I began my research with a practical and urgent problem: traditional sand mold casting remains a dominant manufacturing route for metal components, yet it is also associated with high energy consumption, large material waste, and significant carbon emissions. In my work, I focused on how to select a low-carbon molding process for moldless sand casting by considering casting complexity. I treated casting complexity as a combination of casting structure and production batch. Based on this idea, I developed carbon emission and cost models for three typical molding routes: traditional sand mold casting, sand mold 3D printing, and composite molding. I then used these models to propose a low-carbon process selection method for sand mold casting. My goal was not only to compare processes but also to provide a quantitative tool that can support cleaner production decisions in sand mold casting.
Traditional manufacturing still occupies a central position in the global economy, and sand mold casting is one of its most widely used solidification processes. Because sand mold casting offers simple operation, low tooling requirements, and broad material compatibility, it accounts for a large share of total casting production. However, this route also creates serious environmental pressure. In many foundries, energy efficiency in sand mold casting remains low, and emissions of dust, waste gas, and solid waste are considerable. I found that improving the sustainability of sand mold casting requires more than incremental equipment upgrades. It requires a systematic method that can compare alternative molding processes under different casting complexity conditions.
In recent years, additive and subtractive manufacturing technologies have been introduced into sand mold casting. Sand mold 3D printing, also known as binder jetting of sand, allows molds and cores to be produced directly from digital models without a physical pattern. Sand mold milling, as a subtractive route, can produce sand molds with good surface quality and short lead times. A composite molding method combines traditional sand mold casting with sand mold 3D printing or milling, so that the advantages of each route can be used where they are most effective. Despite these advances, I observed that most existing studies focus on process feasibility, mechanical properties, or cost, while the low-carbon selection problem for sand mold casting has not been fully addressed. In particular, the influence of casting complexity on carbon emissions across different molding methods has not been systematically modeled.
To fill this gap, I proposed a casting complexity model and used it as the basis for carbon emission prediction in sand mold casting. I divided the sand mold casting production process into three stages: molding, pouring, and post-processing. I classified carbon sources into material carbon sources, energy carbon sources, and unexpected carbon sources. I then assigned these sources to the three production stages. This stage-based structure allowed me to build a carbon emission model that can be used for different molding methods. I also built a cost model that includes material cost, labor cost, energy cost, and overhead cost. Using these models, I calculated carbon emissions and costs for a set of case castings. I then applied k-means clustering and multiple linear regression to establish relationships between casting complexity and carbon emissions. Finally, I proposed a low-carbon molding process selection method and validated it with a furnace cover casting example.
Typical Sand Mold Casting Routes and Their Process Characteristics
I studied three typical molding routes for sand mold casting. The first is the traditional molding route, in which sand is mixed, poured into a flask, compacted, cured, stripped, and then assembled. The second is the sand mold 3D printing route, in which a binder is jetted onto successive layers of sand and catalyst, forming the mold and core directly from a digital model. The third is the composite molding route, which combines traditional molding with sand mold 3D printing. In my composite route, the pouring system and other non-critical parts are made by traditional molding, while the mold surfaces that directly contact the casting are made by sand mold 3D printing.
In traditional sand mold casting, the molding stage includes sand filling, vibration compaction, curing, pattern drawing, and surface drying. The pouring stage includes melting, pouring, and cooling. The post-processing stage includes sand reclamation and machining of the casting blank. In sand mold 3D printing, the main difference is in the molding stage. The printing process requires no pattern, so draft angle requirements are eliminated, and highly complex mold geometries can be produced. However, the printing process is slower for large molds and can consume more binder and energy. In composite molding, the high-precision sand mold 3D printing part is used only where dimensional accuracy and surface quality are critical, while the rest of the mold is produced by traditional molding. This reduces the amount of printed sand and shortens the printing time, which can lower both carbon emissions and cost.
I summarized the process stages of the three routes in the following table.
| Production stage | Traditional sand mold casting | Sand mold 3D printing | Composite molding |
|---|---|---|---|
| Molding | Sand mixing, flask filling, compaction, curing, pattern drawing, coating, core setting, mold closing | Sand spreading, binder jetting, curing, layer repetition, depowdering | Traditional molding for non-critical parts plus 3D printing for critical mold surfaces |
| Pouring | Melting, pouring, solidification and cooling | Melting, pouring, solidification and cooling | Melting, pouring, solidification and cooling |
| Post-processing | Sand reclamation, shakeout, shot blasting, machining | Sand reclamation, shakeout, shot blasting, machining | Sand reclamation, shakeout, shot blasting, machining |
The figure below shows an industrial view of sand casting manufacturers and the type of production environment in which these molding routes are applied.

