I investigated a persistent and costly casting defect problem in the production of a 250t oil cylinder for a rubber injection machine. The casting is made of QT500-7 material, has a single-piece pouring weight of 1439 kg, and is produced by resin sand molding and core making in a manual production environment. Over a four-month period, I tracked 47 castings. Among them, 10 castings were rejected, giving a total rejection rate of 21.3%. The direct iron loss associated with these rejected castings was approximately 14.5 t. This level of loss was unacceptable, so I initiated a focused study of the dominant casting defect modes and their causes.
The rejected castings were not random in their casting defect distribution. Sand drop accounted for 6 rejected castings, or 60% of all rejects. Bleed or run-out accounted for 3 rejected castings, or 30% of all rejects. Core eccentricity accounted for 1 rejected casting, or 10% of all rejects. Therefore, sand drop and bleed together represented 90% of the total rejected castings. This meant that any meaningful improvement had to address these two casting defect categories first. I organized a process improvement effort around the specific production steps of molding, core setting, mold closing, clamping, and pouring.
1. Production Context and Casting Defect Statistics
The 250t oil cylinder is a heavy-section casting. More than two thirds of its total height is located in the upper molding box. This geometric feature is important because the static pressure of the molten metal at the parting line increases as pouring progresses. The casting is produced manually, so the quality of mold closing, core positioning, clamping, and pouring control depends strongly on operator technique and discipline. I considered this context essential because many casting defect mechanisms are influenced by both process design and human execution.
I calculated the overall rejection rate using the following relationship:
$$ P_{\text{reject}} = \frac{N_{\text{reject}}}{N_{\text{total}}} \times 100\% $$
Substituting the recorded values:
$$ P_{\text{reject}} = \frac{10}{47} \times 100\% = 21.3\% $$
The share of each casting defect type among the rejected castings was calculated as:
$$ S_i = \frac{N_i}{N_{\text{reject}}} \times 100\% $$
For sand drop:
$$ S_{\text{sand}} = \frac{6}{10} \times 100\% = 60\% $$
For bleed or run-out:
$$ S_{\text{bleed}} = \frac{3}{10} \times 100\% = 30\% $$
For core eccentricity:
$$ S_{\text{core}} = \frac{1}{10} \times 100\% = 10\% $$
The combined contribution of the two main casting defect modes was therefore:
$$ S_{\text{sand+bleed}} = 60\% + 30\% = 90\% $$
Table 1 summarizes the observed casting defect statistics. I used this table as the basis for prioritizing corrective actions.
| Defect type | Rejected castings | Share of rejected castings | Share of total production | Main location |
|---|---|---|---|---|
| Sand drop | 6 | 60% | 12.8% | Upper box square top plane near four small holes |
| Bleed or run-out | 3 | 30% | 6.4% | Parting line |
| Core eccentricity | 1 | 10% | 2.1% | Central bore region |
| Total | 10 | 100% | 21.3% | Multiple locations |
The data showed that the casting defect problem was concentrated. I did not need to solve every possible casting defect at once. Instead, I focused on sand drop and bleed because they accounted for almost all rejected castings. The remaining core eccentricity casting defect was less frequent and did not dominate the loss.
2. Classification and Terminology of the Relevant Casting Defects
Before analyzing causes, I classified the two dominant casting defect modes according to standard terminology. Sand drop is a multi-metal type of casting defect. It is defined as a blocky metal protrusion on the casting surface caused by local sand blocks falling from the mold or core under mechanical force. The shape of the protrusion often resembles the dropped sand block. In other locations of the same casting, sand holes or missing areas may also appear because the fallen sand has displaced or contaminated the mold cavity.
Bleed or run-out is an incomplete type of casting defect. It is defined as a severe depression or incomplete region above the parting line caused by molten metal flowing out from the parting line during pouring. In some cases, a thin fin-like residue remains along the unfilled cavity surface. In my observations, bleed occurred near the end of pouring, when the cavity was nearly full and the internal metal pressure had increased significantly.
