Techn. University, Diss., 2004--Dresden.

Techn. Univ., Diss., 2005--Darmstadt

University, Diss., 2004--Freiburg (Breisgau).

Techn. Universiẗat, Diss., 2006--Darmstadt.

Thesis (MEng)--Stellenbosch University, 2015. / ENGLISH ABSTRACT: Efficient and reliable energy generation capability is vital to any country's economic growth. Many strategic, tactical and operational decisions take place along the energy supply chain. Shortcomings in South Africa's electricity production industry have led to the development of an energy ow simulator. The energy ow simulator is claimed to incorporate all significant factors involved in the energy ow process from primary energy to end-use consumption. The energy ow simulator thus provides a decision support system for electric utility planners. The original aim of this study was to develop a global optimisation model and integrate it into the existing energy ow simulator. After gaining an understanding of the architecture of the energy ow simulator and scrutinising a large number of variables, it was concluded that global optimisation was infeasible. The energy ow simulator is made up of four modules and is operated on a module-by-module basis, with inputs and outputs owing between modules. One of the modules, namely the primary energy module, lends itself well to optimisation. The primary energy module simulates coal stockpile levels through Monte Carlo simulation. Classic inventory management policies were adapted to fit the structure of the primary energy module, which is treated as a black box. The coal stockpile management policies that are introduced provide a prescriptive means to deal with the stochastic nature of the coal stockpiles. As the planning horizon continuously changes and the entire energy ow simulator has to be re-run, an efficient algorithm is required to optimise stockpile management policies. Optimisation is achieved through the rapidly converging cross-entropy method. By integrating the simulation and optimisation model, a prescriptive capability is added to the primary energy module. Furthermore, this study shows that coal stockpile management policies can be improved. An integrated solution is developed by nesting the primary energy module within the optimisation model. Scalability is incorporated into the optimisation model through a coding approach that automatically adjusts to an everchanging planning horizon as well as the commission and decommission of power stations. As this study is the first of several research projects to come, it paves the way for future research on the energy ow simulator by proposing future areas of investigation. / AFRIKAANSE OPSOMMING: Effektiewe en betroubare energie-opwekkingsvermoë is van kardinale belang in enige land se ekonomiese groei. Baie strategiese, taktiese en operasionele besluite word deurgaans in die energie-verskaffingsketting geneem. Tekortkominge in Suid-Afrika se elektrisiteitsopwekkingsindustrie het tot die ontwikkeling van 'n energie-vloei-simuleerder gelei. Die energie-vloei-simuleerder vervat na bewering al die belangrike faktore wat op die energie-vloei-proses betrekking het van primêre energieverbruik tot eindgebruik. Die energie-vloei-simuleerder verskaf dus 'n ondersteuningstelsel aan elektrisiteitsdiensbeplanners vir die neem van besluite. Die oorspronklike doel van hierdie studie was om 'n globale optimeringsmodel te ontwikkel en te integreer in die bestaande energie-vloeisimuleerder. Na 'n begrip aangaande die argitektuur van die energievloei- simuleerder gevorm is en 'n groot aantal veranderlikes ondersoek is, is die slotsom bereik dat globale optimering nie lewensvatbaar is nie. Die energie-vloei-simuleerder bestaan uit vier eenhede en werk op 'n eenheid-tot-eenheid basis met insette en uitsette wat tussen eenhede vloei. Een van die eenhede, naamlik die primêre energiemodel, leen dit goed tot optimering. Die primêre energiemodel boots steenkoolreserwevlakke deur Monte Carlo-simulering na. Tradisionele voorraadbestuursbeleide is aangepas om die primêre energiemodel se struktuur wat as 'n swartboks hanteer word, te pas. Die steenkoolreserwebestuursbeleide wat ingestel is, verskaf 'n voorgeskrewe middel om met die stogastiese aard van die steenkoolreserwes te werk. Aangesien die beplanningshorison deurgaans verander en die hele energie-vloei-simulering weer met die energie-vloei-simuleerder uitgevoer moet word, word 'n effektiewe algoritme benodig om die re-serwebestuursbeleide te optimeer. Optimering word bereik deur die vinnige konvergerende kruis-entropie-metode. 'n Geïntegreerde oplossing is ontwikkel deur die primêre energiemodel en die optimering funksie saam te voeg. Skalering word ingesluit in die optimeringsmodel deur 'n koderingsbenadering wat outomaties aanpas tot 'n altyd-veranderende beplanningshorison asook die ingebruikneem en uitgebruikstel van kragstasies. Aangesien hierdie studie die eerste van verskeie navorsingsprojekte is, baan dit die weg vir toekomstige navorsing oor die energie-vloeisimuleerder deur ondersoekareas vir die toekoms voor te stel.

