In my work on subway axle box bodies, I treat every casting defect as a signal that the process is not fully aligned with the solidification behavior of the part. The axle box body is a safety-relevant component in the bogie running gear. It has a complex, nearly symmetric geometry, a relatively thin main wall, and multiple rib intersections that behave as local hot spots. The material is ZG230-450 cast steel. The net casting mass is 45 kg, the maximum envelope is approximately 700 mm × 350 mm × 170 mm, and the main wall thickness is 16 mm. Because the part is a thin-wall, multi-hot-spot box-type casting, the risk of casting defect formation is high unless the gating system, risers, padding, chills, and molding practice are designed as one integrated system.
I approach the problem from the perspective of defect prevention rather than defect repair. Once a casting defect such as gas porosity, shrinkage porosity, sand inclusion, or a cold shut is exposed after machining, repair is often impossible or uneconomical. Therefore, I focus on the thermal and fluid conditions that create each casting defect. The goal is to produce a dense, pressure-tight, and dimensionally stable axle box body with a low scrap rate and a stable production flow.
Keywords: subway axle box, casting defect, casting process, gating system, riser, shrinkage porosity, gas porosity, solidification, feeding distance, yield.
1. Product Structure and Technical Requirements
The axle box body is almost symmetric about its central plane. It contains a central shaft cylinder, two side spring seats, and a network of ribs that connect the load-bearing regions. The ribs intersect each other and meet the cylinder and spring seats at relatively sharp thermal transitions. These intersections are not simple geometric features; they are thermal masses that cool more slowly than the surrounding 16 mm walls. In my experience, every such intersection is a potential site for a casting defect, especially shrinkage porosity and gas porosity.
The material ZG230-450 is a general engineering cast steel with moderate strength and good weldability. It is suitable for the axle box body, but it also has a relatively wide solidification interval compared with some low-alloy steels. That means the alloy can remain mushy over a significant temperature range, making it sensitive to feeding interruptions. When the local modulus is high and the feeding path is short, the last liquid to freeze can form shrinkage porosity. When the gating system entrains air, the same mushy zone can trap gas bubbles and produce gas porosity.
The technical requirements for the casting are strict. The production and inspection follow the relevant technical requirements and inspection standard for cast steel parts used in locomotives and vehicles. Non-machined surfaces receive primer and topcoat according to the drawing and technical specification. Casting fillets are generally R3 to R5. The lower rib surface carries the manufacturer mark, manufacturing year and month, and serial number. The internal soundness must be acceptable after machining, especially in the shaft cylinder, spring seat, and rib intersection regions.
| Item | Requirement or Value | Process Implication |
|---|---|---|
| Material | ZG230-450 cast steel | Moderate strength, wide solidification interval, needs robust feeding |
| Casting mass | 45 kg net | Riser and gating mass must be controlled for yield |
| Maximum envelope | 700 mm × 350 mm × 170 mm | Requires symmetric mold layout and stable sand support |
| Main wall thickness | 16 mm | Fast cooling; gas and shrinkage defect risk increases |
| Fillet radii | R3-R5 | Reduces stress concentration but can create local hot spots |
| Internal quality | Dense structure, no unacceptable gas porosity or shrinkage porosity | Requires sequential solidification and effective feeding |
| Surface quality | Primer and topcoat on non-machined surfaces | Surface casting defect must be controlled before painting |
| Marking | Manufacturer, date, serial number on lower rib | Lettering areas need chilling or special sand to avoid burn-on and gas defect |
I use a simple geometric parameter to compare the thermal behavior of different regions. For a plate-like section, the thermal modulus is:
$$M = \frac{V}{A}$$
where \(M\) is the modulus, \(V\) is the volume, and \(A\) is the heat-transfer surface area. For a wide plate of thickness \(t\), the modulus simplifies to:
$$M = \frac{t}{2}$$
With a main wall thickness of 16 mm, the nominal wall modulus is:
$$M_{\mathrm{wall}} = \frac{16}{2} = 8 \ \mathrm{mm}$$
At a rib intersection, however, the effective modulus can be much larger than 8 mm. That difference is the root of many casting defect problems. The wall solidifies quickly, while the intersection remains liquid longer. If the intersection cannot be fed through the wall or by a riser, shrinkage porosity forms.
