High pressure molding equipment has been heavily used in modern foundries. This equipment can produce molds that are harder and denser at lower moisture levels and provide castings with better dimensional tolerances; however, these systems also require tighter sand controls. High pressure molding machines use hydraulic pistons to achieve the desired mold hardness. Hydraulic oil can enter a green sand system through these hydraulic pistons or various other equipment in the molding process that utilize oil or grease.

One foundry suspected potential hydraulic leaks in the green sand system, and its internal testing showed evidence of fluctuation in green strength, wet tensile strength, and loss on ignition that seemed to correlate with severe oil consumption. Additionally, foundry personnel perceived degradation of casting surface quality and higher overall scrap associated with oil consumption.
The foundry regularly sends samples to their bentonite supplier for analysis of green sand properties. The supplier was asked to determine whether hydraulic oil was present in the sand and what its potential effects on green sand properties could be. 

A review of available literature found most related topics dated and lacking current testing methodologies, so various foundries collaborated to quantify the impact of oil and how to detect and monitor the levels in their green sand system. 

Design of Experiments (DOE) methodology was used to predict the impact on sand properties as a function of oil contamination; the team also used a modified Hexane Extractable Material (HEM) test to quantify the level of oil and grease contamination.

Hexane Extractable Material (HEM) or Total Oil and Grease (TOG) is a method of detecting hydrocarbon compounds that have boiling points above hexane, the solvent prescribed by the test. This test has been used extensively in the wastewater industry to detect oil, grease, and other hydrocarbons. To apply this test to molding sand, the supplier mixed molding sand with the hexane solvent which dissolved the oil and grease. After mixing, a sample of the solvent was placed on the sample window of a commercially available TOG analyzer. The hexane evaporated and left a residue on the sample window. Infrared light passed through the dried sample and the amount of light absorbed at 2930 cm-1 was measured. Specific details of the test are described later in this study. The presence of hydraulic oil in the foundry’s molding sand was detected using this method. These findings were also confirmed by detecting the presence of oil with other methods including Thermo Gravimetric Analysis (TGA).

The test was able to detect measurable amounts of oil and grease from a foundry with a suspected oil contamination. Initially it was unclear if the values obtained were significantly different from other foundries. To establish a baseline for comparison, a series of commercial molding sands were tested for TOG. Samples were analyzed from different North American foundries and the results appear in Graph 1. The average results, standard deviation and values for a 95% confidence interval appear in Table 1. 

The study focused on contamination levels between 0.0-1.0%. An initial laboratory study was designed in which the foundry’s hydraulic oil was added to its molding sand at various levels from 0.1% to 1.0%. Results from this initial “ladder” study (“Experiment 1”) are presented in Tables 2, 3, and 4. A follow-up experiment (“Experiment 2”) was performed using a two-level, full factorial industrial designed experiment (DOE) to evaluate any statistically significant relationships between the amount of hydraulic oil present in the sand and various responses. 

In the designed experiment, two factors were designated for study: Factor A was the amount of hydraulic oil present in the sand and Factor B was the length of time the mixture was mulled in a laboratory muller. Analysis of Variance (ANOVA) tests were performed on the results using a commercial software package. ANOVA testing uses a hypothesis-based approach to compare sets of variances. The variation observed in the results were compared by assigning what variation was due to changes in the inputs versus random variation in the experiment. The ratios of the variation due to the inputs versus the random factors were compared and evaluated at a confidence level of 95% (p value of <0.05). The null hypothesis was that the average test results were the same regardless of high or low oil levels. The alternate hypothesis was that the average test results were significantly different between the high and low oil levels. The software performed a calculation to determine whether to reject or accept that hypothesis. The result of this experiment, as well as a list of significant relationships between green sand properties and hydraulic oil levels, appear in Tables 5, 6, 7, and 8.

Subsamples of 3,000-gram weight from the foundry were used to prepare individual batches of molding sand for testing. Each subsample was further corrected for weight and placed in a laboratory speed muller with the defined amount of oil and mixed for 10 seconds. After the oil was added, the required amount of water was added to reach the desired compactability target. For Experiment 1, each mixture was mulled for four minutes after the addition of oil and water and then discharged for immediate testing. For Experiment 2, the mulling time was varied according to the test design ranging from two to eight minutes. A list of the green sand responses measured and monitored in the study are listed in Table 9.

