In my work as a foundry process engineer, I treat every axle box body steel casting as a system in which material, gating, feeding, sand-core rigidity, molding practice, and inspection all interact. The axle box body is a thin-wall box-type steel casting and an important load-bearing part in bogie assemblies. As locomotive speed and axle load increase, the demands on casting quality become stricter. In my analysis, the most important objective was not simply to repair individual casting defects, but to understand why casting defects such as shrinkage porosity, sand holes, cracks, sand-core tilting, and black skin after machining appeared and then to redesign the process so that those casting defects would not recur. The following discussion is written from my first-person perspective as a practical process review, with tables and formulas that summarize the decisions and the verification logic.
I began with the product requirements because casting defects in a load-bearing steel casting cannot be evaluated without knowing the service duty, machining allowance, and non-destructive testing acceptance criteria. The axle box body is cast in ZG25MnCrNiMo, a C-grade steel. Its main wall thickness is in the range of 20 to 30 mm, and the rough casting mass is about 95 kg. I used a linear shrinkage allowance of 2%. Molding and cores were made with CO2-hardened sodium silicate sand. The layout was two castings per flask. Melting was performed in a 10 t electric arc furnace. Pouring was bottom-gated, and the pouring temperature was controlled between 1560 and 1580 °C. The design required magnetic particle inspection over the whole casting surface. For locomotive assembly, all machined surfaces and principal load-bearing zones were required to be free from shrinkage porosity and cracks. Ultrasonic inspection was performed by batch. These requirements define the severity of any casting defects that remain after cleaning and machining.
| Requirement category | Value or condition | Process implication for casting defects |
|---|---|---|
| Material | ZG25MnCrNiMo, C-grade steel | High strength and hardenability require controlled cooling and feeding to avoid shrinkage porosity and cracks. |
| Main wall thickness | 20 to 30 mm | Thin walls cool quickly, but local hot spots can still form at ribs, bosses, and cylinder junctions. |
| Rough casting mass | 95 kg | Moderate mass with thin sections makes gating turbulence and sand erosion important. |
| Linear shrinkage | 2% | Pattern dimensions must compensate for contraction without reducing machining allowance. |
| Molding and cores | CO2-hardened sodium silicate sand | Good strength can be achieved, but sand shedding and core collapse must be prevented. |
| Molding layout | Two castings per flask | Flask rigidity and core positioning affect dimensional variation and casting defects. |
| Melting | 10 t electric arc furnace | Steel cleanliness and superheat control influence sand holes and shrinkage porosity. |
| Pouring | Bottom gating | Bottom gating reduces turbulence, but gate design can still create hot spots and cracks. |
| Pouring temperature | 1560 to 1580 °C | Too low causes misruns; too high increases sand erosion and shrinkage defects. |
| Surface inspection | Magnetic particle inspection, full surface | Cracks and surface casting defects must be eliminated before machining. |
| Internal inspection | Ultrasonic inspection by batch | Shrinkage porosity and internal sand holes must be controlled by feeding and cleanliness. |
| Machined and load-bearing zones | No shrinkage porosity or cracks | Feeding distance, riser design, and core stability become critical process variables. |
I use a few simple formulas to connect the observed casting defects with the physical behavior of the steel. The modulus of a casting section is the ratio of volume to cooling surface area:
$$ M = \frac{V}{A} $$
For solidification time, Chvorinov’s rule is useful:
$$ t_s = B M^2 $$
Here, \(t_s\) is solidification time, \(B\) is a mold constant, and \(M\) is the section modulus. For feeding to work, the riser must freeze later than the section it feeds:
$$ t_{s,r} > t_{s,c} $$
$$ M_r \ge 1.2 M_c $$
In my experience, when casting defects such as shrinkage porosity appear in the large axle cylinder, the problem is often not the absence of a riser but the incorrect feeding distance between risers or between the riser and the casting end zone. I therefore calculated the available feeding distance as:
$$ L_{feed} = L_{riser} + L_{end} $$
The required feeding correction for a tall cylinder can be expressed as:
$$ L_{required} = H_c – L_{feed} $$
The original process used a split along the middle of the axle cylinder. A kidney-shaped conformal pad was placed inside the large axle cylinder to promote directional solidification and feeding. At the connection between the small axle cylinder and the axle box ribs, a hot spot existed, but its modulus and radius were relatively small, so a semicircular blind riser was placed near the top connection to ensure internal quality. A 16 mm thick chill plate was used on the lower box of the large axle cylinder and on the spring plate face to strengthen chilling. The internal gate was placed at the central rib. In principle, this original process protected machined surfaces and principal load-bearing zones, but in batch production the main casting defects still appeared in the large axle cylinder, the internal gate, and the spring plate. There were also problems with sand shedding and uneven machining allowance.