In this thesis we use density functional theory (DFT) to study the solvent mediated interactions between solvophobic, solvophilic and patchy nanostructures namely rectangular cross section blocks. We calculate both the density profiles and local compressibility around the blocks and the results obtained for our model system provide a means to understanding the basic physics of solvent mediated interactions between nanostructures, and between objects such as proteins in water, that possess hydrophobic and hydrophilic patches. Our results give an improved understanding of the behaviour of liquids around solvophobic objects and solvophobicity (hydrophobicity) in general. Secondly, we look into the physics incorporated in standard mean-field DFT. This is normally derived by making what appears to be a rather drastic approximation for the two body density distribution function: ρ(2)(r,r′) ≈ ρ(r)ρ(r′), where ρ(r) is the one-body density distribution function. We provide a rationale for why the DFT often does better than this approximation would make you expect. Finally, we develop a lattice model to understand the nature of the pattern formation exhibited by certain systems of particles deposited on liquid-air interfaces and in particular the nature of the transitions between the different patterned structures that are observed. This is done using Monte Carlo computer simulations and DFT and links the observed microphase ordering with the micellisation process seen e.g. in surfactant systems.

Recent technological developments are bringing about substantial changes that are converting traditional distribution networks into "smart" distribution networks. In particular, it is possible to observe seamless integration of Information and Communication Technologies (ICTs), including the widespread installation of automatic equipment, smart meters, etc. The increased automation facilitates active network management, interaction between market actors and demand side participation. If we also consider the increasing penetration of distributed generation, renewables and various emerging technologies such as storage and dynamic rating, it can be argued that the capacity of distribution networks should not only depend on conventional asset. In this context, taking into account uncertain load growth and ageing infrastructure, which trigger network investments, the above-mentioned advancements could alter and be used to improve the network design philosophy adopted so far. Hitherto, in fact, networks have been planned according to deterministic and conservative standards, being typically underutilised, in order for capacity to be available during emergencies. This practice could be replaced by a corrective philosophy, where existing infrastructure could be fully unlocked for normal conditions and distributed energy resources could be used for post fault capacity services. Nonetheless, to thoroughly evaluate the contribution of the resources and also to properly model emergency conditions, a probabilistic analysis should be carried out, which captures the stochasticity of some technologies, the randomness of faults and, thus, the risk profile of smart distribution networks. The research work in this thesis proposes a variety of post fault capacity services to increase distribution network utilisation but also to provide reliability support during emergency conditions. In particular, a demand response (DR) scheme is proposed where DR customers are optimally disconnected during contingencies from the operator depending on their cost of interruption. Additionally, time-limited thermal ratings have been used to increase network utilisation and support higher loading levels. Besides that, a collaborative operation of wind farms and electrical energy storage is proposed and evaluated, and their capacity contribution is calculated through the effective load carrying capability. Furthermore, the microgrid concept is examined, where multi-generation technologies collaborate to provide capacity services to internal customers but also to the remaining network. Finally, a distributed software infrastructure is examined which could be effectively used to support services in smart grids. The underlying framework for the reliability analysis is based on Sequential Monte Carlo Simulations, capturing inter-temporal constraints of the resources (payback effects, dynamic rating, DR profile, storage remaining available capacity) and the stochasticity of electrical and ICT equipment. The comprehensive distribution network reliability analysis includes network reconfiguration, restoration process, and ac power flow calculations, supporting a full risk analysis and building the risk profile for the arising smart distribution networks. Real case studies from ongoing project in England North West demonstrate the concepts and tools developed and provide noteworthy conclusions to network planners, including to inform design of DR contracts.

Frequentist model averaging as a newly emerging approach provides us a way to overcome the uncertainty caused by traditional model selection in estimation. It acknowledges the contribution of multiple models, instead of making inference and prediction purely based on one single model. Functional logistic regression is also a burgeoning method in studying the relationship between functional covariates and a binary response. In this paper, the frequentist model averaging approach is applied to the functional logistic regression model. A simulation study is implemented to compare its performance with model selection. The analysis shows that when conditional probability is taken as the focus parameter, model averaging is superior to model selection based on BIC. When the focus parameter is the intercept and slopes, model selection performs better.

