Thesis (MEng)--Stellenbosch University, 2015. / ENGLISH ABSTRACT: In recent years, the management of physical assets has become increasingly important, even more so, in asset intensive organisations. This research work presents an overall approach to quantify reliability of rolling stock assets in the rail environment. The current maintenance management system in the case studied is over-reliant on cancellations and delays as reliability measure. The objectives of this study were, therefore, to develop a scientific approach to quantify the reliability of the rolling stock fleet and to develop a maintenance planning model based on system reliability. The research methodology followed made use of failure statistics, failure distributions and the interdependence of different systems to determine the impact of component failures on the overall system reliability. This could then be used to determine the reliability of individual train sets in order to better understand their performance. The reliability measure could be used for predicting component and train set failures as well as to better understand the contribution of maintenance towards reliability, hence the term Reliability Based Maintenance. The model, validated with real data, illustrates how the reliability measure can be used to determine maintenance intervals of different train sets. Based on the results, recommendations are made in relation to future planning of the maintenance strategy. / AFRIKAANSE OPSOMMING: Die bestuur van fisiese bates het in die afgelope tyd meer belangrik geword, veral in organisasies wat afhanklik is van hulle fisiese bates. Hierdie navorsing stel ‘n metode voor wat die betroubaarheid van rollende materiaal bates in die spoor bedryf kwantifiseer deur gebruik te maak van falingstatistiek. In die huidige instandhouding bestuurstelsel van die gevallestudie word daar te veel gesteun op kansellasies en vertragings van treine as ‘n betroubaarheids meting. Daarom was die doelwitte van die navorsing om ‘n wetenskaplike benadering te ontwikkel om betroubaarheid van rollende materiaal the kwantifiseer, en om ‘n instandhouding beplannings model voor te stel, gebaseer op sisteem betroubaarheid. Die navorsingsmetodologie is gebaseer op falingstatistieke, falingverspreidings, en die interafhanklikheid van stelsels word gebruik om die invloed van komponent falings op die betroubaarheid van die totale stelsel te bepaal. Hierdie benadering word dan gebruik om die betroubaarheid van individuele treinstelle en die werksverrigting van individuele trainstelle te bepaal. Hierdie instandhouding meting kan gebruik word om falings van komponente en treinstelle te voorspel, asook om die bydrae van instandhouding tot betroubaarheid beter te verstaan, daarom genoem Betroubaarheids Gebaseerde Instandhouding. Dit is ook geïllustreer hoe die betroubaarheid meting gebruik kan word om instandhouding intervalle te bepaal. Die betroubaarheid model is met werklike data gevalideer en aanbevelings word gemaak hoe om betroubaarheid te gebruik om die toekomstige beplanning van instandhouding te doen.

A method able to analyze and optimize subsystems could be useful to reduce project cost, increase subsystem reliability, improve overall aircraft safety and reduce subsystem weight. The earlier the optimization of development of an aircraft in the design phase can be performed, the better the yield of the optimization becomes. This master thesis was formed in order to construct an automatic analysis method, implementing a Matlab script, evaluating devices forming aircraft subsystems using a Genetic Algorithm. In addition to aircraft subsystems, the method constructed in the work is compatible with systems of various industries with minor modifications of the script.

Marine renewable energy promises to provide a significant contribution to the future electricity supply. It is estimated that 17% of today's UK electricity demand could be generated from wave and tidal sources. The ambition to harvest this resource is in the public interest, as it eases the pressures on energy security, holds the potential to reduce carbon emissions and has the prospect to create a new UK industry sector worth £15 billion. From an engineering perspective, marine energy is one of the least developed renewable energy technologies and has to be regarded as unproven. The reliability of components and devices in the harsh marine environment is one of the main engineering challenges. Reliability assessments and the assurance of acceptable reliability levels are dependant on the adequacy of failure information, which is scantily available for marine energy. This thesis shows that large failure rate uncertainties impede the reliability assessment for wave energy converters and how a suite of experimental, numerical and statistical methods can be applied to improve scarcely available reliability information. The analysis of component load conditions identifies fatigue as failure mode of concern and the fatigue life of mooring lines and marine power cables is quantified in a floating wave energy application. A Bayesian statistical approach and dedicated service-simulation component testing is proposed, and implemented to improve the quality of reliability estimates and to provide relevant data and assurance. The methods presented, along with the results, will assist reliability assessment and design during early development stages, and will inform the prediction of maintenance requirements during operation. Reliable marine energy systems will be the technical enabler for the successful transition of prototype devices to a commercially viable marine energy industry.

