In my work with ductile iron castings made by lost foam casting, I have found that the process offers excellent surface reproduction, high dimensional accuracy, and strong pattern yield, but it also creates a narrow process window when the casting contains concentrated thermal nodes and thin-to-thick transitions. The reducer housing I studied weighed 112 kg, had a maximum wall thickness of 54 mm, a minimum wall thickness of 14 mm, and was specified as QT450-10. The part required high strength, toughness, wear resistance, and vibration damping. The geometry was a回转-like body with a thick upper region, a rapid reduction in wall thickness below the middle section, and several bosses around bolt holes. These features produced a strongly non-uniform solidification pattern. In ductile iron castings, that non-uniformity directly influences carbon rejection, gas evolution, liquid feeding, and shrinkage porosity. My initial trials produced acceptable Y-block mechanical properties and a spheroidization rating of 2 to 3, but the actual reducer housing still showed wrinkling on the end face and shrinkage cavities near geometric hot spots. The problem was not the base metallurgy alone. It was the interaction between pattern decomposition, gating hydraulics, local modulus, and vacuum-assisted cooling.
The original process used a top-gating arrangement. Although the pattern was designed with a 4 mm machining allowance on the end face, the actual filling sequence did not behave as a simple top pour. The high-temperature metal entered the top face, penetrated the thick upper wall, then spread sideways and downward through the cavity. In effect, the cavity itself acted as part of the runner. The metal reached the thin wall as a falling stream, spread around the lower circumference, and finally rose to fill upper side regions. This created a mixed top-to-bottom-to-upward flow pattern that can be described as turbulent middle-to-bottom pouring. For ductile iron castings, such turbulence is harmful because it disturbs the decomposition layer of the lost foam pattern, entrains pyrolysis products, and pushes cooler metal into thick dead zones. The final filling locations became cold ends. Those cold ends retained liquid and solid decomposition products, leading to surface folds, carbon-rich skin, and the characteristic wrinkled appearance.
The pouring temperature was controlled between 1370 and 1440 °C, the tapping temperature was 1580 to 1600 °C, and the vacuum level was maintained between -0.06 and -0.04 MPa with a holding time of 900 s. The chemical composition was held within the usual range for QT450-10. The Y-block properties satisfied the specification, and the spheroidization rating was acceptable. That told me the melt treatment and base chemistry were not the primary cause of the defects. I therefore separated the problem into two linked phenomena: wrinkling on the casting surface and shrinkage cavities at thermal nodes. Wrinkling is mainly a filling and decomposition defect. Shrinkage is mainly a feeding and thermal-gradient defect. In lost foam casting of ductile iron castings, however, the two interact because the same pyrolysis gases, vacuum flow, and coating permeability that affect surface quality also affect heat transfer and feeding resistance.
| Element | Target range, mass fraction % | Function in ductile iron castings |
|---|---|---|
| C | 3.5 to 4.0 | Controls carbon equivalent, graphite formation, fluidity, and shrinkage tendency |
| Si | 2.0 to 3.0 | Promotes graphitization, controls matrix, and affects nodularity |
| Mn | ≤ 0.45 | Controls pearlite and hardenability; high levels increase carbides |
| P | ≤ 0.05 | Low melting phase; must be restricted for toughness |
| S | ≤ 0.025 | Consumes nodulizer; high levels reduce spheroidization |
| Mg | 0.02 to 0.06 | Primary nodulizing element in ductile iron castings |
| RE | 0.015 to 0.04 | Controls sulfur and oxygen, supports nodularity, and affects inclusion shape |
I calculated the carbon equivalent to understand the solidification window of the ductile iron castings before changing the tooling. A useful expression for cast iron is:
$$CE = C + \frac{Si}{3} + \frac{P}{3}$$
With the target ranges in the table, the carbon equivalent remained near the eutectic range. That is beneficial for graphitization and feeding, but it does not eliminate shrinkage when the local modulus is high and the thermal gradient is poor. The local solidification time can be approximated by:
$$t_s = k M^n$$
where \(t_s\) is solidification time, \(M\) is the local modulus, \(k\) is a material and mold constant, and \(n\) is an exponent typically between 1.5 and 2. The modulus is the ratio of volume to heat-transfer surface area:
$$M = \frac{V}{A}$$
At a geometric hot spot, the volume is large and the cooling surface area is small, so \(M\) increases. The last liquid to freeze is therefore located at the hot spot. If that region cannot be fed by a riser or by a strong thermal gradient, shrinkage porosity forms. In ductile iron castings, graphite expansion can partially compensate for liquid and solidification shrinkage, but the compensation is effective only when the mold is sufficiently rigid, the graphite expansion is well distributed, and the feeding path remains open. In lost foam casting, the rigid sand and vacuum system provide good mold strength, but the foam pattern and coating can alter local cooling and feeding. That is why I did not rely only on carbon equivalent or nodularity to solve the problem.
