In this study, I focused on the optimization of the sand mold casting process for a tractor saddle, which is a critical connecting component between the tractor and the semi-trailer. The saddle experiences multi-directional impact loads during operation, so its quality requirements are extremely high. Traditionally, the saddle is manufactured from forged steel, which offers good impact resistance but at a higher cost than iron. Therefore, replacing steel with iron through sand mold casting is a promising approach to reduce costs while maintaining performance. However, the transition from steel to iron in saddle production frequently leads to defects such as blowback, cold shut, shrinkage porosity, shrinkage cavity, sand inclusion, and slag inclusion, resulting in significant economic losses. I aimed to solve these problems by designing and optimizing the gating system, exhaust system, and filtration system, and by using numerical simulation and orthogonal experiments. The main research contents and findings are summarized below.

The saddle has an overall dimension of 755 mm × 580 mm × 322 mm, a blank weight of 63 kg, and an average wall thickness of 10 mm, with a maximum wall thickness of 30 mm. It is a large thin-walled casting. I selected ductile iron QT700-6 as the material because of its good ductility, strength, and corrosion resistance, and its lower cost compared to steel. The pouring temperature was initially set between 1395 °C and 1415 °C, and the crystallization temperature is approximately 1150 °C. The chemical composition of QT700-6 is given in Table 1.
| Element | C (%) | Si (%) | Mn (%) | S (%) | P (%) | Mg (%) | Ce (%) | Cr (%) | Cu (%) | Mo (%) |
|---|---|---|---|---|---|---|---|---|---|---|
| QT700-6 | 3.5–3.9 | 1.8–2.1 | 0.35–0.5 | 0.006–0.02 | ≤0.05 | 0.035–0.055 | 0.006–0.03 | ≤0.1 | 0.4–0.6 | ≤0.15 |
The density of QT700-6 increases with temperature up to about 800 °C and then stabilizes, while the thermal conductivity decreases with increasing temperature. The mechanical properties required are tensile strength ≥ 700 MPa, elongation ≥ 6%, and hardness 240–290 HB. According to DIN EN ISO 945, the graphite should be type VI, size 5–8, and flake graphite is not allowed. The raw materials consist mainly of recycled iron, pure ductile iron chips, scrap steel, carburant, ferrosilicon, and composite agents. The molding sand is composed of 80–90% used sand, 10–20% new sand, clay, water, and coal powder. For QT700-6, the charge consists of 50–60% recycled material and 40–50% scrap steel. The carburant and composite agent are added at a ratio of 7:3. After melting, the composition is adjusted, and spheroidization is performed.
I used two-box molding with horizontal parting. Vertical parting would require a complex sand core and a large height difference (794.32 mm), which is unfavorable for production. Horizontal parting ensures uniform height, simpler core structure, and easier mold release. The assembly surface that connects to the vehicle body was placed in the drag to avoid sand and slag defects that tend to float. The linear shrinkage of ductile iron was taken as 1.1% based on free shrinkage. The dimensional tolerance grade is CT 12, and the machining allowance grade is RMAG F–H. The weight tolerance grade is MT 11–14. The draft angle was set to 3°. A single sand core was used to reduce core count, with both vertical and horizontal core prints. The vertical core prints have dimensions of 108 mm, 75 mm, and 42 mm for the upper, and 108 mm, 98 mm, and 42 mm for the lower. Horizontal core prints are symmetric, with a large rear core print of 64 mm and a small front core print of 35 mm, and a draft angle of 3°.
The shrinkage rate formula is:
$$ K = \frac{L_{\text{pattern}} – L_{\text{casting}}}{L_{\text{casting}}} \times 100\% $$
where \(L_{\text{pattern}}\) is the pattern dimension and \(L_{\text{casting}}\) is the casting dimension. For ductile iron with pearlitic matrix, the free shrinkage is 0.9–1.1% and the restrained shrinkage is 0.6–0.8%. I selected 1.1% for design.
