Contact Us
Solve Your Materials Challenges
Find out how we can help
Sampling is a method to obtain assurance that materials are of satisfactory quality. Instead of 100% inspection, a portion of the material is examined to show evidence of the quality of the whole.
Statistical acceptance sampling tables and statistical process-control sampling tables have been developed to meet the needs of sampling method.
The specific influence of the following variables must be considered for proper interpretation of the results obtained:
Sampling is a method to obtain assurance that materials are of satisfactory quality. Instead of 100% inspection, a portion of the material is examined to show evidence of the quality of the whole. There are two important needs for this approach: first, in the final inspection or tests made to ensure that products delivered are in conformance with specification requirements; second, to control parts and assemblies while they are being processed. Statistical acceptance sampling tables and statistical process-control sampling tables have been developed to meet these needs.
Acceptance sampling may be conducted on a go/no-go (or attributes) basis, that is, determining whether or not the units of the sample meet the specification. Examination of the samples may also be conducted on a measurements (or variables) basis, that is, determining actual readings on the units in the sample. The majority of acceptance sampling is carried out on a sampling by attributes basis and the usual acceptance sampling table is designed for go/no-go data.
Process control sampling may be conducted on material during the course of production to prevent large quantities of defective parts being found in the acceptance tests. Many parts and materials are subjected to several successive machining or processing operations before they become finished units.
Parts can be most effectively controlled during production by examining small samples of these parts at regularly scheduled intervals. The object of this process check is to provide a continuous picture of the quality of parts being produced. This helps prevent production of defective parts by stopping and correcting the problem as soon as it begins to appear in the manufacturing process and thereby keeping the process in control. Sampling may be by attributes or by variable and process control sampling tables. The measurements (variables) control chart is by far the most effective process control technique.
Statistical sampling tables have four definite features:
The electromagnetic sorting method is primarily one of comparison between pieces. Empirical data and physical tests determine classification. The calibration and standardization procedure shall be governed by the properties of the sample requiring separation.
When using the absolute coil method, insert the known acceptable calibration standard to a fixed position in or relative to the coil and adjust the test instrument to get an on-scale meter or scope reading, or both. Replace the acceptable standard with a known unacceptable standard in the same exact position and adjust the sensitivity of the instrument to maximize the indicator difference reading without exceeding 90% of the available scale range.
When using the comparative coil method, select a reference piece and place it in the reference coil in such a way that it will not be disturbed, and set this coil and reference piece out of the way. For this method, when confronted with a two-way mix, choose two calibrated standards, one of which represents the acceptable and the other the unacceptable group. Place the acceptable calibration standard to a fixed position in the test coil coinciding with the position of the reference piece in the reference coil and balance the instrument. Replace this acceptable calibration standard with one representing the unacceptable group and adjust the test instrument’s phase, sensitivity, and coil current; then index to maximize the indicator reading without exceeding 90% of the available scale range. Reinsert the acceptable standard and alternately readjust the instrument controls to retain a null value for the acceptable standard and maximum indication for the unacceptable standard.
For a three-way sort, it is best to have three calibration standards, two of which represent the high and low limits of acceptability for one group or one each of the two unacceptable groups. The third standard, of course, represents the acceptable lot of material.
When high current is used in the comparative testing method, the reference piece is likely to heat up and this will change its magnetic properties. It is necessary to provide for cooling or to have several identical reference pieces so that they can be interchanged to prevent drift in the balance point. Another method that may be used is to momentarily turn off power to the coils when pieces are not being tested.
The results of any nondestructive testing procedure are based on the comparison of an unknown with a standard. Unless all of the significant interrelationships of material or product properties are understood and measurable for both standard and unknown samples, the test results may be meaningless.
Electromagnetic sorting is best used for repetitive tests on material "identical" in shape, composition, and metallurgical structure, and not for tests on grossly different materials. Electromagnetic sorting is generally not useful if there is limited knowledge of the properties of the unknown or test material.
Interpretation of data depends upon the degree to which the test materials compare with the reference materials. It is necessary to have all variables, except the one selected as a basis for sorting, under sufficient control if the measured variation is to be properly interpreted. Results can often be interpreted or explained by a processing change, such as in temperature, composition, and inclusions, when the measured property is known to be a function of the processing procedures.
When products grossly different in shape, alloy, permeability, or conductivity are to be measured, only a general interpretation of results can be made. The materials can be said to be different, but the how and the why of the difference usually is not determinable.
When the spread in value of the measured variable is sufficient, electromagnetic sorting can be 100 % effective. However, there may be cases where a single test will not show a clear separation. Often a second test or procedure can be used to further define the separation of materials. For example, a change in test frequency may show the effect of a second variable.
Shape and surface variations can mask the test results. If surface hardness is desired as the basis for sorting, all material should have composition and surface roughness under sufficient control so that effects of variations in hardness can be separated.