| Original process element | Original design choice | Intended benefit | Observed casting defects |
|---|---|---|---|
| Parting line | Along the middle of the axle cylinder | Simplified molding and core placement | Core stability was sensitive to core print fit. |
| Large axle cylinder feeding | Kidney-shaped conformal pad | Promoted sequential solidification | Shrinkage porosity and sand holes remained. |
| Small axle cylinder hot spot | Semicircular blind riser near top connection | Improved local internal soundness | Limited effect on large cylinder feeding. |
| Chills | 16 mm thick chill plates on lower box and spring plate face | Increased local cooling rate | Chill placement and contact could be inconsistent. |
| Internal gate | Located at central rib | Convenient gating layout | Shrinkage and cracks at internal gate. |
| Molding method | CO2-hardened sodium silicate sand | Good strength after gassing | Sand shedding at rib cavities and ear cores. |
| Pouring | Bottom gating, 1560 to 1580 °C | Reduced turbulence and misruns | Sand holes and shrinkage porosity still occurred. |
| Machining allowance | Nominal allowance based on pattern | Expected clean machining | Spring plate allowance was uneven and sometimes insufficient. |
The problems I found in production can be grouped into four families of casting defects. First, sand shedding occurred on the connecting rib plate between the large and small axle cylinders, and similar trouble appeared at the ear-type hanging core. The fallen sand was enveloped by liquid steel and could not be removed, causing a small batch of rejects. Second, the large axle cylinder showed a certain proportion of sand holes and shrinkage porosity. After rough machining, these areas required welding repair, which increased cost and affected product quality. Third, the spring plate core tilted during closing and pouring, producing uneven machining allowance on the spring plate face. Fourth, the internal gate位置 showed shrinkage cracks, and cleaning required gouging and welding repair. Each of these casting defects had a different root cause, but they shared one common feature: the process relied on nominal dimensions and general practice rather than on local rigidity, local feeding distance, and local thermal balance.
| Casting defect | Location | Production consequence | Immediate suspected cause | Process lever |
|---|---|---|---|---|
| Sand shedding | Rib plate between large and small axle cylinders; ear hanging core | Sand inclusion and rejection | Low local sand strength; insufficient core reinforcement; handling impact | Core bone, nails, reserved holes, better handling |
| Sand holes | Large axle cylinder inner wall | Rework after machining | Dirty mold cavity; loose sand; turbulent metal | Better cleaning; larger machining allowance; gating control |
| Shrinkage porosity | Large axle cylinder | Internal reject or weld repair | Insufficient feeding distance; riser spacing too large | More risers; shorter spacing; exothermic sleeves |
| Crack at internal gate | Internal gate and central rib | Gouging and welding repair | Thickness difference causing thermal stress | Relocate gate to machined slot; balance section thickness |
| Uneven machining allowance | Spring plate face | Insufficient allowance; weld repair | Core print fit error; core tilting during closing | Side core prints; smaller core print; better location |
| Black skin after machining | Machined surfaces | Extra cleaning and repair | Sand burn-on and uneven allowance | Cleaner mold; coating control; allowance adjustment |
For sand shedding, I considered several possible mechanisms. The sand might have insufficient strength, poor bonding, or non-uniform mixing. The flask might lack enough bars or hangers. The upper mold surface or hanging sand might not be reinforced with hooks or nails. The sand might be insufficiently rammed, unevenly compacted, or locally missed. A deep mold cavity could cause collision during closing. After closing, pin adjustment or fixture tightening could also cause impact and vibration. In the axle box body, the outer mold used CO2-hardened sodium silicate sand. The two serious sand-shedding locations were the ear sand core and the deep rib cavity. The ear sand core was a hung core made by molding and gluing. During transport and turning, it could be struck, and if the core paste did not reach the required strength, sand would fall. The rib cavity between the large and small axle cylinders was about 50 mm high and relatively deep. After compaction on the vibration table, the local sand strength was still insufficient, and the flowing steel could erode the sand during pouring. I therefore adopted two improvements: first, when making the ear core, I reserved a central hole, and after hanging the core I inserted nails through the reserved hole for reinforcement; second, for the deep rib cavity, I folded welding electrodes into a core bone and placed it in the core to strengthen the sand core. These changes directly attacked the mechanical weakness that produced sand-related casting defects.
