In the competitive landscape of modern manufacturing, ensuring the surface quality and dimensional accuracy of casting parts is paramount. As a supplier specializing in engine components, we have extensively adopted the cold-box resin sand casting process for its efficiency. However, a persistent challenge has been the occurrence of burn-on defects on the final casting parts. This defect manifests as a layer of firmly adhered sand grains on the casting surface, which is difficult to remove. It severely compromises surface finish, increases cleaning and rework costs, reduces productivity, and can lead to more critical issues like non-compliance with cleanliness standards or stress cracks from excessive grinding. This study details a systematic investigation, from our first-person perspective, into the root causes of this defect and the application of Design of Experiments (DOE) to develop an optimized, robust solution for our production line.
1. Mechanism of Burn-On Defect Formation
Burn-on, or sand adhesion, is a common surface defect in resin-bonded sand casting processes. It fundamentally arises from the high-temperature interaction between the molten metal and the mold/core sand. Understanding its mechanism is crucial for effective prevention. The defect is primarily categorized into two types:
- Mechanical Burn-On: This occurs when molten metal, under sufficient pressure and fluidity, physically penetrates into the interstitial pores between sand grains. Upon solidification, the metal mechanically interlocks with the sand matrix, forming a tenacious bond. The primary drivers are high metal pressure, excellent fluidity, and a sand system with high porosity and permeability.
- Chemical Burn-On: This is a more complex phenomenon involving chemical reactions at the metal-mold interface. At elevated temperatures, components within the molten metal (e.g., iron oxides, manganese) can react with silica (SiO2) in the sand or decomposition products from the resin binder. These reactions form low-melting-point silicate phases, such as fayalite (Fe2SiO4), which act as a glue, wetting and bonding the sand grains to the metal surface of the casting part.
In our specific case, involving a complex cylinder block casting part with problematic water jacket cores, tappet bores, and sharp “horn tip” features, detailed examination confirmed the defect as predominantly chemical in nature. The formation is influenced by a confluence of factors often studied in isolation, such as sand granulometry, mold strength, coating refractoriness, coating thickness, and pouring temperature. However, the interaction between these factors is frequently overlooked. We hypothesized that a multi-variate approach was necessary to solve this problem effectively, moving beyond single-factor adjustments.
2. Experimental Design and Methodology
2.1 Material Selection and Noise Factor Control
To ensure a focused study, we first stabilized or eliminated potential noise variables. The molding sand was a consistently sourced scrubbed silica sand with stable chemical and physical properties: SiO2 content ~91%, angularity factor of 1.3, pH 5.6, 3-screen concentration of 93.1%, loss on ignition (LOI) ≤0.5%, and a sintering point of 1,460 °C. The sand grain morphology is critical for packing density and surface finish.
Operator variability was minimized due to a stable workforce, and equipment was maintained under a strict periodic schedule. Environmental factors were monitored and considered constant for the duration of the study. After brainstorming and analyzing production data, three critical and controllable process factors were selected for investigation:
- Pouring Temperature (Tp): Directly influences metal fluidity and the intensity of thermal/chemical attack on the sand mold. Low temperature may cause mist runs, while high temperature exacerbates burn-on.
- Coating Wet Layer Thickness (Ct): The primary barrier between the molten metal and the sand. Thickness affects the insulating and refractory protection offered to the mold surface.
- Phenolic Urethane Cold-Box Resin Addition (Ra): Expressed as a weight percentage of sand. It dictates the cured tensile strength of the core, which influences erosion resistance and the generation of gaseous decomposition products at the metal interface.
