Abstract

Thick-walled cylinders are crucial in various industrial applications, including mechanical, aerospace, naval, offshore, petrochemical, military, and electronics industries. These cylinders function as pressure vessels in diverse structures under different loading conditions. Some applications, such as steam boilers and aerospace propulsion systems, encounter severe cyclic thermo-mechanical loading conditions. Modeling the impact of these cyclic conditions is challenging due to the limited time between successive loads, preventing adequate cooling and resulting in thermal accumulation within the cylinder material. Thus, stress and temperature distributions within the cylinder thickness are altered, affecting mechanical and thermal properties. Existing models commonly assume temperature-independent material properties, utilizing the uncoupled thermo-elasticity approach. However, it is essential to adopt temperature-dependent material properties and a coupled thermo-elasticity approach for a precise estimation of residual temperature and stress distributions throughout the cylinder wall, significantly influencing thick-walled cylinder design.
Moreover, under severe loading conditions, simple virgin cylinders may fail to sustain applied loads without undesirable increases in thickness and weight. Consequently, various surface treatment manufacturing processes, such as shrink-fitting, wire-winding, and autofrettage, have been developed to enhance durability and load-bearing capacity. These processes induce beneficial compressive stresses near the bore region, countering tensile stresses that would normally develop during loading, thus improving their fatigue lifetime. Accurate prediction of residual stresses resulting from these processes is pivotal for optimal cylinder design. However, due to several limitations associated with each individual reinforcement process, different combinations of reinforcement processes are proposed to alleviate these limitations. Estimating residual stresses due to such combinations is complicated, leading many studies to avoid analytical models.
In response to these challenges, this thesis explores the behavior of temperature-dependent thick-walled cylinders treated with various reinforcement processes under cyclic thermomechanical loads. The classical coupled thermo-elasticity approach estimates thermal and mechanical responses, highlighting the significance of considering temperature-dependent material properties. Furthermore, an efficient analytical method is developed for estimating the residual stress profiles in cylinders with diverse reinforcement processes. This method forms the basis for a machine learning-based design optimization, streamlining the process and reducing computational costs significantly. Fatigue life assessment of the optimal configuration underscores the improvement achieved.