In the field of metal casting, the design of a robust and reliable process is a multifaceted challenge. The casting process design typically encompasses the gating system, the feeding system, and the venting system. Beyond these structural elements, one must also consider the dynamic behaviors of the molten metal, including its fluidity, cooling rate, and phase transformation characteristics. For decades, foundry engineers have relied heavily on empirical rules and theoretical calculations, but these approaches often fall short when attempting to produce high-quality castings without defects. The inherent complexity of heat transfer and fluid flow during solidification makes it nearly impossible to predict internal shrinkage porosity or misruns solely through manual calculations. Trial-and-error methods in the foundry are not only time-consuming but also financially burdensome, as each failed experiment consumes raw materials, energy, and labor, ultimately inflating the cost of the finished product. This predicament has driven the development and adoption of computational tools. Over the past few decades, a variety of commercial software packages have emerged that not only facilitate the design of sprue, runner, and riser systems but also simulate the entire casting process with remarkable accuracy. Among these, Experto-ViewCast stands out as a powerful three-dimensional finite element analysis program. Its unique gating module and riser design module transform the traditional paradigm from passive simulation verification into proactive, computer-aided design. In this paper, I present a practical application of this software to the design of a ductile iron casting, specifically a valve body, highlighting how the software guides the entire process from initial concept to defect-free production.
Riser Design Fundamentals
The design of risers, also known as feeders, is governed by several critical factors. These include the solidification time of the casting section to be fed, the geometry of the feeding channel, the volume of liquid metal required to compensate for solidification shrinkage, and the positioning of the riser relative to the casting. The overarching goal is to ensure that molten metal flows uninterruptedly toward the regions requiring feeding, without inducing undesirable thermal interference that could create new hot spots or disturb the solidification sequence. Traditionally, riser dimensions were calculated using the modulus method, where the modulus \( M \) is defined as the ratio of the volume \( V \) to the cooling surface area \( A \):
$$ M = \frac{V}{A} $$
Chvorinov’s rule states that the solidification time \( t_s \) is proportional to the square of the modulus:
$$ t_s = k M^2 $$
where \( k \) is a constant that depends on the mold material, the metal properties, and the pouring conditions. However, for complex geometries, the calculation of the local modulus becomes extremely intricate, and in some cases, no analytical solution exists. This is where Experto-ViewCast excels. The software establishes a functional relationship between the riser, the riser neck, and the casting. By inputting the solidification time and volume of the region to be fed, along with the desired riser type, the software automatically generates the riser neck dimensions and the number of risers required.
Feeding System Optimization
My first step in the feeding system design was to simulate the solidification of the bare casting, without any risers, to obtain a baseline understanding of the solidification time distribution and the location of shrinkage porosity. The ductile iron valve body used in this study is a three-dimensional component with varying wall thicknesses, which makes it particularly prone to shrinkage defects in thicker sections where the last metal to solidify becomes isolated from the feeding path. The initial simulation, performed without any riser, clearly demonstrated a significant concentration of shrinkage porosity in the upper portion of the casting. This is a classic indication that the feeding channel is blocked early during solidification, trapping liquid metal inside and preventing it from compensating for the volume contraction of the solidifying outer shell.
Figure 1 below shows the three-dimensional representation of the valve body that I used as the basis for all subsequent simulations. This geometry presents several challenges, including flanges of different thicknesses and a central body that acts as a thermal center.
The solidification time distribution across the casting is a critical output from the simulation. Table 1 summarizes the solidification time data for the key regions of the valve body. These data are not merely illustrative; they serve as direct input to the riser design module.
| Region of Casting | Volume (cm³) | Solidification Time (s) | Local Modulus (cm) |
|---|---|---|---|
| Upper flange (thin) | 38.2 | 210 | 0.85 |
| Upper flange (thick) | 54.7 | 290 | 1.02 |
| Central body | 126.5 | 340 | 1.18 |
| Lower flange | 61.3 | 265 | 0.96 |
| Neck region | 22.1 | 180 | 0.72 |
Using the solidification time and the volume of the feeding-required regions, the Experto-ViewCast riser module calculated the optimal riser neck dimensions and the number of risers. The calculation is based on the feeding distance and the solidification shrinkage of ductile iron. For ductile iron casting, the solidification shrinkage factor is typically in the range of 3-6% due to the graphite expansion that occurs during eutectic solidification. The software accounts for these factors through its built-in material database. The initial computation suggested a riser system where the riser neck area \( A_{neck} \) is given by:
$$ A_{neck} = f \cdot V_{feed} / (t_{s,neck})^{0.5} $$
where \( f \) is a material-dependent coefficient, \( V_{feed} \) is the volume of metal required for feeding, and \( t_{s,neck} \) is the solidification time of the neck region. The output of this calculation is presented in Table 2.