2. Casting Defect Mechanisms in the Axle Box Body
I classify the relevant casting defect mechanisms into three groups: gas-related defects, shrinkage-related defects, and sand-related defects. Each group has a different cause, but in a thin-wall casting they interact. For example, a gating system that entrains gas can create gas porosity, and the same gating system can produce a low temperature gradient that worsens shrinkage porosity. A sand system that generates gas can create blowholes, and a mold wash or core that spalls can create sand inclusion. Therefore, I do not treat any casting defect in isolation.
The following image shows a representative casting defect pattern that I use as a visual reference when comparing macro- and micro-scale defects in cast steel components.
2.1 Gas Porosity and Entrained Gas
Gas porosity in the shaft cylinder inner surface is one of the most serious casting defect types in this part. After machining, the inner surface can show pores with diameters of about 3 mm to 5 mm. They are often diffuse, spherical or elliptical, with smooth and dark internal walls. They tend to appear in the upper part of the casting, and the pores near the top are usually larger than those lower down. The location and morphology indicate entrained gas rather than simple hydrogen gas porosity. The gas is mechanically entrained by the falling stream and cannot escape before the thin wall solidifies.
The original gating design used a ladle nozzle diameter of 50 mm and a sprue diameter of 60 mm. From a fluid continuity standpoint, the sprue was larger than the inlet stream. The actual flow area at the ladle nozzle was therefore the limiting area. The sprue did not run full. A partially filled sprue is a classic source of air entrainment and casting defect formation. The stream breaks, folds, and carries air into the mold cavity.
The area ratio can be estimated as follows:
$$A_{\mathrm{ladle}} = \frac{\pi d_{\mathrm{ladle}}^2}{4}$$
$$A_{\mathrm{sprue}} = \frac{\pi d_{\mathrm{sprue}}^2}{4}$$
For the original design, \(d_{\mathrm{ladle}} = 50 \ \mathrm{mm}\) and \(d_{\mathrm{sprue}} = 60 \ \mathrm{mm}\). Thus:
$$A_{\mathrm{ladle}} = \frac{\pi (50)^2}{4} = 1963.5 \ \mathrm{mm}^2$$
$$A_{\mathrm{sprue}} = \frac{\pi (60)^2}{4} = 2827.4 \ \mathrm{mm}^2$$
$$\frac{A_{\mathrm{sprue}}}{A_{\mathrm{ladle}}} = \frac{2827.4}{1963.5} = 1.44$$
This ratio greater than 1 means the sprue can accept more metal than the ladle nozzle can supply. The result is a non-pressurized, partially filled sprue and a high probability of entrained gas. Because the wall is only 16 mm thick, the solidification time is short. The bubbles do not have time to float out. They remain as gas porosity casting defect after machining.
I also consider the Bernoulli-type velocity at the sprue inlet. For an ideal fluid, the velocity is approximately:
$$v = C_d \sqrt{2gh}$$
where \(C_d\) is the discharge coefficient, \(g\) is gravitational acceleration, and \(h\) is the metallostatic head. In a real system, the actual velocity is lower because of friction and because the stream is not full. A partially filled sprue increases turbulence and air entrainment. The higher the turbulence, the greater the chance of a gas-related casting defect.
2.2 Shrinkage Porosity and Feeding Interruption
Shrinkage porosity is the second major casting defect in the axle box body. After machining, it appears between the two original waist risers and at the fillet where the curved padding intersects the cylinder wall. The most severe areas are those not covered by the risers. This is a feeding problem. The last liquid to freeze is isolated from the riser, and the natural shrinkage of the steel cannot be compensated.
Chvorinov’s rule gives a first-order estimate of solidification time:
$$t_s = B \left(\frac{V}{A}\right)^n = B M^n$$
For many castings, \(n\) is approximately 2, so:
$$t_s = B M^2$$
If a riser must feed a casting section, the riser modulus \(M_r\) must be larger than the casting modulus \(M_c\). A common design rule is:
$$M_r = 1.2 M_c$$
For a 16 mm wall, \(M_c = 8 \ \mathrm{mm}\), so the required riser modulus is:
$$M_r = 1.2 \times 8 = 9.6 \ \mathrm{mm}$$
The solidification time ratio is then:
$$\frac{t_r}{t_c} = \left(\frac{M_r}{M_c}\right)^2 = (1.2)^2 = 1.44$$
The riser must remain liquid 44% longer than the casting section. If the padding does not provide a sufficiently large and continuous feeding channel, the riser cannot compensate the liquid shrinkage. The original padding was too thin and too shallow. Its thickness was about 15 mm to 20 mm, and its depth was about 80 mm. This was not enough to maintain a hot feeding path between the riser and the cylinder wall. The result was shrinkage porosity casting defect between the risers.
