An efficient method for probabilistic finite element analysis with application to contact assessment of CANDU fuel channels

•Probabilistic finite element analysis of CANDU fuel channels.•Computationally efficient method for probabilistic finite element analysis.•Application to 1D and 3D probabilistic contact assessment of CANDU fuel channels.•Contact assessment based on different pressure tube orientations. In CANDU reactors, irradiation induced creep and growth of the fuel channel (FC) assembly leads to Pressure Tube (PT) sag and to a reduction in the gap between the PT and the Calandria Tube (CT) (PT–CT gap). Excessive PT sag can lead to contact between the PT and the CT. Uncertainties in the parameters that influence PT–CT contact require probabilistic assessment methods to assess risk of PT-CT contact and demonstrate adherence to CSA Standard N285.8 provisions. Improved models in the prediction of PT–CT contact require computationally expensive 3D finite element analysis (3D FEA), a tool that is prohibitively difficult to use in probabilistic assessments. Current probabilistic analyses rely on 1D FEA models for the prediction of PT-CT contact estimates that are generally less accurate. Therefore, an improved and computationally more efficient approach for conducting probabilistic finite element analyses (PFEA) of CANDU FCs is highly desirable. In this paper, such a PFEA approach is proposed by coupling the multiplicative dimensional reduction method (M-DRM) and the polynomial chaos expansion (PCE) method. The analysis that follows studies the PT–CT gap predictions using deterministic 3D FEA by considering two different PT configurations present within the population of the FCs in a CANDU reactor core. The PFEA is then carried out using 1D and 3D FEA by employing the ABAQUS software suite. The accuracy of the probabilistic analysis results using the proposed approach is initially checked against Monte Carlo Simulation (MCS) results using 1D FEA. Following a good agreement, the proposed approach was applied for PFEA based on 3D FEA. The results indicate that the two different PT orientations significantly influence the probabilistic contact results. They also show that the probabilistic results based on 1D FEA, which are currently used by the nuclear industry, are not reliable. Moreover, it is also shown that a sensitivity analysis can be performed without any additional effort. The low computational cost, predictive capability and the generality of the proposed method are important attributes for adopting it in the future in probabilistic assessments of an entire reactor cor