Evaluation of the covariance matrix of the U-235 neutron induced prompt fission spectrum

The new Monte Carlo (MC) method has been applied to produce the covariance matrix of the prompt fission neutron spectra (PFNS) for 235U neutron induced fission. Parameters of the model describing the PFNS were sampled randomly within the given uncertainty intervals, assuming normal distribution. The covariance matrix for the Watt spectrum generated by the MC method was validated by comparison with the matrix derived analytically. Analytical derivation is appropriate in case of linear dependence of the spectra on the underlying parameters, whereas in the MC approach the extension to more complex models is straightforward: PFNS covariance matrices were obtained both for the Watt model and for the phenomenological parameterization of the fission neutron spectra proposed by Kornilov et al. Derived covariance data were used in the cross section sensitivity and uncertainty analyses of several KRITZ UO2 and MOX critical configurations performed in the scope of the OECD international benchmark exercise. Two different approaches to calculate the sensitivity coefficients of keff relative to the fission spectra were used as a means to check and assure the mathematical correctness of the matrices.