ABSTRACT. In the pursuit of mitigating climate change and transitioning towards a sustainable energy landscape, the accurate estimation of greenhouse gas (GHG) emissions from electricity generation plays a critical role in assessing the impact of large-scale adaption of emerging technologies such as energy efficiency measures in buildings. Long-run marginal emission rates (LRMER) represent a refined approach to understanding the environmental impact of changes in electrical demand over a long-term, encompassing both operational adjustments and structural consequences. This paper introduces a data-driven approach utilizing machine learning to instantly predict and apply dynamic marginal GHG emissions without the need of complex simulation and evaluation workflows. The goal is to make dynamic LRMER accessible without the complex workflows involved, and to provide more comprehensive insights into the environmental consequences of emerging technologies in buildings. Two models were developed. The simple linear model only has two coefficients which allows straight-forward implementation in most workflows and provides a coefficient of determination (R2) of 0.78. The fully optimized forest-based machine learning model provides an improved R2 of 0.96 with an overall prediction improvement of 56.9%.

ABSTRACT. Recent efforts to decarbonize power systems have led to the development of carbon-aware optimal dispatch models. Existing approaches typically focus on modeling either the distribution or the transmission system, disregarding the effects of their operation decisions on each other. Nonetheless, there are a few attempts to combine the operational constraints of both transmission and distribution into a single model. However, existing formulations demand sensitive data sharing between transmission and distribution system operators (TSO and DSOs). This paper addresses the joint trade of energy and carbon allowances in a distributed fashion, thus avoiding excessive data sharing between TSO and DSOs. The optimization model is originally formulated as a centralized problem and then decomposed into subproblems, which are solved locally using the Alternating Direction Method of Multipliers (ADMM). Numerical simulations demonstrate that the distributed framework achieves near-optimal economic dispatch, maintaining total operational costs within 0.3% of the optimal solution.

ABSTRACT. The rapid growth of latency- and computationally-critical Industrial Internet-of-Things (IIoT) applications has accelerated the deployment of hierarchical edge-fog-cloud architectures, which raise operational and environmental concerns due to their increased electricity demand. In this paper, we investigate the carbon-aware day-ahead scheduling of hierarchical data centers (HDC) integrated with radial distribution networks. The problem is formulated as a bilevel optimization problem capturing the non-cooperative interaction between the HDC and distribution system operators. Carbon emissions are explicitly modeled using the carbon-emission flow (CEF) framework, enabling accurate tracking of the carbon impact of power consumption at different buses. To address the computational complexity of the resulting non-convex problem, we propose a scalable iterative heuristic algorithm that alternates between the decisions of the two operators. Numerical results and comparisons with baselines demonstrate consistent reductions in carbon emissions and highlight the importance of accounting for power system dynamics in carbon-aware scheduling of HDC.

ABSTRACT. Using reactive power resources installed in the distribution grid for static voltage control in the transmission system requires efficient coordination between two system operators as well as efficient coordination of a high number of decentralized inverter-based resources. This can be achieved by implementing a controller based on Online Feedback Optimization (OFO) with the aim of tracking a set point at the point of common coupling between the two systems. In this work, we implement a controller based on OFO to track an active power set point combined with either a reactive power set point or a voltage set point, essentially turning the distribution grid into a controllable PQ or PV node. By applying the controller in different case studies, this paper investigates how OFO can be used to support voltage control in the transmission system by orchestrating flexible assets in the distribution grid. The results show the set point tracking capability of the controller even under the presence of independent, decentralized voltage controllers.

ABSTRACT. Power distribution systems are expected to host increasing amounts of photovoltaic (PV) generation, often necessitating network reinforcements such as transformer upgrades or capacity enhancements. The costs of these upgrades are typically recovered from the consumers whose PV installations trigger network violations. While this approach is straightforward, it may lead to unfair outcomes: a consumer that initiates an upgrade bears the full cost, whereas subsequent consumers may benefit from the improved infrastructure without contributing to its expense. An alternative approach is to share the upgrade costs among multiple PV installations in proportion to their capacities; however, this method may still fail to capture the true contribution of each participant to the network impact. In this work, we propose a scheme for a fair cost allocation of network upgrade costs for PV project owners associated with new PV installations in power distribution systems. First, we develop an optimization model to determine the total network upgrade cost under varying levels of PV penetration. Then, we compute fair cost allocations for each PV connection using the Shapley Value and Nucleolus methods. These allocations are obtained by evaluating upgrade costs across multiple installation scenarios. The results are compared against conventional proportional and individual cost allocation schemes, demonstrating that the proposed Shapley Value and Nucleolus approaches achieve significantly fairer cost distributions.

