# Sand SQC cuts foundry defects.

Sand producers wash, dry, screen and blend sands to produce different grades. Relative fineness or coarseness can be measured by the percentages of sand retained on standard screens of different mesh sizes.

In common terms, sand is classified as a "three (or a four) screen." This refers to the number of screens (called "major" screens) that have at least 10% retained sand. Sand suppliers control and sell sand based on grain fineness number (GFN), an estimate of the mean sieve size calculated using the percentages of the sand retained on each screen. GFN also can describe a number of different screen distributions.

Casting quality requirements are forcing foundries into tighter process controls that include screen distribution of purchased sands. Studies of core and casting quality are helping foundries establish specifications for sand suppliers. Deciding whose screen distribution to use--the foundry's or the supplier's--is important because of test result variations between laboratories.

Sand Sampling

Sand segregation is a major obstacle to procuring representative sand samples. Handling sands causes segregation; coarse grains tend to move to the outside of the pile while fines tend to stay to the center. Different methods to accumulate a representative sample helps negate this problem.

A recommended procedure to prepare the lab sample uses a precision sample splitter to split the accumulated sand sample. The two halves are poured back into the splitter's hopper and split again for a total of four cycles. Then, one half of the sample is split and split again, until a sample of 40-100 grams is retained in each splitter pan.

One of these samples is weighed and run through the stack of screens. To determine the screen distribution, the sand retained on each screen is weighed. The individual weights are converted to percentages by dividing each weight by the total weight and multiplying by 100.

Each sand sample has a total of at least 10 numbers to record and monitor for statistical analysis. Keeping track of these screens with multiple types, different grades and suppliers is cumbersome even with a computer. Cpk capability values for individual screens will show how well the specification has been met, but it does not indicate acceptability of separate loads.

Figure 1 shows a typical graph of how well the separate screens percentages reflect the established specification limits. Even using just the "major" screens makes this technique lengthy and difficult to monitor. From a statistical process control (SPC) standpoint, a single value for the screen distribution would be beneficial. The statistic, "CHI Square," is a measure of the difference between actual and targeted frequencies of numbers. The formula for this calculation is:

CHI Sq'd = sum of:

(Actual -- Target) for each number/Target

By sharing screen data with one sand supplier, it was determined that a meaningful number was generated by using "major" screen percentages only; "minor" screens skewed the final sum. This number is called "CHI Sq'd" because this calculation is now a derivative of the statistical formula.

Screen Distribution

Accumulated data has shown that different suppliers for the same grade of sand can meet a CHI Sq'd value of less than 1.7. A maximum CHI Sq'd value should be determined for each sand used in any one formula.

Graphing of the CHI Sq'd values is a way to monitor how well a supplier is meeting the required screen distribution. To gain confidence in this one descriptive number, it is suggested that the "major" screens also be graphed for a period of time. Variations in the screen graphs will show those causing variations in CHI Sq'd values.

Depending on sand use, a monthly or quarterly analysis of Cpk for CHI Sq'd values over a period of time will show how well screen distribution specifications are being met. The Cpk values over a period of time will show if continuous improvement is being made.

Some of the benefits of better control of the sand screen distribution control using the CHI Sq'd analysis have shown:

* more consistent sand test results; * reduced binder consumption; * reduced core scrap; * reduced casting scrap; * better dimensional control; * improved casting finish.

In common terms, sand is classified as a "three (or a four) screen." This refers to the number of screens (called "major" screens) that have at least 10% retained sand. Sand suppliers control and sell sand based on grain fineness number (GFN), an estimate of the mean sieve size calculated using the percentages of the sand retained on each screen. GFN also can describe a number of different screen distributions.

Casting quality requirements are forcing foundries into tighter process controls that include screen distribution of purchased sands. Studies of core and casting quality are helping foundries establish specifications for sand suppliers. Deciding whose screen distribution to use--the foundry's or the supplier's--is important because of test result variations between laboratories.

Sand Sampling

Sand segregation is a major obstacle to procuring representative sand samples. Handling sands causes segregation; coarse grains tend to move to the outside of the pile while fines tend to stay to the center. Different methods to accumulate a representative sample helps negate this problem.

A recommended procedure to prepare the lab sample uses a precision sample splitter to split the accumulated sand sample. The two halves are poured back into the splitter's hopper and split again for a total of four cycles. Then, one half of the sample is split and split again, until a sample of 40-100 grams is retained in each splitter pan.

One of these samples is weighed and run through the stack of screens. To determine the screen distribution, the sand retained on each screen is weighed. The individual weights are converted to percentages by dividing each weight by the total weight and multiplying by 100.

Each sand sample has a total of at least 10 numbers to record and monitor for statistical analysis. Keeping track of these screens with multiple types, different grades and suppliers is cumbersome even with a computer. Cpk capability values for individual screens will show how well the specification has been met, but it does not indicate acceptability of separate loads.

Figure 1 shows a typical graph of how well the separate screens percentages reflect the established specification limits. Even using just the "major" screens makes this technique lengthy and difficult to monitor. From a statistical process control (SPC) standpoint, a single value for the screen distribution would be beneficial. The statistic, "CHI Square," is a measure of the difference between actual and targeted frequencies of numbers. The formula for this calculation is:

CHI Sq'd = sum of:

(Actual -- Target) for each number/Target

By sharing screen data with one sand supplier, it was determined that a meaningful number was generated by using "major" screen percentages only; "minor" screens skewed the final sum. This number is called "CHI Sq'd" because this calculation is now a derivative of the statistical formula.

Screen Distribution

Accumulated data has shown that different suppliers for the same grade of sand can meet a CHI Sq'd value of less than 1.7. A maximum CHI Sq'd value should be determined for each sand used in any one formula.

Graphing of the CHI Sq'd values is a way to monitor how well a supplier is meeting the required screen distribution. To gain confidence in this one descriptive number, it is suggested that the "major" screens also be graphed for a period of time. Variations in the screen graphs will show those causing variations in CHI Sq'd values.

Depending on sand use, a monthly or quarterly analysis of Cpk for CHI Sq'd values over a period of time will show how well screen distribution specifications are being met. The Cpk values over a period of time will show if continuous improvement is being made.

Some of the benefits of better control of the sand screen distribution control using the CHI Sq'd analysis have shown:

* more consistent sand test results; * reduced binder consumption; * reduced core scrap; * reduced casting scrap; * better dimensional control; * improved casting finish.

Printer friendly Cite/link Email Feedback | |

Title Annotation: | sand quality control |
---|---|

Author: | Volkmar, Alan |

Publication: | Modern Casting |

Date: | Dec 1, 1992 |

Words: | 678 |

Previous Article: | Conference reveals latest technology in coreless melting. |

Next Article: | Peer review may keep you out of court. |

Topics: |