Development context. I treat this work as a focused study of large ductile iron castings for a heavy-duty diesel engine bearing cover. The part is not a thin, simple bracket. It is a thick, heavy, fatigue-loaded component that supports the crankshaft and receives cyclic alternating loads during engine operation. Because the bearing cover locates and supports the crankshaft, any internal shrinkage porosity, slag inclusion, or sand-related defect near a main bolt hole can become a fatigue initiation site. For ductile iron castings of this size, the process window is narrow: too much riser volume increases yield loss and thermal damage, while too little feeding or chilling allows shrinkage porosity to form in the thick section. I therefore approached the problem as a combined gating, filtration, core, chilling, chemistry, and solidification-control task.
Component and quality requirements. The bearing cover envelope is approximately 700 mm × 450 mm × 150 mm. The rough casting weighs about 190 kg. The material is a ferritic ductile iron grade comparable to QT400-15, with high elongation and good toughness. The maximum local thickness is about 150 mm. The main bolt holes are produced later by machining. The most important acceptance rule is that no defect is permitted within 10 mm of the bolt-hole region after machining. This requirement drives the entire process design because the bolt-hole region is also the thickest region and therefore the natural last-solidifying zone in many ductile iron castings.
| Item | Value or requirement | Process implication |
|---|---|---|
| Part type | Large diesel engine bearing cover | High fatigue duty and dimensional stability required |
| Envelope size | 700 mm × 450 mm × 150 mm | Large thermal mass and long feeding distances |
| Rough casting mass | Approximately 190 kg | High pouring load and significant metallostatic pressure |
| Material | Ferritic ductile iron, QT400-15 type | Graphite expansion can assist feeding if chemistry is controlled |
| Maximum thickness | 150 mm | High modulus and slow cooling in the thick section |
| Critical zone | Within 10 mm of machined bolt holes | No porosity, slag, or crack-like defects allowed |
| Production target | Stable sound castings with improved yield | Robust process rather than selective inspection |
Initial process concept. My first trial followed conventional experience for thick ductile iron castings. The bolt holes were not cast. The thick upper region was fed by an insulated riser, and the corresponding lower region was cooled with chills. I used two ceramic filters to improve slag removal. The intention was reasonable: combine riser feeding and chilling to reduce shrinkage porosity, filter the metal to reduce slag defects, and leave the bolt holes solid so that machining would not encounter a cast hole misalignment or core-related defect. However, this combination also increased the local hot spot because the solid bolt-hole region remained as a thick mass. The riser could not fully compensate the shrinkage demand, and the casting yield was low. The initial process did reduce risk compared with an unfed thick section, but it did not eliminate shrinkage porosity or slag inclusion in ductile iron castings of this geometry.
| Initial process element | Design choice | Expected benefit | Observed limitation |
|---|---|---|---|
| Bolt holes | Not cast; formed by machining | Avoids core shift and black skin in holes | Creates a large solid hot spot |
| Feeding | Insulated riser on thick upper section | Provides liquid feed during solidification | Feeding distance too long for full soundness |
| Chilling | External chills under thick section | Increases local cooling rate | Incomplete coverage of the thermal center |
| Filtration | Two vertical ceramic filters | Traps slag and improves flow | Poor initial filtration before stable filter cake forms |
| Yield | Riser-heavy layout | Lower risk in first trials | Low casting yield |
Solidification analysis of the original design. For any ductile iron castings, the modulus method is a practical first approximation. The thermal modulus is the ratio of volume to cooling surface area:
$$M = \frac{V}{A}$$
where \(M\) is the modulus, \(V\) is the volume, and \(A\) is the effective heat-transfer surface area. A thick plate-like section cooling from both sides has an approximate modulus equal to half its thickness. If the local thick region is treated as a 150 mm plate, then \(M \approx 75\) mm before considering end effects and adjacent mass. This value is large. The solidification time can be estimated by Chvorinov’s rule:
$$t_s = B \left(\frac{V}{A}\right)^n = B M^n$$
where \(t_s\) is solidification time, \(B\) is a mold constant, and \(n\) is usually close to 2 for many castings. Because solidification time increases with the square of modulus, the thick bolt-hole region remains liquid much longer than the surrounding walls. A riser must remain liquid longer than the casting hot spot and must have sufficient feeding channel. A common modulus criterion is:
$$M_r = k M_c$$
where \(M_r\) is the riser modulus, \(M_c\) is the casting modulus at the hot spot, and \(k\) is a safety factor. For ductile iron castings, graphite expansion can reduce the required \(k\), but it cannot compensate for an isolated, massive hot spot surrounded by cooler sections. In my initial design, the uncast bolt holes made \(M_c\) too large and the feeding path too long. The observed shrinkage porosity was therefore predictable.
