A probabilistic physics-of-failure model for prognostic health management of structures subject to pitting and corrosion-fatigue M. Chookah 1 , M. Nuhi, M. Modarres n Department of Mechanical Engineering, University of Maryland-College Park, Glenn L. Martin Hall, College Park, MD 20742, USA article info Article history: Received 31 December 2010 Received in revised form 30 June 2011 Accepted 15 July 2011 Available online 16 August 2011 Keywords: Pitting Fatigue crack growth Corrosion-fatigue Probabilistic modeling abstract A combined probabilistic physics-of-failure-based model for pitting and corrosion-fatigue degradation mechanisms is proposed to estimate the reliability of structures and to perform prognosis and health management. A mechanistic superposition model for corrosion-fatigue mechanism was used as a benchmark model to propose the simple model. The proposed model describes the degradation of the structures as a function of physical and critical environmental stresses, such as amplitude and frequency of mechanical loads (for example caused by the internal piping pressure) and the concentration of corrosive chemical agents. The parameters of the proposed model are represented by the probability density functions and estimated through a Bayesian approach based on the data taken from the experiments performed as part of this research. For demonstrating applications, the proposed model provides prognostic information about the reliability of aging of structures and is helpful in developing inspection and replacement strategies. & 2011 Elsevier Ltd. All rights reserved. 1. Background and motivation Structures such as pipes and steam generator tubes in power plants and oil pipelines subject to changing loads and corrosive agents are susceptible to degradation over the span of their service lives. Corrosion (especially chloride and H 2 S corrosion) is one of the most common degradation mechanisms, but other critical mechanisms, such as stress corrosion cracking (SCC) and creep, are also important. The rate of degradation is influenced by many factors, such as piping material, characteristics of the fluid, process conditions, geometry and location. Based on these factors, a best estimate for the service life (reliability) can be calculated. This estimate serves as a target guideline for prognosis and health management (PHM), preventive maintenance and replacement practices. After a long period of service, however, this estimate requires re-evaluation due to new evidence observed from mon- itoring conditions of the structure. A number of deterministic models have been proposed to estimate the reliability of pipes. Among these models is the ASME B31G code [1], which is the most widely used method for the assessment of pipe corrosion [2]. However, these models are highly conservative and unable to estimate the true life and health of the structure. In addition to the limitations embedded in these deterministic models are the problems associated with inspection techniques and tools that may be inadequate and inaccurate. The research summarized in this paper proposes and compares a simple empirical probabilistic model developed using informa- tion from well-established mechanistic models for pitting-corro- sion and corrosion-fatigue. The parameters of the proposed empirical model are estimated through a formal Bayesian infer- ence from data obtained from pitting and corrosion-fatigue experimental tests. Uncertainties about the structure of the model itself and parameters of the model are characterized. The pro- posed model can capture wide ranges of structural materials [2–10] and accounts for variability in material and size. The existing probabilistic models sufficiently address the corrosion and fatigue mechanisms individually, but are inadequate to capture mechanisms that synergistically interact such as corro- sion-fatigue. Further, the mechanistic models have far too many parameters and are too complex for routine reliability and life estimation practices, especially for routine use in the field. The proposed model alleviates these shortcomings. Admitting the fact that capturing all degradation mechanisms would be a challenging task, the new model addresses two of the most important mechanisms that synergistically interact: pitting- corrosion and corrosion-enhanced fatigue crack growth. This paper first reviews the literature in pitting and corrosion-fatigue and then prepares a simulation technique to inform the empirical model development process. It describes a procedure for estimat- ing the parameters of the proposed model. Finally, it discusses the Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/ress Reliability Engineering and System Safety 0951-8320/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ress.2011.07.007 n Corresponding author. E-mail address: modarres@umd.edu (M. Modarres). 1 Current address: Emirates Nuclear Energy Corporation, Abu Dhabi, United Arab Emirates. Reliability Engineering and System Safety 96 (2011) 1601–1610