Demand and residual demand modelling using quantile regression

Abstract

Abstract. Residual demand, the difference between demand and solar and wind production, is an important variable in predicting the future price and storage requirements. However, little is known about predicting the residual demand itself as well as its quantiles. Therefore, we model both demand and residual demand using both ordinary and quantile regression and compare the results for the hourly electricity consumption in Germany. We find that the residual demand is less predictable than demand. The effect is visible for all hours, and is higher for the lower than the upper quantiles.

Minimum loads of coal‐fired power plants and the potential suitability for energy storage using the example of Germany

“…Due to the increasing amount of renewable energies in Germany and worldwide, the short‐ and long‐term demand for flexible power generation technologies is on the rise. The need for improved flexibility 1,2 is mainly caused by weather‐dependent generation technologies whose grid integration brings new challenges, 3 resulting in steeper and more variable residual loads 4,5 This leads to an increasing amount of low and negative prices at the wholesale markets, which results from nonresponding conventional power plants, especially coal‐fired power plants, during low residual loads causing an irreducible minimum of conventional feed‐in 6 . The reasons for the nonresponding behavior are technical, 7 economical, or stability limitations.…”

Impact of wind and solar production on electricity prices: Quantile regression approach

“…Hence, modelling the probability of extreme prices can be more important than the expected values (Bunn et al, 2016, Hagfors et al, 2016b. In this paper we analyze how electricity price Quantile regressions have been applied in financial risk management and recently in energy market studies: household energy consumption (Kaza, 2010), electricity demand (Do et al, 2016a;He et al, 2019), oil prices (Lee and Zeng, 2011) and CO2 emission allowance price (Hammoudeh et al, 2014). Quantile regression has been successfully applied also to electricity price forecasting, see Jónsson et al (2014), Nowotarski and Weron (2014), Nowotarski and Weron (2015), Juban et al (2016), , Moreira, Bessa and Gama (2016), , Bello et al (2017), Liu et al (2017), Mosquera-López et al (2017), Uniejewski, Marcjasz and Weron (2018).…”