Balance costs for windpower

Abstract

Wind power scheduling is subject to high forecast uncertainty compared to the dominant Norwegian energy source, hydro power. The regulating market, also known as the balance market, is a market solution for optimal operation of the Nordic power system, putting a price on schedule deviations. Schedules, or production/consumption plans, are collected daily, 12 hours prior to delivery by the system operator Statnett. Wind power is especially vulnerable to balance costs due to the relatively high forecast uncertainty. A new feature of the balance market, known as two-price settlement, adds additional incentive for power producers to avoid schedule deviations by, in the end, inducing higher balance costs. This thesis work looks into the specifics of the balance market and the two-price settlement. Emphasis is put on comparing the new settlement system to the original to quantify consequences of the introduction for wind power. Also, an approach for combining short term wind power scheduling with optimal hydro power scheduling is suggested and implemented for analysis. Experiments are carried out using the hydro power plant Røssåga and Smøla wind farm, both property of Norwegian power corporation Statkraft. The purpose of the experiments is to quantify potential benefits of combined scheduling, here referred to as joint operation. The analysis setup is throughout this document referred to as the balance costs model. The main principle of the balance costs model is to compare the costs from balancing wind power alone with costs of including wind power balance power demand in optimal hydro power scheduling. Sintef Energy Research hydro optimization application SHOP represents an essential ingredient to the model, providing optimal hydro schedules for the study cases. In addition, Powel software Sim and SimServer is used for collection of data. Study cases are based on observed historical values and forecasted data from the year of 2007. The analysis does not give a clear indication of whether joint operation will induce a significant economic benefit, as results point in both directions. On average, results show an increase in revenue of no more than 1%. Taken into account that experiments are carried out with access to perfect information, the approach suggested in this thesis do not appear to have significant potential for short term power scheduling.