In the rapidly evolving landscape of the Internet of Things (IoT), ensuring consistency and reliability becomes a top priority for a seamless user experience. In many instances, reliability is assessed through Quality of Service (QoS) metrics, sidelining traditional reliability metrics that thrive on time-dependent failure rates. The lack of a comprehensive framework that fully integrates all layers of an IoT system adds to the complexity. This gap makes it difficult to pinpoint specific areas that need improvement and to conduct a thorough assessment of the system’s reliability. This project addresses this intricate challenge, which holds significant relevance for industry professionals but remains unresolved. The project introduced an IoT architecture spanning the Power, Device, Edge, Application, and Cloud Layers. Within each layer, potential failure points were identified, and the reliability was analysed deploying time-based failure rates with an exponential distribution. Reliability Block Diagrams (RBDs) were employed to map the intricate inter-dependencies, though the framework’s adaptive nature allows for other system reliability methodologies. One of the primary outcomes of this research is the development of a new framework tailored for evaluating the reliability of various IoT system components. This framework yields insights into both system reliability and availability over time, serving as a pivotal tool for stakeholders such as device manufacturers, system integrators, network providers, and research institutions. The results show how the framework emerges as a pivotal starting point for IoT system reliability evaluation. Before this thesis, the feasibility of such a framework was uncertain, with concerns about its potential bias – being either too pessimistic or optimistic. Yet, the tangible results from this work affirm its capability to provide a balanced and reasonable reliability estimation, given the intricacies of IoT devices. This paves the way for subsequent research, enabling a deeper dive into targeted enhancements and fostering a nuanced understanding of IoT reliability. / I det snabbt föränderliga IoT-landskapet (Internet of Things) är det av högsta prioritet att säkerställa konsekvens och tillförlitlighet för en sömlös användarupplevelse. I många fall bedöms tillförlitligheten med hjälp av QoSmått (Quality of Service), vilket innebär att traditionella tillförlitlighetsmått som bygger på tidsberoende felfrekvenser åsidosätts. Avsaknaden av ett heltäckande ramverk som integrerar alla lager i ett IoT-system bidrar till komplexiteten. Denna brist gör det svårt att identifiera specifika områden som behöver förbättras och att göra en grundlig bedömning av systemets tillförlitlighet. Detta projekt tar itu med denna komplicerade utmaning, som har stor relevans för branschfolk men som fortfarande inte har lösts. Projektet introducerade en IoT-arkitektur som spänner över kraft-, enhets-, Edge-, applikationsoch molnlagren. Inom varje lager identifierades potentiella felpunkter och tillförlitligheten analyserades med hjälp av tidsbaserade felfrekvenser med en exponentiell fördelning. Tillförlitlighetsblockdiagram (RBD) användes för att kartlägga de komplicerade ömsesidiga beroendena, även om ramverkets adaptiva natur möjliggör andra metoder för systemtillförlitlighet. Ett av de främsta resultaten av denna forskning är utvecklingen av ett nytt ramverk som är skräddarsytt för att utvärdera tillförlitligheten hos olika IoT-systemkomponenter. Detta ramverk ger insikter om både systemets tillförlitlighet och tillgänglighet över tid, och fungerar som ett viktigt verktyg för intressenter som tillverkare av enheter, systemintegratörer, nätverksleverantörer och forskningsinstitutioner. Resultaten visar hur ramverket framstår som en viktig utgångspunkt för utvärdering av IoT-systemens tillförlitlighet. Före den här avhandlingen var det osäkert om ett sådant ramverk var genomförbart, med farhågor om dess potentiella partiskhet - att vara antingen för pessimistisk eller optimistisk. De konkreta resultaten från detta arbete bekräftar dock ramverkets förmåga att ge en balanserad och rimlig uppskattning av tillförlitligheten, med tanke på IoT-enheternas komplexitet. Detta banar väg för efterföljande forskning, vilket möjliggör en djupare analys av riktade förbättringar och främjar en nyanserad förståelse av IoT-tillförlitlighet.