For those who frequently step foot into different foundries, you will find that each foundry has different ideas and priorities when it comes to sand testing. Some foundries have one or several dedicated sand lab technicians, while others have operators who are sharing many duties in the foundry. Sometimes you see the latest, greatest testing equipment, like the digital compactibility tester, where others are using the tried-and-true 3-ram tester. These foundries all have something in common: They want to capture the quality of their sand systems and make good decisions based off the data they collect. 

Sand testing is a cost-benefit analysis. What personnel resources need to be put toward sand testing to ensure good, reliable process control of the sand system? Foundries don’t sell sand, yet in some ways, sand is still a foundry’s product that they want to ensure has predictable behavior to prevent introducing even more variation—and ultimately scrap—into the casting process. 

Data Quality

To determine the best sand testing frequencies, it is important to consider data quality. Good decisions can only be made from testing data, both short-term and long-term, if it is high-quality and accurately represents the system. 

The following five characteristics of data quality should be considered:

1.           Accuracy. Ensure that recorded data reflect actual events and real measurements. In the sand lab, an example of inaccuracy could be transposing or entering incorrect values into data sets, such as entering a value like 78 instead of 7.8 for a methylene blue test.

2.           Completeness. Data should not be missing records. This may commonly happen if you have a testing schedule that operators are having a hard time meeting or inconsistent management of testing.

3.           Timeliness. The measurement should be up to date to sand production. A compactibility test taken days or weeks ago is likely irrelevant to current sand quality. Each test may require different frequencies to be considered timely. 

4.           Consistency. Testing should be understandable and relevant between operators and datasets. This may involve ensuring sand tests are being taken from the same location and that data is compared between like locations. For foundries with multiple mullers or molding lines, data may or may not be comparable between the two. 

5.           Granularity. This means the level of detail in the dataset. The depth of the data, such as the number of tests being conducted in a standard sand panel test, or how the data is being aggregated. 

It is important to take these data quality characteristics into account when designing your sand lab testing frequencies. The following are two case studies that illustrate problems that can occur with either too frequent or too infrequent testing. 

Foundry A: 
High Frequency Testing Introducing System Variation

Methylene blue (MB Clay) is a test conducted in greensand foundries to determine the amount of active clay available in the system. It is typically one of the top measurements that are used to adjust the sand system, but the test can be fairly operator dependent.

At Foundry A, operators tested MB Clay every 45 minutes and the system turned over about every three hours. The operator’s job was to make individual adjustments to the bond setpoint, and they were required to adjust if they had an out-of-spec value. 

The foundry’s biggest complaint was a low percentage of MB Clay tests within the specification, with the MB Clay data being distributed bimodally—peaking at the specification limits, instead of normally distributed. Figure 1 presents illustrative data that shows the MB Clay test results and corresponding bond setpoint. The vertical dash-dot lines represent the three-hour sand system turnover.

In the initial shaded area of the first graph, the MB clay is low out-of-spec at the beginning of the shift, so the operator increased the bond setpoint. The operator tested four times before the system turned over and after each test increased the bond set point by 0.5. As the system turns over, the MB Clay came into spec, but on the next turnover the bond setting was over-corrected, resulting in high out-of-spec tests. 

The next region shows the next shift, with a new operator. This operator responded by lowering the bond setpoint as the MB Clay continued to read high out of spec. The pattern continues the during the third shift, with the setpoint now being increased to make up for the MB Clay values dropping out. 

The testing data starts to show a pattern of sinusoidal behavior, with increasing waves of out-of-spec data worsened by operator influence. This type of tail chasing, while not always quite as dramatic as this illustrative data shows, can result in poor quality castings from operator input.

The second chart shows what the data looked like the month after Foundry A decreased laboratory testing. Instead of every 45 minutes, tests were taken every 1.5 hours. Operators were also no longer required to act every time an out-of-spec data point was recorded and instead were encouraged to follow normal operating setpoints for the bond addition and consult a supervisor in case of dramatic shifts. 

