SCRO / Journées de l'optimisation

HEC Montréal, 29-31 mai 2023

CORS-JOPT2023

HEC Montréal, 29 — 31 mai 2023

Horaire Auteurs Mon horaire

MMESI Methods and Models for the Energy Sector I

31 mai 2023 10h30 – 12h10

Salle: TAL Gestion globale d'actifs inc. (vert)

Présidée par Seyyedreza Madani

4 présentations

  • 10h30 - 10h55

    Prosumer-Based Residential Electricity Trading with Blockchain

    • Phil Zheng, prés., UCF

    The fast-growing number of electricity prosumers and the rise of blockchain technology have together bestowed a unique opportunity to study a new paradigm of electricity transacting and trading in local- or micro-grid. A major way for decarbonization to fight climate is distributed power generation, especially via household solar photovoltaic panels, which inadvertently transforms traditional electricity consumer to prosumers. Thereafter, it fosters a growing need to design a fair and efficient scheme for prosumers to transact and trade electricity. Coincidentally, at the same time blockchain technology has been prospering ever since the seminal work by Satoshi Nakamoto, i.e., the creator of Bitcoin. This talk is directly responding to this timely and critical research opportunity by discussing two important aspects: (1) building the theoretical foundation by using game theory models for the users’ decisions to buy or sell power in the peer-to-peer energy market and (2) designing and implementing blockchain software stack for deploying such a system. The mathematical proof-of-concept studies show that such local trades provide benefits to both electricity prosumers. In addition, blockchain peer-to-peer platform implementation of the trading scheme shows that it can be managed without a central electricity exchange.

  • 10h55 - 11h20

    Robust Optimal Sizing of a Stand-alone Hybrid Renewable Energy System Using Machine Learning

    • Ali Keyvandarian, prés., Dalhousie University
    • Ahmed Saif, Dalhousie University

    This study provides an adaptive robust method for optimally sizing a hybrid renewable energy system (HRES) consisting of wind turbines, solar photovoltaic panels, a battery bank, and a diesel generator. We employ vector auto-regressive models (VAR) and neural networks (NN) within dynamic uncertainty sets (DUSs) that account for the cross-correlation between uncertain parameters as well as their temporal auto-correlations in the wind and solar energy outputs. The constructed DUSs are compared to the independent DUSs used by robust HRES scaling models to capture the unpredictability of renewable energy sources. A column-and-constraint generation-based iterative technique is applied to efficiently handle the adaptive robust problem. A genuine case study of an isolated community in northern Ontario, Canada is used to evaluate the suggested concept and solution approach. By simulating the system operation using real test data, the superiority of the proposed approach over an ARO with an independent DUS model and a nominal model that uses point estimates of the parameters is demonstrated, highlighting the need to account for the cross-correlations as well as auto-correlations while considering supply and demand uncertainty.

  • 11h20 - 11h45

    Impact of Climate Change on Battery Electrical Energy Storage: A case study of MISO

    • Wu Zhenggao, prés., University of Waterloo
    • Dimitrov Stan, University of Waterloo
    • Pavlin Michael, Wilfrid Laurier University

    Climate change will have multiple impacts on energy markets by changing demand patterns and changing supply patterns --- by indirectly increasing renewable penetration and changing the dynamics of the supply from renewable sources (e.g., when and where it is windy/cloudy). Battery energy storage systems (BESS) are one way to effectively manage intermittency from renewable energy. However, their profitability and incentive to participate in markets are susceptible to both the magnitude and frequency of price variation. This paper investigates the impact of climate change on a BESS operating in a North American deregulated electricity market. We propose a robust optimization model to determine the operating policy over a 70-year period of a BESS under different climate projections. We reformulate the robust optimization model to an equivalent linear program that allows us to numerically explore different climate scenarios. We conclude by discussing the results of our empirical study on the Midcontinent ISO market, U.S., and determining optimal operations of the BESS using our proposed model. Based on the optimal operations, we investigate (1) how the evolution of climate change will impact the operation of the BESS over a long-time horizon, (2) how the realization of different climate scenarios influences the operation of the BESS, and (3) how the profitability of the BESS varies across different geographical locations.

  • 11h45 - 12h10

    Targeted Dynamic Electricity Pricing Using Deep Reinforcement Learning

    • Seyyedreza Madani, prés., HEC Montreal
    • Pierre-Olivier Pineau, HEC Montreal

    Peak loads impose the highest investment cost on the electricity networks. To secure the availability of sustainable electricity for all, demand response strategies are among the most recommended tools on the end user side, as they can effectively shave peaks. This study analyses different types of electricity consumers and investigates the effect of profile characteristics on the effectiveness of demand response. Moreover, using real-world records from Quebec, this study considers the interactions between the distributor, active prosumers, and passive consumers, identifies their optimal strategies, and, calculates their respective costs and revenues. Moreover, since mass peak load shaving offers are prone to create new undesired peak loads, this study introduces a dueling double deep Q-network to propose a novel targeted pricing strategy for the distributor and makes the distributor able to control the amount of the shifted load. Furthermore, the effect of different penetration levels of the DER in the network is investigated and some policy-making insights are offered for future smart grids.

Retour