Wind Power Scenario Generation and Reduction in Stochastic Programming Framework

Kailash Chand Sharma, Prerna Jain, Rohit Bhakar

Electric Power Components and Systems · 2013 · 128 citations · 13 references

Concepts

TL;DR

Wind power trading in pool‑based electricity markets is typically modeled with multi‑stage stochastic programming, but accurately representing the stochastic wind input requires many scenarios, making scenario‑reduction essential. The article proposes simplified algorithms for generating and reducing wind power scenarios. The algorithms use a time‑series ARMA model for scenario generation and a probability‑distance backward reduction technique. Implemented for next‑day scenario generation of a Barnstable, Massachusetts wind farm, the algorithms demonstrate effective wind‑uncertainty modeling and enable optimal wind‑power bids for market trading.

Abstract

Abstract Wind power trading in pool-based electricity markets is a decision-making problem and is generally modeled using a multi-stage stochastic programming approach because of the implicit uncertainty of wind input. In any stochastic programming approach, representation of random input process is a major issue. Due to uncertainty in wind availability, generated power by wind turbines is stochastic and is represented by possible values with corresponding probability of occurrence or scenarios. Accurate representation of uncertainty generally requires the consideration of large number of scenarios, thus necessitating the need for scenario-reduction techniques. This article presents simplified algorithms for wind power scenario generation and reduction. A time series based auto regressive moving average model is used for scenario generation, and probability distance based backward reduction is used for scenario reduction. The algorithms have been implemented for next-day scenario generation of wind farm located at Barnstable, Massachusetts, USA. The results prove the ability of the proposed algorithms in wind uncertainty modeling. These algorithms can successfully be utilized to generate optimal wind power bids for trading in electricity markets.

References

13