ebook img

Identifying Energy Arbitrage Opportunities based on Statistical Analysis of Electricity Prices PDF

63 Pages·2014·3.1 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Identifying Energy Arbitrage Opportunities based on Statistical Analysis of Electricity Prices

eeh power systems laboratory Eren C¸am Identifying Energy Arbitrage Opportunities based on Statistical Analysis of Electricity Prices Master Thesis PSL1407 EEH – Power Systems Laboratory Swiss Federal Institute of Technology (ETH) Zurich Expert: Prof. Dr. G¨oran Andersson Supervisor: Line Roald Zurich, August 31, 2014 Abstract Wholesaleelectricitypricesareexpectedtobecomeincreasinglymorevolatile duetothegrowingshareofintermittentenergysourcesinthetotalelectricity generation [1], [2]. This increased volatility, combined with further advances in storage technologies, is likely to result in energy arbitrage becoming more significant in the wholesale electricity markets in the near future. To maximize the revenue obtained through energy arbitrage, it is impor- tant to install the systems for energy arbitrage at locations with high arbi- trage potential. For this reason, identifying most suitable interconnections aswellastheindividualnodesrelativelyquicklywithoutdoingextensiveop- timization calculations is of great importance. In this thesis, the main goal is therefore to identify the statistical properties which are good indicators of energy arbitrage potential. Furthermore, we also aim to observe the effect of increasing wind penetration on energy arbitrage potential. An energy storage system (ESS) optimization model was implemented to calculate the upper bound on the arbitrage revenue at various nodes of the following U.S. 5-minute interval real-time electricity markets: ISO-NE, CAISO and MISO. The results of the model were compared with relevant statistical methods and the impact of various factors on energy arbitrage revenue was investigated. Finally, the potential impact of increasing share of wind generation on electricity arbitrage potential in the MISO market throughout the years 2008–2013 was investigated. It was found that for a storage system with 0.1 hours of storage capacity theaverage consecutive price difference isthebeststatisticalindicatorofthe capturable arbitrage revenue potential. For storage capacities larger than 0.1 hours, the average intra-day price difference property introduced in this thesis was found to be the best indicator. This property corresponds to the summed up difference between a specified number of hours with the highest prices and those with the lowest prices, where the number of hours are spec- ified flexibly, taking into account the storage capacity and the round-trip ef- ficiencyofthesystem. Moreover, itwasfoundthatlowpriceautocorrelation at a node can indicate that a higher percentage of the available maximum revenue can be captured by a small storage capacity (0.1–0.5 h) system. It was also observed that storage devices with high self-discharge rates tend to waste more of the arbitrage potential in percentage when placed at nodes i ii withhighpriceautocorrelation. Regarding theimpactofwind, MISOnodes locatedatstateswithhighwindpenetrationwerefoundtohaveconsistently higher energy arbitrage revenues throughout the same period compared to other nodes. Acknowledgements Foremost, I would like to express my sincere thanks to my supervisor Line Roaldforhercontinuoussupportthroughoutthisthesis. Thisprojectwould have been much harder to materialize without her guidance, precious feed- back and her overall positive approach to problem solving. I also would like to thank Dr. Johanna L. Mathieu for the fruitful discussions and her help- ful suggestions throughout the thesis. Additionally, I would like to express my gratitude to Prof. Dr. G¨oran Andersson for being my tutor and giving me the opportunity to write my thesis at the Power Systems Laboratory. I would also like to thank my dear friends Vassilis Saplamidis and Iason Avramiotis for taking part in my brainstorming sessions and for their useful tips. Finally, I would like to thank my family for their endless support and encouragement. iii Contents 1 Introduction 1 1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.3 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 Energy Arbitrage Theory 3 2.1 Energy Storage Systems . . . . . . . . . . . . . . . . . . . . . 3 2.2 Demand Response . . . . . . . . . . . . . . . . . . . . . . . . 4 3 Energy Arbitrage Model 6 3.1 Storage Modeling . . . . . . . . . . . . . . . . . . . . . . . . . 6 3.2 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 3.3 Formulation of the Optimization Problem . . . . . . . . . . . 