Harmony Castings (Harmony, Pennsylvania) specializes in V-Process cast A356 castings for military, aerospace, robotics, marine, and other industries. The V-process allows castings to be made with unique features such as zero-degree draft, thinner walls, and tighter tolerances while featuring a greater speed to market, and lower tooling costs with a 150 RMS surface finish. In order to meet the requirement of its customer, the foundry has a total of four heat treat ovens capable of performing T5 or T6 heat treatments per the customer and application requirements. Finally, the facility is outfitted with X-ray capability and a comprehensive metallurgical lab including inhouse mechanical property testing in support of their quality control requirements.

The annual tonnage for the subject foundry is 700 tons melted per year with a running rate for melting ranging from 3,000 to 5,000 lbs. per production hour. Production fits the profile of a job shop as it has nearly 1,000 SKUs and is notorious for making several different pattern changes within a shift. Moreover, the facility makes a wide array of casting sizes ranging from 2-150 lbs. Despite being a job shop, pouring is done robotically, retrieving the metal from one of three 2,000-lb. electric resistance crucible furnaces.

Access to the melt deck is restricted during operation by the cell’s protective guarding; guarding required for robotic pouring. Accordingly, production must be stopped to enter the melt deck to perform the requisite treatments to the melt (e.g. degassing, fluxing, grain refining, elemental additions) and since production does not re-commence until the treatments are complete, the time lapse of the melt treatments is directly linked (in the negative) to production uptime. Shortening the time of melt treatment could have a beneficial impact on the number of castings made each shift.

Specific gravity targets were part specific and were empirically derived based on past experiences and trial/error. Ultimately, castings are made to one of the following specific gravity targets:

• 2.40−2.45

• 2.42−2.55

• 2.50−2.55

• 2.55−2.60

• 2.60−2.65

Because there are so many different specific gravity targets, a straightforward degassing process was not possible. The degassing procedure was unique and modified for each specific gravity range. In general, degassing was completed with a simple hoist-mounted rotary degassing unit utilizing a nitrogen purge. Nominal degassing times were 15-20 minutes. When conditions were not favorable for low specific gravity, typically wood or even potatoes were used to add hydrogen to the melt. Unfortunately, on those target jobs that were notoriously difficult to achieve, repeat cycles as much as 50% of the time were not uncommon. Eliminating repeat cycles and reducing the degas cycle times were two of the primary objectives of this project.

The historical method for grain refining and cleaning the aluminum melts was standardized across the different specific gravity targets and entailed manual additions of metallic form TiBor grain refiner (5%Ti,1%B) and Sr to the melt prior to rotary degassing. Mg additions were also often added during the same time to hold them within specification.

The difficult mix of castings resulted in typical porosity levels between 20% to 25% in typical years, and although this was a profitable level, there was an obvious desire to improve. It was felt that better hydrogen control mixed with improved grain refining and the repeatability of automation could help achieve the project’s other primary objective of lowering porosity scrap.

Grain refinement can beneficially affect feeding and fluidity in aluminum castings. Hence, inadequate grain refinement can yield shrink voids in aluminum castings. Moreover, too much Sr can cause porosity in aluminum-silicon alloys also resulting in porosity in cast aluminum. Accordingly, it was decided to assess the grain refinement and eutectic modification levels to see if an improved practice could minimize the amount of returned scrap.

A THERMATEST 5000 NG III thermal analysis (TA) unit was used to assess the grain refinement (or grain fineness) and eutectic modification (or eutectic structure) of the treated melts. Grain refinement and eutectic modification were both targeted as they can directly influence porosity in aluminum castings.

Thermal analysis involves collecting data from a solidifying melt sample (temperature over time) and comparing the curve to a set of known curves algorithmically.

The TA algorithm analyzes the sample curve liquidus and computes a score on a scale from 1−9 for evaluating grain fineness (GF). A GF score of 1 or less references a curve that compares perfectly with curves exhibiting no grain refining. In contrast, a GF score of 9 is achieved when the sample curve compares with those curves know to have produced “perfect” grain refining of melts with the same alloy composition. A pictorial representation of the subject grain refinement levels is provided in Figure 1.

The results of the thermal analysis testing confirmed an opportunity to improve the grain refining as the experimentally derived results ranged from 6.2 to 7.1 on the 0−9 scale. Fortunately, the incumbent process was shown to be reliable in achieving optimized eutectic structures so the incumbent process for adding Sr (10%Sr, 90%Al) master alloy was maintained.

