Polymer electrolyte membrane fuel cells (PEMFCs) are a promising technology for decarbonizing the transportation sector. However, stack component degradation and the resulting performance ageing remain major barriers to large-scale commercialization. Reliable state-of-health assessment is essential to support the development of durability-oriented control and operation strategies. In this work, a modelling framework coupling a semi-empirical electrocatalyst degradation model, a 1 + 1D physics-based PEMFC performance model, and a 1D in-plane Ce transport model is developed and validated against automotive-representative experimental data. A key innovation is a methodology that converts dynamic voltage profiles into an equivalent sequence of square-wave cycles, enabling the direct application of degradation laws derived from accelerated stress tests to realistic driving conditions. The degradation model accounts for the effects of relative humidity, temperature, upper potential limit, dwell time, and low-voltage excursions on electrochemically active surface area (ECSA) loss. Coupled with the PEMFC model, the framework predicts the heterogeneous ECSA loss in a segmented cell after 1000 h of the automotive ID-FAST driving cycle, which increases from air inlet to air outlet. The average normalized ECSA at end of life is 0.62. The experimentally observed heterogeneous performance ageing, however, is mainly attributed to in-plane Ce redistribution, which accumulates at the air inlet hindering both proton and oxygen transport. After 1000 h, efficiency decreases by 4.5% and 7.5% under low-power and high-power operating modes of the ID-FAST cycle, respectively. The proposed model framework therefore represents a valuable tool for PEMFC durability assessment and state-of-health monitoring under realistic automotive operating conditions.
Low-carbon hydrogen is increasingly considered a key option for decarbonizing hard-to-abate sectors, yet most transition analyses are conducted at national or global scales, with limited attention to metropolitan energy systems where climate policies are implemented. This paper examines how hydrogen can contribute to urban decarbonization under multi-level climate governance by extending the Energy Technology Environment Model (ETEM) to integrate hydrogen production, storage, distribution, and end-use technologies and applying a structured downscaling approach to translate national and provincial data to the metropolitan scale. The framework is applied to the Montréal Metropolitan Community (CMM), a member of the C40 Cities network, to evaluate hydrogen deployment under different decarbonization scenarios. Results show that under full carbon neutrality constraints, hydrogen reaches approximately 8% of total final energy consumption by 2050, with total production of approximately 40 PJ yr−1, yet uptake remains concentrated in heavy industry and hard-to-electrify transport, underscoring hydrogen’s role as a targeted complement to electrification rather than a dominant energy carrier. As climate constraints tighten, hydrogen production shifts toward electrolysis and CCS-enabled pathways, with deployment increasing significantly after 2040. Carbon neutrality further relies on a limited direct air capture capacity, capped at 1 MtCO2 yr−1, to offset residual emissions from hard-to-abate sectors, consistent with the residual permanent-sequestration share of Québec’s net-zero pathway. The proposed modeling framework is designed to be potentially adaptable to other metropolitan regions and C40 cities with similar data infrastructure, offering a structured approach for assessing hydrogen integration in urban energy transitions.
