Researchers hope to improve investment casting shell performance by building a database of thermal properties.

A MODERN CASTING Staff Report

(Click here to see the story as it appears in the January 2015 issue of Modern Casting.)

Reliable and realistic thermal properties data for investment casting shell molds can increase the accuracy of solidification simulations and predictions of shrinkage. Investment casting shells exhibit several phase transformations during firing and pouring that can affect their transient thermal properties. These properties depend on time, temperature and process history.

Mingzhi Xu, Simon Lekakh and Von Richards, Missouri Univ. of Science and Technology, Rolla, Mo., studied the thermal properties (thermal conductivity and specific heat capacity) of seven industrially produced ceramic molds. They used an inverse method, where pure nickel was poured into ceramic molds equipped with thermocouples. Simulation software then was used to simulate virtual cooling curves that resembled the experimental curves by adjusting the temperature dependent thermal properties of the ceramic mold. The thermal properties data obtained from this method were compared with measurement results from laser flash in the hope the dataset will serve to improve the accuracy of investment casting simulation. The paper, “Thermal Property Database for Investment Casting Shells,” provides their analysis of this study.

Question

Will the inverse method (comparing experimental measurements with trial-and-error simulation) provide more accurate measurements for the thermal properties in investment casting shells and help improve simulation?

1. Background

Because of the variety in shell compositions, particle size distribution and processing parameters, ceramic shells may have 10 to 30% porosity, which can provide air permeability but also affects mechanical and thermal properties. Thermal processing history also influences a shell’s thermal properties. Several thermal history stages are involved in the entire process, including pattern removal/de-waxing (176F-572F [80C-300C]); sintering/firing 1,112F-1,832F [600C-1,000C]); preheating (1,472F-2,192F [800C-1,200C]); and pouring (2,732F-2,912F [1,500C-1,600C]). Colloidal silica binder, flour/filler and ceramic stucco have amorphous structures at significant extent. The degree to which the amorphous-to-crystalline transformation takes place during different thermal history conditions affects the shell’s ultimate thermal properties.

The transient nature of the thermal properties of investment shells make them difficult to measure using classical methods, which require steady state conditions. Considering the difficulties of measuring the thermal property of the non-uniform porous shell, researchers may use the inverse method, which characterizes the thermal properties of the shell during the casting process. The shell mold, with a number of thermocouples, is filled with a pure liquid metal with well defined properties. The thermal properties of the shell then are estimated by running multiple computational fluid dynamic (CFD) simulation iterations by varying the thermal conductivity and specific heat capacity to match the calculated cooling curves with the experimental cooling curves for the shell and casting. This inverse method can require a lot of effort to achieve an acceptable match between the two curves.

2. Procedure

The research team introduced a method to correct the specimen thickness used in the laser flash method to obtain more accurate thermal property data. Afterward, the physically measured thermal property data was applied to the inverse method as the starting points to reduce the time and errors induced from extrapolating the optimization algorithm. Seven industrial shells were evaluated. A thermal property database was developed to help increase the accuracy of the investment casting simulations.
Pattern and Shell

A 3 x 3 x 1-in. (76.2 x 76.2 x 25.4mm) expandable polystyrene (EPS) foam pattern was attached to a pouring cup. Patterns were sent to several metalcasting facilities for shelling. Pattern removal, firing and properties analyses were done at Missouri Univ. of Science and Technology. Shells were pre-fired according to requirements from each individual casting facility. Seven different industrial shells were built using the aqueous colloidal silica binder with different mineral fillers as listed in Table 1.

Improved Laser Flash Method

In a laser flash thermal diffusivity test, a small specimen is subjected to a quick, intense radiant laser pulse after thermal equilibration at the test temperature of interest. Typical specimen disc dimensions were 0.5 in. x 0.5 x 0.07 in. (12.7 x 12.7 x 2 mm). The energy of the pulse is absorbed by the front surface and the temperature of the rear face is recorded.

Samples were put into an ambient furnace with 27F (15C)/minute heating rate and laser flash tested from 392F (200C) to 2192F (1,200C) at intervals of 360F (200C). Three runs of each type of specimen were conducted and the average values were reported in the results.

Inverse Method: Setup and Simulation

After firing, one thermocouple (protected by a 0.08-in. [2-mm] diameter OD quartz sheath) was installed in the center of the mold cavity, and the other thermocouple was buried 0.04 in. (1 mm) below the external shell surface. The shells then were entirely wrapped with 0.5-in. (12.7-mm) thick insulation to thermally isolate the shell and limit the influence of the external cooling environment. The shell then was filled with 99.5% nickel at an initial pouring temperature of 2,768F (1,520C). The temperature curves were collected with a 24-bit data acquisition system.

