No description available.

This thesis examines the impact of shocks in spot prices on long-term forward contracts in power markets. A unique comparison of efficiency of German and Hungarian power markets is provided. The risk premium on week-ahead forward contract is scrutinized by both data inspection and by unbiased forward rate hypothesis (UFRH) testing. Additionally, the ex-post market's prediction error for this product is explained by main drivers of spot electricity price, which are presented in section devoted to introduction to power markets. Expectedly, Hungarian forwards with longer time-to-delivery are found to react heavily on spot market shocks after controlling for changes in short-run marginal costs of conventional power plants. Such outcome applies both to intra-day and weekly time horizons. However, this evidence was not found for German market. These results point out to immaturity and the presence of inefficiencies in Hungarian power market. However, Hungarian risk premia on week-ahead and day-ahead forward products turn out to be considerably lower than for Germany. This was confirmed by UFRH tests on week-ahead forward contracts, where a significant risk premium was found in Germany as opposed to Hungarian risk premium. This finding is surprising since Hungarian spot prices are more prone to upward...

Machine learning models has gained traction as an effective tool for short-term electricity price forecasting, namely day ahead and hourly price forecasting. Efficient and accurate forecasting is crucial for demand and capacity planning to ensure stability and optimal use of resources. This project applies two proven machine learning models, LSTM and TCN, to electricity futures contracts in the Swedish pricing areas SE1 and SE3. Future contracts are used to secure the price of electricity in the future. A multivariate time series of fundamental data that correlates with electricity prices is used as input for the forecasting. Fur- thermore, a portfolio approach for hedging is evaluated based on the predictive performance of the models. The forecasting accuracy of the multivariate TCN model outperform the LSTM model. The optimal hedging strategy based on the TCN model indicated potential cost savings of 1.43% compared to a benchmark method. / Maskininlärnings modeller har vunnit mark som effektiva verktyg för att prognosticera kortsiktiga elpriser, för dagen före och timpriser. Effektiv och korrekt prognosticering är viktigt för att skatta behovs- och kapacitetsplanering för optimal resursanvändning. Det här projektet applicerar två välbeprövade modeller, LSTM och TCN, för att prognosticera terminskontrakt i de två svenska pris- områdena SE1 och SE3. Terminskontrakt används för att säkra elpriser i framtiden. En tidsserie, med flera variabler av fundamental data som korrelerar med elpriser, används för att prognosticera elpriser. Vidare utvärderas en portfölj approach för prissäkring baserat på prognoserna från modellerna. TCN modellen gav högre noggrannhet än LSTM modellen. Optimal prissäkringsstrategi baserad på TCN modeller resulterade i 1.43% lägre elpriser jämfört med bench- marks.

A large class of sequential decision making problems under uncertainty with multiple competing decision makers/agents can be modeled as stochastic games. Stochastic games having Markov properties are called Markov games or competitive Markov decision processes. This dissertation presents an approach to solve non cooperative stochastic games, in which each decision maker makes her/his own decision independently and each has an individual payoff function. In stochastic games, the environment is nonstationary and each agent's payoff is affected by joint decisions of all agents, which results in the conflict of interest among the decision makers. In this research, the theory of Markov decision processes (MDPs) is combined with the game theory to analyze the structure of Nash equilibrium for stochastic games. In particular, the Laurent series expansion technique is used to extend the results of discounted reward stochastic games to average reward stochastic games. As a result, auxiliary matrix games are developed that have equivalent equilibrium points and values to a class of stochastic games that are irreducible and have average reward performance metric. R-learning is a well known machine learning algorithm that deals with average reward MDPs. The R-learning algorithm is extended to develop a Nash-R reinforcement learning algorithm for obtaining the equivalent auxiliary matrices. A convergence analysis of the Nash-R algorithm is developed from the study of the asymptotic behavior of its two time scale stochastic approximation scheme, and the stability of the associated ordinary differential equations (ODEs). The Nash-R learning algorithm is tested and then benchmarked with MDP based learning methods using a well known grid game. Subsequently, a real life application of stochastic games in deregulated power market is explored. According to the current literature, Cournot, Bertrand, and Supply Function Equilibrium (SFEs) are the three primary equilibrium models that are used to evaluate the power market designs. SFE is more realistic for pool type power markets. However, for a complicated power system, the convex assumption for optimization problems is violated in most cases, which makes the problems more difficult to solve. The SFE concept in adopted in this research, and the generators' behaviors are modeled as a stochastic game instead of one shot game. The power market is considered to have features such as multi-settlement (bilateral, day-ahead market, spot markets and transmission congestion contracts), and demand elasticity. Such a market consisting of multiple competing suppliers (generators) is modeled as a competitive Markov decision processes and is studied using the Nash-R algorithm.

