Improvement Of PWR Depletion Calculation Method And Research Of Dispersed Particle Fuel Depletion Calculation

In reactor physics,the variation of fuel compostion due to nuclear fission,neutron reaction,radioactive decay and other factors is called “burnup”.The purpose of burnup calculation is to accurately solve the fuel composition change in the whole life cycle.Burnup calculation is an indispensable part of reactor physics calculation.It plays an important role in reactor fuel management,nuclear fuel conversion,breed and spent fuel analysis.With the development of computer technology,high-precision and efficient burnup calculation algorithm is the most fundamental problem of burnup calculation.At present,reactor burnup analysis is based on the traditional coupling framework of burnup and transport,and the calling of transport calculations during the whole life is the bottleneck of the whole core burnup transport coupling calculation.Meanwhile,the pursuit of calculation accuracy in reactor physics high-fidelity calculation is becoming higher.However,due to the large number of burnup regions of core,the memory consumed by using detailed burnup library is still a bottleneck problem.In recent years,with the design of new accident tolerant fuels,how to accurately solve the burnup calculation of dispersed particle fuels has gradually become a difficult problem.A series of problems need to be solved,thus relevant researches are carried out in this paper.First,based on the CRAM algorithm of partial fractional decomposition form and incomplete partial decomposition form,the point-depletion code NES is developed.The code adopts a simplified burnup library of 198 nuclides and a detailed burnup library of1487 nuclides.In the verification of irradiation problem and decay problem,the calculation results of solver NES are in good agreement with the well-know point-depletion program ORIGEN2.In addition,the numerical instability of CRAM algorithm is analyzed in decay problems.The high-order IPF CRAM algorithm has better robustness to the step size in the decay problem.Then,the solver is coupled with the deterministic lattice code ALPHA to implement the burnup calculation capability of the pin cell and assembly.The calculation results by ALPHA in JAERI burnup benchmark are highly consistent with HELIOS1.11.Second,a new coupling method of burnup and transport calculation based on the dynamic mode decomposition algorithm is proposed to reduce the calling times of transport calculation.The feasibility of the model reduction of burnup data is analyzed from the singular value distribution of nuclide atom number density snapshot matrix and correlation matrix of nuclides.Combined with the online prediction model of "Fine calculation + DMD prediction",two prediction methods are proposed.One is to directly predict the nuclide atom number density by extracting the eigenvalues and modes of the nuclide number densities of few burnup steps,and directly construct the expression of nuclide atom number density;the other is to retain the cross-section and flux data in the fine calculation stage and predict the cross-section and flux data respectively,and then the burnup matrix can be reconstructed.In the prediction stage,only point depletion calculation is performed through the reconstructed burnup matrix.Numerical results show that the new burnup calculation method based on dynamic mode decomposition has high-precision prediction results in terms of nuclide atom number density,infinite multiplication factor and power distribution.Third,to reduce the memory usage of burnup calculation,the compression method of burnup library is studied in this paper.Based on the the coupling framework of the Monte Carlo transport code and the point-depletion solver,the burnup library compression platform is built,and the shortcomings of candidate nuclide selection in the traditional contribution function method are analyzed.An improved contribution function method based on target nuclide error iteration is proposed for burnup library compression.In the improved method,the contribution threshold of target nuclide is obtained by approximate repeating burnup calculation iteration,which can further reduce the number of nuclides in compressed burnup library on the premise of ensuring accuracy.At the same time,the idea of problem-dependent burnup library compression is proposed.Different compressed burnup libraries are developed for different fuels,which can further reduce the occupation of computing resources while maintaining high accuracy.Finally,in order to deal with the double heterogeneity(DH)burnup problem of dispersed particles,the detailed burnup calculation capability of the DH system is developed based on the Sanchez-MOC transport framework of transport program ALPHA,which can implement intra-particle burnup calculation.To better model the double heterogeneity of FCM fuel,ALPHA adopts a combination strategy of macroscopic discretization at pin level and microscopic discretization at particle level for burnup region.In addition,the effect of spatial discretization of the burnup calculation on macroscopic and microscopic level is analyzed.In order to accelerate the burnup calculation of the dispersed particle with poisons,a variety of poison reaction rate correction algorithms such as PPC,LLR and high-order correction method are developed.The performance of correction algorithms in different problems is analyzed.In addition,the sensitivity of FCM parameters to the equivalent radius is analyzed based on the conventional reactivity-equivalent physical transformation(RPT)method,and a fitting RPT(FRPT)method based on FCM parameters is proposed,which eliminates the search process of the equivalent radius by Monte Carlo code.To sum up,this study develops CRAM algorithm based on two forms,analyzes the numerical stability of CRAM algorithm.The traditional burnup capability of deterministic lattice code ALPHA and the advanced burnup prediction capablility based on dynamic mode decomposition algorithm are developed;For burnup library compression,an improved contribution function compression method based on target nuclide error iteration is proposed,and the problem-dependent compression idea is proposed;To deal with the DH burnup problem,a high-fidelity burnup model based on Sanchez MOC is developed and a fitting RPT model is proposed.A series of numerical results show that the research in this study not only has practical significance for the improvement of the current lattice code,but also provides a new idea for the acceleration of large-scale burnup calculation in the future.