Abstract:
Management systems developed according to ISO/IEC 17025 must fulfil several requirements,
being particularly relevant the assurance of the measurement results quality. The enhancement of
that quality depends upon several conditions, one of which is the instrumentation metrological
performance and its service conditions along its lifetime.
In this context, the evaluation of long time drifts in the instrumentation metrological performance is
a major concern requiring the development of calibration planning under traceable conditions, to
determine possible corrections and to establish the best measurement capabilities.
The periodicity of the calibration operations is a technical decision with considerable implications in
the system management performance (namely, operational costs and management of decision
risk). Although this decision can be significantly cost effective, many industries and laboratories
apply recommended calibration intervals regardless of any type of data analysis concerning
aspects such as the severity of use and the different accuracy requirements associated with
specific applications.
Considering that there is a relation between the calibration intervals and the expectations of
“failure” of a measurement instrument (“failure” meaning that the instrument will be out-of-tolerance
considering its own usage requirements), to combine the stochastic nature of the problem with
statistical methods to predict the optimized calibration interval is a challenge to many management
systems.
The effort towards this optimization requires prior knowledge of critical variables, the history of
calibration data, methods and models suitable to perform the statistical analysis of the collected
data. The optimized definition of the calibration intervals based in statistical analysis can provide
robust solutions, allowing the improvement of the management system in both economical and
quality performances.
This paper presents concepts, discusses methods, models and their variables that can be used to
define calibration intervals. Moreover, a reliability approach to a cost-effectiveness analysis is also
presented. Some examples concerning the experience of the above mentioned metrological
laboratories are used in order to enhance some of the remarks and conclusions.