ABSTRACT. Growing need for flexibility services in low carbon power systems and the emergence of local flexibility products enable battery storage systems to operate across multiple revenue streams rather than a single service. However, the existing studies examined the revenue-stacking capability of a battery separately for flexibility service and energy arbitrage. This work develops a dual stream revenue stacking for a battery energy storage system (BESS). A jumping model is designed which allows BESS to co-optimize participation in energy arbitrage and a local Distribution System Operator (DSO) flexibility service. The integrated model is expressed as a mixed integer linear program with binary variables that encode service selection decisions. The optimization is implemented in a rolling horizon scheme that advances by updating day-ahead electricity price and flexibility demand. A GB case study evaluates the proposed framework over representative seven-day periods for four seasons, comparing cumulative and daily profits across different scenarios. Seasonal results consistently show that the integrated jumping model delivers a 9% to 23% gain in cumulative net profit with the highest gain in winter compared to a single-service benchmark. The methodology is broadly applicable to BESS who provide for more than one service, enabling service specific adaptation while enabling switching to improve revenue and utilization.

ABSTRACT. Green hydrogen production through electrolysis process has attracted significant interest in recent years with the validation of numerous projects worldwide. This research work proposes a large scale optimization model of a hybrid power plant integrating solar photovoltaic, electrolyzer and hydrogen storage tank to produce and sell electricity and hydrogen. This model considers multiple PV orientations and the plant is connected to the grid for purchasing deficit energy or selling extra energy to the grid. The effect of electricity price cannibalization is consid- ered in the year 2030, where the cannibalized electricity prices are quantified using a data-driven approach based on historical offering and bidding curve data. A novel method of calculating levelized cost of hydrogen and electricity is presented based on the proportion of energy allocated to hydrogen and electricity production. This model is formulated as a mixed integer linear programming problem and solved using a commercial solver.

ABSTRACT. The increasing integration of distributed energy resources (DERs) and multi-energy carriers has accelerated the development of energy hubs (EHs). However, the limited capacity of individual EHs restricts their direct participation in electricity markets. To address this challenge, this paper proposes a low-carbon transactive power and hydrogen trading framework for a virtual energy hub plant (VEHP), where multiple EHs are aggregated to enable coordinated market participation and internal energy exchange. Within the VEHP, EHs participate in the electricity market while engaging in transactive trading of electricity and green hydrogen. A distributed optimization approach based on the alternating direction method of multipliers (ADMM) is developed to preserve the privacy of EHs and ensure scalability. To address renewable energy source (RES) uncertainties, a robust optimization (RO) framework is adopted. The subproblems are reformulated as mixed-integer convex quadratic programming (MICQP). Results demonstrate that the total operation cost of the VEHP is reduced by about 54.18% in light of the proposed model.

ABSTRACT. Equivalent circuit models are commonly used to represent proton exchange membrane fuel cells in system-level simulations due to their computational efficiency and ability to capture essential electrical behavior. However, concentration losses are often simplified or omitted, and their influence on overall system performance remains insufficiently quantified. This study evaluates how different treatments of concentration losses affect the simulated performance of a fuel cell integrated into a renewable hydrogen energy system. A steady-state techno-economic simulation framework is used to compare key operational, energetic, and economic indicators. The results demonstrate that neglecting concentration losses leads to an overestimation of fuel cell efficiency and electricity production and introduces significant deviations in system operation and cost assessment. These findings highlight the importance of appropriately representing concentration losses to ensure reliable system-level analysis and accurate techno-economic evaluation.

ABSTRACT. One promising alternative to stop global warming in our planet is the use of renewable (green) hydrogen and its derivatives. Many countries have already developed policies and strategies for a green hydrogen economy and, accordingly, the private sector has already started developing projects related to green hydrogen. Given the large uncertainty that a project of this type entails, this paper provides a useful methodology to estimate the value of adding both the flexibility of delaying some of the investments associated and the flexibility of expanding the renewable-energy generation capacity and the hydrogen-production capacity in a compound manner. Specifically, we used a method called Compound Least Square MonteCarlo, which can be used to apply real options analysis in multi-stage investment decision processes. We illustrate the proposed methodology using a hypothetical case, located in Northern Chile. Our numerical results show that, in 86.74% of the cases, it is convenient to invest in a hydrogen production plant, making a compound investment in 100% of these scenarios. In the rest of the cases, results suggest investing only in renewable energy generation capacity. Furthermore, the NPV of the project allowing compound investments is 1,186.5% more profitable than the best project without flexibility (that is, that forces the investor to make all investment decisions at the initial period).

