by Paul Pilotte and Ameya Deoras, MathWorks
Wind farm operators selling their power into wholesale electricity markets are looking for ways to maximize their profit and reduce revenue uncertainty.
Like any energy producer, they need software tools to analyze historical data, forecast prices and determine demand on the forward-reserve, day-ahead and real-time spot markets.
Unlike traditional energy producers, however, they also must analyze and forecast difficult-to-predict wind and weather patterns. This type of analysis must consider seasonality, temperature, time of day and directional wind speed.
Managing a wind farm portfolio is a promising renewable energy business, but operators face risks and challenges in a market geared toward predictable, fossil fuel-based generation. Software modeling tools allow better management of wind portfolio risks and financial uncertainty.
Wholesale Electricity Markets
Electricity producers, including wind farm operators, sell electricity through the regional wholesale electricity market or by entering bilateral contracts with utilities. In North America, nine regional transmission organizations (RTOs) administer regional competitive wholesale markets, which set the market clearing price—known as the locational marginal price (LMP)—for electricity at specific locations and delivery times. Trading markets typically include a day-ahead market where hourly LMPs are calculated for the next operating day by clearing generation offers and demand bids and a real-time market where spot LMPs are calculated every five minutes based on real-time grid demand.
Why Wind Power is Different
Wind farm operators must maximize revenue from wind-generated electricity. Wind is unpredictable, often stronger during off-peak hours when demand is low, and difficult to dispatch. The challenge is to find the optimal portfolio for selling energy in the spot market, day-ahead market and long-term contracts. Selling in the spot market has no energy production uncertainty, but operators must accept the prevailing market price.
Selling in the day-ahead market might net a better price, but it requires accurate, next-day price forecasting or hedging activities. Federal Energy Regulatory Commission (FERC)-imposed energy imbalance tariffs issued in 2007 can cut portfolio revenues and increase portfolio volatility. These tariffs apply to imbalances exceeding 1.5 percent of the scheduled energy (for over-delivery or under-delivery). A third alternative is entrance into long-term, bilateral contracts. Here, too, price and wind production forecasts are important tools in anchoring the value and risk for these contracts or hedging strategies.
|The challenge is to find the optimal portfolio for selling energy in the spot market, day-ahead market and long-term contracts. Selling in the spot market has no energy production uncertainty, but operators must accept the prevailing market price.|
Wind Price, Generation Forecasting
The two key components to forecast are the amount of wind energy produced and the energy market price for each available timeframe. These often require numerous forecast points. For example, real-time energy markets issue LMPs every five minutes, or about 8,760 per month for each location. This requires a software system that quickly can perform multiple numerical simulations and is flexible enough to allow a wind farm operator to customize algorithms based on the farm’s wind assets.
The work flow for energy pricing and production forecasting has three main phases: access, explore and discover, and share (see Figure 1).
MATLAB supports this work flow by combining a powerful numeric engine and technical programming environment with interactive modeling, visualization and deployment tools. First, the relevant data sources used for the forecasting model are accessed from historical databases, turbine energy output files and data feeds.
Second, data analysis and visualization tools are used to understand how wind energy production varies by location, weather conditions, time of day and time of year. This can help analysts uncover key relationships and evaluate statistical methods.
Third, results are shared via reports and custom applications.
In the exploration and discovery phase, the data analyst can employ statistical tools to estimate accurately the energy produced per wind turbine. Common choices include autoregressive models and neural networks. An important forecasting environment consideration is the ability to take one or more off-the-shelf statistical models, quickly implement and test them against historical data, and validate them against other models.
Bootstrap aggregation, a type of ensemble learning, also can improve accuracy. Figure 2 shows an example using bootstrap aggregated regression trees. This example was trained with three years of historical data from ISO New England including LMP, temperature and dew point.
The model accurately captures the relationship among the system load and weather, day of the week, hour of the day and holidays to produce next-day forecasts accurate to within 3 percent of the observed load.
