A metalcasting facility manufacturing a complex, rangy aluminum casting experienced a casting quality issue after changing its phenolic urethane binder formula for its molds and cores. Microshrinkage arose in an area that typically did not experience this problem. The initial solidification modeling supported the aggressive design, identifying tight process parameters for the foundry operations, and a review of the foundry process showed no deviation from its process protocols.

Eventually, the foundry found the phenolic urethane coremaking operations experienced a slight change and was the contributing factor to the casting’s quality issues. A series of experiments explored the possibility of manipulating the phenolic urethane binder formula to reestablish the original thermal management within the mold.

Thermal management during casting solidification is becoming more important as the degree of casting geometry becomes more complex—particularly when the castings are thin-walled with pouring temperatures slightly above the decomposition of resin bonded resin systems. As part of this thermal management, foundry designers have traditionally incorporated a range of molding techniques to promote directional solidification. Coupled with solidification modeling tools, casting designs have become aggressive, requiring process control to be equally important factors in casting yield and scrap.

Casting design has changed significantly over the last few decades, incorporating new strategies and materials. A major change has been the incorporation of resin bonded core packages to increase the geometric complexities to meet the engineering design requirements of cast components. As a result, resin bonded sand systems have been developed to meet these challenges during other process operations for the casting process. However, because of the thermal decomposition of resins, along with the demands of the casting process, resin formulations have been designed to meet other factors, such as environmental and handling issues not related to the casting design process. This can have consequences on the design because the alterations can slightly influence the intended thermal management desired during the casting design.

In this case study, the foundry collaborated with its resin supplier and researchers from the University of Northern Iowa foundry to find a solution to the new defect.

Setting Up the Testing
Initially, the team proposed to validate whether subsurface shrinkage could be reduced by varying the resin percentage and/or ratio. An experimental plan was developed to measure the heat flow capabilities of various resin percentages and ratios using differential scanning calorimetry (DSC). Through thermal analysis on a test casting, DSC specific heat flow was confirmed by a comparison of solidification time, metal cooling rate, and mold thermal profile. DSC heat capacity, thermal profile, and dilatometry for each investigated resin level and ratio was coupled with an inverse optimization to generate thermal conductivity data sets for subsequent simulations. Modeling results showed binder content and resin ratios affected the shrinkage porosity rate of the aluminum casting, establishing a resin level and ratios appropriate for the anticipated casting conditions. Trials were conducted at UNI using double plate castings. The results from the DSC and casting trials were used to compare two resin systems, Product A and Product B, at 1% resin and 0.75% resin based on sand weight.

Evaluation of the systems also included conducting dilatometry and differential scanning calorimetry to determine the temperature dependent density and specific heat capacity. These results, in conjunction with the mold and metal thermal profile results, were used in a process simulation software to determine the temperature dependent thermal conductivity and thermal diffusivity using inverse optimization. The temperature dependent properties then were used to create two different datasets in order to simulate the shrinkage defects accurately.

Casting simulation software was used to determine the thermal conductivity and thermal diffusivity of the baseline and best performing sample. The best performing sample was selected using the heat flow and thermal profile results. The experimental thermal profile was replicated in the software to obtain the simulated curves. These curves were optimized by changing the thermal conductivity until a good match was obtained between the simulated and measured curves. Between 1,500-2,000 design iterations were conducted to obtain a good curve match. The specific heat capacity and density results required for datasets were measured directly using instrumentation.

The user datasets created were used to simulate for shrinkage defects in the casting. The simulation software used the casting geometry and process variables sent by the customer. The simulation was run till the maximum temperature of the casting reached the solidus temperature of the A319 aluminum alloy. The resulting simulated porosity was evaluated and compared for the baseline and selected sample.

Results for Product A
Results from the first DSC testing of Product A indicated sand mixed with 1% Product A resin maintained a lower heat flow compared to 0.75%. The DSC testing indicated 1% resin, when heated, requires less energy to raise the temperature of the sand 1 degree. When the resin level in the core sand mix is reduced, more energy is required to raise the sand temperature to the same temperature as sand coated with 1% Product A. Previous research has shown the polymers being used to glue the sand grains together have nearly two times higher heat capacity compared to silica sand. The testing results validate previous research that more energy from the aluminum casting is being used by the higher resin percentage at the mold metal interface before this energy is being transferred further into the mold.

Figure 1 shows the mold thermal profile results for Product A 0.125 inches (3.175 mm) from the mold-metal interface. No significant differences were observed in the temperature between the 1% and 0.75% resin content tests. Both appear to follow a similar relationship. However, 1% 55/45 was slightly lower in the amount of heat flow from the DSC, with the difference being 0.5 W/g approximately.