ABSTRACT. Hydrogen is projected to play a critical role in the global energy transition; however, the commercialisation strategies of firms operating in this emerging sector – including both small and medium-sized enterprises (SMEs) and larger industrial players – remain relatively underexplored. This study develops a multi-source computational methodology to identify, characterise, and compare strategic orientations of hydrogen firms across contrasting national innovation systems, integrating seven publicly available datasets with original firm-level data obtained through web scraping of 113 company websites across the United Kingdom, Germany, and South Korea. Hierarchical clustering reveals four distinct strategic clusters: Technology Specialists (28%), Application Integrators (34%), Multi-Project Developers (21%), and Infrastructure Integrators (17%). Pronounced country-level differences reflect underlying ecosystem configurations. UK firms exhibit a science-based orientation with emerging storage strength (38% university co-patents) but constrained application focus. German firms demonstrate strong industrial application focus (40% in Application Integrator cluster) supported by extensive pipeline networks (45 projects) and apprenticeship systems. South Korean firms show policy-aligned export orientation with 39% of firms concentrated in Technology Specialist roles. Infrastructure integration emerges as a previously underexplored strategic dimension in hydrogen firm commercialisation, with significant correlation between infrastructure score and patent count (r=0.42, p<0.001). Critically, this study demonstrates that Critical Raw Materials (CRM) vulnerability is primarily determined by technology choice and persists across all countries—Technology Specialist manufacturers face High Risk regardless of national context due to direct platinum group metals (PGMs) and rare earth elements (REEs) exposure, while Infrastructure Integrators consistently face Low Risk. Our findings suggest effective hydrogen firm support requires coordinated ecosystem development addressing supply chain resilience, skills availability, investment facilitation, and offtake market maturity alongside traditional innovation policy instruments. For SMEs specifically, targeted measures such as de-risked finance and shared infrastructure access remain essential.

ABSTRACT. Detecting thermal anomalies in wind turbine generators is challenging because fixed temperature thresholds are often insufficient under varying operating conditions. This paper proposes a residual-based early warning framework for a 3 MW wind turbine using a physics-based digital twin and a reduced-order model (ROM). The digital twin was developed in the Modelica language to reproduce the electromechanical behavior of the turbine, and an effective wind correction was introduced to improve consistency with field measurements. A ROM-based generator temperature model was then constructed to estimate the normal thermal baseline. The model achieved MAPEs of 7.41% and 6.29% for output power and generator temperature, respectively. In the absence of fault-labeled field data, three representative thermal-anomaly scenarios were synthetically generated. The results confirmed that the proposed framework can detect both severe and subtle thermal anomalies by identifying sustained positive residuals from the expected thermal response, thereby enabling the identification of abnormal heating patterns that may not be clearly captured by fixed temperature thresholds. Therefore, the proposed framework provides a practical basis for early anomaly detection and predictive maintenance of wind turbine generators and other key components.

ABSTRACT. Research on electric vehicle (EV) charging behavior is persistently constrained by the limited availability of high-fidelity data due to scarcity and privacy restrictions. Existing simulation tools partially address this gap but typically rely on historical data replay or rule-based heuristics, and remain decoupled from fleet-level operational optimization. This paper proposes Data2Charge, an end-to-end framework that, unlike existing tools, integrates generative data synthesis directly with operational optimization, enabling on-demand scenario analysis for fleets not represented in recorded traces. A two-stage architecture is adopted: a conditional tabular generative adversarial network (CTGAN) learns temporal patterns of arrivals and session durations, while a k-nearest neighbor (KNN) module assigns corresponding energy demands to preserve the physical consistency between duration and delivered energy. The synthesized fleet is scheduled through a linear programming (LP) formulation that minimizes charging cost under time-of-use (ToU) pricing, subject to each vehicle's required energy, availability window, and maximum charging power. An uncontrolled first-in-first-out (FIFO) scheme serves as a baseline to quantify the gains of the proposed schedule. Validated on two real-world datasets and scaling up to 50,000 vehicles, the framework is delivered through an interactive web platform that enables on-demand fleet configuration and comparison of FIFO and ToU-optimized profiles. Data2Charge thereby provides a reproducible testbed for fleet operations, tariff evaluation, and infrastructure studies, independently of proprietary data traces.

