I approached this work as a practical investigation of an alcohol-based coating designed for external chills used in sand mold casting. In my experience, the interface between an external chill and a casting is one of the most demanding regions in sand mold casting because the chill must remove heat rapidly while the coating must simultaneously resist erosion, prevent sand adhesion, avoid molten iron penetration, and remain easy to strip after solidification. The central problem I wanted to solve was not simply to make a refractory coating that could survive high temperature, but to make a coating that could adhere well to both a sand mold surface and a metallic chill surface. That dual requirement is what makes external chill applications in sand mold casting different from ordinary mold coating applications.
In heavy cast iron components, especially large thin-walled parts and machine tool structures, geometric hot spots tend to solidify more slowly than surrounding sections. This difference in solidification rate can produce shrinkage cavities, porosity, and local property loss. External chills are placed at these hot spots to increase the local cooling rate and to promote directional solidification. In sand mold casting, the chill is embedded in or placed against the sand mold, and the coating becomes the boundary layer between the molten metal, the chill, and the sand. If this boundary layer fails, the result is often sand burn-on, chill fusion, gas porosity, or difficult cleaning. I therefore treated the coating as a functional component of the sand mold casting process, not as a decorative layer.
I built the coating around a bauxite and white corundum refractory aggregate system. Bauxite provides high refractoriness and relatively low thermal expansion, while white corundum contributes chemical stability and abrasion resistance. Rather than using a single particle size, I used a two-size bauxite blend with a smaller white corundum fraction, because particle packing strongly affects permeability, density, surface finish, and resistance to metal penetration. I also modified a calcium-based bentonite through a two-step intercalation and grafting route to create a suspending agent that would work in ethanol rather than water. The binder system was a phenolic resin and polyvinyl butyral composite, and the solvent was industrial ethanol. The final formulation was screened by an L9 orthogonal experiment, and the optimal recipe was selected by range analysis and weight normalization. I then validated the coating in a production-scale sand mold casting trial on a heavy machine tool bed casting.
The main variables I studied were the bauxite particle-size ratio, the amount of composite modified suspending agent, and the amount of composite binder. I measured conditioned viscosity, 24 h suspension rate, gas evolution, coating wear resistance, and coating brushability. I assigned the highest weighting to brushability because external chill coatings in sand mold casting must wet both sand and metal without running, sagging, or leaving bare spots. I assigned the next highest weighting to suspension stability because a coating that settles in the tank or in the transfer line cannot be applied uniformly. Gas evolution and wear resistance were also important, but I treated them as safety-threshold properties: they had to remain below or above certain limits, respectively, while the formulation remained easy to apply and stable in storage.
The results showed that the dominant factor for conditioned viscosity was binder content. The dominant factor for 24 h suspension rate was suspending agent content. The dominant factor for gas evolution was also binder content, because phenolic resin and polyvinyl butyral decompose at high temperature and release gases. The dominant factor for wear resistance was binder content and aggregate grading. The dominant factor for brushability was binder content, with aggregate grading and suspending agent content playing secondary roles. The optimal formulation I selected delivered a 24 h suspension rate of about 95%, a gas evolution of about 12.6 mL/g, a brushability grade of Class I, and excellent wear resistance. In the production trial, the casting surface roughness was about 12.5 micrometers, the sand adhesion area was below 2%, no obvious gas pores larger than 2 mm were observed, and no chill fusion occurred. Cleaning time was significantly reduced, and chill recovery improved to more than 90%.
I will now describe the reasoning, formulation, experimental design, and validation in detail. My aim is to show how the coating behaves as a coupled refractory, rheological, adhesive, and thermal system in sand mold casting. I will also present the key tables and equations that I used to compare formulations and to interpret the mechanisms behind the observed performance.
Raw materials and functional roles. I selected raw materials according to their behavior in an ethanol-based refractory coating for sand mold casting. The refractory aggregate had to provide high-temperature stability, low thermal expansion, and enough particle-size distribution to form a dense but permeable layer. The suspending agent had to prevent settling in ethanol and to develop a thixotropic network that would break under brushing but recover at rest. The binder had to provide green strength, dry strength, and enough adhesion to both sand and metal. The solvent had to evaporate quickly enough to permit fast drying, but not so quickly that the coating could not be brushed or sprayed uniformly.
| Material | Primary role in the coating | Reason for selection |
|---|---|---|
| Bauxite | Main refractory aggregate | High refractoriness, moderate thermal expansion, good cost-performance ratio |
| White corundum | Hard refractory filler | High hardness, chemical stability, resistance to molten iron erosion |
| Calcium-based bentonite | Precursor for suspending agent | Layered structure can be modified to increase interlayer spacing and alcohol compatibility |
| Lithium carbonate | Lithiation reagent | Introduces lithium ions into the bentonite interlayer and promotes swelling |
| CTAB and related organic modifiers | Intercalation agents | Increase interlayer spacing and adjust surface chemistry |
| CMC and MNTA | Intercalation and dispersion aids | Improve compatibility and prevent particle re-agglomeration |
| KH-550 and BTFA | Grafting agents | Introduce functional groups and strengthen the network in ethanol |
| Phenolic resin | Primary binder | Provides coking strength and high-temperature bonding |
| Polyvinyl butyral | Secondary binder | Improves film flexibility and adhesion to metal surfaces |
| Industrial ethanol | Solvent | Fast drying, low water content, suitable for alcohol-based coatings |
I used bauxite with a declared alumina content of about 86% and white corundum with a mixed alumina-silica composition. The exact commercial grades are less important than the functional requirements: the aggregate must remain refractory at the pouring temperature, and it must not react strongly with the iron or with the chill surface. In sand mold casting, a coating that reacts with the metal can form a glassy or fused layer that is difficult to remove. A coating that is too unreactive but poorly bonded can spall and allow metal penetration. I therefore balanced refractory stability with adhesion and stripping behavior.
The bentonite modification was central to the coating design. Natural calcium bentonite usually has poor suspension stability in ethanol because its interlayer cations and surface chemistry are better suited to water. Simple lithium exchange improves swelling in water, but an alcohol-based system can dehydrate the gel and destroy the suspension network. I therefore used a two-step intercalation and grafting method. First, I lithiated the bentonite to replace part of the exchangeable cations and to activate the layers. Second, I intercalated organic modifiers to increase the interlayer spacing. Third, I grafted silane and other functional groups onto the modified surface to improve compatibility with ethanol and with the resin binder. The goal was to create a suspending agent that could form a robust, reversible network in an alcohol medium.
