Stress Relaxation Of Nickel-Based Superalloy Springs

In recent years there has been a move to increase operating temperatures in steam turbines for power generation. This leads to increased efficiencies but also new challenges for material choice and design. One such challenge is the stress relaxation (reduction of reaction force when held at constant displacement) of materials being far more severe when moving from 6000C to 7000C. This work focuses on helical springs utilised in support of steam turbine casings, which are composed of nickel-based superalloy wire of radius 1.25mm coiled into a spring 28.5 mm in height. This project utilises Finite Element (FE) modelling and mechanical testing to inform development of an analytical model for stress relaxation utilising a modified Dyson creep model, which allows microstructural properties of the nickel superalloy to inform macrostructural creep and stress relaxation phenomena. Dislocation density, γ’ volume fraction and particle radius are included in the models, assuming climb/glide flow mechanism. Preliminary testing of helical samples of Nimonic 90, a medium volume fraction (approx. 20%) γ’ alloy, produced stress relaxation results faster than expected from FE and analytical analysis calibrated from uniaxial creep test data, by up to three orders of magnitude. It is proposed this is a result of the helical geometry containing many Geometrically Necessary Dislocations (GNDs), induced as a result of winding the wire into a coil. By predicting this higher dislocation density improved correlation is found between model and data. Enhancing the model by the addition of back-stress hardening via build-up of dislocations at the γ/γ’ interface improves the correlation further. Microstructural and mechanical testing of stress relaxation specimens is conducted through the use of SEM, EBSD and Vickers hardness tests on the heat treated and thermomechanically treated samples. The geometrically necessary dislocations are calculated at each time period during a stress relaxation test from misorientation maps, and the values are compared with the predicted dislocation density from modelling software, to be generally favourable. Evidence of recovery processes are presented from the EBSD data for annealing twins, dislocation density and grain boundary misorientation varying during relaxation testing. The FE model is used to propose a thermomechanical process map at high temperature. The predictions for the improvement in stress relaxation behaviour are compared with coil springs undergoing the new process and are found to be in agreement. Significant increases in the resistance to stress relaxation are predicted from the proposed process.