Scenario creation and power-conditioning strategies for operating power grids with two-stage stochastic economic dispatch

Multi-Fidelity Stochastic Economic Dispatch for Operating Low Carbon Power Grids

“…As a note to the readers, the problems formulated here are for a single-period analysis. A multi-period analysis of the OPF problem is a simple extension and has been discussed in Reynolds et al (2020) and Panda et al (2021b). Additionally, while we use linear generator costs in the report, it is also possible to use quadratic cost functions in the OPF problem.…”

“…We use the scenario generation capability, described in Reynolds et al 2020, that uses information about the existing data for wind on the grid to inform scenarios for the OPF analysis, and briefly outline the method here. Defining a set of candidate scenarios as S = { s j ( w , t ) } N j = 1 , where t ∈ T denotes time, N is the number of available candidates,…”

“…Several techniques have been proposed for modeling the uncertainty arising in OPF problems including robust optimization (Ben-Tal, El Ghaoui, and Nemirovski 2009;Lorca and Sun 2015), chance constrained optimization (Ben-Tal, El Ghaoui, and Nemirovski 2009;Cheng et al 2018;Shapiro, Dentcheva, and Ruszczyński 2014;Liu and Xu 2010;Zhang, Gatsis, and Giannakis 2013), surrogate modeling (Safta et al 2017;Hu et al 2021;Wang et al 2022), and two-or multi-stage stochastic programming using scenarios (Shapiro, Dentcheva, and Ruszczyński 2014;Birge and Louveaux 2011;King et al 2018;Reynolds et al 2020;Sigler et al 2020). While each of these stochastic methods are viable in certain contexts and have different computational scaling characteristics, all of them can become computationally intractable when considering large interconnection regions, e.g., with O 10 4 buses.…”

Visualization of Multi-Fidelity Approximations of Stochastic Economic Dispatch

“…Multi-fidelity approximation methods [14,20,21] can be used for predicting the generation levels for stochastic economic dispatch and to facilitate rapid deployment of renewable energy without requiring large computational resources or sacrificing accuracy [18]. Our solution builds upon the twostage stochastic programming model for economic dispatch described in [19,23]. The first-stage of the problem optimizes costs for running thermal generation to match demand, considering a predicted amount of renewable energy on the grid, and errors expected in the renewable prediction, and is given by…”

“…We use a modified form of a power grid from the Reliability Test System -Grid Modernization Lab Consortium (RTS-GMLC) [2,7] which contains 73 buses, 104 transmission lines, 96 thermal generators, and 22 wind farms corresponding to 37.28% wind energy penetration [23]. Wind energy scenarios for these 22 generators are drawn from the WIND Toolkit [5] which contains historical high-fidelity wind data.…”