metodologia para avaliação da incerteza associada à curva de transição dúctil-frágil

Abstract

In this work, a methodology was defined and validated to calculate the uncertainty associated with the ductile–brittle transition curve (DBTC) obtained using the symmetric hyperbolic tangent (HT) model. The methodology is based on the guidelines of JCGM 100 and JCGM 101, as well as on the requirements of ISO 148-1. It combines the approaches of the Guide to the Expression of Uncertainty in Measurement (GUM) and the parametric bootstrap method. Charpy V-notched specimens with dimensions of 5 mm × 10 mm × 55 mm were manufactured from six low-alloy carbon steels used in the production of tensile wires for flexible pipes. Charpy impact tests were performed over a temperature range from −196 °C to 200 °C in accordance with ISO 148-1 and ASTM E23. The proposed methodology accounts for the variability of absorbed energy values, the uncertainty associated with the HT model fitting process, and the calibration uncertainty of the Charpy impact machine. Relevant statistical aspects were discussed, including parameter distributions, measurand normality, and residual homoscedasticity, as well as the computational cost associated with each method. The results demonstrated that it was possible to estimate and represent uncertainty bands around the DBTC of the evaluated materials, facilitating their comparison. A systematic comparison between the GUM, Monte Carlo, and parametric bootstrap methods showed the high computational cost of the Monte Carlo approach and the superior statistical representativeness and computational efficiency of the bootstrap method. The methodology was experimentally validated using tests at discrete temperatures, for which the uncertainty bars of the validation tests overlapped with the DBTC uncertainty bands obtained by the bootstrap method. The maximum difference observed between the expanded uncertainties associated with the validation tests and those estimated for the DBTC was −0.93 J.