The gating system, filtration system, and exhaust system were designed based on the theory of liquid metal filling and solidification. The filling process is governed by the conservation of mass, momentum, and energy. The mass conservation equation is:
$$ \frac{\partial \rho}{\partial t} + \frac{\partial (\rho u)}{\partial x} + \frac{\partial (\rho v)}{\partial y} + \frac{\partial (\rho w)}{\partial z} = 0 $$
For steady flow, this reduces to:
$$ \frac{\partial (\rho u)}{\partial x} + \frac{\partial (\rho v)}{\partial y} + \frac{\partial (\rho w)}{\partial z} = 0 $$
The momentum conservation equations in three directions are:
$$ \rho \left( \frac{\partial u}{\partial t} + u \frac{\partial u}{\partial x} + v \frac{\partial u}{\partial y} + w \frac{\partial u}{\partial z} \right) = -\frac{\partial p}{\partial x} + \mu \nabla^2 u + \rho g_x $$
$$ \rho \left( \frac{\partial v}{\partial t} + u \frac{\partial v}{\partial x} + v \frac{\partial v}{\partial y} + w \frac{\partial v}{\partial z} \right) = -\frac{\partial p}{\partial y} + \mu \nabla^2 v + \rho g_y $$
$$ \rho \left( \frac{\partial w}{\partial t} + u \frac{\partial w}{\partial x} + v \frac{\partial w}{\partial y} + w \frac{\partial w}{\partial z} \right) = -\frac{\partial p}{\partial z} + \mu \nabla^2 w + \rho g_z $$
The energy conservation equation is:
$$ \frac{\partial (\rho T)}{\partial t} + \frac{\partial (\rho u T)}{\partial x} + \frac{\partial (\rho v T)}{\partial y} + \frac{\partial (\rho w T)}{\partial z} = \frac{\partial}{\partial x} \left( \frac{k}{c_p} \frac{\partial T}{\partial x} \right) + \frac{\partial}{\partial y} \left( \frac{k}{c_p} \frac{\partial T}{\partial y} \right) + \frac{\partial}{\partial z} \left( \frac{k}{c_p} \frac{\partial T}{\partial z} \right) + S_T $$
For solidification, the latent heat is incorporated as a source term. The relationship between temperature and solid fraction can be linear or quadratic. The defect prediction uses the temperature gradient method. The temperature gradient \(G\) and cooling rate \(R\) are defined as:
$$ G = \sqrt{ \left( \frac{\partial T}{\partial x} \right)^2 + \left( \frac{\partial T}{\partial y} \right)^2 + \left( \frac{\partial T}{\partial z} \right)^2 } $$
$$ R = \frac{T_{\text{upper}} – T_{\text{lower}}}{t_{\text{upper}} – t_{\text{lower}}} $$
The \(G/\sqrt{R}\) criterion is used, where values below 0.8 indicate a high probability of shrinkage defects.
The gating system was designed as a closed system with bottom pouring combined with stratified pouring. The average pressure head \(H_p\) is calculated as:
$$ H_p = H_0 – \frac{h_0}{2} $$
where \(H_0\) is the height difference between the inner gate and the top of the sprue, and \(h_0\) is the height above the inner gate. For the saddle, \(H_0 = 30\) cm and \(h_0 = 32\) cm, so \(H_p = 14\) cm. The minimum remaining pressure head \(H_M\) is:
$$ H_M \geq L \tan \alpha $$
where \(L = 195\) mm and \(\alpha = 8^\circ\), giving \(H_M \geq 27\) mm, so the sprue length \(H_{\text{sprue}} \geq 347\) mm.
The pouring time \(t\) for a casting weighing less than 450 kg is:
$$ t = S \sqrt{G} $$
where \(G = 63\) kg and \(S = 1.85\) (for wall thickness 3.5–8.0 mm), giving \(t = 14\) s. The total weight of molten metal including the gating system is 157 kg. The total inner gate area \(\sum A_{\text{inner}}\) is calculated by:
$$ \sum A_{\text{inner}} = \frac{G}{0.3 \mu t H_p} $$
with \(G = 157\) kg, \(t = 14\) s, \(\mu = 0.64\), and \(H_p = 14\) cm, yielding \(\sum A_{\text{inner}} = 15.6\) cm\(^2\). The closed gating system ratio is \(\sum A_{\text{inner}} : \sum A_{\text{runner}} : \sum A_{\text{sprue}} = 1 : 1.65 : 1.9\). I designed three different inner gate numbers: 4, 6, and 8. The parameters are listed in Table 2.