Casting Complexity Analysis
I defined casting complexity as a combination of casting structure and production batch. Casting structure includes several measurable factors: casting volume, core volume, casting surface area, minimum thickness, maximum thickness, draft depth, and number of cores. Production batch is the number of castings produced in one production run. I selected these factors because they directly affect sand consumption, binder consumption, melting energy, pouring time, and post-processing effort in sand mold casting.
| Symbol | Factor | Role in sand mold casting |
|---|---|---|
| \(V_p\) | Casting volume | Determines metal mass and pouring energy |
| \(V_c\) | Core volume | Affects core sand consumption and core-making energy |
| \(A_p\) | Casting surface area | Influences mold surface area, coating, and cooling |
| \(T_{min}\) | Minimum thickness | Affects solidification time and defect risk |
| \(T_{max}\) | Maximum thickness | Affects feeding and shrinkage behavior |
| \(D_d\) | Draft depth | Affects pattern drawing in traditional molding |
| \(N_c\) | Number of cores | Affects core assembly and molding complexity |
To relate casting structure to carbon emissions, I used a multiple linear regression model. The general form is:
$$CE_N = \beta_0 + \sum_{j=1}^{m} k_j PS_j + \varepsilon$$
Here, \(CE_N\) is the total carbon emission for a production batch of \(N\), \(\beta_0\) is the constant term, \(m\) is the number of structural factors, \(PS_j\) is the \(j\)-th casting structure factor, \(k_j\) is its coefficient, and \(\varepsilon\) is the error term. This model allowed me to quantify how changes in casting volume, surface area, number of cores, and minimum thickness influence carbon emissions in sand mold casting.
Carbon Emission and Cost Modeling for Sand Mold Casting
I modeled the carbon emissions of sand mold casting by dividing the production process into molding, pouring, and post-processing. The total carbon emission is:
$$CE = CE_M + CE_P + CE_{PP}$$
where \(CE_M\) is the molding stage emission, \(CE_P\) is the pouring stage emission, and \(CE_{PP}\) is the post-processing stage emission. I further classified carbon sources into material, energy, and unexpected sources:
$$CE = CE_{(material)} + CE_{(energy)} + CE_{(unexpected)}$$
The material carbon source includes both non-recyclable and recyclable materials. The energy carbon source includes electricity and fuels. The unexpected carbon source includes waste and defective products. I expressed the total carbon emission as:
$$CE = \sum_a m_a f_a + \sum_b m_b (1-\lambda_b) f_b + \sum_c E_c f_c + \sum_d m_d E_d^S f_d$$
In this equation, \(m_a\) is the consumption of non-recyclable material \(a\), \(f_a\) is its carbon coefficient, \(m_b\) is the consumption of recyclable material \(b\), \(\lambda_b\) is its recovery rate, \(f_b\) is its carbon coefficient, \(E_c\) is the consumption of energy \(c\), \(f_c\) is its carbon coefficient, \(m_d\) is the amount of waste \(d\), \(E_d^S\) is the energy consumed to treat unit waste \(d\), and \(f_d\) is the carbon coefficient of that energy.
For cost modeling, I divided the total production cost into material cost, labor cost, energy cost, and overhead cost:
$$C = C_{(material)} + C_{(labor)} + C_{(energy)} + C_{(overheads)}$$
Material cost includes metal charge, sand, binder, coating, and shot. Labor cost is based on working hours. Energy cost is mainly electricity and fuel. Overhead cost includes pattern cost, management cost, and depreciation. These two models formed the basis for comparing the three molding routes in sand mold casting.