The wrinkling mechanism became clearer when I examined the decomposition products of the foam pattern. The pattern is not simply vaporized. It undergoes complex pyrolysis:
$$C_x H_y O_z \rightarrow C(s) + hydrocarbons + tars + gases$$
The solid carbon, liquid tar, and gaseous products must be transported away from the advancing metal front. If the metal front is turbulent, if the coating is too insulating, if the vacuum is too low, or if the local temperature is low, the products remain near the surface. The result is a carbon-rich layer, surface folds, and a wrinkled skin. I observed wrinkling mainly on the upper end face, vertical side walls, and dead corners. Those are exactly the locations where the original top-gating arrangement produced low-temperature metal and poor gas evacuation. The defects were not random. They followed the filling sequence.
| Defect | Typical location | Primary mechanism | Process variables involved |
|---|---|---|---|
| Wrinkling | Top face, upper side walls, dead corners | Pyrolysis product accumulation, cold-end metal, turbulent filling | Gating type, pouring temperature, vacuum, coating permeability, pattern density |
| Surface carbon defect | Last-filled regions and thick sections | Solid carbon and tar retention from foam decomposition | Pattern material, metal velocity, venting, sand permeability |
| Shrinkage cavity | Bolt-hole bosses, thick side sections, geometric hot spots | Insufficient feeding during liquid and solidification shrinkage | Modulus, thermal gradient, riser or chill design, pouring temperature |
| Shrinkage porosity | Hot spot transition zones | Dispersed micro-shrinkage before complete feeding | Carbon equivalent, nodularity, cooling rate, vacuum-assisted heat transfer |
| Cold shut or lap | Thin walls and long flow paths | Low fluidity, premature freezing, oxidized front | Pouring temperature, gating velocity, wall thickness, vacuum |
The original gating system was a top-pour design. In lost foam casting, a top pour often appears simple, but it produces a falling stream that damages the pattern surface and generates a large volume of pyrolysis gas directly under the sprue. The metal then flows through the cavity in an uncontrolled manner. When the reducer housing was poured, the upper thick wall acted as a temporary runner. The metal reached the lower thin wall before the upper side regions were fully filled. The actual filling sequence was therefore not top-down. It was top-penetration, bottom-spread, and upward-backfill. That sequence caused the last metal to arrive at the top face and upper side walls, where the temperature was lowest and the decomposition products were most concentrated. The last metal also had the lowest superheat, so it could not re-melt or push away the carbon-rich layer. The result was wrinkling.
To quantify the original filling condition, I used the conventional pouring time and static pressure relationships. For a cast iron casting, a practical pouring time estimate can be written as:
$$t = S \sqrt{G L}$$
where \(t\) is pouring time in seconds, \(G\) is casting mass, \(L\) is a fluidity and wall-thickness factor, and \(S\) is a coefficient related to alloy and mold. For the original housing, the calculated pouring time was about 24 s. The static pressure head at the bottom of the sprue was estimated from:
$$H_p = H_0 – \frac{C}{2}$$
where \(H_0\) is the initial metal head and \(C\) is the casting height above the ingate. For the original layout, the effective pressure head was about 34 cm. The minimum ingate area was estimated from:
$$A_g = \frac{G}{0.31 t \sqrt{H_p}}$$
That calculation gave approximately 3.46 cm² for a sand mold. However, lost foam casting requires a larger ingate than a conventional sand mold because the pattern decomposition consumes energy and because the gas front must be displaced. In practice, the recommended ingate area for ductile iron castings in lost foam casting is often 1.2 to 2.5 times the sand-mold value, depending on pattern density, coating, and vacuum level. The original top gating did not provide a stable, progressive bottom-to-top filling pattern. Instead, it created a high-velocity jet and a chaotic flow field.
| Parameter | Original top gating | Improved bottom gating |
|---|---|---|
| Filling direction | Top penetration, bottom spread, upward backfill | Progressive bottom-to-top filling |
| Flow character | Turbulent, cavity used as runner | Quiescent, controlled through ingates |
| Last-filled location | Top face and upper side walls | Top machining allowance and upper riser-like allowance |
| Cold-end location | Thick upper dead zones | Top allowance, away from finished surface |
| Pyrolysis product path | Trapped against upper surfaces | Flushed upward toward allowance and vents |
| Wrinkling tendency | High | Low |
| Shrinkage tendency | High at geometric hot spots | Reduced by lower turbulence and targeted heat dissipation |
I redesigned the gating system as a center-closed bottom-pouring system. The upper slag trap was removed. The new design used four ingates arranged centrally. Each ingate had a section of 70 mm by 40 mm with a tolerance allowance, giving a total ingate area between about 11.2 cm² and 12.8 cm². The sprue length was set to 480 mm, and the effective pressure head was about 200 mm. The gating ratio was adjusted to favor a slightly pressurized system, which helps maintain a stable metal front in lost foam casting. A typical ratio can be described as:
$$A_g : A_r : A_s = 1 : 1.2 : 1.4$$
where \(A_g\) is the total ingate area, \(A_r\) is the runner area, and \(A_s\) is the sprue exit area. The exact values depend on the casting and the vacuum system, but the principle is to avoid a high-velocity jet at the ingate and to keep the metal front as continuous as possible. In bottom gating, the metal enters at the lowest part of the cavity and rises steadily. The foam pattern decomposes layer by layer. The decomposition products are pushed upward ahead of the front. Because the top face has a machining allowance, the last metal and the last decomposition products remain in the allowance rather than on the finished surface. This is a key reason why bottom gating solved the wrinkling defect in the ductile iron castings.