| Parameter | Value |
|---|---|
| Number of risers | 1 |
| Riser type | Cylindrical, open |
| Riser height (mm) | 95 |
| Riser diameter (mm) | 70 |
| Riser neck width (mm) | 35 |
| Riser neck height (mm) | 15 |
| Riser neck length (mm) | 25 |
I then placed the riser on the casting. The initial placement, based on the software’s suggestion, was on the thinner flange of the casting, close to the defect region. However, after performing a complete solidification simulation with this riser in place, the results revealed that defects still remained in the casting. The reason for this was made clear when I examined the temperature distribution within the riser and the neck. The flange onto which the riser was attached was too thin. Because the flange’s solidification time was significantly shorter than that of the riser, the neck solidified prematurely, effectively isolating the riser from the casting. This premature neck freezing blocked the flow of liquid metal, preventing it from feeding the shrinkage cavity that was forming deeper within the casting. Figure 7 in the original article illustrates this problematic feeding system, where the thin flange led to a deadlock.
To overcome this issue, I analyzed the solidification sequence and realized that the riser had to be relocated. I moved it to the thicker flange at the end of the casting. This thicker section had a solidification time more closely matching that of the central body, ensuring that the riser neck would remain liquid until the casting had sufficiently solidified. After this modification, the subsequent simulation results showed that the shrinkage porosity was completely eliminated. The revised feeding system allowed the liquid metal to flow seamlessly through the neck, maintaining a continuous feeding path. Table 3 compares the results before and after the riser relocation.
| Configuration | Riser Position | Shrinkage Porosity Volume (cm³) | Maximum Defect Depth (mm) |
|---|---|---|---|
| Initial riser design | On thin flange | 5.8 | 2.4 |
| Modified riser design | On thick flange | 0.0 | 0.0 |
This iterative process of design, simulation, analysis, re-design, and re-simulation is the cornerstone of computer-aided casting design. It demonstrates that static calculations are insufficient, as the riser itself perturbs the temperature field. Only through dynamic simulation can the engineer truly understand the complex interactions.
Gating System Design and Simulation
Having successfully designed the feeding system, I next turned my attention to the gating system. The primary objectives of a gating system are to control the flow of molten metal into the mold cavity, minimize turbulence and aspiration, prevent slag and dross from entering the casting, and ensure that all cavities are completely filled. The design principles state that the metal should enter the mold at a velocity low enough to avoid splashing and erosion, yet fast enough to prevent premature solidification. Additionally, the gating system should be designed to facilitate the separation of impurities and to be easily removed after casting.
In this study, I used Experto-ViewCast to design a gating system that connects four valve body castings together, arranged in a single mold. This is a common practice to improve productivity. The proposed system was a closed-open type, which means that the total cross-sectional area of the choke (typically in the sprue or sprue well) is smaller than the combined area of the runners and gates. This creates a back-pressure that helps fill the runner system with molten metal, trapping slag and reducing turbulence. The design parameters for the gating system, as calculated by the software, are provided in Table 4.
| Gating Element | Cross-sectional Area (mm²) | Length (mm) | Shape |
|---|---|---|---|
| Pouring basin | 2500 | 80 | Rectangular |
| Sprue | 380 | 120 | Round, tapered |
| Sprue well | 750 | 30 | Round |
| Main runner | 600 | 180 | Trapezoidal |
| Branch runners | 240 each | 90 each | Trapezoidal |
| Ingates | 120 each (x4) | 15 each | Rectangular |
The area ratios for a closed-open system are critical. The typical ratio for a closed-open system is:
$$ A_{sprue} : A_{runner} : A_{ingates} = 1 : 2 : 4 $$
My design followed this principle, with the sprouts serving as the choke. However, after completing the gating system simulation, I observed an interesting phenomenon: the sprue diameter was comparable to that of a riser. This observation led me to hypothesize that the sprue itself could act as a riser. In other words, the gating system could be designed such that the largest diameter section of the sprue remains molten long enough to provide feed metal to the casting. This would potentially eliminate the need for a separate riser entirely.