I collected rated power data for typical sand mold casting equipment. The table below summarizes the equipment power used in my calculations.
| Equipment | Rated power (kW) | Equipment | Rated power (kW) |
|---|---|---|---|
| Sand mixer | 25 | Surface drying oven | 104.2 |
| Roller conveyor | 1.5 | Core shooter | 8 |
| Vibrating table | 3 | 3D sand printer | 5 (maximum) |
| Pattern drawing car | 2.2 | Medium-frequency induction furnace | 2500 |
| Flow coater | 1.85 | Shot blasting machine | 38 |
| Sand processing line | 152.05 |
I also used carbon coefficients for different materials and energy sources. The table below lists the main coefficients used in my model.
| Substance | Carbon coefficient | Unit |
|---|---|---|
| Electricity | 0.9300 | kgCO2e/kWh |
| Coating | 6.0232 | kgCO2e/kgce |
| Natural gas | 2.1622 | kgCO2e/kgce |
| Coke | 2.8604 | kgCO2e/kgce |
| Alcohol | 0.4800 | kgCO2e/kgce |
| Silica sand | 0.2543 | kgCO2e/kgce |
| Pig iron | 2.1300 | kgCO2e/kgce |
| Steel | 8.2000 | kgCO2e/kgce |
| Cooling water | 0.1940 | kgCO2e/kgce |
Using these data, I calculated the carbon emissions and costs for forty case castings. The batch size was set to 50 pieces, 500 pieces, and 1000 pieces in different scenarios. I found that the relationship between casting volume and total carbon emission is generally positive in all three molding routes. However, the contribution of each production stage changes significantly. In traditional sand mold casting, small castings have a relatively high molding-stage share, while large castings are dominated by the pouring stage. In sand mold 3D printing, the molding-stage share is more stable. In composite molding, the molding-stage share decreases as casting volume increases, while the pouring-stage share increases.
Carbon Emission Modeling Based on Casting Complexity
Because the carbon emissions of the case castings varied widely, I used k-means clustering to divide the castings into two groups. Group A contained smaller and less emission-intensive castings, and Group B contained larger and more emission-intensive castings. This step improved the accuracy of the regression models. I then performed multiple linear regression for each group and each molding route. I checked multicollinearity using VIF values and removed factors that were not statistically significant.
For traditional sand mold casting at a batch size of 50, the regression model for Group A was:
$$CE_{TM,A,50} = 5.701 + 3.689 V_p + 0.157 A_p + 0.631 N_c – 0.011 T_{min}$$
For Group B at the same batch size, only casting volume was significant:
$$CE_{TM,B,50} = 74.190 + 3.042 V_p$$
For sand mold 3D printing at a batch size of 50, the Group A model was:
$$CE_{AM,A,50} = -0.001 + 3.639 V_p + 0.157 A_p$$
For Group B at the same batch size, the model was:
$$CE_{AM,B,50} = -0.068 + 3.639 V_p + 0.158 A_p$$
For composite molding at a batch size of 50, I introduced the proportion \(P\), which represents the share of sand mold 3D printing in the non-pouring system. The definition is:
$$P = \frac{m_{(sand-AM)}}{m_{(sand-AM)} + m_{(sand-TM)}}$$
For Group A with different \(P\) values, the regression models were:
$$CE_{CM,A,50,P} = \beta_0(P) + k_{Vp}(P) V_p + 0.157 A_p$$
For Group B with different \(P\) values, the regression models were:
$$CE_{CM,B,50,P} = \beta_0(P) + k_{Vp}(P) V_p$$
I also studied the effect of production batch. For traditional sand mold casting at 500 pieces and 1000 pieces, the significant factors remained casting volume, surface area, number of cores, and sometimes minimum thickness. The models for Group A at 500 and 1000 pieces were:
$$CE_{TM,A,500} = 5.575 + 3.341 V_p + 0.157 A_p + 0.621 N_c – 0.011 T_{min}$$
$$CE_{TM,A,1000} = 5.568 + 3.146 V_p + 0.157 A_p + 0.620 N_c$$
For Group B at 500 and 1000 pieces, only casting volume was significant:
$$CE_{TM,B,500} = 49.024 + 2.932 V_p$$
$$CE_{TM,B,1000} = 31.898 + 2.897 V_p$$
For composite molding at 500 and 1000 pieces, I obtained similar equations with \(P\)-dependent coefficients. I then fitted the constant term and the volume coefficient as functions of \(P\). For Group A, the fitting formula was:
$$\beta_0(P) = a_0 P^2 + b_0 P + c_0$$
$$k_{Vp}(P) = a_1 P^2 + b_1 P + c_1$$
For Group B, the same quadratic polynomial form was used. These equations allowed me to predict the carbon emission of composite molding for any chosen proportion of sand mold 3D printing.
