Keywords: cold box process; stacked molding; robotic grinding; production efficiency; ductile iron castings
In our foundry, the small-part workshop is responsible for the batch production of small ductile iron castings. We produce castings weighing less than 50 kg throughout the year. The monthly output is approximately 400 tons, and the quantity of each casting type in a normal order varies from 30 to 500 pieces. These production conditions are the classic definition of a small-batch, multi-variety foundry environment. The original manufacturing route used resin sand pattern-plate molding. For many years, this process served our customers, but it gradually became the bottleneck limiting both quality and delivery performance.
Small ductile iron castings require careful control of mold filling, solidification, and post-casting cleaning. Because ductile iron castings have a spheroidal graphite structure, their solidification behavior differs from that of gray iron. The graphite expansion during eutectic solidification can either be beneficial or harmful, depending on mold rigidity. If the mold is soft, graphite expansion pushes the mold wall outward, creating internal porosity; if the mold is rigid, the expansion helps feed shrinkage and produces sound castings. This important metallurgical characteristic was one of the main reasons why we had to move away from traditional resin sand pattern-plate molding for our small ductile iron castings.
When we examined the entire production process, we realized that the old resin sand process was not suitable for the small and medium batch quantities of ductile iron castings that dominate our workshop. The casting molds were heavy, labor intensive, and difficult to automate. In addition, the dimensional consistency of the resulting ductile iron castings was not ideal. After a detailed investigation, we decided to change the process to a cold box core-making process, using stacked cores and pouring multiple layers of ductile iron castings in a single mold. This paper describes the improvements we made in mold production, pouring, cleaning, and grinding in our small-batch ductile iron castings production system.
Limitations of the Traditional Resin Sand Pattern-Plate Process
The original process used resin sand pattern plates. The pattern equipment was simple, but the productivity was very low. In practice, we could only make six molds per day with one molding station. This was unacceptable because the monthly output target of 400 tons of small ductile iron castings required a much higher molding rate. The traditional process also had a high sand-to-metal ratio, typically from 5:1 to 8:1. That means for every kilogram of finished ductile iron castings, we had to produce and reclamate five to eight kilograms of resin sand molding material.
This high sand-to-metal ratio increased the cost of resin, hardener, sand reclamation, and waste disposal. It also increased the amount of manual work required to move and handle sand. The molding process produced only one casting level per flask, so the metal yield was low. Even though the gating system was simple, the ratio of poured metal to final castings was poor. We had to cut off relatively large runners and risers, and the resulting scrap metal had to be remelted.
The second serious problem was casting loss during shakeout. Because the molds were small and individual castings were buried in a large amount of resin sand, the pattern-plate process allowed castings to be separated and lost in the sand during shakeout and handling. This was very severe for small ductile iron castings, because a small casting can be hidden easily in a pile of sand or in a shot-blasting machine. We frequently had to make replacement castings for lost or damaged parts. This disrupted the production plan and created difficulties with on-time delivery.
The third problem was excessive flash and poor control of parting-line tolerances. Resin sand pattern plates tend to shift or produce uneven mold surfaces. This caused thin fins and flash on the ductile iron castings. The flash was difficult to remove and often created deep gouges when the operator tried to grind it away. The fettling cost was high, and the surface quality of the final ductile iron castings was not consistent.
Finally, the worker environment was not ideal. The resin sand process, manual molding, manual shakeout, and manual grinding all generate dust, noise, and fumes. New environmental regulations and health requirements made it difficult to find workers willing to perform these tasks. We recognized that the only long-term solution was to change the casting process and the cleaning process in a more fundamental way.
Development of the Cold Box Stacked Molding Process
The solution we chose was the cold box core process, also known as the cold box process. In this process, a phenolic urethane resin is mixed with sand and blown into a core box. The core is then gassed with an amine catalyst to cure it rapidly at room temperature. The cold box process is widely used to produce sand cores for internal cavities, but in our case, we use cold box technology to produce the entire mold, including the external cavity, the gating system, and the stacking surfaces.
We designed cold box tooling for a large family of small ductile iron castings. The tooling can be arranged in different ways:
- One cavity per core, when the casting is large relative to the stacking core dimensions.
- Multiple cavities per core, when the casting is small enough to allow several patterns in one core box.
- Different castings grouped in one core assembly, so that a single order of mixed ductile iron castings can be produced together.
This layout flexibility is very important for small-batch ductile iron castings because the order quantity may be only 30 pieces. It is not economical to create a dedicated pattern plate for each type, but it is very economical to design a cold box core box that can produce several types in one core. The number of castings in one stacked mold can be expressed as:
$$N_{\text{total}}=\sum_{i=1}^{p}n_i$$
where \(n_i\) is the number of castings of type \(i\) in one core layer, and \(p\) is the number of different casting types included in the same stack. By increasing \(N_{\text{total}}\), the number of castings produced per mold can be increased dramatically. In the old resin sand process, one flask usually produced one casting. In the cold box stacked process, one flask can produce many dozens of small ductile iron castings at one time.
