Tritium accountancy and unmeasurable inventories in fusion reactors

For the time being fusion technology development involves relatively small quantities of tritium. Consequently, it is sufficient to apply so-called ''conventional'' accountancy tools. However, it is foreseeable that tritium operations - and thus the amount of tritium - will increase substantially. An advanced accountancy methodology will satisfy the resulting new requirements. In this study such an advanced accountancy methodolody is developed and applied to the situation envisaged with idealized experiments of the Karlsruhe Tritium Laboratory (TLK) as well as an idealized ITER-type fuel cycle. Firstly, this task comprises modeling of fuel cycle operations, providing the ''true'' data of the in-process inventories. As both the fuel cycle subsystems and networking themselves are susceptible to changes, a measurement model takes care of the true data, handles data reduction, and applies mathematical methods to confirm the final inventories on a statistical basis. Then, in a third step, the test statistics might verify whether or not a tritium anomaly, e.g. a tritium loss has occurred. Since the statistical analysis generates problems the solutions of which are not part of the standard statistical literature in the Annex the results of the related original work is presented in mathematical-abstract form. (orig.)