Table 2 presents a comparison of the two casting defect modes in terms of definition, appearance, location, and consequence.
| Feature | Sand drop casting defect | Bleed or run-out casting defect |
|---|---|---|
| Defect category | Multi-metal defect | Incomplete defect |
| Typical appearance | Blocky metal protrusion, sand holes, missing areas | Depression, incomplete section, fin-like residue |
| Main location | Upper box square top plane near four small holes | Parting line |
| Timing | Mostly related to mold closing and core damage | Mostly during late pouring |
| Primary consequence | Rejection, rework, sand inclusion | Rejection, iron loss, safety risk |
| Observed frequency | 6 of 10 rejected castings | 3 of 10 rejected castings |
I used this classification to separate the casting defect mechanisms. Sand drop is primarily a mechanical damage phenomenon before pouring. Bleed is primarily a pressure and sealing phenomenon during pouring. Although both casting defects appeared in the same product, their root causes and countermeasures were different. I therefore analyzed them separately and then integrated the findings into a single control plan.
3. Sand Drop Casting Defect: Characteristics and Mechanism
I examined the sand drop casting defect in detail. The defect was concentrated on the upper box square top plane near four small holes. In the rejected castings, I observed blocky metal protrusions that matched the shape of dropped sand fragments. In other regions of the same castings, I found sand holes and incomplete areas. This pattern confirmed that sand blocks had fallen into the mold cavity before or during pouring and had been enveloped by the molten metal.
In my analysis, the sand drop casting defect was not caused by a general sand strength problem. The first five trial castings were acceptable, and the first fifteen production castings did not show any sand drop. This evidence excluded the initial process design and the pattern equipment as the primary root cause. If the sand formulation or core strength had been fundamentally insufficient, the casting defect would have appeared from the beginning. Instead, the defect appeared intermittently after a period of stable production. This pointed to a change in mold closing practice, core setting accuracy, or handling discipline.
The mechanical mechanism of sand drop can be described by the kinetic energy of a falling or colliding sand block. When the upper box is lowered onto the lower box, a sand block or core head may collide with the lower mold or core print. The impact energy is approximately:
$$ E_k = \frac{1}{2} m_s v_c^2 $$
where \( m_s \) is the mass of the sand block or damaged core fragment, and \( v_c \) is the relative closing velocity. If this energy exceeds the local fracture energy of the sand core or mold edge, the core head breaks. The fracture energy can be approximated as:
$$ E_f = \sigma_s A_s \delta_s $$
where \( \sigma_s \) is the local sand strength, \( A_s \) is the effective fracture area, and \( \delta_s \) is the characteristic deformation before fracture. When \( E_k > E_f \), sand blocks fall into the cavity. During pouring, the metal fills the void left by the fallen sand, creating a protrusion. At the same time, the fallen sand itself may be washed into another location, creating a sand hole or inclusion.
I also considered the geometry of the core print and the locating system. The locating sand cone top was lower than the core head top by 150 mm. This meant that the sand cone could not properly locate the core head and core seat during closing. Without reliable lateral and vertical location, the upper box could contact the core head before the parting surfaces were fully aligned. This increased the probability of core head damage and sand drop.
Table 3 lists the possible causes of the sand drop casting defect that I evaluated, together with the evidence and my assessment.
| Possible cause | Evidence from production | Likelihood | My assessment |
|---|---|---|---|
| Upper and lower parting surfaces not parallel during closing | First castings acceptable, later intermittent defects | High | Most likely operational cause |
| Upper and lower boxes misaligned | Locating sand cone top lower than core head top by 150 mm | High | Core head/core seat location ineffective |
| Crane sliding during closing | Not consistently repeated | Low | Not a primary cause |
| Unsuitable core print clearance or taper | Early castings and trial samples were acceptable | Low | Process design likely acceptable |
| Pattern dimension problem | Early castings were acceptable | Low | Tooling likely acceptable |
| Inaccurate position of four small core heads | Observed after closing in some cases | Medium | Contributing factor |
| Insufficient care during core setting | Intermittent nature of the casting defect | High | Human factor |
The evidence strongly suggested that the sand drop casting defect was an operational casting defect rather than a design casting defect. The first acceptable castings proved that the tooling and basic process were capable. The intermittent appearance proved that the process was not stable. The location of the defect near the four small core heads proved that the core head area was vulnerable during mold closing. Therefore, I focused on mold closing parallelism, box alignment, core head protection, and verification before pouring.