The stacking principle is simple. Each core is made with a precise flat face so that the cores can be stacked vertically. The gating system is arranged to connect all cavities through a central runner. When the cores are assembled, the central sprue extends through the entire stack, and the molten metal flows from the bottom to the top. This bottom-filling arrangement allows a quiet and stable filling pattern, which is particularly valuable for ductile iron castings because they can form dross and oxide inclusions.
We also planned the separation of the castings from the runner system. Since the castings remain attached to the runner after solidification, the whole group becomes a casting tree. This tree is easy to handle, and no small casting can be lost in the sand during shakeout. The tree also provides a very convenient handle for shot blasting and for robotic grinding.
Cold Box Core Manufacturing Process
The production flow of the cold box stacked molding process in our workshop can be divided into four main stages: core making, coating, core assembly, and casting.
Core Making
Core making is performed on a cold core machine. The sand is mixed with a two-part resin binder. The mixed sand is then blown into the cold box tooling at a low pressure. After the core is compacted, a triethylamine catalyst is vaporized and passed through the core. The catalyst causes the resin to polymerize quickly, producing a strong, accurate sand core. The core is then stripped from the tool and placed on the assembly line.
Each core must have good dimensional accuracy because the core stack relies on the flatness of the joint faces. If the core faces are uneven, the stack will not be vertical, and the casting dimensions will vary. We therefore pay close attention to the core box maintenance, venting, and blow pressure. The quality of the core directly determines the quality of the final ductile iron castings.
The cold box process is faster than the old resin sand pattern-plate process. Once the core box is filled and gassed, the core is ready immediately. There is no long wait for sand to cure as in the resin sand process. The work of making a complete mold is replaced by making a stack of cores, and this can be done by a small team.
Coating
After the cores are made, we apply an alcohol-based refractory coating. The purpose of the coating is to improve the surface finish of the ductile iron castings and to prevent sand burn-on. We use a dipping method because it gives a uniform coating layer on all surfaces. The coating is applied to the entire core surface, including the cavity and the gating channels. The coated cores are then dried by igniting the alcohol, which evaporates quickly and leaves a hard refractory layer.
Coating is especially important for small ductile iron castings because their surface quality must be high without requiring expensive machining. The coating also helps the sand cores to strip more cleanly after pouring, improving the shakeout operation. When we compare the new process with the old resin sand pattern-plate process, the coating technology is a major reason for the reduction of surface defects and flash on ductile iron castings.
Core Assembly and Stacking
The cores are assembled into a stack. The stack may consist of the same casting repeated many times, or it may consist of different castings arranged in the same order as the customer order. The only rule is that the gating system must be continuous from the bottom to the top. We normally use a bottom sprue and a horizontal runner at each layer. The gates enter the cavity at the lower part, so the melt rises smoothly inside the mold cavity.
This bottom-gating design is beneficial for ductile iron castings in several ways. First, it reduces turbulence and prevents air entrapment. Second, it avoids cold metal splashing on the core surface. Third, it promotes a directional solidification pattern because the bottom part of the casting remains hot while the upper part of the stack is slightly cooler. Fourth, the entire stack provides a large casting yield because the runner system is shared by many castings.
Sand Backup and Mold Rigidity
After the cores are stacked, they are placed into a steel flask. The flask is larger than the stack, and the space between the stack and the flask is filled with a support material. For ductile iron castings, mold rigidity is critical. If the mold is not rigid, the graphite expansion during eutectic solidification will push the mold walls outward. This movement prevents the graphite expansion from compensating for liquid contraction and can create internal shrinkage porosity in the ductile iron castings.
We therefore selected resin sand as the backup material. The resin sand is packed around the core stack and allowed to cure. Once cured, it forms a strong, rigid envelope around the stack. This rigid assembly ensures that the effective modulus of the mold is high enough to withstand both the ferrostatic pressure and the graphite expansion pressure.
There are also two alternative backup methods that we considered. The first is the dry-sand vacuum process, in which the flask is filled with dry silica sand and a vacuum is applied to make the sand mass rigid. The second is filling the flask with steel shot, which has a high density and high heat absorption. Each method is suitable for certain types of ductile iron castings, but we found that resin sand backup gave the best combination of rigidity, cost, and convenience for our product range.
| Backup Method | Rigidity | Cost | Operational Features |
|---|---|---|---|
| Resin sand | High | Moderate | Good surface support; requires curing time; our chosen method |
| Dry sand with vacuum | Medium-high | Lower | Fast filling; requires vacuum equipment; less rigid than resin sand |
| Steel shot | High | Higher | Very dense; high weight; good cooling; more expensive handling |
These backup methods were evaluated specifically for small ductile iron castings. Since our castings are relatively small and are stacked in multiple layers, the backup material must be able to fill the narrow gaps between the stack and the flask. Both resin sand and steel shot perform well, but resin sand gives a more stable and predictable assembly. The dry-sand vacuum process is excellent for large, complex molds, but the vacuum level must be controlled carefully to avoid deformation of the core stack.
Pouring Height and Metal Pressure
One of the most important consequences of the stacked cold box process is the increase in pouring height. In the old resin sand pattern-plate process, the flask height was only about 500 mm. The mold could hold only one layer of ductile iron castings. In the new cold box stacked process, the total stack height is greater than 1,200 mm. The metal pressure at the bottom of the mold is now much higher.