| Parameter | Approximate value | Meaning |
|---|---|---|
| Maximum section thickness | 150 mm | Main thermal center |
| Plate modulus estimate | 75 mm | Very high local modulus |
| Solidification scaling | \(t_s \propto M^2\) | Small modulus changes strongly affect cooling |
| Riser criterion | \(M_r = k M_c\) | Riser must remain active longer than hot spot |
| Original risk | High | Long feeding distance and large hot spot |
Observed defects. After cleaning, I inspected the casting surface and found inclusion-type defects. I used electron microscopy and energy-dispersive spectroscopy to identify the local chemistry of the defect. The analysis showed significant oxygen, calcium, aluminum, silicon, sulfur, manganese, iron, and zirconium. The presence of calcium, aluminum, and oxygen indicated a slag or oxide inclusion, while sulfur and manganese suggested that sulfide and oxysulfide phases may also have been present. The surface defect was not a simple sand burn. It had the characteristic chemistry of a slag-related inclusion in ductile iron castings.
| Element | Weight percent | Atomic percent | Possible source or phase |
|---|---|---|---|
| O | 49.17 | 71.32 | Oxide, slag, surface oxidation |
| Na | 1.34 | 1.35 | Minor contaminant |
| Mg | 0.70 | 0.67 | Residual magnesium from nodularization |
| Al | 3.34 | 2.87 | Oxide, refractory, or alloy contamination |
| Si | 11.03 | 9.11 | Silicate slag or matrix contribution |
| S | 0.79 | 0.57 | Sulfide inclusion |
| Ca | 11.26 | 6.52 | Slag former or refractory reaction product |
| Mn | 5.05 | 2.13 | Sulfide or matrix segregation |
| Fe | 6.52 | 2.71 | Matrix contribution |
| Zr | 10.82 | 2.75 | Refractory, coating, or filter-related material |
Internal shrinkage porosity. After sectioning the casting, I found dispersed cavity-like defects in the bolt-hole region. These were not surface slag. Their morphology and location indicated shrinkage porosity. In ductile iron castings, shrinkage porosity often forms in the last-solidifying liquid pockets when liquid feeding, graphite expansion, and mold dilation are not balanced. The bolt-hole area had the highest modulus, so it solidified last. The riser was above it, but the feeding path was not sufficient to maintain liquid metal all the way into the thermal center. The result was a distributed porosity field that could violate the 10 mm bolt-hole exclusion zone after machining.
| Defect type | Location | Character | Main cause |
|---|---|---|---|
| Slag inclusion | Surface and near-surface | Oxide and calcium-rich phases | Insufficient filtration and slag control |
| Shrinkage porosity | Bolt-hole thermal center | Dispersed cavities | Large hot spot and limited feeding |
| Sand-related risk | Core region in later design | Potential deformation or breakage | Core strength and support |
| Machining risk | Near bolt holes | Exposed porosity or inclusion | Defect too close to finished surface |
Root cause of the filtration problem. The original gating system was compact. The distance between the sprue and filter was short, and the ceramic filter was placed vertically. A ceramic filter works through mechanical interception, flow straightening, and filter-cake formation. A stable filter cake takes time to develop. During the initial pouring stage, the filter efficiency is lower because the cake has not yet formed and the flow through the filter is not fully uniform. A vertical filter can also allow metal to pass through only part of its area, especially when the flow impinges directly on the filter surface. If the sprue is close to the filter, the high-velocity jet may carry slag and oxide through the filter before the filtration layer becomes effective. For large ductile iron castings, I need a gating system that presents the metal to the filter in a controlled, low-turbulence manner and then allows the filtered metal to fill the mold from the bottom upward.
The residence time in the runner and filter can be approximated by:
$$t_r = \frac{L_r A_r}{Q}$$
where \(t_r\) is residence time, \(L_r\) is the effective runner length, \(A_r\) is the runner cross-sectional area, and \(Q\) is the volumetric flow rate. A short runner reduces \(t_r\), giving less time for slag to float or be trapped. The filter area can be estimated by:
$$A_f = \frac{Q}{v_f}$$
where \(A_f\) is the active filter area and \(v_f\) is the permissible superficial velocity through the filter. If the vertical filter is only partially active, the effective \(A_f\) is lower than the geometric area, so the local velocity is higher and filtration is less stable. This explains why the vertical filter did not eliminate the slag defects in the early ductile iron castings.