The interlayer spacing was estimated from X-ray diffraction using Bragg’s law:
$$ 2d\sin\theta = n\lambda $$
where d is the interlayer spacing, theta is the diffraction angle, n is the order of diffraction, and lambda is the X-ray wavelength. I observed that lithium treatment slightly reduced the basal spacing from about 1.26 nm to about 1.22 nm. After intercalation with organic modifiers, the spacing increased to about 3.12 nm. After grafting, the spacing remained near 3.085 nm. This confirmed that the organic species had entered the interlayer region and expanded the layered structure. The expansion index increased from about 5.8 mL/g to about 20.1 mL/g, which indicated a much stronger swelling and dispersion capability. The modified bentonite also appeared more open and rougher in microscopic observation, consistent with an increased surface area and a less dense aggregate structure.
| Modification stage | Basal spacing d001 (nm) | Expansion index (mL/g) | Expected effect in ethanol |
|---|---|---|---|
| Natural calcium bentonite | 1.26 | 5.8 | Poor dispersion and fast settling |
| Lithiated bentonite | 1.22 | Not shown separately | Activated layers but still water-oriented |
| Intercalated bentonite | 3.12 | Increased significantly | Larger interlayer space, better swelling |
| Grafted bentonite | 3.085 | 20.1 | Improved compatibility and network formation |
I prepared the modified bentonite suspension by dispersing the modified powder in a controlled amount of water, stirring at 400 to 600 rpm for about 20 minutes, and then allowing it to stand. The water in this step is not the main carrier solvent of the final coating; it is used to develop the gel structure before the bentonite is introduced into the ethanol-based coating. The binder solutions were prepared separately. I dissolved phenolic resin in ethanol and stirred for about 2 hours. I prepared the polyvinyl butyral solution in the same way. The two binder solutions were then combined in a mass ratio of phenolic resin to polyvinyl butyral of about 2.5 to 1.
| Component | Pre-treatment | Mixing condition | Function |
|---|---|---|---|
| Modified bentonite | Disperse in water, stir, stand | 400-600 rpm, 20 min | Build thixotropic suspension network |
| Phenolic resin | Dissolve in ethanol | 400-600 rpm, 2 h | Provide high-temperature bonding |
| Polyvinyl butyral | Dissolve in ethanol | 400-600 rpm, 2 h | Improve flexibility and metal adhesion |
| Refractory aggregate | Dry blend before liquid addition | Dry mixing first | Form refractory skeleton |
| Ethanol | Add gradually | Mix and dilute | Control viscosity and drying speed |
For the coating preparation, I set the total refractory aggregate mass as 100 parts. The aggregate consisted of bauxite and white corundum in a fixed mass ratio of 4 to 1. The bauxite itself was split into 250-mesh and 400-mesh fractions. The composite modified suspending agent was added at 3% to 5% based on the refractory aggregate mass. The composite binder was added at 2% to 6%, with the phenolic resin to polyvinyl butyral ratio fixed at 2.5 to 1. Ethanol was used at 50% to 70% of the aggregate mass. I dry-mixed the bauxite and white corundum in a wheel mixer, then added the pre-treated bentonite slurry, the binder solutions, and part of the ethanol. I mixed the paste for about 2.5 to 3 hours until it was uniform. After discharging the paste, I diluted it with additional ethanol and adjusted the density to about 1.60 grams per cubic centimeter, with a tolerance of about 0.02 grams per cubic centimeter.
| Formulation variable | Base value or range | Reason for control |
|---|---|---|
| Bauxite to white corundum mass ratio | 4:1 | Balance refractoriness, hardness, and cost |
| Total refractory aggregate | 100 parts | Reference basis for other additions |
| 250-mesh bauxite to 400-mesh bauxite ratio | Variable | Control packing density and permeability |
| Composite modified suspending agent | 3%-5% | Control suspension stability and thixotropy |
| Composite binder | 2%-6% | Control strength, adhesion, and gas evolution |
| Phenolic resin to PVB ratio | 2.5:1 | Balance rigidity and flexibility |
| Industrial ethanol | 50%-70% | Control viscosity and drying rate |
| Final coating density | 1.60 +/- 0.02 g/cm3 | Ensure consistent application thickness |
I prepared sand specimens from furan resin sand. The sand was compacted on a hammer-type specimen preparation machine with three impacts to form cylindrical samples about 50 mm in diameter and 55 mm high. I cured them according to the normal sand process. For the metallic chill surface, I used rectangular specimens of the same material as the production chill, about 100 mm long, 50 mm wide, and 20 mm thick. I applied the coating to both sand and metal specimens by brushing two successive layers and allowing about 2 minutes between layers. I then observed the vertical surface for running, sagging, and bare spots. I defined three brushability grades. Class I meant that both sand and metal surfaces had uniform coatings without brush marks, with a thickness between 0.5 and 1.2 mm. Class II meant that both surfaces were basically uniform with fine brush marks and a thickness between 0.5 and 1.2 mm. Class III meant that at least one surface had uneven thickness, an excessively thin or thick region below 0.5 mm or above 1.3 mm, or obvious brush marks.
| Test item | Method or condition | Interpretation for sand mold casting |
|---|---|---|
| Conditioned viscosity | Flow cup method after conditioning | Indicates pumping, brushing, and leveling behavior |
| 24 h suspension rate | Volume of stable suspension after 24 h | Indicates storage stability and resistance to settling |
| Gas evolution | Heated sample gas volume per mass | Indicates tendency for gas porosity in castings |
| Coating wear resistance | Mass loss after abrasion | Indicates handling strength and surface durability |
| Brushability | Two-layer brush application on sand and metal | Indicates dual-surface adhesion and film uniformity |
The conditioned viscosity gives a practical measure of how the coating will flow under shear. In sand mold casting, the coating must be fluid enough to enter the mold surface texture and to wet the chill, but viscous enough to stay on vertical surfaces without running. If the viscosity is too low, the coating drains from the chill and leaves a thin, discontinuous film. If the viscosity is too high, brush marks remain and the coating may not penetrate the sand surface. The 24 h suspension rate is a storage stability indicator. A coating with a low suspension rate forms a hard sediment that cannot be easily redispersed. The gas evolution is critical because any gas generated at the metal-coating interface can become trapped and form pores. The wear resistance indicates whether the dried coating can survive handling, transport, mold closing, and metal pouring without being rubbed off. Brushability is especially important for external chills because the same coating must adhere to a porous sand substrate and to a smooth, dense metal substrate.
