PSI - Issue 64

Abdalla Elhadi Alhashmi et al. / Procedia Structural Integrity 64 (2024) 1990–1996 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

1992

3

Fig. 1. Schematic of the proposed framework of analysis.

An automated MATLAB-Python code is developed to apply and generate the input parameters required for the concrete. Stress-strain curves for tension and compression, as well as the corresponding damage curves, are developed in MATLAB. These curves are generated for specified nodes, which are part of a random field mesh. The rule of thumb is that the random field mesh is always coarser or equal to the defined FE mesh, depending on the specified correlation length ( Li and Der Kiureghian, 1993 ). The predefined field holds the material properties at certain coordinates within a geometry, referred to as an analytical field, which does not rely on a FE mesh. ABAQUS uses its built-in functionality for interpolation to determine material properties of a given FE element based on the node’s values and the element's centroid (ABAQUS CAE, 2021). Subsequentially, the image of the deficient structural member is processed, scaled, and incorporated into the RFE model. The image processing task is carried out using the MATLAB image processing toolbox (The MathWorks Inc., 2022), while the transfer and scaling of the image is performed utilizing DIGISIM software (Meng et al., 2020). The main role of DIGISIM is to convert the processed image (i.e., binary image) to geometric data. This is because FE software is designed to work with geometric data rather than direct image information. Finally, the safety of the structure is assessed by constructing a resistance model and a load model to determine the reliability index, β . The resistance model is determined by generating resistance histograms from N numbers of RFE models to obtain the mean, coefficient of variation (COV), and type of distribution. The load model is determined from load statistics, which can be found in Bartlett et al. (2003). The reliability index is determined using Monte Carlo simulation for a defined limit state function. The limit state function, denoted as ( ) , corresponds to the difference between the resistance model ( ) multiplied by a model factor, and the load model, denoted as ( ) , as expressed in Equation 2: ( ) = ∗ ( ) − ( ) < 0 (2)