The feeding distance can be approximated as:
$$L_f = k M_r$$
where \(L_f\) is the effective feeding distance and \(k\) is a material and geometry coefficient. For cast steel in this type of section, \(k\) can be on the order of 10 to 15. If \(k = 12\) and the effective riser modulus is about 10 mm, then:
$$L_f = 12 \times 10 = 120 \ \mathrm{mm}$$
This simple estimate explains why a padding depth of 80 mm was insufficient and why a depth of 120 mm was more appropriate. The original design did not maintain the feeding distance required by the thermal modulus of the shaft cylinder and rib intersection region.
I also use the Niyama criterion as a qualitative indicator of shrinkage porosity:
$$N_y = \frac{G}{\sqrt{\dot{T}}}$$
where \(G\) is the temperature gradient and \(\dot{T}\) is the cooling rate. A low \(N_y\) value indicates a high risk of shrinkage porosity. The original process had low gradients between the two waist risers and a high cooling rate in the thin walls. That combination is unfavorable. The optimized process increases the thermal gradient toward the risers and reduces the cooling rate in the feeding zone, raising \(N_y\) above the critical level for this material and section thickness.
2.3 Sand Inclusion and Mold-Related Casting Defect
Although the main reported defects were gas porosity and shrinkage porosity, sand inclusion and sand burn-on are always possible in a sodium silicate sand system. The original process used CO2-hardened sodium silicate sand for molding and core making. Chromite sand was used at fillets and lettering positions where chills or rivets could not be placed. This is a reasonable approach, but sodium silicate sand is sensitive to moisture, hardening uniformity, and mold coating practice. Poorly hardened sand can spall, and loose sand can enter the metal stream. The result is a sand inclusion casting defect.
I therefore treat sand control as part of casting defect prevention. The mold and core must be uniformly hardened, properly vented, and coated where necessary. The gating system should not impinge directly on fragile core surfaces. The ingate location and flow direction must be chosen so that the metal stream does not erode the mold or core. In the optimized design, I moved the ingates to a more favorable location and increased the ingate thickness to reduce jet velocity. This reduces the risk of sand inclusion and gas porosity at the same time.
3. Original Casting Process and Its Defect Outcomes
The original process used a one-box, two-piece layout with symmetric placement. The gating system was a center-pour system. The ladle nozzle diameter was 50 mm, the sprue diameter was 60 mm, the runner was a trapezoid of 65 mm/60 mm × 30 mm, and the ingate was a trapezoid of 45 mm/40 mm, with height 30 mm transitioning to 10 mm at the casting. The metal entered at the axle box rib position. The intent was to reduce direct erosion of the sand mold.
Risers were placed at the central shaft cylinder and at the tops of the two side spring seats. The risers were insulating exothermic waist-shaped risers. Inside the central shaft cylinder, a shaped padding matching the waist-riser curvature was added. The padding depth was selected from handbook recommendations. Chills and rivets made from carbon steel plate were placed at small hot spots to work together with the risers. Chromite sand was used at fillets and lettering positions. The top surface had 8 mm machining allowance, and other machined positions had 5 mm to 6 mm allowance.
| Original Process Parameter | Value or Description | Observed Effect |
|---|---|---|
| Mold layout | 1 box, 2 pieces, symmetric | Good productivity but limited riser space |
| Gating type | Center-pour | Simple but prone to turbulence if sprue is not full |
| Ladle nozzle | 50 mm | Limiting flow area |
| Sprue | 60 mm | Larger than nozzle, caused partial filling and air entrainment |
| Runner | Trapezoid 65/60 mm × 30 mm | Moderate flow capacity |
| Ingate | Trapezoid 45/40 mm, height 30 mm to 10 mm | Thin ingate increased jet velocity and mold erosion risk |
| Ingate location | Rib position | Reduced direct sand erosion but did not solve gas entrainment |
| Risers | 2 waist-shaped insulating exothermic risers | Insufficient coverage between risers |
| Padding | Shaped, thickness about 15-20 mm, depth about 80 mm | Too small for effective feeding distance |
| Chills | Carbon steel plate and rivets at small hot spots | Helpful locally but insufficient globally |
| Molding sand | CO2-hardened sodium silicate sand | Good strength but sensitive to moisture and hardening |
| Special sand | Chromite sand at fillets and lettering | Reduced local hot-spot risk |
| Machining allowance | Top 8 mm; other machined surfaces 5-6 mm | Acceptable for dimensional control |
The original process produced a significant number of castings with gas porosity in the shaft cylinder and shrinkage porosity between the risers. The gas porosity was a casting defect that appeared after machining, so it was not visible on the raw casting. The shrinkage porosity was also internal and was revealed only after machining or sectioning. These two casting defect types are especially costly because they are discovered late in the value stream.
