4 Presentations

• 
  01:30 PM - 01:55 PM

  Generation Planning of Renewable Energy Systems Using Stochastic Dual Dynamic Programming: A Case Study on Quebec's Energy Network

  • Soroosh Sabaei, presenter, PhD student
  • Michel Denault, Professor, HEC Montreal
  • Pierre-Olivier Pineau, Professor, HEC Montreal

  Our research focuses on scheduling electricity generation of hydro and wind power utilizing a multi-stage stochastic optimization approach since the variability of inflows and wind is a major concern. Using Quebec's hydro and wind power networks as a case study, we investigate long-term reservoir management strategies to meet the demand while minimizing the total costs. We applied a state-of-the-art technique Stochastic Dual Dynamic Programming (SDDP) to find the optimal generation policy. We also compare the results with a deterministic policy that utilizes the expected values to find the optimal policy. The last part of our analysis is a numerical investigation of expanded generation capacity. In particular, we examine the trade-offs between the costs of new capacity construction and the reduction of load shedding and power purchasing costs.

• 
  01:55 PM - 02:20 PM

  Energy, power and cost impacts of different electrification pathways

  • Florian Mitjana, presenter, HEC Montréal
  • Michel Denault, HEC Montréal
  • Pierre-Olivier Pineau, HEC Montréal

  The challenge of decarbonization calls for the electrification of many sectors of the economy, leading to profound changes in the electricity sector and a growing demand for electricity.

  A load profile generator has been developed to simulate consumption in the building, industry and transport sectors. Various parameters are considered, such as population size, building surface area, space heating and cooling characteristics, heat pump penetration, and the number and consumption of electric vehicles.

  A first scenario simulating the complete electrification of all Québec sectors by 2050 shows a 2-fold increase in peak load and a 1.5-fold increase in total load compared with 2020 levels. This profile is then incorporated into our capacity expansion and dispatch model. It highlights the significant need for additional capacity, the importance of load shedding during peak loads, and the tripling of provincial power system costs compared to the 2020 system.

  Finally, two scenarios involving improved building insulation and a reduction in floor space by 2050 highlight solutions that reduce peak load by 40% and total demand by 13% compared to the first scenario. System costs still increase by 70% compared to the 2020 system, but they correspond to a complete decarbonization of our energy system and therefore generate savings in other areas.

• 
  02:20 PM - 02:45 PM

  Hierarchical optimization of cooperative distributed energy resource aggregations in power distribution systems

  • Loreley Sepho, presenter, Polytechnique Montreal
  • Miguel F. Anjos, GERAD, Polytechnique Montréal
  • Amira Dems, Institut de recherche d’Hydro-Québec
  • Antoine Lesage-Landry, Polytechnique Montréal
  • Hanane Dagdougui, Ecole Polytechnique de Montréal

  The rise in incentives has increased the prevalence of distributed energy resources (DERs), posing both challenges and opportunities for electric grid. We propose a cooperative approach that optimizes individual loads to provide on-demand flexibility to the distribution system operator (DSO). This work introduces a hierarchical optimization strategy, firstly with a day-ahead phase to coordinate loads and set individual setpoints with the objective of adjusting the aggregated consumption to follow DSO regulation signal. The regulation signal represents the optimal aggregation consumption, enabling the DSO to ensure efficient network management. Uncertainties are accounted through chance constraints at the day-ahead level to mitigate the inherent unpredictability of DER, customer baseload, and DSO setpoint. Power flow constraints are integrated into the day-ahead level to ensure that the physical network limits are not exceeded. Secondly, model predictive control (MPC) is employed during the real-time phase to track the individual setpoint while subject to limited local information and uncertainties. Finally, a cooperative game theoretic mechanism utilizing the marginal contribution is proposed, to provide a scalable approximative for the fair and efficient distribution of aggregation payoffs. The approach is numerically implemented to demonstrate its ability to provide flexibility to modern power systems.

• 
  02:45 PM - 03:10 PM

  A 3-spaces dynamic programming heuristic for the Cubic Knapsack Problem (CKP)

  • Ibrahim Dan Dije, presenter, UQAM

  The Cubic Knapsack Problem (CKP) is respectively a generalization of Quadratic Knapsack Problem (QKP) and Knapsack Problem (KP). CKP is known as a combinatorial optimization problem in which, a cubic function of binary variables is maximizing subject to one dimension knapsack constraint. It has many applications in biology, project selection, capital budgeting problem, etc. The CKP is a NP-hard problem in the strong sense, because it's a generalization of QKP that is NP-hard, then there is no polynomial exact algorithm that can solve it. In this work, we introduced a 3-spaces dynamic programming heuristic that explore the linear space variables, quadratic space and cubic space to solve CKP by finding a near optimal solution at each space. The solution of dynamic programming heuristic is compared with an exact solution and another heuristic find in the literature.