The ferrostatic pressure can be expressed as:
$$\Delta p = \rho g H$$
where \(\rho\) is the density of the molten ductile iron, \(g\) is the gravitational acceleration, and \(H\) is the vertical height of the liquid metal above a given reference point. If the height is increased from 0.5 m to 1.2 m, the pressure ratio becomes:
$$\frac{\Delta p_2}{\Delta p_1}=\frac{H_2}{H_1}=\frac{1.2}{0.5}=2.4$$
This means that the bottom layers of the ductile iron castings in the stacked mold experience more than twice the metallostatic pressure that they experienced in the old low flask mold. Higher pressure is helpful for feeding solidification shrinkage, especially in ductile iron castings with heavy sections or with casting junctions.
High pouring height also improved our pouring operation. With the old low flasks, a 5-ton ladle was difficult to control at the beginning of the pour because the ladle was full and the pouring basin was low. The metal stream was difficult to keep steady, and if it was interrupted, the cold metal formed a cold shut. Near the end of the pour, the metal temperature in the ladle was lower, which made the problem worse. With the taller stack, the pouring basin is higher above the floor, and the pourer can maintain a continuous stream even when the ladle is relatively full. The number of cold-shut rejections decreased by more than 90% compared with the old process.
We can write the cold-shut rate as:
$$R_{\text{cs}}=\frac{N_{\text{cs}}}{N_{\text{poured}}}\times 100\%$$
where \(N_{\text{cs}}\) is the number of castings rejected because of cold shuts, and \(N_{\text{poured}}\) is the total number of ductile iron castings poured. The reduction in cold shut defects is:
$$\text{Reduction}=\frac{R_{\text{cs,old}}-R_{\text{cs,new}}}{R_{\text{cs,old}}}\times 100\% \approx 90\%$$
This was one of the most important quality improvements. Cold shuts are dangerous because they are often invisible on the surface and are only discovered after machining or pressure testing. Reducing cold shuts saved significant rework cost and improved the reliability of our small ductile iron castings.
Solidification and Shrinkage Behavior of Ductile Iron Castings
Ductile iron castings have a unique solidification behavior. As the molten metal cools, primary austenite forms, and the remaining liquid becomes enriched in carbon. When the eutectic temperature is reached, graphite nodules nucleate and grow. The formation of graphite is accompanied by an increase in volume. In a rigid mold, this graphite expansion can compensate for liquid shrinkage and reduce the need for large risers. In a soft mold, the expansion is lost through mold wall movement.
The net volume change during solidification of ductile iron castings can be described as:
$$\Delta V = V_{\text{cast}}\left(\beta_{\text{shrinkage}}-\beta_{\text{graphite}}\right)$$
where \(\beta_{\text{shrinkage}}\) represents the volumetric shrinkage coefficient of the liquid and solidified metal, and \(\beta_{\text{graphite}}\) represents the volumetric expansion coefficient associated with graphite precipitation. If the mold is rigid, the casting can be self-feeding to a large degree. The cold box stacked process with resin sand backup provides the rigidity needed for this self-feeding effect.
In the old resin sand pattern-plate process, the mold rigidity was not always sufficient. The side walls of the mold could yield when graphite expansion occurred, causing internal porosity. We therefore had to add larger risers to compensate. This reduced casting yield and increased the amount of remelted material. With the cold box stacked process, the high rigidity of the mold allowed us to reduce riser size and increase the yield of ductile iron castings.
Sand-to-Metal Ratio and Yield Improvement
The sand-to-metal ratio is one of the most important parameters in sand casting. It is defined as the weight of sand used in the mold divided by the weight of good castings produced. For the old resin sand pattern-plate process, the sand-to-metal ratio for our small ductile iron castings was typically:
$$R_{\text{s/m,old}}=\frac{W_{\text{sand,old}}}{W_{\text{castings}}}=5\sim 8$$
For the new cold box stacked process, the sand-to-metal ratio has decreased to:
$$R_{\text{s/m,new}}=\frac{W_{\text{sand,new}}}{W_{\text{castings}}}=2\sim 3$$
The reduction in sand-to-metal ratio can be expressed as:
$$\Delta R_{\text{s/m}}=\frac{R_{\text{s/m,old}}-R_{\text{s/m,new}}}{R_{\text{s/m,old}}}\times 100\%$$
Depending on the old ratio and the new ratio, the sand consumption has been reduced by roughly 40% to 75%. This is a very large saving because the sand system includes not only the sand itself, but also the binder, the reclamation energy, and the waste disposal cost. Lower sand consumption also means less mold weight and less labor required to handle the molds.
Casting yield is defined as the weight of the castings divided by the total weight of metal poured:
$$\eta=\frac{W_{\text{castings}}}{W_{\text{poured}}}\times 100\%=\frac{W_{\text{castings}}}{W_{\text{castings}}+W_{\text{riser}}+W_{\text{runner}}+W_{\text{sprue}}}\times 100\%$$
In the old process, each casting had its own runner and riser system. In the stacked cold box process, many castings share one central sprue and one horizontal runner per layer. This reduces the runner weight per casting and increases the casting yield. The higher yield reduces metal melting energy, inoculation cost, and finishing cost.