| Casting Defect | Location | Typical Size | Morphology | Primary Cause |
|---|---|---|---|---|
| Gas porosity | Shaft cylinder inner surface | 3-5 mm | Spherical or elliptical, smooth dark walls, diffuse | Entrained gas from partially filled sprue and turbulent flow |
| Shrinkage porosity | Between risers and at padding-wall fillet | Variable, often internal | Spongy or dendritic internal voids | Insufficient padding thickness and depth; feeding distance too short |
| Sand inclusion | Mold-facing surfaces and core edges | Variable | Irregular dark inclusions | Sand spalling, poor hardening, or excessive metal jet velocity |
| Local hot-spot defect | Rib intersections and spring seat transitions | Variable | Shrinkage or porosity | High modulus and insufficient local chilling |
I summarize the root causes in a cause-and-effect matrix. This matrix helps me avoid the common mistake of treating each casting defect as a separate problem. In reality, the gating system, riser system, padding, chills, and sand system are coupled.
| Factor | Effect on Casting Defect | Evidence | Countermeasure |
|---|---|---|---|
| Sprue larger than ladle nozzle | Air entrainment, gas porosity | 60 mm sprue vs. 50 mm nozzle | Reduce sprue to 50 mm |
| Thin ingate | High jet velocity, mold erosion, sand inclusion | 10 mm ingate at casting | Increase ingate thickness to 25 mm |
| Two waist risers | Insufficient feeding between risers | Shrinkage between risers | Use three circular insulating exothermic risers |
| Shallow padding | Short feeding distance | 80 mm depth, 15-20 mm thickness | Use circumferential padding, 120 mm depth, 45 mm thickness |
| Insufficient local chilling | Local shrinkage porosity | Hot spots at intersections | Optimize chills, rivets, chromite sand |
| Sand moisture and hardening variation | Blowholes and sand inclusion | Sodium silicate sand sensitivity | Control CO2 hardening and mold coating |
4. Quantitative Process Analysis
Before changing the process, I performed a quantitative check of the gating and feeding design. The purpose was not to replace experience but to make the casting defect mechanisms visible in engineering terms. The two most important calculations were the gating area ratio and the feeding distance.
4.1 Gating Area Ratio and Flow Continuity
For a stable, non-turbulent fill, the gating system should be designed so that the sprue runs full and the gating ratio promotes smooth flow. The continuity equation is:
$$Q = A_s v_s = A_r v_r = A_i v_i$$
where \(Q\) is the volumetric flow rate, \(A_s\) is the sprue area, \(A_r\) is the runner area, \(A_i\) is the ingate area, and \(v_s\), \(v_r\), and \(v_i\) are the corresponding velocities. If the sprue is not full, the effective sprue area is less than the geometric area, and air is entrained. The original design had:
$$A_{\mathrm{sprue}} > A_{\mathrm{ladle}}$$
This condition is unacceptable for a quality casting. In the optimized design, I set:
$$A_{\mathrm{sprue}} \le A_{\mathrm{ladle}}$$
With a 50 mm ladle nozzle and a 50 mm sprue, the areas match:
$$A_{\mathrm{ladle}} = A_{\mathrm{sprue}} = \frac{\pi (50)^2}{4} = 1963.5 \ \mathrm{mm}^2$$
This change alone reduces the risk of entrained gas casting defect. I also enlarged the runner and ingate to reduce velocity and turbulence. A larger ingate thickness of 25 mm provides a more open flow path and reduces the metal jet velocity at the mold wall.
| Gating Parameter | Original | Optimized | Reason |
|---|---|---|---|
| Ladle nozzle diameter | 50 mm | 50 mm | Keep supply rate stable |
| Sprue diameter | 60 mm | 50 mm | Match nozzle area and keep sprue full |
| Sprue-to-nozzle area ratio | 1.44 | 1.00 | Eliminate air entrainment |
| Runner | Trapezoid 65/60 mm × 30 mm | Enlarged | Reduce flow velocity |
| Ingate thickness at casting | 10 mm | 25 mm | Open gating, lower jet velocity |
| Ingate location | Rib position | Rear square position on parting line | Improve filling and venting |
| Riser between ingates | None | One riser for venting and slag collection | Reduce gas and inclusion casting defect |
4.2 Modulus and Feeding Distance
The shaft cylinder and rib intersection region has a larger effective modulus than the 16 mm wall. I estimated the casting modulus \(M_c\) in the critical region and compared it with the riser modulus \(M_r\). The original waist risers had a certain modulus, but the padding did not provide a continuous feeding path. The effective feeding distance was therefore shorter than the geometric distance between the riser and the last-feeding region.
