PREDICTIVE CALORIMETRIC DETERMINATIONS FOR THE INVESTIGATIONS OF NUCLEAR MATERIALS

Among all the techniques that are available for the characterization of nuclear materials, calorimetry is considered to be one of the most accurate. The calorimetric technique provides a precise measurement, not subject to matrix effects as it is the case with emission or absorption methods. No initial sample preparation is needed for this non destructive technique. The size of the calorimeters is such that they are directly adapted for use with standard containers used in the nuclear industry. The assay is independent of the sample geometry, the nuclear material distribution and especially adapted for the investigations of nuclear wastes. In the case of tritiated wastes, determining the amount of tritium is difficult because beta counting of a sample can only be realized by a destructive method. In this case the counting is affected by large uncertainties as far as the wastes are generally heterogeneously tritiated (due to different matrix absorption of tritium) and as far as the container is often large. For such a tritium detection, calorimetry is a more precise method. For the plutonium control and accountability, calorimetry in association with gamma and/or neutron spectrometry for isotopic composition measurement, is the most accurate NDA measurement technique for bulk materials, as standard methods for measuring particle emissions are very often liable to attenuation problems due to matrix effects from packaging (glass, plastic, metal…). However the calorimetric technique suffers from the time (several hours) needed to obtain a steady-state signal. This parameter depends on the time of response of the calorimeter, but mainly the thermal inertia due to the large heat capacity of the container. To reduce the duration of measurement, the problem has to be numerically solved. For its range of large volume calorimeters, Setaram has developed a software for a predictive determination of the calorimetric signal. The calculation is based on the analysis of the calorimetric curve that is described by an exponential relation S(t) = A . exp (–t/t) ,with t, the time of response for the calorimeter. The calorimetric signal is corrected to get the predictive value Sc (t) = S(t) + t . dS/dt. With such a software, it is no more needed to wait for the steady-state signal, and the duration of the test is drastically reduced without affecting the accuracy of the measurement.