| Sand shedding factor | Original condition | Improved action | Expected effect on casting defects |
|---|---|---|---|
| Ear core handling strength | Hung core with glue; impact during transport | Reserve central hole; insert nails after hanging | Prevents core breakage and sand inclusion casting defects |
| Rib cavity sand strength | 50 mm high, deep cavity; vibration compaction only | Fold welding electrodes into core bone | Improves local tensile and erosion resistance |
| Core paste strength | Could be below requirement before handling | Control glue and curing before movement | Reduces falling sand and sand holes |
| Ramming uniformity | Possible uneven compaction or missed areas | Operator control and card-based inspection | Reduces sand shedding during pouring |
| Closing impact | Possible collision during mold closing | Smooth lifting and guided closing | Prevents mechanical damage to cores |
| Flask vibration | Pin adjustment and fixture tightening | Reduce knocking and shock after closing | Reduces loose sand and core shift |
The erosion of sand by liquid steel can be summarized by a simple relation. The scouring stress from the metal stream increases with density and with the square of velocity:
$$ \sigma_{scour} = C \rho v^2 $$
For the sand to survive, its local strength must exceed the scouring stress:
$$ \sigma_{sand} \ge \sigma_{scour} $$
This is why I paid attention not only to bulk sand strength but also to local reinforcement. A core bone or embedded nail increases \(\sigma_{sand}\) exactly where the metal velocity and thermal load are highest. When I later reviewed the casting defects, the elimination of sand shedding was one of the clearest improvements.
For the large axle cylinder, the main casting defects were shrinkage porosity and sand holes. Sand holes after machining were mainly caused by incomplete cleaning of the mold cavity. Since the inner wall of the cylinder is a machined surface, most sand holes can be removed if the machining allowance is large enough. I therefore increased the machining allowance in that area to make the process more robust against minor sand-related casting defects. This was not a substitute for cleaning, but it was a practical risk reduction. Shrinkage porosity, however, could not be solved by machining allowance because it is an internal feeding defect. I analyzed the feeding distance. The large cylinder wall thickness before machining was 25 mm. The end zone of feeding was about 65 mm, and the riser zone was about 50 mm. The vertical feeding distance was therefore about 115 mm. The axle cylinder height was 245 mm. To achieve full sequential solidification and feeding, the required pad depth was about 130 mm. The original two insulating risers were spaced 120 mm apart, and the feeding distance between them was about 100 mm. This was insufficient. The root cause of shrinkage porosity in the large axle cylinder was therefore an inadequate feeding distance between risers, combined with the difficulty of controlling pouring temperature. I changed the original two kidney-shaped insulating risers into three circular insulating exothermic risers. The riser spacing was reduced to 90 mm. In theory, this gives a better feeding field and reduces the tendency for shrinkage porosity without harming the process yield.
| Feeding parameter | Original value | Calculated or target value | Comment |
|---|---|---|---|
| Large cylinder wall thickness before machining | 25 mm | 25 mm | Thin wall but locally hot at junctions. |
| End zone feeding distance | About 65 mm | About 65 mm | Dependent on section modulus and chill effect. |
| Riser zone feeding distance | About 50 mm | About 50 mm | Improved by exothermic riser sleeves. |
| Vertical feeding distance | 115 mm | 115 mm | Sum of end zone and riser zone. |
| Axle cylinder height | 245 mm | 245 mm | Requires enough feeding along height. |
| Required pad depth | Not fully achieved | About 130 mm | Needed for sequential solidification. |
| Original riser spacing | 120 mm | 100 mm feeding distance | Insufficient for the local modulus. |
| New riser spacing | 120 mm | 90 mm | Three circular exothermic risers. |
| Riser type | Two kidney-shaped insulating risers | Three circular insulating exothermic risers | Better feeding and easier placement. |
The feeding distance calculation can be stated directly. The original spacing was too large relative to the available feeding distance:
$$ L_{feed, original} \approx 100 \text{ mm} < L_{required} \approx 130 \text{ mm} $$
After redesign, the spacing was reduced:
$$ s_{new} = 90 \text{ mm} < s_{original} = 120 \text{ mm} $$
This gives a more continuous feeding path and reduces the volume of isolated liquid that cannot be fed. When I connect this to casting defects, the logic is straightforward: isolated liquid zones become shrinkage porosity; continuous feeding zones remain sound. The change from two large risers to three smaller exothermic risers also allowed the riser positions to follow the thermal centers more closely.