Using the feeding distance formula:
$$L_f = k M_r$$
I found that the original padding depth of 80 mm was below the required feeding distance for the critical section. The optimized padding depth of 120 mm matched the estimated requirement when \(k \approx 12\) and \(M_r \approx 10 \ \mathrm{mm}\). This does not mean that every casting will follow the same coefficient, but it explains why the original design produced shrinkage porosity casting defect between the risers.
I also reviewed the number and spacing of risers. For a circular shaft cylinder, the maximum distance between adjacent risers can be approximated as:
$$L_{\mathrm{max}} = \frac{\pi D}{n} – w_r$$
where \(D\) is the effective cylinder diameter, \(n\) is the number of risers, and \(w_r\) is the riser contact width. When \(n\) increases from 2 to 3, the maximum inter-riser distance decreases by roughly one third. This reduction is significant because the feeding distance required by the casting does not change. By using three circular insulating exothermic risers instead of two waist risers, I reduced the distance that the liquid steel must travel and improved the probability of sequential solidification.
| Feeding Parameter | Original | Optimized | Effect on Casting Defect |
|---|---|---|---|
| Number of risers | 2 waist-shaped | 3 circular | Shorter feeding distance, better liquid compensation |
| Riser type | Insulating exothermic waist | Insulating exothermic circular | Higher thermal efficiency, lower metal consumption |
| Padding geometry | Shaped to waist riser | Circumferential | Continuous feeding path |
| Padding thickness | 15-20 mm | 45 mm | Higher thermal modulus in feeding channel |
| Padding depth | 80 mm | 120 mm | Meets estimated feeding distance |
| Local chilling | Chills and rivets | Optimized chills and chromite sand | Reduces local hot-spot shrinkage |
4.3 Solidification Sequence and Defect Risk Index
I use a qualitative defect risk index to compare process alternatives. The index combines the thermal modulus ratio, the feeding distance ratio, and the gating turbulence factor:
$$R_d = \frac{M_c}{M_r} \cdot \frac{L_{\mathrm{required}}}{L_f} \cdot F_t$$
where \(R_d\) is the casting defect risk index, \(M_c\) is the casting modulus, \(M_r\) is the riser modulus, \(L_{\mathrm{required}}\) is the required feeding distance, \(L_f\) is the available feeding distance, and \(F_t\) is a turbulence factor. A value of \(R_d\) greater than 1 indicates a high risk of a casting defect. The original process had a high modulus ratio, an insufficient feeding distance, and a high turbulence factor. The optimized process reduced all three contributions.
| Risk Factor | Original Process | Optimized Process | Defect Risk Trend |
|---|---|---|---|
| Modulus ratio \(M_c/M_r\) | High | Lower | Improved feeding |
| Feeding distance ratio \(L_{\mathrm{required}}/L_f\) | Greater than 1 | Close to or less than 1 | Reduced shrinkage porosity |
| Turbulence factor \(F_t\) | High | Low | Reduced gas porosity |
| Local hot-spot control | Moderate | Improved | Reduced local casting defect |
| Sand-related risk | Moderate | Lower | Reduced sand inclusion |
5. Optimized Casting Process
After analyzing the casting defect mechanisms, I redesigned the process with three main changes: a new gating system, a new riser and padding scheme, and improved local chilling and sand control. The optimized process was designed to be robust in production, not just in a single trial. I also considered the economic effect, because a casting defect prevention plan that consumes excessive riser metal is not sustainable.
5.1 Gating System Redesign
I reduced the sprue diameter from 60 mm to 50 mm. This change matches the ladle nozzle diameter and forces the sprue to run full. A full sprue is essential for minimizing air entrainment and gas porosity casting defect. I also enlarged the runner and ingate. The ingate thickness at the casting increased from 10 mm to 25 mm, and the cross-section became fully open. The ingate location moved from the rib area to the rear square position on the parting line. This location provides a smoother metal path and better venting. Between the two ingates, I placed a riser that acts as a vent and slag collector. This riser also helps reduce gas and inclusion-related casting defect.
