Method for activity measurement in large packages of radioactive wastes. Is the overall activity stored inside a final repository systematically under-estimated?

The activity of a rad waste package is usually evaluated from gamma spectrometry measurements or dose rates emitted by the package, associated with transfer functions. These functions are calculated assuming that both activity and mass distributions are homogeneous. The proposed method, OPROF-STAT (patented) evaluates the error arising from this homogeneous assumption. This error has a systematic part, leading to an over or underestimation of the overall activity in a family of similar waste packages, and a stochastic part, whose mean effect on the overall activity of the family is null. The method consists in building a virtual family of packages, by numeric simulation of the filling of each package of the family. The simulated filling has a stochastic part, so that the mass and activity distributions inside a package are different from one package to another. The virtual packages are wholly known, which is not the case for the real family, and it is then possible to compute the result of a measurement, and the associated error, for each package of the virtual family. A way to fit and demonstrate the representativeness of the virtual is described. The main trends and parameters modifying the error are explored: a systematic underestimation of the activity in a large family of rad waste packages is possible. (author)