| Filtration factor | Original design | Effect |
|---|---|---|
| Filter orientation | Vertical | Partial flow and unstable cake formation |
| Sprue-filter distance | Short | High-velocity jet reaches filter early |
| Runner length | Compact | Less flotation and slag separation time |
| Initial pouring stage | Poor filtration | Slag passes before filter cake develops |
| Result | Surface inclusions | Oxide and calcium-rich defects |
Root cause of the shrinkage problem. The uncast bolt holes created a very large solid mass. Although this avoided core-related hole defects, it also created a thermal center that could not be fed from the riser within the available distance. Feeding distance in ductile iron castings depends on the modulus, carbon equivalent, temperature gradient, and graphite expansion. A simplified form is:
$$L_f = c M^{m}$$
where \(L_f\) is the effective feeding distance, \(c\) is an empirical coefficient, and \(m\) is an exponent, often between 0.5 and 1.0 depending on geometry and gradient. Increasing the local modulus \(M\) increases the solidification time, but it also increases the required feeding distance. The riser may remain liquid long enough, but the narrow feeding path can freeze before the last liquid pocket is compensated. In the original design, the riser plus chill combination did not cover the entire thermal center. The result was a mixed condition: some regions were chilled, some were fed, and the center remained isolated. I concluded that the process had to reduce the hot spot itself, not merely attempt to feed it.
| Shrinkage factor | Original condition | Consequence |
|---|---|---|
| Bolt-hole geometry | Solid, not cast | Large isolated thermal center |
| Local modulus | Very high | Late solidification and high shrinkage demand |
| Riser | Insulated top riser | Limited feeding distance |
| Chills | Lower side only | Incomplete thermal coverage |
| Graphite expansion | Not fully utilized | Self-feeding insufficient |
Improved process strategy. I changed the process in three main areas: filtration and gating, casting structure, and cooling. The goal was not simply to add more metal or more chills. The goal was to make the solidification sequence more uniform, reduce the hot spot, filter the metal more effectively, and use graphite expansion as a controlled feeding mechanism. This is the core principle I now apply to large ductile iron castings.
Gating and filtration redesign. I moved the sprue to increase the distance to the filter and lengthened the runner. I changed the filter from vertical to horizontal. With a horizontal filter, the metal passes from bottom to top through the filter. This has several advantages. First, the metal fills the runner more evenly before reaching the filter. Second, the filter is wetted over a larger area. Third, the filter cake forms more uniformly. Fourth, gravity helps separate slag and oxide before the metal enters the mold cavity. Fifth, the filtered metal enters the mold with less turbulence. For ductile iron castings, lower turbulence also reduces oxidation and gas entrainment.
| Gating parameter | Original design | Improved design | Benefit |
|---|---|---|---|
| Sprue location | Close to filter | Moved to increase distance | Lower jet velocity at filter |
| Runner length | Short | Longer | More time for slag flotation |
| Filter orientation | Vertical | Horizontal | Full-area flow and stable cake |
| Flow direction through filter | Lateral | Bottom to top | Improved slag interception |
| Mold filling | More turbulent | More stable and bottom-up | Reduced oxide entrainment |
Filter sizing and pouring control. I sized the filter using the estimated pouring rate. If the casting mass is \(W\) and the target pouring time is \(t_p\), then the average flow rate is:
$$Q = \frac{W}{\rho t_p}$$
where \(\rho\) is the density of liquid ductile iron. The filter area must satisfy:
$$A_f \ge \frac{Q}{v_f}$$
I selected a filter grade and area that kept the superficial velocity within the recommended range while providing enough capacity for the total poured mass. I also controlled the pouring temperature and pouring time. Too low a pouring temperature increases the risk of cold shut and filter blockage. Too high a pouring temperature increases shrinkage and slag formation. For large ductile iron castings, I aim for a stable pouring temperature window and a smooth, non-interrupted pour.
| Parameter | Target range | Reason |
|---|---|---|
| Pouring temperature | Controlled window for thick ductile iron castings | Balance fluidity, shrinkage, and slag formation |
| Pouring time | Long enough for stable flow | Avoid filter erosion and mold erosion |
| Filter velocity | Below recommended maximum | Maintain filtration efficiency |
| Filter capacity | Above total poured mass | Prevent late-stage blockage |
| Runner design | Pressurized or controlled flow | Reduce turbulence and air entrainment |
Casting structure optimization. I changed the bolt holes from a solid, machined-only feature to cast holes formed by sand cores. This was the most important change for shrinkage control. By casting the holes, I removed a large volume of metal from the thermal center. The local modulus dropped significantly. A solid section of thickness \(t\) has a modulus approximately \(t/2\) when cooled from two sides. If the bolt hole reduces the local wall thickness to \(t_w\), the modulus becomes approximately \(t_w/2\). For example, reducing an effective 150 mm solid mass to a 40–60 mm wall changes the modulus from about 75 mm to about 20–30 mm. Because solidification time scales with \(M^2\), this change greatly shortens the local solidification time and reduces the feeding demand.