The new gating system follows the principle:
$$A_{\mathrm{sprue}} \le A_{\mathrm{runner}} \le A_{\mathrm{ingate}}$$
This is a non-pressurized, open gating system. The larger downstream areas reduce velocity and turbulence. The lower velocity reduces mold erosion and sand inclusion. The full sprue reduces air entrainment and gas porosity. The rear square ingate location reduces direct impingement on cores and fillets. All these changes are aimed at casting defect prevention.
5.2 Riser and Padding Optimization
I replaced the two waist-shaped insulating exothermic risers with three circular insulating exothermic risers. The circular shape is more efficient because it has a lower surface-area-to-volume ratio for the same modulus. The three-riser arrangement provides more uniform liquid shrinkage compensation around the shaft cylinder. It also promotes sequential solidification from the thin wall toward the risers. This is a direct countermeasure to shrinkage porosity casting defect.
The padding was changed from a shaped padding that matched the waist riser to a circumferential padding around the shaft cylinder. The padding thickness increased from 15-20 mm to 45 mm, and the padding depth increased from 80 mm to 120 mm. The larger thickness raises the thermal modulus of the feeding channel, and the larger depth extends the effective feeding distance. The result is a more continuous hot zone between the riser and the last-feeding region. The riser can therefore feed the casting for a longer time and compensate the liquid shrinkage. At the same time, the smaller circular risers consume less liquid metal than the original waist risers, so the yield improves.
| Optimization Area | Original Design | Optimized Design | Expected Effect on Casting Defect |
|---|---|---|---|
| Sprue diameter | 60 mm | 50 mm | Reduce air entrainment and gas porosity |
| Runner and ingate | Smaller, thin ingate | Enlarged, 25 mm ingate | Reduce turbulence and sand inclusion |
| Ingate location | Rib area | Rear square on parting line | Improve filling and venting |
| Riser count | 2 waist risers | 3 circular risers | Reduce shrinkage between risers |
| Riser type | Insulating exothermic waist | Insulating exothermic circular | Improve feeding efficiency and yield |
| Padding | Shaped, 15-20 mm thick, 80 mm deep | Circumferential, 45 mm thick, 120 mm deep | Extend feeding distance and prevent shrinkage porosity |
| Local chilling | Carbon steel chills and rivets | Optimized chills plus chromite sand | Reduce local hot-spot casting defect |
| Sand practice | CO2 sodium silicate sand | Controlled hardening and coating | Reduce sand inclusion and blowholes |
5.3 Mold and Core Practice
I kept the CO2-hardened sodium silicate sand system because it provides good strength and dimensional stability for this box-type casting. However, I tightened the process control. The sand must be uniformly mixed and hardened. The CO2 gas must be introduced uniformly, and the mold must be vented. The chromite sand at fillets and lettering positions remains important because it chills the local hot spots and reduces the risk of burn-on and shrinkage casting defect. I also reviewed the mold coating to ensure that it does not generate excessive gas. A coating that is too thick or not fully dried can cause gas porosity and surface casting defect.
I maintained the top machining allowance at 8 mm and other machined allowances at 5 mm to 6 mm. These allowances are sufficient for dimensional control and for removing any minor surface casting defect. The goal is not to rely on machining allowance to remove internal defects, because internal casting defect such as shrinkage porosity and gas porosity cannot be corrected by machining.
6. Verification and Production Results
I verified the optimized process through sectioning, batch production, and post-machining inspection. The first verification step was to section castings and examine the internal density in the shaft cylinder, riser contact zones, and rib intersections. The sectioned samples showed a dense structure with no visible shrinkage porosity or gas porosity. The feeding path from the risers to the casting was continuous, and the padding geometry was effective.
