Metodologia integrada e robusta de projeto de estruturas em betão reforçado com fibras de aço assistida por modelos de análise não linear baseados no método dos elementos finitos e de inteligência artificial

Besides the task corresponding to the management, this proposal has more 5 tasks, each one with a main objective and sub-objectives. However, the main objective of this proposal is the development of integrated research for giving to FEM-based users the possibility of designing with confidence HSFRC structures to really explore the potential of fibre reinforcement for a more sustainable built environment. Therefore, the main milestone is the development of the AIpFRC web-platform containing modules with the following relevant functionalities: 1) for certain SFRC mix composition, using an AI framework that integrates the COFIT inverse analysis approach, determine the most accurate data to define the mechanical properties of this SFRC, with special focus on the constitutive laws that characterize the fracture process (Fig.2). Besides the database for the SFRC mix compositions and their properties at fresh and hardened states (DB_Mat), this module will also include files with the experimental results from the 3PNBBT, RPT and FIP. A computational tool will be developed for automatically feed this database by ethically extract reliable information from reputed journals of public access using web scraping techniques, text mining procedures, and target keywords. 2) perform global sensitivity analysis to assess the significance of each parameter on the outputs (predictive performance), by simultaneously varying all input parameters. This analysis allows the assessment of the dependence of the model's output on the selected values of the input parameters (available in DB_Mat). To provide a sufficiently large and random sample from a general statistical population while maintaining the overall statistical characteristics, sampling methods, like Monte Carlo, Latin Hypercube, Halton, Sobol, will be adopted. The use of genetic algorithms-based optimization procedure will also be explored to diminish the number of simulations that provide best fitting. Eventual correlations between the parameters defining the constitutive laws of SFRC fracture process and the input parameters will be assessed for a faster and simple definition of the constitutive laws of a SFRC to be used by a FEM-based model. 3) Implement the concept of fibre orientation tensor in the PDMFSCM available in FEMIX, to model the fibre orientation and distribution effects without the necessity of simulating explicitly the fibres in the mesh, since this last option is too time consuming in the analysis of real scale structures [35]. 4) Include in the AIpFRC a database (DB_Str) with relevant information from experimental tests with prototypes of HSFRC, namely: displacements in critical sections, strains in the constituent materials, strain fields and crack width measurements from DIC, photos and videos for the crack patterns and failures models. This information will be used in task 5 for determining, assisted by AI, the best values of the constitutive models on the simulation of HSFRC structures. 5) Design methodology of HSFRC structures using GRM by accounting for the aleatory uncertainties associated to the material properties, model robustness and failure mode brittleness of the structure, and target reliability index (Figs.4 and 5). 6) Develop of a postprocessing tool with a user-friendly framework for a comprehensive interpretation of the results in the FEM-based design of a HSFRC structure.