Nuclear Data Sensitivity And Uncertainty Analysis In The Burnup Calculation Based On RMC Code

As the development of computational tools and analysis methods,the uncertainty of results caused by methods becomes smaller and smaller,and nuclear data turns into the main uncertainty source.There are two main kinds of methods for uncertainty analysis: the first-order uncertainty qualification method and the stochastic sampling method.Sensitivity coefficients is the base of the first-order uncertainty qualification method.However,solving sensitivity coefficients in Monte Carlo codes would face the challenge of huge memory consumption.On the other hand,applying the stochastic method in Monte Carlo codes would face the challenge of severe time consumption.The first-order uncertainty qualification method is suitable for transport calculation due to its high efficiency while the stochastic sampling method is suitable for burnup calculation due to its generality.In order to solve the problem of huge memory consumption in sensitivity analysis and the problem of serious time consumption of the stochastic sampling method,this paper is focus on research of advanced methods and algorithms,as well as the development of the capability of nuclear data sensitivity and uncertainty analysis in both the transport and burnup calculation in the Reactor Monte Carlo code RMC.The main content includes:(1)Sensitivity analysis of keff to nuclear data in the transport calculation.After comparing the existing adjoint-flux-based sensitivity analysis methods,this paper proposed several memory reduction schemes,including data decomposition strategy,fission-based CLUTCH method,adjoint superhistory method.(2)Sensitivity uncertainty analysis of generalized response function,including linear response functions as well as but also bilinear response f unctions,to nuclear data in the transport calculation.Firstly,this paper proves the theoretical principle of the collision history-based method and demonstrates the feasibility of applying this method in the Ray-Tracking transport process.After that,in order to reduce the memory consumption,the collision superhistory-based method has been put forward.(3)Uncertainty analysis of computational results caused by nuclear data uncertainty in the transport calculation and the burnup calculation.In order to solve the problem of overestimation of nuclear data uncertainty due to the underestimation of Monte Carlo variance,this paper proposes the fast RMC method which is based on the real variance estimation.To reduce the memory consumption of storing perturbed nuclear data library,the user-friendly integrated fast stochastic sampling flow which puts the nuclear perturbation module inside the RMC code has been put forward.This integrated fast stochastic sampling flow can increase computational efficiency,a utomatically process the results and perform uncertainty analysis of different response functions due to nuclear data uncertainty in both the transport calculation and the burnup calculation.In addition,the capability of uncertainty analysis of keff due to nuclear data in the transport calculation has been developed based on the first-order uncertainty qualification method.(4)Finally,this paper performs calculation and analysis on two benchmarks from the benchmarks for uncertainty analysis in modelling(Phase II),i.e.,the BWR full-core transport case and the PWR pin cell burnup case and shows the capability of nuclear data sensitivity analysis and uncertainty analysis in RMC.This paper is focus on not only the innovation of methodology of sensitivity and uncertainty analysis in the Monte Carlo transport and burnup calculations,but also making an advanced capability of sensitivity and uncertainty analysis in RMC code through the breakthrough of algorithm.