Machining Allowance Reduction
Another important advantage of the cold box process is the improvement in dimensional accuracy. The cold box core is made in a precision core box with a smooth surface. The dimensional repeatability of the resulting ductile iron castings is much better than that of castings made in resin sand pattern plates. Because of this improved accuracy, we were able to reduce the machining allowance from 4 to 5 mm down to 2 to 3 mm.
The machining allowance reduction has a direct effect on cost. The volume of metal removed by machining is approximately:
$$V_{\text{machining}} = A\cdot a$$
where \(A\) is the machined surface area and \(a\) is the machining allowance. If the allowance is reduced from 5 mm to 3 mm, the volume of chips is reduced by 40%. The mass of metal saved per casting is:
$$m_{\text{saved}}=\rho_{\text{iron}}\sum_{j} A_j \left(a_{\text{old},j}-a_{\text{new},j}\right)$$
where \(\rho_{\text{iron}}\) is the density of ductile iron, and \(A_j\) is the area of each machined surface. This saving reduces the load on the machining equipment, the tool wear, and the energy required for cutting. It also shortens the machining time and increases the output of the machine shop.
The reduced machining allowance was possible only because the cold box stacked process produced ductile iron castings with consistent dimensions. If the castings were distorted or shifted, a small machining allowance would lead to black spots or insufficient clean-up. The rigid stack and the precise core geometry eliminated most of these problems.
Shakeout, Identification, and Finishing Advantages
In the old resin sand pattern-plate process, each small ductile iron casting was separated from its mold individually. During shakeout, the casting often fell into the sand and was mixed with other castings. It was difficult to identify the part number, especially when several similar products were made in the same week. Some castings were lost until the shot-blasting machine was cleaned. Replacing lost castings was expensive and delayed customer shipments.
In the cold box stacked process, the castings remain attached to the central runner system as a tree. The entire tree is removed from the flask as one unit. We can identify the part number and the heat number on the runner system. The individual castings are then cut off from the tree. This completely solved the problem of lost castings. It also improved inventory tracking and reduced the risk of mixing different grades of ductile iron castings.
The flash problem was also improved. The cold box cores have tight parting lines, so the flash on the castings is much thinner and more regular. Thin flash is easier to grind and does not leave deep notches in the casting surface. The grinding operation is faster and the final appearance of the ductile iron castings is more uniform.
Shot blasting efficiency also increased. In the old process, each casting was blasted individually, and small castings often fell through the grates into the shot-blasting chamber. In the new process, the whole casting tree is blasted at once. The shot can reach almost all surfaces because the castings are attached only at the gate positions. This reduced the shot-blasting time per piece and increased the capacity of the shot-blasting machine.
Cost Comparison Between Old and New Process
The combination of lower sand-to-metal ratio, higher rigidity, higher yield, reduced machining allowance, fewer cold shuts, and lower labor cost leads to a substantial reduction in total manufacturing cost. Based on our internal records, the total production cost for the affected small ductile iron castings has been reduced by about 50% compared with the old resin sand pattern-plate process.
We can express the cost saving as:
$$C_{\text{saving}}=\frac{C_{\text{old}}-C_{\text{new}}}{C_{\text{old}}}\times 100\%\approx 50\%$$
where \(C_{\text{old}}\) is the total cost per casting using the traditional process, and \(C_{\text{new}}\) is the total cost per casting using the cold box stacked process. The cost saving comes from many sources, but the largest contributions are sand consumption, resin consumption, energy, fettling labor, and rework.
| Parameter | Original Resin Sand Pattern-Plate | Optimized Cold Box Stacked Process |
|---|---|---|
| Productivity | Only 6 molds per day | Multiple castings per mold; stack height above 1,200 mm |
| Sand-to-metal ratio | 5:1 to 8:1 | 2:1 to 3:1 |
| Machining allowance | 4–5 mm | 2–3 mm |
| Casting loss during shakeout | Frequent | Eliminated because castings remain on runner tree |
| Flash control | Difficult, thick flash | Thin and regular flash |
| Cold shut defects | Baseline | Reduced by more than 90% |
| Total production cost | Baseline | Reduced by about 50% |
| Pouring control | Difficult with low flasks | Easier with tall stack and stable stream |
| Shrinkage porosity | More frequent | Reduced by higher pressure head and rigid mold |
It is important to note that the cost saving of 50% is not achieved by simply reducing the price of any single material. It is achieved by changing the process architecture, so that many small ductile iron castings can be produced in the same flask, with less sand, less binder, less riser metal, less grinding labor, and less scrap. This type of process innovation is more valuable than piecemeal cost cutting.
Robotic Grinding of Ductile Iron Castings
The final bottleneck in many foundries is the cleaning and grinding operation. For small ductile iron castings, the flash, gate stubs, and feeder remnants are often removed manually. Manual grinding is slow, expensive, and dangerous. The work is noisy and produces fine metallic dust. It is difficult to find operators who are willing to perform manual grinding for many hours per day. As environmental and occupational health standards became stricter, we realized that manual grinding was limiting our production capacity.