The second verification step was production. More than 300 castings were produced with the optimized process. After machining, the shaft cylinder inner surface was inspected for gas porosity. The previous diffuse 3-5 mm gas pores were no longer present. The region between the risers, which had previously shown shrinkage porosity, was also sound. Sand inclusion and sand burn-on were reduced because the gating velocity was lower and the mold and core practice was more stable. The overall scrap rate was below 4%. This is a significant improvement for a thin-wall, multi-hot-spot casting.
| Verification Item | Method | Original Process Result | Optimized Process Result |
|---|---|---|---|
| Gas porosity in shaft cylinder | Post-machining inspection | Frequent 3-5 mm pores | Not observed in production batch |
| Shrinkage porosity between risers | Machining and sectioning | Frequent internal porosity | Dense structure, no visible porosity |
| Sand inclusion | Visual and machining inspection | Occasional | Reduced |
| Internal density | Sectioning | Unsound in feeding zones | Dense and continuous |
| Yield | Mass balance | 57.6% | 62.3% |
| Scrap rate | Batch statistics | High | Below 4% |
| Production quantity | Batch record | Not optimized | More than 300 pieces |
The yield improvement can be calculated from the casting mass and the total poured mass:
$$Y = \frac{W_{\mathrm{casting}}}{W_{\mathrm{poured}}} \times 100\%$$
The original yield was:
$$Y_{\mathrm{original}} = 57.6\%$$
The optimized yield was:
$$Y_{\mathrm{optimized}} = 62.3\%$$
The absolute increase is:
$$\Delta Y = 62.3\% – 57.6\% = 4.7\%$$
The relative increase is:
$$\frac{62.3 – 57.6}{57.6} \times 100\% = 8.16\%$$
This means that the optimized process produces more castings from the same amount of liquid steel. At the same time, the casting defect rate is lower. The combination of higher yield and lower scrap rate improves both productivity and quality.
The scrap rate is defined as:
$$S = \frac{N_{\mathrm{defective}}}{N_{\mathrm{total}}} \times 100\%$$
With \(S < 4\%\), the optimized process is stable enough for regular production. The casting defect types that previously dominated, gas porosity and shrinkage porosity, are no longer the main drivers of scrap. This is the result of aligning the gating system, riser system, padding, and mold practice with the solidification behavior of the axle box body.
7. Discussion
From my perspective, the axle box body project demonstrates that casting defect prevention is a system problem. A single change, such as increasing the riser size, would not have solved both gas porosity and shrinkage porosity. The gas porosity was caused mainly by the gating system. The shrinkage porosity was caused mainly by the feeding system. The sand inclusion risk was related to the mold and core practice. Only by addressing all three areas together did the process become robust.
The most important lesson is that a thin-wall casting cannot be treated as a simple plate. The thermal modulus of the wall is low, but the thermal modulus of the rib intersections and the shaft cylinder is high. The casting defect risk is highest where the modulus changes abruptly. I therefore focus on the hot spots and the feeding paths between them. The padding is not just extra metal; it is a thermal conduit. The riser is not just a reservoir; it is a pressure source and a heat source. The gating system is not just a pipe; it is a flow conditioner that determines whether gas is entrained and whether the mold is eroded.
I also learned that production control is as important as process design. A robust design can still produce a casting defect if the sand is not hardened properly, if the mold coating is wet, or if the pouring temperature is inconsistent. For this reason, I include process control checks for sand moisture, CO2 hardening, mold venting, coating thickness, pouring temperature, pouring speed, and ladle nozzle condition. These checks are practical and low-cost, and they protect the process from variation.
| Control Item | Target or Check | Why It Matters for Casting Defect Prevention |
|---|---|---|
| Sand moisture | Within specified range | Excess moisture creates gas porosity and blowholes |
| CO2 hardening | Uniform and sufficient | Weak mold can spall and cause sand inclusion |
| Mold venting | Clear vents and proper venting path | Trapped gas causes gas porosity casting defect |
| Coating thickness | Uniform and fully dried | Wet coating generates gas and surface defect |
| Pouring temperature | Within approved window | Too low causes cold shut; too high increases shrinkage |
| Pouring speed | Controlled and steady | Unstable stream entrains gas and erodes mold |
| Ladle nozzle | Correct diameter and clean | Mismatch with sprue causes partial filling and gas entrainment |
| Riser placement | Accurate and secure | Riser movement or gap reduces feeding and causes shrinkage porosity |
| Chill placement | Correct position and contact | Poor contact reduces chilling and leaves local hot spots |
| Chromite sand use | At specified fillets and lettering | Reduces local burn-on and shrinkage casting defect |
Another important point is the role of the gating ratio. The original sprue was larger than the ladle nozzle, which seems counterintuitive to some operators because a larger sprue appears to allow faster filling. In reality, a larger sprue does not increase the flow rate if the ladle nozzle is the limiting area. It only causes the sprue to run partially full. A partially full sprue is one of the most common causes of gas porosity casting defect in steel castings. The optimized design matches the sprue to the ladle nozzle and enlarges the runner and ingate downstream. This creates a full sprue and a lower velocity in the mold cavity. The result is less entrained gas and less mold erosion.