4. Sand Drop Casting Defect: Preventive Measures
I implemented a set of preventive measures to eliminate the sand drop casting defect. These measures were directed at the specific mechanism identified in the previous section. I did not rely on a single action. Instead, I combined training, mechanical checks, procedural verification, and a possible hardware improvement.
First, I raised the level of responsibility and care during mold closing. The operators were instructed to treat the core head area as a protected zone. They were required to slow down the final lowering stage and to observe the gap between the upper and lower boxes before full contact. This reduced the closing velocity \( v_c \) in the kinetic energy expression \( E_k = \frac{1}{2} m_s v_c^2 \). Because kinetic energy depends on the square of velocity, even a moderate reduction in closing speed produced a significant reduction in impact energy.
Second, I required the lower box to be placed horizontally before closing. The upper box was adjusted through the lifting rigging to ensure that its parting surface was horizontal. The parallelism deviation was checked and controlled. I defined the parallelism deviation as:
$$ \Delta h = h_{\max} – h_{\min} $$
where \( h_{\max} \) and \( h_{\min} \) are the maximum and minimum gap heights between the upper and lower parting surfaces. I set an internal acceptance criterion:
$$ \Delta h \leq \delta_{\text{allow}} $$
For this casting, I used a strict allowable deviation to prevent local contact with the core head before full parting surface contact. When the upper box is not parallel, one side may contact the core head while the other side is still high. This creates a lever effect that breaks the core head. Ensuring parallelism prevents this local overload.
Third, I required multi-person operation during mold closing. One person controlled the crane, while another person guided the upper box and observed the alignment. The relative position of the upper and lower boxes was checked before lowering. This reduced the chance of lateral misalignment. Lateral misalignment can cause the upper box to strike the core head from the side, which is especially damaging because the core head has limited bending strength.
Fourth, I introduced verification before pouring. When necessary, a trial closing or empty closing was performed. A template or card was used to verify the position of the four small core heads. This verification step was simple but effective. If a core head was damaged or displaced, the mold was not poured. This prevented the sand drop casting defect from reaching the final casting.
Fifth, I considered a resin sand guide pin and bushing system for mold closing. A guide pin and bushing provide positive alignment before the parting surfaces contact the core head. This reduces reliance on operator judgment. The guide system can be designed with a lead-in taper to gradually align the boxes. The alignment force can be estimated from the lateral displacement resistance of the guide pin. A properly designed guide system does not replace careful operation, but it makes correct operation easier and more repeatable.
Table 4 summarizes the sand drop casting defect preventive measures, their target mechanism, and the expected effect.
| Preventive measure | Target mechanism | Expected effect |
|---|---|---|
| Increase care during mold closing | Reduce impact energy on core head | Lower probability of core head fracture |
| Level lower box and adjust upper box rigging | Ensure parting surface parallelism | Prevent local contact and lever loading |
| Use multi-person operation | Improve lateral alignment | Reduce side impact on core head |
| Perform trial closing or template verification | Detect damaged or displaced core heads | Prevent pouring defective molds |
| Use resin sand guide pin and bushing | Provide positive alignment | Make correct closing repeatable |
| Protect four small core heads | Reduce local stress concentration | Lower sand drop casting defect rate |
After implementing these measures, I observed a clear reduction in sand drop. The casting defect had been intermittent, so I monitored several consecutive production batches. The results are discussed later together with the bleed casting defect results.
5. Bleed or Run-Out Casting Defect: Characteristics and Mechanism
I next analyzed the bleed or run-out casting defect. In the rejected castings, the defect appeared along the parting line. The most severe cases showed missing metal above the parting line. In less severe cases, a fin-like residue remained. The operators reported that near the end of pouring, when the mold was nearly full, a loud popping sound occurred, followed by a large amount of molten metal overflowing from the parting line. This created a safety risk for the pouring crew and caused significant iron loss.