The following table summarizes the regression coefficients for the traditional sand mold casting models at different batch sizes.
| Batch | Group | Constant | \(V_p\) | \(A_p\) | \(N_c\) | \(T_{min}\) |
|---|---|---|---|---|---|---|
| 50 | A | 5.701 | 3.689 | 0.157 | 0.631 | -0.011 |
| 50 | B | 74.190 | 3.042 | — | — | — |
| 500 | A | 5.575 | 3.341 | 0.157 | 0.621 | -0.011 |
| 500 | B | 49.024 | 2.932 | — | — | — |
| 1000 | A | 5.568 | 3.146 | 0.157 | 0.620 | — |
| 1000 | B | 31.898 | 2.897 | — | — | — |
For sand mold 3D printing, the coefficients were nearly independent of batch size because the printing process is digitally driven and does not require a pattern. The following table shows the coefficients for sand mold 3D printing.
| Group | Constant | \(V_p\) | \(A_p\) | Other factors |
|---|---|---|---|---|
| A | -0.001 | 3.639 | 0.157 | Not significant |
| B | -0.068 | 3.639 | 0.158 | Not significant |
Low-Carbon Molding Process Selection Method
Based on the carbon emission models and cost models, I proposed a low-carbon molding process selection method for moldless sand casting. The method considers three candidate routes: traditional sand mold casting, sand mold 3D printing, and composite molding. For composite molding, I used \(P = 100\%\) as the reference case in the initial comparison. The selection procedure is as follows. First, determine the casting structure factors and production batch. Second, predict the carbon emission for each route using the regression models. Third, predict the production cost for each route. Fourth, select the route that provides the lowest carbon emission and acceptable cost. If two routes have similar carbon emissions, the lower-cost route is preferred. If two routes have similar costs, the lower-carbon route is preferred.
For a batch size of 50 pieces, the carbon emissions of the three routes in Group A are:
$$CE_{TM,A,50} = 5.701 + 3.689 V_p + 0.157 A_p + 0.631 N_c – 0.011 T_{min}$$
$$CE_{AM,A,50} = -0.001 + 3.639 V_p + 0.157 A_p$$
$$CE_{CM,A,50} = \beta_0(P) + k_{Vp}(P) V_p + 0.157 A_p$$
For Group B at 50 pieces, the carbon emissions are:
$$CE_{TM,B,50} = 74.190 + 3.042 V_p$$
$$CE_{AM,B,50} = -0.068 + 3.639 V_p + 0.158 A_p$$
$$CE_{CM,B,50} = \beta_0(P) + k_{Vp}(P) V_p$$
I compared these expressions and found that for small castings, sand mold 3D printing usually has the lowest carbon emission. For large castings, traditional sand mold casting can become more favorable. The boundary depends on casting volume and surface area. For Group B at 50 pieces, the condition for sand mold 3D printing to be lower than traditional sand mold casting is approximately:
$$V_p < 124.385 – 0.265 A_p$$
For a batch size of 500 pieces in Group A, the comparison becomes:
$$CE_{TM,A,500} = 5.575 + 3.341 V_p + 0.157 A_p + 0.621 N_c – 0.011 T_{min}$$
$$CE_{AM,A,500} = -0.001 + 3.639 V_p + 0.157 A_p$$
Ignoring the small minimum thickness term, the condition for sand mold 3D printing to be lower is:
$$V_p < 2.084 N_c + 18.711$$
For a batch size of 1000 pieces in Group A, the condition becomes:
$$V_p < 1.258 N_c + 11.296$$
For Group B at 500 and 1000 pieces, traditional sand mold casting is usually more favorable when casting volume is large. These boundary equations provide a simple decision rule for selecting a low-carbon molding route in sand mold casting.