| Feature | Original solid design | Improved cored design | Effect on solidification |
|---|---|---|---|
| Bolt-hole region | Solid metal | Sand core forms hole | Removes thermal mass |
| Approximate local modulus | High, about 75 mm | Lower, about 20–30 mm | Much shorter solidification time |
| Hot spot size | Large and isolated | Reduced and more distributed | Lower shrinkage risk |
| Feeding distance | Long | Shorter | Riser and chill action more effective |
| Machining allowance | Solid stock | Controlled core allowance | Less machining stock and lower cost |
Core strength and dimensional control. Casting the bolt holes introduced a new risk: core breakage, core shift, or core deformation. I placed a steel core rod inside the bolt-hole sand core to increase its stiffness. The core rod prevented the sand core from bending under the metallostatic pressure and from breaking during handling and pouring. I also controlled core venting, coating, and location. A deformed core would create an eccentric bolt hole and could reduce the wall thickness around the hole, which would be unacceptable in a fatigue-critical bearing cover. For ductile iron castings with cored holes, the core system must be designed with the same care as the gating system.
| Core design factor | Control method | Purpose |
|---|---|---|
| Core strength | Steel core rod | Resist bending and breakage |
| Core location | Robust core prints | Prevent shift and eccentricity |
| Core coating | Controlled refractory coating | Improve surface finish and reduce sand burn |
| Core venting | Proper vent channels | Reduce gas defects |
| Core removal | Good collapsibility | Avoid cracking during shakeout |
Cooling system optimization. I replaced the original riser-heavy approach with conformal chills placed directly on the thick hot-spot regions. The chills were shaped to follow the casting contour so that the thermal coverage was complete. The purpose of the chills was to increase the cooling rate at the hot spot, promote directional solidification, and allow graphite expansion to compensate the remaining liquid shrinkage. In ductile iron castings, the graphite expansion can be beneficial only if the casting has enough stiffness and if the thermal gradient is controlled. If the hot spot is isolated, the expansion cannot feed the center. If the cooling is too uneven, it can create internal stress or cold shuts. I therefore designed the chill layout to balance cooling rather than simply maximize it.
The cooling rate can be approximated as:
$$R = \frac{T_p – T_m}{t_s}$$
where \(R\) is the average cooling rate, \(T_p\) is the pouring temperature, \(T_m\) is the mold temperature near the interface, and \(t_s\) is the local solidification time. Increasing the heat extraction with a chill reduces \(t_s\), which reduces the size of the liquid pocket. The chill modulus can be expressed as:
$$M_{ch} = \frac{V_{ch}}{A_{ch}}$$
where \(V_{ch}\) and \(A_{ch}\) are the chill volume and effective contact area. For a given casting modulus, the chill must have enough thermal mass to absorb heat without becoming saturated too early. I selected the chill thickness, contact area, and coating to achieve the required cooling without causing premature freezing of the feeding channel.
| Cooling parameter | Original approach | Improved approach | Result |
|---|---|---|---|
| Hot-spot coverage | Partial | Full conformal coverage | More uniform solidification |
| Riser dependence | High | Reduced | Higher yield |
| Cooling rate | Moderate | Controlled and local | Smaller liquid pocket |
| Graphite expansion | Underused | Better utilized | Improved self-feeding |
| Residual stress risk | Moderate | Controlled by balanced layout | Lower distortion risk |
Graphite expansion and self-feeding. Ductile iron castings differ from many other cast irons because the graphite grows during solidification and can partially compensate for liquid shrinkage. The volume change associated with graphite formation can be represented conceptually as:
$$\Delta V_g = V_l \left( \frac{\rho_l}{\rho_s} – 1 \right) + V_g \epsilon_g$$
where \(\Delta V_g\) is the volume change, \(V_l\) is the liquid volume, \(\rho_l\) and \(\rho_s\) are liquid and solid densities, \(V_g\) is the graphite volume, and \(\epsilon_g\) is the expansion strain associated with graphite growth. A high carbon equivalent promotes graphite expansion, but too high a carbon equivalent can cause graphite flotation in very thick sections. I therefore adjusted the carbon and silicon content to a narrow range that supports expansion without excessive flotation or nodule degeneration.