The bleed casting defect is driven by the static pressure of the molten metal. The pressure at the parting line can be expressed as:
$$ p_s = \rho_m g h $$
where \( \rho_m \) is the density of the molten metal, \( g \) is the gravitational acceleration, and \( h \) is the height of the molten metal above the parting line. As pouring progresses, \( h \) increases. Near the end of pouring, \( h \) reaches its maximum, so \( p_s \) reaches its maximum. If the parting line has a weak seal or a local gap, the molten metal can escape.
The flow rate of molten metal through a parting line gap can be approximated by a parallel-plate flow model:
$$ Q = \frac{w h_g^3 \Delta p}{12 \mu L} $$
where \( Q \) is the volumetric leakage rate, \( w \) is the gap width, \( h_g \) is the gap height, \( \Delta p \) is the pressure difference driving the flow, \( \mu \) is the dynamic viscosity of the molten metal, and \( L \) is the flow length along the gap. This equation shows that the leakage rate is extremely sensitive to the gap height \( h_g \). If the gap height doubles, the leakage rate increases by a factor of eight. Therefore, even a small parting line gap can cause a severe bleed casting defect.
The upper mold can also be lifted by the molten metal. The lifting force is the sum of buoyancy and static pressure acting on the projected area of the upper mold:
$$ F_{\text{lift}} = \rho_m g V_{\text{mold}} + A_{\text{proj}} p_s $$
where \( V_{\text{mold}} \) is the displaced volume of the mold, and \( A_{\text{proj}} \) is the projected area of the upper mold exposed to the metal pressure. The clamping force must exceed this lifting force with a safety factor:
$$ F_{\text{clamp}} \geq k F_{\text{lift}} $$
where \( k \) is a safety factor. If the clamping force is insufficient, the upper mold lifts, the parting line gap increases, and the bleed casting defect occurs. If the clamping force is sufficient but the parting surface is not flat or the seal is poor, leakage can still occur.
I evaluated the possible causes of the bleed casting defect. Table 5 presents the causes, the evidence, and my assessment.
| Possible cause | Evidence | Assessment |
|---|---|---|
| Poor sand sealing on parting line | Bleed occurred along parting line | High likelihood |
| Upper and lower parting surfaces not flat, warped, or with gaps | Intermittent bleed, similar to sand drop pattern | High likelihood |
| Insufficient clamping force | Bleed occurred near end of pouring when pressure was highest | Medium to high likelihood |
| Improper clamping position | Local parting line opening | Medium likelihood |
| Too light clamping weight | Possible uplift of upper box | Medium likelihood |
| Too fast pouring at the end | Pressure surge and dynamic impact | Medium likelihood |
| Core collision causing box wedging | Similar root cause as sand drop | High likelihood |
The analysis showed that the bleed casting defect and the sand drop casting defect shared a common root cause pattern. Both were associated with mold closing accuracy. If the upper and lower boxes were not properly aligned, the core could be struck, or the boxes could become wedged with a local gap. That local gap became a leakage path when the metal pressure increased. Therefore, the countermeasures for bleed overlapped with those for sand drop, but they also required additional attention to clamping and pouring control.
6. Bleed Casting Defect: Preventive Measures
I implemented a set of preventive measures for the bleed casting defect. These measures addressed sealing, clamping, parting surface condition, and pouring practice.
First, I improved the responsibility and care during sealing and clamping. The operators were instructed to inspect the parting line before pouring. Any loose sand, gap, or damaged edge was repaired. A sand sealing strip or refractory paste was applied along the parting line where needed. This increased the flow resistance \( L \) and reduced the effective gap height \( h_g \) in the leakage equation. Because leakage is proportional to \( h_g^3 \), even a small reduction in gap height produced a large reduction in leakage risk.
Second, I ensured that the core was not collided with during mold closing. This prevented the upper and lower boxes from becoming wedged or tilted. If the core head is struck, the boxes may not close completely, leaving a gap that is difficult to see from the outside. By preventing core collision, I removed a major cause of local parting line gaps.
Third, I checked and maintained the parting surfaces. Warped or damaged boxes were repaired or replaced. The parting surfaces were cleaned and inspected before each use. A flatness check was performed. I defined the local gap as \( h_g(x,y) \) over the parting surface. The maximum gap \( h_{g,\max} \) was controlled to be below an allowable value:
$$ h_{g,\max} \leq h_{\text{allow}} $$
This controlled the worst-case leakage path.
