Non Invasive Water Level Monitoring on Boiling Water Reactors Using Internal Gamma Radiation: Application of Soft Computing Methods

Abstract

To provide best knowledge about safety-related water level values in boiling water reactors (BWR) is essentially for operational regime. For the water level determination hydrostatic level measurement systems are almost exclusively applied, because they stand the test over many decades in conventional and nuclear power plants (NPP). Due to the steam generation especially in BWR a specific phenomenon occurs which leads to a water-steam mixture level in the reactor annular space and reactor plenum. The mixture level is a high transient non-measurable value concerning the hydrostatic water level measuring system and it significantly differs from the measured collapsed water level. In particular, during operational and accidental transient processes like fast negative pressure transients, the monitoring of these water levels is very important. In addition to the hydrostatic water level measurement system a diverse water level measurement system for BWR should be used. A real physical diversity is given by gamma radiation distribution inside and outside the reactor pressure vessel correlating with the water level. The vertical gamma radiation distribution depends on the water level, but it is also a function of the neutron flux and the coolant recirculation pump speed. For the water level monitoring, special algorithms are required. An analytical determination of the gamma radiation distribution outside the reactor pressure vessel is impossible due to the multitude of radiation of physical processes, complicated non-stationary radiation source distribution and complex geometry of fixtures. For creating suited algorithms Soft Computing methods (Fuzzy Sets Theory, Artificial Neural Networks, etc.) will be used. Therefore, a database containing input values (gamma radiation distribution) and output values (water levels) had to be built. Here, the database was established by experiments (data from BWR and from a test setup) and simulation with the authorised thermofluid code ATHLET.