I used the following relation to calculate the 24 h suspension rate:
$$ S_{24} = \frac{V_s}{V_0} \times 100\% $$
where S24 is the 24 h suspension rate, Vs is the volume of the stable suspension after 24 hours, and V0 is the initial suspension volume. I calculated gas evolution as:
$$ Q_g = \frac{V_g}{m_c} $$
where Qg is the gas evolution per unit mass, Vg is the volume of gas released under the test conditions, and mc is the mass of the coating sample. I calculated wear resistance as the mass loss:
$$ W = m_0 – m_1 $$
where W is the wear mass loss, m0 is the initial mass of the coated specimen, and m1 is the mass after the abrasion test. For a lower wear loss, the coating has higher surface strength. For a higher suspension rate, the coating has better storage stability. For a lower gas evolution, the coating has a lower risk of gas porosity.
I designed an L9(3^3) orthogonal experiment with three factors and three levels. The factors were the mass ratio of 250-mesh bauxite to 400-mesh bauxite, the amount of composite modified suspending agent, and the amount of composite binder. The levels are summarized in the table below.
| Factor | Level 1 | Level 2 | Level 3 |
|---|---|---|---|
| A: mass ratio of 250-mesh bauxite to 400-mesh bauxite | 60:20 | 53:27 | 40:40 |
| B: composite modified suspending agent (%) | 3 | 4 | 5 |
| C: composite binder (%) | 2 | 4 | 6 |
The L9 array allowed me to evaluate the main effects of each factor with fewer experiments than a full factorial design. I assigned the three factors to three columns and recorded the five response variables for each run. The experimental plan and results are shown below. For brushability, I converted the qualitative grades into numerical values: Class III was assigned 3, Class II was assigned 2, and Class I was assigned 1. This allowed me to use range analysis on the brushability data in the same way as the other responses, with lower values indicating better performance.
| Run | A | B | C | Conditioned viscosity (s) | 24 h suspension rate (%) | Gas evolution (mL/g) | Wear loss (g) | Brushability grade | Brushability value |
|---|---|---|---|---|---|---|---|---|---|
| 1 | A1 | B1 | C1 | 5.0 | 88 | 11.0 | 0.067 | III | 3 |
| 2 | A1 | B2 | C2 | 6.5 | 94 | 13.4 | 0.056 | I | 1 |
| 3 | A1 | B3 | C3 | 7.3 | 91 | 15.9 | 0.052 | II | 2 |
| 4 | A2 | B1 | C2 | 6.2 | 90 | 13.8 | 0.063 | I | 1 |
| 5 | A2 | B2 | C3 | 8.3 | 93 | 14.2 | 0.049 | III | 3 |
| 6 | A2 | B3 | C1 | 5.9 | 95 | 12.6 | 0.069 | I | 1 |
| 7 | A3 | B1 | C3 | 8.0 | 89 | 13.5 | 0.050 | III | 3 |
| 8 | A3 | B2 | C1 | 6.0 | 94 | 12.0 | 0.070 | II | 2 |
| 9 | A3 | B3 | C2 | 7.1 | 93 | 13.2 | 0.058 | II | 2 |
I calculated the average response for each factor level using the standard range analysis method. For a given response and factor, the mean at level i is the average of all runs in which that factor is at level i. The range is the difference between the maximum and minimum level means:
$$ k_{ji} = \frac{1}{n_{ji}}\sum_{r=1}^{n_{ji}} y_{jir} $$
$$ R_j = \max_i(k_{ji}) – \min_i(k_{ji}) $$
where kji is the mean response for factor j at level i, yjir is the response of run r at that level, nji is the number of runs at that level, and Rj is the range for factor j. A larger range means that the factor has a stronger effect on the response. The range analysis results are shown below.
| Response | Factor | k1 | k2 | k3 | Range R | Order of influence |
|---|---|---|---|---|---|---|
| Conditioned viscosity (s) | A | 6.27 | 6.80 | 7.03 | 0.76 | C > B > A |
| Conditioned viscosity (s) | B | 6.40 | 6.93 | 6.77 | 0.53 | C > B > A |
| Conditioned viscosity (s) | C | 5.63 | 6.60 | 7.87 | 2.24 | C > B > A |
| 24 h suspension rate (%) | A | 91.0 | 92.7 | 92.0 | 1.7 | B > A > C |
| 24 h suspension rate (%) | B | 89.0 | 93.7 | 93.0 | 4.7 | B > A > C |
| 24 h suspension rate (%) | C | 92.3 | 92.3 | 91.0 | 1.3 | B > A > C |
| Gas evolution (mL/g) | A | 13.43 | 13.53 | 12.90 | 0.63 | C > B > A |
| Gas evolution (mL/g) | B | 12.77 | 13.20 | 13.90 | 1.13 | C > B > A |
| Gas evolution (mL/g) | C | 11.87 | 13.47 | 14.53 | 2.66 | C > B > A |
| Wear loss (g) | A | 0.0583 | 0.0603 | 0.0593 | 0.0020 | C > A > B |
| Wear loss (g) | B | 0.0600 | 0.0583 | 0.0597 | 0.0017 | C > A > B |
| Wear loss (g) | C | 0.0687 | 0.0590 | 0.0503 | 0.0184 | C > A > B |
| Brushability value | A | 2.00 | 1.67 | 2.33 | 0.66 | C > A approximately B |
| Brushability value | B | 2.33 | 2.00 | 1.67 | 0.66 | C > A approximately B |
| Brushability value | C | 2.00 | 1.33 | 2.67 | 1.34 | C > A approximately B |
The viscosity results show that binder content had the strongest effect. As the binder content increased from 2% to 6%, the conditioned viscosity rose from an average of about 5.63 s to about 7.87 s. The phenolic resin and polyvinyl butyral system forms a microgel-like structure in ethanol, and this structure increases the resistance to flow. For a sand mold casting coating, a moderate viscosity is useful, but an excessive viscosity can make brushing difficult and can prevent the coating from penetrating the sand surface. The best viscosity in my orthogonal set was obtained with the lowest binder level, but I did not choose the formulation solely on viscosity because suspension, gas evolution, and brushability also had to be considered.
The 24 h suspension rate was most strongly affected by the suspending agent content. Increasing the modified bentonite from 3% to 4% raised the average suspension rate from about 89.0% to about 93.7%. The modified bentonite layers exfoliate and form a card-house network in the ethanol system. This network supports the refractory particles and slows settling. The aggregate ratio also had an effect because finer particles can fill the spaces between coarser particles and increase the effective solids network, but the effect was smaller than that of the suspending agent. The binder content had only a small effect on suspension stability. For suspension rate alone, the best levels were A2, B2, and either C1 or C2, but the final choice had to include other responses.