CFD inverse modeling was done using the optimization module of the simulation software. Initially, a base simulation was completed to represent the actual casting conditions by using initial properties. The processing information for initial shell and liquid metal temperatures, pouring time and insulating wrap locations were used in the simulation definition (Fig. 1). The nickel dataset was created from the known pure nickel data. Initially, the property dataset measured by laser flash was used as a starting point. An insulating wool dataset was obtained from thermo-physical data available in the product datasheet. The heat transfer coefficient (HTC) assumed between the casting and shell was 3,500 W/m2K (HTC1) and between the shell and insulating wool was 1,000 W/m2K (HTC2).

The inverse method’s goal was to match the computationally simulated curves with the experimentally measured temperature curves. The initial simulation setup was the baseline for the curve to be compared with the temperature curves obtained from the experimental castings. Figure 2 shows an example of the good match between calculated and experimental temperature curves after hundreds of simulations.

The specific heat capacities, thermal conductivity of the shell and insulating material, and the external heat transfer coefficient (HTC3) were the main parameters that influenced the temperature curves of the casting and shell. Preliminary modeling showed solidification time and the coordinates of the point where the shell reached the highest temperature were mainly influenced by the specific heat capacity and thermal conductivity of the shell.

Density and Porosity

To evaluate the shell density and porosity, pieces of the shell were examined. The overall bulk density and open porosity were measured. In addition, a shell specimen was crushed to 100 mesh to obtain the theoretical density. The total porosity and closed porosity then were calculated.

3. Results and Conclusions

During the project, the research team used the laser flash method to reduce discrepancies due to open porosity by determining the effective thickness of the sample with the help of a 3-D optical profiler. The inverse method was used to generate a thermal properties database for investment casting shells. Using a combination of laser flash and the inverse methods, the researchers could accurately determine the thermal properties for the seven industrial shell systems.

Table 2 shows the densities and porosities of the seven industrial shells after prefiring at 1,562F (850C) for an hour. The silica-based shells (1 and 3) are less dense compared to the aluminosilicate-based shells (4 and 6). The alumina-based shells (5) had the highest density. Total porosity mostly depended on the shell-building process (particle sizes, slurry viscosity, etc.), but shell 7, made by a rapid shelling process, was nearly 40% porous.

Thermal Properties from Inverse Method

Figure 3 shows the specific heat capacity and thermal conductivity data estimated by the inverse method. Temperature-dependent specific heat capacities in all shells had a similar trend, but the average and maximum values mainly depended on the phase of starting materials and the thermal processing’s reactions and transformations, which were not readily predicable.

The investment casting shells, where colloidal silica was used as a binder in most cases and a significant amount of fused silica was utilized as flour and stucco, more often showed increasing thermal conductivity at higher temperatures.

Porosity had a significant influence on thermal conductivity. Between the two aluminosilicate shells (4 and 6), 6 had the higher total porosity (37.65%) and exhibited lower thermal conductivity values throughout the measured temperature range.

Another good example is the weak temperature dependence of conductivity in the alumina-based shell (5). Since the photon radiation in alumina is not significant until 1,832F (1,000C), this radiation compensates phonon scattering in alumina and the porosity effects, and consequently the thermal conductivity didn’t change much over the elevated temperature range.

Thermal Properties From Laser Flash

The thermal conductivity and specific heat capacity values measured from laser flash are listed in Fig. 4. Shell 7 (rapid shelling technique) was highly porous and broke apart when being ground during laser flash sample preparation. Effective density calculated from sample surface topolography was used to calculate these values. Laser flash showed a similar trend to the inverse method on both thermal conductivity and heat capacity values.

Comparing Inverse Method, Laser Flash and Theoretical Values

The thermal conductivity values were fairly close between the inverse method and laser flash. However, the inverse method presented higher specific heat capacity values. In the inverse method, the shell was heated rapidly when metal was poured and cooled at a relatively slower rate during solidification.

The laser flash method showed similar total reaction enthalpy (i.e., defined thermodynamic potential) to the theoretically calculated values, because the thin specimen used in the laser flash method was under partially thermally stabilized condition which was closer to thermal equilibrium. Nevertheless, the shell in reality was hardly in thermal equilibrium conditions. Therefore, the inverse method provided more realistic effective heat capacity values for modeling. However, thermal property data measured from laser flash could be used as starting points in the automatic optimization process, which greatly reduces the number of simulation cases needed and decreases the potential error in iteration step estimates.

The theoretical thermal conductivity of shell 1, shell 3 and pure silica with 33% porosity are shown in Fig. 5. The shells’ measured and theoretical values of thermal conductivity were similar at a lower temperature (<752F [<400C]), but were more heat conductive at a higher temperature. 

This article is based on a paper (14-023) that was presented at the 2014 AFS Metalcasting Congress.