Optimization of Measurement Control

Keeping materials-balance variances within acceptable limits requires some form of measurement control. For processes operating near steady-state with small fluctuations in inventory (the desired case in high-throughput commercial facilities) the net transfer measurement errors soon become the dominant component of the materials balance variance. Controlling the net transfer measurement error so that systematic errors do not propagate over long time intervals is possible through judicious recalibration of the transfer measurement instrument. However, if the process contains many such instruments, with a variety of error variances and calibration costs, then the best allocation of limited recalibration resources may be uncertain. This paper proposes a simple method for allocating resources that is optimal in the sense of achieving the minimum net transfer variance within a resource constraint. Use of the method requires only a common measure of recalibration costs and the knowledge of systematic error variance of each instrument. This method allows: a proposed process to be analyzed for the minimum net transfer variance attainable at any level of recalibration resource investment, an existing process to be analyzed for the impact on the net transfer variance of proposed instrumentation changes, and specific instrument recalibration schedules to be developed for use in process simulations. The utility of the method is illustrated by an example drawn from a commercial reprocessing plant in which reductions in the net transfer variance attained with an optimal recalibration strategy are compared with some other intuitively reasonable strategies.