Modeling of Fatigue Behavior due to Shot Peening Conditions

Author:  M.K. Tufft, GE Aircraft Engines, Cincinnati
Source:  Conf Proc: ICSP-8 Sept. 16-20, 2002 Garmisch-Partenkirchen, Germany
Doc ID:  2002069
Year of Publication:  2002
Abstract:  
Author Marsha K. Tufft, GE Aircraft Engines, Cincinnati, OH, USA 1 Introduction The beneficial effects of shot peening have long been recognized. One of the major reasons for shot peening is to induce a beneficial surface condition (compressive stress layer and altered microstructure) that acts to retard the development and propagation of surface cracks. If surface crack formation and propagation can be suppressed, longer component operating lives can often be attained. Dorr and Wagner [1] demonstrated that shot peening was effective in retarding crack propagation of existing cracks, even when peening was applied after the development of cracks. Luetjering and Wagner [2], and others have recognized, however, that shot peening can also cause the equivalent of fatigue damage. This effect has received considerably less attention. There is increasing interest in methods to predict life capability of shot peened parts, and in the use of models that enable a designer to select a robust level of shot peening that will optimize the life benefit, minimize manufacturing costs, and avoid potential life degradation from "overpeening". This paper examines six different approaches to assessing shot peen impact on life. Four approaches focus on fatigue crack initiation life, or rather the life to failure in the absence of pre-existing cracks. One method deals with crack propagation life. The final approach attempts to correlate surface residual stress state with residual life remaining at the time of inspection. Of these six approaches, only two offer general predictive tools: one of fatigue initiation life, the other of crack propagation life. The other methods provide alternate ways for analyzing and using specific fatigue and/or residual stress data. This paper examines some of the challenges and limitations in using each of these methods. Where possible, these methods are demonstrated using data from a shot peening Design of Experiment (DOE) conducted on Rene' 88DT, a nickel-base superalloy, as documented in references [3,4,5,6]. It must be noted that life prediction methods are engineering attempts at modeling complex physical processes, and will therefore always be limited by inadequate understanding and inability to model the significant elements of physical reality. All models are wrong - by definition they are approximations at best - but some are useful. The most useful are substantiated by data covering the relevant conditions of interest. One must be cautious when trying to apply a model to conditions outside the validated set of conditions - physical reality is often complex and non-linear, and does not always cooperate with attempts at extrapolation.