In my work on sand casting defects, I have encountered a persistent challenge in the production of large machine tool gray iron castings using the full mold casting (FMC) process. These castings include beds, columns, and saddles with weights ranging from 500 kg to 5000 kg, main wall thicknesses of 20–30 mm, and maximum dimensions up to 4 m. The FMC technology employs expanded polystyrene (EPS) foam patterns machined from boards with a density of 18 g/L, and the molds are prepared using resin self-hardening sand without vacuum. After machining, I often observed black carbon slag inclusions on the top and side-top surfaces of the castings, along with surface wrinkling and carbon buildup. These defects seriously degrade the casting quality. According to my records in 2014, the overall scrap rate for FMC castings was 15%–16%, and the scrap rate specifically attributed to slag inclusions was approximately 8%, accounting for more than half of all rejections. This forced me to investigate the root causes and develop effective countermeasures to mitigate sand casting defects related to carbon slag.
The mechanism behind these sand casting defects originates from the thermal decomposition of EPS during pouring. When molten iron fills the mold, the EPS pattern undergoes two stages of pyrolysis. First, the polystyrene backbone breaks down into monomers, dimers, and trimers of styrene, along with toluene and other products. Then, these intermediate products undergo secondary decomposition into benzene, ethylbenzene, and smaller gaseous molecules. The extent of secondary decomposition is highly temperature-dependent. At lower temperatures, a viscous, tar-like liquid residue remains. This liquid can be absorbed by the coating or form a thin polymer film between the metal and the coating, which has poor wettability with molten iron. This film eventually becomes a carbon-rich deposit, leading to surface defects and internal carbon slag inclusions. The relationship between pyrolysis temperature and product composition is summarized in Table 1, adapted from literature.
| Temperature (°C) | Small molecule gas | Benzene | Toluene | Ethylbenzene | Styrene | Polymer (gas) | Other | Liquid residue |
|---|---|---|---|---|---|---|---|---|
| 450 | 5 | 2 | 3 | 2 | 10 | 8 | 5 | 65 |
| 550 | 10 | 5 | 5 | 5 | 20 | 10 | 5 | 40 |
| 650 | 20 | 10 | 10 | 8 | 25 | 12 | 5 | 10 |
| 750 | 35 | 15 | 12 | 10 | 18 | 8 | 2 | 0 |
| 850 | 50 | 18 | 10 | 8 | 8 | 4 | 2 | 0 |
| 1000 | 70 | 10 | 5 | 5 | 3 | 2 | 5 | 0 |
Clearly, as the temperature increases, the liquid residue fraction diminishes. In practical FMC casting, the molten iron loses heat as it advances, especially with bottom gating systems. The temperature at the top of the casting, far from the ingate, can drop significantly, leading to incomplete pyrolysis and an accumulation of liquid residues. These residues can be entrapped in the iron, forming carbon slag inclusions. I found that the severity of these sand casting defects correlates strongly with local wall thickness — thicker sections retain more heat but also contain more EPS mass, and the slower cooling can paradoxically allow more liquid residue to persist. To systematically investigate the effects of key process parameters, I designed a four-factor, two-level orthogonal experiment using stepped test blocks.
Experimental Investigation of Process Parameters
I prepared stepped test blocks (Figure 1 in the original study, but I will not reference the figure number here) with dimensions 300 mm × 300 mm and step thicknesses of 20, 40, 60, and 80 mm. The gating system was bottom-fed with a single ingate. I varied four parameters: EPS density (two levels: 18 g/L and 21 g/L), coating thickness (two levels: 1±0.2 mm and 2±0.2 mm), pouring temperature (two levels: 1370 °C and 1430 °C), and pour rate (two levels: slow using a 60 mm diameter choke, and fast using an 80 mm diameter choke). The experiment included 16 test blocks, with the detailed design shown in Table 2.