| Scheme | Number of inner gates | Single inner gate area (cm²) | Total inner gate area (cm²) | Runner area (cm²) | Sprue area (cm²) |
|---|---|---|---|---|---|
| 1 | 4 | 3.9 | 15.6 | 25.74 | 29.64 |
| 2 | 6 | 2.6 | 15.6 | 25.74 | 29.64 |
| 3 | 8 | 2.0 | 16.0 | 25.74 | 29.64 |
For filtration, I used a combination of choke, filter screen, and slag trap. A 90 mm × 90 mm × 22 mm (10 ppi) filter screen was placed in the runner. For the exhaust system, I designed vent holes in both the mold cavity and the sand core. The total vent area \(S\) was calculated as:
$$ S = (1.5 \sim 4) \frac{G}{2 \rho t h_p \mu} \times 2.26 $$
with \(G = 157\) kg, \(\rho = 6.984\) kg/cm\(^3\), \(t = 14\) s, \(h_p = 140\) mm, and \(\mu = 0.35\), giving \(S = 11\) cm\(^2\). I placed one large vent hole of 8 cm\(^2\) and two small vent holes of 1.5 cm\(^2\) each. The sand core vent area \(v_a\) is determined by:
$$ v_a = \frac{a_c m}{a \rho} – A $$
where \(A = \frac{H}{K G C} \sqrt{\frac{\gamma_1}{\gamma_2 \rho}}\). Increasing the distance from the metal surface to the core top and reducing the gas evolution of the sand mixture can reduce the required vent area.
I used AnyCasting software for numerical simulation. The mesh was divided into different regions: the core had a mesh volume of 2.2 mm\(^3\), the gating system runner 4.5 mm\(^3\), and the mold cavity 9.1 mm\(^3\). The initial conditions were: pouring temperature 1395 °C, gravity 9.8 m/s\(^2\) along +x, pouring time 14 s, and natural cooling in the sand box. The thermophysical properties of FCD 700 (equivalent to QT700-6) were used.
For the filling process, I compared the velocity fields of the three schemes. In Scheme 1 (4 inner gates), the flow was unstable, with a maximum velocity of 179 mm/s and a minimum of 0.699 mm/s, causing turbulence and air entrapment. In Scheme 2 (6 inner gates), the flow was smooth, with no large velocity differences, and the mold filled completely. In Scheme 3 (8 inner gates), the flow was mostly smooth but slightly faster near the end. The filling percentages at key stages are shown in Table 3.
| Scheme | Filling 25% | Filling 35% | Filling 55% | Filling 95% |
|---|---|---|---|---|
| 4 inner gates | 25% | 35% | 55% | 95% |
| 6 inner gates | 24% | 36% | 56% | 94% |
| 8 inner gates | 25% | 35% | 55% | 95% |
For the solidification process, I analyzed the temperature fields. In Scheme 1, at 65 s the solid fraction was 5.7%, and at 516 s it was 85.9%, with small isolated liquid regions. In Scheme 2, at 50 s the solid fraction was 2.4%, and at 734 s it was 91.9%, with minimal isolated liquid regions. In Scheme 3, at 67 s the solid fraction was 5.4%, and at 506 s it was 85.4%, but large and dispersed isolated liquid regions appeared, leading to shrinkage defects. The solidification results are summarized in Table 4.
| Scheme | Time (s) at 5% solid | Time (s) at 15% solid | Time (s) at 65% solid | Time (s) at 85% solid | Isolated liquid regions |
|---|---|---|---|---|---|
| 4 inner gates | 65 | 97 | 294 | 516 | Small and dispersed |
| 6 inner gates | 50 | 117 | 302 | 734 | Very small and few |
| 8 inner gates | 67 | 127 | 301 | 506 | Large and dispersed |
For defect probability, I used the combined defect parameter and probability defect parameter in AnyCasting. In Scheme 1, there were many defect points on the top surface and in the cavity. In Scheme 2, only a few linear defects appeared near the lock hole, and the rest was clean. In Scheme 3, there were many dense defects in the cavity and linear defects on the top and bottom. Therefore, Scheme 2 with 6 inner gates was selected as the optimal gating system.
The formation of shrinkage porosity and shrinkage cavity in ductile iron is influenced by the volume change during solidification. The volume change \(\Delta V\) is:
$$ \Delta V = -S_L + E_{GP} + S_r + E_{Ge} $$
where \(S_L\) is liquid shrinkage, \(E_{GP}\) is graphite expansion, \(S_r\) is volume shrinkage, and \(E_{Ge}\) is eutectic graphite expansion. If \(\Delta V < 0\), shrinkage defects occur. To prevent these, I controlled the carbon and silicon content, kept manganese between 0.2% and 0.5%, and minimized rare earth, phosphorus, and magnesium. A lower pouring temperature is beneficial for thick-walled castings, while thin-walled castings require sufficient superheat to allow graphite precipitation and residual liquid transformation. The relationship between wall thickness and pouring temperature is shown in Figure 13 of the original study, but I will not reproduce it here.