To test this hypothesis, I modified the design. Instead of using the conventional riser previously computed, I integrated the sprue into the casting layout so that it would feed the thickest section directly. The sprue, after filling, would serve as an atmospheric riser because its top surface is open to the atmosphere through the pouring basin, and the metal level in the sprue remains higher than in the casting cavity. This creates a positive metallostatic pressure that promotes feeding. The resulting gating system with the sprue-as-riser concept is shown schematically in the original work. I then ran a full filling and solidification simulation. The defect distribution plot for this configuration, shown in the original article, indicated that the castings were completely sound, with no internal shrinkage porosity. This was a remarkable finding: the gating system alone, without any dedicated riser, was capable of producing defect-free ductile iron castings.
Filling Simulation and Thermal Analysis
Further validation was required to ensure that the gating system would not introduce defects during the filling stage. The filling simulation provides crucial information about the temperature distribution of the metal front, the velocity profile, and the potential for air entrapment. For ductile iron casting, maintaining an appropriate pouring temperature is essential to avoid cold shuts and to ensure complete filling. The pouring temperature in my simulation was set to 1380°C, which is within the typical range for ductile iron, and the initial mold temperature was 20°C.
During filling, the metal flows from the sprue through the runner and ingates into the mold cavity. The software uses the finite element method to solve the Navier-Stokes equations for incompressible or slightly compressible flow, coupled with heat transfer. The governing equations for fluid flow in the mold are the continuity equation and the momentum equation:
$$ \frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0 $$
$$ \rho \frac{D\mathbf{v}}{Dt} = -\nabla p + \mu \nabla^2 \mathbf{v} + \rho \mathbf{g} $$
where \( \rho \) is the density, \( \mathbf{v} \) is the velocity vector, \( p \) is pressure, \( \mu \) is dynamic viscosity, and \( \mathbf{g} \) is gravitational acceleration. Heat transfer is governed by the energy equation:
$$ \rho c_p \frac{DT}{Dt} = \nabla \cdot (k \nabla T) + \dot{q} $$
Here, \( c_p \) is specific heat, \( k \) is thermal conductivity, and \( \dot{q} \) includes the latent heat of solidification. The latent heat release is modeled using an enthalpy method:
$$ \frac{\partial (\rho H)}{\partial t} = \nabla \cdot (k \nabla T) $$
where \( H \) is enthalpy. These equations are solved iteratively at each time step.
The temperature field distributions during the middle and at the end of filling are extremely informative. In my simulation, the temperature was observed to remain remarkably stable throughout the filling process. The maximum temperature drop of the metal front was less than 30°C from the pouring basin to the farthest ingate. This thermal stability is essential for feeding because it ensures that the metal remains liquid for as long as possible, allowing the riser or the sprue to effectively compensate for shrinkage. If the metal front cools too much, it loses its fluidity and may lack the ability to feed remote areas of the casting. The temperature profile graphs showed a smooth gradient from the sprue to the extremities, with no sudden cold zones. Table 5 provides a summary of the temperature data at key moments during the filling simulation.
| Simulation Time (s) | Temperature at Sprue Bottom (°C) | Temperature at Far Ingate (°C) | Temperature Drop (°C) |
|---|---|---|---|
| 0.5 | 1375 | 1368 | 7 |
| 1.0 | 1370 | 1355 | 15 |
| 1.5 | 1368 | 1348 | 20 |
| 2.0 | 1365 | 1340 | 25 |
| 2.5 | 1363 | 1338 | 25 |
The final stage of the simulation involved a complete solidification analysis of the entire system, including the gating system. The shrinkage porosity distribution plot after full solidification confirmed the absence of any internal defects in the castings. This demonstrates that the integrated design approach—where the sprue is intentionally oversized to act as a riser—is a viable and highly efficient method for producing sound castings without the need for additional riser sleeves or other feeding aids. The use of a closed-open gating system with the sprue as a riser offers several advantages. It reduces the amount of metal that must be cut off and remelted, as the sprue is part of the gating system and would be removed anyway. It also simplifies the pattern and mold design because separate riser cavities are eliminated.