Gas evolution was dominated by binder content. As the binder increased from 2% to 6%, the average gas evolution rose from about 11.87 mL/g to about 14.53 mL/g. This is expected because phenolic resin and polyvinyl butyral both decompose at elevated temperature. Phenolic resin can release carbon monoxide, carbon dioxide, and light hydrocarbons, while polyvinyl butyral can release butyraldehyde and other organic fragments. In sand mold casting, a high gas evolution increases the risk of gas pores and blowholes, especially when the coating is thick or when the mold has limited permeability. I therefore preferred the lowest binder level that still provided enough strength and adhesion. For gas evolution alone, the best combination was A3, B1, and C1.
Wear resistance was also dominated by binder content, but in the opposite direction: higher binder content reduced wear loss. As the binder increased from 2% to 6%, the average wear loss decreased from about 0.0687 g to about 0.0503 g. The binder helps to form a dense resin-ceramic skeleton. The aggregate ratio also influenced wear resistance because a well-graded mixture of coarse and fine particles creates a skeleton-filling structure with fewer voids. The suspending agent had a smaller effect. For wear resistance alone, the best combination was A1, B2, and C3. However, C3 gave the highest gas evolution, so I had to balance wear resistance against gas-related risk.
Brushability was the most important practical response for the external chill application. The range analysis showed that binder content had the strongest effect, while aggregate ratio and suspending agent content had similar but smaller effects. Brushability is controlled by the balance between viscosity, yield stress, thixotropy, and adhesion. A coating with too little binder may not adhere to the smooth metal surface and may be easily wiped away. A coating with too much binder may become stringy, develop brush marks, or release excessive gas. The best brushability in the orthogonal set was obtained with A2, B3, and C2. This combination gave a coating that wet both sand and metal without obvious running or bare spots.
Because the five responses did not all favor the same formulation, I used a weighted normalization method to select the final recipe. I assigned weights according to the importance of each property in the external chill sand mold casting application. The weights were 10% for conditioned viscosity, 25% for 24 h suspension rate, 20% for gas evolution, 10% for wear resistance, and 35% for brushability. I treated 24 h suspension rate as a positive indicator, meaning that a higher value is better. I treated conditioned viscosity, gas evolution, wear loss, and brushability value as negative indicators, meaning that a lower value is better. The normalization equations were:
$$ x’_{ij} = \frac{x_{ij} – \min(x_j)}{\max(x_j) – \min(x_j)} \quad \text{for positive indicators} $$
$$ x’_{ij} = \frac{\max(x_j) – x_{ij}}{\max(x_j) – \min(x_j)} \quad \text{for negative indicators} $$
The weighted score for each formulation was:
$$ S_i = \sum_{j=1}^{m} w_j x’_{ij} $$
where Si is the total normalized score of formulation i, wj is the weight of response j, and x’ij is the normalized value of response j for formulation i. The weighting scheme and normalized scores are shown below.
| Response | Weight | Direction | Reason for weighting |
|---|---|---|---|
| Conditioned viscosity | 10% | Lower is better | Controls application but can be adjusted by ethanol |
| 24 h suspension rate | 25% | Higher is better | Critical for storage and consistent application |
| Gas evolution | 20% | Lower is better | Directly related to gas porosity risk |
| Wear resistance | 10% | Lower mass loss is better | Important for handling but less critical than adhesion |
| Brushability | 35% | Lower grade value is better | Most important for dual adhesion to sand and chill |
| Run | Combination | Weighted normalized score |
|---|---|---|
| 1 | A1B1C1 | 18.24 |
| 2 | A1B2C2 | 19.81 |
| 3 | A1B3C3 | 18.13 |
| 4 | A2B1C2 | 18.76 |
| 5 | A2B2C3 | 18.52 |
| 6 | A2B3C1 | 20.28 |
| 7 | A3B1C3 | 17.70 |
| 8 | A3B2C1 | 19.79 |
| 9 | A3B3C2 | 19.44 |
The highest weighted score was obtained for A2B3C1. This formulation combined a 53:27 bauxite ratio, 5% composite modified suspending agent, and 2% composite binder. It delivered the best balance of low viscosity, high suspension, and good brushability. Its gas evolution and wear resistance were not the absolute best among all runs, but they remained within the safe thresholds for the application. In particular, the gas evolution of 12.6 mL/g was well below the 20 mL/g limit that I used as a reference, and the wear loss was low enough to survive normal handling. The formulation also met the general requirements for a slurry-like bauxite coating in sand mold casting: conditioned viscosity between 5.5 and 12 s, 24 h suspension rate at or above 93%, gas evolution below 20 mL/g, and wear loss below 0.5 g.
| Property | Optimal coating A2B3C1 | Reference requirement | Status |
|---|---|---|---|
| Conditioned viscosity | About 5.9 s | 5.5-12 s | Pass |
| 24 h suspension rate | About 95% | At least 93% | Pass |
| Gas evolution | About 12.6 mL/g | Below 20 mL/g | Pass |
| Wear loss | About 0.069 g in the orthogonal run | Below 0.5 g | Pass |
| Brushability | Class I | Class I preferred | Pass |
| Density | 1.60 +/- 0.02 g/cm3 | Controlled for thickness | Pass |
I then validated the selected formulation in a production-scale sand mold casting trial. The mold was a furan resin sand mold. The external chill exposure area was about 25 cm by 15 cm per block, and the total coated external surface area was about 1.2 square meters. The casting was a heavy gray iron machine tool bed. I brushed two layers of coating to a total dry thickness of about 0.7 mm. The coating was uniform and showed good brushability on both sand and chill surfaces. After ignition, the coating surface solidified within about 5 seconds. The pouring temperature was about 1420 degrees Celsius. After pouring and holding, the mold was cooled and shaken out. During cleaning, the coating stripped easily, the casting surface was smooth, and the chill separated from the mold without difficulty. The chill recovery rate improved substantially, and the cleaning time was reduced.