A difference in temperature occurred for Product A at different overall resin ratios at 0.25 in. (6.25 mm) from the mold-metal interface thermocouple location (Fig. 2). The 1% resin sample continued to heat up in the mold and displayed a higher peak in the mold compared to the 0.75% resin sample. Based on the heat flow characteristics, the 1% does not require as much energy to heat up the sand. It could be theorized heat is being absorbed and transferred at a higher rate, maintaining the higher temperature in the mold. This results in energy being pulled away from the casting at a higher rate, in theory chilling the metal at a faster rate.

Results for Product B
A modification was made to the part 2 resin to understand its influence, and in the 1% resin, it resulted in a slightly higher amount of energy being needed to heat up the sand mold compared to the 0.75% resin sample. The difference in heat flow was measured to be approximately 0.2 W/g. This trend was opposite what was seen with the Product A heat flow results. The DSC results for Product B samples are shown in Figure 3.

Figure 4 shows the mold temperature for Product B 0.125 in. (3.175 mm) from the mold-metal interface. A similar trend was seen in Product A, with little difference in temperature between the 1% and 0.75% resin samples.

The mold temperature results for Product B 0.25 in. (6.35 mm) from the mold-metal interface are shown in Figure 5. The variation between peak samples wasn’t as significant compared to Product A. The difference in peak temperatures between the two resin contents in the original system (Product A) was 60F (16C) while the difference in peak temperatures between the two resin contents with the new part 2 was measured to be 44F (7C). Additionally, the 0.75% resin sample showed a slightly higher temperature for Product B, which correlates with the DSC results.

Comparing Product A and Product B
When comparing the DSC results from Product A to Product B at 1% resin content, the modified part 2 in Product B showed a significant increase in the amount of energy required to increase the temperature of the sand mold. Product B was observed to have a significantly higher heat flow.

A similar trend was observed in the 0.75% resin samples from the two products, though the difference in heat flow is lower when compared to the 1.0% resin samples. The heat flow difference was measured to be approximately 0.75 W/g for the 1% resin samples and 0.20 W/g for the 0.75% resin samples.

Little difference was seen in the mold thermal profile 0.125 in. (3.175 mm) from the mold-metal interface between the two products. A similar trend was seen for both products at 0.75% resin and 0.125 in. (3.175 mm) distance.

The results for both products at 1% resin at a distance of 0.25 in. (6.35 mm) from the mold-metal interface are shown in Figure 6. Product A shows a higher temperature in the mold from approximately 50 seconds to 4,000 seconds, with the maximum temperature difference being approximately 75F. This correlates with the DSC heat flow results, where Product A was observed to have a lower heat flow, resulting in a lower amount of energy being required to raise the temperature of the sand mold. This would result in a lower amount of energy, in the form of heat, being pulled from the casting for Product B, since this product heats up at a slower rate, allowing the casting to retain its heat longer. Based on this, it could be theorized that the reduction of heat being removed from the liquid metal will reduce and/or eliminate shrink issues if 1% resin was maintained while the part 2 was changed.

Figure 7 shows the mold temperature results for both products at 0.75% resin 0.25 in. (6.175 mm) from the mold-metal interface. A trend similar to the 1% resin is seen, with Product A having slightly higher mold temperatures. A difference in maximum temperature of approximately 30F was measured, which is lower than the difference seen for the 1% resin.

The solidification times and temperature were measured from the metal temperature. Product B generally showed a higher solidification time (Table 1). This agreed with the DSC and thermal profile results. Additionally, for Product A, reducing the resin content also increased the solidification time slightly. Reducing the resin content for Product B decreased the solidification time.

Defect Prevented
Results from testing the effects of reducing resin level showed only a slight change in physical properties. The DSC results indicated the higher the resin content, the more energy required to raise the sand temperature to the same level as a lower resin level at a given period of time.

The results for both products at 1% resin at a distance of 0.25 in. (6.35 mm) from the mold-metal interface, where Product A had a higher temperature in the mold from approximately 50 seconds to 4,000 seconds, correlated with the DSC heat flow results and indicated less energy is required to raise the temperature of the sand mold. This would result in less heat being pulled from the casting for Product B. The simulations results correlated to the data collected. A reduction of shrink was found when adjusting the resin level and product.

The data also showed that a lower resin percentage could result in lower heat being removed from the liquid metal and reduce and/or eliminate shrink issues.

Field trials were conducted based on this research, and when using the 0.75% Product A, the occurrence of the shrinkage defect was below the original scrap levels reported with Product A at 1%. Because of this success, Product B has not been tested in casting trials yet.  

Click here to see this story as it appears in the March 2019 issue of Modern Casting