Scheduling of hydrothermal power systems using the augmented Lagrangian decomposition and coordination technique

When the standard Lagrangian relaxation method is used to solve hydrothermal scheduling problems, many subproblems have linear or piecewise linear cost functions. A well recognized difficulty is that the solutions to these subproblems may oscillate between maximum and minimum generations with slight changes of the multipliers. This may result in large differences between subproblem solutions and the optimal primal schedule. In this paper, the augmented Lagrangian decomposition and coordination technique (AL) is applied to the scheduling of a hydrothermal power system. By appending the quadratic penalty term to the Lagrangian, the AL method reduces the subproblem solution oscillations and results in a smoother dual function. Numerical results based on Northeast Utilities Service Company data show that the AL method speeds up algorithm convergence and requires less computation time.