Liquid Organic Hydrogen Carriers (LOHC) offer a practical solution to overcome the storage and transportation challenges hindering large-scale adoption of green hydrogen. By leveraging existing fuel infrastructure, LOHC eliminates the need for high-pressure or cryogenic conditions, significantly reducing logistical complexity and cost. This paper presents an advanced optimal sizing framework for a renewable-powered, grid-connected LOHC generation facility designed to simultaneously meet transportation-sector hydrogen demand and participate in the ancillary services market. The framework integrates a detailed non-linear electrolyzer degradation-recovery model and incorporates stack replacement cost and carbon pricing directly into the optimization to incentivize re newable energy utilization. It also accounts for seasonal variations in ancillary service requirements. Embedding degradation and replacement effects within the optimization improves electrolyzer efficiency management, reduc ing annual efficiency degradation from 2.1% to 1% and extending stack lifetime from 5 to 10 years. Consequently, the facility achieves a substantially higher net present value of $88.38 million compared with a base case that neglects these effects during optimization. The results highlight the economic and operational advantages of degradation-aware optimization and comprehensive market modeling in the long-term planning of hydrogen infrastructure.

The rapid advancement of Artificial Intelligence (AI) is driving unprecedented computational demands, posing significant challenges to datacenter infrastructure and threatening the stability and resilience of modern power grids. This study presents an open-access dataset featuring a diverse set of AI training sessions recorded at sub-second resolution, designed to advance research on the energy consumption profiles of AI workloads and their interactions with power grid dynamics in datacenter environments. The dataset contains 32 training sessions on high-performance H100 and B200 8-GPU nodes and 40 sessions on consumer-grade NVIDIA GeForce RTX 3060 GPUs, encompassing over 1.8 million samples. Each session records power demand, CPU and GPU utilization, per-GPU power, memory usage, and temperature across diverse AI tasks (at the node scale, temperature refers to the GPUs temperature), including forecasting, classification, reinforcement learning, and text and image generation. Data quality was verified through detailed technical validation, including timing accuracy, hardware limit conformance, and cross-metric correlation analysis. Measurements remained within manufacturer-specified thermal and power envelopes, and observed correlations among power, utilization, temperature, and current were consistent with established processor and GPU behavior. The dataset provides a robust foundation for modeling AI datacenter energy behavior, system-level performance analysis, and power grid connection impact assessment studies.

Decentralized sustainable microgrids are emerging as a promising approach for addressing the increasing complexity of modern power systems while ensuring reliable and efficient operation. A fundamental driver of this transition is the partitioning of distribution networks into self-sufficient microgrids supported by the effective integration of Distributed Energy Resources (DERs) and Energy Storage Systems (ESSs), enabling improved power flow management and enhanced voltage stability. In this regard, this paper proposes a tri-stage optimization framework designed to segment power distribution systems into multiple self-sustaining microgrids while maintaining optimal network performance. In the first stage, the distribution grid is partitioned into microgrid clusters based on electrical distance metrics and bus correlation analysis. The second stage focuses on the optimal sizing and operational management of DERs and ESSs within each identified microgrid to ensure energy self-sufficiency and minimize greenhouse gas (GHG) emissions. In the third stage, an optimal resource allocation strategy is implemented, where the resources determined in the previous stage are optimally placed within the distribution network to achieve optimal power flow, reduce system losses, and maintain voltage stability under worst-case operating conditions. The proposed framework is validated using the IEEE 33-bus test system. Simulation results demonstrate its effectiveness in multi-microgrid classification, coordinated planning, and resource allocation, highlighting its superiority in enhancing system performance and resilience.

Remote communities commonly depend on diesel generators for electricity supply and gasoline-powered vehicles for transportation, despite having abundant solar and wind resources. Limited renewable energy deployment and the high cost of charging infrastructure restrict the transition to cleaner systems.Therefore, this paper proposes a Mixed-Integer Nonlinear Programming (MINLP) framework that jointly optimizes the sizing and scheduling of Renewable Energy Sources (RESs) such as Photovoltaic (PV) and wind, along with Diesel Generation (DG), Stationary Battery Storage System (SBSS), and Mobile Battery Storage System (MBSS). The model also coordinates MBSS routing between community zones to support both EV charging and local load demand. The objective minimizes the total capital and operation costs of the system infrastructure while simultaneously reducing diesel generator contribution, associated greenhouse gas (GHG) emissions, and renewable energy curtailment. Simulation results show that the coordinated use of RESs, SBSS, and MBSS significantly lowers system cost, mitigates renewable curtailment, and achieves substantial decarbonization compared to conventional diesel-only operation and partial decarbonization scenarios.