One of the approaches considered for lowering the scrap was improving and automating the grain refining treatment process. Accordingly, a salt-based cleaning and grain refining flux was targeted as a replacement for the metallic TiBor rod additions historically used.

A metal treatment station (MTS) was targeted to assist in automating the cleaning and grain refining flux to maximize the potential impact. An MTS is the predominant way for adding salt-based grain refiner/flux in ladles, crucible, or continuous flow well applications. In an MTS, a vortex is temporarily created by rotating or withdrawing a vortex breaker baffle board and increasing RPMs of the graphite shaft and rotor used in the rotary impeller degassing (or rotary degassing) process. The PLC-controlled additions of treatment flux were added into the vortex and mixed to complete reaction prior to the vortex breaker baffle board re-engaging the melt, effectively stopping the vortex. After the vortex has been stopped, the MTS returns to a standard rotary degassing process. In low hydrogen (high specific gravity) applications, the MTS degasses until it completes its cycle. In intermediate or high hydrogen (low specific gravity) applications, the MTS unit will upgas according to the following methodology, prior to cycling complete.

Adding intermediate amounts of hydrogen to assist in offsetting volumetric shrink in difficult-to-cast aluminum alloy casting designs is a common industrial practice with potatoes, wood, and gassing pills being the most common historical methods of adding hydrogen to melts.

More recently, upgassing has evolved to using blend gasses purged through the same units used for rotary degassing. The gases used are nonammonium blends of nitrogen and hydrogen.

SMARTT is an acronym for SelfMonitoring, Adaptive, Re-calculating Treatment Technology. SMARTT utilizes an Industry 4.0 type degassing model that captures environmental conditions (ambient temperature, ambient humidity, melt temperature, etc.) along with input parameters (alloy, crucible size, degassing rotor design, etc.) to determine the optimized set of output conditions (rpm, purge flow rate, cycle time) for rotary degassing.

The SMARTT technology allows the user to choose one of four optimization schemes listed below: • High-speed degassing (shortest cycle time)

• Low gas degassing (lowest purge consumption)

• Long life degassing (lowest rpm for low wear)

• Standard degassing (hybrid)

The high-speed degassing, low gas degassing, and long-life schemes are obvious on the intent. The standard degassing is a hybrid option that uses a proprietary algorithm that balances tradeoffs in cycle time, purge consumption, and graphite component rpms based on experiential derivations. SMARTT-enabled MTS units are all equipped with a datalogging feature that allows reports on each cycle to be downloaded into csv file or viewed through enabled software packages.

These reports capture as many as 41 cycle datapoints such as start time, start date total cycle times, the aforementioned environmental and parameters conditions, and other informational items such as whether a cycle was completed or aborted due to a general fault or other (e.g. Estop). These reports can prove invaluable when troubleshooting issues or identifying opportunities (e.g., times of the day or melt temperatures when cycles can be shortened).

An article published in the April 2020 issue of Modern Casting documents the efforts of a low-pressure aluminum foundry in the Midwest and its efforts to add blended gasses to add hydrogen into the melt. In the article, it was established that utilizing SMARTT versus a manual process side by side to upgas aluminum to a single specific gravity point target achieved several benefits. Among the benefits achieved when using SMARTT versus manual was a nearly fourfold reduction in standard deviation of specific gravities, a sevenfold reduction in distance of mean to the target specific gravity, and a reduction in the average achieved cycle time of 19% (Figures 2 and 3).

Because of the potential metallurgical and productivity benefits possible, a trial was scheduled to see if the intended benefits could be realized.

The compactness of the production melt deck layout provided the first hurdle to the trial. A photo of the melt deck is provided in Figure 4. Hence, the first order decision was identifying an MTS unit design compact enough to perform treatments in the three adjacent crucible furnaces. Another consideration was the robot guarding, which mandated that the control panel be remote from the mechanical body. Additionally, it was clear that the unit would have to be outfitted with quick connects/disconnects to permit transportation in/out, prior to and after treatments, without damaging the electrical connections.

Ultimately, a unit design was chosen that could be transported via an overhead hoist that can be set on the furnace but controlled from a remote panel. A dimensional schematic for the mechanical portion of the chosen MTS unit is shown in Figure 5.

The hydrogen trials were more complicated than the grain refining evaluation. The user of a SMARTT machine does not input a specific gravity target but rather a concentration of hydrogen (ml/100g) as a set point. Accordingly, onsite efforts had to be made to derive a conversion scale between the specific gravity test used and the concentration of hydrogen level for the various targets. For this exercise, both the aforementioned degassing model and a novel in situ hydrogen measurement apparatus were used to assist in setting the hydrogen concentration targets.

