The API 579 MPC Project Omega Multiaxial Creep Ductility Ratio

Abstract The fitness for service standard API 579 provides a very prescriptive parametric model for the assessment of creep. Using this API 579 model it is possible to perform creep simulations in finite element software. The API 579 model considers the deviatoric stress magnitude, the relative hydrostatic stress, and the material temperature. Finite element software is then able to simulate creep strain flow and the resulting load redistribution. The parametric degree of freedom (which is interpreted as a voiding/rupture unity failure criteria) enables the API 579 model to accumulate creep damage and to consider ongoing creep damage and creep rate interactions. In the API 579 model the creep damage rate is taken to be linearly related to the creep strain rate, where the scale factor is the state dependent Ωm coefficient (from which the model gets its name; MPC Project Omega). Stress magnitude sensitivity effects are incorporated into the Ωm parameter through the omega function and the (instantaneous) Norton-Bailey coefficient, while stress multiaxiality effects are incorporated into a power-exponent. In effect the low damage API 579 multiaxial creep ductility ratio is ∼ max(Ω−n, 3)−δ | n:= Norton-Bailey coefficient, δ:= multiaxiality parameter, Ω:= Omega damage function. In order to better fit the MPC Project Omega model to experimental creep data, API 579 has expanded the original minimal coefficient Project Omega model to include additional (non-physical) fit coefficients (log-space polynomials of up to 3rd order). One set of fit coefficients defines the creep strain rates sensitivity to stress, while another defines the creep damage rates sensitivity to stress. Including so many non-physically motivated fit terms in the API model can improve the fit of the material model, but also significantly relax the bounding under extrapolation. As there is very little (and for some materials listed in API 579, apparently zero) experimental data that considers creep in scenarios other than uniaxial loading, it is assumed API 579 is extrapolating from that single point to all other stress states. It would appear that, for some materials in particular, this has led to wholly unreasonable stress multiaxiality sensitivity. This paper aims to provide a thorough review of the MPC Project Omega (API 579) stress multiaxiality sensitivity. First the paper outlines the numerical issues with the way the model is constructed and the resulting difficulties that appear to have arisen when the uniaxial experimental data fitting has been performed. Second the paper considers alternative models for capturing the stress multiaxiality effects, in particular the Spindler model adopted by R5. Lastly the paper discusses some implementation issues associated with the limitations of common commercial finite element software packages.