This paper proposes a new multi-objective optimal design and energy management framework for low density housing Fast Charging Stations (FCS) in multi-EV households. The system integrates Renewable Energy Sources (RES) in the form of Photovoltaic (PV) and Hybrid Battery-Supercapacitor Energy Storage Systems (HBSC ESS) to reduce both annual Capital (CAPEX) and Operational Expenditures (OPEX), while ensuring Electric Vehicle (EV) user convenience. The proposed framework is formulated as a Mixed-Integer Nonlinear Programming (MINLP) problem, and incorporates several critical and practical system constraints including 1) power ramp limits of distribution transformers and ESS, 2) EVs fast-charging behavior modeled through State-of-Power (SoP) versus State-of-Charge (SoC) curves, and 3) transformer thermal capacity constraints based on IEEE C57.91 standard. The framework leverages high-resolution time intervals of 10-sec to capture rapid power variations in generation and load profiles, enabling accurate converter and SC sizing. A detailed sensitivity analysis was conducted across various factors, including daily driven mileage, SoP-SoC characteristics, number of EVs per household, SoC mismatch penalties, transformer thermal overload limits, and different time interval resolution. Results show that neglecting the SoP-SoC relation leads to innacurate design and overestimation of achievable departing SoC, adversely affecting scheduling decisions. Additionally, utilizing transformer thermal capacity allowed a 3.5–4.5% reduction in transformer rating without compromising performance. The number of EVs significantly influenced system design, where coordination becomes critical as the household EV count increases. Additionally, utilizing dual objective allows to maximize user convenience while prevents oversizing the FCS, distribution transformer and HBSC ESS.

Integrating unitized reversible fuel cells (URFCs) in microgrid (MG) frameworks offer a unique opportunity for bidirectional services. While URFC's hold significant potential, coordinating its operation within MGs remains challenging. To address this challenge, this paper proposes a thorough investigation and coordination scheme of detailed dynamic URFCs in a green hydrogen based islanded MGs. The proposed scheme enables the URFC to participate in supporting the DC link voltage, thereby redefining the role of the dispatchable source in the MG. It also coordinates multiple URFCs in a MG enabling them to switch modes according to the operating conditions. Several case studies have been conducted using simulation and real time analysis platforms that integrate URFC into an islanded MG, highlighting the operational advantages that the proposed scheme offers.

Protecting the environment from emissions requires widespread adoption of renewable energy resources alongside both short-term and long-term storage solutions like hydrogen-based systems. However, its crucial to carefully evaluate the economic and operational aspects of the hydrogen system, to ensure proper degradation monitoring and cost management. Most studies in literature tend to focus on either cost or degradation independently. Therefore, this work introduces an optimal framework to simultaneously minimize cost of hydrogen (CoH) and voltage degradation rates within a scalable, modular electrolyzer system by achieving optimal energy management and power allocations among electrolyzer stacks in the islanded microgrid. The framework incorporates variations in Faradaic efficiency to reflect the impact of renewable energy intermittency and employs a look-ahead strategy for proactive decision-making under uncertainty. The resulting non-linear, multimodal optimization problem is solved using the Artificial Hummingbird Algorithm (AHA), a metaheuristic technique known for its strong global search capability and resistance to local optima. The proposed approach is evaluated versus several state-of-the-art approaches using key performance metrics including CoH, voltage degradation rates and various system efficiencies, are calculated to provide detailed insights into the performance of the proposed method compared to existing approaches. The results mainly show that the proposed approach optimally allocates power across system components and among the modular electrolyzer stacks in the islanded microgrid, achieving CoH within 0.15%-0.75% of the approach that focuses primarily on the economic aspect. Furthermore, it significantly reduces voltage degradation rates by 43% and improves various system efficiencies by 2.65%.

