Considerations for Determining the Frequency of Instrument Measurement Control Checks

A new Quality Culture is slowly taking hold in the United States. The premise of this culture is that by considering quality from the start and by continually striving to improve, costs can be reduced while producing a better product. It is not always recognized that this is true whether the product is a manufactured item, a service or a measurement. Thus, the principles of Statistical Process Control (SPC) apply to processes that produce measurements as well as to those that produce items. In DOE facilities, a major concern is the assurance of quality measurements of nuclear materials. One consideration in the application of SPC to these measurement systems is the choice of the sampling interval, the frequency at which measurement control (MC) checks are performed. This paper gives several considerations related to the choice of the sampling interval for these systems. The considerations include the cost of an MC check, cost of failing to discover a system problem, regulations, specifications, and available resources. One factor that often is not adequately considered in the choice of a sampling interval is the cost of a response to a measurement system failure. When a system fails, the cause of the problem must be sought and the system fixed. In addition, all measurements made since the last successful MC check become suspect. Moreover, if that last "successful" check is marginal, measurements before that check may also be suspect. Providing assurance of the quality of these measurements, or remeasuring the items if necessary, can be a considerable cost. The procedures for addressing a system failure should be documented in a failure response plan, and the results of the investigation of a system failure should be kept for audit.