Stainless steel profiled walls have increasingly been used in the process and other industries to protect people and personnel against hydrocarbon or chemical explosions. The existence of various uncertainties, in particular the ones associated with explosion loading and parameters, make the current design and assessment which are based on single degree of freedom (SDOF) and deterministic approach, very complicated and in many cases leading to unreliable design assessment. Therefore, developing an appropriate reliability approach for assessing and designing blast wall structures would greatly assist in improving the safety of personnel and plant facilities. The objective of this research study is to develop a practical framework for performance based design of stainless steel profiled barrier blast walls, with specific focus on reliability assessment by implementing stochastic finite element analysis (SFEA). Initially, the current traditional SDOF method is reviewed to identify the related issues and weaknesses and accordingly an appropriate method for structural assessments of the blast walls is proposed. Furthermore, a comprehensive investigation on various available methods is carried out to identify a suitable probabilistic approach for the reliability assessments. The corresponding reliability of these structures is evaluated with a MCS method, implementing the Latin Hypercube Sampling (LHS) approach. A programming package is developed using Ansys Parametric Design Language (APDL), to generate parametric finite element models and to perform automated reliability assessments. The significant uncertainties are combined with an advanced analysis model to investigate the influence of loading, material and geometric uncertainties on the response of these structures under realistic boundary conditions and connection configuration. Effective implementation of the framework is achieved by the development of a combined programming package to deal with both finite element and reliability analyses. A further development for this research study is associated with the development of performance based design approach, using the results of the probabilistic and finite element assessments. This can be utilised for optimum and appropriate design of the blast wall structures, based on the defined performance levels. Application and practicality of the developed approach and associated programming package is demonstrated through a number of case studies of realistic stainless steel profiled barriers subject to explosion loading. The results of the preliminary probabilistic case studies confirm that the explosion loading is the main influential input parameter and also nonlinearities are more critical than dynamic effects for unstiffened profiled barrier blast walls. An appropriate dynamic load factor (DLF) is proposed for the preliminary stage of the design and assessments. It is observed that using the probabilistic approach can help identify the important variables and parameters to optimize the design of profiled blast walls, and to perform risk assessments for these structures. The study is expanded to develop a performance based design methodology, linking the probabilistic results with various performance levels and associated parameters (e.g., damage levels). The results and discussions of the case studies associated with performance based design assessments confirm the suitability of the proposed framework, and also highlight the complications in defining intermediate levels, without preliminary investigations. This shows that QRA approach and involvement of professionals can play an important role to develop performance levels and the associated objectives. The developed programming package and associated framework are expected to provide valuable guidance to professional design engineers and researchers, by obviating the need for complex computational requirements.

Thesis (M. S.)--Mechanical Engineering, Georgia Institute of Technology, 2009. / Committee Chair: Dr. Richard W. Neu; Committee Member: Dr. David L. McDowell; Committee Member: Dr. Itzhak Green.

Fretting, which can occur when two bodies in contact undergo a low amplitude relative slip, can drastically reduce the fatigue performance of a material. The extent of fretting damage is dependent on the material combination and is affected by many parameters, making it difficult to design against fretting. Some of these parameters include contact force, displacement amplitude, and contacting materials. This work develops a method for quantifying the extent of damage from fretting as a function of these parameters for a thin sheet of AISI 301 stainless steel in the full hard condition in contact with both ANSI A356 aluminum and AISI 52100 steel contacting bodies. Fretting experiments were conducted on a Phoenix Tribology DN55 Fretting Machine using a fixture which was developed for holding thin specimens. The displacement amplitude and normal force were systematically varied in order to cover a range that could typically be experienced during service. The tribological behavior was studied by analyzing friction force during cycling and inspecting the resulting surface characteristics. Fretting damaged specimens were cycled in tension in a servohydraulic test system to failure. The decrease in fatigue life caused by fretting damage was determined by comparing the stress-life (S-N) response of the fretted specimens to the S-N response of the virgin material, thus characterizing the severity of the fretting damage. The conditions that lead to the greatest reduction in life were identified in this way. Using the fracture mechanics based NASGRO model, an Equivalent Initial Flaw Size (EIFS) was used to quantify the level of fretting damage, thus separating the life of the component into crack nucleation and subsequent propagation. This method and data will allow engineers to design more robust components that resist fretting damage, thus increasing the safety and reliability of the system.