Fourth, I optimized clamping. The number, strength, and position of clamps were reviewed. The clamping force was calculated using the lifting force model. For a simplified estimate, the static pressure at the parting line was:
$$ p_s = \rho_m g h $$
Using \( \rho_m = 7000 \, \text{kg/m}^3 \), \( g = 9.81 \, \text{m/s}^2 \), and a metal height above the parting line of \( h = 0.4 \, \text{m} \), the pressure was:
$$ p_s = 7000 \times 9.81 \times 0.4 = 27468 \, \text{Pa} \approx 27.5 \, \text{kPa} $$
If the projected area of the upper mold exposed to this pressure was \( A_{\text{proj}} = 0.25 \, \text{m}^2 \), the pressure-induced lifting force was:
$$ F_p = p_s A_{\text{proj}} = 27468 \times 0.25 = 6867 \, \text{N} \approx 6.87 \, \text{kN} $$
Adding a safety factor of \( k = 1.5 \), the required clamping force was approximately:
$$ F_{\text{clamp}} \geq 1.5 \times 6.87 = 10.3 \, \text{kN} $$
This estimate showed that the clamping system had to provide a reliable force greater than about 10 kN, with proper distribution around the mold. In practice, I used multiple clamps and checked that each clamp was tightened correctly. The actual clamping force depends on the clamp type and tightening torque. For a bolt clamp, the preload can be estimated as:
$$ F_b = \frac{T}{K d} $$
where \( T \) is the tightening torque, \( K \) is the nut factor, and \( d \) is the nominal bolt diameter. This formula helped the team understand that clamp force is not guaranteed by simply installing a clamp; it must be generated by proper tightening.
Fifth, I controlled the pouring speed, especially at the end of pouring. A slower final pouring stage reduced the dynamic pressure surge and gave the sand seal time to resist the static pressure. The pouring rate can be expressed as:
$$ Q_{\text{pour}} = \frac{dV}{dt} $$
where \( V \) is the volume of metal in the mold. A controlled decrease in \( Q_{\text{pour}} \) near the end reduced the rate of pressure increase at the parting line. This was particularly important because the bleed casting defect occurred at the moment of maximum pressure.
Table 6 summarizes the bleed casting defect preventive measures and their functions.
| Preventive measure | Physical function | Expected result |
|---|---|---|
| Improve sealing and clamping care | Reduce gap height and increase flow resistance | Lower leakage rate |
| Avoid core collision | Prevent box wedging and local gaps | Maintain full parting surface contact |
| Check and maintain parting surfaces | Control maximum gap \( h_{g,\max} \) | Eliminate severe leakage paths |
| Optimize clamp number, strength, and position | Increase clamping force and distribution | Prevent upper mold lifting |
| Control final pouring speed | Reduce pressure surge | Lower dynamic load on seal |
| Use sand sealing strip or refractory paste | Fill small gaps and improve seal | Prevent metal escape |
After applying these measures, the bleed casting defect was also controlled. The improvements were consistent with the physical models. The most important factor was the elimination of local parting line gaps caused by poor closing alignment. The second most important factor was reliable clamping. The third was controlled pouring at the end of the fill.
7. Integrated Process Control and Verification
I integrated the sand drop and bleed countermeasures into a single process control plan. The plan included pre-closing checks, closing procedure, clamping verification, and pouring control. I used a checklist to make the plan repeatable. The checklist items are shown in Table 7.
| Stage | Check item | Acceptance criterion | Action if not met |
|---|---|---|---|
| Before closing | Lower box level | No visible tilt | Adjust support |
| Before closing | Core head position | Matches template | Reset core |
| Before closing | Parting surface cleanliness | No loose sand or debris | Clean surface |
| During closing | Upper box parallelism | \( \Delta h \leq \delta_{\text{allow}} \) | Adjust rigging |
| During closing | Lateral alignment | Box edges aligned | Reposition upper box |
| After closing | Core head condition | No damage or displacement | Open and repair |
| After closing | Parting line gap | \( h_{g,\max} \leq h_{\text{allow}} \) | Seal or reclose |
| Before pouring | Clamp force and position | \( F_{\text{clamp}} \geq k F_{\text{lift}} \) | Add or tighten clamps |
| During pouring | Final pouring speed | Controlled reduction | Adjust pouring rate |
I verified the results by tracking subsequent production. The intermittent nature of the casting defect required a sufficient number of castings to confirm improvement. I compared the period before the corrective actions with the period after the corrective actions. Table 8 presents the before-and-after comparison.