The carbon equivalent is commonly estimated as:
$$CE = C + \frac{Si}{3} + \frac{P}{3}$$
I targeted a carbon equivalent high enough to promote graphite expansion and good fluidity, but not so high that graphite flotation or excessive dross would occur in the thick bearing cover. I also controlled magnesium and rare-earth additions to achieve good nodularity without creating excessive slag or dross. For large ductile iron castings, the magnesium content must be sufficient for nodularization but not so high that it increases shrinkage and inclusion risk.
| Element | Target range | Role in ductile iron castings |
|---|---|---|
| C | Controlled high range | Promotes graphite expansion and fluidity |
| Si | Moderate range | Controls matrix and carbon equivalent |
| Mn | Low | Reduces carbide formation and segregation |
| P | Very low | Avoids phosphorus eutectic and embrittlement |
| S | Very low | Necessary for magnesium treatment control |
| Mg | Narrow residual range | Nodularization; excess increases dross |
| RE | Low controlled range | Supports nodularization and counters impurities |
Chemistry and mechanical property targets. The final material must satisfy the QT400-15 type requirement: tensile strength around 400 MPa and elongation around 15%. For thick ductile iron castings, the cooling rate is slower than in thin sections, so the matrix can become coarser and nodule count can decrease. I controlled inoculation and magnesium treatment to maintain nodularity and nodule count. I also kept manganese and phosphorus low to preserve ductility. The combination of high carbon equivalent, low manganese, and controlled magnesium gave me a material that could exploit graphite expansion while still meeting the mechanical property requirements.
| Property or feature | Target | Reason |
|---|---|---|
| Tensile strength | Approximately 400 MPa | Meets grade requirement |
| Elongation | Approximately 15% | Fatigue and toughness requirement |
| Nodularity | High | Ensure ductile iron behavior |
| Nodule count | Sufficient for thick section | Reduce segregation and shrinkage tendency |
| Carbides | Low or absent | Preserve machinability and elongation |
| Graphite flotation | Avoided | Maintain uniform properties |
Simulation of filling and solidification. I used simulation to compare the original and improved designs. The filling simulation showed that the improved gating system filled the mold in a stable, bottom-up manner. The metal front advanced evenly, and the filter was fully wetted before the main flow entered the cavity. The original design showed more local turbulence and early flow through the filter before a stable cake could form. The solidification simulation showed that the bolt-hole region in the improved design had a much lower porosity tendency. The hot spot was smaller and located farther from the critical bolt-hole surface. The final solidification occurred in a region that could be fed or compensated by graphite expansion, rather than in an isolated central mass.
A useful local criterion for shrinkage prediction is the Niyama criterion:
$$N_y = \frac{G}{\sqrt{\dot{T}}}$$
where \(G\) is the temperature gradient and \(\dot{T}\) is the cooling rate. Low \(N_y\) values indicate a high probability of shrinkage porosity. In the original design, the bolt-hole region had low gradient and low cooling rate, producing low \(N_y\). In the improved design, the cored holes reduced the local modulus, the chills increased the cooling rate, and the gating change improved temperature uniformity. The simulated \(N_y\) field therefore shifted away from the critical zone.
| Simulation item | Original design | Improved design |
|---|---|---|
| Filling stability | Local turbulence | Stable and bottom-up |
| Filter wetting | Partial and early | Full and progressive |
| Hot-spot size | Large | Reduced |
| Shrinkage tendency near bolt holes | High | Low |
| Solidification sequence | Isolated center | More directional |
| Porosity risk | High | Acceptable |
Simulation results in more detail. The filling simulation reached a stable filling state at approximately 70% of the mold volume after about 18.2 seconds. The final solidification was completed after approximately 4 hours 34 minutes 45 seconds. The porosity field showed that the highest-risk region in the original design was concentrated near the bolt-hole thermal center. In the improved design, the high-risk region moved away from the bolt-hole surface. The maximum porosity value in the improved simulation was lower and was located in a non-critical region. I used these results to set the final chill layout, core design, and gating dimensions before cutting the trial castings.
| Simulation parameter | Result | Interpretation |
|---|---|---|
| Stable filling progress | Approximately 70% at 18.2 s | Controlled mold filling |
| Final solidification time | Approximately 4 h 34 min 45 s | Long cooling cycle of thick ductile iron castings |
| Hot-spot location | Shifted away from bolt-hole surface | Critical zone protected |
| Porosity tendency | Reduced near bolt holes | Improved feeding and cooling |
| Process window | Wider than original | Better production stability |
Validation by casting inspection. After pouring the improved design, I inspected the casting surface. I found no obvious slag inclusion defects. I then sectioned a validation casting to check the internal quality. The shrinkage porosity, where present, was located more than 76 mm away from the bolt-hole region. This distance was sufficient to keep the critical 10 mm zone free from defects after machining. The sectioned result confirmed that the revised process had moved the last-solidifying region away from the bolt-hole surface and had reduced the porosity level to an acceptable condition.