We therefore introduced a robotic grinding cell for the finishing of small ductile iron castings. The robot is equipped with a grinding spindle and a positioner. The casting tree is cut into individual castings, each of which is loaded into a fixture. The robot then performs a programmed sequence of grinding operations to remove flash, gates, and other unwanted features. The robot uses force control to maintain consistent contact between the grinding wheel and the casting surface.
The improvement in productivity is very significant. Before using the robot, one skilled worker could grind about 20 pieces per day in an 8-hour shift. With the robotic grinding cell, the same type of small ductile iron castings can be ground at a rate of about 200 pieces per 12-hour shift. The average productivity can be expressed as:
$$P_{\text{manual}}=\frac{20\ \text{pieces}}{8\ \text{hours}}=2.5\ \text{pieces/hour}$$
$$P_{\text{robot}}=\frac{200\ \text{pieces}}{12\ \text{hours}}=16.7\ \text{pieces/hour}$$
The productivity improvement factor is:
$$\xi=\frac{P_{\text{robot}}}{P_{\text{manual}}}=\frac{16.7}{2.5}\approx 6.7$$
This means that the robotic grinding cell can do the work of roughly six to seven manual operators, while working a longer shift and maintaining the same level of quality. For our batch production of ductile iron castings, this is a very large improvement. It released capacity in the cleaning shop and removed the grinding bottleneck that has been limiting our overall output.
In addition to productivity, robotic grinding improves consistency. A manual operator may remove thin flash on one casting but leave a thicker fin on the next casting. The robot applies the same grinding program to every casting, so the final dimensions and surface finish are more repeatable. This is especially important for ductile iron castings that are later machined on automated machining lines. If the grinding consistency is poor, the casting will not locate properly in the machining fixture, and the machining allowance may be exceeded.
Robotic grinding also improves the working environment. The robot is enclosed in a cell with a dust extraction system. Operators are not exposed to the high noise of grinding wheels and the airborne iron dust. This helps us comply with health and safety regulations and makes it easier to retain workers in the foundry.
| Method | Shift Length | Output per Day | Output per Hour | Quality Consistency |
|---|---|---|---|---|
| Manual grinding | 8 hours | 20 pieces | 2.5 pieces/hour | Variable |
| Robotic grinding | 12 hours | 200 pieces | 16.7 pieces/hour | High and repeatable |
The robotic grinding system is also more flexible than a traditional fixed-purpose machine. When the production order changes from one type of ductile iron casting to another, the operator selects the corresponding program in the robot controller. The fixture may be changed quickly if necessary. The robot can handle a family of parts with similar geometry, which fits well with our small-batch production model.
Furthermore, robotic grinding reduces the risk of over-grinding. In manual grinding, an operator can accidentally remove too much metal from the casting surface, creating a recess that is difficult to repair. This problem is minimized in the robotic process because the grinding path and the force are controlled precisely. The amount of material removed is limited to the flash and gate stub thickness, while the base casting surface remains protected.
Practical Implementation and Lessons Learned
Converting from resin sand pattern-plate molding to cold box stacked molding required changes in tooling design, process control, and worker skill. We learned several important lessons during the implementation. First, the design of the core box must consider the stacking direction and the shape of the parting line. If the casting has deep external undercuts, the core box must be designed with movable slides, which increases the tool cost but is still economical for small-batch ductile iron castings.
Second, the gating system must be balanced for the entire stack. Because the stack can be more than 1,200 mm high, the lower cavities receive metal at a higher pressure than the upper cavities. We therefore designed the gates with a carefully chosen cross-section so that the metal flow distribution is as uniform as possible. The bottom gates are slightly smaller, while the upper gates are slightly larger, to equalize the filling of the different layers.
Third, the venting of the cores is essential. In a stacked mold, the gas generated by the resin sand must be able to escape. If the gas is trapped, it can cause gas holes in the ductile iron castings. We added vertical vent channels at the outer edges of the core stack. These vents connect to the atmosphere and allow the air and gas to escape as the metal rises.
Fourth, the rigidity of the backup material must be checked before every pour. If the resin sand backup is not fully cured, the stack can move when the flask is lifted. We established a standard curing time and a simple test method to confirm that the backup sand is hard enough. This is particularly important for ductile iron castings because of the graphite expansion pressure mentioned earlier.
Quality Results and Defect Reduction
The quality data collected after the introduction of the cold box stacked process shows consistent improvement across all the main defect categories of small ductile iron castings.
Cold Shut Defects
Cold shuts were one of the most frequent defects in the old process. They were caused by interrupted pouring, low pouring temperature, and inadequate venting. In the cold box stacked process, the taller pouring cup allows a continuous metal stream, and the bottom-gating system fills the mold smoothly. Cold shut rejections fell by more than 90%.
Shrinkage Porosity
Shrinkage porosity was reduced because of the higher metallostatic pressure and the rigid mold. The temperature gradient in the stack also improved feeding. The upper part of the stack remains slightly hotter than the lower part, which promotes directional solidification. The graphite expansion of the ductile iron is more effectively used to feed the casting because the mold walls do not yield.
