Exploração, análise comparativa e modelagem da deformação permanente de materiais granulares da microregião de Mossoró/RN

• Ano de defesa: 2021
• Autor(a) principal: Cabral, Wendel Silva
• Orientador(a): Barroso, Suelly Helena de Araújo
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Deformação permanente; Ensaio triaxial; Redes neurais artificiais; Solo; Pavimentação
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/56860

Due to the new requirements of mechanical analysis of Brazilian pavements motivated by the development of new sizing methods, becames important to promote research for knowledge and control of the occurrence of Permanent Deformation (PD) in soils. The analysis of the behavior of regional materials used in the layers of the pavement, as for irreversible deformations, provides essential information for mechanistic-empirical dimensioning, as well as optimizes resources and meets the structural needs of the pavement. From this, the present thesis presents as general objective to analyze the PD of soils used in pavements in Mossoró microregion for the development of models to be used in methods of mechanistic-empirical sizing of pavements in the light of faster tests and the artificial neural networks (RNA) technique. Thus, the experimental program was developed in three stages: (i) step 1 consisted of research and data acquisition in the study region; definition of the sampling effort; selection and collection of materials; performance of characterization and mechanical tests on the selected samples, as well as preliminary evaluation of the PD test method to be used using the Brazilian (DNIT 179/2018) and European (CEN 13286–7/2004) standards; (ii) step 2 covered a pre-processing of the data obtained in the tests and the development of the dp and MR forecastmodels using RNA and software such as LAB Fit; and (iii) step 3 understood the analysis of empirical (through the CBR) and mechanistic-empirical (by the medina and cap 3D-D) computational programs. The results showed that the multi-stage triaxial assay, based on the European standard, presented operational advantages and was recommended as a method to obtain PD from the samples. In addition, the models developed using the RNA technique were considered satisfactory to predict soil PD, as well as the equations generated to predict the resilience module. It was considered that the pavements dimensioned by the mechanistic-empirical methods were able to predict the performance of the structures considering the occurrences and influences of stresses in the appearance of ruptures by permanent deformation and cracked area. It was noticed, with the exception of the pavement with low traffic volume, the undersizing of the structures by the DNIT method, which was proven in the early deterioration observed in the highways of the region. Therefore, it is recommended that the European PD testing technique and models developed based on artificial intelligence can be updated and tested to scale pavements in the Mossoró/RN microregion.