Enhancing Energy Systems and Rural Communities through a System of Systems Approach: A Comprehensive Review

Abstract

Today’s increasingly complex energy systems require innovative approaches to integrate and optimize different energy sources and technologies. In this paper, we explore the system of systems (SoS) approach, which provides a comprehensive framework for improving energy systems’ interoperability, efficiency, and resilience. By examining recent advances in various sectors, including photovoltaic systems, electric vehicles, energy storage, renewable energy, smart cities, and rural communities, this study highlights the essential role of SoSs in addressing the challenges of the energy transition. The principal areas of interest include the integration of advanced control algorithms and machine learning techniques and the development of robust communication networks to manage interactions between interconnected subsystems. This study also identifies significant challenges associated with large-scale SoS implementation, such as real-time data processing, decision-making complexity, and the need for harmonized regulatory frameworks. This study outlines future directions for improving the intelligence and autonomy of energy subsystems, which are essential for achieving a sustainable, resilient, and adaptive energy infrastructure.

1. Introduction

In today’s constantly evolving landscape of complex technological advances and interconnected infrastructures, the concept of systems of systems (SoSs) has emerged as a paradigm designed to encompass the intricacies of highly integrated and interdependent systems [1]. Defined as an arrangement of individual and autonomous systems working together to achieve a common goal [2], SoSs transcend the traditional boundaries of isolated systems, opening the way to a holistic and synergistic approach to problem solving [3]. This paradigm shift is becoming increasingly relevant in contemporary engineering, where the demand for scalable, adaptable, and interconnected solutions is paramount [4]. The interconnectivity inherent in SoSs presents innovative challenges and opportunities, from managing emerging behaviors to optimizing collaboration between diverse components [5].

This concept extends to various areas such as mobility, healthcare, security, energy and beyond [6,7,8,9], attracting interest across different sectors and prompting the International Organization for Standardization (ISO) to develop standards in 2016 [10]. These standards include considerations throughout the stages of a system’s lifecycle (ISO/IEC/IEEE 21839) [11], guidelines for using existing standards in SoS contexts (ISO/IEC/IEEE 21840) [12], and a taxonomy describing the classification of SoSs (ISO/IEC/IEEE 21841) [13]. Within this framework, four types of SoSs—directed, recognized, collaborative, and virtual—are distinguished according to the degree of independence between the constituent systems [14]. These types manifest themselves in diverse applications such as business, education, government agencies, healthcare, media, and transportation, illustrating the adaptability and ubiquity of the SoS concept in different sectors [15]. The ultimate goal of the SoS architecture is to extract maximum value by understanding the interactions, interfaces, and use of each smaller system within a larger, evolving context [16].

The main contributions of SoSs in energy systems focus on improving coordination between disparate subsystems, such as renewable energy sources, power grids, and energy storage systems [17,18]. Studies such as [19,20] have explored how SoSs can improve the reliability and efficiency of energy systems by implementing control and optimization strategies. Research such as [21,22,23] has developed SoSs for smart grid development, focusing on communication and decision-making protocols that guarantee real-time optimization of energy flows. In addition, studies such as [24,25] have explored the role of SoSs in integrating renewable energy sources such as solar and wind power into existing energy infrastructures.

In addition, the transition to renewable energy systems has introduced many challenges, including variable operational characteristics of energy subsystems and the need for intermittent energy supply [26,27]. Studies such as [28,29,30] have addressed these challenges by proposing dynamic control models and intelligent energy management systems (EMSs) that use SoS frameworks to manage these complexities. Other studies [31,32] have focused on decentralized energy storage solutions, showing how SoS approaches can balance supply and demand on distributed energy resources.

Furthermore, the development of SoSs in energy systems is also linked to the growing use of machine learning and artificial intelligence (AI) for predictive analysis and optimization [33]. Certain research studies [34,35] have highlighted the potential of AI-based SoS approaches for autonomously adjusting subsystem behavior based on real-time data, which should impressively enhance the efficiency and adaptability of energy systems.

Tekinerdogan et al. [36] discussed several challenges inherent to SoS engineering. These obstacles included decentralized engineering processes, the continuous evolution and deployment requirements of SoSs, the lack of explicit modeling formalisms leading to reliance on traditional models, the need for a multi-paradigm modeling approach, and the requirement for a stable understanding of interdisciplinary engineering. In addition, the paper highlighted the lack of a socio-technical approach addressing the interaction of technical and human components, the misalignment of current software ecosystem architectures with the SoS scale, and the need for new design optimization approaches to balance global and local considerations. Furthermore, it highlighted the difficulties posed by emerging behaviors, diverse governance structures, heterogeneity, interoperability challenges, and the inadequacy of traditional evaluation methods for the unique characteristics of SoSs. Finally, the paper highlighted the essential role of software engineering in very large-scale SoSs, pointing out that current approaches fail to meet the evolving demands of intelligent systems at this level.

