Fatigue Monitoring and Remaining Lifetime Prognosis Using Operational Vibration Measurements

Abstract

A framework is presented for real-time monitoring of fatigue damage accumulation and prognosis of the remaining lifetime at hotspot locations of new or existing structures by combining output-only vibration measurements from a permanently installed, optimally located, sparse sensor network with the information build into high-fidelity computational mechanics models. To produce fatigue damage accumulation maps at component and/or system level, valid for the monitoring period, the framework integrates developments in (a) fatigue damage accumulation (FDA) and (b) stress time histories predictions under loading and structural modeling uncertainties based on monitoring information (Papadimitriou et al., Struct Control Health Monit 18(5):554–573, 2011). Methods and computational tools include, but are not limited to, the use of Kalman-type filters for state and stress response reconstruction based on the sensor information (Eftekhar Azam et al., Mech Syst Signal Process 60:866–886, 2015; Lourens et al., Mech Syst Signal Process 29:310–327, 2012), as well as stress cycle counting techniques, S-N curves and fatigue damage accumulation laws (Miner, Appl Mech Trans (ASME) 12(3):159–164, 1945; Palmgren, VDI-Z 68(14):339–341, 1924) to estimate fatigue from the reconstructed stress time histories at numerous hot spot locations. The FDA maps provide realistic fatigue estimates consistent with the actual operational conditions experienced by an individual structure. Combined with models of future loading events and their uncertainties, assumed or rationally estimated during the long-term monitoring period, the continuously updated FDA maps can be used to predict the remaining fatigue lifetime maps and associated uncertainties. Developments are valuable for planning cost-effective maintenance strategies, eventually reducing the life-cycle maintenance cost.