I verified the new design by producing 2000 housings. The surface quality was inspected after shot blasting and before machining. The batch showed no repeated wrinkling on the end face or side walls. The surface was smooth and uniform. The top allowance contained the last metal and any minor residue, and that allowance was removed during machining. The improvement confirmed that the wrinkling was not an unavoidable property of ductile iron castings in lost foam casting. It was a consequence of the original filling pattern.
| Verification item | Original process result | Bottom-gated process result |
|---|---|---|
| Batch size for comparison | Small-batch trial and early production | 2000 pieces |
| End-face wrinkling | Frequent and batch-related | No repeated wrinkling |
| Upper side-wall folds | Occurred in dead zones | Eliminated or confined to allowance |
| Surface after blasting | Orange-peel and rough patches | Smooth and consistent |
| Machining allowance condition | Defects sometimes reached finished surface | Defects captured in allowance |
| Process stability | Variable | Stable within the tested batch |
After the wrinkling problem was controlled, I turned to the shrinkage cavities. These appeared in the thick side sections and around the bolt-hole bosses. Those regions were geometric hot spots. They had a larger modulus than the surrounding thin walls. I measured and modeled the local volume-to-area ratio. The hot spots had a modulus perhaps two to three times that of the nearby thin wall. Under normal solidification, the thin wall freezes first, and the hot spot remains liquid. If the feeding path from the gating system or from a riser closes before the hot spot finishes freezing, a shrinkage cavity forms. The problem is made more severe in ductile iron castings because the graphite expansion that can compensate for shrinkage is not uniform. If the hot spot is isolated, graphite expansion may increase internal pressure, but it may also cause mold wall movement or localized porosity if the surrounding skin is weak.
The conventional solutions are risers and chills. A riser provides liquid metal to compensate for volume contraction. A chill increases the local cooling rate and creates an artificial end zone. Both can work in sand castings, but they are difficult in lost foam casting. A riser increases pattern complexity, reduces pattern yield, and adds a cutting and cleaning operation. A chill placed in the sand is difficult to position reliably because the pattern is coated and buried. During compaction and vacuum, the chill can shift, fall, or create a gap. That gap changes local heat transfer and may cause distortion. External chills also use valuable space and can interfere with the foam pattern. For these reasons, I did not choose risers or conventional chills as the primary solution.
Instead, I developed a heat-dissipation process that uses foamed heat-sink pieces attached directly to the pattern at the hot spots. I call these pieces heat-sink fins. They are made of the same or a similar foam material and are bonded to the pattern surface before coating. During coating, drying, sand filling, and pouring, they remain in place. After pouring, they decompose and leave no solid residue in the casting. Their function is not to act as a metal chill. Their function is to increase the effective surface area of the hot spot during solidification and to create a local flow path for cooling air under vacuum. The vacuum pump continuously draws air through the sand. The air enters from the top of the flask, passes through the sand, reaches the casting and the heat-sink fins, exchanges heat, and exits through the vacuum system. The fins increase the surface area for convection and conduction. The local mold sand around the fins forms a microchannel network. The result is a strong local cooling effect.
The heat transfer can be approximated by:
$$Q = h A (T_s – T_\infty)$$
where \(Q\) is the heat transfer rate, \(h\) is the convective heat-transfer coefficient, \(A\) is the surface area, \(T_s\) is the surface temperature, and \(T_\infty\) is the cooling-air temperature. When I attach a fin, the area \(A\) increases. The fin efficiency can be estimated from:
$$\eta_f = \frac{\tanh(mL)}{mL}$$
where
$$m = \sqrt{\frac{h P}{k A_c}}$$
Here \(P\) is the fin perimeter, \(k\) is the thermal conductivity of the fin material, \(A_c\) is the fin cross-sectional area, and \(L\) is the fin length. In the lost foam process, the fin is consumed by the metal, so the effective cooling occurs mainly before and during the early stages of solidification. Even so, the additional surface area and the vacuum-driven air flow reduce the local solidification time and lower the local modulus. The modulus after adding fins can be written as:
$$M_f = \frac{V}{A + A_f}$$
where \(A_f\) is the additional effective heat-transfer area created by the fins. Because \(A_f > 0\), \(M_f < M\). A lower modulus means a shorter solidification time and a smaller hot spot. If the fins are placed correctly, the original geometric hot spot is converted into a region of higher cooling rate, and the thermal gradient becomes more favorable for directional solidification.