What can be observed in this data is there was a real process shift at the 14:00 hour mark—perhaps due to higher mold-metal ratio castings being run in the previous hours, which before may have been masked or exacerbated by the excessive operator influence on the MB Clay. In this case, less frequent testing, and less rigid decision-making, revealed true sand system behavior. 

Foundry B: 
Low Frequency Testing Hiding Critical Problems

Foundry B faced issues many foundries do, with high turnover and reduced staff. As a result, sand testing was affected. Sand lab testing staff, and the frequencies and tests that were being conducted, decreased. 

The sand sieve analysis, which looks at the distribution of sand sizes across a series of decreasing mesh sizes, previously conducted once a week, started to be conducted at best once a month. Other typically more frequent tests, such as the percent moisture, compactibility, and MB Clay, started being taken once a day or less. 

Throughout the next several months, Foundry B started to see a worsening casting surface finish with pinhole and gas defects. They also had several personnel changes, including multiple changes in management of the sand lab. This meant testing and analysis of testing fell by the wayside. 

Foundry B resorted to metallurgical additions to try to correct the gas issues, with little improvement. Overall, the sand system problems started to get worse and those daily tests that were being conducted became increasingly hard to interpret, as critical information such as sand and bond addition rates were not being recorded. How was the foundry to determine what changes had impact to the casting quality? 

Once staff turnover settled down and sand lab staff were re-trained and more frequent testing resumed, it was found that the sand sieve distribution showed an abnormally high, and increasing, grain fineness number. As more investigation was done, it was determined that new sand additions were not being made regularly, as there was not a system in place for automated additions of new sand to the system. This, along with other neglected testing, resulted in the higher grain fineness number (GFN) and higher moisture sand, which caused casting defects. Foundry B’s example demonstrates how low testing frequency testing can quickly affect casting quality. 

Establishing Your Foundry’s Testing Frequency

While it may be tempting to look at another well running foundry’s sampling plan and apply it to your own system, the best testing frequencies will likely look different at every foundry, even in companies with multiple foundries. When working to determine if your foundry’s sand testing plan is appropriate, consider the following factors:

Shifts. Consider operating shifts and if testing can be conducted on all shifts. Start-up sand behavior may differ from already warmed sand, but it can be common for testing to be neglected on off-shifts. 

Work Schedule. When days of operation are adjusted, such as adding a day of production in the week, see if your testing schedule and sand lab staffing accounts for these changes. 

Job Instructions. A foundry should outline the tests and frequencies in their job instructions. Often, foundries may rely on institutional knowledge, but when employee turnover inevitably happens, testing frequencies can slip, which can result in missing important system changes. 

Training. Train new and existing operators in job instructions, both when changes are made and on a regular basis. Assess and work to improve their understanding of the sand system, especially how their job tasks could affect it. 

Routines. Operator routines can also easily obscure system issues. If an operator always collects a sample at the same time each day, they could be missing activities that occur at other times of the day. One common example is belt spill sand being added back into the system, which could impact casting quality. It can be beneficial to occasionally deviate from the routine.

System Turnover. As discussed in Foundry A, your system turnover time can be an important factor in determining when to take sand samples. Calculating the sand system tonnage can help a foundry determine whether changes being made to the system are having an effect or if they are a product of natural variation. 

Conclusion

It’s important to select appropriate sand testing frequencies and regularly evaluate whether current testing is meeting your foundry’s needs. 

Foundries need to have high-quality, meaningful data to make good decisions for casting quality, but balance the cost and time it takes to sample the sand system. The sand lab should aim to measure and respond to variation, without inducing it as seen in the Foundry A example. 

Foundry B demonstrated how reduction of testing frequencies and forgetting the history of system changes can quickly affect casting quality and the sand system performance—and that sand testing still needs to be made a priority during tumultuous times. 

Foundries must work to understand their individual sand systems, including aspects such as personnel, system capacity, and inputs and outputs to optimize their sand production.