ABSTRACT. Highway electric vehicle (EV) charging systems experience severe congestion during peak-demand periods, where fixed charging station (CS) capacity is often insufficient to satisfy temporally clustered requests. This paper proposes a coordinated charging framework integrating fixed CSs with mobile charging stations (MCSs) under a pricing-aware operational model. A state-of-charge (SoC)-based tariff mechanism is introduced, where EV user acceptance is explicitly modeled through price-dependent decision rules. The problem is formulated as a mixed-integer linear programming (MILP) model that jointly optimizes charging schedules, service allocation, and MCS battery operation to maximize a quality-of-service (QoS)-oriented objective based on demand satisfaction and service coverage. The proposed framework is evaluated under a highway peak-demand scenario with realistic EV characteristics and three pricing schemes. Results demonstrate that coordinated CS–MCS operation significantly improves service coverage, energy demand satisfaction, and operator profitability compared to standalone CS operation, while an appropriately designed pricing strategy enables a balanced trade-off between user participation and economic performance.

ABSTRACT. This paper addresses the Electric Vehicle Charging Scheduling Optimization Problem (EV-CSOP) using PLUS ULTRA, a hybrid metaheuristic combining constructive heuristics with adaptive Large Neighborhood Search (LNS). The method integrates multi-factor dynamic priority scoring with seven specialized destroy-repair operators. Computational experiments on 17 realistic instances (5-60 vehicles, 3-10 chargers) demonstrate 88.24\% average satisfaction and 92.67\% demand fulfillment within 147.2 seconds. The approach uses energy cost minimization objective with demand penalties lambda = 1000 dólares por kWh, validated against MILP at 99.2\% cost quality, and is designed for rolling-horizon periodic reoptimization.

ABSTRACT. The widespread adoption of Plug-in Electric Vehicles (PEVs) creates significant challenges for low-voltage (LV) distribution networks, such as voltage deviations, load imbalance, and increased stress on grid assets. This paper proposes a multi-objective optimization framework for coordinated PEV charging that simultaneously considers load smoothing and rapid charging. An Energy Management System (EMS) dynamically allocates charging power according to network conditions and user preferences. A Pareto-front analysis characterizes the trade-off between charging speed and grid performance, while Jain’s Fairness Index and the Gini coefficient assess the equity of charging among users. Simulation results show that the proposed strategy smooths the power profile, maintains fair and consistent charging across all vehicles, and affects the voltage at the most critical bus under high PEV penetration, confirming the suitability of the approach for managing LV networks with growing PEV demand.Electric vehicles, load smoothing, fairness metrics, multi-objective optimization, smart charging

ABSTRACT. The electrification of transport is accelerating the adoption of electric vehicles (EVs) and increasing the need for smart charging strategies. Dynamic tariffs can unlock EV flexibility by incentivizing charging during periods of lower carbon intensity or reduced grid congestion. However, the optimization models used to design such tariffs are often complex and opaque, limiting transparency, trust, and practical adoption. This work proposes a large language model (LLM)-guided evolutionary framework based on verbal reinforcement learning to generate interpretable surrogate models for the EV tariffs optimization problem via imitation learning. By combining LLM-driven rule generation with multi-objective selection, the proposed approach balances imitation accuracy with structural interpretability. The resulting rule-based models closely approximate the outputs of the original optimization problems while remaining compact and human-interpretable, achieving competitive performance relative to standard machine learning baselines. These results highlight the potential of LLMs to extract interpretable representations from conventional mathematical optimization models, which are often perceived as "black-box" by human decision-makers.

ABSTRACT. In this study, we simulate a bilevel reinforcement learning-based grid-to-vehicle (G2V) charging framework with multiple electric vehicles (EVs) and a single EV Load Aggregator (EVLA) which announces hourly electricity prices, and the EVs respond by optimizing their charging schedules accordingly. The model includes grid-level and customer-level objectives with trade-offs, such as load distribution, revenue maximization, and customer satisfaction. Three reinforcement learning algorithms (TD3, SAC, and PPO) are evaluated and compared with baseline non-adaptive policies, including Flat, Time-of-Use (TOU), and Valley pricing, to analyze their performance as decision-making algorithms for the agents' next actions. The simulation results show that reinforcement learning (RL) models perfom better compared to baseline methods, with lower peak-to-average ratio (PAR) and much higher revenue, thanks to high grid utilization and adaptive pricing. Among all the models used, Soft Actor-Critic (SAC) achieves the best overall performance with 92\% of EVs reach target target State of Charge (SOC) and the revenue of 3550\$, compared to the best baseline model (TOU) with 618\$ , balancing high EV satisfaction with low energy not served (ENS) rates and a flat pricing profile, while effectively balancing demand and maximizing revenue. In contrast, TD3 comes as second with moderate load distribution but performs worse in price stability and user satisfaction. PPO agents fail to optimize their charging and pricing policies and cannot meet charging demand, resulting in significantly lower revenue. The results emphasize that RL algorithms are powerful for simulating bilevel Stackelberg games, as they converge both EVLA and EV policies to the optimal, within 2-way interactions involving pricing and charging signals.