I placed 12 heat-sink fins on each reducer housing. Each fin measured 50 mm by 30 mm by 7 mm. The fins were bonded to the pattern at the bolt-hole bosses and the thick side sections. The other process steps remained the same. After coating, drying, sand filling, vacuum application, pouring, cooling, and cleaning, the castings were machined. The bolt-hole areas were inspected after machining. No shrinkage cavities were found in the tested batch of 2000 pieces. The fins had decomposed completely, and no residue or surface defect was observed. The process yield remained high because no large risers were added. The cleaning operation was simple because the fins left no solid metal to cut.
| Heat-dissipation parameter | Value or description |
|---|---|
| Fin material | Foam compatible with lost foam pattern |
| Fin dimensions | 50 mm × 30 mm × 7 mm |
| Number of fins per housing | 12 |
| Attachment locations | Bolt-hole bosses and thick side hot spots |
| Attachment method | Bonded to pattern before coating |
| Coating | Applied over pattern and fins; dried before sand filling |
| Vacuum condition | Continuous negative pressure during pouring and solidification |
| Cooling mechanism | Increased surface area, microchannel air flow, local convection, reduced modulus |
| Effect on pattern yield | Very small; no large riser required |
| Effect on cleaning | No metal fins; simple cleaning |
The heat-sink process is especially suitable for ductile iron castings because it addresses the local thermal condition without changing the global chemistry. In ductile iron castings, carbon equivalent and nodularity are important, but they do not by themselves determine whether a hot spot will shrink. The local cooling rate, the feeding path, and the pressure condition during solidification are equally important. By reducing the local modulus, the heat-sink fins allow the hot spot to freeze earlier or at least at a rate closer to the surrounding wall. This reduces the volume of isolated liquid and lowers the shrinkage risk. The vacuum-driven air flow also helps remove heat from the sand, so the effect is not limited to the fin contact area. The surrounding sand becomes a cooler heat sink, and the thermal gradient becomes more directional.
I also considered the gas pressure balance during filling, because it affects both wrinkling and shrinkage. In lost foam casting, the metal front must displace the pattern decomposition products. A simplified pressure balance at the front can be written as:
$$P_{metal} + P_{vacuum} = P_{gas} + P_{sand} + P_{coating} + \Delta P_{flow}$$
where \(P_{metal}\) is the metallostatic pressure, \(P_{vacuum}\) is the vacuum pressure, \(P_{gas}\) is the gas pressure generated by pattern decomposition, \(P_{sand}\) is the pressure drop through the sand, \(P_{coating}\) is the pressure drop through the coating, and \(\Delta P_{flow}\) is the pressure loss in the gating system. If the gas pressure becomes too high, the metal front can become unstable, and gas can be trapped. If the vacuum is too high, the metal front can be drawn forward too quickly, causing turbulence and cold shuts. If the coating is too insulating, the gas cannot escape. The bottom-gating design helped because it produced a more uniform front and pushed the gas upward. The heat-sink fins helped because they created local pathways for air flow and heat exchange.
The vacuum level was kept between -0.06 and -0.04 MPa. The holding time was 900 s. The pouring temperature remained between 1370 and 1440 °C. These values were not changed dramatically. The main changes were the gating direction and the local thermal design. This is important because it shows that the defects were not solved by simply raising the pouring temperature. A higher pouring temperature can improve fluidity, but it also increases liquid shrinkage and gas generation. In ductile iron castings, an excessively high pouring temperature can worsen shrinkage and carbon defects. The better approach is to control the filling pattern and the local cooling rate.
| Process variable | Original range | Optimized range or change | Reason for control |
|---|---|---|---|
| Tapping temperature | 1580 to 1600 °C | Kept in same range | Ensures proper spheroidization and melt treatment |
| Pouring temperature | 1370 to 1440 °C | Kept in same range | Balances fluidity against shrinkage and gas evolution |
| Vacuum level | -0.06 to -0.04 MPa | Kept in same range | Maintains sand rigidity and gas evacuation |
| Vacuum holding time | 900 s | Kept at 900 s | Allows complete solidification before release |
| Gating type | Top gating | Center-closed bottom gating | Reduces turbulence and controls filling sequence |
| Ingate area | Small sand-mold estimate | Four ingates, total 11.2 to 12.8 cm² | Provides stable flow for lost foam decomposition |
| Sprue length | Original layout | 480 mm | Provides sufficient pressure head |
| Hot-spot cooling | None | 12 foam heat-sink fins | Reduces local modulus and increases heat transfer |
I used the Niyama criterion as a qualitative guide for shrinkage porosity. The criterion can be expressed as:
$$N_y = \frac{G}{\sqrt{R}}$$
where \(G\) is the temperature gradient and \(R\) is the cooling rate. A low \(N_y\) value indicates a higher risk of shrinkage porosity. In the original process, the hot spots had a low temperature gradient and a relatively low cooling rate, so \(N_y\) was low. Bottom gating improved the overall filling temperature and reduced cold-end effects, but it did not by itself eliminate the geometric hot spots. The heat-sink fins increased the local cooling rate \(R\) and improved the thermal gradient \(G\) around the hot spot. Therefore, \(N_y\) increased locally, and the shrinkage risk decreased. This explains why the combination of bottom gating and heat-sink fins was more effective than either change alone.
