Applying Markov chains for the determination of the capacity credit of wind power

Abstract

Investments in wind power occur everywhere in the world. The value of these investments for integration in an electricity generation system cannot be determined in the same way as conventional electricity sources due to the variable and relative unpredictable nature of wind power. Wind power can only to some limited extend be centrally dispatched. To look at the long term value of investments in wind power, the term capacity credit can be used. It defines the level of conventional generation that can be replaced by wind power generation. Using four adequacy indices, namely Loss-of-load Expectancy (LOLE), Loss of Energy Expectation (LOEE), Loss-of-load Frequency (LOLF) and Expected Interruption Cost (EIC), the Peak load Carrying Capability (PLCC) is established for different sizes and locations of wind power in a system. The PLCC can be seen as a way to quantify the capacity credit of wind power since it determines how much the load can be increased for a given level of wind power investment, while maintaining the system reliability. The adequacy indices are found to vary depending on size and location of wind power investments, therefore causing the PLCC to change accordingly. A Monte Carlo approach is used for determining the indices. Expected and unexpected outages of system elements are simulated and evaluated against system load. Wind power data are generated through Markov chains, based on actual meteorological data from Belgian weather measurement sites, thereby preserving the same statistical properties as the original data.