Local Energy Communities (LECs) face significant challenges in managing uncertain electricity demand and renewable generation while participating in energy and reserve markets. This study proposes a robust optimization model for LECs equipped with shared battery energy storage systems (BESS) and photovoltaic (PV) generation. Uncertainties in PV output and demand are represented using bounded deviation intervals around nominal forecasts, controlled by a budget-of-uncertainty parameter that tunes the level of conservatism. Linear decision rules (LDRs) are employed to approximate adaptive real-time recourse decisions as affine functions of normalized uncertainty realizations, enabling explicit and tractable adjustments of energy and reserve provision as deviations occur. The model is formulated as an adaptive robust optimization (ARO) problem with a min-max structure. The objective function is decomposed and solved efficiently using the column-and-constraint generation (C&CG) algorithm. The upper-level problem determines optimal energy market bids and reserve commitments under worst-case uncertainty, while the lower-level problem identifies the most adversarial realizations within the uncertainty set. The integration of LDRs with C&CG ensures computational tractability across diverse LEC sizes and configurations, delivering reliable performance despite uncertainty. The budget-ofuncertainty parameter provides transparent control over conservatism, enabling the LEC to balance hedging against severe demand/PV deviations while maintaining efficient market participation at varying uncertainty levels. Numerical results demonstrate the effectiveness of shared BESS in reducing grid dependency, enhancing self-consumption, and enabling robust reserve provision under uncertainty.

Distribution networks face increasing stress from electrification and decentralization, with rising adoption of electric vehicles, heat pumps, and distributed generation reshaping local load profiles. Traditionally, network reinforcement has been the default response to growing demand, but it is capital-intensive, slow to implement, and subject to planning risk under uncertain demand growth. Non-Network Solutions (NNS), including demand response, distributed storage, energy efficiency, and hybrid portfolios, offer an alternative means to manage peak demand and defer costly upgrades. This paper develops and applies a simplified scenario-based framework to evaluate the effectiveness of NNS over a 30-year horizon. The methodology combines load growth modeling, scenario-adjusted trajectories, reinforcement trigger analysis, and economic evaluation through Net Present Value (NPV), including a per-year deferral value to reflect broader system benefits. A stylized case study of a medium-voltage feeder demonstrates the approach across 17 scenarios. Results show that all NNS delay reinforcement relative to business-as-usual, with modest single measures extending capacity by 1–6 years and hybrid portfolios achieving deferrals of up to 15 years. Economically, only low-cost efficiency and demand response options yield positive NPVs under the assumed parameters, while storage-heavy portfolios remain unattractive despite longer deferrals. The findings underscore that while financial viability is sensitive to implementation costs and deferral valuation, the technical and strategic benefits of NNS extend beyond narrow economic metrics. By providing flexibility, reducing planning risk, and enhancing resilience, NNS can play a complementary role in distribution planning when evaluated through a broader lens.

In China, the urban expansion is projected to increase carbon dioxide emissions from buildings. The development of zero-carbon buildings is a strategy for deep decarbonization in the building sector. However, a comprehensive framework to evaluate mitigation potential, implementation pathways, and cost-benefit analyses for future zero-carbon building deployment is lacking. Here, we combine data on life cycle carbon dioxide emissions for 278 cities in China from 2006 to 2022 with a building carbon emission simulation model, socio-economic pathways scenarios, and an optimization model to assess the mitigation potential and associated costs of deploying zero-carbon buildings. The results indicate that under a plausible baseline scenario, total building carbon dioxide emissions are projected to increase to 6.42 gigatonnes by 2030 and reach 8.73 gigatonnes by 2060. An optimal deployment strategy with a 1.5% annual adoption rate of zero-carbon buildings could achieve a 1.21 gigatonne reduction in annual carbon dioxide emissions at implementation costs of $446.8 billion. Our study offers insights into the design and implementation of zero-carbon dioxide building strategies in China, and may inspire similar actions in other countries. In China, adopting zero-carbon buildings at a 1.5 percent annual rate could reduce 1.21 gigatonnes of carbon dioxide emissions annually, with cost reductions through the use of recycled materials, according to socio-economic data and a building carbon emission simulation model.