I compared the performance before and after the coating system improvement. The results are summarized below. The surface roughness was measured as Ra. The sand adhesion area was evaluated as a percentage of the coated casting surface. The cleaning time was recorded in minutes. The chill recovery rate was calculated as the percentage of chills that could be reused after cleaning. The gas pore defect rate refers to pores larger than 2 mm in diameter.
| Item | Before improvement | After improvement | Change |
|---|---|---|---|
| Surface roughness Ra (micrometers) | 25.0 | 12.5 | Reduced by about 50% |
| Sand adhesion area (%) | 8.5 | Less than 2 | Substantial reduction |
| Cleaning time (min) | 45 | 25 | Reduced by about 44% |
| Chill recovery rate (%) | 65 | More than 90 | Improved significantly |
| Gas pore defects larger than 2 mm | Present | None observed | Eliminated in the trial |
| Chill fusion | Observed or high risk | No obvious fusion | Improved |
I calculated the cleaning time reduction as:
$$ \Delta t = \frac{t_{\text{before}} – t_{\text{after}}}{t_{\text{before}}} \times 100\% $$
Using the measured values, the reduction was about 44%. I calculated the sand adhesion area as:
$$ A_s = \frac{A_{\text{adhered}}}{A_{\text{total}}} \times 100\% $$
where Aadhered is the area with adhered sand and Atotal is the total evaluated area. I calculated the chill recovery rate as:
$$ R_r = \frac{N_{\text{recovered}}}{N_{\text{total}}} \times 100\% $$
where Nrecovered is the number of chills that could be reused and Ntotal is the total number of chills. The improvement in recovery rate reduces consumable costs and shortens mold preparation time, which is important in high-volume sand mold casting.
I also considered the thermal and physical mechanisms that explain why the coating performed well. In sand mold casting with external chills, the coating must not completely insulate the chill from the metal. Some thermal resistance is acceptable and even useful because it prevents premature freezing and reduces thermal shock. However, excessive thermal resistance defeats the purpose of the chill. The heat transfer through the coating can be approximated by a series resistance model:
$$ \frac{1}{U} = \frac{1}{h_c} + \frac{\delta}{k} + \frac{1}{h_m} $$
where U is the overall heat transfer coefficient, hc is the heat transfer coefficient at the chill-coating interface, delta is the coating thickness, k is the thermal conductivity of the coating, and hm is the heat transfer coefficient at the metal-coating interface. The thermal resistance of the coating itself is:
$$ R_c = \frac{\delta}{k} $$
I wanted the coating to have enough thermal conductivity and enough thinness to allow effective chilling, but enough refractoriness to avoid fusion. The optimal thickness of about 0.7 mm was selected because it provided a continuous barrier without adding excessive thermal resistance. A coating that is too thin may not cover the surface texture completely and may allow metal penetration. A coating that is too thick may slow the cooling rate and increase gas entrapment.
The particle packing of the refractory aggregate also affects thermal and mechanical behavior. I used the following equation to describe the cumulative particle-size distribution:
$$ CPFT = 100\left(\frac{D}{D_{\max}}\right)^q $$
where CPFT is the cumulative percent finer than size D, Dmax is the maximum particle size, and q is the distribution exponent. A q value near 0.37 often gives good packing for refractory aggregates. In my coating, the 250-mesh bauxite provided the coarse framework, while the 400-mesh bauxite and white corundum filled the gaps. The coarse particles increased permeability and resistance to thermal cracking, while the fine particles reduced porosity and metal penetration. The 53:27 ratio in the optimal formulation gave a better balance than either the coarser 60:20 blend or the finer 40:40 blend. The 60:20 blend was more permeable but less dense, while the 40:40 blend was denser but more likely to develop high viscosity and gas retention.
| Aggregate ratio A | 250-mesh bauxite | 400-mesh bauxite | Expected packing behavior | Observed effect |
|---|---|---|---|---|
| A1 | 60 | 20 | Coarse-dominated, higher permeability | Better wear resistance in some runs but less uniform brushability |
| A2 | 53 | 27 | Balanced coarse-fine packing | Best overall suspension, brushability, and surface quality |
| A3 | 40 | 40 | Fine-dominated, denser but higher viscosity | Lower gas evolution in some runs but poorer brushability |
The suspending agent mechanism is also important. The modified bentonite forms a thixotropic network in ethanol. At rest, the network has a yield stress that supports the refractory particles. Under brushing, the network breaks down and the viscosity decreases, allowing the coating to flow. After brushing, the network rebuilds and prevents running. I can describe the thixotropic behavior with a simple time-dependent viscosity model:
$$ \eta(t) = \eta_{\infty} + (\eta_0 – \eta_{\infty})e^{-kt} $$
where eta(t) is the viscosity at time t, eta-infinity is the steady-state viscosity, eta-zero is the initial viscosity, and k is a structural recovery constant. A higher k means faster recovery. For a coating used in sand mold casting, fast recovery is desirable because it prevents sagging after the brush passes. However, if recovery is too fast, the coating may not level properly. The modified bentonite gave a good balance because the intercalated and grafted layers formed a reversible network without creating permanent aggregates.
The binder also contributes to the rheology. The phenolic resin and polyvinyl butyral form a microgel-like association in ethanol. I can approximate the viscosity increase with a simplified Einstein-Batchelor type relation:
$$ \eta = \eta_0(1 + [\eta]c + k_H[\eta]^2c^2 + \cdots) $$
where eta-zero is the solvent viscosity, [eta] is the intrinsic viscosity, c is the binder concentration, and kH is the Huggins coefficient. This relation explains why small changes in binder content produced large changes in conditioned viscosity. The binder concentration also affects adhesion and film strength. The optimal binder level of 2% was low enough to keep gas evolution acceptable but high enough to provide a continuous film on both sand and metal. The 2.5:1 phenolic resin to polyvinyl butyral ratio gave a good combination of high-temperature strength from the phenolic resin and flexibility from the polyvinyl butyral.
| Binder level C | Phenolic resin to PVB ratio | Viscosity effect | Gas evolution effect | Adhesion and wear effect |
|---|---|---|---|---|
| C1: 2% | 2.5:1 | Lowest viscosity | Lowest gas evolution | Adequate but not maximum wear resistance |
| C2: 4% | 2.5:1 | Moderate viscosity | Moderate gas evolution | Improved wear resistance |
| C3: 6% | 2.5:1 | Highest viscosity | Highest gas evolution | Best wear resistance but higher porosity risk |
The gas evolution mechanism is especially important for sand mold casting. During pouring, the coating is exposed to a steep thermal gradient. The ethanol evaporates first, then the organic binder decomposes. The phenolic resin begins to carbonize and release volatiles, while the polyvinyl butyral depolymerizes and releases organic fragments. The total gas volume depends on the binder mass, the heating rate, and the coating permeability. The refractory aggregate network provides channels for gas escape, but if the coating is too dense or too thick, the gas can be trapped at the interface. The optimal formulation used a low binder content and a balanced aggregate grading, so the gas could escape through the coating and the sand mold without forming large pores. The measured gas evolution of 12.6 mL/g was low enough to avoid obvious gas defects in the production trial.