7 3.4 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 4 Analyzed Markets and Data Sets 11 4.1 Introduction to Locational Marginal Pricing and U.S. elec- tricity markets . . . . . . . . . . . . . . . . . . . . . . . . . . 11 4.2 ISO New England (ISO-NE) . . . . . . . . . . . . . . . . . . . 12 4.3 California ISO (CAISO) . . . . . . . . . . . . . . . . . . . . . 14 4.4 Midcontinent ISO (MISO) . . . . . . . . . . . . . . . . . . . . 15 4.5 Data-Sets Comparison . . . . . . . . . . . . . . . . . . . . . . 17 5 Analysis of Storage Parameters 19 5.1 Energy Storage Capacity . . . . . . . . . . . . . . . . . . . . . 19 5.2 Round-trip Efficiency . . . . . . . . . . . . . . . . . . . . . . . 20 5.3 Self-discharge Rate . . . . . . . . . . . . . . . . . . . . . . . . 21 6 Statistical Analysis and Results 23 6.1 Investigated Statistical Properties. . . . . . . . . . . . . . . . 23 6.1.1 Standard Deviation . . . . . . . . . . . . . . . . . . . 24 6.1.2 Coefficient of Variation . . . . . . . . . . . . . . . . . 24 6.1.3 Kurtosis . . . . . . . . . . . . . . . . . . . . . . . . . . 24 6.1.4 Autocorrelation . . . . . . . . . . . . . . . . . . . . . . 25 iv CONTENTS v 6.1.5 Average Intra-Day Price Difference . . . . . . . . . . . 25 6.1.6 Average Consecutive Price Difference . . . . . . . . . 27 6.2 Results of the Statistical Analysis . . . . . . . . . . . . . . . . 28 6.2.1 Case Study Set-Up . . . . . . . . . . . . . . . . . . . . 28 6.2.2 Price Statistics in Different Markets . . . . . . . . . . 29 6.2.3 CorrelationbetweenPriceStatisticsandArbitrageRev- enue . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 6.2.4 Correlation between Price Differences and Arbitrage Revenue . . . . . . . . . . . . . . . . . . . . . . . . . . 30 6.3 Effect of Autocorrelation . . . . . . . . . . . . . . . . . . . . . 34 6.3.1 Autocorrelation and the percentage of total arbitrage potential captured . . . . . . . . . . . . . . . . . . . . 34 6.3.2 Autocorrelationimpactondeviceswithhighself-discharge rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 6.4 Detecting Cyclic Patterns with Fourier Analysis . . . . . . . . 38 7 Wind Power Effect on Energy Arbitrage 42 7.1 Wind Power in the U.S. and the Analyzed Regions . . . . . . 42 7.2 Wind Power Penetration and Affected Price Statistics . . . . 43 7.3 Wind Power Impact on Energy Arbitrage Potential in MISO 45 8 Conclusions and Outlook 49 8.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 8.2 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . 50 Bibliography 51 List of Figures 3.1 Illustration of a generic energy storage system and its inter- action with the grid . . . . . . . . . . . . . . . . . . . . . . . 9 3.2 A sample daily operation of a storage system with 5 hours of storage at CAISO ARMSTRNG node . . . . . . . . . . . . . 10 4.1 ISO-NE operation areas [21] . . . . . . . . . . . . . . . . . . . 12 4.2 ISO-NE generation fuel-mix in 2000 and 2013. Plotted with data from [22] . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 4.3 NaturalGaspriceimpactonreal-timeLMPsinISO-NE.Plot- ted with data from [22] . . . . . . . . . . . . . . . . . . . . . 13 4.4 CAISO map and the three main trading hubs [24] . . . . . . 14 4.5 CAISOgenerationfuel-mixin2001and2013(calculatedfrom gross energy generation values from [26]) . . . . . . . . . . . 15 4.6 MISO regions and the major hubs [29] . . . . . . . . . . . . . 16 4.7 MISO generation fuel-mix in 2009 and 2013. Plotted with data from [30] . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 4.8 Marginal fuels in MISO in the years 2010–2013 [32] . . . . . . 17 5.1 Total yearly energy arbitrage revenue in 2013 with varying energy storage capacity in hours (E /P ), using hourly max max and 5-minute interval LMPs . . . . . . . . . . . . . . . . . . . 20 5.2 Total yearly energy arbitrage revenue in 2013 with varying round-trip efficiency for the cases of 0.1 and 12 hours storage capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 5.3 Total yearly energy arbitrage revenue in 2013 with varying self-discharge rate for the cases of 0.1 and 12 hours storage capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 6.1 Weekly revenue for the CAISO, MISO and ISO-NE nodes plotted against the average daily standard deviation for the corresponding week for a storage capacity of 0.5 h . . . . . . 31 6.2 Realized energy arbitrage revenues for a storage capacity of 0.1hoursandtheadditionalarbitragepotentialavailablewith 1 and 9 hours of storage options . . . . . . . . . . . . . . . . 35 vi LIST OF FIGURES vii 6.3 Weekly captured percentage of total arbitrage potential ver- sustheaveragedailyautocorrelationfortherespectiveweeks, with a storage capacity of 0.1 h . . . . . . . . . . . . . . . . . 