TY - JOUR
T1 - Stochastic security-constrained unit commitment
AU - Wu, Lei
AU - Shahidehpour, Mohammad
AU - Li, Tao
PY - 2007/5
Y1 - 2007/5
N2 - This paper presents a stochastic model for the long-term solution of security-constrained unit commitment (SCUC). The proposed approach could be used by vertically integrated utilities as well as the ISOs in electricity markets. In this model, random disturbances, such as outages of generation units and transmission lines as well as load forecasting inaccuracies, are modeled as scenario trees using the Monte Carlo simulation method. For dual optimization, coupling constraints among scenarios are relaxed and the optimization problem is decomposed into deterministic long-term SCUC subproblems. For each deterministic long-term SCUC, resource constraints represent fuel and emission constraints (in the case of vertically integrated utilities) and energy constraints (in the case of electricity markets). Lagrangian relaxation is used to decompose subproblems with long-term SCUC into tractable short-term MIP-based SCUC subproblems without resource constraints. Accordingly, penalty prices (Lagrangian multipliers) are signals to coordinate the master problem and small-scale subproblems. Computational requirements for solving scenario-based optimization models depend on the number of scenarios in which the objective is to minimize the weighted-average generation cost over the entire scenario tree. In large scale applications, the scenario reduction method is introduced for enhancing a tradeoff between calculation speed and accuracy of long-term SCUC solution. Numerical simulations indicate the effectiveness of the proposed approach for solving the stochastic security-constrained unit commitment.
AB - This paper presents a stochastic model for the long-term solution of security-constrained unit commitment (SCUC). The proposed approach could be used by vertically integrated utilities as well as the ISOs in electricity markets. In this model, random disturbances, such as outages of generation units and transmission lines as well as load forecasting inaccuracies, are modeled as scenario trees using the Monte Carlo simulation method. For dual optimization, coupling constraints among scenarios are relaxed and the optimization problem is decomposed into deterministic long-term SCUC subproblems. For each deterministic long-term SCUC, resource constraints represent fuel and emission constraints (in the case of vertically integrated utilities) and energy constraints (in the case of electricity markets). Lagrangian relaxation is used to decompose subproblems with long-term SCUC into tractable short-term MIP-based SCUC subproblems without resource constraints. Accordingly, penalty prices (Lagrangian multipliers) are signals to coordinate the master problem and small-scale subproblems. Computational requirements for solving scenario-based optimization models depend on the number of scenarios in which the objective is to minimize the weighted-average generation cost over the entire scenario tree. In large scale applications, the scenario reduction method is introduced for enhancing a tradeoff between calculation speed and accuracy of long-term SCUC solution. Numerical simulations indicate the effectiveness of the proposed approach for solving the stochastic security-constrained unit commitment.
KW - Lagrangian relaxation
KW - Mixed integer program
KW - Monte Carlo simulation
KW - Random power outages
KW - Scenario aggregation
KW - Security-constrained unit commitment
KW - Subgradient method
KW - Uncertainty
UR - http://www.scopus.com/inward/record.url?scp=34248512634&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=34248512634&partnerID=8YFLogxK
U2 - 10.1109/TPWRS.2007.894843
DO - 10.1109/TPWRS.2007.894843
M3 - Article
AN - SCOPUS:34248512634
SN - 0885-8950
VL - 22
SP - 800
EP - 811
JO - IEEE Transactions on Power Systems
JF - IEEE Transactions on Power Systems
IS - 2
ER -