ESTIMATION OF REACTIVITY COEFFICIENTS OF VVER-1200 REACTORS ON THE BASIS OF MEASUREMENTS MADE AT POWER LEVELS
29th Symposium of AER on VVER Reactor Physics and Reactor Safety (2019, Energoland, Mochovce NPP, Slovakia)
reactor dynamics and safety analysis
K. Gomozov, A. Pinegin, S. Semenov, A. Scheglov, NRC “Kurchatov institute”, Russian Federation
For start-up of new types of reactors wide series of experiments are performed. The purpose of such experiments is to obtain experimental data that allow to estimate real values of some neutronic parameters of the core. In particularly, during the start-up of VVER-1200 reactor the experiments of initiating axial and diametral xenon oscillations were performed with help of moving CR CPS group and specific CR CPS correspondingly. After CR CPS had been moved the damped oscillations of spatial power distribution were observed. Spatial power distribution came gradually into equilibrium state due to negative feedbacks. The course of these oscillations, their amplitudes, damping decrements, periods, phases depend primarily on the values of reactivity coefficients from fuel temperature and of moderator density, as well as on the cross section of xenon-135, the effectiveness of the SC RC, the geometry of the core and a number of other factors. The parameters of the core model, which ensure the conformity of the calculated and experimental data on the oscillations, are taken as their experimental values. The values of parameters of the core model that ensure conformity of calculated and experimental data on oscillations are considered as close to experimental data.
Since there are a lot of significant model parameters,the attempt of estimation all the significant parameters of the core for this process on the basis of one experiment often leads to degeneracy of task. In this case the conformity of calculated and experimental data may take place for different sets of calculated model parameters. That’s why the attempt of estimation of model parameters on the basis of two experiments (initiating axial and diametral xenon oscillations) was performed. Such approach is also effective due to reactivity coefficients (in particularly, a reactivity coefficient by coolant density) have different effects on the course of these processes.
For the estimation of probability density function Bayes method was used. The Bayes method was implemented in DRAM method. Priori information for unknown parameters was formed on the basis of their estimations. These estimations were obtained during similar experiments on other VVER reactors. The estimation of error in experimental data was based on the estimation of measurements error in VVER-1200 reactor.
During the construction of Markov chain the procedure of multidimensional interpolation on non-uniform mesh was used. For this purpose the direct simulation of series of processes under consideration by the code of mutual neutronic and thermophysical calculations NOSTRA had been performed before calculations by DRAM method was performed. Further, the series of direct calculations through equal number of steps of Markov chain were added with new calculations. This ensured the high accuracy of interpolation procedure in the point of maximum of probability density function.
In the paper the most probable values of six unknown model parameters of the core including reactivity coefficients, and twodimensional cuts for probability density function for these parameters provided.