Local electricity markets (LEMs) operate under low liquidity and high behavioral sensitivity, yet most existing analyses assume static or perfectly rational bidding. This paper proposes an evolutionary game-theoretic (EGT) framework to examine the long-term stability of pricing heuristics under bounded rationality. A uniform-price LEM with grid fallback is simulated over a 96-step representative day, and discrete-time replicator dynamics are used to model strategy adaptation among heterogeneous prosumers. Four interpretable strategies (marginal-cost, fixed markup, scarcity-based markup (SM), and cooperative low-markup) are evaluated under increasing PV penetration levels $(\gamma \in 0.5,1,2,5)$. Evolutionary dominance is identified through asymptotic population shares at convergence. Results reveal a structural transition as PV penetration increases: mixed equilibria prevail under low and moderate liquidity, while low-markup strategies dominate under high surplus conditions. In saturated scenarios, the SM strategy becomes evolutionarily stable, prioritizing dispatch certainty during surplus periods while retaining limited margins under scarcity. Despite a compression of clearing prices toward the feed-in tariff floor, average prosumer profits increase substantially due to scale effects. These findings highlight that LEM stability depends on adaptive pricing rules that reconcile margin extraction with competitive clearing, offering insights into the design of resilient decentralized market mechanisms.

Rapid data center expansion intensifies the tension between deep power-sector decarbonisation and the need for highly reliable, continuous electricity supply. Under grid congestion and limited interconnection capacity, annual renewable claims cannot guarantee hour-by-hour deliverable clean power. We develop an integrated planning framework for private-wire-supplied data centers. This framework determines the optimal within-city siting of the data center hub and its matched wind, solar photovoltaic, and battery resources, subject to point-ofcommon-coupling exchange limits and dedicated-line deliverability constraints. Candidate renewable sites are identified through spatial screening; hourly generation is simulated using ERA5 reanalysis; and an 8760-h load profile is reconstructed for a standardised 25,000-rack campus using a climate-driven power usage effectiveness model. A mixed-integer linear programming model then co-optimises siting, capacities, line sizing, and grid transactions under renewable supply targets of 30%, 50%, 70%, and 95%. Using Qinghai, China, as an illustrative case, we find that hub siting depends mainly on interconnection access and coupling conditions, whereas renewable siting increasingly concentrates in the strongest resource corridors as the target tightens. Solar photovoltaics form the main scalable backbone, while wind is used selectively only in the most advantageous zones. Across candidate-city cases, total annual cost rises from 1.76 to 1.96 billion USD at 30% to 8.56-9.56 billion USD at 95%. Annual displacement of grid electricity purchases increases from about 0.76 TWh to 2.17-2.26 TWh per city, and implied emissions reductions rise from about 0.14 to 0.39-0.41 Mt. COQ per city. The key finding is that, at high clean-electricity shares, the main constraint is not annual renewable availability but hour-by-hour deliverability within the project boundary.

This paper investigates how negative electricity prices influence the operation of peer-to-peer (p2p) energy communities. Building upon a previously validated optimization model, we simulate twelve scenarios combining three tariff structures, a triple-tariff, and indexed-positive tariff (indexed tariff with prices always above the feed-in tariff), and indexed-negative tariff (indexed tariff with prices consistently below the feed-in tariff). We also assess four operational setups: Business-as-Usual (BAU), p2p-only, energy storage only (ESS-only), and p2p with ESS (p2p+ESS). The analysis is carried out for a representative day using high-resolution data from a ten-household community. Results show that while p2p trading and ESS improve system performance under conventional price conditions, their effectiveness is significantly reduced under negative price signals. In particular, the indexed-negative tariff leads to increased grid imports and exports driven by arbitrage, undermining the benefits of local trading and flexibility. The household billing analysis using the mid-market rate (MMR) local price confirms that local trading becomes less relevant in such pricing environments, with household bills converging across configurations. These findings call for a reassessment of local market design assumptions in the context of increasing price volatility and renewable penetration.

The energy sector necessitates strategies to minimise energy losses and enhance energy efficiency. Artificial intelligence has emerged as the most viable solution for forecasting tasks owing to its capability to predict energy consumption and generation patterns. It enables decision-makers to make more informed decisions, with regard to demand response programs (which rely on forecasting algorithms to align demand with supply or system operator requests). These applications often involve short-term forecasts (from 1 min to one day anticipation). This results in forecasting performances that rely on the accuracy of the training and target datasets and the parameterisation of the forecasting algorithms. This concept is presently related to explainable artificial intelligence (XAI). Numerous studies have evaluated this subject and presented various applications and challenges. However, these studies identify several deficiencies, particularly in short-term forecasting. These include the absence of focus on short-term scenarios and insufficiency of explainable algorithms for forecasting applications, particularly in buildings. Therefore, this research conducts a literature review on applications that generate explanations regarding the forecasting performance of energy demand for short periods in buildings with the support of XAI algorithms.