The riser and padding optimization also has an economic dimension. The original two waist risers were large and consumed a significant amount of liquid steel. The optimized three circular risers are smaller individually and together consume less metal, yet they provide better feeding because they are better distributed and the padding is more effective. The yield improved from 57.6% to 62.3%. This improvement is not just a cost saving; it also reduces the thermal load on the mold and the risk of shrinkage porosity casting defect. A smaller riser that is well placed and well connected is better than a larger riser that is poorly connected.
The verification results support the process changes. After more than 300 castings, the scrap rate was below 4%, and the internal quality met the acceptance criteria. The sectioned castings showed a dense structure. The post-machining inspection did not find the previous gas porosity or shrinkage porosity casting defect. This confirms that the optimized process is capable of producing a sound axle box body in regular production.
8. Practical Design Rules I Apply to Similar Castings
Based on this work, I have developed a set of practical design rules for thin-wall, multi-hot-spot steel castings. These rules are not a substitute for simulation or trial, but they provide a strong starting point for casting defect prevention.
| Rule | Design Guideline | Reason |
|---|---|---|
| Gating rule 1 | Keep the sprue area equal to or smaller than the ladle nozzle area | Ensures a full sprue and reduces gas porosity casting defect |
| Gating rule 2 | Use an open gating system with increasing areas downstream | Reduces velocity, turbulence, and sand inclusion |
| Gating rule 3 | Place ingates away from fragile cores and fillets | Reduces mold erosion and sand-related casting defect |
| Riser rule 1 | Use the minimum number of risers that still satisfies feeding distance | Balances yield and shrinkage porosity prevention |
| Riser rule 2 | Prefer circular risers over elongated risers when possible | Better volume-to-surface ratio and thermal efficiency |
| Riser rule 3 | Use insulating or exothermic risers for high-modulus zones | Prolongs liquid time and improves feeding |
| Padding rule | Make the padding thick enough and deep enough to maintain feeding distance | Prevents shrinkage porosity casting defect between risers |
| Chill rule | Use chills, rivets, or chromite sand at local hot spots | Reduces local thermal modulus and shrinkage |
| Sand rule | Control moisture, hardening, venting, and coating | Prevents gas porosity and sand inclusion |
| Inspection rule | Section the first castings to verify internal density | Detects internal casting defect before production release |
I also use a simple decision flow for casting defect analysis. First, identify the defect location and morphology. Second, classify the defect as gas-related, shrinkage-related, or sand-related. Third, calculate the gating area ratio and the feeding distance. Fourth, compare the calculated values with the design. Fifth, modify the gating, riser, padding, or mold practice. Sixth, verify by sectioning and production. This flow keeps the analysis focused and prevents random changes.
9. Conclusion
The subway axle box body is a thin-wall, multi-hot-spot steel casting with strict internal quality requirements. In the original process, gas porosity in the shaft cylinder and shrinkage porosity between the risers were the dominant casting defect types. The gas porosity was caused by a partially filled sprue and entrained gas. The shrinkage porosity was caused by insufficient padding thickness and depth, which shortened the effective feeding distance. The original two waist risers did not provide enough liquid shrinkage compensation between them.
I optimized the process by matching the sprue diameter to the ladle nozzle, enlarging the runner and ingate, moving the ingate to a rear square position, and adding a venting riser between the ingates. I replaced the two waist risers with three circular insulating exothermic risers and changed the shaped padding to a circumferential padding with a thickness of 45 mm and a depth of 120 mm. I also tightened the mold and core practice, including sand hardening, venting, coating, and local chilling with chromite sand.
The optimized process produced a dense internal structure with no observed gas porosity or shrinkage porosity after machining. The yield increased from 57.6% to 62.3%. More than 300 castings were produced, and the scrap rate was below 4%. The process changes reduced the casting defect rate and improved product quality. The approach can be applied to other axle box bodies and similar thin-wall steel castings with multiple hot spots.
In my view, the key to casting defect prevention is to design the gating system, riser system, padding, and mold practice as one integrated thermal and fluid system. When these elements are aligned with the solidification behavior of the casting, internal defects such as gas porosity and shrinkage porosity can be controlled. The result is a reliable, economical, and high-quality casting process.