| Period | Castings produced | Sand drop rejects | Bleed rejects | Core eccentricity rejects | Total rejects | Rejection rate |
|---|---|---|---|---|---|---|
| Before corrective actions | 47 | 6 | 3 | 1 | 10 | 21.3% |
| After corrective actions | 30 | 0 | 0 | 1 | 1 | 3.3% |
The after-action results showed that the two dominant casting defects were eliminated in the monitored batch. The remaining core eccentricity casting defect was not related to the sand drop or bleed mechanisms. I continued to monitor core setting accuracy separately.
I also calculated the defect reduction using the defect rate before and after:
$$ \Delta P_{\text{reject}} = P_{\text{reject,before}} – P_{\text{reject,after}} $$
$$ \Delta P_{\text{reject}} = 21.3\% – 3.3\% = 18.0\% $$
The relative reduction in rejection rate was:
$$ R_{\text{reduction}} = \frac{P_{\text{reject,before}} – P_{\text{reject,after}}}{P_{\text{reject,before}}} \times 100\% $$
$$ R_{\text{reduction}} = \frac{21.3\% – 3.3\%}{21.3\%} \times 100\% \approx 84.5\% $$
This was a significant improvement. The iron loss avoided can be estimated from the single-piece pouring weight and the reduction in rejected castings. If the after-action period had continued at the before-action rejection rate, the expected number of rejects would have been:
$$ N_{\text{expected}} = 30 \times 21.3\% \approx 6.4 $$
The actual number of rejects was 1. Therefore, approximately 5.4 reject castings were avoided. The iron loss avoided was approximately:
$$ M_{\text{avoided}} = 5.4 \times 1439 \, \text{kg} \approx 7770 \, \text{kg} $$
This estimate showed the economic benefit of the casting defect reduction. The improvement also reduced the safety risk associated with bleed during pouring.
8. Discussion of Root Causes and Process Control Principles
The analysis of this 250t oil cylinder production showed that the dominant casting defects were not caused by a fundamental process design error. The first five trial castings and the first fifteen production castings were acceptable. This strongly indicated that the tooling, sand system, and basic process were capable of producing sound castings. The casting defects appeared later and intermittently. This pattern is typical of operational casting defects caused by variation in human execution.
I concluded that the sand drop casting defect was primarily caused by core head damage during mold closing. The core head area near the four small holes was vulnerable. When the upper box was not parallel or not aligned, it could strike the core head. The kinetic energy of the closing motion broke the sand core. The fallen sand created a protrusion and a sand inclusion. The corrective actions focused on reducing impact energy and improving alignment.
The bleed casting defect was primarily caused by molten metal escaping through a parting line gap under high static pressure. The gap could be created by poor sealing, warped parting surfaces, insufficient clamping, or box wedging caused by core collision. The leakage equation showed that the flow rate is proportional to the cube of the gap height. This means that even a very small gap can cause a serious bleed casting defect. The corrective actions focused on reducing the gap height, increasing clamping force, and controlling the final pouring speed.
Both casting defects shared a common process control principle: mold closing must be accurate and repeatable. When the upper and lower boxes are properly aligned and parallel, the core heads are protected, and the parting line is closed. When the boxes are misaligned, both sand drop and bleed become more likely. Therefore, I treated mold closing as a critical control point.
I also found that visual inspection alone is not sufficient. Some parting line gaps are very small and difficult to see, but they can still cause bleed when the metal pressure is high. Similarly, some core head damage is hidden inside the mold after closing. Therefore, I introduced verification steps such as trial closing, template checks, and gap measurement. These steps made the invisible visible before pouring.
