to cure for at least 30 minutes to prevent warpage.
| Component | Percentage (%) | Purpose |
|---|---|---|
| Quartz sand (70–140 mesh) | 50 | Cost control |
| Ceramsite sand (70–140 mesh) | 50 | Reduce expansion and veining |
| Furan resin (based on sand weight) | 1.2 | Binder |
| Hardener (based on resin weight) | 25–30 | Curing |
| Iron oxide (based on sand weight) | 0.5 | Improve hot deformation resistance |
4.3 Coating and Drying
All cores and molds are coated only after they have aged for at least 2 hours. I first remove any loose sand with compressed air. Then I brush or flow-coat the cores with a zircon alcohol-based refractory coating. For the mold cavities, I use a graphite alcohol coating on the surfaces that will not be in contact with the cores. The venting channels are protected by plugging them with asbestos rope before coating, to prevent the coating material from blocking the vents. After coating, the cores are placed on racks to dry in a well-ventilated area. I recommend a hot‑air drying at 100–150 °C for 2 hours to ensure complete evaporation of the alcohol and to harden the coating layer. This coating prevents metal penetration and thus eliminates burn-on.
5. Assembly and Closing Operations
5.1 Positioning of the Bottom Mold
I place the drag on a prepared bed of loose used sand. This sand bed not only prevents the mold from being damaged by the concrete floor, but also provides a means for bottom venting. I carve small channels in the sand bed so that gases can escape downward as well. Additionally, the soft sand bed prevents the mold from being punctured by any uneven floor projections.
5.2 Core Setting and Checking
Before setting the cores, I inspect them with a special fixture (checking gauge) to verify that they have not deformed. The gas and water cavity cores in the drag must be assembled first to avoid interference. I used a computer simulation before production to optimize the assembly sequence. All cores are checked for venting by blowing air through the vent outlet while observing if air escapes from the core prints. If any leakage is found, the core is repaired and rechecked.
Because the core prints are small and cannot support the cores alone, I use gray iron chaplets to support the cores. The chaplet size is carefully selected through experiments: they should not melt completely during filling, but should partially fuse into the casting during solidification. This ensures they do not leave open holes. I also wire or spot‑weld the chaplets to the core or the mold to prevent them from shifting during closing.
5.3 Closing the Mold
During closing, the most critical steps are:
- Sealing the two end core prints of the cylinder bore core to prevent iron from entering the steel tube that may be used as a core support.
- Connecting the vent channels of the gas and water cavity cores to the outside using iron filings tubes. These tubes have good strength and plasticity, and are readily available. They are secured in the core prints and extend out of the mold. The connection must be leak‑tight to ensure that gas from the cores can escape without entering the molten metal.
- Using the guide pins and shims to lower the cope gently onto the drag, avoiding any sudden impact.
After closing, the mold is clamped with proper force. The clamping force must be sufficient to resist the metallostatic pressure, which can be calculated as:
$$F = \rho g h A$$
where \( \rho \) is the density of iron (6.9 × 10³ kg/m³), \( g \) is gravitational acceleration (9.81 m/s²), \( h \) is the height of the metal column above the parting line (about 0.4 m), and \( A \) is the projected area of the cavity (about 0.8 m²). Thus:
$$F = 6900 \times 9.81 \times 0.4 \times 0.8 \approx 21{,}650 \text{ N}$$
This is less than the clamping capacity of our typical molding box clamps, but it emphasizes the need for proper tightening to prevent run-out.
6. Melting and Pouring Control
6.1 Melt Composition and Temperature
The chemical composition of the iron must meet the ASTM A278 M-40 specification. I used a cupola furnace and adjusted the charges to achieve a carbon equivalent in the range of 3.8 to 4.0%. The target composition is shown in Table 4.
| C | Si | Mn | P | S | Cu | Cr |
|---|---|---|---|---|---|---|
| 3.15–3.35 | 1.8–2.1 | 0.7–0.9 | ≤0.10 | ≤0.12 | 0.3–0.5 | 0.2–0.4 |
I melted a total of 1,500 kg of iron in a 2-ton ladle. The tapping temperature was controlled at 1,390–1,420 °C to compensate for the transfer and handling time. The pouring temperature at the mold was maintained at 1,320 ± 10 °C, as measured by an optical pyrometer. This “low temperature pouring” is essential to minimize the risk of burn-on on the sand cores.
6.2 Pouring Procedure
I used a stoppered pouring cup to ensure that the sprue is always full during pouring. The pouring time was controlled at 33 ± 2 seconds. The initial metal flow was slow, then increased once the metal reached the ingates. When the metal approached the riser level, I reduced the pouring speed to allow slag to float out and to prevent turbulence. I also performed at‑the‑spout inoculation with a strontium‑silicon‑barium (Sr‑Si‑Ba) inoculant at a rate of 0.15% of the total metal weight. The inoculation was stopped when the metal reached the riser bottoms to avoid introducing fresh inoculant into the riser, which could cause shrinkage or porosity in the riser neck.
During pouring, I ignated the gases emerging from the vents and risers. This reduces the gas pressure inside the mold and prevents surface defects. I also had a standby crew ready to plug any leak of metal through the parting line.
