Electricity markets may become more sensitive to weather conditions because of higher penetration of renewable energy sources and climatic changes. Albania is 100% reliant on hydropower for its domestic energy generation, making this country compelling to investigate as it is highly sensitive to changing weather conditions. We use an ARMA-GARCH model to investigate whether weather and economic factors had a relationship with monthly hydroelectricity prices in the Albanian Energy Market in the period 2013-2018. We find that electricity price is affected by variations in weather and is not utterly robust to extreme hydrological changes. Generally, our dependent variable appears to be particularly influenced by air pressure followed by temperature and rainfall. We also perceive that there is a relationship between economic factors and hydroelectricity prices, where residual supply appears to have a significant negative relationship with our dependent variable. However, we were originally anticipating a higher dependency of electricity prices on weather conditions, due to the inflated hydro-power reliance for electricity production in the Albanian Energy Market. This effect is offset by several factors, where the state monopolized behaviour of the energy sector occupies a predominant influence on our results.

Electricity is an essential good, which can hardly be replaced. It can be produced thanks to a wide rangeof sources, from coal to nuclear, not to mention renewables such as wind and solar. In order to meetdemand at the lowest cost, an optimisation is made on electricity markets between the differentproduction plants. This optimisation mainly relies on the electricity production cost of each technology.In order to include long-term constraints in the short-term optimisation, a so-called use value (oropportunity cost) can be computed and added to the production cost. One long-term constraint thatEDF, the main French electricity producer, is facing is that its gas plants cannot exceed a given numberof operation hours and starts between two maintenances. A specific software, DiMOI, computes usevalues for this double constraint but its parameters needs to be tested in order to improve thecomputation, as it is not thought to work properly.DiMOI relies on dynamic programming and more particularly on an algorithm called Bellman algorithm.The software has been tested with EDF R&D department in order to propose some modellingimprovements. Electricity and gas market prices, together with real plant parameters such as startingcosts, operating costs and yields, were used as inputs for this work, and the results were checkedagainst reality.This study gave some results but they appeared to be invalid. Indeed, an optimisation problem wasdiscovered in DiMOI computing core: on a deterministic context, a study with little degrees of freedomwas giving better profits than a study with more degrees of freedom. This problem origin was notfound precisely with a first investigation, and the R&D team expected the fixing time to be very long.The adaptation of a simpler tool (MaStock) was proposed and made in order to replace DiMOI. Thisproject has thus led to DiMOI giving up and its replacement by MaStock. Time was missing to testcorrectly this tool, and the first study which was made was not completely positive. Further studiesshould be carried out, for instance deterministic ones (using real past data) whose results could becompared to reality.Some complementary studies were made from a fictitious system, in order to study the impact of someparameters when computing use values and operations schedules. The conclusions of these studiesare the little impacts that changes in gas prices and start-up costs parameters have on the global resultsand the importance of an accurate choice in the time periods durations used for the computations.Unfortunately these conclusions might be too specific as they were made on short study periods.Further case studies should be done in order to reach more general conclusions.

<p>Online decision making with switching costs has received considerable attention in many practical problems that face uncertainty in the inputs and key problem parameters. Because of the switching costs that penalize the change of decisions, making good online decisions under such uncertainty is known to be extremely challenging. This thesis aims at providing new online algorithms with strong performance guarantees to address this challenge.</p> <p><br></p> <p>In part 1 and part 2 of this thesis, motivated by Network Functions Virtualization and smart grid, we study competitive online convex optimization with switching costs. Specifically, in part 1, we focus on the setting with an uncertainty set (one type of prediction) and hard infeasibility constraints. We develop new online algorithms that can attain optimized competitive ratios, while ensuring feasibility at all times. Moreover, we design a robustification procedure that helps these algorithms obtain good average-case performance simultaneously. In part 2, we focus on the setting with look-ahead (another type of prediction). We provide the first algorithm that attains a competitive ratio that not only decreases to 1 as the look-ahead window size increases, but also remains upper-bounded for any ratio between the switching-cost coefficient and service-cost coefficient.</p> <p><br></p> <p>In part 3 of this thesis, motivated by edge computing with artificial intelligence, we study bandit learning with switching costs where, in addition to bandit feedback, full feedback can be requested at a cost. We show that, when only 1 arm can be chosen at a time, adding costly full-feedback is not helpful in fundamentally reducing the Θ(<em>T</em>2/3) regret over a time-horizon <em>T</em>. In contrast, when 2 (or more) arms can be chosen at a time, we provide a new online learning algorithm that achieves a significantly smaller regret equal to <em>O</em>(√<em>T</em>), without even using full feedback. To the best of our knowledge, this type of sharp transition from choosing 1 arm to choosing 2 (or more) arms has never been reported in the literature.</p>