I also examined the volume contraction during solidification. For a simplified estimate, the total volume change can be written as:
$$\Delta V = V \left( \beta_l \Delta T_l + \beta_s \right)$$
where \(\beta_l\) is the liquid volumetric contraction coefficient, \(\Delta T_l\) is the superheat above the liquidus, and \(\beta_s\) is the solidification contraction coefficient. In ductile iron castings, graphite expansion can offset part of \(\Delta V\), but the offset depends on the timing and distribution of graphite growth. If graphite grows late or in isolated regions, it cannot feed the hot spot. If the mold is too soft, graphite expansion can expand the mold instead of feeding the liquid. In lost foam casting, the sand is under vacuum and is relatively rigid, which is favorable. However, the pattern decomposition layer and the coating can create local soft spots or gas paths. The heat-sink fins helped by reducing the hot-spot volume and by increasing the cooling rate, so the graphite expansion occurred earlier and more uniformly.
The chemical composition of the ductile iron castings was kept within the specification. I did not try to solve the shrinkage by simply increasing carbon. Although a higher carbon equivalent can improve graphitization and reduce shrinkage in some castings, it can also increase the risk of carbon defects in lost foam casting. The pattern pyrolysis already introduces carbon. Adding more carbon to the melt may increase graphite flotation, carbides, or surface carbon defects. The better strategy was to maintain the carbon equivalent near the eutectic range and to control the local solidification condition. The heat-sink process allowed me to keep the composition stable and to improve the hot spot without changing the melt treatment.
| Shrinkage factor | Effect on ductile iron castings | Control action |
|---|---|---|
| Liquid contraction | Volume decrease as superheat is lost | Control pouring temperature and filling time |
| Solidification contraction | Volume decrease during austenite and graphite growth | Promote graphite expansion and feed hot spots |
| Local modulus | High modulus delays freezing and creates hot spots | Use heat-sink fins to increase surface area |
| Temperature gradient | Low gradient promotes dispersed porosity | Bottom gating and local cooling improve gradient |
| Feeding path | Blocked path prevents liquid compensation | Design gating and hot-spot location to maintain feeding |
| Graphite expansion | Can compensate shrinkage if well distributed | Control carbon equivalent, nodularity, and cooling rate |
| Mold rigidity | Soft mold reduces expansion efficiency | Maintain vacuum and proper sand compaction |
| Gas pressure | High gas pressure can disturb feeding | Control coating permeability, vacuum, and pattern density |
The pattern material also influenced the defects. For ductile iron castings, I selected a copolymer pattern material rather than a simple EPS pattern. The copolymer balances gas evolution and solid carbon generation. EPS produces a large amount of gas and carbon, while EPMMA produces more gas and less solid carbon. A copolymer blends the two behaviors. This reduces the risk of carbon defects and gas porosity while still providing good pattern strength and surface finish. The pattern density was controlled to avoid excessive decomposition products. A low-density pattern decomposes faster but may produce more gas per unit volume. A high-density pattern produces more solid carbon and tar. The copolymer gave a better balance for the reducer housing.
The coating was also important. The coating must allow gas to pass through but must not allow metal to penetrate the sand. If the coating is too permeable, metal can infiltrate and cause surface roughness or burn-on. If the coating is too insulating, gas cannot escape and carbon defects form. I used a coating with controlled permeability and applied it uniformly over the pattern and the heat-sink fins. The coating was dried thoroughly before sand filling. Moisture in the coating can generate steam and cause gas defects. The drying process was therefore part of the defect-control strategy, not just a preparation step.
