Predictive Limits of Curve Fitting Creep Models Under Non-Stationary Operating Conditions in High-Temperature Metallic Alloys

The characterization of high-temperature plastic deformation behavior in centrifugally cast alloys has historically been based on curve fitting of experimentally obtained creep data. Although this approach provides acceptable results within the range of experimental conditions used for curve fitting, performance predictions under conditions outside conventional tests—typically derived from constant temperature and stress—remain insufficiently accurate for petrochemical engineering applications. In this study, constant-load creep tests were performed across a stress range of 5–33 MPa at 950–1100 °C. The minimum creep rate was calculated from the creep curves and used to fit Norton-law parameters by log–log regression. A global Norton-law fit was first carried out for each temperature, and an additional segmented fit was then performed by separating the low- and high-stress domains. The results reveal a systematic variation in the apparent Norton stress exponent (n) with the stress range considered. This variation suggests that the creep response changes from one apparent stress domain to another, contradicting the assumption of a single-valued Norton exponent inherent in standard curve-fitting procedures. Although the experimental database was obtained from conventional constant-load and constant-temperature creep tests, the results are discussed in terms of their implications for creep modeling under non-stationary operating conditions, where local stress and temperature fields may evolve during service. Parameters fitted over a broad stress range may produce systematic local errors when applied to stress domains with different apparent sensitivities. Furthermore, it is well known that classical steady-state creep models may be insufficient when their fitted parameters are transferred to non-stationary loading conditions, where thermal transients during start-up and shutdown generate differential thermal strains and high local stress levels. Geometric constraints, combined with these peak stresses, may lead to values exceeding the yield strength, a condition under which classical methods fail to adequately describe material behavior and stress relaxation mechanisms. This manuscript directly addresses this limitation through the analysis of three heats of the same centrifugally cast alloy.