2.2 Defining Factor Levels and Response Variable
The levels for each factor were set based on historical process windows and preliminary trials to ensure they were practically significant and likely to reveal a response.
| Factor | Symbol | Low Level (-1) | Center Point (0) | High Level (+1) |
|---|---|---|---|---|
| Pouring Temperature | Tp | 1,420 °C | 1,440 °C | 1,460 °C |
| Coating Thickness | Ct | 0.4 mm | 0.5 mm | 0.6 mm |
| Resin Addition | Ra | 1.1 % | 1.2 % | 1.3 % |
The response variable, or metric for success, was defined as the Percentage Burn-On Area (PBA). This was calculated for each experimental casting part by visually assessing and measuring the area affected by burn-on on critical surfaces (water jacket, tappet bore, horn tip) relative to the total area of those features. A lower PBA indicates a better-quality casting part.
2.3 DOE Matrix Construction
A full three-factor, three-level factorial design would require 27 runs (33). To improve efficiency while retaining the ability to model interactions and curvature, a two-level factorial design augmented with center points and replicates was employed. A power analysis confirmed that a design with 17 total runs (a 23 full factorial with 2 replicates and 1 center point) provided a statistical power of approximately 87% to detect significant effects. The experimental matrix is shown below.
| Run Order | Tp (°C) | Ct (mm) | Ra (%) | PBA (%) | Primary Defect Location on Casting Part |
|---|---|---|---|---|---|
| 1 | 1,420 | 0.4 | 1.1 | 12 | Water Jacket, Horn Tip |
| 2 | 1,460 | 0.4 | 1.1 | 15 | Water Jacket, Tappet, Horn Tip |
| 3 | 1,420 | 0.6 | 1.1 | 9 | Water Jacket |
| 4 | 1,460 | 0.6 | 1.1 | 13 | Water Jacket, Tappet, Horn Tip |
| 5 | 1,420 | 0.4 | 1.3 | 13 | Water Jacket, Horn Tip |
| 6 | 1,460 | 0.4 | 1.3 | 14 | Water Jacket, Horn Tip |
| 7 | 1,420 | 0.6 | 1.3 | 8 | Water Jacket |
| 8 | 1,460 | 0.6 | 1.3 | 12 | Water Jacket, Horn Tip |
| 9 | 1,420 | 0.4 | 1.1 | 11 | Water Jacket, Horn Tip |
| 10 | 1,460 | 0.4 | 1.1 | 16 | Water Jacket, Tappet, Horn Tip |
| 11 | 1,420 | 0.6 | 1.1 | 10 | Water Jacket |
| 12 | 1,460 | 0.6 | 1.1 | 14 | Water Jacket, Horn Tip |
| 13 | 1,420 | 0.4 | 1.3 | 13 | Water Jacket, Tappet, Horn Tip |
| 14 | 1,460 | 0.4 | 1.3 | 13 | Water Jacket, Tappet, Horn Tip |
| 15 | 1,420 | 0.6 | 1.3 | 9 | Water Jacket |
| 16 | 1,460 | 0.6 | 1.3 | 14 | Water Jacket, Tappet, Horn Tip |
| 17 | 1,440 | 0.5 | 1.2 | 11 | Water Jacket, Horn Tip |
3. Statistical Analysis and Results
The data from Table 2 was analyzed using Analysis of Variance (ANOVA) to determine the significance of the main effects and their interactions on the burn-on area of the casting part. The model was refined through sequential model reduction, removing non-significant terms (p-value > 0.05) while ensuring the predictive power (R2pred) remained above 70%.
The final ANOVA revealed that both Pouring Temperature (Tp) and Coating Thickness (Ct) had statistically significant effects on the Percentage Burn-On Area (PBA). The effect of Resin Addition (Ra) was found to be not statistically significant within the range studied. Furthermore, a significant interaction between Tp and Ct was identified, meaning the effect of coating thickness depends on the pouring temperature, and vice-versa.
The main effects can be summarized as follows:
- Pouring Temperature (Tp): As Tp increases, PBA significantly increases. Higher temperature intensifies the thermal and chemical attack on the sand mold, promoting both metal penetration and silicate formation on the casting part.
- Coating Thickness (Ct): As Ct increases, PBA significantly decreases. A thicker coating provides a more effective barrier, isolating the sand from the molten metal for a longer duration and reducing interface reactions.