| Pattern and coating factor | Influence on ductile iron castings | Optimized condition |
|---|---|---|
| Pattern material | Controls gas and solid carbon during pyrolysis | Copolymer balanced for gas and carbon |
| Pattern density | Affects decomposition rate and gas volume | Controlled within a stable production range |
| Pattern surface | Affects coating adhesion and casting surface | Smooth, defect-free, properly bonded |
| Heat-sink bonding | Prevents fins from falling during coating and sand filling | Strong bond with compatible adhesive |
| Coating permeability | Controls gas evacuation and metal penetration | Uniform, matched to vacuum level |
| Coating thickness | Affects local insulation and gas path | Uniform, within specification |
| Coating drying | Prevents steam defects and coating spalling | Complete drying before sand filling |
| Sand compaction | Affects mold rigidity and heat transfer | Controlled vibration and vacuum |
I also paid attention to the filling velocity. In bottom gating, the metal rises from the bottom, but if the ingate velocity is too high, the front can still become turbulent. The ingate velocity can be estimated from:
$$v_g = \frac{Q}{A_g}$$
where \(Q\) is the volumetric flow rate and \(A_g\) is the total ingate area. For a given pouring time, the flow rate is related to the casting mass and density:
$$Q = \frac{G}{\rho t}$$
By increasing the total ingate area to about 11.2 to 12.8 cm², I reduced the ingate velocity compared with a small sand-mold ingate. The lower velocity produced a more stable metal front. The four ingates also distributed the flow around the center, which reduced the tendency for a single high-velocity jet to cut through the pattern. The result was a smoother filling sequence and less pyrolysis product entrapment.
The Reynolds number can be used as a qualitative indicator of turbulence:
$$Re = \frac{\rho v D}{\mu}$$
where \(\rho\) is density, \(v\) is velocity, \(D\) is a characteristic dimension, and \(\mu\) is dynamic viscosity. A lower ingate velocity reduces \(Re\). In lost foam casting, the critical condition is not only the Reynolds number in the runner but also the stability of the metal front in the cavity. A lower velocity and a larger ingate area help maintain a coherent front. This is one reason why the bottom-gated design produced better surface quality.
The heat-sink fins also influenced the local Reynolds number and heat transfer in the sand. The vacuum draws air through the sand pores. The flow can be approximated by Darcy’s law:
$$Q = \frac{\kappa A}{\mu} \frac{\Delta P}{L}$$
where \(\kappa\) is the permeability of the sand, \(A\) is the cross-sectional area, \(\Delta P\) is the pressure difference, and \(L\) is the flow length. The fins increase the local surface area and create channels with lower flow resistance. The air flow removes heat from the sand and from the casting surface. The effect is strongest near the fins and decreases with distance. By placing fins at the hot spots, I targeted the cooling effect where it was needed. The surrounding sand became cooler, which increased the local temperature gradient and promoted directional solidification.
The heat-sink process is simple, but it must be controlled. If the fins are too large, they can chill the metal too quickly and cause cold shuts or incomplete filling. If they are too small, they do not provide enough cooling. If they are placed incorrectly, they may create a new hot spot or interrupt the coating. I determined the fin size and number by considering the local modulus, the wall thickness, the distance to the thin wall, and the available pattern surface. The 50 mm × 30 mm × 7 mm fin was a practical size for the reducer housing. Twelve fins were enough to cover the main hot spots without over-cooling the surrounding walls. The fins were placed symmetrically where possible to avoid distortion. The coating was applied carefully around the fins to maintain a continuous gas path.
| Fin design consideration | Reason | Chosen condition |
|---|---|---|
| Fin size | Controls cooling area and decomposition volume | 50 mm × 30 mm × 7 mm |
| Number of fins | Must cover all critical hot spots without over-cooling | 12 per housing |
| Placement | Must target geometric hot spots and bolt-hole bosses | Directly bonded at hot spots |
| Symmetry | Reduces thermal distortion | Balanced around the housing |
| Bonding | Prevents movement during coating and sand filling | Compatible adhesive, full contact |
| Coating coverage | Maintains gas evacuation and prevents sand penetration | Uniform coating over fins and pattern |
| Drying | Prevents steam and coating defects | Complete drying before sand filling |
| Inspection | Ensures fins remain in place before pouring | Visual check after coating and sand filling |
The results after implementing both changes were consistent. The bottom gating eliminated the surface wrinkling. The heat-sink fins eliminated the shrinkage cavities at the bolt-hole bosses and thick side sections. The process yield remained high because no large risers were added. The cleaning and machining operations were simplified because there were no metal risers to cut and no chill marks to remove. The casting dimensions and surface quality met the requirements. The Y-block mechanical properties continued to satisfy QT450-10, and the spheroidization rating remained within 2 to 3. This showed that the process changes did not harm the metallurgical quality of the ductile iron castings.
| Quality characteristic | Requirement or original condition | After optimization |
|---|---|---|
| Material | QT450-10 | QT450-10 maintained |
| Tensile behavior | Y-block met specification | Y-block continued to meet specification |
| Spheroidization rating | 2 to 3 | 2 to 3 maintained |
| Surface wrinkling | Present on end face and side walls | Eliminated in 2000-piece batch |
| Shrinkage cavity | Present at bolt-hole bosses | Not found after machining in 2000-piece batch |
| Process yield | Reduced by potential risers | High; no large risers required |
| Cleaning | Potential riser cutting | Simple; fins decompose completely |
| Dimensional accuracy | Requirement met but defects interfered | Requirement met with stable surface |
From a practical standpoint, I recommend that any lost foam casting process for ductile iron castings be evaluated with three questions. First, what is the actual filling sequence? A top-gating design may appear to fill from the top, but the cavity can become the runner, producing a mixed flow that traps pyrolysis products. Second, where are the last-filled regions and geometric hot spots? Those locations determine both surface defects and shrinkage. Third, what local thermal control is available? Risers and chills are not always compatible with lost foam casting. Heat-sink fins provide a simple alternative that increases surface area and uses the vacuum-driven air flow to remove heat.