The integration of Inverter-Interfaced Distributed Generators (IIDGs) into islanded microgrids presents significant fault ride-through (FRT) challenges, particularly under asymmetrical fault conditions. Key issues include limiting per-phase fault currents, protecting healthy phases, and restoring voltage balance after fault clearance. Existing approaches that rely on fixed or symmetrical virtual impedance often lack the flexibility to address dynamic unbalanced scenarios and fail to ensure post-fault voltage stability. To address these limitations, this paper proposes an adaptive unsymmetrical virtual impedance fault current limiter (UVIFCL), which dynamically regulates each phase’s fault current by modifying its voltage reference. The UVIFCL is governed by a voltage-dependent droop function, enabling adaptive response to fault severity and damping of post-fault oscillations. A frequency-freezing technique is also incorporated to enhance transient stability during faults. The controller is implemented in the natural (abc) frame to allow phase-specific control under both symmetrical and asymmetrical conditions. The proposed strategy is validated using PSCAD simulations on a 4-bus unbalanced islanded microgrid (UBIMG), the IEEE 34-bus distribution system, and a real-time simulation of a 6-bus UBIMG using OPAL-RT. Results confirm the controller ability to limit inverter output currents, preserve voltage in healthy phases, and improve dynamic system performance, offering a robust and communication-free FRT solution for UBIMGs.

The rapid expansion of the Artificial Intelligence (AI) industry has driven large-scale growth of AI-oriented Data Centers (AIDCs) and their integration into power systems. In addition to their high energy demand, AIDCs rely on accelerator-based infrastructure, including GPUs, TPUs, and NPUs, which exhibit significant power variability due to the synchronized nature of AI training workloads. This synchronized operation produces fast and large power fluctuations that propagate to the regional grid and affect power system frequency stability. Therefore, this paper investigates the impact of ultra-fast varying AIDC loads on power system stability under different power smoothing strategies and highlights the new challenges that distinguish AIDC loads from traditional large industrial demand. Results show that AIDC operation increases frequency deviation and Rate-of-Frequency-Change (ROFC) by 20.15% and 153.29%, respectively, even when the AIDC represents only 40% of a traditional load. Applying local smoothing reduces these increases to 4.43% and 5.26%, respectively, but requires oversizing dedicated local Energy Storage Systems (ESS). In contrast, Centralized Remote Power Smoothing reduces ESS capacity requirements but exhibits a notable performance degradation due to the higher magnitude and frequency of AIDC power fluctuations. These findings demonstrate that conventional smoothing approaches are not fully adequate for the emerging dynamic behavior of AIDC loads.

Rapid battery electric vehicle adoption requires direct-current fast-charging infrastructure that is both sufficient in capacity and spatially aligned with mobility-driven demand. Conventional planning relies on administrative or grid-based spatial units that fail to capture the corridor-based nature of urban travel. This study introduces a mesolevel, mobility-aware spatial framework that uses arterial road networks as the fundamental unit of analysis. BEV ownership data are redistributed from postal zones to 234 arterial zones derived from Toronto's road network, yielding a planar adjacency graph that supports spatial smoothing and ridge-path extraction; the extracted ridges act as proxies for high-activity corridors of concentrated charging demand. A graph-based Weighted Proximity Distance metric, jointly weighted by BEV density and charger capacity, quantifies the alignment between chargers and demand corridors. The observed FLO charging network is benchmarked against 1050 randomized config urations preserving network size and capacity. A complementary equity-coverage benchmark, built around a suppressed-demand criterion, adds a fairness dimension to the same null-distribution experiment. On the de mand axis, the existing network performs moderately, sitting at the 31st percentile of the null distribution and outperforming 68.9% of random placements; a further 49.5% reduction in proximity distance remains attainable. On the equity axis, however, the same network is a near-outlier, covering only 7.1% of high-equity zones against a random mean of 51.6% (bottom 0.19%), revealing a structural fairness gap that the demand-only metric does not surface. The framework offers a scalable, data-efficient foundation for demand-aware, equity-conscious DC fast charging planning.

This paper presents a novel formulation for optimizing the design and operation of the grid-connected Proton Exchange Membrane Water Electrolyzer (PEMWE)-based Hydrogen Plants (PEMWE-HP), considering their lifetime degradation impacts. First, the PEMWE voltage and membrane degradation are modeled, and their effects on PEMWE efficiency are analyzed. Then, a degradation-aware optimization framework is developed, encompassing optimizing the ratings of the PEMWE-HP components (i.e., PEMWE, AC/DC converter, compressor, and hydrogen storage) and the PEMWE-HP hourly energy consumption concurrently with the internal parameters of the PEMWE (i.e., cell area, membrane thickness, and cathodic pressure). The proposed formulation is validated using the IEEE 30-bus benchmark system. Results demonstrate that neglecting the degradation models leads to suboptimal sizing of the PEMWE-HP and the PEMWE internal design. Moreover, this omission results in a misleading estimation of the Levelized Cost of Hydrogen (LCOH), with inaccuracy reaching up to 30%.