<p>In the US, seven regional transmission organizations (RTOs) operate wholesale electricity markets within three largely independent transmission systems, the largest of which includes five RTO regions and many vertically integrated utilities.</p> <p>RTOs operate a day-ahead and a real-time market. In the day-ahead market, generation and demand-side resources are optimally scheduled based on bids and offers for the next day. Those schedules are adjusted according to actual operating conditions in the real-time market. Both markets involve a unit commitment calculation, a mixed integer program that determines which generators will be online, and an economic dispatch calculation, an optimization determines the output of each online generator for every interval and calculates locational marginal prices (LMPs).</p> <p>The use of LMPs for the management of congestion in RTO transmission systems has brought efficiency and transparency to the operation of electric power systems and provides price signals that highlight the need for investment in transmission and generation. Through this work, we aim to extend these efficiency and transparency gains to the coordination across RTOs. Existing market-based inter-regional coordination schemes are limited to incremental changes in real-time markets. </p> <p>We propose a multi-regional unit-commitment that enables coordination in the day-ahead timeframe by applying a distributed approach to approximate a system-wide optimal commitment and dispatch while allowing each region to largely maintain their own rules, model only internal transmission up to the boundary, and keep sensitive financial information confidential. A heuristic algorithm based on an extension of the alternating directions method of multipliers (ADMM) for the mixed integer program is applied to the unit commitment. </p> The proposed coordinated solution was simulated and compared to the ideal single-market scenario and to a representation of the current uncoordinated solution, achieving at least 58% of the maximum potential savings, which, in terms of the annual cost of electric generation in the US, could add up to nearly $7 billion per year. In addition to the coordinated day-ahead solution, we develop a distributed solution for financial transmission rights (FTR) auctions with minimal information sharing across RTOs that constitutes the first known work to provide a viable option for market participants to seamlessly hedge price variability exposure on cross-border transactions.

