Abstract
A method for solving the stochastic network constrained hydro-thermal scheduling problem based on the progressive hedging algorithm (PHA) is presented. The quadratic penalty terms in the PHA are dynamically linearized to retain a linear problem structure. Furthermore, a technique for efficient treatment of linear power flow constraints in scenario-based decomposition methods is proposed and tested. The method is applied to a case study, emphasizing on computational performance. It is demonstrated how the PHA convergence properties can be compared with those of the multi-stage Benders decomposition method by using bounds.