Search

A Considerable amount of scenarios and constraints were produced by the basic systems-theoretic process analysis (STPA) application. It increased the difficulty to allocate resources for problem-solving. While these problems also applied to all other traditional hazard analysis tools, only STPA had not implement prioritization in the process. The purpose of this study is to develop a prioritization approach for STPA. Prioritization approach for systems-theoretic process analysis (PA-STPA) was proposed based on risk-based decision making (RBDM) approach. The proposed approach consisted of five main steps, four main steps from the basic STPA process and one additional step for solution proposal. Additionally, assessment processes were added to each step to quantify the qualitative assessment results. The obtained values were used to rank the results. PA-STPA has been applied to "the subsea gate box (SGB)" system and compared to the result from basic STPA. The study found that PA-STPA had higher analysis efficiency than basic STPA process while having the risk of reducing the safety level of the results by screening out high-risk scenarios. This effect might differ, depending on the unsafe control action (UCA) screening criteria used. The subjectivity of the assessment, formula used for calculation and result safety issue had been identified as weaknesses of PA-STPA. Several countermeasures had been done to resolve these issue. The unsolved problems were discussed and proposed as further works to refine the proposed PA-STPA.

Traditional reliability methods such as fault tree analysis and reliability block diagrams were not designed to account for the complex dynamic behavior of most industrial systems. These traditional reliability methods make simplifications in the modelling of many engineering systems, by assuming static operational and environmental conditions. Therefore, they are not able to account for the influence of the dynamics of the operations, environment or interactions between components of the system over the failure rates or degradation process of these, neither for the reciprocal influence of the failures and degradation processes over the process and system performance. For this purpose, dynamic reliability has been conceived as an extension of traditional reliability methods, by combining these with process modelling, automation and the theory of dynamic systems. This thesis focuses in hybrid modelling, as a mean to approach dynamic reliability problems. In hybrid modelling, the problem is described as dependent processes that are modelled separately and then coupled by means of shared variables, describing the dynamic behavior of the system in terms of deterministic physical laws and stochastic models describing failures and/or degradation processes. Driven by the goal in the oil and gas industry of finding new solutions for moving entire topside processing facilities to the seabed, a separator case study is presented, exploring the potential of hybrid models and demonstrating the modelling and simulations procedure and characteristics. The dynamic hybrid model of the case study is constructed using Simulink® as a tool for conducting simulations. Discussions about the modelling procedure and the hybrid model capabilities are included, pointing at the potential and challenges or drawbacks. The results show that hybrid models allow to integrate systems design with reliability analysis, from early stages of new technology development, giving a mean to find optimal solutions regarding safety, cost and performance, while providing confidence in the qualification of such technology. Dynamic reliability assessments are meant to fill the areas in which the traditional methods fall short, complementing these rather than replacing them.

Subsea production and processing systems have become a hot topic among research institutes and industries. While highlighting the advantages on production and economy, the reliability issues show a different picture with limited access, difficulty of maintenance and possibly lower availability. The influence of these issues on the system performance is thus to be studied, to evaluate the benefit of subsea systems. A literature review was carried out to prepare the background knowledge of the subsea system and its features. The main elements in subsea systems were introduced. Since the study case was about implementing a subsea separator, a more detailed introduction on subsea separation was provided. The drivers and challenges for subsea were also discussed, together with considerations for reliability modeling. Production availability analysis was selected as the approach for analysis. By conducting such an analysis, the system performance can be predicted and the critical components can be identified for system optimization. Two different types of approaches, the analytical approach and the simulation approach, were discussed in this thesis. The analytical approaches have restricted use and provide less precise results, but require less effort. The simulation approaches are more flexible and can provide more detailed predictions, but are rather time and cost consuming, and a solid mathematical and programming basis is often needed. A set of software tools were developed to simplify this approach. Since this thesis was done in collaboration with DNV GL, ExtendSim is suggested for the case study. The basic principles of Discrete Event Simulation, especially of ExtendSim, was introduced for a better understanding of how simulation is performed in these tools. After all preparations were ready, a case study was carried out to demonstrate the analysis process. Two difference cases, namely a reference case (with all separation on FPSO) and a subsea case (with the use of subsea separator), were modeled in ExtendSim based on a library from DNV GL. The modeling process was demonstrated in detail. The production availability and loss contributors for the two cases were derived from simulations. A discussion was then carried out on the model and the potential use of the results. Recommendations for future work were included at the end of this thesis.