Sand Inclusions
Sand inclusions decreased because the cold box cores are coated with a stable refractory coating. The coating separates the liquid metal from the sand and prevents the sand from being eroded by the metal flow. The bottom gating system also reduces the velocity of the metal at the gate, minimizing the erosion of the gate area.
Dimensional Variations
Dimensional variations decreased because the cold box cores have precise dimensions and the stack is rigidly supported. In the old resin sand pattern-plate process, the mold cavities could shift slightly during assembly and pouring. In the new process, the core stack is assembled with positive location features, so the relative position of each layer is fixed.
Lost Castings and Mixing
The problem of lost castings was completely solved. Since the castings remain attached to the runner tree until the final cutting operation, they cannot be lost in the shakeout. This also prevents accidental mixing of different grades of ductile iron castings, which is a serious risk when many similar small parts are produced in the same workshop.
Environmental and Energy Benefits
The process improvement also brings environmental benefits. The amount of new sand consumed per ton of good ductile iron castings is significantly lower. Since the sand-to-metal ratio dropped from 5:1–8:1 to 2:1–3:1, the amount of sand entering the reclamation plant has also dropped. This reduces the energy used in sand reclamation and the amount of waste sand sent to landfill.
The energy consumption for melting is also reduced because the casting yield is higher. In the old process, a large fraction of the poured metal became runner and riser scrap. Each kilogram of remelted scrap requires energy to melt it again. In the new process, the total runner weight is shared by many ductile iron castings, so the metal yield improves. This reduces the energy required per ton of finished castings.
The lower machining allowance also reduces energy in the machine shop. When 2 mm to 3 mm of material is removed instead of 4 mm to 5 mm, the machining time is shorter, the tool wear is lower, and the machine tool consumes less power. These savings are often overlooked, but they are significant for a foundry that supplies machined ductile iron castings to customers.
Robotic grinding also reduces energy and material waste. The robot operates in a more controlled way than a manual grinder, so fewer grinding wheels are broken and less metal is removed by accident. The dust collection system captures the grinding dust and prevents it from entering the environment. The work cell is more compact and safer than a manual grinding bench.
Production Planning and Delivery Performance
One of the main reasons for changing the process was to improve delivery performance. In the old resin sand pattern-plate process, small orders of ductile iron castings often had to wait for available flasks and molding boxes. The daily production rate of six molds was too low to support the required monthly output of 400 tons. When castings were lost during shakeout, we had to schedule an extra molding operation and an extra pouring operation, which delayed the entire production plan.
The cold box stacked process has changed this completely. Because one flask can contain dozens of small ductile iron castings, the number of flasks needed per order is greatly reduced. This means that the molding section can handle a larger number of orders in a shorter time. The fact that castings are not lost during shakeout means that we do not have to reserve extra capacity for replacement castings. The production plan is therefore more reliable.
In addition, the combination of different casting types in one stack allows us to make mixed orders in a single pouring operation. This is very useful for customers who order a small quantity of several different ductile iron castings. Instead of making one mold for each type, we make one stack that contains all the required types. This reduces setup time and improves the overall efficiency of the small-batch production system.
Production planning can be expressed as:
$$T_{\text{manufacturing}}=\frac{N_{\text{pieces}}}{N_{\text{stack}}\times M_{\text{stacks/day}}}$$
where \(N_{\text{pieces}}\) is the order quantity, \(N_{\text{stack}}\) is the number of castings in each stack, and \(M_{\text{stacks/day}}\) is the number of stacks produced per day. If the stack contains 10 castings and we produce 10 stacks per day, the daily output is 100 castings. In the old process, if each mold contained one casting and we produced 6 molds per day, the daily output was only 6 castings. This explains why the cold box stacked process is much more suitable for batch production of small ductile iron castings.
Tooling Design Considerations
The tooling design for the cold box stacked process is different from that of conventional resin sand pattern plates. Each core box must be designed with a vertical parting line, or at least a parting line that allows the core to be stripped without damaging the stack joints. The core box must also contain the runner and gate cavities, so that they are formed in the same core.
For small ductile iron castings, the tooling can be made of aluminum or cast iron. Aluminum is easier to machine and lighter to handle, while cast iron is more wear-resistant for long production runs. Since our production quantities are usually 30 to 500 pieces per order, aluminum tooling is often sufficient. The tooling cost is lower than that of a high-pressure die casting mold, and the tooling can be modified if the casting design changes.
The core box design must also include an ejection system. After the cold box sand core is cured, it must be ejected from the tool without breaking. We use ejection pins and sometimes a stripping plate, depending on the geometry of the core. The surface of the core box must be polished to produce a smooth core surface, which directly improves the surface finish of the ductile iron castings.
Process Control Parameters for Cold Box Cores
The quality of the cold box core depends on several process parameters. We control the sand temperature, the resin dosage, the catalyst dosage, and the gassing time. The sand temperature should be stable because high sand temperature can cause premature curing of the resin in the sand mixer. Low sand temperature can cause incomplete curing of the core.