The cost comparison shows a similar trend. For small batches and small castings, sand mold 3D printing and composite molding often have lower total cost because they avoid pattern costs and reduce labor. For large batches and large castings, traditional sand mold casting becomes more economical because pattern cost is amortized over many parts and sand consumption per part is lower. I summarized the cost results in the following table for the case castings at different batch sizes.
| Batch size | Casting group | Lowest-cost route | Highest-cost route | Typical cost difference |
|---|---|---|---|---|
| 50 | A | Sand mold 3D printing | Traditional sand mold casting | Large |
| 50 | B | Composite molding | Traditional sand mold casting | Moderate |
| 500 | A | Sand mold 3D printing | Traditional sand mold casting | Moderate |
| 500 | B | Traditional sand mold casting | Sand mold 3D printing | Moderate |
| 1000 | A | Sand mold 3D printing | Traditional sand mold casting | Small |
| 1000 | B | Traditional sand mold casting | Sand mold 3D printing | Moderate |
Case Study: Furnace Cover Casting
I applied the proposed method to a furnace cover casting. The casting volume was \(843.02 \, \text{cm}^3\), the material was gray cast iron HT250, the batch size was 500 pieces, the maximum dimension was \(142.5 \, \text{mm} \times 142.5 \, \text{mm} \times 60.5 \, \text{mm}\), the maximum wall thickness was 35 mm, and the average wall thickness was 11.42 mm. The casting contained an internal cavity, so the mold needed sufficient strength and hardness to avoid cracking or damage during transportation and pouring. The quality requirement was that the casting should not contain shrinkage porosity, shrinkage cavities, or other defects.
For traditional sand mold casting at 500 pieces, the predicted carbon emission and cost were:
$$CE_{TM} = 5.575 + 3.341 V_p + 0.157 A_p + 0.621 N_c – 0.011 T_{min}$$
$$C_{TM} = 119.176 \, \text{CNY}$$
For sand mold 3D printing, the predicted carbon emission and cost were:
$$CE_{AM} = -0.001 + 3.639 V_p + 0.157 A_p$$
$$C_{AM} = 86.352 \, \text{CNY}$$
For composite molding, the predicted carbon emission and cost were:
$$CE_{CM} = \beta_0(P) + k_{Vp}(P) V_p + 0.157 A_p$$
$$C_{CM} = 88.167 \, \text{CNY}$$
I also optimized the proportion \(P\) for composite molding. The optimal value was found to be \(P = 80.20\%\). At this proportion, the carbon emission and cost were:
$$CE_{CM,opt} = 36.53 \, \text{kgCO2}$$
$$C_{CM,opt} = 88.17 \, \text{CNY}$$
The following table summarizes the measured carbon emissions for the three routes.
| Molding route | Molding stage (kgCO2) | Pouring stage (kgCO2) | Post-processing (kgCO2) | Total (kgCO2) |
|---|---|---|---|---|
| Traditional sand mold casting | 8.882 | 28.917 | 2.012 | 39.81 |
| Sand mold 3D printing | 5.388 | 27.326 | 1.739 | 34.45 |
| Composite molding | 6.695 | 28.071 | 1.769 | 36.54 |
I used Procast software to simulate the pouring process. The sand box was cooled by forced air, the molding material was self-hardening resin sand, and the alloy was EN-GJL-250. The simulation showed that defects were almost entirely concentrated in the riser, and no obvious defects appeared in the casting itself. This confirmed that the furnace cover and its pouring system were acceptable for actual production. The material consumption for one casting is shown in the following table.
| Production stage | Material | Traditional (kg) | 3D printing (kg) | Composite (kg) |
|---|---|---|---|---|
| Molding | Furan resin sand | 57.719 | 18.256 | 22.480 |
| Molding | Wood | 0.343 | — | 0.107 |
| Molding | Pig iron | 0.999 | 0.893 | 0.918 |
| Pouring | Returned iron | 4.150 | 3.710 | 3.811 |
| Pouring | Scrap steel | 5.015 | 4.483 | 4.605 |
| Pouring | Coke | 0.113 | 0.101 | 0.104 |
| Pouring | Inoculant | 0.078 | 0.061 | 0.072 |
| Post-processing | Steel shot | 0.012 | 0.012 | 0.012 |
The cost breakdown for one casting is shown below.