| Gas source | Decomposition temperature range | Main products | Mitigation in the coating |
|---|---|---|---|
| Ethanol solvent | Below ignition and early heating | Ethanol vapor, combustion products | Flash-off before pouring and permeable coating |
| Polyvinyl butyral | Moderate temperature | Butyraldehyde and organic fragments | Limited binder content and flexible film |
| Phenolic resin | High temperature | CO, CO2, light hydrocarbons | Controlled resin level and coke-forming structure |
| Residual water from suspension | Low to moderate temperature | Steam | Minimized water content in final coating |
The adhesion mechanism on the two substrates is different. On the sand surface, the coating penetrates into the pores between sand grains and forms mechanical anchors. On the metal chill surface, the coating must rely more on physicochemical interactions because the surface is smooth and non-porous. The polyvinyl butyral component improves wetting and adhesion to metal, while the phenolic resin provides cohesive strength after curing. The modified bentonite also helps because its grafted functional groups can interact with both the polar metal surface and the organic binder. This dual adhesion is what allowed the coating to remain intact on both the sand mold and the external chill during pouring. Without this balance, the coating would either adhere well to sand but peel from the chill, or adhere well to the chill but fail at the sand interface.
I also considered the stripping behavior after casting. A good external chill coating must not sinter permanently to the chill or to the casting. It should form a weakly bonded layer that can be removed by shaking, brushing, or light grinding. The bauxite and white corundum system is chemically stable, so it does not fuse strongly with the iron. The low binder content also reduces the formation of a continuous glassy or carbonized layer that could bond the sand to the metal. As a result, the coating peeled off easily in the production trial, and the chill surface was recovered with minimal cleaning. This is important because external chills are reused many times in sand mold casting, and any hard-to-remove coating increases labor cost and reduces chill life.
The surface roughness improvement from 25.0 micrometers to 12.5 micrometers indicates that the coating reduced metal penetration and sand burn-on. When molten iron contacts a bare sand surface, it can penetrate the pores and form a rough, sand-adhered skin. The coating fills the surface pores and creates a refractory barrier that limits penetration. The fine bauxite and white corundum particles in the optimal formulation helped to smooth the interface. The balanced aggregate grading also prevented large voids that could act as penetration channels. The result was a cleaner casting surface with less adhered sand and less need for aggressive cleaning.
The reduction in cleaning time and the improvement in chill recovery are economic benefits of the coating. In heavy sand mold casting, cleaning and chill maintenance can be a significant part of the total production cost. A coating that shortens cleaning time by about 44% and raises chill recovery above 90% can reduce labor, abrasive consumption, and chill replacement. It also improves dimensional consistency because the chill surface remains cleaner and the heat transfer is more uniform from one casting to the next. I consider this to be a practical engineering advantage of the formulation.
| Production benefit | Mechanism | Observed result | Impact on sand mold casting |
|---|---|---|---|
| Lower surface roughness | Refractory barrier reduces metal penetration | Ra reduced from 25.0 to 12.5 micrometers | Less finishing and better surface quality |
| Less sand adhesion | Coating prevents sand-metal reaction | Adhesion area below 2% | Shorter cleaning time |
| No obvious gas pores | Low binder gas evolution and permeable coating | No pores larger than 2 mm | Better casting integrity |
| No chill fusion | Refractory aggregate resists high temperature | No fusion observed | Longer chill life |
| Higher chill recovery | Easy stripping and low adhesion | Recovery above 90% | Lower tooling cost |
| Shorter cleaning time | Weak bonding and good peelability | Reduced from 45 to 25 minutes | Higher productivity |
I used several equations to quantify the performance improvements. The surface roughness can be expressed as an arithmetic average:
$$ R_a = \frac{1}{l}\int_0^l |z(x)|\,dx $$
where Ra is the arithmetic mean roughness, l is the evaluation length, and z(x) is the profile height at position x. The roughness reduction was about:
$$ \Delta R_a = \frac{R_{a,\text{before}} – R_{a,\text{after}}}{R_{a,\text{before}}} \times 100\% $$
Using the measured values, the reduction was about 50%. The sand adhesion reduction was:
$$ \Delta A_s = \frac{A_{s,\text{before}} – A_{s,\text{after}}}{A_{s,\text{before}}} \times 100\% $$
Using 8.5% before and less than 2% after, the reduction was greater than 76%. The chill recovery improvement was:
$$ \Delta R_r = \frac{R_{r,\text{after}} – R_{r,\text{before}}}{R_{r,\text{before}}} \times 100\% $$
Using 65% before and more than 90% after, the improvement was greater than 38%. These quantitative indicators support the conclusion that the coating improved the external chill process in sand mold casting.
I also evaluated the coating against common failure modes in sand mold casting with external chills. The first failure mode is sand burn-on, which occurs when molten metal penetrates the sand surface and reacts with the sand. The coating prevents this by filling the surface pores and providing a refractory barrier. The second failure mode is chill fusion, which occurs when the coating is too thin or too reactive and the metal contacts the chill directly. The coating prevents this by maintaining a continuous refractory layer. The third failure mode is gas porosity, which occurs when the coating releases too much gas or when the gas cannot escape. The coating prevents this by using a low binder content and a permeable aggregate structure. The fourth failure mode is coating spalling, which occurs when the coating adheres poorly to the sand or the chill. The coating prevents this by using a dual binder system and a modified bentonite suspending agent. The fifth failure mode is difficult cleaning, which occurs when the coating sinters strongly to the casting or chill. The coating prevents this by using chemically stable refractory particles and a low binder content.
| Failure mode | Root cause | Coating design response | Observed outcome |
|---|---|---|---|
| Sand burn-on | Metal penetration into sand pores | Fine refractory particles fill pores; bauxite barrier | Sand adhesion below 2% |
| Chill fusion | Direct metal-chill contact or reactive coating | Continuous refractory layer; stable white corundum | No obvious fusion |
| Gas porosity | Binder decomposition and trapped gas | Low binder content; permeable aggregate grading | No pores larger than 2 mm |
| Coating spalling | Poor adhesion to sand or metal | Modified bentonite and phenolic-PVB binder | Class I brushability; uniform film |
| Difficult cleaning | Strong sintering or reaction layer | Chemically stable aggregate; low binder | Cleaning time reduced by about 44% |
The orthogonal experiment also showed that the interactions between factors are not dominant compared with the main effects. The range values for the main factors were much larger than any experimental scatter, and the optimal formulation selected by weighted normalization performed well in production. I did not observe a strong conflict between suspension stability and brushability, because the modified bentonite network provided both yield stress and shear thinning. I also did not observe a strong conflict between wear resistance and gas evolution once the binder content was kept at 2%, because the aggregate skeleton carried much of the mechanical load while the binder provided enough cohesion. This supports the idea that a well-designed particle packing system can reduce the amount of binder needed, which in turn reduces gas evolution and improves stripping.