The resin content is usually between 0.8% and 1.5% by weight of sand, depending on the required core strength. The cold box process uses a two-part resin system: Part I is a phenolic resin and Part II is a polyisocyanate. These two components are mixed with the sand before blowing. The amine gas is passed through the core to catalyze the reaction and to cure the binder.
The gassing time and purge time must be adjusted for each new core box. If the gas flow is too low, the core may be soft in the center. If the purge air is not sufficient, the amine gas may remain in the core and cause odor or casting defects. We document the optimal parameters for every core in a standard operating sheet.
The core strength can be tested by measuring the bending strength of a standard test specimen. We also perform a visual inspection of the core edges and surface. If a core has a broken edge, the resulting ductile iron casting will have a flash or a dimensional defect. We therefore handle the cores carefully after ejection and place them on flat support boards.
Coating Process Control
The alcohol-based coating used in our process is applied by dipping. This is faster than brushing, and it gives a more uniform coating thickness. However, the dipping process must be controlled carefully because the coating thickness depends on the immersion time, the withdrawal speed, and the viscosity of the slurry.
If the coating is too thick, it can block the fine details of the core and reduce the dimensional accuracy of the ductile iron castings. If it is too thin, it may not provide sufficient resistance to sand burn-on. We therefore maintain a viscosity control chart for the coating slurry and measure the density with a hydrometer. The coated cores are placed on a drying rack and ignited to burn off the alcohol.
The coating also improves the stripping of the sand from the casting after solidification. When the molten metal comes into contact with the coated core surface, the refractory coating forms a barrier that prevents the sand from fusing with the metal. This makes the shakeout process easier and improves the surface quality of the ductile iron castings.
Pouring Practice for Stacked Ductile Iron Castings
We use a 5-ton ladle for pouring the stacked molds. The first important rule is to skim the slag thoroughly before pouring. Ductile iron melt contains magnesium treatment products that form a viscous slag on the surface of the iron. If this slag is poured into the mold, it can create inclusions in the ductile iron castings. The tall pouring cup helps the slag float on the top of the sprue and stay in the pouring basin, but we still try to pour in a continuous manner.
The second rule is to keep the pouring basin full. Because the stack is tall, the metal must have enough pressure head to fill the upper cavities. If the pouring basin becomes empty, air will enter the sprue and the stack may stall. We trained the pouring operators to pour steadily and to avoid overfilling the basin.
The pouring temperature is selected according to the minimum wall thickness of the ductile iron castings. For small thin-walled ductile iron castings, we use a higher pouring temperature to avoid cold shuts. For heavier castings, we use a lower pouring temperature to avoid burn-on and reduce shrinkage. The cold box stacked process gives us a wide window of pouring temperatures because the filling is smooth and the metal pressure is high.
Solidification Modeling and Stack Design
During the development of the cold box stacked process, we used simple solidification modeling to check the feeding behavior of the stack. The modeling showed that the key to producing sound ductile iron castings is to ensure that the hottest metal remains at the top of the stack. This is because the top of the stack acts as an effective riser for the lower layers.
We use the modulus concept to evaluate the feeding requirements. The solidification modulus of a casting section is defined as:
$$M=\frac{V}{A_{\text{cooling}}}$$
where \(V\) is the volume of the section and \(A_{\text{cooling}}\) is the cooling surface area. A section with a higher modulus takes longer to solidify. To avoid shrinkage, the runner system and the feeding portions of the stack should have a higher modulus than the casting sections that they feed.
In the stacked mold, the central sprue has a very large modulus because it is surrounded by sand and is connected to multiple layers. This means that the sprue remains liquid longer than the individual castings. As the castings begin to solidify, liquid metal can flow from the sprue to compensate for shrinkage. This is similar to a large vertical riser. After solidification, the sprue is easily removed from the casting tree.
The use of the central sprue as a feeder reduces the need for individual risers on each casting. This is one of the reasons why the yielded metal increased and the cost decreased. For ductile iron castings, the graphite expansion also contributes to feeding, so the combined effect of the central sprue and the rigid mold is very positive.
Heat Treatment and Metallurgical Properties
The metallurgical quality of the ductile iron castings produced by the cold box stacked process is fully equivalent to that of castings produced by the conventional process. The nodule count in the microstructure is similar, and the ferrite/pearlite ratio can be controlled by the chemical composition and cooling rate. If the customer requires ferritic ductile iron castings, we perform an annealing heat treatment. If the requirement is pearlitic ductile iron castings, we control the cooling rate and adjust the copper or tin content.
The stacking process does not have a negative effect on graphite nodularity. The bottom-gating system avoids turbulence, which is important because turbulent filling can create slag and dross, and the nitrogen in the air can be absorbed by the melt. The quiet filling of the cold box stack helps maintain a high nodule count and a low inclusion content.
One potential concern with tall stacks is that the lower castings may cool more slowly because they are surrounded by hot sand and by other castings. In our product range, this has not caused any problem. The slow cooling actually improves the diffusion of carbon and helps reduce carbide formation in thin sections. Since the sand is a poor conductor of heat, the cooling rate in all layers is uniform enough to maintain consistent hardness.