Renewable cannibalization refers to the phenomenon where the increasing penetration of zeromarginal cost renewable energy sources, such as solar and wind power, leads to a decline in their market value. By extension, this threatens to reduce investment incentives in wind and solar. Based on theory of supply and demand, the extent to which cannibalization is experienced should increase as the penetration of wind and solar increases. Since electricity prices, and therefore cannibalization, are set with considerations to domestic system dynamics, regulations and policies, cannibalization research are typically limited to investigate it for a specific country or region. For this thesis, Spain has been chosen for the case study. Coupling an already high penetration of both wind and solar with ambitious goals for wind and solar capacity expansion, Spain constitutes an interesting case study for cannibalization research. The investigation is centered around two factors, the market capturing price (MCP) and the cannibalization factor (CF). The MCP is the generation weighted electricity price and measures absolute cannibalization, while the CF is the ratio between the MCP and the average electricity price, thus constituting a relative cannibalization measurement. These will be calculated using hourly day-ahead wind and solar forecasts, and day-ahead electricity prices. A time series econometric study is then conducted to quantify the cannibalization effect together with potentially influential factor that, in theory, should be the driving factors behind the cannibalization phenomenon. To investigate dynamic affects across these factors, temporal regressions are conducted. In these regressions, data is isolated in different groups based on their characteristics. Furthermore, carbon pricing and granular guarantees of origin (GOs) are investigated and assessed based on their potential for alleviating the effect of cannibalization.  The study finds that both wind and solar cannibalizes their own market values, and that cannibalization occurs across technologies. The results indicate that there is a negative marginal relation between wind and solar infeed, suggesting the presence of both self-cannibalization and cross-cannibalization effects on their respective MCPs. The same can be said for the CF of solar, for which there is a negative marginal effect with the infeed of wind and solar. These statements hold true across all ranges of wind and solar penetration investigated in the temporal regression analysis. Moreover, the negative relations increase as the penetration range increases, indicating that cannibalization effects are stronger at high renewable penetration. For wind power, this is not entirely the case. Although the regression results yielded a low coefficient of determination (R2), indicating weak explanatory power in the regressions, it is possible to interpret whether the marginal effects are positive or negative. The temporal regression results indicate that there is a positive marginal effect between solar infeed and the CF of wind when the penetration of solar is lower than 10%. Thus, considering the ambitious wind and solar capacity targets of Spain, the economic viability of wind and solar could be threatened. Furthermore, the results from the study indicate that although carbon pricing helps increasing the MCP for both wind and solar by adding an increment to the day-ahead price, it reduces the CF. Furthermore, carbon pricing is a limited tool for alleviating cannibalization, considering that it requires. Thus, once the system is fully decarbonized, carbon pricing is rendered obsolete. On the contrary, the implementation of granular (hourly) GOs provide extra revenue in addition to the revenue from sold electricity. Furthermore, its immediate effect is that it increases the revenue while not impacting the price of electricity. Thus, it helps counteracting both absolute and relative cannibalization effects. / Förnybar kannibalisering hänvisar till fenomenet där en ökande markandspenetration av förnybar energi med noll marginalkostnad, så som vind- och solkraft, leder till en minskning i deras marknadsvärde. Detta hotar i förlängningen att minska incitament för fortsatta investeringar inom vind- och solkraft. Baserat på teorier om utbud och efterfrågan bör graden av kannibalisering öka i takt med att marknadspenetrationen av vind- och solenergi ökar inom ett givet system. Eftersom elpriser, och därmed graden av kannibalisering, beror på nationella dynamiker inom kraftsystemet, föreskrifter och riktlinjer, utförs ofta studier om kannibaliserade effekter inom förnybar energi från fallstudier inom ett land eller en regions system. I denna avhandling har Spanien valt som fallstudie. Med en redan hög grad av förnybart inom elsystemet, samt ambitiösa mål för fortsatt utbyggnad av vind- och solkraft, utgör Spanien en intressant fallstudie för forskning om förnybar kannibalisering.  Undersökningen utgår från två faktorer; det produktionsviktade elpriset (MCP) och kannibaliseringsfaktorn (CF). MCP utgör en absolut faktor och är det produktionsviktade elpriset för en viss teknologi, medan CF utgör ett mått på relativ kannibalisering och beräknas genom att dividera MCP med det genomsnittliga elpriset. Dessa två faktorer beräknas genom att använda timvisa prognoser från dagen-före marknaden för produktion av vind- och solkraft, samt priser från spotmarknaden. Därefter utförs en ekonometrisk studie baserad på tidseriedata för att kvantifiera de kannibaliserande effekterna med avseende på utomstående faktorer som bör kunna ses som drivkrafter bakom kannibaliseringsfenomenet. Dessutom undersöks och bedöms koldioxidprissättning och timvisa ursprungsgarantier baserat på deras förmåga att lindra effekterna av kannibalisering i Spanien. Studien visar att både vind- och solenergi kannibaliserar sina egna marknadsvärden, samt att kannibalisering sker mellan kraftslag. Resultaten indikerar att det finns en negativ marginaleffekt mellan produktion av vind- och solel, vilket tyder på förekomsten av både självkannibaliserande och korskannibaliserande effekter på deras respektive produktionsviktade elpriser (MCP). Samma kan sägas gälla för solkraftens kannibaliseringsfaktor (CF), där det finns en tydlig negativ marginaleffekt gentemot produktion av vind- och solel. Dessa påståenden stämmer över alla intervall av vind- och solpenetration som undersökts i den temporala regressionsanalysen. Dessutom ökar de negativa relationerna desto högre intervall som avses, vilket tyder på att kannibaliseringseffekten är starkare vid hög penetration av förnybar energi. För vindkraft stämmer inte detta helt. Även om regressionsresultaten gav en låg determinationskoefficient (R2), vilket indikerar svag förklarande kraft i regressionerna, är det möjligt att tolka om marginaleffekterna är positiva eller negativa. De temporala regressionsresultaten visar att det finns en positiv marginal- effekt mellan produktionen av solel och CF för vindkraft när solkraft utgör mindre än 10% av den dagliga produktionen. Med tanke på Spaniens ambitiösa mål för vind- och solkapacitet kan den ekonomiska livskraften för vind- och solenergi hotas. Vidare visar resultaten från studien att även om koldioxidprissättning hjälper till att öka MCP för både vind- och solenergi genom att skapa ett tillägg på spotmarknadspriset, minskar det CF. Dessutom är koldioxidprissättning ett begränsat verktyg för att lindra kannibalisering, med tanke på att det kräver. När systemet är fullständigt avkarboniserat blir koldioxidprissättning överflödig. Å andra sidan ger implementeringen av granulära (timvisa) ursprungsgarantier (GOs) extra intäkter utöver intäkterna från såld el. Dessutom ökar det omedelbart intäkterna utan att påverka elpriset. På så sätt hjälper det till att motverka både absoluta och relativa kannibaliseringseffekter.

