CALINS: A New ASNR Software for Nuclear Data Sensitivity Analysis, Uncertainty Propagation, and Assimilation of Integral Experiments

Sensitivity and uncertainty (S/U) analyses play a key role in criticality safety by enabling the quantification of nuclear data (ND)—induced effects on integral responses. In this context, flexible and transparent computational tools are essential to support advanced analyses and methodological developments. CALculations and Investigations on Nuclear data uncertainties and Sensitivities (CALINS) is a new in-house Autorité de sûreté nucléaire et de radioprotection (ASNR) open-source software, written in Python, designed to perform S/U analyses in support of nuclear criticality safety assessments. CALINS also implements the generalized linear least squares (GLLS) method for the estimation of bias and uncertainties. For greater flexibility, uncertainty propagation and the GLLS method can be performed using different ND covariances such as the one available in the standard ND libraries or in the SCALE code system or the Covariance Matrices from Cadarache (COMAC) covariance matrices developed by the French Alternative Energies and Atomic Energy Commission (CEA). Additionally, CALINS allows the identification of representative benchmarks of a specific application case from the sensitivity coefficients of the MORET6 code validation database and the Database from the 2021 International Handbook of Evaluated Criticality Safety Benchmark Experiments (DICE 2021). To validate the computational methods implemented in CALINS, a comparison was performed with the Tool for Sensitivity/Uncertainty analysis of Response Functionals using Experimental Results (TSURFER) (the uncertainty propagation tool from the SCALE 6.3.1 code system) using 46 International Criticality Safety Benchmark Evaluation Project benchmark categories. These benchmarks were selected to cover a wide range of energy spectra, fissile materials, chemical forms, and structural materials. The data assimilation was carried out using sensitivity coefficients and keff values from DICE 2021. The validation included calculation of similarity coefficients Ck, prior uncertainty using the propagation matrix product formula also called “sandwich formula,” and posterior bias and uncertainty based on the GLLS method. The results demonstrate perfect agreement between CALINS and TSURFER, with differences lying within the numerical precision of the TSURFER outputs. This paper outlines the main features of the CALINS software and its validation process. ASNR’s motivations for developing a new Python-based software dedicated to uncertainties and sensitivities are presented. CALINS is under BSD-3-Clause license to allow anyone to freely use, modify, and distribute the code. CALINS is accessible through this link: https://asnr-fr.github.io/CALINS.