The filling sequence can be mapped with a simple mass balance. If the metal front rises uniformly, the fill height \(h(t)\) can be approximated by:
$$h(t) = \frac{Q t}{A_c}$$
where \(A_c\) is the horizontal cross-sectional area of the cavity at the filling height. If the cavity area changes suddenly, the front velocity changes. A large increase in area, such as at a thick upper wall, slows the front and can create a cold end. A decrease in area, such as at a thin wall, accelerates the front and can cause turbulence. The reducer housing had both features. The original top gating exaggerated the effect. The bottom gating made the front more uniform because the metal entered at the bottom and rose through the changing cross-section in a controlled way. The top allowance acted as a buffer for the last metal and decomposition products.
The solidification sequence can also be estimated by comparing local moduli. If the modulus of the hot spot is \(M_h\) and the modulus of the surrounding wall is \(M_w\), the solidification time ratio is:
$$\frac{t_{s,h}}{t_{s,w}} = \left( \frac{M_h}{M_w} \right)^n$$
A ratio greater than 1 means the hot spot freezes later. If the ratio is large, a riser or local cooling is needed. The heat-sink fins reduce \(M_h\) by increasing the heat-transfer area. If the fin area is \(A_f\), the new modulus is:
$$M_{h,f} = \frac{V_h}{A_h + A_f}$$
The ratio becomes:
$$\frac{t_{s,h,f}}{t_{s,w}} = \left( \frac{M_{h,f}}{M_w} \right)^n$$
By reducing \(M_{h,f}\), the ratio moves closer to 1. This reduces the risk of isolated shrinkage. The exact fin area depends on the local geometry and the heat-transfer coefficient, but the principle is clear. The heat-sink fins change the local modulus and the local cooling rate without changing the global gating or the alloy composition.
In ductile iron castings, the graphite expansion also depends on the cooling rate. A faster cooling rate can refine the graphite and increase the number of nodules. A slower cooling rate in a hot spot can produce coarse graphite, segregation, and shrinkage. The heat-sink fins increase the local cooling rate, so they promote a more uniform graphite distribution. This is beneficial for mechanical properties and for shrinkage control. The vacuum-driven air flow also removes heat from the sand, which reduces the mold temperature and increases the temperature gradient. The combination of these effects explains why the heat-sink process worked so well for the reducer housing.
| Physical effect | Mathematical indicator | Influence of heat-sink fins |
|---|---|---|
| Local modulus | \(M = V/A\) | \(M\) decreases because \(A\) increases |
| Solidification time | \(t_s = k M^n\) | \(t_s\) decreases at the hot spot |
| Temperature gradient | \(G\) | \(G\) increases near the fin |
| Cooling rate | \(R\) | \(R\) increases due to local heat extraction |
| Niyama criterion | \(N_y = G / \sqrt{R}\) | \(N_y\) increases, reducing shrinkage risk |
| Convective heat transfer | \(Q = h A (T_s – T_\infty)\) | \(Q\) increases because \(A\) and flow increase |
| Darcy air flow | \(Q = \kappa A \Delta P / (\mu L)\) | Fins create local channels and increase flow |
| Graphite expansion | \(\Delta V_g\) | More uniform cooling supports distributed expansion |
I also found that the heat-sink fins did not cause any significant problem during pouring. Because they are made of foam, they decompose before the metal reaches them. They do not create a metal-to-metal chill that could cause cold shuts. Their cooling effect is mainly through the sand and the vacuum air flow. This is different from a metallic chill, which extracts heat very quickly and can cause local chilling, carbides, or cracks. The foam fins provide a gentler and more distributed cooling effect. They are also easy to attach and do not require special tooling. This makes the process practical for production.
The gating system and the heat-sink process should be designed together. If the gating system is changed without considering the hot spots, the shrinkage may remain. If the heat-sink fins are added without stable filling, the surface defects may remain. In the reducer housing, the bottom gating solved the wrinkling by controlling the filling sequence and by moving the last metal to the allowance. The heat-sink fins solved the shrinkage by controlling the local thermal condition at the hot spots. The two changes complemented each other. The bottom gating reduced turbulence and gas entrapment, which improved the effectiveness of the fins. The fins reduced the hot-spot modulus, which reduced the risk of shrinkage even if the gating system could not fully feed the thick sections.