ABSTRACT. The SHS31 project implements a physical, modifiable low-voltage microgrid (MG) within an existing institute building. Unlike purely inverter-based grid-forming units, a dedicated Motor-Generator-Unit (MGU) serves as the backbone and primary grid-forming element in the SHS31 grid. This guarantees real, physical inertial behavior. A decisive unique selling point of our topology is the transfer switching system: specific, real test bench loads can be selectively and modularly switched between the public grid and the MG to create a universal, galvanically isolated, and scalable testbed. The electrical sizing is based on the institute’s measured load profile, ensuring economic scaling while maintaining the full performance capability of the MGU.

ABSTRACT. The increasing penetration of photovoltaic (PV) systems in distribution networks (DNs) necessitates their active participation in voltage and frequency regulation schemes. This paper presents the real-world implementation and validation of voltage-related and frequency-related ancillary services (AS) provided by energy community (EC) PV plants in a 20 kV medium-voltage DN in Greece. The study utilizes high-resolution operational data to validate decentralized droop-based control schemes deployed at PV plant level. Voltage regulation is achieved through Q(V) and Q(P) characteristics compliant with the IEEE 1547-2018 Standard, while frequency support is provided through operation under Limited Frequency Sensitive Mode–Overfrequency (LFSM-O) in accordance with EU Regulation 2016/631 (RfG). The field results confirm reliable reactive power support for voltage regulation, proportional active power curtailment during overfrequency events, and stable, coordinated multi-plant operation of the EC. The work constitutes one of the first documented real-world demonstrations of AS provision by EC PV plants at the DN level in Greece.

ABSTRACT. Today, reducing the carbon footprint is an important target to be attained also in the leisure yachting sector. The adoption of renewable energy is most effective in specific mooring infrastructures, where controlled power systems can store and then route on-land green energy from PV or wind. Due to the large presence of interconnected battery units in marinas, the DC paradigm proves to be the most suitable for powering the controlled smart recharging systems. The paper aims to perform Control-Hardware-In-the-Loop tests to validate the management of a DC yacht microgrid. An optimization problem developed in Python is executed in real-time to minimize the external energy purchase, thus forcing the affordable utilization of marina stored energy. The references for the power converters, output of the optimization algorithm, are then imposed on the HIL platform. The Modbus protocol is adopted to close the communication loop between the optimizer and the microgrid.

ABSTRACT. This paper presents a Decision Focused Learning (DFL) framework for optimal energy management of photovoltaic systems, battery storage, and V2G enabled electric vehicles in energy communities under uncertainty. The proposed method directly trains a neural network policy by minimizing operational cost rather than prediction error. LSTM and XGBoost models are used to forecast PV generation, load demand, and electricity prices, and these forecasts are used as inputs to the control policy. Unlike conventional predict then optimize approaches, the proposed framework learns cost effective decisions through differentiable system dynamics and cost evaluation within the training loop, learning to take the best decision. The method is tested on multiple energy communities with heterogeneous load profiles and compared with a rule based baseline and a MILP benchmark. Results show that the proposed approach reduces operational cost and improves coordination of distributed energy resources under uncertainty.

ABSTRACT. The rapid proliferation of Distributed Energy Resources (DERs) and Electric Vehicles (EVs) has accelerated the adoption of Low-Voltage Direct Current (LVDC) microgrids (MGs). Efficient control of DC fast chargers within these networks is critical for maintaining system stability and efficeint operation. This paper demonstrates a primary control strategy for EV chargers that utilizes the ISO~15118-20 communication protocol to support LVDC microgrid operation. By leveraging the standard's Dynamic Charging Mode and Bidirectional Power Transfer (BPT) capabilities, the proposed system adheres to Current/OS droop control rule setpoints without requiring session renegotiation with the vehicle. Experimental results validate the charger's ability to provide rapid power-flow adjustments, transitioning between power consumption and injection to ensure reliable power distribution in a low-inertia DC environment.