After selecting the 6-inner-gate scheme, I performed an orthogonal experiment to further reduce defects. The factors were pouring temperature (A), filling pressure (B), pouring speed (C), and molten metal weight (D). Each factor had three levels, as shown in Table 5. The experimental design was L9(3\(^4\)). The responses were total pore volume and solidification time.
| Level | Pouring temperature (°C) | Filling pressure (Pa) | Pouring speed (m/s) | Molten metal weight (kg) |
|---|---|---|---|---|
| 1 | 1395 | 1.0 | 0.34 | 157 |
| 2 | 1385 | 0.9 | 0.30 | 155 |
| 3 | 1405 | 1.1 | 0.40 | 159 |
The orthogonal experiment results are shown in Table 6. The total pore volume and solidification time were obtained from AnyCasting simulations.
| Experiment | A | B | C | D | Total pore volume (cm³) | Solidification time (h) |
|---|---|---|---|---|---|---|
| 1 | 1 | 1 | 1 | 1 | 0.986 | 0.5575 |
| 2 | 1 | 2 | 2 | 2 | 0.654 | 0.5581 |
| 3 | 1 | 3 | 3 | 3 | 0.473 | 0.5542 |
| 4 | 2 | 1 | 2 | 3 | 0.813 | 0.5458 |
| 5 | 2 | 2 | 3 | 1 | 0.634 | 0.5411 |
| 6 | 2 | 3 | 1 | 2 | 0.746 | 0.5417 |
| 7 | 3 | 1 | 3 | 2 | 0.682 | 0.5708 |
| 8 | 3 | 2 | 1 | 3 | 0.914 | 0.5728 |
| 9 | 3 | 3 | 2 | 1 | 0.753 | 0.5739 |
I performed range analysis for the total pore volume. The mean values \(K_1, K_2, K_3\) for each factor are shown in Table 7. The range \(R\) indicates the influence order: pouring speed (C) > filling pressure (B) > molten metal weight (D) > pouring temperature (A). The optimal combination for minimum pore volume is A1B3C3D2 (1395 °C, 1.1 Pa, 0.40 m/s, 155 kg).
| Factor | K1 | K2 | K3 | Range R | Optimal level |
|---|---|---|---|---|---|
| A (Pouring temperature) | 0.704 | 0.731 | 0.783 | 0.079 | A1 |
| B (Filling pressure) | 0.827 | 0.734 | 0.657 | 0.170 | B3 |
| C (Pouring speed) | 0.882 | 0.740 | 0.596 | 0.286 | C3 |
| D (Molten metal weight) | 0.791 | 0.694 | 0.733 | 0.097 | D2 |
For solidification time, the range analysis showed that pouring temperature had the largest influence, followed by pouring speed, filling pressure, and molten metal weight. The optimal combination for shortest solidification time was A2B3C3D2 (1385 °C, 1.1 Pa, 0.40 m/s, 155 kg). The mean values are given in Table 8.
| Factor | K1 | K2 | K3 | Range R | Optimal level |
|---|---|---|---|---|---|
| A (Pouring temperature) | 0.5566 | 0.5429 | 0.5725 | 0.0296 | A2 |
| B (Filling pressure) | 0.5580 | 0.5573 | 0.5566 | 0.0014 | B3 |
| C (Pouring speed) | 0.5573 | 0.5593 | 0.5554 | 0.0039 | C3 |
| D (Molten metal weight) | 0.5575 | 0.5569 | 0.5576 | 0.0007 | D2 |
Since the two responses gave slightly different optimal combinations, I used matrix analysis to handle the multi-response problem. The matrix method constructs an indicator layer matrix \(M\), a factor layer matrix \(T\), and a level layer matrix \(E\). For a smaller-is-better indicator, the elements are \(K_{ij} = 1/k_{ij}\). The weight matrix is \(\omega_i = M_i T_i E_i\). The results are summarized in Table 9.
