Optimal scheduling of spinning reserve

Effective control of spinning reserve can provide substantial cost reductions in large power systems. This paper suggests that this goal can he achieved by tightly integrating a probabilistic reserve assessment with the unit commitment (UC) function. This enables the reliability of individual units and the uncertainty of the load forecast to be considered for short-term scheduling. The generation can then be scheduled to meet a given risk index. It is further argued that the optimal value of this risk index should be selected on the basis of a tradeoff between the total schedule cost obtained from the UC solution and the expected cost of energy not served which can be derived from the probabilistic reserve assessment. Finally, the problem of over-commitment of reserve in Lagrangian relaxation based UC is reduced by means of a new unit decommitment technique incorporating a look-ahead feature.