Safety Assessment of Ultra Thin Precast Concrete Shells

• Main Author: Vanzeller, Joana Boléo
• Publication Date: 2024
• Format: Master thesis
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10362/178094
• topic: Structural Reliability; Ultra Thin Concrete Shells; Structural Safety; Failure Probability; Neural Network; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Civil

The aim of this dissertation is to analyze the safety and reliability of ultra-thin concrete shells reinforced with fibers, focusing on assessing various thicknesses, tensile strengths of concrete, and angles of wind incidence. The proposed approach involves the use of probabilistic methods due to the scarcity of specific data for the application of partial safety coefficient methods. A neural network was developed to predict the failure probability of a structure and was trained to handle multiple input variables, including shell thick- ness, concrete strength, and wind incidence angles. Visual Studio Code (VSC) was utilized as the integrated development environment, along with the Py- thon programming language, to develop the neural network. The results indicate that the analyzed shells demonstrate a satisfactory level of safety and reliability for thicknesses exceeding 200 mm, particularly when high-strength and high-performance concrete is used. Moreover, they exhibit safer behavior at a wind incidence angle of 60°, which significantly depends on the tensile strength of the concrete. This study significantly contributes to the understanding of the structural reliability of ultra-thin concrete shells, emphasizing the importance of probabilistic approaches in safety analysis, particularly for unique structures such as shells.