| Factor | Level 1 | Level 2 | Level 3 | Optimal level |
|---|---|---|---|---|
| A | 0.1605 | 0.1632 | 0.1545 | A2 |
| B | 0.0463 | 0.0513 | 0.0566 | B3 |
| C | 0.0799 | 0.0916 | 0.1095 | C3 |
| D | 0.0271 | 0.0305 | 0.0290 | D2 |
Therefore, the final optimal combination is A2B3C3D2: pouring temperature 1385 °C, filling pressure 1.1 Pa, pouring speed 0.40 m/s, and molten metal weight 155 kg. I then validated this optimized scheme using AnyCasting. The filling process was smooth, the solidification showed sequential freezing with minimal isolated liquid regions, and the defect probability was significantly reduced. The total pore volume decreased to 0.401 cm³. The isolated liquid regions at 4.9%, 24.7%, 68.2%, and 92.2% solid fractions were much smaller than in the original 6-inner-gate scheme.
I also compared the optimized scheme with the original production process. The original process used 5 vent holes, a pouring temperature of 1420 °C, and an inner gate area of 14 cm². This led to blowback, cold shut, and shrinkage defects. The optimized scheme with 6 inner gates and the orthogonal parameters eliminated these defects. I produced two batches of 14 saddles and performed ultrasonic testing. No unacceptable defects were found. The microstructure after etching with nitric acid alcohol showed pearlite content ≥80% and ferrite ≤20%, meeting the requirements. The mechanical properties are listed in Table 10.
| Sample | Elongation A (%) | Yield strength (MPa) | Tensile strength (MPa) | Hardness (HBW) | Graphite size (grade) | Pearlite content (%) | Nodularity |
|---|---|---|---|---|---|---|---|
| 01K | 10.0 | 424 | 739 | 245 | 6 | 80 | VI |
| 02K | 10.3 | 412 | 725 | 247 | 6 | 81 | VI |
| 03K | 10.7 | 426 | 747 | 248 | 6 | 80 | VI |
| 04K | 10.0 | 411 | 749 | 240 | 6 | 75 | VI |
| 05K | 10.2 | 436 | 728 | 246 | 6 | 80 | VI |
| 06K | 10.1 | 432 | 735 | 249 | 6 | 81 | VI |
| 07K | 10.3 | 416 | 736 | 247 | 6 | 81 | VI |
| 08K | 10.6 | 414 | 745 | 245 | 6 | 80 | VI |
| 09K | 10.5 | 433 | 742 | 247 | 6 | 82 | VI |
| 10K | 10.2 | 421 | 728 | 249 | 6 | 82 | VI |
| 11K | 10.0 | 424 | 726 | 246 | 6 | 81 | VI |
| 12K | 10.1 | 410 | 731 | 248 | 6 | 80 | VI |
| 13K | 10.1 | 433 | 744 | 246 | 6 | 83 | VI |
| 14K | 10.5 | 424 | 742 | 247 | 6 | 82 | VI |
| 15K | 10.2 | 423 | 729 | 241 | 6 | 82 | VI |
All samples met the requirements: elongation >10%, tensile strength >700 MPa, hardness 240–250 HBW, graphite size grade 6, pearlite content >80% (except sample 04K), and nodularity grade VI. The qualified rate increased from 60% in the original process to 95% with the optimized scheme. This demonstrates that the optimized sand mold casting process for the tractor saddle is effective and reliable.
In conclusion, I successfully optimized the sand mold casting process for a tractor saddle by replacing steel with ductile iron QT700-6. I designed a closed gating system with 6 inner gates, a combined filtration system, and an effective exhaust system. Numerical simulation using AnyCasting revealed that the 6-inner-gate scheme provided smooth filling, minimal isolated liquid regions, and low defect probability. Orthogonal experimentation further optimized the parameters to pouring temperature 1385 °C, filling pressure 1.1 Pa, pouring speed 0.40 m/s, and molten metal weight 155 kg. The final castings passed ultrasonic testing and metallographic examination, achieving a 95% qualified rate. This work provides a theoretical and technical basis for the production of tractor saddles using sand mold casting with iron instead of steel.
Future research could explore different ductile iron materials and heat treatments to further improve performance. More accurate thermophysical properties of QT700-6 would enhance simulation accuracy. Regression analysis could be applied to the orthogonal experiment to reduce experimental error. With the continuous development of numerical simulation, sand mold casting will become even more efficient and reliable for large thin-walled complex castings.
