Optimization of peak loads among multiple provincial power grids under a central dispatching authority

• The study comes from the real demand of operations in Chinese regional power grids.
• An optimization model of peak loads among multiple provincial power grids under a CDA is developed.
• The model is solved via convex quadratic optimization and applied to the ECG.
• The model can coordinate generations among provinces to meet different peak demands.
• The model produces more rational generation schedules than the conventional method.

There is a lack of the capacity to respond to peak loads in most provincial power grids of eastern China and coastal regions. A CDA (central dispatching authority), which is usually a regional power grid, is responsible to dispatch its own plants and allocate power generation to multiple subordinate provincial power grids for responding to their peak loads simultaneously. Hence, this paper develops an optimization model to determine the quarter-hourly generation schedules allocated for provincial power grids under a CDA. To meet the need for peak shaving, the selected objective function in the model requires minimization of the variance of remaining load that is obtained by subtracting power generation from the original load of each provincial power grid. The objectives of multiple power grids are first reduced to an equivalent scalar objective through the weighted sum method. The general quadratic and linear formulations applying to the conventional plants and pumped-storage plants are developed to respectively deal with complex composite objective and spatial-temporal coupling constraints such as multilateral electricity contracts constraints and load balance constraints. The resulting problem is finally solved via the convex quadratic optimization technique. The proposed method is applied to scheduling plants owned by the East China Grid which covers five provinces. The simulations show that the method can rationally coordinate electric power among provinces to meet different peak demands. The comparison with the conventional method further demonstrates the advantages of our method.