A presente dissertação devolveu uma metodologia para determinar a probabilidade de escape de nêutrons conforme a energia e posição no reator. Para tanto, simulou-se um reator qualitativo semi-infinito de secção quadrada composto por três regiões homogêneas distintas. O domínio do reator foi subdividido em cem subcamadas concêntricas uniformes para a análise da fuga de nêutrons. Desta maneira, o nascimento de cada nêutron em cada camada foi registrado, bem como sua energia inicial. Os cálculos das trajetórias dos nêutrons foram efetuados usando o Método de Monte Carlo Físico. Assim, o código gerou a história paralela de 4x106 nêutrons, armazenando a energia final, posição final e fluxo angular na superfície. Desta forma, foi possível atribuir a probabilidade de escape de nêutrons provenientes de diferentes camadas conforme suas energias e posições iniciais. O método foi capaz de estabelecer o espectro de fuga, relações de dependência entre energia inicial e probabilidade de escape, além de observar que, sob certas condições, a probabilidade de escape possui crescimento exponencial ao longo do domínio. / This work presents a methodology to determine the neutron escape probability according to its energy and start position in the reactor. A semi-infinite qualitative reactor was simulated by a C++ program. This reactor has three distinct homogeneous regions. It was subdivided into hundred uniform concentric layers for a statistical analysis, allowing to record the birth and initial energy of each neutron in each layer. The neutron's path calculation was performed using Monte Carlo. The program generated 4x106 parallel neutron stories and has stored the final energy, position and angular flux. Thus, we determined the neutron escape probability from different layers. The method was able to estimate the leakage dependency with initial energy and position and it showed that the escape probability has a exponential growth tendency along the domain in certain conditions.

A disputa por parcelas de mercado impõe condições severas às empresas sob diversas perspectivas. Dentre elas salienta-se a crescente demanda por alta variedade de produtos, que por sua vez cria um ambiente de decisões gerenciais complexas e por vezes conflitantes. Neste contexto, dois pontos relativos a processos produtivos tornam-se cada vez mais importantes na implantação de estratégias diferenciadas: a programação da produção e a gestão de estoques. Esta dissertação apresenta uma sistemática que visa embasar decisões relativas a tais pontos, aprimorando o processo produtivo. Como primeira etapa, trata-se o problema relativo à programação da produção diária. Para tanto, é apresentada uma sistemática de seleção de variáveis de clusterização para agrupamento de produtos, a qual é integrada à Simulação de Monte Carlo (SMC) com objetivo de maximizar lucro. Os cenários propostos são aplicados em clusters (famílias de produtos) e não nos produtos de forma individual, simplificando e agilizando a programação da produção. O erro percentual em relação à situação real foi de 1%. A segunda etapa desta dissertação foca na seleção de variáveis de clusterização com vistas à gestão de estoques. Desta forma, é apresentada uma abordagem de seleção de variáveis para clusterização de 76 produtos em três clusters, sendo que para cada cluster são geradas políticas simultâneas de reposição dos produtos. Tais políticas são confrontadas, em termos de custos de colocação de pedidos e guarda de estoques, com os resultados gerados pelo Lote Econômico de Compras (LEC). A redução do volume de pedidos anuais se aproximou de 90%, enquanto que o incremento de custos relativos à guarda de produtos e processamento de pedidos foi de 0,2% frente ao custo gerado pelo LEC. / The dispute for larger market shares imposes hard conditions to companies in several perspectives. The growing demand for high variety of product models gives rise to complex productive scenarios, requiring precise managerial decisions. In this context, two points relating to production processes become increasingly important when implementing managerial strategies: production scheduling and inventory management. This dissertation presents an approach aimed at supporting decisions related to such points. As a first step, we tackle the daily scheduling problem presenting a systematic for selecting the most relevant variables for clustering products with similar features into groups; such groups are then integrated to a Monte Carlo Simulation (MCS) tailored to maximizing profit. In our propositions, managing clusters of products leads to simpler and faster managerial decisions regarding the production schedule. A proper training of the MCS parameters yielded a 1% deviation when compared to the real situation. The second part of this dissertation focuses on variable selection for clustering tailored to inventory management. For that matter, we present a variable selection approach for clustering 76 products into three clusters; such clusters are then integrated to a simultaneous inventory policy. The simultaneous policy aims at reducing costs of orders placement and simplifying the inventory management. When compared to the Economic Quantity Order (EOQ), our propositions reduced the number of order placements in 90%, while increasing costs related to inventory keeping in 0.2%.