The development of advanced power plants requires alloys to operate at elevated temperature and pressure for an extended period of time. It is critical to consider creep during the design process to avoid catastrophic failure. Creep rupture data are often not available for desired operating conditions. Accurate extrapolation of creep life is necessary. One of the earliest and most widely used life prediction model is the classic Larson-Miller Parametric (LM) model. Over time numerous time-temperature parametric (TTP) models have been proposed such as Manson-Haferd, Orr-Sherby-Dorn, Manson-Succop, Graham-Walles, Goldhoff-Sherby parametric models. Non-TTP models such as the Wilshire equation is available. The prediction models vary in mathematical form, and number of material constants but shares a common calibration approach. Each model is calibrated against data for every available isotherm. A recently proposed model calibration approach is the parametric numerical isothermal datum (P-NID) method that can be applied to an existing model for improved long-term extrapolation. The P-NID approach is different than the traditional approach as the data are transferred to a datum temperature followed by model calibration against the transferred data at the datum temperature. The calibrated model is then transferred back to the original temperatures.
In this study, the P-NID method is applied to the LM model to perform extrapolation for Inconel 617 alloy. Creep rupture data for five isotherms ranging from 800 to 1000°C and stress levels from 9MPa to 170 MPa are used. A detail step by step procedure is provided for the application of the P-NID method to calibrate the LM model (LM-NID). The extrapolation performance of the classic LM and LM-NID models are compared. Normalized Mean Squared Error (NMSE) is used to analyze prediction accuracy. It is observed that the LM-NID model provides a realistic inflection free prediction compared to the LM model. A 10% data-cull from the lowest stress data is performed to assess the reliability of extrapolation. Based on the comparison a recommendation is provided.