Accurate forecasting of national energy dependency is critically important for sustainable energy planning, energy security, and strategic policy development. However, national energy systems exhibit highly complex temporal dynamics, interdependent energy relationships, and nonlinear consumption–import interactions, making reliable forecasting a challenging task. Existing forecasting approaches generally focus on temporal modeling while insufficiently capturing dynamic relational dependencies among multiple energy components. To address these limitations, this study proposes a Physics-Informed Multi-Scale Graph Transformer Network (PMGT-Net) framework for national energy dependency forecasting using monthly EMRA/EPDK energy data. The proposed framework integrates multi-scale temporal learning, dynamic graph construction, Graph Transformer-based relational modeling, and physics-informed optimization within a unified forecasting architecture. Unlike existing energy forecasting approaches that primarily emphasize temporal modeling or address graph learning, physics-informed optimization, and explainability as separate components, PMGT-Net jointly integrates these complementary mechanisms within a unified end-to-end framework specifically designed for national energy dependency forecasting. In addition, explainable artificial intelligence (XAI) mechanisms are incorporated to improve interpretability and analyze the contribution of individual energy variables to forecasting decisions. The framework models complex interactions among electricity, natural gas, petroleum, and LPG production–consumption–import variables while simultaneously learning long-term temporal dependencies and energy-balance constraints. Extensive experiments were conducted using comparative deep learning and Transformer-based baseline models, including LSTM, GRU, Bi-LSTM, Transformer, Informer, Autoformer, FEDformer, PatchTST, iTransformer and ST-GNN architectures. Experimental results demonstrate that the proposed PMGT-Net framework achieves superior forecasting performance with lower RMSE, MAE, and MAPE values compared with all comparative models. Ablation analyses confirm the effectiveness of each architectural component, while robustness analyses demonstrate stable forecasting capability under noisy, missing-value, and seasonal-shift scenarios. Furthermore, statistical significance tests verify that the observed performance improvements are statistically meaningful. Explainability analyses reveal that natural gas imports, petroleum-related indicators, and electricity consumption are among the most influential variables affecting national energy dependency forecasting. Overall, the proposed PMGT-Net framework provides an effective, interpretable, and physically informed forecasting solution for national energy dependency analysis. The framework can support energy policy development, strategic energy planning, sustainable resource management, and decision-support systems for future smart energy infrastructures.
Industrial decarbonization requires autonomous renewable microgrids capable of reliable operation without utility-grid support. This study develops a reliability-constrained planning framework to identify when battery energy storage (BES), green hydrogen energy storage (GHES), or their hybrid combination is technically justified for industrial autonomy. Three off-grid configurations, namely photovoltaic (PV)–BES, PV–GHES, and PV–BES–GHES, are evaluated using 30-min load profiles from Malaysian plastic, food, and ceramic factories over a 10-year horizon. A rule-based energy management strategy is embedded in a non-dominated sorting genetic algorithm II (NSGA-II) multi-objective optimization framework to minimize cost of energy (COE) and renewable energy curtailment (REC), while energy not supplied (ENS) is imposed as a hard feasibility constraint of 0.1%. Results show that storage selection is governed mainly by load–PV temporal alignment. For the daytime-dominant food factory, all configurations satisfy the reliability limit, and PV–BES achieves the lowest COE of 0.186 $/kWh. For the continuous plastic and ceramic factories, single-storage configurations are infeasible despite lower cost or lower curtailment. PV–BES–GHES is the only feasible option for these cases, achieving ENS values of 0.0900% and 0.0940% respectively. Dispatch-level diagnosis shows that PV–BES fails when the battery remains near minimum state of charge, whereas PV–GHES fails when low surplus is rejected by the electrolyzer, the hydrogen tank is depleted, and fuel-cell generation cannot be activated. Therefore, GHES is most valuable as a complementary long-duration reserve, not as a substitute for high-efficiency battery balancing.
With the rapid growth of flexible loads in power systems, virtual power plants (VPPs) have emerged as a key technological platform for aggregating distributed resources to participate in demand response (DR) programs. The accuracy of aggregated baseline load (ABL) prediction is critical to the effectiveness of DR programs and the fairness of market incentives. However, this remains challenging since the ABL cannot be directly measured and differs from the actual load during DR programs. Additionally, the varying aggregation scales of VPPs create a heterogeneous customer base, further complicating ABL prediction. To address these challenges, this study proposes a demand response adaptive decomposition and prediction (DRADP) system for different types of power customers with varied load levels. Specifically, we first perform data extension and mode decomposition for various resource types, reconstructing ABL data into trend and fluctuation branches. Upon the completion of data construction, a model incorporating a patch temporal convolutional network, an encoder–decoder structure, and a multi-source attention mechanism is developed to capture hidden features and volatile trends. The forecasting results are then integrated and regressed into point and interval predictions. Finally, experiments based on real customer load data from VPP operators validate the effectiveness of the proposed system in both deterministic forecasting and uncertainty analysis. In representative 60-min DR forecasting scenarios, DRADP achieves the mean absolute percentage error values of MAPEDM=1.534%, MAPEBAC=2.520%, and MAPECBSF=5.862%, respectively, significantly outperforming the benchmark models.