Power industry is undergoing a transition from the traditional regulated environment to the competitive power market. To have a reliable state estimator (SE) in the power market environment, two major challenges are emerging, i.e. to keep SE running reliably even under a contingency and to run SE over a grid with extremely large size. The objective of this dissertation is to use graph theory to address the above two challenges. To keep SE running reliably under a contingency, a novel topological approach is first proposed to identify critical measurements and examine network observability under a contingency. To advance the classical topological observability analysis, a new concept of contingency observability graph (COG) is introduced and it is proven that a power system network maintains its observability under a contingency if and only if its COG satisfies some conditions. As an application of COG, a two-stage heuristic topological approach is further developed based on the new concept of qualified COG (QCOG) to minimize the number of measurements and RTUs under the constraint that the system remains observable under any single contingency. To overcome the disadvantages of existing SE over extremely large networks, a textured distributed state estimator (DSE), which consists of the off-line textured architecture design and the on-line textured computation, is proposed based on COG and a new concept of Bus Credibility Index (BCI). The textured DSE is non-recursive, asynchronous and avoids central controlling node. Numerical tests verify that the performance of the new textured DSE algorithm improves greatly compared with existing DSE algorithms in respect of bad data detection and identification. Furthermore, the software implementation for DSE is formulated as an information integration problem over regional power markets, and is very challenging because of its size and complexity. A new concept of semantic knowledge warehouse (SKW), together with the proposed concepts of semantic reasoning software component (SRSC) and deduction credibility, is developed to implement such an information integration system.

The research presented in this thesis examines the use of electric vehicles and renewable energy to reduce emissions of CO₂, SO₂ and NO[subscript x], and within the state of Texas. The analysis examines the impact of increased renewable energy output and electric vehicle charging on the emissions of fossil fuel electric generators used to serve the bulk power system within Texas. The analysis then compares those impacts to alternative scenarios in which fossil fuel generation replaces some renewable energy generation, and Internal Combustion Engine (ICE) vehicles of varying efficiency are used instead of electric vehicles. This research uses temporally-resolved regression analysis combined with a unit commitment and dispatch model that incorporates several different scenarios for EV charging and fuel mixes to evaluate emissions outcomes based on a variety of conditions. Hourly historical generation and emission data for each fossil fuel generator, combined with hourly output data for non-fossil fuel units aggregated by fuel type (i.e. nuclear, wind, hydro-electric) within the Electric Reliability Council of Texas (ERCOT) footprint is regressed to assess the impact of wind generation output on fossil-fuel generation emissions. The regression analysis is used to assess potential increases in emissions resulting from the ramping of fossil-fuel Electric Generation Units (EGUs) to compensate for variability in wind generation output due to changing weather conditions. The unit commitment dispatch model is used to evaluate the impact of changes in customer demand due to increased usage and charging of electric vehicles on the ERCOT system and any resulting increase in emissions from generation used to meet this new demand. The model uses detailed cost, performance and emissions data for EGUs in the ERCOT footprint to simulate the impact of a variety of charging scenarios and fuel mixes on EGU dispatch patterns and any resulting change in system-wide emissions. The results of this model are combined with the results of the regression analysis to present a more complete analysis of the combined impacts of increase EV and renewable energy usage on the emissions of CO₂, SO₂ and NO[subscript x] within the ERCOT footprint. Based on these analyses the increases in renewable energy generation demonstrate clear benefits in terms of emission reductions when the impacts of increased emissions due to more frequent ramping of fossil-fuel units are taken into account. This analysis also finds that EV charging generally has emissions benefits across a range of charging patterns and bulk power system fuel mixes, although in certain circumstances EV charging might result in higher emissions than the use of ICE vehicles. This research finds when future ICE vehicles with reduced emissions are taken into account, approximately half of the modeled scenarios show net emissions benefits from EV charging, while half show net emissions costs when emissions impacts across pollutants are taken into account. / text

