abstract: Cohesive zone model is one of the most widely used model for fracture analysis, but still remains open ended field for research. The earlier works using the cohesive zone model and Extended finite element analysis (XFEM) have been briefly introduced followed by an elaborate elucidation of the same concepts. Cohesive zone model in conjugation with XFEM is used for analysis in static condition in order to check its applicability in failure analysis. A real time setup of pipeline failure due to impingement is analyzed along with a detailed parametric study to understand the influence of the prominent design variable. After verifying its good applicability, a creep model is built for analysis where the cohesive zone model with XFEM is used for a time dependent creep loading. The challenge in this simulation was to achieve coupled behavior of cracks initiation and propagation along with creep loading. By using Design of Experiment, the results from numerical simulation were used to build an equation for life prediction for creep loading condition. The work was further extended to account for fatigue damage accumulation for high cycle fatigue loading in cohesive elements. A model was conceived to account for damage due to fatigue loading along within cohesive zone model for cohesive elements in ABAQUS simulation software. The model was verified by comparing numerical modelling of Double cantilever beam under high cycle fatigue loading and experiment results from literature. The model was also applied to a major industrial problem of blistering in Cured-In-Plane liner pipelines and a demonstration of its failure is shown. In conclusion, various models built on cohesive zone to address static and time dependent loading with real time scenarios and future scope of work in this field is discussed. / Dissertation/Thesis / Masters Thesis Mechanical Engineering 2016

京都大学 / 新制・課程博士 / 博士(工学) / 甲第24295号 / 工博第5068号 / 新制||工||1791(附属図書館) / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 CRUZ Ana Maria, 教授 渦岡 良介, 教授 肥後 陽介 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

The transportation of oil and gas and their products through the pipelines is a safe and economically efficient way, when compared with other methods of transportation, such as tankers, railroad, trucks, etc. Although pipelines are usually well-designed, during construction and later in service, pipelines are subjected to a variety of risks. Eventually, some sections may experience corrosion which can affect the integrity of pipeline, which poses a risk in high-pressure operations. Specifically, in pipelines with long history of operation, the size and location of the corrosion defects need to be determined so that pressure levels can be kept at safe levels, or alternatively, a decision to repair or replace the pipe section can be made. To make this decision, there are several assessment techniques available to engineers, such as ASME B31G, MB31G, DNV-RP, software code called RSTRENG. These assessment techniques help engineers predict the remaining strength of the wall in a pipe section with a corrosion defect. The corrosion assessment codes in the United States, Canada and Europe are based on ASME-B31G criterion for the evaluation of corrosion defects, established based on full-scale burst experiments on pipes containing longitudinal machined grooves, initially conducted in 1960s. Because actual corrosion defects have more complex geometries than machined grooves, an in-depth study to validate the effectiveness of these techniques is necessary. This study is motivated by this need. The current study was conducted in several stages, starting with the deformation behavior of pipe steels. In Phase 1, true-stress-true plastic strain data from the literature for X42 and X60 steel specimens were used to evaluate how well four commonly used constitutive equations, namely, those developed by Hollomon, Swift, Ludwik and Voce, fit the experimental data. Results showed that all equations provided acceptable fits. For simplicity, the Hollomon equation was selected to be used in the rest of the study. In Phase 2, a preliminary finite element modeling (FEM) study was conducted to compare two failure criteria, stress-based or strain-based, performed better. By using data from the literature for X42 and X60 pipe steels, experimental burst pressure data were compared with predicted burst pressure data, estimated based on the two failure criteria. Based on this preliminary analysis, the stress-based criterion was chosen for further FEM studies. In Phase 3, failure data from real corrosion pits in X52 pipe steels with detailed profiles were used to develop a FEM scheme, which included a simplified representation of the defect. Comparison of actual and predicted burst pressures indicated a good fit, with a coefficient of determination (R2) level of 0.959. In Phase 4, burst pressure levels were estimated for real corrosion pits for the experiments from the same study as in Phase 3, but only with corrosion pit depths and length and without corrosion widths. Widths were estimated from the data used in Phase 3, by using an empirical equation as a function of pit length. There was significant error between experimental and predicted burst pressure. Errors in Phases 3 and 4 were compared statistically. Results showed that there is a statistically significant difference in the error when the width of the corrosion pit is unknown. This finding is significant because none of the assessment techniques in the literature takes width into consideration. Subsequently, a parametric study was performed on three defect geometries from the same study in Phase 3. The pit depths and lengths were held constant but widths were changed systematically. In all cases, the effect of the pit width on burst pressure was confirmed. In Phase 5, the three assessment techniques, ASME B31G, MB31-G and DNV-RP were evaluated by using experimental test results for X52 pipe. Synthetic data for deeper pits were developed by FEM and used along with experimental data in this phase. Two types of the error were distinguished to classify defects. Type I errors (α) and Type II errors (β) were defined using Level 0 evaluation method. Results showed that although ASME B31G is the most conservative technique, it is more reliable for short defects than MB31G and DNV-RP. The least conservative technique was DNV-RP but it yielded β error, i.e., the method predicted a safe operating pressure and pipe section would fail. Therefore, DNV-RP is not recommended for assessment of steel pipes, specifically for X52 pipes.

No description available.