- Resin Addition (Ra): The effect was minor and non-significant. While higher resin increases core strength, it may also increase the volume of gaseous decomposition products; these competing effects likely canceled out within our operating window.
The relationship between the significant factors and the response was quantified using a regression model. The final empirical model for predicting the burn-on area percentage on the casting part is:
$$ P_{BA} = 80.8 – 0.0437T_p – 371C_t + 0.250T_pC_t $$
Where \( T_p \) is in °C and \( C_t \) is in mm. This model clearly shows the negative linear coefficients for Tp and Ct (indicating their opposing directional effects) and the positive interaction coefficient for \( T_pC_t \), which moderates the combined effect.
4. Optimization and Validation
4.1 Determination of Optimal Parameters
Using the derived model within a numerical optimization routine, the goal was to minimize PBA. A practical upper specification limit of 11% was set as the target, beyond which post-casting cleaning becomes economically burdensome. The optimization response solver indicated that the minimum burn-on for the casting part is achieved at the lower bound of pouring temperature and the upper bound of coating thickness. Since resin addition was not significant, it was set to the middle level for cost-effectiveness and robustness.
Optimal Process Parameters:
- Pouring Temperature (Tp): 1,420 °C
- Coating Thickness (Ct): 0.6 mm
- Resin Addition (Ra): 1.2 %
The model prediction for this setting was a PBA of approximately 8.9%, with a 95% confidence interval between 7.8% and 10.0%. The response surface clearly illustrates that to produce a high-integrity casting part, one must employ a lower pouring temperature coupled with a sufficiently thick protective coating.
4.2 Confirmation Trials
To validate the optimization, three separate production trials (a total of 120 casting parts) were run using the optimal parameter set. The burn-on area was meticulously measured for each critical casting part.
| Trial # | Tp (°C) | Ct (mm) | Ra (%) | Average PBA per Batch (%) |
|---|---|---|---|---|
| 1 | 1,420 | 0.6 | 1.2 | 7.0 |
| 2 | 1,420 | 0.6 | 1.2 | 10.0 |
| 3 | 1,420 | 0.6 | 1.2 | 9.0 |
| Grand Average | 1,420 | 0.6 | 1.2 | 8.67 |
The average PBA of 8.67% from the confirmation trials fell well within the predicted confidence interval and was significantly lower than the typical values observed before the study (often above 12-15%). The visual improvement on the casting part was dramatic: water jacket passages, tappet bores, and horn tips showed markedly less adhered sand, translating directly to reduced cleaning time, improved surface quality, and lower scrap rates.
5. Conclusion
This study successfully demonstrates the power of a structured, data-driven approach to solving a complex manufacturing defect. By applying Design of Experiments (DOE), we moved beyond anecdotal adjustments and quantitatively identified the key factors controlling burn-on defects in our cold-box resin sand casting process for engine block casting parts.
- The most significant factors affecting burn-on on the casting part were identified as Pouring Temperature and Coating Thickness, with a notable interaction between them. Resin Addition in the range of 1.1% to 1.3% showed no statistically significant impact on this specific defect.
- A predictive regression model was established, enabling us to understand and quantify the process behavior: $$ P_{BA} = 80.8 – 0.0437T_p – 371C_t + 0.250T_pC_t $$
- The optimal process parameters to minimize burn-on were determined to be a Pouring Temperature of 1,420 °C and a Coating Thickness of 0.6 mm, with Resin Addition at a standard level of 1.2%. Validation trials confirmed the effectiveness, reducing the average burn-on area percentage to 8.67%, a substantial improvement that enhances the quality and manufacturability of the final casting part.
This methodology provides a reliable framework for process optimization. While this study focused on three primary variables, the surface quality of a casting part is influenced by a broader system including sand compaction, gating design, and part geometry itself. Future work could expand the DOE to incorporate these additional factors, employing more advanced designs like Response Surface Methodology (RSM) to model non-linear relationships and find global optima for even more challenging casting parts.