In my dissertation, I present ATLAS — the Auto-Training Look-Ahead Scheduler. ATLAS improves service to applications with regard to two non-functional properties: timeliness and overload detection. Timeliness is an important requirement to ensure user interface responsiveness and the smoothness of multimedia operations. Overload can occur when applications ask for more computation time than the machine can offer. Interactive systems have to handle overload situations dynamically at runtime. ATLAS provides timely service to applications, accessible through an easy-to-use interface. Deadlines specify timing requirements, workload metrics describe jobs. ATLAS employs machine learning to predict job execution times. Deadline misses are detected before they occur, so applications can react early.:1 Introduction 2 Anatomy of a Desktop Application 3 Real Simple Real-Time 4 Execution Time Prediction 5 System Scheduler 6 Timely Service 7 The Road Ahead Bibliography Index

[pt] A possível mudança no paradigma de formação de preço no Brasil do modelo vigente por custos auditados para o modelo por oferta, com o objetivo de modernizar o Setor Elétrico e buscar práticas que incentivem a competição, implica a necessidade de estudos prévios para auxiliar o processo de transição e a definição do desenho de mercado adequado à realidade brasileira. Nesse sentido, o uso de modelos de equilíbrio, notadamente o Equilíbrio de Nash, desponta como uma poderosa ferramenta ex-ante que permite analisar o comportamento dos competidores para identificar possíveis ineficiências a serem mitigadas. Com esse fim, o presente trabalho modela o processo decisório de ofertas ótimas de um competidor em um mercado de energia elétrica de dia-seguinte de base hidrotérmica por um modelo de otimização binível, possibilitando a identificação do Equilíbrio de Nash do mercado através de um algoritmo baseado em Gauss-Seidel. Adicionalmente, o método é aplicado a dois experimentos numéricos: a um sistema-teste de três barras e a um caso representativo do sistema brasileiro completo, permitindo a análise do comportamento dos competidores a partir da comparação dos resultados com os modelos de Custos Auditados e Equilíbrio Competitivo. Foram observados os impactos das afluências e das cascatas de usinas hidrelétricas de múltiplos proprietários nas receitas e, consequentemente, no comportamento dos competidores. Por fim, foi verificada a ocorrência de competição exclusivamente pelas quantidades, além da prática de retenção de ofertas por parte dos competidores para a modificação do preço de equilíbrio de mercado, aumentando suas receitas. / [en] The potential shift in Brazil s energy pricing paradigm from the current Audited Costs model to the Bid-Bases model, with the aim of modernizing the electricity sector and seeking practices that encourage competition, implies the need for preliminary studies to assist the transition process and define the appropriate market design for the Brazilian reality. In this sense, the use of equilibrium models, notably the Nash Equilibrium, emerges as a powerful ex-ante tool that allows the analysis of competitors behavior to identify possible inefficiencies to be mitigated.To this end, this thesis models the optimal bidding decision process of a competitor in a hydrothermal day-ahead electricity market using a bilevel optimization model, enabling the identification of the Nash Equilibrium of the market through an algorithm based on the Gauss-Seidel. Additionally, the method is applied to two numerical experiments: a three-bus test system and a representative case of the complete Brazilian system, allowing for the analysis of competitors behavior by comparing the results with the Audited Costsand Competitive Equilibrium models. The impacts of water inflows and hydro plants in cascade with distinct ownership on revenues and consequently competitors behavior were observed. Finally, the occurrence of competition exclusively in quantities was verified, as well as the practice of quantity bids retention by competitors in order to modify the market equilibrium price, increasing their revenues.

