AAOM Handbook

INTERNAL

MP.14 Define the sampling interval

Context

Storage capacity within, or between, processes provides a natural filter that smooths variation (noise) within a process, and buffers connected processes against the effects of variation occurring in one of the connected elements. The capacity of these storage elements, coupled with the maximum throughput rate of the process, help us to determine how quickly a significant change in the process can occur and hence how frequently we need to look at a measure - the sampling interval. For example, if you must maintain the concentration of a solution within specification limits, and the accuracy of the measuring device is 5%, you will probably wish to be able to identify any 5% change in concentration. The minimum time interval in which this can occur is the determined by the volume of the solution and the maximum rate at which a chemical can be added. If we take a measurement more frequently than the minimum time in which a 5% change can occur then we will always detect the change. If we take the measure much less frequently than the minimum time in which a 5% change can occur then variation that occurs between sampling intervals will not be seen. Selecting the sampling interval for a measure is a trade-off between the maximum rate of variation that you will identify and the effort required to collect the measurement.

Purpose

To determine the sampling frequency for each measure

Quantity One sampling plan for each measure

Quality

The Nyquist criteria state that the maximum frequency of variation that can be represented by the data has a period of twice the sampling rate. Stated another way – you must sample at less than half the time interval in which the change you wish to detect can occur.

Select a sampling interval after considering storage capacities, maximum throughput and the implication of the Nyquist criteria.

• If the sampling interval is equal to or greater than the Nyquist rate then reconstructing a graph by connecting the instantaneous values at the sampling interval represents the relevant performance of the process between samples.

© McAlear Management Consultants 2007

Page 61 of 66