A hybrid stochastic/interval approach to transmission-constrained unit commitment

Yury Dvorkin, Hrvoje Pandžić, Miguel A. Ortega-Vazquez, Daniel S. Kirschen

Research output: Contribution to journalArticlepeer-review


This paper proposes a new transmission-constrained unit commitment method that combines the cost-efficient but computationally demanding stochastic optimization and the expensive but tractable interval optimization techniques to manage uncertainty on the expected net load. The proposed hybrid unit commitment approach applies the stochastic formulation to the initial operating hours of the optimization horizon, during which the wind forecasts are more accurate, and then switches to the interval formulation for the remaining hours. The switching time is optimized to balance the cost of unhedged uncertainty from the stochastic unit commitment against the cost of the security premium of the interval unit commitment formulation. These hybrid, stochastic, and interval formulations are compared using Monte Carlo simulations on a modified 24-bus IEEE Reliability Test System. The results demonstrate that the proposed unit commitment formulation results in the least expensive day-ahead schedule among all formulations and can be solved in the same amount of time as a full stochastic unit commitment. However, if the range of the switching time is reduced, the hybrid formulation in the parallel computing implementation outperforms the stochastic formulation in terms of computing time.

Original languageEnglish (US)
Article number6847238
Pages (from-to)621-631
Number of pages11
JournalIEEE Transactions on Power Systems
Issue number2
StatePublished - Mar 1 2015


  • Interval optimization
  • stochastic optimization
  • uncertainty
  • unit commitment

ASJC Scopus subject areas

  • Energy Engineering and Power Technology
  • Electrical and Electronic Engineering


Dive into the research topics of 'A hybrid stochastic/interval approach to transmission-constrained unit commitment'. Together they form a unique fingerprint.

Cite this