| Inspection item | Original result | Improved result | Acceptance status |
|---|---|---|---|
| Surface slag inclusion | Present | Not obvious | Accepted |
| Shrinkage porosity location | Near bolt-hole center | More than 76 mm from bolt hole | Accepted |
| Critical 10 mm zone | At risk | Protected | Accepted |
| Core-related defects | Not applicable in original | No breakage or shift observed | Accepted |
| Dimensional quality | Machining stock high | Controlled core allowance | Accepted |
Small-batch production validation. I validated the improved process with a small batch of 20 castings. All 20 castings passed surface inspection and sectioning checks. No shrinkage porosity or slag inclusion was found in the critical bolt-hole region. The casting process yield increased from 56% in the original design to 82% in the improved design. This yield improvement came from reducing the riser size, casting the bolt holes, using conformal chills, and controlling the gating system. The result was not only a quality improvement but also a significant reduction in metal consumption and machining cost.
| Production metric | Original process | Improved process | Change |
|---|---|---|---|
| Validation batch size | Not applicable | 20 castings | Small-batch verification |
| Slag inclusion defects | Observed | Not found | Eliminated |
| Shrinkage porosity near bolt holes | Observed | Not found in critical zone | Eliminated from critical zone |
| Casting process yield | 56% | 82% | +26 percentage points |
| Machining risk | High | Low | Improved |
| Process stability | Marginal | Stable | Improved |
Why the improved design works. The improved design works because it addresses the two competing defects at their source. Slag inclusion is controlled by lengthening the runner, moving the sprue, and using a horizontal filter so that the metal passes through the filter from bottom to top. This gives the filter cake time to form and allows the filter to operate over its full area. Shrinkage porosity is controlled by removing the solid bolt-hole mass, reducing the local modulus, adding conformal chills, and adjusting the chemistry so that graphite expansion can assist feeding. The casting no longer has a single isolated hot spot that must be fed from a distant riser. Instead, the solidification sequence is more distributed and more directional. For large ductile iron castings, this combination is more effective than simply increasing riser size.
Process control plan for production. I established a production control plan covering melting, nodularization, inoculation, pouring, core making, mold assembly, and inspection. The most important variables are carbon equivalent, magnesium residual, pouring temperature, pouring time, filter condition, core rod placement, chill contact, and mold hardness. I also specified that the critical bolt-hole zone must be inspected after machining. If any indication appears in that zone, the casting is rejected. The control plan is not based on a single trial; it is based on the validated process window from the simulation and small-batch production.
| Process stage | Key variable | Control target | Risk if uncontrolled |
|---|---|---|---|
| Melting | Carbon equivalent | Narrow high-CE range | Shrinkage or graphite flotation |
| Nodularization | Mg residual | Controlled range | Poor nodularity or dross |
| Inoculation | Inoculant amount and timing | Consistent late inoculation | Carbides and low nodule count |
| Pouring | Temperature and time | Stable window | Cold shut, slag, or shrinkage |
| Filtration | Filter orientation and condition | Horizontal, full-area flow | Slag inclusion |
| Core making | Core rod and venting | Correct placement | Core shift, breakage, gas defects |
| Chilling | Contact and coverage | Full conformal coverage | Shrinkage porosity |
| Inspection | Critical zone | No defect within 10 mm | Fatigue risk |
Thermal and modulus comparison. The original design had a high modulus because the bolt holes were solid and the riser was the main feeding method. The improved design reduced the modulus by casting the holes, and it increased the local cooling rate with chills. The relationship between modulus and solidification time is nonlinear, so the reduction in modulus is more significant than it first appears. If the modulus is reduced from approximately 75 mm to approximately 25 mm, the solidification time ratio is:
$$\frac{t_{s,new}}{t_{s,old}} \approx \left(\frac{M_{new}}{M_{old}}\right)^2 = \left(\frac{25}{75}\right)^2 = \frac{1}{9}$$
This means the local solidification time can be reduced by nearly an order of magnitude, depending on the actual geometry and mold conditions. A shorter local solidification time means a smaller liquid pocket, less shrinkage demand, and a greater chance that graphite expansion can compensate the remaining liquid contraction. This is why casting the bolt holes was so effective for the large ductile iron castings.