Europe aims for a transition towards less greenhouse gas emission and dependency on fossil fuels. The integration of intermittent renewable energy sources, as wind or solar power, can be facilitated by, among others, temporally decoupling demand and supply of electricity. If technologies for electrical energy storage are profitable, wide implementation could support the transition. Therefore, this study assesses the revenues and costs of long-term storage technologies and evaluates the possibility to stack several services provided by a storage unit to fully utilise it. Key market characteristics influential on storage potential are outlined and used to classify and compare the demand for storage in European markets. Considering them market clusters are formed comprising of countries with similar electricity market and from these two contrasting archetypal countries are chosen for further evaluation, Sweden and Germany. Storage technologies, as pumped hydro or compressed air, are delineated by essential technical specifications and used to determine the compatibility of them to corresponding services. A process for combining multiple services for a single storage unit is designed and employed to develop five cases. The results show that providing multiple services device can improve the profitability of the designed business cases by generating multiple revenue streams and increase the value to the electricity system. The stacking also minimises the storage’s idle time. Furthermore, the results demonstrate the influence of the key characteristics on the economic viability of the electrical energy storage in European markets. / Europa eftersträvar en övergång från att släppa ut mindre växthusgaser och minska eroendet av fossila bränslen. Att implementera förnybara energikällor, som vindkraft ller solenergi, kan underlättas genom att bland annat genom att tillfälligt rikoppla efterfrågan och leveransen av el. Om tekniken för energilagring är lönsam, an genomförande därför stödjas. Denna studie beskriver därför både intäkter ch kostnader för långsiktiga lagringstekniker och utvärderar även möjligheten att mplementera flera tjänster som tillhandahålls av en lagringsenhet till att kunna tnyttja denna till fullo. Marknaden av lagring kännetecknas redan ha potentiella förutsättningar och att den an användas för att klassificera och jämföra efterfrågan för lagring på de europeiska arknaderna. En marknadsanalys har valts att genomförts av länder med likartad lmarknad och de två länder som har välts ut för en vidare utvärdering är Sverige ch Tyskland. Lagringsteknik som vattenkraft som pumpas eller tryckluft, beskrivs ärmare med tekniska förutsättningar och används för att beskriva kompatibiliteten v tekniken och tjänsterna. Ett förfarande för att leverera flera tjänster från en enda innesenhet är utformad och används för att utveckla fem beskrivande fall. Resultatet visar att tillhandahålla flera tjänster från en enda lagringsenhet förbättrar önsamheten för affärsidéer genom att generera flera intäktsströmmar och ökar ärdet till elsystemet. Stapling minimerar också lagring är ledig tid. Dessutom visar esultatet om hur de viktigaste egenskaperna påverkas av de ekonomiska affärsmodellerna.

Includes vita. / Title from PDF of title page. / Document formatted into pages; contains 111 pages. / Thesis (Ph.D.)--University of South Florida, 2003. / Includes bibliographical references. / Text (Electronic thesis) in PDF format. / ABSTRACT: A large class of sequential decision making problems under uncertainty with multiple competing decision makers/agents can be modeled as stochastic games. Stochastic games having Markov properties are called Markov games or competitive Markov decision processes. This dissertation presents an approach to solve non cooperative stochastic games, in which each decision maker makes her/his own decision independently and each has an individual payoff function. In stochastic games, the environment is nonstationary and each agent's payoff is affected by joint decisions of all agents, which results in the conflict of interest among the decision makers. In this research, the theory of Markov decision processes (MDPs) is combined with the game theory to analyze the structure of Nash equilibrium for stochastic games. In particular, the Laurent series expansion technique is used to extend the results of discounted reward stochastic games to average reward stochastic games. / ABSTRACT: As a result, auxiliary matrix games are developed that have equivalent equilibrium points and values to a class of stochastic games that are irreducible and have average reward performance metric. R-learning is a well known machine learning algorithm that deals with average reward MDPs. The R-learning algorithm is extended to develop a Nash-R reinforcement learning algorithm for obtaining the equivalent auxiliary matrices. A convergence analysis of the Nash-R algorithm is developed from the study of the asymptotic behavior of its two time scale stochastic approximation scheme, and the stability of the associated ordinary differential equations (ODEs). The Nash-R learning algorithm is tested and then benchmarked with MDP based learning methods using a well known grid game. Subsequently, a real life application of stochastic games in deregulated power market is explored. / ABSTRACT: According to the current literature, Cournot, Bertrand, and Supply Function Equilibrium (SFEs) are the three primary equilibrium models that are used to evaluate the power market designs. SFE is more realistic for pool type power markets. However, for a complicated power system, the convex assumption for optimization problems is violated in most cases, which makes the problems more difficult to solve. The SFE concept in adopted in this research, and the generators' behaviors are modeled as a stochastic game instead of one shot game. The power market is considered to have features such as multi-settlement (bilateral, day-ahead market, spot markets and transmission congestion contracts), and demand elasticity. Such a market consisting of multiple competing suppliers (generators) is modeled as a competitive Markov decision processes and is studied using the Nash-R algorithm. / System requirements: World Wide Web browser and PDF reader. / Mode of access: World Wide Web.

