Banca de DEFESA: JOAO LUCAS AUSTREGESILO NEPOMUCENO

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : JOAO LUCAS AUSTREGESILO NEPOMUCENO
DATE: 23/08/2023
TIME: 14:00
LOCAL: Pós-Graduação Engenharia Civil
TITLE:

Computational Investigation of the Nepo Factor for Failure Pressure Prediction of Curved Pipelines with Internal Corrosion Defects

KEY WORDS:

Curved pipes. Corrosion defect. Failure pressure. Finite Element Method. Lorenz Factor. Nepo Factor

PAGES: 64
BIG AREA: Engenharias
AREA: Engenharia Civil
SUMMARY:

In this research, the variation of failure pressures of curved and straight pipelines with corrosion defects was investigated, considering their geometric parameters. The application of curved pipes is extensive, mainly in logistic solutions when facing physical obstacles. Some studies have already been conducted comparing failure pressure results obtained through the Finite Element Method (FEM) with those obtained by the Lorenz Coefficient (LC), and the findings indicated potential improvements to the factor. Modeling and analysis were performed using FEM through the PIPEFLAW software, developed by the PadMec group at UFPE, which was modified to automate the pipe bending process, and the modeling and analysis of multiple cases in sequence. Parametric studies were conducted, considering the external diameter of the cross-section, length, depth and width of the defect, aiming to find an interrelation with the Target Factor (TF), which indicates the ratio between failure pressures of curved and straight pipes. Initially, 40 discretized cases were modeled, 10 for each evaluated variable, providing initial indications of relevant parameters. Additional 120 random cases were generated for linear and logarithmic regressions, which indicated that the defect depth (d) was the most relevant parameter for predicting the TF for Intrados cases. Subsequently, 100 discretized cases were generated, varying the defect depth and pipe relative radius, and 5 new proposals, called Nepo Factor (NF), were made to improve the prediction of TF. The last proposal showed a Root Mean Square Error (RMSE) of 34.5e-3 and Mean Absolute Percentage Error (MAPE) of 1.0%, contrasting with the results of the LC, which presented an RMSE of 1.8 and MAPE of 20%. These results reinforce the possibility of a correlation between the defect depth parameter and the prediction of TF.

COMMITTEE MEMBERS:
Externo ao Programa - 2164698 - JOSE MARIA ANDRADE BARBOSA - nullExterno à Instituição - JULIO TENORIO PIMENTEL
Interno - 2994651 - TIAGO ANCELMO DE CARVALHO PIRES DE OLIVEIRA