| Riser design comparison | Original design | Improved design | Effect on casting defects |
|---|---|---|---|
| Number of risers | 2 | 3 | Shorter feeding paths, lower shrinkage porosity risk. |
| Shape | Kidney-shaped | Circular | More predictable modulus and easier exothermic sleeve fit. |
| Riser spacing | 120 mm | 90 mm | Improves feeding between risers. |
| Sleeve type | Insulating | Insulating exothermic | Extends riser solidification time. |
| Process yield impact | Baseline | Maintained within acceptable range | Fewer repair hours and lower rejection from casting defects. |
The spring plate machining allowance problem was related to core location, not to feeding. The original spring plate core was a circular disc core with a lower core print. The core box print position and the core seat draft had errors. The wooden pattern also deformed during production and use, so after core placement the gap between the core print and the core seat could become too large. During mold closing, the core tilted forward. After pouring, the spring plate face had uneven machining allowance, and the upper part sometimes had insufficient allowance and required welding repair. I changed the lower core print to left and right side core prints. The core print size was reduced, and the fit between the core print and the core seat became better. The positioning effect was better than the original design. This change directly reduced dimensional variation and the casting defects associated with insufficient machining allowance.
| Spring plate core issue | Original condition | Improved action | Result for casting defects |
|---|---|---|---|
| Core print type | Lower core print | Left and right side core prints | Better lateral location and less tilting. |
| Core print size | Large, sensitive to gap | Reduced size | Improved fit with core seat. |
| Pattern deformation | Wooden pattern deformed over time | Corrected print fit and process control | Reduced uneven allowance. |
| Closing behavior | Core tilted forward | Stable side location | More uniform spring plate thickness. |
| Machining allowance | Uneven; upper part insufficient | Target allowance improved | Less weld repair and fewer dimensional casting defects. |
The machining allowance can be expressed as a tolerance stack-up. The minimum allowance must remain positive after core error, mold error, core shift, and deformation are considered:
$$ a_{min} = a_{nom} – (\delta_{core} + \delta_{mold} + \delta_{shift} + \delta_{deform}) $$
For the spring plate, the target is:
$$ a_{min} \ge 3 \text{ mm} $$
In the improved trial, the spring plate position had about 7 mm machining allowance and the thickness was uniform. This is a comfortable margin for a casting with previous casting defects related to core movement.
The internal gate shrinkage and crack problem came from a local thickness difference. The internal gate size was 40/35 mm by height H. The height transitioned from the runner to the casting rib position at 15 mm. The rib thickness at the internal gate was 7.5 mm. The thickness ratio between the gate and the rib was therefore:
$$ R_t = \frac{t_{gate}}{t_{rib}} = \frac{15}{7.5} = 2 $$
Because of this difference, the rib solidified before the internal gate. The contraction of the rib produced tensile stress at the contact location, and a crack could form during solidification. To solve shrinkage porosity, the gate and the casting rib would need similar thickness, and the gating system would need to be open and flat. However, steel has a high melting point and poor fluidity. If the internal gate were made too flat, the pouring temperature could be difficult to control and misruns could occur. I therefore considered relocating the internal gate. I moved it to the left slot. The slot is a machined surface with a thickness of about 25 mm. This location is more suitable because even if a minor casting defect appears, machining can remove it. The change reduced the thermal stress concentration and improved the ability to remove any remaining casting defects by machining.
| Internal gate parameter | Original design | Improved design | Effect on casting defects |
|---|---|---|---|
| Gate size | 40/35 mm by H | Adjusted to suit new location | Better filling without local hot spot. |
| Gate height at casting | 15 mm | Moved to left slot | Reduces thickness mismatch. |
| Rib thickness | 7.5 mm | 7.5 mm | Original cause of thermal stress. |
| Thickness ratio | 2.0 | Closer to balanced local section | Lower crack risk. |
| Machining access | Central rib | Left slot, machined surface | Any small casting defect can be removed. |
| Thermal stress | High at gate-rib junction | Reduced by relocation | Fewer shrinkage cracks. |
The thermal stress driving the crack can be estimated by:
$$ \sigma = \frac{E \alpha \Delta T}{1-\nu} $$
If the local constraint is high and the section thickness changes abruptly, the stress can exceed the hot strength of the steel:
$$ \sigma \ge \sigma_{hot} $$
Then a crack becomes a casting defect. Relocating the gate to a thicker machined slot reduces the abrupt constraint and makes the stress easier to accommodate. This is a good example of how a casting defect can be solved by process geometry rather than by post-casting repair.