Similar models can help develop day-ahead and longer-term forecasts for energy produced, market price and projected revenue. In customizable environments, other variables such as the penalty for over- and under-production can be included to help operators evaluate scenarios for selling their generated power.
The third phase in the work flow is sharing the results. This may include publishing reports as documents or on the Web, deploying forecasting applications or integrating algorithms into production systems. Automating manual steps in this phase can reduce errors and free analysts for strategy and decision-making activities.
UniÃ³n Fenosa, the fourth-largest Spanish energy company, successfully uses a technical computing environment to project energy production capacity and electricity demand to optimize a portfolio of generation assets. The company’s diverse portfolio includes coal, combined-cycle gas, nuclear, wind and hydroelectric power. UniÃ³n Fenosa analysts developed forecasting models in MATLAB to project energy production capacity and demand, which enabled them to optimize their asset portfolio. The models incorporate historical usage patterns, weather forecasts, production costs and other factors. For wind farm modeling, they used MATLAB to correlate historical wind strength measurements with actual electricity production, first using simple linear correlation and then using other advanced statistical models to enhance the predictive accuracy. To maximize the potential value of custom statistical models, they must be deployed across the organization. In UniÃ³n Fenosa’s case, this means sharing the results with analysts who optimize production resources, minimize production costs and verify infrastructure capacity. Programs are deployed as standalone applications that run automatically day and night. For UniÃ³n Fenosa, a key benefit of this approach is the flexibility with which it can develop, update and optimize its forecasting models in response to changing needs.
Wind Farm Risk Analysis Tools
Wind farm operators also can use risk analysis to better manage the price and volume volatility of wind generation portfolios and minimize the financial risk this volatility represents to the enterprise, customers and shareholders. Weather variations cause volume volatility, and the potential of negative prices and tariffs for under-delivery and over-delivery contribute to price volatilities.
Automated risk-forecasting systems, such as the one developed by Horizon Wind Energy, can quantify value at risk, revenue at risk and projected revenue for a wind farm portfolio. Typical systems link price forecasts to volumetric forecasts for the wind power generated for each wind farm. Horizon’s system replaced a complex network of calculations, which used 15 spreadsheets, some with as many as 500,000 rows. Monthly price forecasts for more than 20 of its wind farm sites are produced using historical data from market prices, daily forward contracts and option implied volatility. Wind power production is forecasted for each site using statistical distributions to fit local historical wind-level data. The results form part of a risk management process Horizon uses to hedge its exposure, saving millions of dollars through better price and congestion management.
Software Modeling, Forecasting Tools
Wind farm operators may choose from several software modeling and forecasting approaches. These include informal spreadsheets such as those created in Excel, turnkey trading and risk management systems and technical computing languages and environments. The UniÃ³n Fenosa and Horizon Wind case studies illustrate how wind production and price forecasting were automated using software tools that let operators apply their own experience and insights. The tools allow analysts to enhance accuracy by prototyping statistical modeling methods, quickly adapting models in response to changing needs or regulations, and sharing applications with other analysts and operations managers.
Wind power and energy trading forecasting tools are essential for wind farm operators, but these same modeling and forecasting tools may be used extensively by other organizations in the energy market, including electricity producers, electric utilities, power marketers and RTOs.
Horizon Wind Energy Develops Revenue Forecasting, Risk Analysis Tools for Wind Farms
Seeking to forecast revenue and quantify risk for wind farms across multiple geographic locations, Horizon Wind Energy used MATLAB to develop and deploy an automated production system that analyzes historical, current and forward-looking price and wind-level data.
“By creating standalone operational programs and running them automatically, we can provide up-to-date forecasts and projections to Horizon analysts on a daily basis,” said Manuel Arancibia, Horizon market operations manager. “The results raised awareness of the risk magnitude within our wind generation portfolio. That led the company early on to adopt a risk management process and hedge our exposure, which resulted in better price and energy congestion management at Horizon.”
Paul Pilotte is a technical marketing manager at MathWorks.
Ameya Deoras is an application engineer at MathWorks.