Total Oil and Grease Content (TOG) testing was performed by combining 40 grams of molding sand at “as received’ moisture with 40 mls of HPLC (high pressure liquid chromatography) grade Hexane in a polycarbonate Erlenmeyer flask with cap. The sample was mixed by hand with a swirling motion for two minutes. After agitating, a small aliquot of the supernatant from the mixture was placed on the sample stage of an instrument capable of measuring Total Oil and Grease, such as a InfraCal TOG/TPH analyzer. For this testing, the sample was allowed to evaporate for five minutes on the sample window before a reading was taken. The results obtained from the instrument were multiplied by 10 and recorded in parts per million (ppm) per manufacturer’s instruction. The accuracy of the calibration is +/- 20 ppm. Samples were run in duplicate and averaged. There may be variation from instrument to instrument; please refer to the instrument manual. Special safety precautions are required when handling and disposing of solvents such as hexane; please consult the Safety Data Sheet (SDS) and conduct a full safety review before testing.

Wet tensile strength (WTS) was of particular interest to include in this study. WTS has been shown to be sensitive to changes in green sand systems compared to other tests and can detect changes quickly. As described in several sources, WTS can be influenced by several factors, and it is often concluded that lower WTS values correlate to an increase in certain types of casting defects. There are ample examples where the wet tensile test has shown to detect changes in green sand systems quicker than other test results. 

In Experiment 1, a series of molding sands was prepared with known additions of the foundry’s hydraulic oil. The results from Experiment 1 showed a linear correlation between the amount of oil added and the total oil and grease content values (TOG) (Figure 3). Thermo gravimetric analysis (TGA) was run on the prepared samples and the results appear in Figure 2. The overlay of the TGA results show an increase in the amount of weight loss associated in the temperature range of 250-350C (482-662F) as a function of increased oil. In Experiment 1, the composite with no additional oil added was estimated to have 580 ppm TOG, which is significantly higher than the estimated average TOG from the 95% confidence interval exercise in Table 1. Review of the data in Tables 2, 3 and 4 indicate the green compression strength (GCS), permeability, wet tensile strength (WTS) and working bond of the sands decrease in a linear relationship with the amount of oil added. A comparison between GCS at 0% and 1.0% oil added shows a decrease of approximately 13%, while WTS shows a decrease of approximately 25%. The moisture of the batches controlled at 40% compactability resulted in minimal variation in GCS, yielding only small changes to the working bond, available bond and muller efficiency. WTS, TOG, dry compression strength (DCS), and LOI all exhibited significant changes between 0% and 1% hydraulic oil added. Figures 3 and 4 show the respective correlations between TOG and WTS and the amount of hydraulic oil added to the test batch. The data also shows that LOI and the ratio of LOI and active clay increased linearly with the addition of hydraulic oil.

Experiment 2, an extension of Experiment 1, evaluated the impact of the amount of oil added to the molding sand, as well as the mulling time. The experiments were set up with a high and low level for each factor, and the average results were evaluated using a commercially available statistical software program to look for statistically significant relationships between the factors and the responses. Like findings in Experiment 1, several responses were found to move predictably with changes in the amount of hydraulic oil present in the sand. The ANOVA exercise in Experiment 2 produced a strong predictive model for the relationship between the permeability, TOG, WTS, LOI and the amount of hydraulic oil present in the sand. Figures 5 and 6 show contour graphs generated with the commercial software. The graphs show the predicted change in TOG and WTS and a function of increased oil concentration and mulling time. The ANOVA results indicate a high probability that changes in TOG, WTS, and LOI are the result of changes in the amount of hydraulic oil added to the sands and not due to other factors. The models predicted decreases in WTS and increases in TOG and LOI as the amount of hydraulic oil increased. All significant models and factors found for Experiment 2 are listed in Table 8. This table outlines which factors in the model were found to be statistically significant based on the data found in Tables 5, 6, and 7.