| Parameter | Original design | Improved design | Approximate effect |
|---|---|---|---|
| Local modulus | About 75 mm | About 25 mm | Reduced to one third |
| Relative solidification time | Reference | About 1/9 | Smaller liquid pocket |
| Feeding demand | High | Lower | Less riser dependence |
| Graphite expansion benefit | Limited | Enhanced | Improved self-feeding |
| Shrinkage risk | High | Low | Critical zone protected |
Filtration efficiency comparison. The original vertical filter had a smaller effective flow area and a shorter residence time. The improved horizontal filter had a larger effective area, more stable cake formation, and a better flow direction. If the same volumetric flow rate \(Q\) passes through the filter, the superficial velocity is:
$$v_f = \frac{Q}{A_f}$$
For the vertical filter, the effective area was only part of the geometric area, so \(v_f\) was higher. For the horizontal filter, the effective area was closer to the full area, so \(v_f\) was lower. Lower velocity improves mechanical interception and reduces the chance that small slag particles are pushed through the filter. It also reduces filter erosion. This is a simple but important reason why the horizontal filter performed better in the ductile iron castings.
| Filter factor | Vertical filter | Horizontal filter | Result |
|---|---|---|---|
| Effective area | Partial | Nearly full | Lower velocity |
| Cake formation | Slow and uneven | Fast and uniform | Better filtration |
| Slag interception | Less effective | More effective | Fewer inclusions |
| Flow stability | Moderate | High | Less turbulence |
| Observed defects | Surface slag | No obvious slag | Process improved |
Economic and manufacturing benefits. The improvement in casting yield from 56% to 82% is a major economic benefit. Higher yield means less liquid metal is poured for each accepted casting, less riser removal, less energy consumption, and lower melting cost. Casting the bolt holes also reduces machining stock. The need for a large insulating riser is reduced, which simplifies mold assembly and reduces the risk of riser-related defects. The conformal chills and core rods add some process complexity, but the overall manufacturing cost is lower because the rejection rate and machining cost are reduced. For large ductile iron castings, the economic optimum is not always the simplest mold. It is the process that reliably produces a sound critical zone with the lowest total cost.
| Cost or efficiency factor | Original process | Improved process | Effect |
|---|---|---|---|
| Pouring weight per casting | High | Lower | Less melted metal |
| Riser removal | Large riser | Reduced riser | Less labor and energy |
| Machining stock | High near bolt holes | Controlled | Less machining time |
| Rejection risk | High | Low | Lower scrap cost |
| Casting yield | 56% | 82% | +26 percentage points |
| Process reproducibility | Marginal | Stable | Better planning |
Key lessons for large ductile iron castings. I have drawn several lessons from this development. First, filtration is not just a matter of installing a filter. The filter orientation, location, upstream runner length, and initial flow conditions determine whether the filter can work effectively. Second, a large solid hot spot cannot always be solved by a larger riser. Reducing the hot spot by casting holes or using cores is often more effective. Third, chills must be designed with full coverage and balanced cooling. Partial chilling can move the shrinkage defect rather than eliminate it. Fourth, chemistry must support graphite expansion, but the carbon equivalent must remain within a safe range to avoid flotation and dross. Fifth, simulation is most useful when it is combined with defect analysis and production validation. The simulation helps explain the thermal field, but the final process must be verified by sectioning and small-batch production. Sixth, for ductile iron castings, the critical zone must be protected by process design, not by inspection alone.
| Lesson | Original assumption | Improved understanding |
|---|---|---|
| Filtration | Any filter will remove slag | Orientation and flow conditions are critical |
| Feeding | Larger riser solves shrinkage | Reducing hot spot is often better |
| Chilling | Local chill is enough | Coverage and balance are essential |
| Chemistry | Higher CE is always better | CE must be controlled for thick sections |
| Simulation | Simulation alone is sufficient | Simulation plus trials and sectioning is necessary |
| Inspection | Inspection can screen defects | Process design must eliminate critical defects |
Final process sequence. In the final process, I first prepare the mold with the cored bolt holes and steel core rods. I place conformal chills in the thick regions. I assemble the gating system with the sprue at the revised location, the longer runner, and the horizontal ceramic filter. I melt and treat the ductile iron to the target carbon equivalent and magnesium residual. I inoculate the metal just before pouring. I pour within the validated temperature and time window. After solidification and shakeout, I clean the casting and inspect the critical surfaces. I then section validation castings periodically to confirm that the shrinkage porosity remains outside the bolt-hole exclusion zone. This sequence has produced sound large ductile iron castings with a yield of 82% and no observed shrinkage or slag defects in the small-batch validation.
