Given the strong reliance on long-term datasets containing multiple heats of the same material (i.e., different bathes), there is a pressing need to obtain reliable predictions of creep properties that are representative of all such batches. This paper presents a new estimation framework for the theta methodology which enables a distribution of creep properties to be predicted which should be representative of any batch of material. When this new statistical approach was applied to a number of batches of 1Cr-1Mo-0.25V steel, it was found that allowing the parameters describing the shape of the creep curve to be random variables that follow a joint normal distribution enabled an accurate representation of creep scatter to be made at various stresses for all batches. For example, the extrapolations to 155 MPa suggested that the times to failure will follow a non-normal distribution with a median time of 6.65E+08 seconds over all possible batches and a 95 pct confidence limit of 4.98E+6 to 1.29E+11 seconds.
Metallurgical and Materials Transactions A, 2013, Vol. 44A(5), (2013), Pages 2431-2444. doi: 10.1007/s11661-012-1555-0.