After completing the cause analysis, I designed a new casting process. The new process was simulated, and the simulation results agreed with the analysis. I then made tooling and performed trial production. During production, I strengthened control of the molding stage. The operators placed chills, sprue sticks, and insulating riser sleeves according to the process card. Chills were not allowed to be tilted or bedded on loose sand. The sand thickness between the core bone and the mold wall, or the eat-sand distance, was controlled between 15 and 50 mm, or above that range where required. Lifting the pattern had to be smooth. The compactness of the mold and core had to be uniform and meet the process card. If no process card value was specified, the compactness was controlled above 45 units. When coating the mold and core, the coating depth had to be appropriate, without accumulation, flow marks, or thick and thin patches. Before hanging the core, the mold cavity had to be cleaned. Loose sand around chills and vent holes had to be scraped flat and blown clean. Vent passages had to be open. Oxygen addition holes in the cavity had to be plugged with molding sand and smoothed. The core surface had to be clean, fins and burrs removed, and sharp edges rounded.
| Production control point | Requirement | Reason for casting defects control |
|---|---|---|
| Chill placement | According to process card; no tilt or sand bedding | Ensures local cooling and prevents shrinkage porosity. |
| Sprue stick and riser sleeve placement | Exact position and alignment | Maintains gating and feeding path. |
| Eat-sand distance | 15 to 50 mm, or as specified | Balances core strength and cooling. |
| Pattern lifting | Smooth and perpendicular | Avoids mold wall damage and loose sand. |
| Mold and core compactness | Uniform; at least 45 units unless otherwise specified | Prevents sand shedding and sand holes. |
| Coating | Proper depth, no accumulation, no flow marks, uniform thickness | Reduces burn-on and black skin. |
| Cavity cleaning before core hanging | Remove all loose sand and debris | Prevents sand holes and inclusions. |
| Chill and vent area cleaning | Scrape flat and blow clean | Prevents loose sand entrapment. |
| Vent passages | Open and unobstructed | Reduces gas-related casting defects. |
| Oxygen addition holes | Plug with sand and smooth | Prevents metal penetration and surface casting defects. |
| Core surface | Clean; fins and burrs removed; sharp edges rounded | Reduces local sand breakage and inclusion casting defects. |
For compactness, I used a simple relative density expression:
$$ C = \frac{\rho – \rho_0}{\rho_0} \times 100\% $$
For cavity cleanliness, I also used a qualitative index:
$$ I_c = 1 – \frac{A_{residue}}{A_{cavity}} $$
Here, \(A_{residue}\) is the area of residual loose sand or debris, and \(A_{cavity}\) is the cavity area. The closer \(I_c\) is to 1, the lower the risk of sand-related casting defects. This index helped communicate to operators that cleaning is not a minor housekeeping step but a direct casting defect prevention measure.
After producing 10 sample pieces with the improved process, the castings were cleaned and ground, then inspected for appearance. There was no sand shedding in the improved castings. Dimensional layout inspection showed that the main inspection dimensions were good. The spring plate position had about 7 mm machining allowance and uniform thickness. After cleaning, the axle cylinders were inspected by ultrasonic testing, and no shrinkage porosity was found inside any casting. External inclusions and sand holes were significantly reduced. The internal gate position was inspected by magnetic particle testing, and there were no cracks or other surface casting defects; the inspection passed. The first batch was inspected and tested piece by piece, and all passed. This verified the reasonableness of the cause analysis and the feasibility of the improvement measures.
