Prediction Algorithms for Performing Calorimetry Measurements in about One Hour

摘要

Calorimetry provides a precise and accurate measurement of the mass of radioisotopes with significant heat output. Calorimetry is frequently used for safeguard measurements of plutonium, but its use in treaty verification has not been practical due to the long measurement times. Calorimetry is currently being considered as an NDA option for attribute measurements in a treaty verification scenario. To that end, mathematical methods for extrapolating calorimetry data to obtain an equilibrium prediction in one hour have been investigated. Three different types of prediction algorithms will be discussed. These include weighted curve fits, numerical convergence algorithms and spectral analysis of the heat-up curve. The successful algorithms are capable of predicting the equilibrium values to within 2-20% in one hour, which is fit for the purpose of treaty verification, depending on size and mass of the item. Authentication of the respective algorithms will be discussed. The algorithms are currently being implemented into a deployable analysis routine for further testing on real-time data for realistic treaty verification scenarios. The application of the prediction algorithms to standard calorimetry applications will also be discussed. The spectral analysis method provides a direct measurement of the exponential components which are important physical parameters of the system. The mathematical treatment has the potential for increasing the precision of calorimetry measurements (albeit with increased measurement time), and of significantly decreasing measurement times with minimal decrease in accuracy.