Risk-averse stochastic model predictive control-based real-time operation method for a wind energy generation system supported by a pumped hydro storage unit


Yildiran U., Kayahan I.

APPLIED ENERGY, vol.226, pp.631-643, 2018 (Peer-Reviewed Journal) identifier identifier

  • Publication Type: Article / Article
  • Volume: 226
  • Publication Date: 2018
  • Doi Number: 10.1016/j.apenergy.2018.05.130
  • Journal Name: APPLIED ENERGY
  • Journal Indexes: Science Citation Index Expanded, Scopus
  • Page Numbers: pp.631-643
  • Keywords: Stochastic model predictive control, Real-time operation, Day-ahead bidding, Wind energy, Pumped hydro storage, Conditional value at risk, VIRTUAL POWER-PLANTS, ELECTRICITY MARKET, PROBABILISTIC FORECASTS, OFFERING STRATEGIES, INTEGRATED APPROACH, RESERVE MARKETS, OPTIMIZATION, SPEED, FARM, PROFIT

Abstract

A wind energy producer participating in deregulated markets needs to make contracts on the energy it will supply in the next day. Deviations from the contracts, which could occur due to wind uncertainties, are compensated in real-time balancing markets at a considerable cost. Therefore, developing advanced day-ahead bidding and real-time operation strategies minimizing such imbalance costs constitutes an important problem. There are several works on finding optimal day-ahead bids but the real-time operation problem is not studied well. Motivated by this fact, we propose a new strategy in which the day-ahead bids are computed by solving a risk-averse stochastic program, and real-time operation is performed by a stochastic model predictive control based algorithm with a risk control capability. The algorithm is applied to a realistic system composed of wind farms and a pumped hydro storage plant. Its performance is compared to a number of approaches appearing in the literature. Because the problem considered has two conflicting objectives of profit maximization and risk minimization, a Pareto optimality analysis is also conducted. Finally, the validity of a common practice followed in the literature, which is estimating the economic performance by bidding optimization, is investigated by comparing the estimate with the actual performance achieved by real-time operation methods.