Energy management systems can avoid energy waste, increase savings on the energy bill, increase energy efficiency in the building, and optimize the usage of renewable energy sources to increase the building's sustainability. One of the great challenges, and opportunities, of energy management systems is the human-in-the-loop interaction, enabling users to monitor, control, and operate the facilities as part of the system. This paper proposes a novel human-in-the-loop solution for electric water heater optimization supported by artificial neural networks to predict energy consumptions based on daily and monthly data reading. The proposed solution uses as an application environment a Smart Home that with the use of the Internet of Things establishes the connection between the devices, the use of machine learning to predict the activation of the electric water heater and an Intelligent Virtual Assistant (Alexa) for human-in-the-loop interaction. The case study demonstrates the potential of the proposed solution, as it achieved 47.50

Microgrids (MGs) play a key role in enhancing the resilience of power systems and accommodating more renewable energy sources. Resilience is often measured using tentative indices such as the Independence Performance Index (IPI), which aims to enhance the capability of MGs to cope with contingencies caused by utility interruptions. In this study, the unintended impacts of applying the IPI in the resilience-oriented operation of an MG are analyzed. A stochastic optimization model is developed to minimize operating costs and maximize the IPI, and applied to a modified IEEE 33-bus test system. The results reveal that grid dependence is reduced by increasing the IPI; however, this approach can lead to a significant curtailment of renewable energy, exceeding 20% in the case study. This outcome is driven by increased reliance on distributed generators (DGs) within the MG, which alters power flow and creates contingencies, leaving less room for renewables to generate. These findings highlight the importance of carefully designing resilience metrics to ensure alignment with sustainability and decarbonization goals.

As the demand for cleaner and more resilient energy systems continues to rise, there is a growing need for smarter energy management strategies—especially as natural gas-powered systems and various renewable sources become more prominent. Energy hubs (EHs) have emerged as an effective solution, providing a unified framework for managing and coordinating different energy carriers such as electricity, natural gas, and heat. In this paper, we introduce a novel risk-averse optimization framework based on Information Gap Decision Theory (IGDT), aimed at addressing the uncertainty surrounding natural gas prices and its effect on both the cost and reliability of EH operations. The focus is on optimizing the day-ahead operation of a hydrogen-integrated energy hub, which combines multiple technologies for energy generation, conversion, and storage. These include Combined Heat and Power (CHP) units, various renewable energy sources (RESs), energy storage systems (ESSs), electrolyzers, and hydrogen storage facilities. We provide a detailed mathematical model of the entire system, with all key variables and parameters clearly defined for better clarity and reproducibility. This study highlights how the proposed IGDT-based approach can improve the economic performance and resilience of future energy systems in the face of uncertain market conditions. According to the IGDT analysis, prioritizing robustness through a risk-averse strategy strengthens the system’s ability to handle uncertainty, but this comes with the trade-off of higher expenses.

Energy communities play an important role in enhancing local energy resilience and enabling cost-effective energy management. By aggregating consumers, these communities can leverage collective market participation to achieve economic and operational benefits. This paper proposes an optimal strategy for energy communities to participate in the day-ahead and intraday markets while ensuring cost savings for community members. To address the uncertainties in intraday market prices and the energy consumption of community members, an adaptive robust optimization approach is deployed. The proposed model ensures that, in the worst-case scenario, the energy cost for community members does not exceed the cost of individual participation. The resulting formulation is a nonlinear mixed-integer min-max-min problem, which is recast as a nonlinear min-max problem using duality theory. The outer and inner minimizations determine the optimal strategy of the energy community in the day-ahead and intraday markets, respectively. The worst-case realization of uncertain parameters is captured by the middle maximization. The Big-M method and a decomposition technique are employed to linearize the nonlinear terms. Simulation results demonstrate the effectiveness of the proposed model in a community of 20 consumers, reducing the total energy cost from & euro;515.33 under three daily tariffs to & euro;326.64 per day. Moreover, these simulation results highlight the significant impact of seasonal variation on intraday costs, which are driven by uncertainties in energy demand and market prices.

