A CALIBRATION PROBLEM IN NUCLEAR WASTE MANAGEMENT

The nature of nuclear waste makes its quantification a statistical problem. Moreover, for the detection of low levels of plutonium in waste creates, the standard regression techniques do not apply and can lead to costly mistakes. In this paper, the problem of labeling waste crates for offsite shipping and storage is discussed in terms of building a calibration model. It is shown that the most common approach using ordinary least squares regression is not suitable. A suitable approach using weighted least squares regression is given and a procedure for implementing the results outlined. Also included is a limited simulation study that provides support of the main guidelines in designing a calibration experiment of this type.