[pt] A utilização de modelos de inteligência computacional para tarefas de previsão Multi-Step de séries temporais tem apresentado resultados que permitem considerar estes modelos como alternativas viáveis para este tipo de problema. Baseados nos requerimentos computacionais e a melhora de desempenho, recentemente novas áreas de pesquisa têm sido apresentadas na comunidade científica. Este é o caso do Reservoir Computing, que apresenta novos campos de estudo para redes neurais do tipo recorrentes, as quais, no passado, não foram muito utilizados devido à complexidade de treinamento e ao alto custo computacional. Nesta nova área são apresentados modelos como Liquid State Machine e Echo State Networks, que proporcionam um novo entendimento no conceito de processamento dinâmico para redes recorrentes e propõem métodos de treinamento com baixo custo computacional. Neste trabalho determinou-se como foco de pesquisa a otimização de parâmetros globais para o projeto das Echo State Networks. Embora as Echo State Networks sejam objeto de estudo de pesquisadores reconhecidos, ainda apresentam comportamentos desconhecidos, em parte pela sua natureza dinâmica, mas também, pela falta de estudos que aprofundem o entendimento no comportamento dos estados gerados. Utilizando como fundamento o modelo Separation Ratio Graph para análise do desempenho, é proposto um novo modelo, denominado ESN-GA-SRG, que usa como base redes ESN com otimização de parâmetros globais utilizando GA e seleção de topologias para Reservoir por meio de análise de estados empregando SRG. O desempenho deste novo modelo é avaliado na previsão das 11 séries que compõem a versão reduzida do NN3 Forecasting Competition e em 36 séries da competição M3, selecionadas segundo características de periodicidade na amostragem, assimetria, sazonalidade e estacionaridade. O desempenho do modelo ESN-GA-SRG na previsão dessas séries temporais foi superior na maioria dos casos, com significância estatística, quando comparado com outros modelos da literatura. / [en] The use of computational intelligence models for Multi-Step time series prediction tasks has presented results that allow us to consider these models as viable alternatives for this type of problem. Based on computational requirements and performance improvement, new areas of research have recently been presented in the scientific community. This is the case of Reservoir Computing, which presents new fields of study for recurrent-type neural networks, which in the past were not widely used because of training complexity and high computational cost. In this new area are presented models such as Liquid State Machine and Echo State Networks, which provide a new understanding of the concept of dynamic processing for recurring networks and propose methods of training with low computational cost. In this work, we determined the optimization of global parameters for the Echo State Networks project. Although Echo State Networks are the object of study by recognized researchers, they still present unknown behavior, partly due to their dynamic nature, but also due to the lack of studies that deepen behavior understanding of the generated states. Based on the Separation Ratio Graph model for performance analysis, a new model, called ESN-GA-SRG, is proposed, which uses ESN networks with global parameter optimization using GA and selection of topologies for Reservoir through analysis of States employing SRG. The performance of this new model is evaluated to forecast the 11 series that made up the reduced version of the NN3 Forecasting Competition and for 36 series of the M3 competition, selected according to characteristics of periodicity in sampling, asymmetry, seasonality and stationary. The performance of the ESN-GA-SRG model in predicting these time series was superior in most cases, with statistical significance when compared with other models in the literature.

No description available.

No description available.