| Verification item | Target | Observed result on 10 samples | Status |
|---|---|---|---|
| Sand shedding after cleaning | None | No sand shedding found | Passed |
| Main dimensions | Within drawing requirements | All main inspection dimensions good | Passed |
| Spring plate machining allowance | Uniform and sufficient | About 7 mm, uniform thickness | Passed |
| Ultrasonic inspection of axle cylinder | No shrinkage porosity | No shrinkage porosity found | Passed |
| External inclusions and sand holes | Significantly reduced | Clearly reduced compared with original process | Passed |
| Magnetic particle inspection of internal gate | No cracks or surface casting defects | No cracks or surface defects | Passed |
| Batch inspection | All pieces accepted | All first-batch pieces accepted | Passed |
To summarize the improvement in quantitative terms, I used defect rate and process yield. The defect rate for a given casting defect type is:
$$ D_i = \frac{N_i}{N_{total}} \times 100\% $$
The reduction in a casting defect type is:
$$ \Delta D = \frac{D_{old} – D_{new}}{D_{old}} \times 100\% $$
The process yield is:
$$ Y = \frac{W_{casting}}{W_{poured}} \times 100\% $$
When the riser design changes from two larger risers to three smaller exothermic risers, the yield must be watched, but the reduction in repair and rejection often compensates for the additional riser metal. In my trial, the main benefit was not a dramatic yield increase but a strong reduction in casting defects and repair work. The process became more stable, and the operators had clearer control points.
| Key performance indicator | Original process tendency | Improved process tendency | Interpretation |
|---|---|---|---|
| Sand shedding casting defects | Small batch rejects at ear core and rib cavity | No sand shedding in 10-piece trial | Core reinforcement worked. |
| Shrinkage porosity in large axle cylinder | Regular occurrence requiring repair | No ultrasonic indications in trial | Feeding distance improved. |
| Sand holes after machining | Present and required rework | Significantly reduced | Cleaning and allowance improved. |
| Crack at internal gate | Required gouging and welding | No magnetic particle indications | Gate relocation reduced thermal stress. |
| Spring plate allowance | Uneven, sometimes insufficient | About 7 mm and uniform | Side core prints improved location. |
| Black skin after machining | Occasional | Reduced | Better coating and cleaner cavity. |
| Repair cost | High due to weld repair | Lower | Fewer casting defects reaching machining. |
| Process confidence | Reactive | More preventive | Cause analysis linked to control points. |
From my perspective, the most important lesson is that casting defects in axle box body steel castings should not be treated as isolated events. Sand shedding, sand holes, shrinkage porosity, cracks, and uneven machining allowance are all symptoms of process variables that can be measured, controlled, and improved. The large axle cylinder needed a feeding design based on modulus and feeding distance. The rib cavity needed a core bone because local sand strength was insufficient. The ear core needed mechanical retention because handling loads were high. The spring plate core needed better core print location because core shift changed the machining allowance. The internal gate needed relocation because local thickness difference produced thermal stress. When these changes were combined, the casting defects decreased and the casting passed inspection.
I also learned that the relationship between gating and casting defects is not only about filling. A gate that fills the mold can still create a hot spot, a crack, or a shrinkage defect if it is placed at a thin rib or if it creates an abrupt section change. For this reason, I now review gating design together with feeding design and core design. The gate must deliver clean steel without eroding sand, but it must also allow the resulting thermal field to solidify in a way that avoids casting defects. In the axle box body, moving the internal gate to the left slot was a simple change with a strong effect because the slot is a machined surface and has a more suitable thickness.
The core design is equally important. A core is not only a shape former; it is a load-bearing structure during transport, closing, pouring, and solidification. If the core is weak or poorly located, the result will be casting defects such as sand inclusions, sand holes, core shift, and uneven wall thickness. The changes I made, including reserved holes with nails, welding-electrode core bones, and side core prints, were low-cost but highly effective. They improved the local stiffness and the positioning accuracy of the sand cores. This reduced casting defects and made the process more repeatable.
Molding and pouring controls remain the foundation. Even a well-designed feeding system can fail if the mold cavity is dirty, if chills are poorly placed, if coating is uneven, or if compactness is not uniform. I therefore use process cards and operator checks for the critical points. The casting defects that once appeared after machining were often related to small details: loose sand around a chill, a blocked vent, a coating flow mark, or a core fin. These details matter because the axle box body has machined surfaces and load-bearing zones that cannot accept shrinkage porosity or cracks. By controlling these details, I reduced the probability that a minor sand or gas defect would become a major reject.