Researchers observed in Experiment 2 that WTS and GCS trended slightly higher than what was observed in testing of weekly samples. The difference was believed to be due to the additional mulling compared to the weekly samples from the foundry. It is believed that the additional mulling, while not a statistically significant factor, had an impact on some of the green sand responses. The regression equation from Experiment 1 for hydraulic oil contamination versus WTS (Graph 3) and the predictive model generated from Experiment 2 were used to estimate the amount of oil contamination. The calculations using WTS as the input resulted in typical estimates between 2% and 3% oil in the sand. The slopes in the regression equations for WTS were different and are believed to be influenced by the difference in mulling time. The regression equations generated for Experiment 1 and Experiment 2 for hydraulic oil added versus the TOG have similar slopes and suggest the base molding sand, with a TOG value of 580 ppm, contained approximately 0.10% residual oil. An estimate of 0.10% residual TOG is a better fit with reported values by the foundry of approximately 0.19% total oil leakage.

The data shows that while WTS may be influenced by the hydraulic oil content, other factors also influence WTS. Conversely, TOG data is only influenced by the amount of hydraulic oil present and can be used to directly estimate the oil concentration in the sand. The GCS data did not produce a statistically significant relationship but did show interesting results. The decrease in GCS between 0% oil added and 0.7% oil added was between 0.7 and 1.1 psi (between 2.5% and 5% decrease), and resulted in small decreases to working bond, available bond, and muller efficiency. The data again indicated the sensitivity of the WTS test. The WTS dropped over 18% with the addition of 0.7% hydraulic oil compared to less than 5% decrease in GCS. In Experiment 2, the LOI was also sensitive to the addition of hydraulic oil; the LOI increased approximately 18% with the addition of 0.7% hydraulic oil.

In summary, the lab data supported that increased hydraulic oil contamination led to increased LOI, and decreased GCS, permeability, working bond, and muller efficiency. Additionally, the lab data shows that WTS can be a predictor of an increase of hydraulic oil content in molding sands. The results of the study also show that the TOG test can be used as a direct indicator of oil and grease content and can be used to model and estimate total oil contamination in a molding sand.

Over the course of the eight months covered in this study, a concerted effort was made at the participating foundry to identify and reduce oil leaks into the sand system. The foundry continued to conduct regular green sand testing, in addition to sending samples to their bentonite bond supplier for analysis. The test results from both foundry and supplier showed similar trends of higher WTS as the foundry focused on reducing the oil contamination. The best correlation was found between the foundry’s lab LOI data and the supplier’s TOG results (Fig. 7). Given the foundry’s consistent new sand addition, nearly constant clay load, and no core sand additions, the overall system remained very stable. This allowed the LOI test to become a useful predictor of the TOG results.

With the focus on mitigating oil leaks, overall maintenance of the production line also improved, and the foundry achieved an overall scrap reduction of over 50% directly related to these maintenance efforts.

While the wet tensile data showed a strong inverse correlation during the initial reduction of oil, the correlation did not hold during later months (Fig. 8). Wet tensile testing is a new test for this foundry, in use for less than 18 months. During that time some seasonality has been found with the results. Additional factors, specifically related to inconsistent water quality, could be causing this fluctuation in wet tensile strength. The wet tensile test, in this case, may be detecting changes in multiple variables that impact the correlation to TOG. 

The LOI test served as a good indicator of increases in hydraulic oil for this foundry because there were few other factors besides carbon, clay, and oil level that influence LOI.

Since other factors can influence the wet tensile strength or LOI of a molding sand from foundry to foundry, changes to these properties would need be evaluated relative to typical or baseline levels at a facility, which can make comparison across foundries difficult. The total oil and grease measurement is related only to the amount of oil and grease in the molding sand. Values can be compared across different foundries without needing to account for clay type, active clay levels or levels of carbonaceous additives. Testing of a variety of molding sands from commercial foundries in North America showed that levels of oil and grease above 100 ppm may indicate the presence of hydraulic oil and potential negative side effects.

Both the experimental lab data and routine green sand data supported the foundry’s initiative to find and repair leaks. The foundry found that tracking the wet tensile data, loss on ignition, permeability, and ratio of loss on ignition to active clay within its operation provided information on directional changes in hydraulic oil contamination. The foundry also found good correlation between its internal LOI data, and the oil and grease measurements performed by the supplier. The foundry concluded the total oil and grease content measurement was the most direct measure of suspected oil contamination, but that LOI correlated well and is useful to monitor changes in their system related to oil concentration.

Click here to view the article in the October 2022 digital edition.