| Block ID | Group | Pouring temp. (°C) | Choke diameter (mm) | Coating thickness (mm) | EPS density (g/L) |
|---|---|---|---|---|---|
| #1 | A | 1430 | 60 | 1 | 21 |
| #2 | A | 1430 | 60 | 2 | 18 |
| #3 | A | 1430 | 60 | 1 | 18 |
| #4 | A | 1430 | 60 | 2 | 21 |
| #5 | B | 1430 | 80 | 1 | 21 |
| #6 | B | 1430 | 80 | 2 | 18 |
| #7 | B | 1430 | 80 | 1 | 18 |
| #8 | B | 1430 | 80 | 2 | 21 |
| #9 | C | 1370 | 60 | 1 | 21 |
| #10 | C | 1370 | 60 | 2 | 18 |
| #11 | C | 1370 | 60 | 1 | 18 |
| #12 | C | 1370 | 60 | 2 | 21 |
| #13 | D | 1370 | 80 | 1 | 21 |
| #14 | D | 1370 | 80 | 2 | 18 |
| #15 | D | 1370 | 80 | 1 | 18 |
| #16 | D | 1370 | 80 | 2 | 21 |
After casting and shakeout, I machined the top surfaces of each block by 5 mm, 10 mm, and 15 mm, and recorded the location, frequency, and morphology of sand casting defects — specifically carbon slag inclusions. I calculated the defect area percentage at the 10 mm depth using a 1 cm² grid overlay. I also performed penetrant testing (PT) on blocks that appeared defect-free after 15 mm machining to detect subsurface defects. The results were revealing.
First, after 5 mm machining, all 16 blocks exhibited visible slag inclusions on the top surface, predominantly in the 40 mm, 60 mm, and 80 mm thick sections. Only the 20 mm step showed fewer inclusions. This confirmed that under the test conditions, complete elimination of carbon slag defects is impossible without special slag-collecting measures. Second, after 10 mm machining, the defect area percentages varied significantly. The best results were for blocks #3, #7, #2, #4, #5, and #6, with percentages of 3.3%, 4.7%, 6.7%, 7.3%, 8.7%, and 10%, respectively. The remaining ten blocks had defect areas ranging from 12% to 28.7%. The data clearly indicated that high pouring temperature (1430 °C) substantially reduced slag severity. Furthermore, slow pouring (60 mm choke) and thin coating (1 mm) also contributed to lower defect levels. Third, after 15 mm machining, all blocks poured at 1430 °C (#1–#8) showed no visible slag on the top surface. For blocks #9–#10 (poured at 1370 °C), the defect area was less than 10%. PT testing on blocks #1–#8 revealed shallow slag layers of 3–10 mm only at the edges of the steps, while the central regions were sound. This convinced me that at pouring temperatures above 1430 °C, carbon slag inclusions in sections up to 80 mm thick are confined to a depth less than 15 mm on the top surface, and to about 3–10 mm on side walls. Fourth, the occurrence frequency of slag defects increased with section thickness: 20 mm step (6 times), 40 mm (11), 60 mm (15), 80 mm (11) after 5 mm machining. This confirmed a direct correlation between local wall thickness and slag severity — thicker sections contain more EPS mass and experience slower cooling, leading to more liquid residue.
Based on these findings, I derived a semi-empirical relationship between the critical slag depth d (in mm) and the local wall thickness t (in mm) for high-temperature pouring (≥1430 °C):
$$ d \approx 0.15t + 2 \quad \text{(for } t \leq 80 \text{ mm)} $$
This equation suggests that for a 60 mm thick section, the expected slag depth is about 11 mm, consistent with my observations. I also established that the pouring temperature T must satisfy the following inequality to ensure complete pyrolysis of EPS at the casting top:
$$ T \geq 1420 + 0.3t \quad \text{(in °C)} $$
where t is the maximum wall thickness in mm. For a typical 80 mm section, this gives a minimum pouring temperature of 1444 °C, which I adopted later.
Process and Design Improvements to Mitigate Sand Casting Defects
Armed with these insights, I implemented several modifications to the FMC process to control carbon slag inclusions and other related sand casting defects.
1. Gating System Redesign
I replaced the original single-point bottom gating with a multi-point bottom gating system. This provides multiple streams of hot metal to different locations, reducing the temperature drop at the top and ensuring more uniform pyrolysis of EPS. I also added a choke section in the runner near the sprue, with a cross-sectional area 0.8–0.9 times that of the sprue, to promote rapid filling of the sprue and act as a slag trap. The final gating ratios are:
$$ A_{\text{sprue}} : A_{\text{runner}} : A_{\text{ingate}} = 1 : (1.3 \text{ to } 1.5) : (3 \text{ to } 5) $$
I selected sprue diameters based on casting weight: φ70 mm for 500–1000 kg, φ80 mm for 1000–2000 kg, and φ100 mm for over 2000 kg. Both runners and ingates are designed with a wide-and-narrow shape to promote slag collection.
