|
||||||||||
|
| DataLyzer.com |
|
Gauge Repeatability & Reproducibility at DataLyzer.com Gauge Repeatability & Reproducibility by Stephen Computer Services, Inc. The measuring process often contributes a significant amount of variation to the total variation of a manufacturing process. This measuring variation can be expressed in terms of a percentage of total variation or a percentage of the engineering tolerance range. There are several standardized ways of studying this variation. Selecting one over another depends on what type of measuring system is being used (variable or attribute–pass/fail gages). Beyond that, how much discrimination and precision is required vs. the amount of time is available to do the study. Gauge Repreatability & Reproducibility studies were developed by several divisions of GM and adopted as standards by the AIAG. Application Short Variable study (average and range method) This method is used for variable gauges. Discrimination is not as precise as long variable study. Only 2 operator/appraisers can be used instead of 3 in the long study. This study requires 2 appraisers to make one measurement on the same characteristic of five parts. To begin the study, select five production parts at random (or of sizes ranging through the specifications) and any 2 operator/appraisers who measure the parts in process. Each appraiser should measure each part once, independently, so neither appraiser knows the results of the other’s test (i.e. a blind test). The measurements are arranged in two columns so the results from each part are side by side. The range or difference between the 2 columns is calculated and printed in a 3rd column. The third column ranges are added together resulting in a column sum. The column sum is divided by 5 to render an average range (Rbar). This average range is divided by a constant from the d2 tables (1.19) get the estimated standard deviation. This value is multiplied by 5.15 to calculate gage error. (5.15 sigma covers 99% of the area under a normal curve. One can therefore have 99% confidence in the results of the study.) To make the results clearer, this gauge error is divided by the total engineering tolerance for the characteristic and expressed as a percentage of the tolerance. Long variable study (average and range method) This method is used for variable gages. It is the most accurate and most time consuming Variable gage study.  The long study has flexible parameters. Two or three appraisers/operators may used and each can make 2 or three trial measurements on 8 to 10 sample-product/characteristics. The sample products are chosen randomly from production (or of sizes ranging through the specifications). The appraisers are asked to to make the desired number of measurements (8-10) using their normal measuring techniques. Appraisers should take measurements independent of each other, so results between appraisers have no influence on each other. The data table will be filled in as shown in fig C. Readings are added down the columns and the sum entered at the bottom. The average range (Rbar) and Appraiser average (Xbar) are calculated for each appraiser and entered into the boxes indicated. The average of the average ranges (R double bar) and the difference in the highest and lowest operator averages are determined (Xbarmax-Xbarmin). A Control Limit for the ranges are calculated and any values found to exceed this limit should be re-measured and table re-calculated. The average range and the differences in the averages can be used to determine repeatability and reproducibility from the equations shown. (Fig. D) A high repeatability value (also known as equipment variation) can point to the need for gage maintenance, repair or redesign, the addition of fixturing, etc. A high reproducibility value (also known as appraiser variation) points to the need to increased measurement consistency between operators, possibly better measuring procedures or additional appraiser training to develop uniform measuring techniques between operators.  Either Total Tolerance (USL-LSL) or Total Process Variation (6sigma range) can be used for calculating the proportion used in measurement error depending on customer requirements. Page 2... Gauge Repeatability Reproducibility at DataLyzer.com |
| Copyright © 2004 Stephen Computer Services, Inc. All rights reserved. | DataLyzer.com |