In my research, I focused on the rapid sand mold casting of an engine cylinder head made from ZL105 aluminum alloy. The cylinder head is one of the most complex and critical components in an internal combustion engine. It must withstand high thermal and mechanical loads, maintain accurate dimensions, and provide reliable gas exchange, cooling, and lubrication paths. Because the component has thin walls, many internal cavities, and strict quality requirements, conventional trial-and-error development is slow and expensive. I therefore combined three-dimensional modeling, casting process CAD, numerical simulation, and rapid prototyping to build a practical route for rapid sand mold casting. My main objective was to design a reliable gating and feeding system, obtain accurate interfacial heat transfer conditions, simulate filling and solidification, and finally produce a qualified cylinder head by rapid sand mold casting.
The use of aluminum alloys in automobiles has grown continuously because of the need for weight reduction, lower fuel consumption, and reduced emissions. Aluminum castings offer low density, good strength, excellent thermal conductivity, and acceptable corrosion resistance. In engine components, aluminum cylinder heads and blocks have become standard in many passenger vehicles. The following table summarizes typical applications and characteristics of aluminum castings in automobiles.
| Component | Application Level | Main Characteristics |
|---|---|---|
| Aluminum cylinder block | More than 45% | Low cost, good lightweight effect |
| Aluminum cylinder head | More than 90% | Good heat dissipation, complex internal cavities |
| Aluminum radiator | More than 90% | High durability and strong heat transfer ability |
| Aluminum wheel | About 80% abroad, about 50% in China | Good balance, variable styling, sufficient strength |
| Chassis parts | Increasing | Light weight, long service life |
ZL105 alloy belongs to the Al-Si-Mg system. It has good casting performance, high strength, and favorable hot plasticity. The liquidus temperature is about 622 °C and the solidus temperature is about 536 °C. These thermal properties strongly influence the filling and solidification behavior during rapid sand mold casting. I selected ZL105 because it is suitable for gravity casting of complex automotive components and because its properties are well matched to resin-bonded sand molds.
Rapid sand mold casting is a foundry method that uses rapid prototypes, such as stereolithography apparatus (SLA) patterns, instead of conventional wooden or metal patterns. The process combines casting process CAD, rapid prototyping, and traditional sand molding. In my work, the general route was as follows: I first created a three-dimensional model of the cylinder head, then used a casting process CAD system to design the parting surfaces, sand cores, gating system, risers, and process parameters. After that, I imported the model into ProCAST for numerical simulation. Finally, I selected molding materials, produced the SLA patterns, assembled the sand cores, and poured the casting. This route shortened the design cycle and improved the overall efficiency of rapid sand mold casting.
The cylinder head used in my study had overall dimensions of approximately 426 mm × 200 mm × 131 mm, a volume of about 4581 cm³, and a mass of about 12.4 kg. It is a thin-walled box-type part with a minimum wall thickness of about 4 mm. For single-piece or small-batch production, a dimensional tolerance grade of about 10 CT is acceptable. The internal structure includes water jackets, oil galleries, intake ports, and exhaust ports. These features require carefully designed sand cores and accurate core assembly. The main sand cores I designed included lower left oil gallery cores, lower right oil gallery cores, water jacket cores, upper oil gallery cores, intake port cores, and exhaust port cores. Each core had to be strong enough for handling, positioned accurately, and easy to remove after casting.
When designing the parting and core scheme for rapid sand mold casting, I followed several practical rules. First, every sand core must be easy to manufacture and remove. Second, the number of parting surfaces should be minimized to preserve dimensional accuracy. Third, flat parting surfaces are preferred because they reduce deformation and simplify pattern making. For the cylinder head, I divided the internal cavities into separate cores and then assembled them with locating features. The lower oil gallery cores were placed on the base core. The intake and exhaust cores were located by core prints and side positioning blocks. The water jacket core and lower oil gallery cores were connected into an integrated assembly. The upper oil gallery core was positioned on the mold by three-point location. Core bones were used in slender sections to increase strength.