| Step | Action | Control point |
|---|---|---|
| 1 | Prepare sand cores with steel rods | Core dimensions and rod position |
| 2 | Place conformal chills | Contact and full coverage |
| 3 | Assemble gating and horizontal filter | Runner length and filter seating |
| 4 | Melt and nodularize ductile iron | CE, Mg, S, and temperature |
| 5 | Inoculate and pour | Pouring time and temperature |
| 6 | Cool and shake out | Solidification and stress control |
| 7 | Clean and inspect | Surface, critical zone, and sectioning |
Quality assurance for the critical zone. The bolt-hole region is the most important quality zone. I use a combination of process control and verification. The process control ensures that the hot spot is reduced, the cooling is balanced, and the metal is clean. The verification includes surface inspection, dimensional check, and periodic sectioning. If a section shows porosity close to the critical zone, I adjust the chill layout, core design, or pouring temperature before continuing production. I do not rely on random inspection to catch a systematic process problem. For ductile iron castings that carry cyclic loads, the critical zone must be protected by the process itself.
| Quality activity | Frequency | Purpose |
|---|---|---|
| Chemical analysis | Every heat | Confirm carbon equivalent and residual elements |
| Pouring record | Every casting | Trace temperature and time |
| Visual inspection | Every casting | Detect surface slag or sand defects |
| Dimensional check | Sampling | Confirm core position and machining stock |
| Sectioning | Periodic validation | Verify internal soundness near bolt holes |
| Mechanical testing | Per heat or as required | Confirm grade properties |
Comparison with alternative approaches. One alternative would be to keep the solid bolt-hole design and increase the riser size further. I rejected this because it would increase yield loss, increase the thermal center, and still might not feed the isolated hot spot. Another alternative would be to add more chills without casting the holes. This would improve cooling but would not remove the large liquid volume at the center. A third alternative would be to use a different gating system without changing the filter orientation. This might reduce turbulence but would not solve the initial filtration problem. The selected approach combines all necessary changes: cast holes, conformal chills, horizontal filtration, longer runner, and chemistry control. The combination is more robust than any single change.
| Alternative | Advantage | Limitation | Decision |
|---|---|---|---|
| Larger riser only | Simple | Low yield; hot spot remains | Rejected |
| More chills only | Faster cooling | Does not remove central liquid mass | Rejected |
| Gating change only | Reduces slag | Does not solve shrinkage | Insufficient |
| Cast bolt holes only | Reduces hot spot | Requires core control | Accepted with core rods |
| Combined redesign | Addresses both defects | More process variables | Selected |
Conclusion from the first-person process development. I began with a conventional riser-and-chill process for thick ductile iron castings. The process reduced risk but did not eliminate slag inclusion or shrinkage porosity. The defect analysis showed that the surface inclusions were slag-related and the internal defects were shrinkage porosity. The root causes were an ineffective vertical filtration system and an excessively large bolt-hole hot spot. I changed the gating system to a longer runner with a horizontal filter, cast the bolt holes with reinforced sand cores, replaced the riser-heavy feeding strategy with conformal chills, and adjusted the chemistry to support graphite expansion. Simulation showed more stable filling and a lower shrinkage tendency. Production validation of 20 castings showed no shrinkage or slag defects, with the shrinkage location more than 76 mm from the bolt hole. The process yield increased from 56% to 82%. For large ductile iron castings, the most effective solution is a balanced combination of filtration, core design, chilling, chemistry, and solidification control.
| Final result | Value |
|---|---|
| Validation batch | 20 castings |
| Observed slag defects | None |
| Observed shrinkage in critical zone | None |
| Minimum distance of residual porosity from bolt hole | More than 76 mm |
| Casting yield before improvement | 56% |
| Casting yield after improvement | 82% |
| Main process changes | Horizontal filter, longer runner, cast bolt holes, core rods, conformal chills, controlled chemistry |
In my final assessment, the large bearing cover is a good example of why ductile iron castings require an integrated process design. A single change, such as a larger riser or a different filter, is rarely enough. The defect mechanisms are linked. Slag inclusion is linked to gating and filtration. Shrinkage porosity is linked to modulus, feeding, chilling, and graphite expansion. Machining risk is linked to the location of the last-solidifying region. When these factors are designed together, the casting can be produced with high yield and reliable internal quality. The process I now apply to similar ductile iron castings uses a longer runner, a horizontal filter, cored holes with steel reinforcement, conformal chills, and a controlled carbon equivalent. This combination has proven effective for the thick bearing cover and provides a practical route for other large ductile iron castings with critical bolt-hole zones.