This study presents a systematic literature review using the Web of Science (WoS) database covering the period from 2020 to July 2024. Our approach to the selection of research articles was based on the main important parameters studied, namely, the impact of emergent behaviors on the effectiveness of SoS control and optimization, particularly in energy systems. As shown in Figure 1, search criteria were defined using databases such as Google Scholar, IEEE Xplore, Scopus, and Web of Science and specific keywords related to SoS, optimization, and control. The study selection process used the TAK (Title, Abstract, and Keywords) approach [37], with keywords such as “system of systems”, “emerging behaviors”, and “control” or “optimization” to filter relevant articles followed by a full-text review to check for relevance. Inclusion criteria focused on peer-reviewed articles, research articles, and case studies published within the last five years, while non-English articles, non-peer-reviewed articles, and irrelevant topics were excluded. Data extraction included study objectives, methodologies, and the application of SoSs in the energy field. Finally, the results were synthesized to identify common themes, trends, and gaps, leading to the final selection of articles.

According to a search of the Web of Science database over the last five years (2020–2024), focusing on “systems of systems” and “emerging behaviors”, the total number of publications amounted to 3086, including 114 specifically on “control or optimization” in the energy sector. Of these, the highest number was recorded in 2023, with 24% of SoS publications and 29% of publications on energy applications. This peak suggests an increased interest in research in that year. In contrast, the lowest figures were observed in 2020, with 19% of SoS publications and 13% of publications on energy applications, reflecting less attention paid to these topics at the start of the period. While the percentage of publications on energy applications as a proportion of total SoS publications was consistently low, ranging from 2.5% to 6.3%, the gradual increases each year indicates a growing emphasis on integrating SoS principles into energy research, in line with global trends towards sustainable and optimized energy systems (Figure 2).

Figure 3 provides a distribution of annual Indexed Web of Science (WoS) publications by publisher title from 2020 to 2024, focusing on SoSs and SoS energy applications. Elsevier had the highest number of publications in both categories, totaling 20% for SoSs and 63% for SoS energy applications. IEEE followed with 10% SoS publications but only 3% for SoS energy applications. Other notable publishers include Springer Nature (8% SoSs, 1% energy) and MDPI (6% SoSs, 16% energy).

This article aims to provide an overview of control and optimization systems in the energy field. It is structured as follows: Section 2 presents an overview of studies focusing on SoS control and optimization. Section 3 presents an overview of SoSs and their importance for energy systems, including photovoltaics, electric vehicles, power systems, energy storage, renewables, smart cities, and carbon emissions. Section 4 discusses the main challenges and presents future directions in this field.

2. Overview of Systems of Systems (SoSs) and Their Relevance to Energy Systems

In this context, various studies have collectively advanced the knowledge of energy systems, engineering, and biomedical fields by introducing innovative methods and technologies. Figure 4 illustrates an example of an SOS, showcasing how two separate systems may be combined, and as a result of their interaction, new behaviors might arise. Figure 5 illustrates an example of an SoS approach in home automation, where a photovoltaic panel subsystem captures solar energy and transfers it to a battery; the stored energy is subsequently distributed for household consumption and used to recharge electric vehicles.

As part of the research, economic model predictive control has been used to improve the performance of chemical reactors [38]. Hybrid solar-biomass systems have been designed to dry crops more efficiently [39], and fuzzy-based intelligent EMSs have been created to lower peak demand in Saudi Arabian residential buildings [40]. Other studies have focused on enhancing organic Rankine cycle systems [41], proposing a dual-layer game optimization method for secure peer-to-peer energy transactions [42], and optimizing geothermal energy extraction by analyzing fracture connectivity [43]. Additionally, significant progress was made in designing compact heat exchangers [44], predicting coal seam explosions using infrasound [45], and improving fluidization processes in chemical industries [46]. Advances in high-temperature heat pumps [47], the auto-ignition of fuel mixtures [48], Tesla turbine design for heat pumps [49], and gelled foam technologies for oil recovery [50] were also highlighted. In agriculture, optimizing greenhouse climates [51], and in material science, enhancing drilling fluid stability with nanocomposites [52], were key contributions. Finally, innovations in thin-film thermoelectric coolers for chip cooling [53], blockchain risk management in carbon trading [54], and cell alignment mechanisms [55] further underscore the breadth of these advancements.