The transition toward decentralized energy systems requires intelligent, resilient, and user-centered infrastructures capable of supporting both autonomy and cooperation across buildings. This paper presents Caravels, which is a distributed system architecture designed to enable intelligent communities through container-based orchestration, focusing in two new features: personalized user modeling, and inter-building service and data sharing. Building upon a modular microservice deployment paradigm, Caravels integrates a graph-based user preference module for context-aware personalization and a data sharing mechanism that facilitates peer-to-peer cooperation. This paper presents a three-part case study evaluating the system across heterogeneous environments deployed in a community setting. Results demonstrate real-time adaptability through dynamic service deployment based on user-defined preferences, effective inter-building collaboration via shared IoT access and service consumption, and personalized automation through reinforcement learning embedded in graphs. The system maintains low-latency operation and minimal resource usage, validating its applicability to low-cost, edge-constrained environments. Caravels innovations enable distributed cooperation on edge, sharing data while preserving sovereignty, and context aware personalization at the building-level.

The Smart Python Agent Development Environment (SPADE) is widely adopted for multi-agent development in Python, yet robust interoperability with heterogeneous frameworks, such as the Java Agent Development Framework (JADE), remains costly. This paper presents PEAK-ACL (Python-based Framework for Heterogeneous Agent Communities - Agent Communication Language), a communication package compliant with the Foundation for Intelligent Physical Agents specifications. It provides a message parser, envelope handling, conversation utilities, and a message transport layer. PEAK-ACL integrates with Python environments, automates Directory Facilitator registration, and supports symmetric exchanges between SPADE and JADE. The public repository has reproducible examples as well as unit and integration tests. By replacing improvised bridges with a tested, standards-based path, PEAK-ACL enables researchers and developers to connect heterogeneous agents more quickly and with less effort.

This article proposes an optimization model based on Mixed-Integer Linear Programming (MILP) for Flexible Job Shop Scheduling (FJSSP) that accounts for participation in electricity markets, including retail tariffs and the Day-Ahead Market (DAM), to understand the behavior between these approaches. It optimizes the assignment of work to machines and the corresponding sequencing, accounting for constraints, while simultaneously managing photovoltaic systems and batteries in the presence of price variations. The results demonstrate reductions in energy costs, particularly in the DAM, thereby enabling a more effective and economic approach to planning the next day’s work. A comparative analysis between ToU and DAM scenarios reveals a reduction of approximately $50 \%$ in the total energy costs, demonstrating a significant economic advantage over traditional tariff structures.

A promising opportunity to optimize energy control and storage is the use of prediction. Several forecasting algorithms from the artificial intelligence area like the Neural Networks or from the machine learning field as K-Nearest Neighbors and XGBoost are recommended for prediction tasks involving the estimation of energy patterns ahead of time. Additionally, it is recommended to apply wisely the forecasting algorithm relying on unique contexts that define different target periods. The method in this paper targets the prediction of consumptions of a building scheduled for all periods of five minutes of a week:•with the support of K-Nearest Neighbors and Neural Networks•a decision tree identifies unique contexts according to rules that rely on all the patterns with energy and sensors data of a building for periods of five minutes•a Multiarmed Bandit algorithm gifted with reinforcement learning capabilities selects the algorithm more convenient for prediction tasks.The results and conclusions indicate that identification of contexts through decision rules results in higher confidence bounds while evaluating the most effective forecasting algorithm. The SMAPE forecasting errors obtained in the third context were 3.54% with KNN and 4.79% with ANN. The obtained SMAPE forecasting errors in the fourth context were 4.91% with KNN and 4.54% with ANN.

A pervasive threat to power systems is the occurrence of False Data Injection Attacks (FDIA), which can disrupt the underlying control infrastructure and endanger the reliability of the system. Many efforts have been made to combat FDIAs. However, most of the reported methods have some shortcomings that prevent them from being used in real-world applications, such as scalability issues, resource constraints, and privacy concerns, which hinder their practical applicability. In this paper, a novel Fusion Deep Learning (FDL) method is applied to the detection of FDIAs, enabling real-time deployment with high accuracy, low memory overhead, and fast model convergence. The proposed framework employs a confidence-aware, memory-based fusion strategy that enhances robustness to non-IID data, concept drift, and realistic disturbances. It demonstrates strong generalization by accurately distinguishing cyber-attacks from natural variations such as sensor noise and topological changes, thereby reducing false positives. More importantly, no data is transmitted to the control center during optimization of the FDIA detection model to maintain data privacy. The performance of the proposed method is compared with the Centralized Deep Learning model based on Convolutional Neural Network as a representative of centralized approaches. The evaluation of the proposed method was performed on the IEEE 118-bus power network, a real dataset of the New York Independent System Operator. Through numerical evaluation, our findings not only affirm the superiority of FDL in accuracy, data storage efficiency, and convergence speed but also underscore its potential to revolutionize security strategies in modern power infrastructures.