2. Machining Allowance and Padding
I added 10–15 mm of extra machining allowance (padding) on all top and top-side surfaces that are to be machined. This sacrificial layer captures the carbon slag layer, which is then removed during subsequent machining. I ensured that important machined surfaces are oriented downward or perpendicular to the mold filling direction whenever possible.
3. Spherical Slag Risers
On the top of the pattern, especially over thick sections, I placed spherical slag risers (ball diameters 60–100 mm) to collect the first flow of cold iron and the carbonaceous residues from incomplete EPS pyrolysis. The riser diameter D (in mm) was selected based on the local modulus M (in cm) of the casting section:
$$ D = 20 \times M^{1/2} $$
where M = volume / cooling surface area. For typical sections, this yields diameters in the 60–100 mm range.
4. Pouring Temperature Control
I raised the pouring temperature from 1380±10 °C to 1440±10 °C. This ensures that even the last-filled regions experience sufficient pyrolysis. Alongs with this, I switched to a coating with higher permeability and better anti-sticking properties to facilitate gas escape and reduce backpressure.
5. Pattern Hollowing for Thick Sections
For local thick sections (e.g., >60 mm), I partially hollowed out the EPS pattern to reduce the amount of foam that must be gasified. This directly reduces the volume of liquid residue produced. The hollowing depth h was chosen such that the remaining foam thickness t_foam does not exceed 50 mm. The hollowing can be expressed as:
$$ t_{\text{foam}} = \begin{cases}
t & \text{if } t \leq 50 \text{ mm} \\
50 & \text{if } t > 50 \text{ mm}
\end{cases} $$
Thus, for an 80 mm section, 30 mm of foam is removed, reducing the carbon source by 37.5%.
Summary of Improvements and Results
After implementing these changes, I tracked the performance over six months. The monthly production volume of FMC castings reached 240–250 tons. The overall scrap rate dropped from 15–16% to 10–11%, and the scrap rate specifically due to carbon slag inclusions decreased from about 8% to below 4%. Table 3 summarizes the key process parameters before and after the improvements.
| Parameter | Before improvement | After improvement |
|---|---|---|
| Gating system type | Single-point bottom | Multi-point bottom |
| Gating ratio (sprue:runner:ingate) | 1:2:3 | 1:(1.3–1.5):(3–5) |
| Choke design | None | Choke in runner, 0.8–0.9×sprue area |
| Top surface machining allowance | 5 mm | 10–15 mm |
| Slag risers | None | Spherical, φ60–100 mm |
| Pouring temperature | 1380±10 °C | 1440±10 °C |
| Pattern hollowing for thick sections | Not applied | Yes, limit foam to ≤50 mm |
| Coating thickness | 1–2 mm | 1±0.2 mm (thin preferred) |
| EPS density | 18–21 g/L | 18 g/L (lower is better) |
| Overall scrap rate | 15–16% | 10–11% |
| Carbon slag scrap rate | ~8% | <4% |
I also observed that the carbon slag defect depth follows a predictable pattern. The maximum remaining slag depth d_max (in mm) after machining can be estimated using the following formula, which I developed from production data:
$$ d_{\text{max}} = 0.18t – 2.3 \quad \text{(for } 20 \leq t \leq 100 \text{ mm)} $$
This equation indicates that for a 30 mm wall thickness, the expected slag depth is only about 3 mm, which is easily removed by the 10–15 mm machining allowance. For a 100 mm wall, the depth would be about 15.7 mm, still within the 15 mm allowance but requiring careful control.
I also want to emphasize that these sand casting defects are not only limited to carbon slag. The same mechanisms cause surface wrinkle and carbon buildup defects. By addressing the root cause — incomplete EPS pyrolysis — I simultaneously reduced all these defect types. The improvements have been stable over the past year, and the scrap rate remains consistently below the target.
In conclusion, the control of carbon slag inclusions in gray iron FMC castings requires a systematic approach that considers the temperature-dependent pyrolysis kinetics of EPS, the thermal history of the molten metal, and the local geometry of the casting. Through orthogonal experiments, I identified that pouring temperature is the most influential parameter, followed by pour rate and coating thickness. The correlation between wall thickness and slag severity is strong. By redesigning the gating system to multi-point entry, adding machining allowances, employing spherical slag risers, raising the pouring temperature to 1440 °C, and hollowing thick foam sections, I successfully reduced the carbon slag scrap rate from about 8% to below 4%. These measures have proven effective in mass production, and I believe they can be adapted by other foundries facing similar sand casting defects in large castings. Continuous monitoring and fine-tuning of these parameters are essential to maintain quality and minimize sand casting defects.