We perform tensile tests and Brinell hardness tests on sample castings from each batch of ductile iron castings. The results are generally more consistent than before because the casting process is more stable. The dimensional accuracy and the microstructure stability enable us to predict the mechanical properties of the ductile iron castings with greater confidence.
Comparison of Core Making Capacity
The cold box core machine has a much higher production capacity than the old resin sand molding station. In the old process, each mold had to be made manually, rammed with sand, and allowed to cure. In the new process, the core machine blows and cures cores automatically in a cycle time of only a few minutes. The core production is no longer the bottleneck.
We can define the capacity of the core machine as:
$$C_{\text{core}}=\frac{3600}{t_{\text{cycle}}}$$
where \(t_{\text{cycle}}\) is the cycle time in seconds per core. If the core machine cycle time is 60 seconds, the capacity is 60 cores per hour. If each core layer contains 4 castings, the hourly production rate of cavities is:
$$C_{\text{cavity}}=C_{\text{core}}\times n_{\text{cavity}}$$
where \(n_{\text{cavity}}\) is the number of castings in each core. This multiplication of capacity is the fundamental reason why the cold box stacked process is so efficient for small ductile iron castings.
The old resin sand process was not only slow, but it was also physically demanding. The workers had to handle heavy flasks, mix sand, and manually compact the mold. The cold box process reduces the physical workload because the cores are lighter and easier to handle. This is particularly important for older workers and for retention of skilled labor in the foundry industry.
Future Improvements and Automation Potential
After the successful conversion to the cold box stacked process, we are now looking for further improvements. One area is the automation of core assembly. Currently, the cores are stacked manually. In the future, we plan to use a robot or a special stacking machine to place the cores into the flask. This will reduce the labor requirement and increase the consistency of assembly.
Another area is the automatic pouring system. The tall stacks are suitable for an automatic pouring furnace that can pour exactly the same amount of metal at the same rate. This would further reduce cold shuts and inclusions in ductile iron castings. Automatic pouring also eliminates the influence of operator fatigue on casting quality.
The robotic grinding cell will also be expanded. We plan to add a vision system to recognize the position of each ductile iron casting and to select the correct grinding program automatically. This will reduce the setup time when changing from one casting type to another. The vision system can also inspect the castings for major defects before grinding, so that defective castings are rejected earlier.
The sand reclamation system will be optimized to handle the lower sand volume and the different sand type used in the cold box process. Because the sand-to-metal ratio is much lower, the thermal load on the reclamation system is lower. This means that the reclamation plant can operate at a lower temperature and consume less energy. We are also studying the possibility of reusing the cold box sand as a core sand after thermal reclamation, which would reduce the amount of new sand purchased.
Conclusion
In conclusion, the optimization of our small-batch ductile iron castings production has been successful in every important aspect: productivity, quality, cost, and working environment. The old resin sand pattern-plate process was replaced by a cold box core-making process with stacked molds. This change allowed us to produce many small ductile iron castings in a single flask, with a lower sand-to-metal ratio and a much higher production efficiency.
The main conclusions of this work are as follows:
- The cold box stacked process is highly suitable for small ductile iron castings in order quantities of 30 to 500 pieces. It allows one mold to contain multiple identical castings or multiple different types, which shortens the production cycle and improves the delivery performance.
- The sand-to-metal ratio was reduced from 5:1–8:1 to 2:1–3:1. This reduced sand consumption, binder consumption, reclamation cost, and waste disposal cost.
- The machining allowance was reduced from 4–5 mm to 2–3 mm. This reduced machining time, tool wear, and material loss, while still ensuring that the final machined dimensions of the ductile iron castings are clean and fully machined.
- The cold shut defect rate was reduced by more than 90% because of the taller pouring height, the full-ladle pouring condition, the bottom-gating design, and the continuous metal stream.
- The loss of castings during shakeout was eliminated because the castings remain attached to the central runner as a tree. This reduced rework and prevented the accidental mixing of different part numbers.
- The rigid resin-sand backup around the stacked cores improved the feeding behavior of ductile iron castings by allowing graphite expansion to be used for internal feeding. Shrinkage porosity was therefore reduced.
- The total production cost was reduced by approximately 50%. This comes from the combined effect of all the improvements described above.
- The introduction of robotic grinding increased the grinding productivity by a factor of about 6.7, from 2.5 pieces per hour to 16.7 pieces per hour. The reject rate caused by over-grinding and inconsistent flash removal was also reduced.
- The robotic grinding system improved the consistency and quality of the final ductile iron castings and solved the labor shortage problem in the cleaning shop.
Our experience confirms that process innovation is not always about large, high-cost equipment. In the production of small ductile iron castings, a clever combination of cold box core making, stacked molding, and robotic finishing can bring enormous benefits. We believe this approach can be adapted by other foundries that produce small and medium batch ductile iron castings. The key is to understand the solidification behavior of ductile iron, to design the stacking gating system carefully, and to integrate the casting and cleaning processes as a complete production system. We will continue to refine this process and explore further automation in the future.