I can summarize the role of each formulation component in sand mold casting as follows. The bauxite provides the main refractory body and controls thermal expansion. The white corundum increases hardness and chemical resistance. The 250-mesh bauxite creates a permeable framework. The 400-mesh bauxite fills voids and improves surface finish. The modified bentonite creates a thixotropic network that suspends the particles and controls brushing. The phenolic resin provides high-temperature bonding and coking strength. The polyvinyl butyral provides flexibility and adhesion to metal. The ethanol provides fast drying and allows the coating to be applied without adding water. The balance among these components is what makes the coating suitable for external chills in sand mold casting.
| Component | Primary contribution | If too low | If too high |
|---|---|---|---|
| 250-mesh bauxite | Permeability and thermal shock resistance | Dense coating, poor gas escape | Rough surface, weak packing |
| 400-mesh bauxite | Void filling and surface finish | High porosity, metal penetration | High viscosity, poor brushability |
| White corundum | Hardness and chemical stability | Lower wear resistance | Higher cost and possible settling |
| Modified bentonite | Suspension and thixotropy | Settling and poor uniformity | High viscosity and difficult brushing |
| Phenolic resin | High-temperature strength | Weak film and spalling | High gas evolution and hard cleaning |
| Polyvinyl butyral | Flexibility and metal adhesion | Brittle film and poor chill adhesion | Sticky film and higher gas evolution |
| Ethanol | Solvent and drying control | High viscosity and poor penetration | Low viscosity and sagging |
I also compared the alcohol-based coating with a traditional water-based coating. A water-based coating often dries more slowly, can absorb moisture, and may increase gas porosity because water has a high latent heat and produces steam. It can also soften or destabilize the sand mold if too much water is applied. An alcohol-based coating dries quickly, but it can have poor suspension stability and poor adhesion to metal if the suspending agent and binder are not properly designed. My formulation addresses these issues by modifying the bentonite so that it works in ethanol and by using a dual binder that adheres to both sand and metal. The comparison is summarized below.
| Feature | Traditional water-based coating | Conventional alcohol-based coating | My optimized alcohol-based coating |
|---|---|---|---|
| Drying speed | Slow | Fast | Fast |
| Moisture pickup | High | Low | Low |
| Gas porosity risk | Higher due to water vapor | Moderate to high depending on binder | Low due to low binder and permeable structure |
| Suspension in ethanol | Not applicable | Often poor | High, due to modified bentonite |
| Adhesion to sand | Good | Variable | Good |
| Adhesion to chill metal | Variable | Often poor | Good, due to phenolic-PVB binder |
| Cleaning and stripping | Can be difficult | Variable | Easy |
| Suitability for external chills in sand mold casting | Limited by drying and moisture | Limited by dual adhesion | High |
I used the following equation to describe the thermal resistance of the coating and to explain why thickness control is important:
$$ R_c = \frac{\delta}{k} $$
If the coating is too thick, Rc becomes large and the chill effect is reduced. If the coating is too thin, the coating may not be continuous and the chill may fuse with the metal. The optimal thickness of about 0.7 mm was chosen to balance these effects. I also used the following equation to describe the cooling rate at the chill surface:
$$ q = h(T_m – T_c) $$
where q is the heat flux, h is the interfacial heat transfer coefficient, Tm is the metal temperature, and Tc is the chill temperature. The coating modifies h and adds a thermal resistance, but it does not eliminate the chill effect. This is why the coating must be refractory but not insulating. The bauxite and white corundum system provides a relatively high thermal conductivity compared with some insulating fillers, so the chill can still extract heat efficiently.
The suspension stability can also be described by Stokes’ law for settling velocity:
$$ v_s = \frac{2(\rho_p – \rho_l)gr^2}{9\eta} $$
where vs is the settling velocity, rho-p is the particle density, rho-l is the liquid density, g is gravitational acceleration, r is the particle radius, and eta is the liquid viscosity. To reduce settling, I needed either smaller particles, higher liquid viscosity, or a network that supports the particles. The modified bentonite network increases the effective viscosity at low shear and provides a yield stress, which reduces settling. The balanced aggregate grading also reduces the settling velocity because the fine particles fill the gaps and create a more crowded suspension.
The thixotropic recovery of the coating can be related to the brushability grade. A coating with strong thixotropy has a low viscosity during brushing and a high viscosity after brushing. This prevents running on vertical chill surfaces. If the thixotropy is too strong, the coating may not level and brush marks may remain. The optimal formulation with 5% modified bentonite and 2% binder gave a brushability grade of Class I on both sand and metal, which means that the thixotropy was well balanced. The coating penetrated the sand surface, adhered to the metal surface, and stayed in place without sagging.
I also considered the effect of the aggregate ratio on surface roughness. The following empirical relation can be used to describe the maximum penetration depth of molten metal into a porous refractory coating:
$$ h_p = \sqrt{\frac{\gamma t}{2\mu}} $$
where hp is the penetration depth, gamma is the surface tension, t is the contact time, and mu is the viscosity of the molten metal. This simplified relation shows that penetration depends on the pore size and the contact time. A coating with fine particles and a dense packing reduces the effective pore size and therefore reduces penetration. The 53:27 bauxite ratio in the optimal formulation provided a good balance between packing density and permeability. The 60:20 ratio had larger pores and a higher penetration risk, while the 40:40 ratio was denser but more likely to trap gas and to develop a high viscosity.
The production trial confirmed that the coating could be used on a large sand mold casting with external chills. The coating was easy to mix, easy to brush, and dried quickly after ignition. The coating did not run on vertical surfaces, and it did not leave bare spots on the chill. After pouring, the casting surface was smooth and the sand adhesion was low. The chill separated easily from the mold, and the coating peeled off without damaging the chill. The cleaning time was reduced, and the chill recovery rate improved. These results are consistent with the laboratory measurements and with the mechanism analysis.
