|
|
|
|
Session:
|
8C - Calibration Laboratory Management
|
Date & Time:
|
Wednesday August 9, 2:30 PM - 4:00 PM
|
Paper Title:
|
Simplified Calibration Interval Analysis
|
|
|
Speaker:
|
Allen Bare, Metrology Engineer
|
CoAuthors:
|
|
Speaker Info
|
Company:
|
Westinghouse Savannah River Company
|
Address:
|
736-A
Aiken, SC, 29808, United States
|
Phone:
|
803-725-1131
|
Email:
|
Allen.Bare@srnl.doe.gov
|
|
|
Abstract:
|
There are many methods of assigning calibration recall cycles. Unfortunately many are complicated and cumbersome. A simplified method has been developed for laboratories with limited calibration histories. The method proposed may easily be modified to meet metrology program reliability goals. Many factors can influence decisions regarding the assignment of calibration intervals. In many instances customers will assign their own calibration intervals; ISO 17025 directs laboratories to not make any recommendation on the calibration interval except where it has been agreed on with the customer. However many calibration labs are interested in how to maximize their efficiency through adjusting their own calibration recall cycles or are mandated by a primary customer's quality department to assign calibration intervals. This method was developed to analyze calibration history with regard to both the simple as found condition (in or out of tolerance) and the degree of an out of tolerance condition should one exist (out of tolerance greater than 1x the tolerance, greater than 2x the tolerance, greater than 3x the tolerance, etc.). Analysis of these two variables indicates key factors that when evaluated historically may be developed into an algorithm for adjusting calibration cycles in accordance with the unique requirements of an individual metrology program. The method presented here is a modification of existing algorithms stated in NCSL Recommended Practice 1 (RP 1). This paper will discuss the selection of calibration intervals and the application of this interval algorithm in relationship to observed results and in relationship to the algorithms of NCSL RP 1.
|
|
|