The gating system is a key part of rapid sand mold casting. It must guide molten metal into the mold cavity smoothly, avoid sand erosion, prevent turbulence and air entrapment, and promote directional solidification. I considered three main pouring positions: single-side top pouring, symmetrical top pouring, and single-side bottom pouring. The gating system was designed as an open system with the cross-sectional ratio:
$$
A_{\text{sprue}} : A_{\text{runner}} : A_{\text{ingate}} = 1 : 2 : 4
$$
This type of system has its throttling section at the sprue, which helps the metal fill the mold smoothly and reduces the impact on the sand mold. For non-ferrous alloys such as ZL105, an open gating system is often preferred because it lowers the risk of oxidation and turbulence. The pouring time was estimated by:
$$
t = \delta \sqrt{G_L}
$$
where \(t\) is the pouring time, \(G_L\) is the mass of liquid metal passing through the throttling section, and \(\delta\) is an empirical coefficient. The total ingate area was calculated from:
$$
A_{\text{ingate}} = \frac{G_L}{\rho_L \mu t \sqrt{2gH_p}}
$$
where \(\rho_L\) is the density of the liquid metal, \(\mu\) is the flow loss coefficient, \(g\) is the gravitational acceleration, and \(H_p\) is the average static pressure head. For a three-element gating system, the average static pressure head can be expressed as:
$$
h_p = \frac{k_2^2}{1+k_2^2}H_p
$$
where \(k_2\) is the area ratio between the sprue and the runner. For a four-element system, an additional term is included. I used a flat ingate cross-section and a trapezoidal runner cross-section. The flat ingates helped distribute the metal evenly, while the trapezoidal runner reduced heat loss and was easy to machine in the pattern.
I compared three gating layouts in detail. The following table summarizes their advantages and disadvantages for rapid sand mold casting.
| Scheme | Pouring Position | Advantages | Disadvantages |
|---|---|---|---|
| Scheme 1 | Single-side top pouring | Good filling ability; ingates at the top favor directional solidification and feeding; simple mold construction. | Strong impact; possible splashing, oxidation, and air entrapment; sand holes, gas pores, and inclusions may form. |
| Scheme 2 | Symmetrical top pouring | More uniform ingate distribution; liquid level rises more smoothly than in Scheme 1. | Metal streams meet in the middle; inclusions, gas entrapment, and turbulence are likely. |
| Scheme 3 | Single-side bottom pouring | Avoids direct impact on the sand mold; liquid level rises relatively smoothly; fewer turbulence-related defects. | Horizontal static pressure may be insufficient; risk of misrun if the pouring temperature is too low. |
Riser design was based on the modulus method. The modulus of a casting section is:
$$
M = \frac{V}{A}
$$
where \(V\) is the volume and \(A\) is the heat-transfer surface area. According to Chvorinov’s rule, the solidification time is:
$$
t_c = \left(\frac{M_c}{K_c}\right)^2
$$
for the casting, and:
$$
t_r = \left(\frac{M_r}{K_r}\right)^2
$$
for the riser. To ensure effective feeding, the riser must solidify later than the casting section it feeds, so:
$$
t_r \geq t_c
$$
I selected standard kidney-shaped open risers located in the upper mold. Two riser types were placed according to the feeding requirements of the thick sections. The exhaust system was also important. Because resin-bonded sand releases gas during pouring, I designed vent holes and vent plugs in the cores and mold surfaces. These vents allowed gas to escape without causing blowholes or backpressure defects.
The process parameters for rapid sand mold casting were set as follows. The machining allowance was 4 mm for general surfaces and 5 mm for the top surface connected to the risers. The linear shrinkage of the casting was taken as 1%. These values were chosen to ensure dimensional accuracy after cooling and to provide enough material for subsequent machining.
After the casting process design, I imported the three-dimensional model into ProCAST. The mesh was generated with tetrahedral elements. The element size was controlled so that the minimum wall thickness contained at least two to three elements. For the three schemes, the mesh statistics were approximately as follows.
| Scheme | Number of Nodes | Number of Elements |
|---|---|---|
| Scheme 1 | 107,385 | 471,675 |
| Scheme 2 | 107,035 | 470,289 |
| Scheme 3 | 115,990 | 508,256 |
Mesh quality was evaluated using dihedral angle, radius ratio, and edge length ratio. The ideal tetrahedral element has a dihedral angle close to 60°, a radius ratio close to 1, and an edge length ratio close to 1. I optimized the mesh until no severely distorted elements remained. This step was necessary because poor mesh quality can cause inaccurate temperature and flow predictions in rapid sand mold casting simulation.