As variable renewable generation increases, power systems require new strategies to mitigate supply fluctuations and maintain grid stability. Hydrogen electrolysis facilities (HEFs) offer a promising dual-purpose solution, producing clean hydrogen while acting as flexible loads that support grid balancing. However, prior studies have largely relied on simplified steady-state models and colocated configurations, overlooking the dynamic limitations of electrolyzers and the potential of coordinating geographically distributed HEFs. This article presents the first renewable power smoothing (RS) framework that coordinates local and remote HEFs, based on Alkaline (Ak) and proton exchange membrane (PEM) technologies, while accounting for their distinct dynamic response capabilities. A computationally efficient, real-time setpoint allocation algorithm is developed to assign smoothing tasks based on proximity to the renewable source(s) while enforcing HEFs operational constraints. The proposed framework is validated on the New England 39-bus benchmark system in MATLAB/Simulink. Results show that PEM electrolyzers are well-suited for rapid fluctuations, while Ak units contribute effectively to slower variations when appropriately coordinated. Communication delays are also analyzed, underscoring the need for low-latency channels. This work, thus, provides the first practical coordination strategy for distributed HEFs, enabling their effective participation in grid-scale RS and supporting their integration into resilient, low-carbon power systems.

The exponential growth in the usage of textual data across industries and data sharing across institutions underscores the critical need for frameworks that effectively balance data utility and privacy. This paper proposes an innovative agentic AI-based framework specifically tailored for textual data, integrating user-driven qualitative inputs, differential privacy, and generative AI methodologies. The framework comprises four interlinked topics: (1) A novel quantitative approach that translates qualitative user inputs, such as textual completeness, relevance, or coherence, into precise, context-aware utility thresholds through semantic embedding and adaptive metric mapping. (2) A differential privacy-driven mechanism optimizing text embedding perturbations, dynamically balancing semantic fidelity against rigorous privacy constraints. (3) An advanced generative AI approach to synthesize and augment textual datasets, preserving semantic coherence while minimizing sensitive information leakage. (4) An adaptable dataset-dependent optimization system that autonomously profiles textual datasets, selects dataset-specific privacy strategies (e.g., anonymization, paraphrasing), and adapts in real-time to evolving privacy and utility requirements. Each topic is operationalized via specialized agentic modules with explicit mathematical formulations and inter-agent coordination, establishing a robust and adaptive solution for modern textual data challenges.

The increasing integration of intermittent renewable energy sources and the large load variations introduce fast power fluctuations. These fluctuations can impact grid stability and lead to significant frequency deviations, emphasizing the need for effective power smoothing strategies to ensure reliable grid operation. This paper introduces a novel centralized remote power smoothing (CRS) framework, where smoothing facilities (SF) are located away from the sources of power fluctuations. The CRS is proposed as a novel market ancillary service that can be fully integrated within the existing automatic generation control (AGC) of power system operators. The feasibility of the CRS is evaluated in comparison to conventional on-site smoothing and increasing AGC reserve capacity. First, a real-world proof of concept is demonstrated in the electricity system of Ontario, Canada, where a 2-MW flywheel storage facility remotely smooths the output power of a 200-km away transmission-connected wind plant. Subsequently, several simulation studies are conducted on the New England IEEE 39-Bus System, using real high-resolution renewable power and load profiles with 2–second granularity.The proposed CRS achieves up to 84% of the on-site smoothing performance using only 56% of its SFs' capacities. Yet, CRS large-scale practical implementation could be limited to 80% ofits theoretical performance.

This paper presents a dynamic multi-physics model for large-scale proton exchange membrane electrolyzers (PEMEs), explicitly capturing the interactions between the electrical, mass transfer, and thermal domains. The model accounts for the impact of PEME configuration-specifically, the number of series cells and parallel stacks-on the electrical dynamics. The developed framework is first used to evaluate the performance of a conventional decoupled control strategy, revealing its limitations under dynamic operation, particularly with respect to safety constraints such as hydrogen-to-oxygen ratios. To address these limitations, the paper proposes a coordinated control approach designed to improve safety and performance by explicitly accounting for the coupling between physical domains.

