On the accuracy of prediction models for the collapse strength of worn casing

Resumo

Oil and gas wells are typically built in highly complex and harsh environments. Casing is a crucial part of the well structure, undergoing a wide range of loads during the well life cycle, such as internal and external pressures, and axial force. After installation, these tubulars may develop wear grooves on their inner wall, caused by contact with the tool joints of the drill string. This reduction in wall thickness, combined with initial geometric imperfections, such as ovality and eccentricity, as well as residual stresses, may cause a significant reduction in tubular resistance, especially under external pressure (collapse). Thus, models that can estimate accurate collapse pressure, validated with realistic data, are extremely relevant. In this context, some collapse prediction models of worn casing are proposed in the literature, based on experimental, analytical, or numerical approaches. However, in several cases, the data used in deriving the models are limited in quantity and variety of cases. The present study aims to investigate and propose improvements in the equations for the collapse pressure of worn casing by using a large database of numerical simulations. Several Finite Element (FE) simulations are performed to generate a large database of collapsed pipes. Then, the accuracy of collapse prediction models from the literature is evaluated, while calibrating new model parameters to enhance precision. The study is carried out adopting material and geometric configurations that are commonly observed in worn casing tubulars of oil and gas wells. Two subsets of the large databases were generated: one is used for fitting the models and the other for testing them. The exploratory analysis of the FE database provides insights into the collapse strength deration in relation to relevant parameters, such as damage depth, tool joint radius and tube slenderness. The results compare the accuracy of the models, and a discussion about features influence is carried out in parallel.