In the pre-processing stage, I used a virtual mold to reduce computation time. The virtual mold was defined as a rectangular box surrounding the casting. The mold material was set as phenolic urethane resin sand, and the interface heat transfer coefficient was assigned between the casting and the mold. The outer boundary of the virtual mold was treated as adiabatic. The following table lists the initial and boundary conditions I used.
| Parameter | Value or Range |
|---|---|
| Pouring temperature | 690–720 °C |
| Pouring rate | 0.75–1.5 kg/s |
| Casting initial temperature | 690–720 °C |
| Mold initial temperature | 20–30 °C |
| Gravity | 9.8 m/s² |
| Interfacial heat transfer coefficient | 200–1000 W/(m²·K) |
To improve the accuracy of the simulation, I determined the interfacial heat transfer coefficient between ZL105 aluminum alloy and phenolic urethane resin sand by an inverse method. This coefficient is one of the most important boundary conditions in rapid sand mold casting. It depends on surface temperature, air gap formation, mold material, coating, and casting geometry. Direct measurement is difficult, so I used measured temperature data and a nonlinear inverse approach.
The experimental model was simplified as a one-dimensional heat-transfer problem. A cylindrical casting with a symmetric structure was used. Thermocouples were placed in the casting and the sand mold. One thermocouple measured the casting temperature near the interface, and four thermocouples measured the mold temperature at distances of 5 mm, 10 mm, 15 mm, and 20 mm from the interface. The casting material was ZL105 aluminum alloy, and the mold material was phenolic urethane resin sand. The top and bottom of the experimental assembly were insulated to minimize heat loss to the environment.
The heat conduction equation in cylindrical coordinates is:
$$
\frac{\partial T}{\partial t} = \alpha \left( \frac{\partial^2 T}{\partial r^2} + \frac{1}{r}\frac{\partial T}{\partial r} \right)
$$
where \(T\) is temperature, \(t\) is time, \(r\) is radial distance, and \(\alpha\) is thermal diffusivity. The interfacial heat flux and heat transfer coefficient are related by:
$$
h(t) = \frac{q(t)}{T_1 – T_2}
$$
where \(q(t)\) is the heat flux across the casting–mold interface, \(T_1\) is the casting surface temperature, and \(T_2\) is the mold surface temperature. I used Beck’s nonlinear inverse method to estimate the heat flux from the measured temperatures. The finite-difference form of the one-dimensional heat conduction equation can be written as:
$$
T_i^{p+1} = T_i^p + F_0 \left( T_{i+1}^p – 2T_i^p + T_{i-1}^p \right)
$$
where \(F_0\) is the Fourier number:
$$
F_0 = \frac{\alpha \Delta t}{\Delta r^2} \leq 0.5
$$
The inverse procedure updated the heat flux iteratively until the calculated temperatures matched the measured temperatures within a specified tolerance. The convergence criterion was:
$$
\left| \frac{\Delta q}{q} \right| \leq \varepsilon
$$
The thermal properties of ZL105 aluminum alloy varied with temperature. The solid fraction changed from 1 to 0 between the solidus and liquidus temperatures. The density decreased as temperature increased, with a steeper slope in the mushy zone. The enthalpy and thermal conductivity also changed significantly. These temperature-dependent properties were included in the simulation to improve accuracy.
The measured cooling curves showed that the casting center cooled from about 690 °C to about 635 °C in the first 60 s. After that, cooling became slower. The casting temperature reached the solidus temperature of 536 °C at about 2300 s. The mold temperature at 5 mm from the interface rose rapidly and reached a maximum of about 571 °C at 300 s. The temperatures at 10 mm, 15 mm, and 20 mm followed similar trends but with smaller amplitudes and longer delays. These data were used to calculate the heat flux and interfacial heat transfer coefficient.