Table 9 presents a summary of the root causes and countermeasures for the two main casting defects.
| Casting defect | Primary root cause | Secondary root cause | Key countermeasure | Verification method |
|---|---|---|---|---|
| Sand drop | Core head damage during mold closing | Inaccurate core head location | Parallel closing, alignment, core head protection | Trial closing, template check |
| Bleed or run-out | Parting line gap under high metal pressure | Insufficient clamping or poor sealing | Gap control, clamping force, final pouring control | Gap measurement, clamp check, pouring record |
The broader lesson is that casting defect analysis must connect general causes to specific production steps. General lists of possible causes are useful, but they do not solve the problem by themselves. The analyst must evaluate each possible cause against the actual production evidence. In this case, the early acceptable castings excluded many design causes. The intermittent appearance pointed to operational variation. The defect location pointed to specific mechanisms. This logical elimination process made the corrective actions targeted and effective.
9. Economic and Safety Effects of Casting Defect Reduction
The economic effect of reducing the two dominant casting defects was substantial. The total rejection rate fell from 21.3% to 3.3% in the monitored period. The sand drop and bleed casting defects were eliminated. The iron loss avoided was approximately 7.77 t based on the estimate in the previous section. This does not include the additional costs of rework, labor, energy, and delayed delivery. The safety effect was also important. A bleed casting defect can cause a sudden release of molten metal, which poses a serious risk to operators. By eliminating the parting line leakage, I reduced this risk.
I calculated the defect opportunity cost avoided as:
$$ C_{\text{avoided}} = N_{\text{avoided}} \times C_{\text{unit}} $$
where \( N_{\text{avoided}} \) is the number of avoided reject castings, and \( C_{\text{unit}} \) is the unit loss per rejected casting. Even without assigning a monetary value to \( C_{\text{unit}} \), the reduction in iron loss alone demonstrated the benefit. The process improvement also increased confidence in the manual molding line.
Table 10 presents the economic and safety effects in a structured format.
| Effect category | Before improvement | After improvement | Change |
|---|---|---|---|
| Rejection rate | 21.3% | 3.3% | Reduced by 18.0 percentage points |
| Sand drop rejects | 6 per 47 castings | 0 per 30 castings | Eliminated in monitored batch |
| Bleed rejects | 3 per 47 castings | 0 per 30 castings | Eliminated in monitored batch |
| Estimated iron loss avoided | Not applicable | About 7.77 t | Significant reduction |
| Safety risk | High during bleed | Low | Improved operator safety |
10. Standardization and Sustaining the Improvement
To sustain the improvement, I standardized the corrective actions into the work instructions. The key elements were:
1. Mold closing must be performed with a level lower box and a parallel upper box. The closing speed must be reduced at final contact.
2. At least two operators must participate in mold closing. One controls the crane, and one observes alignment and core head condition.
3. The four small core heads must be verified with a template or trial closing when any doubt exists.
4. The parting surface must be clean and free of loose sand before closing.
5. The clamping force must be sufficient and evenly distributed. The number and position of clamps must follow the work instruction.
6. The final pouring stage must be controlled to avoid a pressure surge.
7. Any bleed or sand drop occurrence must be recorded with the specific process conditions, so that the root cause can be traced if the casting defect reappears.
I also proposed a periodic audit of the parting surface flatness and the core print dimensions. The audit can use the following acceptance criteria:
$$ \Delta h \leq \delta_{\text{allow}} $$
$$ h_{g,\max} \leq h_{\text{allow}} $$
$$ F_{\text{clamp}} \geq k F_{\text{lift}} $$
These criteria connect the dimensional checks to the physical mechanisms. They are simple enough to be used on the shop floor and rigorous enough to prevent the two major casting defects.
Table 11 presents a standardized control plan for the two casting defects.
