Banca de QUALIFICAÇÃO: RAFAEL AREDES COUTO

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : RAFAEL AREDES COUTO
DATE: 22/09/2023
TIME: 09:00
LOCAL: https://conferenciaweb.rnp.br/sala/peter-ludvig
TITLE:

Proposal for computational approach for service life prediction of reinforced concrete structures exposed to carbonation through Machine Learning

KEY WORDS:

Artificial Intelligence; Machine Learning; Durability of Structures; Reinforced Concrete; Carbonation

PAGES: 65
BIG AREA: Engenharias
AREA: Engenharia Civil
SUBÁREA: Construção Civil
SPECIALTY: Materiais e Componentes de Construção
SUMMARY:

Given the increasing concentration of carbon dioxide (CO2) in the atmosphere and the need for sustainable construction, the development of a tool capable of estimating the service life of reinforced concrete (RC) structures subject to carbonation is relevant. The cement and steel industries are responsible for a significant amount of global CO2 emissions. Therefore, studies on the durability and prediction of the service life of RC structures are important for quantifying and optimizing the use of resources in their construction. Carbonation is one of the main aging and deterioration mechanisms of RC structures, responsible for limiting their service life. This mechanism is influenced by various factors related to the material (compressive strength of concrete, water-to-binder ratio, type of cement/binder) and the environment (CO2 concentration in the atmosphere, relative humidity, exposure to rain, and ambient temperature). Due to the high number of variables, the uncertainties associated with the process are significant, making it challenging to quantify the service life of a carbonated RC structure. Therefore, artificial intelligence (AI), through machine learning (ML), presents itself as a probable tool for developing algorithms capable of predicting service life with greater accuracy and precision. ML is a subfield of AI that teaches machines to handle data more efficiently. Thus, the present research aims to develop an ML-based algorithm capable of predicting the service life of RC structures subject to carbonation. For algorithm development, the Python language and the supervised ML model, Random Forest, will be used. Data for model development will be obtained in two stages: i. synthetically, through Monte Carlo Simulation, using probabilistic algorithms developed in Couto (2017), and ii. through a literature review of data on the subject, resulting in a robust database. The SHAPley Additive exPlanations (SHAP) technique will be used for model interpretability, determining the impact of each variable in the process. This research aims to provide a tool capable of predicting the service life of carbonated reinforced concrete structures with reduced uncertainties. 

COMMITTEE MEMBERS:
Presidente - PETER LUDVIG
Interno - PAULO HENRIQUE RIBEIRO BORGES
Interna - FLAVIA SPITALE JACQUES POGGIALI
Externo ao Programa - DANIEL HASAN DALIP
Externa à Instituição - SOFIA MARIA CARRATO DINIZ - UFMG
Externa à Instituição - MICHELE AMARAL BRANDÃO - IFMG