Mechanistic evaluation of fatigue mitigation approaches to extend the fatigue lifetime of low-pressure steam turbine blades

Abstract

The increasing use of renewable energy sources requires the energy output demand from fossil-fuel fired power plants to become increasingly sporadic. Thus, the Steam Turbines (ST) used in fossil fuel power plants are now subjected to a higher frequency of start-stop cycles and the use of ageing plant needs to be safely extended for the remaining proposed lifetime. ST blades are typically manufactured using martensitic stainless steel (MSS) due to the corrosion resistant properties and high achievable strength and fatigue resistance. This increase of loading frequency leads to premature end of predicted life compared to the original design, leading to more frequent maintenance cycles and potentially sooner than expected blade replacements. Also, due to an earlier than predicted decommissioning of power plants, the blades are too costly to replace hence an enhanced maintenance schedule was considered. The proposed maintenance process consists in grinding out of any observed fatigue short cracks in the fir tree notch root and shot peening. The process was evaluated by previous studies as potentially effective in increasing fatigue lifetimes in some MSS alloys, FV566 and FV448 namely. The focus of the current study is to evaluate the blade-to-blade variability effects of another MSS, FV520B, on shot-peening and its effects on a lifing model proposed to reduce the conservatism of fatigue lifetime prediction of ST blades. The current study focuses on FV520B, a stainless-steel alloy with a precipitation hardened martensitic microstructure. The delivered material was from ex-service blades, hence an initial analysis to verify the provided material matched expected FV520B features was conducted. Chemical and microstructural analysis confirmed the material to be FV520B. Mechanical and hardness testing was performed to analyse potential variability between blades. Results from microstructural and mechanical testing showed FV520B having scatter in fine microstructural features, mainly lath size, which correlated to hardness variations explaining scatter observed in both hardness and mechanical properties.Fatigue testing was performed to study the effect of the variation in properties and microstructural features on baseline fatigue properties and to compare FV520B with prior literature. Results showed consistency in stage II fatigue growth rates between FV520B samples from different blades, as well as consistency with FV566 performance. The threshold values from different blade of FV520B were found to be different and dependent on fine microstructural features like lath size. Crack tip opening displacement results indicated some effects from yield stress variation on fatigue crack opening effects4near threshold but in stage II the consistency between samples remained. Fracture surface observations between tested samples indicated a higher dependency of crack growth on laths at near threshold ΔK levels. The crack was observed to grow more tortuously at ΔKth levels progressively transitioning to smoother more transgranular growth as ΔK increased, the phenomena was observed consistently between blades.Short crack testing was performed on FV520B U-notched samples in 3-point bend under load control with two different surface conditions, polished and shot peened. The peening process was consistent with prior research on similar martensitic alloys from turbine blades. The residual compressive stresses after peening on FV520B were found to be consistent with the peening of FV566, indicating the peening process is equally effective on different blade alloys. Short fatigue crack growth testing was carried out at different strain ranges to evaluate the beneficial effects of peening, compared to polished unpeened samples, in terms of lifetimes and crack growth rates. Shot peening was found to increase fatigue lifetime in FV520B even at quite high local strain levels in a notch root under bending. Results also suggest FV520B was able to retain residual compressive stresses and have increased lifetimes at higher strain ranges when compared to FV566. Crack growth rate analysis in terms of ΔK showed strain range to have little effect on crack growth rate in polished samples, with growth rates being comparable to rates derived from long crack testing when characterised in terms of ΔK. The crack growth rate in peened samples showed a reduced growth rate at relatively low ΔK levels with some increase for higher ΔK compared to the unpeened case. Indicating residual compressive stresses retarded crack growth in the near-surface residual stress layer but had a decreased effect as they grow further into the depth. Paris law constants were also extrapolated from short crack tests and long crack tests.A mechanistic based lifetime prediction model originally proposed by Cunningham was applied with some modifications to FV520B. The model accounted for different phases of fatigue lifetimes, initiation, short crack growth and long crack growth. The number of cycles in each phase was either extrapolated from empirical data or iteratively calculated based on crack growth rate information gathered experimentally. The model was found to be conservative in lifetime prediction of fatigue in polished samples, but significantly overestimated lifetime for peened conditions. The conservative prediction for polished samples were attributed to the unaccounted-for arrested cracks and the assumptions on growth rate homogeneity. The overestimations in peened cases were likely due to the equally spaced initiation site assumption which strongly affects the expected coalescence phase, and given the greater number of initiation sites in the shot peened case this is likely to overestimate the onset of coalescence. Another factor to consider was the expected variations in crack growth rate through the residual compressive stress layer were not taken into account. Further data collection and analysis will be necessary to improve the model as well as highlighting the need to account for the spacing of initiation sites and the onset of coalescence as well as the varying crack growth rates in peened samples.

More information

Identifiers

ORCID for Philippa Reed: orcid.org/0000-0002-2258-0347

ORCID for Andrew Hamilton: orcid.org/0000-0003-4627-849X

Catalogue record

Export record