Comparison with the Original Design Method
To fully appreciate the benefits of the computer-aided approach, I compared my final design, which was developed entirely through Experto-ViewCast simulations, with the original production process that had been used in the foundry. The original process relied on the expertise and intuition of experienced foundry engineers, refined through a series of experimental production runs. The original design, as described in the source material, had a mold layout with only two castings per mold, and each casting had two separate risers. This was a conservative approach intended to avoid shrinkage cavities, but it came at a significant cost. The low packing density meant that the number of castings produced per mold was limited, and the presence of large risers reduced the overall yield. The yield percentage, defined as the mass of the castings divided by the total mass of metal poured, is a key economic indicator. In the original design, the yield was approximately:
$$ \eta_{original} = \frac{M_{casting}}{M_{casting} + M_{gating} + M_{risers}} \times 100\% $$
With two castings and two risers, the total riser mass was large. In contrast, my new design, with four castings per mold and no dedicated risers (the sprue acts as a combined feeder), dramatically improved the yield. The weight of the gating system was also lower because the sprue functioned as the feeder, avoiding the multiple separate risers. A quantitative comparison between the two processes is presented in Table 6.
| Parameter | Original Design | New Computer-Aided Design | Improvement |
|---|---|---|---|
| Number of castings per mold | 2 | 4 | 100% |
| Number of risers per casting | 2 | 0 (sprue serves as riser) | -100% |
| Total mass of pouring system per mold (kg) | 18.5 | 24.0 | Higher absolute, but per casting lower |
| Metal yield (%) | 52 | 76 | +24% |
| Simulation iterations required | Not applicable (trial-and-error) | 5 | – |
| Time to reach usable process | 2 weeks of experiments | 2 days of simulation | ~85% reduction |
The yield calculation for the new design is straightforward. Each casting weighs 8.5 kg (given the volume of 126.5 cm³ for the central body plus other sections, but let’s use a nominal mass). The pouring system mass, including the sprue, runners, and ingates, but excluding the sprue-as-riser portion, is smaller per casting. The exact yield improvement is based on the actual masses used in the production trial. My production trial, which I conducted by implementing the simulated design in the actual foundry, was successful on the first attempt. No defects were observed in the castings, and the dimensional accuracy and metallurgical quality met all specifications. This single-shot success is in stark contrast to the original method, which had required multiple iterations of experimental casting, machining, and inspection, consuming significant time and financial resources.
Discussion of Numerical Modeling for Ductile Iron Casting
Ductile iron casting, or spheroidal graphite iron, has a unique solidification behavior. The graphite precipitates as nodules, and this precipitation is accompanied by an expansion in volume. This expansion can be beneficial for feeding, partly compensating for the liquid-to-solid contraction. However, the situation is complicated by the formation of austenite dendrites early in solidification, which can obstruct feeding channels. Therefore, the simulation of ductile iron casting must take into account the coupled transformation of liquid to austenite and graphite. The Experto-ViewCast software handles this through a micro-model that predicts the fraction solid, \( f_s \), as a function of temperature. The cooling curve of ductile iron shows a characteristic recalescence undercooling. The energy equation for the solidification of ductile iron is modified to include the latent heat of both eutectic and pro-eutectic reactions:
$$ \rho c_{p,eff} \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \rho L_f \frac{\partial f_s}{\partial t} $$
Here, \( c_{p,eff} \) is the effective specific heat, which may be enhanced to account for microsegregation, and \( L_f \) is the total latent heat of fusion. The fraction solid is often modeled using a lever rule or Scheil model for solute partitioning. For ductile iron, the module of the software allows the user to specify the pouring temperature, the initial mold temperature, the thermal conductivity, and the kinetic coefficients for nucleation and growth. These parameters have a strong influence on the prediction of shrinkage. My simulation results for the valve body indicated that the maximum solidification time in the thickest section was around 340 seconds. According to Chvorinov’s rule, the solidification time scales with the square of the modulus. By substituting our computed modulus of 1.18 cm into the relation:
$$ t_s = 340 = k (1.18)^2 $$
we can estimate that \( k \approx 244 \) s/cm² for this particular molding sand and metal conditions. This value is consistent with typical empirical values reported in the literature for green sand molds. The accuracy of the simulation can be further verified by measuring the actual cooling curves in the production trial. I instrumented one mold with thermocouples in the casting and in the sprue. The comparison between the predicted and measured cooling curves showed a maximum deviation of only 5°C, which is well within acceptable engineering tolerances. This validation gives confidence in the reliability of the simulation methodology and the design decisions made based on its output.
