Integrated diagnostic/prognostic tools for small cracks in structures

Abstract An integrated system for diagnosis of the ‘health’ of a structural component subjected to high-cycle fatigue (HCF) consists of sets of embedded or emplaced sensors at various locations, extracting information related to the generation of material defects, the presence of crack-like discontinuities and their progression and changes in system dynamics that may relate to this progression. Conceptually, signals from these sensors are fed into a processing environment that can project deleterious conditions related to the onset of loss of function or propagation of cracks to critical dimensions. Since the idea is to monitor the gradual changes of component performance and various local related indices before catastrophic failure to enable the operator to respond with a maintenance hold, it is essential to couple the diagnostics with prognostic capability; this facilitates a prediction of how much time remains within the window of viable servicing or repair. In the HCF regime, the dominant fraction of total fatigue life may be spent at crack lengths of the order of 20-500 um. The detection of longer cracks near the end of component life is critical since component failure may lead to failure of the overall structure. This necessitates the identification of (a) algorithms for identifying component ‘hot spots’ where failure is likely to occur, (b) development of appropriate crack growth laws for cracks of different length scales, ranging from the order of grain size to the order of component dimensions, including consideration of contacting components (fretting fatigue) and environmental effects and (c) development of algorithms for identifying the progression of component degradation on the basis of multiple-sensor inputs at different time and length scales, providing feedback to support cause for maintenance shutdown. This paper discusses each of these issues.