The increasing complexity of modern electric power systems, driven by rising demand, technological advancements, and the imperative to decarbonize, underscores the need for sustainable, cost-effective, and decentralized distribution systems. Microgrid-based architectures that incorporate distributed energy resources (DERs) have emerged as a transformative development in this evolution of power networks. A key enabler of this transition is the clustering of power systems into microgrids, which facilitates optimized deployment of DERs and microgridlevel energy planning. This paper presents a clustering approach based on electrical distance, integrated with a comprehensive mixed-integer linear programming (MILP) optimization framework for the optimal sizing and planning of photovoltaic (PV) systems, battery storage systems (BSS), and power conversion systems (PCS) in multi-microgrid distribution networks. The proposed methodology addresses multiple performance objectives, including minimizing capital costs and reducing reliance on the utility grid. By incorporating scenario-based stochastic modeling, the framework captures uncertainties in load and solar generation, enabling robust resource planning to enhance self-sufficiency. The effectiveness of the proposed clustering and optimization strategy is demonstrated through a case study based on the IEEE 13-bus system, validating its capability to support autonomous, low-emission energy planning in modern distribution networks.

This paper derives a dynamic multiphysics model for large-scale proton exchange membrane electrolyzers (PEMEs). The proposed model considers the impact of PEME configuration (i.e., number of series cells and number of parallel stacks) on its electrical domain dynamic model and captures the interactions between the electrical, mass transfer, and thermal domains. As such, it provides a robust foundation for future research aimed at coordinating the controllers of these domains.

Electric bus transit is crucial in reducing greenhouse gas (GHG) emissions, decreasing fossil fuel reliance, and combating climate change. However, the transition to electric-powered buses demands a comprehensive plan for optimal resource allocation, technology choice, infrastructure deployment, and component sizing. This study develops system configuration optimization models for battery electric buses (BEBs) and hydrogen fuel cell buses (HFCBs), minimizing all related costs (i.e., capital and operational costs). These models optimize component sizing of the charging/refueling stations, fleet configuration, and energy/fuel management system in three operational schemes: BEBs opportunity charging, BEBs overnight charging, and electrolysis-powered HFCBs overnight refueling. The results indicate that the BEB opportunity system is the most economically viable choice. Meanwhile, HFCB requires a higher cost (134.5%) and produces more emissions (215.7%) than the BEB overnight charging system. A sensitivity analysis indicates that a significant reduction in the HFCB unit and electricity costs is required to compete economically with BEB systems.

Deep learning (DL) models have been proven to be highly effective in detecting false data injection attacks (FDIAs) in smart grids relying only on the measurements of the secured component. However, the vulnerability of these DL models to Trojan attacks poses significant security risks. This article introduces a model-independent Trojan attack (MITA) targeting DL models used for FDIA detection in critical smart grid components, such as transformer differential relays (TDRs). MITA injects a carefully crafted trigger signal into the TDR's measurements, which evades detection by DL-based FDIA detection methods, causing FDIAs to be misclassified as faults. This leads to unnecessary tripping of the attacked TDR, potentially compromising grid reliability. Key features of MITA include model independence, enabling it to target diverse DL architectures without retraining, high success rates exceeding 99%, and stealthiness. The attack is validated against multiple DL models trained to detect FDIAs on TDRs using FDIA and fault scenarios generated in a realistic OPAL-RT environment. Our results show that MITA maintains the original model's performance in the absence of the trigger while significantly misclassifying FDIAs as faults when the trigger is present. These results highlight the threat Trojan attacks can cause on industrial smart grid systems.

This paper presents an optimization model for enhancing the operational efficiency and expanding the safe operating range of Proton Exchange Membrane Electrolyzers (PEMEs). The approach focuses on dynamically adjusting cathodic pressure and operating temperature in response to real-time variations in input power, electricity prices, and hydrogen demand. The method involves hourly updates to these parameters to maximize PEME efficiency while maintaining system reliability. The study begins by analyzing the impact of cathodic pressure and temperature variations on PEME performance. Building on these insights, an integrated optimization is developed, encompassing the operating set points, sizing, and scheduling of a PEME-based Hydrogen Production Plant (PEME-HP). The proposed methodology is validated using the IEEE 30-bus benchmark system. Results demonstrate that the optimized framework can achieve up to a 19% reduction in the Levelized Cost of Hydrogen (LCOH) compared to conventional non-optimized commercial operating conditions, underscoring its potential to enhance the economic viability of hydrogen production.