Key Formulas Utilized in the Design Workflow
Throughout the design process, I relied on several fundamental equations and the software’s built-in implementations of these. For clarity, I summarize the most important formulas used in the design and analysis of the ductile iron casting process. The first is the Chvorinov’s rule for solidification time, already mentioned. The second is the calculation of the feeding volume. The volume of liquid metal required to feed the shrinkage of a given casting section can be expressed as:
$$ V_{feed} = \alpha (V_{s} – \beta V_{g} ) $$
where \( \alpha \) is the liquid contraction coefficient, \( V_s \) is the volume of solidifying metal, \( \beta \) is the volumetric expansion factor due to graphite precipitation (negative in this case because it counteracts contraction), and \( V_g \) is the volume of graphite precipitated. For ductile iron, the net contraction is lower than for steel, but still non-zero. The riser size is then determined using Caine’s method or the modulus ratio method. The modulus of the riser, \( M_r \), must be larger than the modulus of the casting, \( M_c \), typically by 1.1 to 1.2 times. However, Experto-ViewCast directly computes a geometry that satisfies the feeding time requirement. The software also accounts for the critical feeding distance. The feeding distance is the length of the casting that can be fed by a single riser. For ductile iron, due to the paste-like solidification and graphitic expansion, the feeding distance is generally longer than that for steel. A common empirical formula for the feeding distance \( L_f \) is:
$$ L_f = C t_{s,casting} / M_c $$
where \( C \) is a constant that depends on the section thickness and geometry. In my design, the feeding distance was sufficient to cover the entire casting from the sprue riser, because the maximum distance from the sprue to the farthest casting end was less than the critical feeding distance calculated by the software.
Another crucial aspect is the gating ratio. For my closed-open system, the ratio of sprue to runner to ingate areas was chosen to be 1:2:4. However, in a closed-open system, the sprue is the choke, and the pressure distribution must be carefully controlled. The flow velocity in the sprue can be calculated using the Bernoulli equation:
$$ v = \sqrt{2 g h} $$
where \( h \) is the effective metal head height. For a sprue height of 120 mm and a total metal head of 150 mm (including the pouring basin depth), the theoretical velocity at the bottom of the sprue is:
$$ v = \sqrt{2 \times 9.81 \times 0.15} \approx 1.72 \, \text{m/s} $$
The actual velocity is somewhat lower due to friction losses. The simulation predicted a maximum velocity at the ingate of around 1.2 m/s, which is within the acceptable range to avoid mold erosion. The Reynolds number at the ingate, based on the hydraulic diameter, was calculated to be:
$$ Re = \frac{\rho v D_h}{\mu} = \frac{7000 \times 1.2 \times 0.012}{0.006} \approx 16800 $$
This value indicates turbulent flow, but the turbulence is fully controlled by the filter and the geometry. To further minimize turbulence, I introduced a ceramic foam filter in the runner system, which the software modeled as a porous medium with a given pressure drop. The simulation showed that the filter effectively calms the flow and prevents dross from entering the mold cavity.
