PSI - Issue 39

Nabam Teyi et al. / Procedia Structural Integrity 39 (2022) 608–623 Author name / Structural Integrity Procedia 00 (2019) 000–000

612

5

reinforced, cracks form above the top due to compression failure. In the case of over reinforcement, the beam has the potential to withstand greater bending stress; however, if the top reinforcement provided is insufficient to withstand compressive stress, the beam’s top cracks. Another prominent failure in beams occurs due to flexural cracks (Fig. 5.). The word “flexure” denotes “bending”. In beams subjected to bending, cracking normally begins in the tensile zone. Flexural cracks on a beam’s sides begin at the tension face and continue up to the neutral axis at most. In general, the fractures will be evenly spaced throughout the area of the beam that is most strongly loaded, such as in the mid-span in sagging or over the supports in hogging. In the short term, the width of flexural cracks may remain narrow. Long-term continuous loading, on the other hand, may cause the crack width to widen and become more uniform across the member. In the last decade, Thatoi et al. (2012) did an overview of various methodologies used in fault diagnosis and condition monitoring of beam structures. Their review was on methodologies that included classical method; WT & FEM; AI techniques; and numerical & experimental techniques.

b) Double cracks

c) Multiple cracks

a) Single crack

Fig. 4. Cantilever beam with single crack, double cracks and multiple cracks.

Fig. 5. Simply supported beam with shear and flexural cracks.

4. Review For crack investigation on shafts and beams, this section reviews selected relevant literature from authors who have published their works utilizing data science approaches. The instruments and procedures used in crack analysis are highlighted in this section. A brief introduction of the technique is followed by a review of the selected published publications that apply that technique. Each subsection builds on this foundation. 4.1. Artificial Neural Network Many authors have used ANN tool in crack analysis. In order for computers to learn and make judgments in a humanlike manner, ANNs try to imitate the brain’s network of neurons (Fig. 6.). Programming ordinary computers to