The most common two target-specific gravities were chosen for the preliminary evaluation. After some trial and error, the specific gravity results were hitting target on every cycle, so the process evaluation was complete.

The data from the SMARTT reports from the new unit were extracted on February 10, 2021. This new unit was installed on July 9, 2020, so the data encompasses cycles completed over an eight-month period. The data was very telling in terms of the variation in environmental and application conditions experienced in standard production. Histograms for some key environmental and application conditions appear in Figures 6-8. In these figures, another histogram appears in the upper right-hand corner that is the same histogram prepared from the results in the aforementioned Modern Casting article. The reference histograms are provided as a basis of comparison. A bar chart showing how many cycles for each of the various pre-programmed programs/targets was run over the eight-month period is presented in Figure 9.

A charting of eight-month humidity, ambient temp, melt temperature and cycle time data from Harmony Castings in this article and the other plant in the Modern Casting article is provided in Table 1. This information suggests the ambient temperature in Harmony, Pennsylvania, is nearly half the standard deviation as the other foundry dataset.

Moreover, the average humidity was nominally 6% RH lower, and the melt temperature control was much better with nearly half the standard deviation. On the downside, the standard deviation of the cycle time was nearly 2.5 times the other foundry; although, that is to be expected since many more different specific gravity levels are targeted in this recent study.

A simple F-test analysis was used within population. The results of the F-test are provided in Table 2.

In all the F-test evaluations, the F value exceeded F Critical so the null hypotheses were rejected. Accordingly, the datasets have a statistically different variance to a 5% confidence level. This data suggests that even sister facilities running operations in different locations should expect to see statistically different weather conditions variances; and accordingly, require different optimized melt treatment/degassing/upgassing treatment parameters/practices between them. The most fascinating result from the statistical testing is the ambient temperatures, since the arithmetic mean difference was less than 0.06F between the two datasets. Despite the similar mean temperatures, the dataset variations differ due to the large scatter of measure temperatures from the 2020 article dataset. The authors believed that a large part of the scatter from this dataset is due to the foundry layout. The foundry layout from the earlier Modern Casting article entails melt treatment/ degassing/upgassing being performed near two large shipping doors that will often remain open (when hot) while fluctuating between open and closed during those periods where the temperature is cooler. Harmony Castings differs greatly in that there are no shipping doors near the melting/treatment location; hence, air movement is minimized and likely sources more stable ambient conditions.

The same TA used to evaluate the incumbent process was used to evaluate several trial melts. As expected, the trial melts evaluated achieved a perfect grain fineness reading of 9.0, confirming the effectiveness of the grain refining flux. Similar to SMARTT, the TA device generates csv files of the tabularized results as well as for quick reference, and these were cataloged for future reference. Representative TA curves from before and after the new process implementation are shown in Figures 10a and 10b.
 

The results also showed immediate improvement in several areas, so a permanent unit was ordered with some modifications to replace the demonstration unit used for the trials. The first improvement noticed was in the mechanical properties (Table 3).

SMARRT reports were also used to establish the mean, mode, and median for the cycle times. A histogram of cycle times extracted from the same SMARTT data reported on in the experimental procedure is provided in Figure 11. Moreover, a bar chart showing the minimum, maximum, median, mode, and mean cycle times all versus the reference (most common cycle time of the old process) is presented in Figure 12.

Finally, porosity was reduced—meeting one of the primary objectives of this project. A histogram graphic of average porosity levels for 2018, 2019, and 2020 is displayed in Figure 13. Despite a similar casting mix, the porosity scrap was reduced from around 25% in 2019 to 10% in 2020. A further reduction in 2021 was anticipated due to an anomalously high scrap month in March 2020 attributed to staffing issues caused by the COVID-19 pandemic, and in fact, the reported porosity scrap in the first quarter of 2021 was about 5%.

The results of this project establish that using an Industry 4.0-style model integrated into a metal treatment station has full applicability into an aluminum alloy job shop environment, when many different hydrogen levels are required due to product mix. Incorporating this integrated degassing model with improved flux from grain refining and oxide cleaning can have many synergistic benefits. A successful implementation of this type in this project, led to the following benefits to the foundry: • Mechanical property improvements observed in Brinell hardness, ultimate tensile strength, yield strength, and, especially, elongation.

• A reduction in porosity scrap to less than half the original amount.

• An improvement in productivity by reducing the degas cycle by 400%.

• A calculated payback of savings realized when factoring the above project metrics of just under nine months.