ABSTRACT. This paper develops an optimization framework for energy communities that explicitly accounts for greenhouse gas emissions through time-varying grid emission factors. A mixed-integer linear program is formulated with either a cost-minimizing or an emissions-minimizing objective under identical operational constraints and optional peer-to-peer trading. Real 15-minute data from a Portuguese community are used to evaluate eight scenarios combining two tariff schemes, two objectives, and two trading modes. Results show that peer-to-peer exchange consistently lowers both costs and emissions by matching surplus PV locally, while the choice of objective strongly shapes storage scheduling. Cost-oriented cases align battery use with price signals, whereas emission-oriented cases align with cleaner periods, reducing emissions by up to 14% at the expense of higher bills. Emission-oriented schedules are insensitive to tariff structure, confirming that dispatch follows the emission factor profile rather than prices. The study quantifies operational trade-offs between cost and emissions and provides evidence supporting emission-aware community design.

ABSTRACT. The Spanish Secondary Regulation Service (SRS), launched in 2024, introduced a new settlement mechanism for automatic Frequency Restoration Reserve (aFRR) that significantly impacts the remuneration and compliance obligations of Balancing Service Providers (BSPs). This paper provides a comprehensive analysis of the SRS settlement model, detailing the calculation of marginal prices, accepted aFRR energy, and associated penalties for non-compliance. A case study illustrates the practical application of these concepts, highlighting discrepancies between delivered and accepted energy and their economic implications for BSPs. The results demonstrate that accurate bidding strategies and real-time monitoring are essential to maximize revenues and avoid penalties. Furthermore, the proposed framework is generalizable to any system operator implementing reserve markets, supporting harmonization efforts such as those promoted by the PICASSO project.

ABSTRACT. Fuel-poverty mitigation increasingly requires support mechanisms that are both targeted and implementable, rather than broad ex-post affordability corrections. However, data-driven vulnerability screening and tariff design are often treated as separate tasks, limiting the operational value of predictive risk signals. This paper proposes a unified risk-to-design framework that connects Lower Super Output Area (LSOA)-level fuel-poverty prediction with equity-oriented residential tariff redesign. First, an XGBoost classifier is trained on historical fuel-poverty indicators, and isotonic calibration is applied to ensure the model output serves as a decision-relevant probability rather than merely a ranking score. SHAP analysis is then used to identify the dominant socio-economic and housing-related drivers behind the learned risk signal. Second, the calibrated probabilities are propagated to a synthetic household population, on which a genetic algorithm designs a two-block lifeline tariff under explicit revenue-feasibility and affordability-equity criteria. On the test set, the predictive layer achieves ROC-AUC values of 0.976 before calibration and 0.974 after calibration, indicating that probability refinement is obtained with negligible loss of discrimination. In the tariff-design layer, the optimized structure preserves revenue almost exactly, with a relative deviation of 6.29×〖10〗^(-5), while reducing the lowest-to-highest income-decile burden gap by 9.88%. Thus, the results show that calibrated vulnerability estimation can be translated into an auditable, practically constrained tariff design workflow.

ABSTRACT. As renewable energy integration accelerates, continuous intraday electricity markets have become essential for short-term balancing. However, the simultaneous trading of overlapping contracts with varying granularities creates complex, hidden liquidity dependencies that are critical to market stability but poorly understood. This paper proposes a multivariate State-Dependent Hawkes Process (SDHP) to model cross-product liquidity dynamics between Quarter-Hourly and Hourly intraday continuous contracts. To determine how cross-product effects vary under different market conditions, we classify the trading sessions into two distinct regimes: 'Normal' and 'Stressed', based on a composite metric of the bid-ask spread and inverse order book depth. The model is calibrated using order book data (level 3) data of the Intraday continuous electricity market in the Netherlands in 2024 from EPEX SPOT. Our results reveal that cross-excitation parameters surge during high-stress periods. Notably, Quarter-Hourly cancellations trigger Hourly cancellations at a roughly 5:1 ratio, confirming that short-term liquidity shocks rapidly propagate across the market. By uncovering these regime-dependent liquidity dynamics, this modeling approach provides a quantitative basis for understanding cross-product spillovers, offering valuable insights for execution strategies and risk management in volatile intraday electricity markets.

