Stochastic Planning of Dual-provision Power-to-hydrogen Systems

Abstract

The transportation sector is a major source of global greenhouse gas (GHG) emissions, contributing 24% in Canada, 27% in the USA, and about a quarter in the EU in 2020. A promising solution is the adoption of battery electric vehicles (BEVs) and fuel cell electric vehicles (FCEVs) to replace internal combustion engine (ICE) vehicles and support the Paris Agreement’s 2050 net-zero target. FCEVs use hydrogen to power an onboard fuel cell, where hydrogen reacts with oxygen, to generate electricity for the motor. Hydrogen is commonly produced via water electrolysis, which splits water into hydrogen and oxygen using electricity. The resulting hydrogen can be stored and distributed to end users. Power-to-hydrogen (P2H) systems can be centralized—producing hydrogen at a hub and transporting it via liquid organic hydrogen carriers (LOHC)—or distributed, with stations generating and supplying hydrogen locally. While essential for large-scale FCEV adoption, relying solely on transportation demand may challenge the economic feasibility of P2H systems. Hence, providing ancillary services, in response to signals from the Independent System Operator (ISO), has been explored in the literature as a potential additional revenue source for these facilities. This thesis presents a novel stochastic planning framework for centralized dual-provision P2H systems that supply hydrogen to the transportation sector while providing ancillary services to the power grid. The framework mirrors the full project life cycle, from conceptual planning to grid interconnection, and aims to optimize the central LOHC hub location and size to ensure economic viability and attract investment, thereby advancing hydrogen infrastructure and supporting net-zero goals. The research includes four parts: 1) developing stochastic models capturing essential FCEVs characteristics based on actual data to be used in the planning framework, 2) developing a stochastic optimal siting model for the hydrogen refueling stations (HRSs), 3) introducing a stochastic planning model to determine the optimal size and location of the central LOHC hub considering an investor’s perspective, and 4) introducing a multi-objective stochastic planning model for the central LOHC hub incorporating both perspectives of the investor and the utility/ISO. The first part develops two stochastic models that represent key transportation attributes of FCEVs. These models are used as key inputs to the planning framework. The first model estimates FCEV drivers’ convenience level by calculating the trip success percentage (TSP) for a given hydrogen refueling station service range (HRSSR). The second model develops the probability distribution of the transportation sector hydrogen demand, thereby capturing its stochastic nature. Both models rely on real-world FCEV data from a National Renewable Energy Laboratory (NREL) project, under the U.S. Department of Energy (DOE), incorporating factors such as distance and days between refueling, fuel economy, tank level at refueling, number of daily trips, and hydrogen tank capacity. Monte Carlo simulation (MCS) captures the variability in these parameters, where the resulting TSP–HRSSR relationship informs optimal HRS siting, ensuring alignment with drivers’ needs, while the stochastic hydrogen demand model provides realistic demand profiles for more robust and reliable infrastructure planning. The second part presents an optimal stochastic planning model for locating liquid hydrogen HRSs (LH-HRSs). The problem is formulated as a transportation coverage problem that seeks to maximize FCEV coverage within a defined HRSSR. Using the TSP stochastic model to determine HRSSR based on a specific drivers’ convenience level, the model optimizes LH-HRS spatial distribution while accounting for demand variability. This approach ensures efficient refueling infrastructure that meets demand fluctuations. Sensitivity analysis is conducted to further examine how planning for higher fueling reliability, i.e., higher TSP, affects the optimal LH-HRS locations. The third part introduces a novel stochastic planning model to optimize the size and location of the central LOHC hub. This planning model ensures the hub can satisfy FCEVs hydrogen demand, while simultaneously maintaining responsiveness to requests from the ISO for ancillary services to the power grid. The formulation focuses on the investor’s objective, which is maximizing profits from hydrogen sales and ancillary service participation. The last part introduces a novel multi-objective stochastic planning model for the central LOHC hub. It incorporates two objectives: 1) maximizing investor’s profit from hydrogen sales and ancillary service provision, and 2) minimizing costs from the utility/ISO perspective achieved by optimal power flow (OPF). The resulting Pareto front illustrates the trade-off between these two objectives, offering decision-makers a range of optimal solutions from which the most appropriate solution, that aligns with certain strategies, can be selected. A decision-making technique is also implemented to select the most suitable solution based on a defined criterion.