Modeling long correlation times using additive binary Markov chains: applications to wind generation time series

TL;DR

This paper introduces an additive binary Markov chain model for wind power generation time series, effectively capturing temporal correlations and aiding in system backup and storage planning.

Contribution

The paper presents a novel application of additive binary Markov chains to model wind generation, accurately reproducing temporal correlations with minimal input data.

Findings

Model accurately reproduces wind power autocorrelation functions.

The approach estimates backup and storage needs effectively.

Temporal dynamics of wind events are well captured.

Abstract

Wind power generation exhibits a strong temporal variability, which is crucial for system integration in highly renewable power systems. Different methods exist to simulate wind power generation but they often cannot represent the crucial temporal fluctuations properly. We apply the concept of additive binary Markov chains to model a wind generation time series consisting of two states: periods of high and low wind generation. The only input parameter for this model is the empirical autocorrelation function. The two state model is readily extended to stochastically reproduce the actual generation per period. To evaluate the additive binary Markov chain method, we introduce a coarse model of the electric power system to derive backup and storage needs. We find that the temporal correlations of wind power generation, the backup need as a function of the storage capacity and the resting time distribution of high and low wind events for different shares of wind generation can be reconstructed.