ABSTRACT. The transition toward decentralized and consumer-centric electricity systems requires market designs that allow end users to participate more actively in local energy trading. Peer-to-peer (P2P) electricity trading has emerged as a promising approach, yet many existing models remain either highly theoretical or insufficiently aligned with practical market operation and regulatory constraints. This paper presents the design and functional validation of a consumer-centric P2P energy marketplace that combines forecasting-driven market activation, mobile-based user interaction, continuous order-book trading, and structured post-trade settlement within a unified digital framework. The proposed platform operates in two tightly coupled layers: a forecasting-driven participation layer, where short-term demand and renewable generation forecasts identify prospective surplus and deficit conditions and trigger user-side notifications, and a continuous trading layer, where validated offers are submitted, matched, recorded, and settled. The market workflow incorporates gate closure, price-time-priority matching, trade lifecycle management, participant-level settlement positions, and invoice generation. Validation is carried out using real consumption and behind-the-meter generation data from living laboratory communities established under the EU-DREAM project. The results confirm that forecasted market opportunities can be translated into structured participant actions, that the trading logic operates consistently across delivery products, and that executed trades can be carried through to auditable post-trade accounting. The proposed platform therefore provides an integrated and practically implementable pathway toward consumer-centric P2P electricity trading, bridging local market participation with formal market operation.

ABSTRACT. The transition of power systems toward higher shares of renewable energy sources, gradual withdrawal of conventional generation and increasing participation of distributed energy resources has been significantly influencing electricity market outcomes. In Greece, major changes are expected by 2030, including substantial renewable capacity expansion, retirement of thermal generation units and the full interconnection of the island of Crete with the Greek mainland system. These developments are anticipated to highly affect price levels, volatility and overall market dynamics. This paper presents a comparative analysis of outcomes in the Greek day-ahead market under current and projected system conditions and investigates the participation and impact of energy communities. A market simulator capable of reproducing the clearing process is used under two system configurations: 2025 and estimated 2030 conditions. Additionally, the participation of an EnC through stepwise bidding curves under both scenarios is examined. The energy community-market interaction is assessed by introducing the optimized bids into the market simulator and re-clearing the market. The results indicate a notable decline in the average clearing price, from 108.27 €/MWh to 80.95 €/MWh (25.23%) over an examined period of two months, while price variability increases, with the coefficient of variation rising from 39.73% to 59.91%. The expected costs of the EnC during a one-week examination period decrease from -24728.78C to -20059.79C (18.88%), demonstrating a measurable improvement in cost efficiency under the projected system conditions.

ABSTRACT. European electricity markets have recently transitioned to 15-minute trading and imbalance settlement periods, replacing the previous 60-minute resolution. This shift strengthens incentives for maintaining balance between supply and demand at shorter time intervals, and is expected to reduce power imbalances. This paper quantifies the impact of this change on power imbalances and discusses its implications for reserve dimensioning and activation. Focusing on the Nordic synchronous area, which is characterized by a high share of variable renewable generation and HVDC interconnections, a model is used to generate 1-minute resolution power imbalance time series for both 60-minute and 15-minute trading periods. Both deterministic and stochastic variations in generation and demand are represented in the model. The results show that 15-minute trading reduces both the occurrence and magnitude of large imbalances, and mitigates those associated with hourly schedule shifts. This implies a lower requirement for balancing reserves and a shift toward more continuous reserve activation. These findings highlight how market design, and in particular its temporal resolution, shapes reserve dimensioning and activation.

ABSTRACT. Renewable energy communities (RECs) offer a flexible structure for coordinating distributed energy resources; however, electric vehicle (EV) charging adds operational complexity due to time-varying demand and electricity prices. This paper proposes a market-aware optimization framework for the cost-optimal operation of RECs with mobile charging stations (MCSs), which are modelled as controllable, flexible charging assets. The proposed model jointly optimizes community energy exchange, grid transactions, and MCS charging/discharging schedules over a time-coupled horizon to minimize total REC operating cost. The framework is evaluated under three configurations: conventional fixed charging under time-of-use (ToU) pricing, MCS scheduling under flat pricing, and market-aware MCS scheduling under ToU pricing. Simulation results show that the proposed market-aware MCS strategy reduces total operating cost by 21.9% compared with the fixed-charging benchmark and by 15.6% compared with flat-price MCS scheduling. The findings demonstrate that MCS flexibility can improve price-responsive EV charging coordination and enhance the economic operation of RECs.