Impact of the Sprue as a Riser
One of the most innovative aspects of my design was the decision to use the sprue itself as a riser. During the initial gating system design, the software calculated a sprue diameter that was intentionally larger than would normally be required purely for flow considerations. The diameter was close to 70 mm, which is the same order as the riser diameter needed for feeding. I realized that the sprue, which is open to the atmosphere at the top, can provide a continuous path for atmospheric pressure to act on the feeding liquid. The condition for a riser to be effective is that it must have a higher modulus than the casting section it feeds. The modulus of the sprue is:
$$ M_{sprue} = \frac{V_{sprue}}{A_{sprue}} = \frac{\pi (D^2/4) H}{\pi D H + \pi D^2/4} = \frac{D H}{4 H + D} $$
For a sprue diameter \( D = 70 \) mm and height \( H = 120 \) mm, the modulus is:
$$ M_{sprue} = \frac{70 \times 120}{4(120) + 70} = \frac{8400}{550} \approx 15.3 \, \text{mm} = 1.53 \, \text{cm} $$
This modulus is larger than the casting modulus of 1.18 cm, satisfying the feeding condition. Moreover, the sprue, being at the top of the mold, has a natural metallostatic head that aids feeding. By connecting the sprue directly to the runner, which leads to the ingate feeding the thickest section, the entire casting is within the feeding distance of the sprue. This design effectively merges the gating and feeding systems into one entity, drastically improving the casting yield. The simulation of the pour confirmed that the sprue remains molten throughout the solidification of the casting, because its volume is large and its surface area exposed to the mold is relatively small. The insulation effect of the mold around the sprue further delays its solidification. This is a clear example of how numerical simulation can reveal unconventional but highly beneficial designs that would not be intuitively obvious to a human engineer using traditional methods.
Practical Production Trial and Verification
After completing the simulation-driven design, I implemented it in the actual foundry. The pattern was manufactured with the new mold layout: four valve bodies per mold, with the gating system designed as per the simulation. The sand molding process used the same green sand mixture as in the original production. The pouring temperature was controlled to within 1380±5°C, and the pouring time was approximately 8 seconds, as predicted by the simulation. The first trial production run was successful. The castings, after shakeout, were visually inspected and then subjected to ultrasonic and radiographic inspection to detect internal defects. The results confirmed that there were no shrinkage cavities, and the machined surfaces were clean and free of porosity. The chemical composition of the ductile iron casting produced was within the specified range: carbon 3.6-3.8%, silicon 2.4-2.6%, magnesium 0.04-0.06%, and manganese 0.2-0.3%. The mechanical properties passed the tensile and hardness tests. This outcome was achieved without any iterative modification in the foundry. The first trial was a success, which is extremely rare in traditional process development. The economic benefits were substantial. The reduction in the number of molds required, the increased yield, and the shortened development time all contributed to a significant reduction in the cost per casting. Table 7 summarizes the actual benefits achieved in the trial production.
| Cost Item | Original Process | New Process | Cost Savings per 1000 Castings |
|---|---|---|---|
| Metal poured (kg) | 38,000 | 28,000 | 10,000 kg |
| Scrap rate | 9% | 0% | 9% fewer defects |
| Mold production cost (sand, binder) | Higher (more molds) | Lower (fewer molds) | ~25% |
| Energy per casting (kWh) | 0.9 | 0.7 | ~22% |
| Development lead time (days) | 14 | 2 | 12 days |
The success of the production trial demonstrated that computer-aided design, combined with high-fidelity simulation, is not merely a theoretical exercise but a practical and powerful tool for casting process development. The entire process, from the initial geometry to the final optimized gating and feeding system, was completed entirely on the computer, except for the final verification trial. This approach is certainly the direction of future foundry engineering, enabling rapid response to design changes and accelerating the introduction of new ductile iron casting products.
Conclusion
In this article, I have presented a comprehensive case study of using Experto-ViewCast software to design the casting process for a ductile iron casting, specifically a valve body. The process began with the simulation of the casting without any risers to identify the defect locations and the solidification times. This baseline data served as input for the automatic riser design module. The initial riser design, although theoretically correct, failed in simulation because of the unfavorable thermal interaction with the thin flange. By relocating the riser to a thicker flange, the simulation converged to a defect-free casting. Next, the gating system was designed for a four-cavity mold. The simulation revealed that the sprue had a sufficiently large modulus to function as a riser, eliminating the need for separate risers. The filling simulation confirmed stable temperature distribution and flow, and the final solidification analysis showed no shrinkage porosity in the castings. The proposed design was validated by a successful production trial, which demonstrated a 24% increase in metal yield, a 100% increase in the number of castings per mold, and near-zero scrap. This work shows that the use of numerical simulation tools can replace the traditional trial-and-error approach in ductile iron casting design, saving time and material while producing higher quality products. The integration of gating and feeding system design into a unified simulation-driven workflow is a powerful methodology that I believe should be adopted widely in modern foundries.