Asynchronous stochastic dual dynamic programming algorithm applied to hydrothermal coordination

Abstract

Optimizing power generation planning is very important to achieve the lowest possible costs while keeping an adequate trade-off with electrical network security. High computational resources are required to solve this problem, which is a multistage, stochastic, complex and large scale problem. In some cases, the use of parallel schemes becomes mandatory. A widely used method to solve long-term energy planning problems is an extension of Dual Dynamic Programming (DDP) called Stochastic Dual Dynamic Programming (SDDP) which makes use of sampling techniques to be able to deal with high-dimensional state-spaces. In this work we propose an asynchronous SDDP parallel scheme capable of overcoming the intrinsic synchronism of the traditional parallel version of the SDDP method, thus allowing to better exploit the parallel resources and decrease the overall CPU time to solve the problem. Consistency and performance tests were applied to evaluate the proposed asynchronous SDDP approach and one variant of this algorithm in a problem equivalent to the real Brazilian system, where was verified advantages over using the traditional parallel schema.