In terms of inspection, magnetic particle testing and ultrasonic testing are not merely final gates. They are feedback tools. When magnetic particle testing finds a crack near the internal gate, it tells me that the thermal stress or section change is still too severe. When ultrasonic testing finds shrinkage porosity in the large axle cylinder, it tells me that the feeding distance or riser modulus is still insufficient. When dimensional inspection finds uneven spring plate allowance, it tells me that the core print or closing operation is still unstable. I used this feedback to refine the process until the trial castings passed. The successful 10-piece trial was not the end of the work, but it confirmed that the direction of improvement was correct.
| Cause analysis summary | Physical mechanism | Process change | Resulting casting defect reduction |
|---|---|---|---|
| Sand shedding at ear core | Handling impact and insufficient core retention | Reserved hole and embedded nails | Fewer sand inclusions and sand holes. |
| Sand shedding at rib cavity | Low local sand strength and metal erosion | Welding-electrode core bone | Improved sand retention and lower sand-related casting defects. |
| Shrinkage porosity in large cylinder | Insufficient feeding distance between risers | Three circular exothermic risers at 90 mm spacing | No ultrasonic shrinkage indications in trial. |
| Sand holes in large cylinder | Dirty cavity and minor loose sand | Better cleaning and larger machining allowance | Sand holes removed by machining or reduced. |
| Spring plate allowance variation | Core tilt due to poor core print fit | Side core prints with reduced size | Uniform allowance of about 7 mm. |
| Internal gate crack | Thickness difference and thermal stress | Relocate gate to left slot | No magnetic particle indications. |
| Black skin after machining | Burn-on and uneven coating | Coating control and cavity cleanliness | Cleaner machined surfaces. |
If I express the feeding condition as a design inequality, it becomes a useful rule for future work:
$$ M_r \ge 1.2 M_c $$
$$ L_{feed} \ge L_{required} $$
$$ \sigma_{hot} \ge \sigma_{thermal} $$
These inequalities summarize my approach. The riser must be large enough, the feeding path must be short enough, and the hot strength of the steel at the critical section must be greater than the thermal stress. When any of these conditions is violated, casting defects appear. When all are satisfied, the process is much more robust. The axle box body improvement showed that this way of thinking can be applied to a real production casting.
In future production, I would continue to monitor the casting defects that matter most for this part: shrinkage porosity in the large axle cylinder, sand holes on the machined inner wall, cracks at the internal gate, and dimensional variation at the spring plate. I would keep the improved core reinforcement and core print design. I would maintain the three-riser feeding layout unless simulation or production data show a better arrangement. I would also continue to record pouring temperature, compactness, coating thickness, and inspection results so that the process can be adjusted with evidence rather than guesswork. In my view, the best casting process is not the one with the most complex design, but the one in which every casting defect has a known cause and every cause has a controlled countermeasure.
| Future control item | Why it matters for casting defects | Monitoring method |
|---|---|---|
| Pouring temperature | Too high increases sand erosion; too low causes misruns | Thermocouple or optical pyrometer record |
| Mold compactness | Low compactness causes sand shedding and sand holes | Compactness tester and operator checks |
| Core print fit | Poor fit causes core shift and uneven allowance | Pre-closing gauge and visual check |
| Riser sleeve condition | Damaged sleeve reduces feeding and causes shrinkage porosity | Inspection before placement |
| Chill contact | Poor contact reduces cooling and creates local shrinkage | Visual check after placement |
| Cavity cleanliness | Residual sand causes sand holes and inclusions | Cleanliness index and final blow-off |
| Coating uniformity | Uneven coating causes burn-on and black skin | Visual and thickness check |
| Inspection feedback | Links casting defects to process variables | Magnetic particle and ultrasonic records by batch |
In conclusion, my improvement of the axle box body casting process focused on the root causes of casting defects rather than on isolated repairs. The original process produced sand shedding, sand holes, shrinkage porosity, cracks, and uneven machining allowance. I analyzed each defect family. I reinforced the ear core with a reserved hole and nails. I strengthened the deep rib core with a welding-electrode core bone. I reduced the riser spacing from 120 mm to 90 mm by replacing two kidney-shaped insulating risers with three circular insulating exothermic risers. I increased machining allowance where sand holes could remain. I changed the spring plate core from a lower core print to left and right side core prints. I relocated the internal gate to the left slot to reduce thermal stress. I tightened molding, cleaning, coating, and compactness controls. The trial production of 10 pieces showed no sand shedding, good dimensions, uniform spring plate allowance, no ultrasonic shrinkage porosity, and no magnetic particle cracks at the internal gate. The casting defects were brought under control, and the process became more reliable. This experience reinforced my belief that casting defects are not random; they are the visible result of process variables that can be understood, calculated, and improved.