The estimated heat flux increased rapidly in the first 60 s and reached a maximum of about 256,000 W/m². It then decreased quickly to about 4,300 W/m² at 500 s and gradually stabilized. The interfacial heat transfer coefficient between ZL105 aluminum alloy and phenolic urethane resin sand was then obtained. The results can be summarized as follows.
| Temperature Range | Interfacial Heat Transfer Coefficient |
|---|---|
| Above liquidus, about 690 °C | About 811 W/(m²·K) |
| At liquidus, about 622 °C | About 790 W/(m²·K) |
| Mushy zone, 622–536 °C | Decreasing rapidly |
| At solidus, about 536 °C | About 310 W/(m²·K) |
| After solidification, below 536 °C | Stabilizing near 230 W/(m²·K) |
I validated the inverse result by comparing measured and simulated temperatures. The calculated casting center temperature was slightly lower than the measured value, and the calculated mold temperature was slightly higher than the measured value. The maximum deviation was about 11 °C, which is acceptable for rapid sand mold casting simulation. After 5000 s, the measured and calculated curves almost overlapped. This confirmed that the inverse method provided a reliable interfacial heat transfer coefficient for ZL105 aluminum alloy and phenolic urethane resin sand.
Using the obtained boundary conditions, I simulated the filling and solidification of the cylinder head for the three gating schemes. The filling process was analyzed using the temperature field and the free surface position. The solidification process was analyzed using the temperature field, solid fraction, and shrinkage defect prediction. I also monitored the filling time distribution and the formation of isolated liquid regions.
For Scheme 1, single-side top pouring, the mold cavity filled completely in about 15 s. The metal entered through multiple ingates and initially avoided excessive sand erosion. After about 8 s, the lower part of the liquid metal became nearly horizontal, and the free surface rose smoothly. However, because the ingates were located on one side, the filling time at the same horizontal level was not uniform in the early stage. The side near the ingates filled earlier than the far side. In the solidification stage, the temperature gradient initially favored directional solidification toward the ingates and risers. The ingates remained hot and provided feeding for a short time. After about 120 s, the riser began to solidify, while the lower thick sections were still in the mushy state. This created isolated liquid regions. The shrinkage prediction showed several concentrated shrinkages in the lower thick sections near the ingates. Therefore, Scheme 1 was not optimal.
For Scheme 2, symmetrical top pouring, eight ingates were distributed symmetrically on the top. The filling time was about 14 s. The metal entered the cavity more uniformly than in Scheme 1, and the liquid level rose relatively smoothly. However, the two streams met in the middle of the mold, which increased the risk of turbulence, gas entrapment, and inclusion formation. The filling time distribution was less uniform than expected because the flow paths were long. During solidification, the risers and ingates initially fed the thick sections. After about 125 s, isolated liquid regions still formed in the lower thick sections. The predicted shrinkage volume was somewhat smaller than in Scheme 1, but a significant shrinkage defect remained. Therefore, Scheme 2 also required improvement.
For Scheme 3, single-side bottom pouring, the mold cavity filled completely in about 16 s. The metal entered from the bottom, so it did not directly impact the sand mold. The liquid level rose smoothly, and the filling time distribution became uniform after the first 15% of filling. This reduced turbulence, oxidation, and inclusion defects. During solidification, the temperature field followed a favorable sequence. The risers fed the upper thick sections, and the bottom ingates fed the lower thick sections. At 240 s, no large isolated liquid region was observed. The predicted shrinkage defects were finely dispersed rather than concentrated. Therefore, Scheme 3 was the best among the three schemes for rapid sand mold casting.
The following table compares the three schemes based on filling time, filling stability, and shrinkage defects.
| Scheme | Filling Time | Filling Stability | Shrinkage Defects |
|---|---|---|---|
| Scheme 1: Single-side top pouring | About 15.7 s | Moderate; early stage not uniform | Concentrated shrinkages in lower thick sections |
| Scheme 2: Symmetrical top pouring | About 14.0 s | Moderate; metal streams meet in the middle | Noticeable shrinkage, but smaller than Scheme 1 |
| Scheme 3: Single-side bottom pouring | About 15.4 s | Best; smooth and uniform after initial stage | Only dispersed micro-shrinkage; best quality |
Based on the simulation results, I selected Scheme 3 for the rapid sand mold casting experiment. The preferred pouring temperature was about 690 °C, and the preferred pouring speed was about 0.22 m/s. These conditions produced smooth filling without turbulence or interruption. The temperature field accurately reflected the solidification sequence, and the predicted shrinkage defects were minimized.
