CORS / Optimization Days
HEC Montréal, May 2931, 2023
CORSJOPT2023
HEC Montreal, 29 — 31 May 2023
OMMEP Optimization Methods for Modern Electric Power Systems
May 30, 2023 03:30 PM – 05:10 PM
Location: TAL Gestion globale d'actifs inc. (green)
Chaired by Antoine LesageLandry
4 Presentations

03:30 PM  03:55 PM
Uncertainty modelling for planning and operations of multimicrogrids
We present a Two Stage stochastic Programming (TSSP) model for the planning of MultiMicrogrids (MMGs) in Active Distribution Networks (ADNs) and an intervalbased method for their operations. The models aim to minimize the total costs while benefiting from interconnections of Microgrids (MGs), considering uncertainties associated with electricity demand and Renewable Energy Sources (RESs). In the planning model, the associated uncertainties are analyzed using Geometric Brownian Motion (GBM) and their associated probability distribution functions (pdfs). The planning model includes longterm purchase decisions and shortterm operational constraints, using Geographical information Systems (GIS) to realistically estimate rooftop solar limits, and is used to study the feasibility of implementing an MMG system consisting of 4 individual MGs at an ADN in a municipality in the state of Sao Paulo, Brazil. Using the same test system, the uncertainties in the operational model are formulated in an Affine Arithmetic (AA) domain to obtain an Energy Management System (EMS) model that is robust for a range of realizations of the uncertain parameters, with no need of statistical assumptions or repeated calculations, which can be solved with relatively low computational burden, as opposed to other approaches such as Monte Carlo Simulation (MCS).

03:55 PM  04:20 PM
Measurementbased Locational Marginal Pricing in Power Distribution Systems
Locational marginal pricing is widely used in wholesale electricity markets in the power transmission system, and it represents a promising solution to establish local electricity markets within distribution networks. We present a measurementbased method for calculating realtime distribution locational marginal prices (DLMPs) without the use of an offline network model. Instead, the proposed method relies only on online measurements collected at a subset of distribution system buses to estimate a linear sensitivity model mapping bus voltages to injections, which in turn is embedded in an optimal power flow (OPF) problem as an equality constraint. The proposed method completely obviates the need for an accurate distribution network model that may not be available, especially for active distribution networks with faster variations in operating point. Also, the proposed method renders the original OPF problem with nonlinear constraints a computationally efficient quadratic programming problem (with linear constraints) and provides sufficiently accurate DLMPs at buses where measurements are collected. Via numerical simulations involving a 33bus test system, we demonstrate that the proposed method yields similar DLMPs as solving the OPF problem with an uptodate model and greatly outperforms it when the model is out of date.

04:20 PM  04:45 PM
Featuredriven trading of wind power and hydrogen
This talk introduces a datadriven approach to support the decisions and enhance the profits of hybrid power plants trading wind energy, in forward and balancing electricity markets, and hydrogen, in bilateral contracts. The proposed approach leverages valuable explanatory variables (socalled features) of the uncertain parameters in a datadriven model. Given historical observations of the uncertain parameters (e.g. wind power forecast error) and the corresponding explanatory variables (e.g. wind power forecast values), we fit a piecewise linear policy to those data, which is optimized to maximize the profits of the hybrid power plant under constraints on its operation. We show on a realistic case study, using historical wind production data from a wind farm in Denmark, that the proposed featuredriven approach achieves significant economic gains compared to traditional approaches.

04:45 PM  05:10 PM
Online Dynamic Submodular Optimization for Power Systems
We propose new algorithms with provable performance for online combinatorial optimization in dynamic settings and subject to general binary constraints. We consider the subset of problems in which the objective function is submodular. First, we propose online submodular greedy algorithm (OSGA) which solves to optimality an approximation of the previous round loss function to avoid the NPhard original problem. We then extend OSGA to a generic approximation function. We show that OSGA have a dynamic regret bound that is of the same order as the tightest bounds in OCO. Then, for cases where no approximation exists or when an efficiencyoriented implementation is desired, we formulate the online submodular projected gradient descent (OSPGD) by leveraging the Lovász extension. We obtain a dynamic regret bound that is similar to the conventional online gradient descent. Finally, we numerically test our algorithms in two power system applications: realtime distribution network reconfiguration and fasttimescale demand response.