Back

G-2024-46

Improved generalized Benders decomposition for stochastic unit commitment models with demand response

, , and

BibTeX reference

The increasing penetration of renewable electricity generation as well as the implementation of demand response programs has led to new challenges in the operation of the power grid. The output of renewable sources fluctuates, and this adds uncertainties to the problem. The distributed nature of the demand response resources is an additional operational challenge that is normally addressed by the creation of aggregators that manage these resources. The impacts of the power transmission system must also be taken into account. We propose a short-term unit commitment model to allocate demand response resources considering the variability of renewable sources and the needs of the grid. We formulate this as a mixed nonlinear integer optimization problem that is challenging to solve, which motivates, first, to relax the problem by applying a semidefinite relaxation to it, and, second, the use of Generalized Benders Decomposition (GBD) to tackle it. It is well known that the GBD algorithm can suffer from slow convergence to an optimal solution, therefore we use a Benders-based Branch-and-Cut with various enhancement methods to improve its performance. In order to choose which enhancement methods should be used, we analyze their impact on the performance of the GBD algorithm using the IEEE RTS-96 network. We conclude that while all of the enhancement methods considered improve the convergence rate and solution time for our model, the Pareto-Optimal cuts are the most significant improvement, both in terms of convergence rate and computational time.

, 16 pages

Research Axes

Research application

Document

G2446.pdf (400 KB)