For the rapid sand mold casting experiment, I selected phenolic urethane resin-bonded sand as the molding material. This material has good flowability, high strength, good collapsibility, and excellent dimensional accuracy. It is suitable for single-piece and small-batch production, especially when SLA patterns are used. The original sand was natural washed silica sand with high SiO₂ content and a regular grain shape. The grain size was controlled between 40 and 70 mesh. The sand also had to meet strict requirements for moisture, clay content, and temperature.
The performance requirements for the original sand in phenolic urethane resin sand mold casting were as follows. The SiO₂ content should be as high as possible, generally above 75% for non-ferrous alloys. The grain shape should be rounded or sub-rounded to reduce resin consumption. The grain size should be about 40–70 mesh. The moisture content should be less than 0.2%, because excess water slows curing and can cause gas defects. The clay content should be less than 0.05%, because clay increases binder demand and reduces strength. The sand temperature should be controlled between room temperature and about 35 °C, because temperature strongly affects the curing rate.
I performed orthogonal experiments to determine the optimal ratio of the two resin components. The sand mass for each trial was 400 g. The mixing time was about 2 min, and the demolding time was about 40 min. The experiments were conducted at room temperature. The factor levels are shown in the following table.
| Level | Component I (g) | Component II (g) |
|---|---|---|
| 1 | 2.5 | 2.5 |
| 2 | 5.0 | 5.0 |
| 3 | 7.5 | 7.5 |
| 4 | 10.0 | 10.0 |
The results showed that when either component was less than 0.8% of the sand mass, the mold strength was insufficient, regardless of how much of the other component was added. When one component was sufficient and the other was more than 1%, the sand could meet the basic molding requirements. The tensile strength increased with increasing resin content, but the rate of increase slowed when Component I exceeded about 2% and Component II exceeded about 2.2%. For rapid sand mold casting, excessively high strength makes sand removal and repair difficult and increases cost. Therefore, I selected an ideal ratio of Component I to Component II of about 1:1, with Component I accounting for about 0.8%–1% of the original sand mass. Under this condition, the tensile strength reached about 1.1 MPa, which is sufficient for molding while saving resin and reducing cost.
After preparing the sand and patterns, I assembled the sand cores, closed the mold, and poured the ZL105 aluminum alloy. The production steps included core installation, mold closing, sand repair, coating, and pouring. The intake and exhaust cores were installed first, followed by the water jacket core. The assembled mold was then cleaned and coated. The pouring temperature was controlled near 690 °C. After cooling, the casting was removed and cleaned. The resulting cylinder head showed good overall contour and no major visible defects. The rapid sand mold casting route was therefore technically feasible for single-piece and small-batch production.

From my study, I can draw several conclusions. First, the combination of UG NX 8.0, a casting process CAD system, and ProCAST provided an effective workflow for rapid sand mold casting of a complex cylinder head. Second, the inverse method successfully determined the interfacial heat transfer coefficient between ZL105 aluminum alloy and phenolic urethane resin sand. The coefficient was about 811 W/(m²·K) above the liquidus, decreased through the mushy zone, and stabilized near 230 W/(m²·K) after solidification. Third, the single-side bottom pouring scheme produced the smoothest filling and the best solidification sequence, with only dispersed micro-shrinkage. The preferred pouring temperature was about 690 °C and the pouring speed was about 0.22 m/s. Fourth, the optimal ratio of phenolic urethane resin components was about 1:1, with Component I at 0.8%–1% of the original sand mass. This ratio provided sufficient strength and good molding performance for rapid sand mold casting.
For future work, I would like to improve the material property database and include more realistic boundary conditions, such as mold permeability and coating effects. I would also like to perform stress analysis to optimize the mold design further. Finally, the mold structure could be adjusted to make core assembly and disassembly easier for workers, which would improve the practical application of rapid sand mold casting in small-batch production.