| Cost category | Traditional (CNY) | 3D printing (CNY) | Composite (CNY) |
|---|---|---|---|
| Material cost | 27.502 | 31.997 | 32.044 |
| Labor cost | 38.858 | 37.990 | 37.479 |
| Energy cost | 7.695 | 5.707 | 7.597 |
| Overhead cost | 45.121 | 10.658 | 11.047 |
| Total cost | 119.176 | 86.352 | 88.167 |
Comparison Between Predicted and Measured Results
I compared the predicted carbon emissions and costs with the measured results. For the furnace cover casting, sand mold 3D printing gave the lowest carbon emission among the three routes. The total carbon emission of sand mold 3D printing was 34.45 kgCO2, which was 5.36 kgCO2 lower than traditional sand mold casting, a reduction of about 15.56%. Compared with composite molding, sand mold 3D printing reduced carbon emission by 2.08 kgCO2, about 6.04%. In terms of cost, sand mold 3D printing was also the lowest. It reduced cost by 32.83 CNY compared with traditional sand mold casting, about 38.02%. Compared with composite molding, it reduced cost by 1.82 CNY, about 2.11%. These results show that for this furnace cover casting, sand mold 3D printing is the best low-carbon and low-cost molding route.
I also analyzed the maximum and minimum carbon emissions and costs among the three routes for the case castings at a batch size of 500 pieces. The difference between routes could reach 96.19% in carbon emission and 171.10% in cost. As casting volume increased, the differences generally decreased. This means that for small and medium castings, process selection has a large influence on carbon emission and cost, while for very large castings, the differences between sand mold casting routes become smaller.
Results and Discussion
My research shows that casting complexity has a strong influence on the carbon emission of sand mold casting. Casting volume is the most important structural factor in almost all regression models. Surface area and number of cores are also significant in many cases. Minimum thickness has a negative effect in some traditional sand mold casting models, which means that a larger minimum thickness can slightly reduce carbon emission because it reduces defect risk and improves yield. Production batch affects traditional and composite molding more than sand mold 3D printing. In traditional sand mold casting, increasing batch size reduces the per-casting carbon emission because pattern and setup burdens are shared. In sand mold 3D printing, the per-casting carbon emission is relatively stable because the process is driven by digital data and does not require a pattern. In composite molding, the proportion \(P\) controls the trade-off between printing precision and traditional molding efficiency. By optimizing \(P\), a lower carbon emission can be achieved without losing mold quality.
The low-carbon molding process selection method I proposed can be used as a decision-support tool for foundries. It requires only basic casting structure data and production batch information. It then provides quantitative predictions of carbon emission and cost for traditional sand mold casting, sand mold 3D printing, and composite molding. The method can help foundries choose a molding route that reduces environmental impact while maintaining economic feasibility. It also supports the broader goal of sustainable manufacturing in sand mold casting.
Conclusions
I developed a comprehensive method for low-carbon molding process selection in moldless sand mold casting based on casting complexity. The main contributions of my work are as follows. First, I analyzed the process characteristics of traditional sand mold casting, sand mold 3D printing, and composite molding, and I divided the production process into molding, pouring, and post-processing stages. Second, I proposed a casting complexity model that includes casting structure and production batch. Third, I established carbon emission and cost models for the three molding routes and calculated their values for forty case castings. Fourth, I used k-means clustering and multiple linear regression to build carbon emission models based on casting complexity. Fifth, I introduced the proportion \(P\) to quantify the use of sand mold 3D printing in composite molding and modeled its effect on carbon emission. Sixth, I proposed a low-carbon molding process selection method and validated it with a furnace cover casting.
The results show that sand mold 3D printing is often the best choice for small and medium castings with complex structures and small to medium batch sizes. Traditional sand mold casting becomes more favorable for large castings and large batch sizes. Composite molding provides a flexible middle ground, and its performance can be optimized by adjusting the proportion of printed sand. For the furnace cover casting, sand mold 3D printing reduced carbon emission by about 15.56% and cost by about 38.02% compared with traditional sand mold casting. These findings confirm that the proposed method can effectively support low-carbon process selection in sand mold casting.
Future Work
In future work, I plan to extend the method by including additional production indicators such as production efficiency, product quality, and resource utilization. I also intend to study larger castings and larger batch sizes to broaden the applicability of the method. Furthermore, I will investigate how different binder systems, sand reclamation strategies, and energy sources affect the carbon emission of sand mold casting. By integrating more factors into the complexity model, I hope to develop a more complete sustainable production optimization framework for sand mold casting.