35 6.4 Percentage of the weekly captured revenue potential plotted against average daily autocorrelation a) for a self-discharge rate of 0.2%, b) for a self-discharge rate of 1% . . . . . . . . 37 6.5 FFTplotfortheCAISOARMSTRNGnode,withafrequency unit of [1/day] . . . . . . . . . . . . . . . . . . . . . . . . . . 39 6.6 Results of the FFT analysis for the investigated nodes; intra- day cycle amplitude and the arbitrage revenue for 9 hours . . 39 6.7 Total yearly arbitrage revenue for the 1 h and 9 h of storage capacity cases versus the intra-day cycle amplitude for the 19 nodes considered in this analysis . . . . . . . . . . . . . . . . 41 6.8 Total yearly arbitrage revenue for the 1 h and 9 h of storage capacity cases versus the inter-day cycle amplitude for the 19 nodes considered in this analysis . . . . . . . . . . . . . . . . 41 7.1 Yearly additional wind capacity installed; 2014 value covers the capacity installed only during the first quarter. Data from [37] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 7.2 Standard deviation plotted against the mean for the nodes analyzed; note that blue indicates ISONE nodes, red mark- ers stand for CAISO and yellow markers represent the MISO nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 7.3 Coefficientofvariationplottedagainstthemeanforthenodes analyzed; note that blue indicates ISONE nodes, red mark- ers stand for CAISO and yellow markers represent the MISO nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 7.4 Normalizedarbitragerevenueinyears2008-2013plottedagainst wind penetration for various MISO nodes . . . . . . . . . . . 48 List of Tables 4.1 Percentage of positive spikes, negative spikes, negative price and zero price occurrences in the data-sets considered (for 2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 6.1 Statistical values for the investigated transmission regions . . 29 6.2 Correlations of the respective statistical properties with the weekly arbitrage revenue for varying storage sizes . . . . . . . 30 6.3 Correlations of average consecutive price differences (Cons. Diff.),andaverageintra-daypricedifferences forvarioushours oft withtheweeklyenergyarbitragerevenuefortherespec- H tive storage sizes . . . . . . . . . . . . . . . . . . . . . . . . . 33 6.4 Correlation of captured arbitrage potential and autocorrela- tion for varying storage capacities . . . . . . . . . . . . . . . . 36 6.5 Average inter- and intra-day cycle amplitudes for the investi- gated markets . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 7.1 The states where the analyzed nodes are located and the corresponding installed wind power capacity as well as wind power share in generation . . . . . . . . . . . . . . . . . . . . 44 viii Chapter 1 Introduction 1.1 Motivation Wholesaleelectricitypricesareexpectedtobecomeincreasinglymorevolatile due to the growing share of intermittent energy sources such as wind energy in the total electricity generation [1], [2]. This new market structure is likely to provide for more suitable circumstances for energy arbitrage possibilities, i.e. buying and storing energy during low price periods and selling it back when the prices are sufficiently high. Due to this increased price volatil- ity phenomenon combined with further advances in storage technologies it is expected that energy arbitrage will have an increased significance in the wholesale electricity markets in the near future. Electricity markets with nodal pricing are generally more suitable for energy arbitrage since various individual nodes are likely to exhibit excep- tional volatility compared to wholesale power markets with uniform pricing. Furthermore, many nodal markets operate real-time markets with 5-minute intervals, which are ideally better candidates for energy arbitrage due to in- creased volatility. Therefore, in this thesis North American wholesale elec- tricity markets employing nodal pricing with 5-minute interval real markets are investigated. 1.2 Goal To maximize the revenue potential obtained through energy arbitrage, it is important to install the systems for energy arbitrage at locations with high arbitrage potential. For this reason, identifying the most suitable intercon- nection as well as the individual nodes with the highest arbitrage potential is of utmost importance. In previous literature, the focus has been generally on assessing the ar- bitrage potential of various storage technologies at individual nodes by con- ducting extensive optimization calculations. Relevant price statistics indi- 1

Description:
EEH – Power Systems Laboratory. Swiss Federal Institute of In this context, the impact of the following ESS properties on the captured energy
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.