I verified the process over a batch of 2000 pieces. The castings were inspected for surface quality, machined at the bolt-hole areas, and checked for shrinkage. The results were consistent. There were no repeated wrinkling defects and no shrinkage cavities at the critical locations. The process yield remained high, and the cleaning operation was simple. The mechanical properties and microstructure remained within specification. This confirmed that the optimized process is stable and suitable for production of ductile iron castings with similar geometry.
| Production verification | Result |
|---|---|
| Batch size | 2000 reducer housings |
| Gating system | Center-closed bottom gating with four ingates |
| Hot-spot control | 12 foam heat-sink fins per housing |
| Surface wrinkling | No repeated defects |
| Shrinkage at bolt holes | No defects after machining |
| Mechanical properties | QT450-10 requirements met |
| Spheroidization | Rating 2 to 3 maintained |
| Process yield | High; no large risers |
| Cleaning | Simple; no metal fins or risers |
| Production stability | Stable within the verified batch |
The main lesson I take from this work is that defects in lost foam casting of ductile iron castings should not be treated as isolated problems. Wrinkling and shrinkage may appear in different locations, but they often share the same root causes: unstable filling, poor gas evacuation, and unfavorable local thermal gradients. The original top gating created a turbulent filling pattern that produced cold ends and carbon-rich surface defects. The geometric hot spots created slow-freezing regions that could not be fed. By changing to bottom gating, I stabilized the filling sequence and moved the last metal to the machining allowance. By adding foam heat-sink fins, I increased the local cooling rate and reduced the hot-spot modulus. Together, these changes eliminated the defects without reducing process yield or complicating cleaning.
For future work, I would extend the same approach to other ductile iron castings with thick-to-thin transitions. The heat-sink fin size and placement can be optimized by thermal simulation. The gating system can be designed using the same principles: bottom filling, controlled ingate velocity, sufficient total ingate area, and a top allowance to capture the last metal and decomposition products. The vacuum level and coating permeability should be matched to the pattern material and the casting geometry. The carbon equivalent and nodularity should be kept within specification, but they should not be expected to compensate for poor local thermal design. In ductile iron castings, the best results come from combining good metallurgy with good filling and good local cooling.
In summary, I solved the wrinkling defect by changing the pouring system from top gating to center-closed bottom gating. The new system used four ingates with a total area of about 11.2 to 12.8 cm², a sprue length of 480 mm, and an effective pressure head of about 200 mm. The metal filled the cavity progressively from the bottom, and the last metal and pyrolysis products were captured in the top allowance. I solved the shrinkage defect by attaching 12 foam heat-sink fins, each 50 mm × 30 mm × 7 mm, to the geometric hot spots. The fins increased the local heat-transfer area, reduced the local modulus, and used the vacuum-driven air flow to remove heat. The result was a stable process that produced sound ductile iron castings with high yield and simple cleaning. The process was verified on 2000 housings, and the defects did not reappear. This approach is practical, low-cost, and suitable for lost foam casting of ductile iron castings with concentrated thermal nodes.
| Final process summary | Specification or result |
|---|---|
| Casting mass | 112 kg |
| Maximum wall thickness | 54 mm |
| Minimum wall thickness | 14 mm |
| Material | QT450-10 |
| Original gating | Top gating |
| Optimized gating | Center-closed bottom gating |
| Number of ingates | 4 |
| Ingate section | 70 mm × 40 mm each |
| Total ingate area | 11.2 to 12.8 cm² |
| Sprue length | 480 mm |
| Effective pressure head | About 200 mm |
| Heat-sink fin size | 50 mm × 30 mm × 7 mm |
| Number of heat-sink fins | 12 |
| Vacuum level | -0.06 to -0.04 MPa |
| Vacuum holding time | 900 s |
| Pouring temperature | 1370 to 1440 °C |
| Tapping temperature | 1580 to 1600 °C |
| Wrinkling result | Eliminated in 2000-piece batch |
| Shrinkage result | Eliminated at bolt-hole hot spots |
| Mechanical properties | QT450-10 requirements met |
| Spheroidization | 2 to 3 |
| Process yield | High |
| Cleaning | Simple |
The equations and tables I used in this study provide a framework for understanding and controlling defects in lost foam casting of ductile iron castings. They show that filling, solidification, and gas evacuation are coupled. A change in gating affects the temperature field. A change in local cooling affects the shrinkage risk. A change in pattern material or coating affects the gas pressure and carbon defect tendency. When these factors are considered together, the process becomes more robust. The reducer housing case demonstrates that a practical solution can be achieved without complex tooling or expensive materials. Bottom gating and foam heat-sink fins are simple, low-cost, and effective. They can be applied to other ductile iron castings where wrinkling and shrinkage are controlled by geometric hot spots and filling instability.