In this review article, we analyzed 114 articles from the WoS database, including 8 review articles, which focused on various aspects of renewable energy systems and technologies. Collectively, these articles highlighted the various approaches and innovations needed to advance renewable energy systems and address the associated challenges. The first article [56] explored applications of reinforcement learning (RL) in energy systems, noting significant performance improvements despite the underutilization of advanced RL techniques. The second article [57] examined recent advances in modeling and simulation for renewable and sustainable energy systems (RSESs), highlighting the need for further research into model validation and system behavior. The third article [58] discussed the potential and challenges of occupant-centered controls (OCCs) in building energy management (EM), presenting a repository of 58 OCC case studies to aid standardization and practical implementation. The fourth article [59] examined the impact of emissions trading schemes on global low-carbon energy transitions, highlighting the complexities and effectiveness of such schemes compared to traditional carbon pricing methods. The fifth article [60] systematically examined the use of reinforcement learning in occupant-centered building EM, proposing a new framework for its application. The sixth article [61] analyzed the socio-psychological factors influencing the intention to adopt electric vehicles (EVs) in India, suggesting the need for policy shifts beyond subsidies towards attitudinal and norm-based incentives. The seventh article [62] provided a comprehensive overview of anion exchange membrane water electrolysis (AEMWE) for green hydrogen production, discussing recent advances in electrode design and system optimization. Finally, the eighth paper [63] compared solar thermal and PV thermal technology in multi-source energy systems, demonstrating significant efficiency improvements with the use of PV thermal panels in various European climates.

2.1. Photovoltaic Systems

This section explores advancements in the control and optimization of SoS approaches for energy applications, particularly in renewable energy integration and energy management. Boubii et al. [64] highlighted advances in SoS control and optimization in the energy sector, focusing on photovoltaic (PV) systems as a key solution to the global energy crisis through the promotion of renewable energies. They presented an innovative PV system designed for grid connection, focusing on dynamic modeling and control to improve reliability and power quality. The system incorporated model predictive control (MPC) integrated with a high-gain DC-DC converter to improve maximum power point tracking (MPPT) efficiency, taking advantage of the incremental conductance (IN-C) method and predictive control supported by particle swarm optimization (PSO). The article mentioned that future research should develop advanced control algorithms integrating real-time weather forecasts and sensor-based feedback, explore energy storage solutions to improve PV system reliability, and optimize the use of solar energy for sustainable power generation and grid integration.

Battery storage is a fundamental element in improving the efficiency of renewable energy systems such as solar PV and wind power. It is therefore essential to model batteries accurately to predict their performance under various conditions and develop advanced grid management applications. For this purpose, Shabani et al. [65] explored the impact of two battery modeling scenarios on the optimization of a grid-connected PV battery system. The first scenario used a simple battery model without considering dynamic behavior, while the second used a complex model that took into account the current voltage characteristics of the battery under various conditions. A rule-based operational strategy and a genetic algorithm for non-dominated sorting were used for optimization, focusing on minimizing battery lifecycle cost and maximizing the self-sufficiency rate. The results indicated that the complex model led to more accurate and efficient battery sizing, lower lifecycle costs, and higher self-sufficiency rates compared to the simple model. Scenario 1 systematically underestimated and overestimated battery state-of-charge (SOC) during charging and discharging, respectively, by neglecting internal battery parameters. Scenario 2, on the other hand, provided more realistic SOC forecasts and enabled a better energy supply to the PV system. In economic terms, scenario 2 results in a lower battery capacity requirement and a 9.4% reduction in lifecycle cost, demonstrating the benefits of using an accurate battery model for system optimization.

Kang et al. [66] addressed the challenge of optimizing battery energy storage systems (BESSs) to manage the intermittency and volatility of PV systems in residential buildings. Recognizing the limitations of existing BESS planning due to insufficient consideration of diverse user objectives, the researchers developed an optimal planning model based on reinforcement learning (RL). Four RL algorithms were evaluated, with proximal policy optimization (PPO) emerging as the most effective for managing BESS with PV systems in the context of real uncertainties. The case study of a residential building demonstrated that the PPO-based RL model outperformed other algorithms in decision making for optimal BESS planning, maximizing self-sufficiency and economic benefits. The RL-based model had the potential for application in virtual power plants, enabling continuous electricity sharing between multiple buildings. However, the study noted limitations due to data availability and suggested that future research should include broader scenarios, different battery sizes, building types, and regions to improve the scalability and adaptability of the model. This future work would provide more in-depth information on the model’s effectiveness and improve its practical utility in EM and distribution, taking into account real-world uncertainties such as market price volatility and BESS degradation. In addition, ongoing efforts will focus on sensitivity analysis and long-term economic considerations to further refine the RL model’s performance.

