Adaptive mixed-integer programming unit commitment strategy for determining the value of forecasting

Erik Delarue

doi:10.1016/j.apenergy.2007.07.007

Return

Adaptive mixed-integer programming unit commitment strategy for determining the value of forecasting

2008-04-01

PRE

Erik Delarue

William D’haeseleer *

DOI：10.1016/j.apenergy.2007.07.007

Original

Original request for help

Save

Abstract

En 中文

This paper presents the development of a method to determine the value of forecasting (for load, wind power, etc.) in electricity-generation. An adaptive unit commitment (UC) strategy has been developed for this aim. An electricity generator faces demand with a given uncertainty. Forecasts are made to meet this load at the lowest cost. The adaptive UC strategy can be described as follows. Each hour, the generating company constructs a new forecast for a fixed number of hours. We assume that the first forecasted hour is in fact predicted correctly. For these forecasted hours, an optimal UC schedule is determined (given the on/off states of power plants for the current hour). The solution for the first hour (i.e., the one that was predicted correctly) is retained, and a new forecast is made. A 15,000 MW power system is used in a 168 hour (one-week) schedule. The UC problems presented in this work are solved through a Mixed-Integer Linear Programming (MILP) approach. In the first case, the effect of limited (correct) forecasting is investigated. Forecasts are made 100% correctly, but the UC scheme is built modularly and compared with the reference case, where the UC problem is solved for the one-week problem as a whole. Depending on the number of forecasted hours, solutions differ by up to 0.5% with the reference case. In a second case, when a certain error is imposed on the forecasts made (up to 5%), the deviations from the optimal solution become larger and amount in certain cases to almost 1%. (C) 2007 Elsevier Ltd. All rights reserved.

Keywords:

electricity-generation modelling

unit commitment (UC)

mixed-integer linear programming (MILP)

forecasting

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.