The proliferation of internet of things solutions has been gaining importance, which are now widely adopted in residential and commercial buildings. These solutions can benefit from the adoption of artificial intelligence models to create intelligent solutions and provide supported actions and decisions to assist users and building operators. The evolution of technology, hardware, and software enabled the capability of having artificial intelligence models deployed in the edge layer and on the internet of things’ devices. However, further studies to test these capabilities are needed to validate the feasibility of having artificial intelligence-based models near the devices, resources, and users. The proposed solution presented in this paper will assess the feasibility of having in the same microcontroller three machine learning models while being able to read and measure multiple signals from sensors and communicate the sensor’s data and the models’ outputs to a streaming communication protocol. In this work, two neural networks will be used for temperature forecast and CO2 forecast, and a random forest model will be used to classify, with true or false, the occupancy of a room. All the data will be communicated using the message queuing telemetry transport protocol. The results seem very promising, and the microcontroller was able to perform the given tasks.

The increasing adoption of renewable energy sources, introduces variability and intermittency challenges for power system operation. Virtual Power Plants (VPPs) have emerged as a promising solution by aggregating distributed energy resources (DERs) and enabling coordinated operation, effectively behaving as a single controllable entity in energy markets. This paper presents an optimization framework for VPP participation in the day-ahead energy market, explicitly accounting for battery energy storage system (BESS) degradation. Unlike traditional approaches, the proposed model incorporates the interaction between the battery’s state of charge (SOC) and its C-rate, capturing the SOC-dependent effects on degradation. By integrating these degradation costs into operational decisionmaking, the framework enables more accurate and economically efficient scheduling of BESS assets. Simulation results demonstrate that the battery degradation values affect the scheduling results and the participation in the energy markets.

Electric power systems are undergoing rapid evolution driven by increasing loads, widespread renewable energy integration, distributed generation, sector liberalization, and the rise of emerging technologies like electric vehicles. These transformations necessitate intelligent and efficient management of distribution networks, marking the transition to Smart Grids. This study introduces a novel optimization framework utilizing Benders’ Decomposition to tackle network reconfiguration and self-healing challenges in medium-voltage distribution networks during contingency scenarios. The proposed methodology supports decision-making by optimizing network topology and balancing supply-demand dynamics, minimizing operational costs while ensuring system resilience and reliability. Key contributions include the development of a robust tool capable of delivering optimal reconfiguration solutions with low computational latency, adaptable to networks of various sizes and topologies. Simulations on both 13-bus and 180-bus networks demonstrated the model’s scalability and effectiveness, ensuring operational continuity even under severe contingencies. Additionally, this approach accommodates modern network elements such as energy storage systems, electric vehicle charging infrastructure, and distributed renewable generation, enabling a comprehensive Smart Grid framework. The study highlights the potential for integrating this tool into real-time operational systems, ensuring proactive network management and enhanced resilience.

This work presents CLASH (Computational Load Assessment daSHboard), a software service that operationalizes the methodology proposed by [1] for assessing the computational and energy costs of building load forecasting. CLASH extends the original approach by providing a service with an API and interactive dashboard for configuring the featured applications parameters, running the complete prediction pipeline, and visualizing the trade-offs between accuracy, execution time, and energy consumption. The backend implements REST endpoints that wrap the forecasting scripts, including context creation, outlier handling, dataset splitting, and prediction execution, while enabling parameter control through JSON configuration or API calls. Users can define the context periods (hourly or 5-min periods), the context moments (All day moments; Only activity times; Only night periods; Only night periods), and the processing unit (CPU/GPU). Additionally, users can upload datasets, and trigger both complete and stepwise forecasting. The frontend dashboard allows intuitive parameter selection, execution sequencing, and results exploration, including forecast outputs and computational metrics. As such, CLASH enables direct comparison of forecasting scenarios, highlighting cases where increased model accuracy incurs high computational or monitoring energy costs, confirming and extending the findings of the original study. Initial tests show that end-users can dynamically evaluate forecasting strategies, balancing predictive performance with sustainability considerations. By embedding green computing principles into a ready-to-use, API-accessible service, CLASH bridges the gap between research methodology and practical deployment. It empowers researchers to make informed, sustainability-oriented forecasting decisions without manual script execution or code modification.