PSI - Issue 41

Mohamed Amine Belyamna et al. / Procedia Structural Integrity 41 (2022) 372–383 Mohamed Amine Belyamna et al. /Structural Integrity Procedia 00 (2022) 000–000

374

3

Various papers in the literature addressed the probabilistic failure analysis of components subjected to SCC. Zhang et al. (1997) carried out experimental investigations to determine the time to crack initiation and crack propagation velocity for IG-SCC in sensitized type AISI304 stainless steel in dilute sulfate solutions. Ting (1999) analyzed the crack growth due to IG-SCC in stainless steel piping of BWR plants. Probabilistic failure analysis of nuclear piping of BWR plant was carried out by You and Wu (2002). Failure probabilities of a piping component subjected to SCC was computed using MCS technique by (Guedri et al., 2012; Guedri,2013a; Guedri, 2013b; Boutelidja et al., 2019). In reliability analysis of structures, the MCS method is particularly applicable when an analytical solution isn't attainable and the failure domain can't be expressed or approximated by an analytical form. The traditional MCS method is time consuming for the low possibilities of the pipeline failure but the ANN is preferable for the complex nonlinear situation (Wen, 2019). Neural network learning models and the cumulative results of expertise have found their way into practical applications in many fields. Papadrakakis and Lagaros (2002) examine the application of ANN to reliability-based structural optimization of large scale structural systems. Cardoso et al. (2008) examine a methodology for computing the probability of structural failure by combining ANN and MCS. In recent years, there has been a growing interest in using ANN in engineering applications such as oil and gas production (Pourahmadi et al., 2022), safety, and reliability, to evaluate the nonlinear relationship between a variety of characteristic variables and output variables in the process of reliability evaluation. (Solanki et al., 2020) describes ANN-based response surface methodology developed for reliability assessment of passive systems. Failure analysis of corroded high-strength pipeline subject to hydrogen damage based on finite element method and neural network is given by (Zhang, 2022). In the present study, an integrated reliability method is proposed to conduct the IG-SCC line pipe’s reliability analysis considering the initiation in multiple sites and residual stress effects. The proposed method integrates several numerical approaches, including the probabilistic fracture mechanics (PFM) model, sensitivity analysis, MCS, and the ANN method. The sensitivity analysis approach is used to quantify the contribution of each parameter to the pipe’s reliability and simplify the ANN structure. MCS is implemented to generated reliability data and input parameters for ANN training, and uncertainties in geometric features, material properties and operating conditions The piping reliability model was developed based on concepts of PFM. The calculation procedure for estimating the probability of failure combines several random variables, such as the initial crack size distribution, crack detection probability, crack growth, and deterministic stress history. The calculation begins with an initial size of the defects in the form of cracks at a given location. These growing cracks are detected with some probability during pre-and in-service inspections. Cracks that escape detection and repair can develop as a result of growth under conditions such as SCC and the spread of fatigue cracks. The critical crack size for a leak can be defined using an appropriate criterion. The probability of failure at one location of the analyzed pipe is equal to the probability that a crack will reach the corresponding critical size within the specified time. SCC can be intergranular or transgranular in nature depending on the material, level of stress, and environment. The methodology recommended in PRAISE for modeling IG-SCC in stainless steel pipe is presented in this section. PRAISE separates the overall time to pipe leak into three steps (Harris, 1992): a) Time to initiate a very small crack, b) Time spent growing the small cracks at an initiation velocity v 1 , c) Time spent growing larger cracks at fracture mechanics velocity v 2 to become through—wall cracks. 2.1. Initiation and Growth of Cracks The time to crack initiation under static load conditions has been found to be a function of the damage parameter IG SCC D  as presented in equation (2). Therefore, the time to crack initiation t I for a given IG SCC D  is taken to be log normally distributed. The mean and standard deviation of log (t I ) are given in (Harris, 1992) by: are considered in the simulation procedure. 2. Original Probabilistic IG-SCC Model