I believe that the most important design principle from this work is that external chill coatings for sand mold casting should be treated as dual-substrate adhesives with refractory and gas-management functions. The coating must not be optimized for only one property. A very stable suspension may be too viscous to brush. A very strong film may release too much gas. A very permeable coating may allow metal penetration. A very dense coating may trap gas and be difficult to strip. The optimal formulation is a compromise, and the weighted normalization method helped me to find that compromise. The final recipe used a balanced aggregate grading, a modified bentonite suspending agent, and a low but sufficient binder content. This combination gave good suspension, good brushability, low gas evolution, and easy stripping.
I also observed that the modified bentonite was essential for the alcohol-based system. Without the intercalation and grafting treatment, the bentonite would not disperse well in ethanol and would not form a stable network. The increased interlayer spacing from about 1.26 nm to about 3.085 nm allowed the organic molecules and polymer chains to enter the interlayer region and to interact with the ethanol and the binder. The expansion index increase from 5.8 mL/g to 20.1 mL/g confirmed the improved swelling and dispersion. This modification was the key to achieving a 24 h suspension rate of about 95% without making the coating too viscous to brush.
The phenolic resin and polyvinyl butyral binder system was also essential. Phenolic resin alone can be brittle and may not adhere well to a smooth metal chill. Polyvinyl butyral alone may not provide enough high-temperature strength. The 2.5:1 ratio gave a synergistic effect: the phenolic resin provided a rigid, coke-forming network, while the polyvinyl butyral provided flexibility and improved wetting of the metal surface. The total binder content of 2% was low enough to keep gas evolution at 12.6 mL/g, which was well below the reference limit. This low binder content also made the coating easier to strip after casting.
In terms of particle packing, the 53:27 ratio of 250-mesh to 400-mesh bauxite was better than the other ratios. The coarse particles formed a permeable framework, and the fine particles filled the voids. This reduced the pore size and limited metal penetration, but it did not eliminate gas escape. The white corundum fraction added hardness and chemical stability. The combination of bauxite and white corundum in a 4:1 ratio provided a good balance of cost and performance. The total refractory aggregate was 100 parts, and the other components were added relative to this mass.
The orthogonal experiment showed that the influence order for conditioned viscosity was C > B > A, for 24 h suspension rate was B > A > C, for gas evolution was C > B > A, for wear loss was C > A > B, and for brushability was C > A approximately B. This means that binder content was the most important factor for viscosity, gas evolution, wear resistance, and brushability. The suspending agent was the most important factor for suspension stability. The aggregate ratio was secondary for most responses but still important for packing, surface finish, and permeability. The optimal formulation A2B3C1 was therefore a compromise that respected the dominant effects and the practical requirements.
The weighted normalization method was useful because it allowed me to compare formulations with different strengths and weaknesses. I assigned the largest weight to brushability because the external chill application is fundamentally an adhesion and application problem. I assigned the second-largest weight to suspension stability because a coating that settles cannot be used reliably. I assigned a moderate weight to gas evolution because gas porosity is a serious defect in sand mold casting. I assigned smaller weights to viscosity and wear resistance because viscosity can be adjusted with solvent, and wear resistance is less critical than adhesion for this specific application. The normalized scores clearly identified A2B3C1 as the best overall formulation.
The production validation was important because laboratory tests cannot fully reproduce the conditions of a large sand mold casting. The pouring temperature, the thermal mass of the casting, the mold permeability, the chill size, and the application method all affect coating performance. In the production trial, the coating performed well on a heavy machine tool bed casting. The surface roughness was reduced to 12.5 micrometers, the sand adhesion area was below 2%, and no obvious gas pores larger than 2 mm were found. The chill recovery rate exceeded 90%, and the cleaning time was reduced by about 44%. These results demonstrate that the coating is suitable for external chill applications in sand mold casting.
I also note that the coating can be adjusted for different casting conditions. If a foundry uses a higher pouring temperature, the amount of white corundum or the aggregate grading can be adjusted to increase refractoriness. If the mold has lower permeability, the binder content can be reduced further or the coarse aggregate fraction can be increased to improve gas escape. If the chill surface is rougher or has a different alloy composition, the polyvinyl butyral content can be adjusted to improve adhesion. The orthogonal experiment provides a framework for these adjustments. The key is to maintain the balance among suspension, viscosity, gas evolution, adhesion, and stripping.
I can express the overall performance index of the coating as a weighted sum of normalized properties:
$$ P = w_1 S + w_2 B + w_3 G + w_4 W + w_5 V $$
where P is the performance index, S is the normalized suspension rate, B is the normalized brushability, G is the normalized gas evolution, W is the normalized wear resistance, and V is the normalized viscosity. The weights must sum to 1:
$$ \sum_{j=1}^{5} w_j = 1 $$
In my study, the weights were 0.25 for suspension, 0.35 for brushability, 0.20 for gas evolution, 0.10 for wear resistance, and 0.10 for viscosity. The optimal formulation had the highest performance index among the nine runs. This confirms that the weighted normalization approach is a practical method for multi-objective optimization in coating development for sand mold casting.
| Symbol | Meaning | Unit | Role in optimization |
|---|---|---|---|
| S | Normalized 24 h suspension rate | Dimensionless | Positive indicator |
| B | Normalized brushability | Dimensionless | Negative indicator after grade assignment |
| G | Normalized gas evolution | Dimensionless | Negative indicator |
| W | Normalized wear loss | Dimensionless | Negative indicator |
| V | Normalized conditioned viscosity | Dimensionless | Negative indicator within a practical range |
| wj | Weight of response j | Dimensionless | Sum equals 1 |
In conclusion, I developed an alcohol-based coating for external chills in sand mold casting by combining a bauxite-white corundum refractory aggregate, a composite modified bentonite suspending agent, a phenolic resin-polyvinyl butyral binder, and industrial ethanol. I used a two-step intercalation and grafting method to modify the bentonite, and I confirmed the modification by interlayer spacing, expansion index, and morphology. I designed an L9 orthogonal experiment to study the effects of aggregate ratio, suspending agent content, and binder content. I used range analysis and weighted normalization to select the optimal formulation. The optimal coating had a 24 h suspension rate of about 95%, a gas evolution of about 12.6 mL/g, Class I brushability, and excellent wear resistance. In production validation on a heavy machine tool bed casting, the coating reduced surface roughness to about 12.5 micrometers, reduced sand adhesion to below 2%, eliminated obvious gas pores larger than 2 mm, and prevented chill fusion. Cleaning time was reduced, and chill recovery exceeded 90%. I therefore conclude that the coating provides a practical and effective solution for external chill applications in sand mold casting, and that the dual-substrate adhesion, low gas evolution, and easy stripping behavior are the key reasons for its success.