Leak rate testing in the range of leak detection systems

Highlights: • Leak tests in the range of typical detection systems are done. • Modular leak specimen allow to study slits, cracks and holes. • Model uncertainties are quantified and ranked. • The impact on leak-before-break assessment is discussed. Leak-before-break assessments of piping in nuclear power plants require an accurate prediction of leakage rates through thin slits and the understanding of uncertainties. Two test rigs at the University of Stuttgart are dedicated to the measurement of flow rates in the order of the typical leak monitoring specification of 1 gal/min. Several hundred measurements of flow rates (single phase and two-phase) for different leak geometries and fluid conditions are performed, which fosters the data amount in this range significantly. Different established models for the prediction of leak flow rates have been applied and compared to the measured flow rates. The homogeneous data of the tests allows a quantitative understanding of the prediction uncertainties of the individual models. In the two-phase flow regime, about 90% of the tests are within 40% of deviation between measurement and computations, and for about 40% of the tests the deviation is even only 10% or less. By comparing five different models with different accuracy quantifications, the homogeneous equilibrium approaches with friction (Estorf model, Pana model) have the best overall result, while the Henry model has the worst one.