| Control point | Parameter or check | Standard | Frequency | Responsibility |
|---|---|---|---|---|
| Core setting | Core head position | Template match | Every mold | Molding operator |
| Mold closing | Upper box parallelism | \( \Delta h \leq \delta_{\text{allow}} \) | Every mold | Closing operator |
| Mold closing | Lateral alignment | Box edges aligned | Every mold | Closing operator |
| After closing | Parting line gap | \( h_{g,\max} \leq h_{\text{allow}} \) | Every mold | Closing operator |
| Clamping | Clamp force and position | \( F_{\text{clamp}} \geq k F_{\text{lift}} \) | Every mold | Closing operator |
| Pouring | Final pouring speed | Controlled reduction | Every pour | Pouring operator |
| Maintenance | Parting surface flatness | Below allowable gap | Weekly | Maintenance team |
| Maintenance | Core print dimension | Drawing tolerance | Monthly | Tooling team |
11. Additional Observations on Casting Defect Analysis
I made several additional observations during this study. First, casting defect analysis should always begin with accurate data. The total number of castings, the number of rejects, the defect type, and the defect location must be recorded. Without this data, it is easy to focus on the wrong casting defect. In this case, the data showed that sand drop and bleed accounted for 90% of the rejects. This allowed the improvement effort to be concentrated.
Second, casting defect mechanisms should be described quantitatively whenever possible. The kinetic energy formula for sand drop, the static pressure formula for bleed, the leakage flow formula, and the clamping force inequality all helped explain why the casting defects occurred. These formulas also helped prioritize the corrective actions. For example, the leakage flow formula showed that reducing the gap height is much more effective than reducing the gap width. This justified the emphasis on parting surface parallelism and sealing.
Third, human factors must be taken seriously in manual casting production. The casting defects in this study were intermittent. They did not occur in every mold. This is a classic sign of process variation caused by human execution. Training, checklists, and verification steps are therefore essential. A robust process design cannot compensate for inconsistent operation in a manual line.
Fourth, the two dominant casting defects were connected. Both were influenced by mold closing accuracy. This means that improving mold closing addressed both casting defects at the same time. This is an important principle in casting defect reduction: look for shared root causes. When multiple casting defects share a common cause, a single improvement can produce a large overall effect.
Fifth, the corrective actions should be measurable. I used the rejection rate, the number of sand drop rejects, the number of bleed rejects, the parallelism deviation, the maximum gap height, and the clamping force ratio as measurable indicators. These indicators allowed the team to verify that the process was under control. If any indicator deviates from the standard, corrective action can be taken before a casting defect occurs.
12. Conclusion
I analyzed the casting defect problem in the production of a 250t oil cylinder for a rubber injection machine. The casting is made of QT500-7, weighs 1439 kg per piece, and is produced by resin sand molding and core making in a manual line. Over four months, 47 castings were produced, 10 were rejected, and the rejection rate was 21.3%. Sand drop accounted for 60% of the rejected castings, and bleed or run-out accounted for 30%. Together, these two casting defects accounted for 90% of the rejects.
I identified the sand drop casting defect as a mechanical damage phenomenon during mold closing. The core head near the four small holes was vulnerable to impact when the upper and lower boxes were not parallel or not aligned. The fallen sand created a metal protrusion and sand inclusion. I implemented measures to reduce closing impact energy, improve parallelism, ensure lateral alignment, verify core head position, and consider a guide pin and bushing system. These measures eliminated the sand drop casting defect in the monitored production.
I identified the bleed casting defect as a pressure-driven leakage phenomenon at the parting line. Near the end of pouring, the static pressure of the molten metal reached its maximum. If a parting line gap existed, the metal flowed out according to the parallel-plate flow relationship. The leakage rate was proportional to the cube of the gap height, so even a small gap was dangerous. Insufficient clamping could also lift the upper mold. I implemented measures to improve sealing, avoid core collision, maintain parting surfaces, optimize clamping, and control the final pouring speed. These measures eliminated the bleed casting defect in the monitored production.
The combined corrective actions reduced the rejection rate from 21.3% to 3.3% in the monitored period. The estimated iron loss avoided was about 7.77 t. The safety risk associated with bleed was also reduced. The study showed that targeted casting defect analysis, supported by quantitative models and standardized work instructions, can solve stubborn casting defect problems in manual casting production. The approach can be used as a reference for similar heavy-section castings and for other production lines facing sand drop and bleed casting defects.
In my future work, I will continue to monitor the process, audit the critical control points, and update the standards as needed. Casting defect prevention is not a one-time activity. It requires continuous attention to detail, data collection, and process discipline. The experience from this 250t oil cylinder project confirmed that even a high rejection rate caused by severe casting defects can be brought under control when the root causes are correctly identified and the countermeasures are consistently applied.