ABSTRACT. This study investigates the economic potential of vehicle-to-grid energy arbitrage in public electric vehicle charging infrastructure. In contrast to residential or workplace electric vehicle charging, public charging sessions are typically shorter, limiting the temporal flexibility required to exploit differentials in dynamic pricing. Uncontrolled charging, smart charging, and vehicle-to-grid strategies are compared using openly available data from real public charging sessions. The results show that, under typical public charging conditions, the economic advantage of vehicle-to-grid over smart and uncontrolled charging remains limited, with charging cost savings ranging between 1% and 6%. The influence of operational factors on the potential savings, such as connection time, charging power and presence of incentives, is investigated. The findings demonstrate that short connection duration is the primary limiting factor for energy arbitrage in public charging, highlighting the need for additional value streams beyond arbitrage alone.

ABSTRACT. The provision of flexibility by electric vehicles requires a certain level of reliability in order to be considered for applications in the energy system. Therefore, it is essential to determine a generally applicable flexibility potential that remains reliable despite uncertainties in driving and charging behavior. In this work, a Monte Carlo simulation is used to generate multiple aggregated flexibility potentials based on energy and power boundaries. These boundaries are then used to derive reliable flexibility potentials using an intersection approach. In addition to the reliability potential that accounts for all simulated uncertainties, confidence intervals are applied to represent different levels of reliability. The subsequent analyses show that using a reliable flexibility potential leads to a systematic underestimation of flexibility, as this represents a worst-case approach.

ABSTRACT. Abstract- The electrification of transportation introduces ‎both opportunities and operational challenges for the power ‎system, as electric vehicles (EVs) with unidirectional (G2V) and ‎bidirectional (V2G) capabilities interact with variable renewable ‎supply. This paper investigates the role of EV flexibility and ‎availability on power system operation using projected 2030 ‎data for the Republic of Ireland. The study examines four ‎combined scenarios of solar generation and load levels (HPHD, ‎HPLD, LPHD, LPLD) over a one-week period, alongside three ‎EV availability cases, from fully available to daytime-restricted ‎operation. A mixed-integer linear programming (MILP) model is ‎used for the evaluation of operational costs and generation ‎dispatch under varying EV participation and renewable ‎conditions. Results indicate that limiting EV availability can ‎increase total system costs by over 8% in the HPHD scenario, ‎reducing EV participation by approximately 80%, while ‎including EV discharging costs mitigates cost increases to 1.7–‎‎3% and lowers participation reductions. Furthermore, ‎bidirectional V2G operation consistently outperforms ‎unidirectional G2V charging, reducing operational costs by up ‎to 6.4% and enhancing system flexibility. Sensitivity analyses ‎highlight the strong dependence of operational benefits on both ‎availability and economic assumptions. Overall, the findings ‎demonstrate that EVs can significantly enhance optimal resource ‎dispatch and system resilience, particularly under high ‎renewable penetration, while emphasizing the importance of ‎accurately modeling user behavior, connection patterns, and ‎network integration for realistic power system planning.‎

ABSTRACT. Electric vehicle (EV) carsharing fleets with Vehicle-to-Grid (V2G) capabilities can provide flexibility to power systems while meeting urban mobility needs. Their operation, however, is strongly influenced by stochastic mobility patterns, especially in one-way systems where vehicle availability and state-of-charge vary significantly between days. Deterministic optimization approaches based on fixed mobility assumptions may therefore lead to unrealistic flexibility estimates and suboptimal decisions.

This work proposes a data-driven stochastic extension of an aggregated V2G optimization framework using Monte Carlo simulation. Real-world mobility data from a carsharing operator, covering several hundred days of operation, are used to generate daily mobility scenarios by random sampling. For each sampled day, an aggregated mixed-integer linear programming model is solved to optimize charging, discharging, and energy procurement over a single-day horizon. The ensemble of solutions is then analyzed to derive robust operational strategies that remain feasible under a wide range of mobility realizations. The proposed approach captures inter-day mobility variability while preserving computational tractability for large urban fleets.

ABSTRACT. In this work, we investigate the DC-side load profiles for battery electric vehicles under the secondary use case of smart charging and V2G. Our focus lies on highly dynamic frequency stabilization control schemes. The power response of three different stabilization services (frequency containment reserve and two different virtual synchronous generators) a derived for a grid incident in the Continental European Synchronous Area. We use a SimScape Electrical simulation model of the EV and charging station combination to assess the current and power profiles for a 400 V class EV for each control strategy. We then apply the power profiles to a series production EV (Volkswagen ID.3) connected to our experimental EVSE via CCS. The measured power profiles are compared to the control setpoints to discuss the suitability of the standard charging power negotiation loop for highly dynamic power profiles. Finally, we suggest improvements to this configuration as well as our own testing setup and give an outlook on our planned future activities in this field.