Prediction of creep behavior of Zr-Nb alloy under dual-phase condition using data driven models

Abstract

Pressure tubes (PTs) play an important role in the safe and efficient operation of Nuclear Power Plants (NPPs) as they contain the fuel bundles and provide structural integrity. Creep has been identified as one of the main degradation mechanisms of PTs, which are made widely of Zr-Nb alloys. The creep curve of a material gives an insight into the nature of its creep behavior. In the present investigation, accelerated creep experiments were conducted on Zr-2.5Nb PT alloy in the stress and temperature range of 22–58 MPa and 600–850 °C, respectively. Two data-driven models, namely Radial Basis Function Neural Network (RBFNN) and Least Square Fit (LSF) were developed to simulate the non-linearity of the creep curves. Applied stress, test temperature, and time to failure were taken as the input parameters for the models. It was observed that although the LSF could predict the primary creep zone, it failed to predict the transition between the secondary and tertiary creep region. However, the creep curves predicted by the RBFNN model were in close agreement with the experimental results, having a confidence level of ≈ 0.99. Two separate sets of creep experiments were also done later to verify the accuracy of the proposed models. The results from the study established the ability of the RBFNN technique to simulate the complex behavior of the creep curves.