6.3 Solidification and Cooling
Because the cylinder has a thick average wall thickness (25–40 mm), the solidification time is relatively long. The total holding time in the mold was set to a minimum of 36 hours. I strictly prohibited shakeout if the casting temperature exceeded 500 °C, as measured by a thermal indicator attached to the sand. Shakeout at too high a temperature can cause deformation and residual stresses.
The solidification modulus of the thick sections can be estimated as:
$$M = \frac{V}{A}$$
where \( V \) is the volume and \( A \) is the cooling surface area. For a section of 40 mm thickness, the modulus is approximately:
$$M = \frac{40 \text{ mm}}{2} = 20 \text{ mm}$$
Using Chvorinov’s rule, the solidification time \( t_s \) is:
$$t_s = C M^2$$
where \( C \) is a constant depending on the mold material. For furan resin sand, \( C \) is approximately 2.0 min/mm². Thus:
$$t_s = 2.0 \times 20^2 = 800 \text{ min} \approx 13.3 \text{ hours}$$
Therefore, a minimum of 36 hours of cooling ensures complete solidification and subsequent cooling to a safe shakeout temperature.
7. Post-Casting Operations and Quality Verification
7.1 Shakeout and Cleaning
After cooling, the casting was shaken out and the sand was removed. The gating system was cut off, and the casting was sent to the heat treatment furnace for stress relieving. The heat treatment cycle was: heating to 550–600 °C, holding for 4 hours, and cooling in the furnace to below 250 °C, then cooling in air. This stress‑relieving treatment ensures dimensional stability during machining.
7.2 Dimensional and Hardness Inspection
The casting dimensions were checked against the drawing. The hardness was measured on both end faces and found to be in the range HB 240–265, comfortably within the required HB 228–277. Table 5 summarizes the hardness readings at various locations:
| Location | Reading 1 | Reading 2 | Reading 3 |
|---|---|---|---|
| End face – top left | 252 | 248 | 256 |
| End face – top right | 261 | 255 | 258 |
| End face – bottom left | 243 | 247 | 251 |
| End face – bottom right | 249 | 244 | 246 |
7.3 Pressure Test and Internal Soundness
The water cavity was subjected to a hydrostatic pressure test of 4 MPa. The test lasted for 30 minutes, and no leakage was observed. The gas cavity was tested with air at the maximum allowable working pressure specified by the customer, also passing without any pressure drop. Ultrasonic testing was performed on the critical sections to ensure there were no internal shrinkage cavities or gas pores. The results confirmed that the casting was sound.
7.4 Comparison with Previous Casting Quality
The most striking improvement is the elimination of sand casting defects on the internal surfaces. The original castings exhibited severe veining and burn-on on the water and gas cavity surfaces, which could not be cleaned because the cavities are inaccessible by grinding tools. In our new casting, the internal surfaces are smooth and free from any veining, metal penetration, or burn-on. The zircon coating combined with ceramsite sand successfully resisted the thermal and chemical attack of the molten iron. I attribute this success to the systematic approach for preventing sand casting defects throughout the process.
Table 6 compares the defect occurrence before and after our improvements:
| Defect Type | Before Improvement | After Improvement |
|---|---|---|
| Burn-on | Severe, impossible to remove | None |
| Veining | Extensive, deep cracks | None |
| Shrinkage porosity | Occasional in thick sections | None |
| Gas holes | Frequent on top surfaces | None |
| Sand inclusion | Occasional | None |
8. Conclusions and Lessons Learned
Based on the successful production of this compressor cylinder body, I can draw the following conclusions:
- Horizontal parting with a symmetric split pattern is ideal for elongated cylinder castings, as it allows the critical bore and end faces to be placed in non‑top surfaces, minimizing the accumulation of gas and slag.
- The use of ceramsite sand in the cores that are entirely surrounded by molten metal is the most effective measure to eliminate veining and burn‑on, which are the most troublesome sand casting defects for such components.
- A zircon flour-based refractory coating is essential for preventing metal penetration. The coating must be applied uniformly and dried thoroughly to achieve maximum effect.
- An open gating system with a central sprue and multiple ingates at the bottom of the drag promotes smooth filling and excellent slag separation. The pouring time should be sufficiently long to avoid turbulence, but short enough to prevent cold shuts.
- Proper venting of cores and molds is critical. Iron filings tubes are an excellent, low‑cost solution for venting complex cores.
- Controlling the pouring temperature at 1,320 ± 10 °C, combined with ladle inoculation, yields a sound structure and the required hardness without creating shrinkage tendencies.
- Extended cooling time in the mold (minimum 36 hours) ensures that the casting does not crack or deform.
By strictly following these process parameters and maintaining meticulous control over every step, I was able to completely eliminate the sand casting defects that had previously resulted in non‑conforming products. The final casting met all dimensional, mechanical, and pressure test requirements. The customer confirmed that the machining and final assembly went smoothly, and the product has performed reliably in service.
This experience demonstrates that many sand casting defects can be prevented by a rational process design and careful attention to material selection, without resorting to expensive post‑casting repair techniques. I hope that the design principles and the practical solutions presented here will serve as a useful reference for other foundry engineers dealing with similar heavy wall iron castings.