Electricity production from renewable energy resources such as wind energy and photovoltaics is variable. Integration of these intermittent resources into the electricity system leads to new challenges in how to manage imbalance between supply and demand on the grid. One way to meet these challenges is to develop so-called smart grid solutions. One idea, called demand response, is to adjust the amount or timing of energy consumption, e.g. by control of household appliances, to provide flexibility that could be used to balance the grid. In aggregate, when applied to many units across the system, large volumes of energy could be made available when needed and this grid flexibility can be used as a product on the electricity regulation market. Despite the potential benefits, the number of demand response bids is currently low. The aim of this thesis is to identify barriers in the Swedish regulation market, and togive Sweden's transmission system operator, Svenska kraftnät, recommendations on how to facilitate implementations of the technique. This has been done throughliterature studies and a wide range of interviews with people within the electricity market sector. The results indicate that a combination of several elements in the complex energy system impede the introduction of demand response. The main issues are related to market regulations and profitability difficulties. The authors recommend that Svenska kraftnät lowers the minimum bid size in all of the Swedish bidding areas, and adjusts the balance responsibility agreement and the system of balancing settlement.

Orientador: Luiz Koodi Hotta / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Computação Cientifica / Made available in DSpace on 2018-08-12T12:42:20Z (GMT). No. of bitstreams: 1 Felizatti_HenriqueLeme_M.pdf: 5869463 bytes, checksum: 2d5c3f23857df348a4d6b755f4481c8e (MD5) Previous issue date: 2008 / Resumo: Contratos bilaterais celebrados no ambiente de balcão (ACL) do mercado de eletricidade brasileiro geralmente possuem em seu corpo derivativos embutidos que podem fazer com que o desempenho de carteiras de contratos fique exposto à variação dos preços do mercado à vista de energia. Muitos desses instrumentos derivativos, denominados flexibilidades contratuais, possuem mecanismos de funcionamento semelhantes a Take-or- Pay, Swing Options e Swaptions, os quais são funções de processos de consumo e de preços de mercado. A presença desses tipos de contratos nas carteiras de agentes faz com que o processo de gestão de risco dentro do mercado de energia brasileiro seja uma tarefa complexa. Nesta dissertação são propostas metodologias para tratar os principais passos relacionados com o processo de comercialização de contratos contendo flexibilidades de aumento e redução de montantes, Take-or-Pay ou sazonalização de energia. É mostrado como desmembrar esses contratos em instrumentos derivativos conhecidos e como modelar estatisticamente cada um desses instrumentos levando em conta seu mecanismo de funcionamento. São apresentadas também metodologias para a derivação de diretrizes de preços e prêmios para cada produto estudado. Após, são introduzidos procedimentos para avaliação, decisões de exercício e derivação de estratégias comerciais, para cada instrumento derivativo estudado, dentro de um contexto de risco versus desempenho. Avalia-se o desempenho da metodologia proposta utilizando carteiras de contratos com dados reais, e dois cenários de preços de mercado. Conclui-se que os modelos respondem de maneira coerente às mudanças de mercado e que a estrutura proposta pode ser aplicada para realizar gestão de risco e acompanhamento de carteiras com contratos contendo flexibilidades contratuais. / Abstract: Bilateral contracts traded over-the-counter in the Brazilian energy free markets (ACL) have, in most cases, embedded derivatives that can generate an exposure to spot price and its volatility. Many of these derivatives, called contract flexibilities, have the same effect as Take-or-Pay clauses, Swing Options and Swaptions, that are exercised depending on the energy consumption/needs of the counterparties, and on the market prices. The existence of these flexibilities in contracts makes risk management of portfolios a difficult task. This thesis proposes a methodology to assess the impacts of each flexibility in power supply contracts in the Brazilian Electricity Market and their associated risks. Each specific flexibility is broken-down to a combination of commonly known derivative instruments and consequently a statistical modeling is proposed to specifically evaluate each derivative. The use of these methodologies allows us to estimate risks and fair value prices of derivatives and consequently can be used for multiple purposes including: the "Mark-to- Model" accounting procedure, exercise decisions and portfolio monitoring. The portfolio used as example reflects current traded products and the price scenarios were selected to show the response of the methodology. The results were satisfactory and show the applicability of this methodology. / Mestrado / Mercado de Energia / Mestre em Estatística