As the world’s population grows, rural areas, particularly isolated ones, are struggling with inadequate energy infrastructure, making essential services such as electricity inaccessible. Rodriguez et al. [67] presented the design of a novel energy management system (EMS) for isolated microgrids that included a PV system, a diesel generator (DLG), and a battery energy storage system (ESS). Given the effectiveness of fuzzy logic control (FLC) in managing microgrid non-linearities and its rare application in isolated microgrids, the study proposed an FLC-based EMS. The proposed system used production and demand forecasts to optimize the operation of the DLG and the use of solar resources while preserving the longevity of the ESS. FLC parameters were refined using particle PSO and cuckoo search (CS) algorithms to improve EMS performance. A battery degradation model estimated the state of health (SOH) of the ESS. A case study in a rural Ecuadorian community demonstrated the superiority of the proposed EMS over a previous, non-optimized EMS. The optimized EMS, in particular with PSO tuning, showed better DLG management, more effective battery state-of-charge (SOC) control, and lower operating costs by maximizing the use of PV energy and minimizing waste. Validation using Matlab^® and the Typhoon HIL-402 hardware-in-the-loop device (with the Typhoon HIL Control Center v2021.4 platform) confirmed that the proposed EMS effectively limited DLG usage, optimized solar energy utilization, and extended battery life by preventing overcharging and deep discharge.

Further, Haffaf et al. [68] explored the experimental operation and performance evaluation of a grid-connected hybrid microgrid (MG) system integrating PV panels, batteries, and an EV to enhance PV self-consumption and demand-side management (DSM). The MG, deployed at the Institut Universitaire de Technologie in Mulhouse, France, in August 2018, comprised two subsystems: subsystem 1 with a 2.16 kWp polycrystalline PV array and subsystem 2 with a 2.4 kWp monocrystalline PV array, plus an EV with a 6.1 kWh lithium-ion battery. The study analyzed the behavior of system components, including inverters, batteries, and the power grid, under conditions with and without an EV connection. The results showed that subsystems 1 and 2 injected 3466.82 kWh and 5836.58 kWh, respectively, into the grid, with subsystem 2 producing 5597.65 kWh of energy and emitting 4.17 metric tons of CO₂ over its lifetime. Detailed assessments of PV power output, energy efficiency, injection power, and self-consumed energy demonstrated the MG’s ability to increase the self-consumption of PV energy and reduce dependence on the conventional electricity grid. However, the inclusion of local energy storage maximized the use of solar energy, effectively reducing electricity costs despite higher installation costs. In addition, the integration of DSM activities such as load transfer and optimal consumption scheduling further optimized self-consumption, underscoring the MG’s potential for sustainable and efficient EM.

Fischer et al. [69] presented a bottom-up stochastic model designed to assess the flexibility of residential electric load profiles, considering user behavior and the relevant technologies characterized by their various sizes and controller parameters. They mapped the flexibility of these technologies into an equivalent linear electrical storage model, dynamically parameterized during load profile simulation, making it compatible with optimization methods commonly used in optimal controls. The model was applied to examine the flexibility of individual technologies in a residential area in Germany, taking into account technology penetration rates. The results revealed that EVs and PV battery systems provided the highest values of switchable power and storage capacity. However, when considering penetration rates, heat pumps (HPs) emerged as important flexibility providers. EVs offered the highest flexibility potential as a single technology, with average annual values of 5.4 kW per apartment and a maximum storage capacity of 11.4 kWh, but their availability varied due to limited permanent connections to the domestic grid and energy demand. However, when technology penetration rates were considered, HPs provided up to 55% of maximum switchable power and storage capacity if storage temperatures were increased. Appliances, although present in every apartment, contributed less to maximum power and energy. Micro-cogeneration heat and power (μ-CHP) units played a minor role due to their current base-load operating size and low storage. This suggests the need for adjusted sizing procedures for increased flexibility. The flexibility of HP and PV battery systems showed seasonal and daily variations, with potential competition between PV battery systems and HP flexibility in winter when the battery charge was lower.

Nezamabadi et al. [70] proposed a new arbitrage strategy for renewable energy (RE)-based microgrids that addressed the volatile nature of renewable energy sources (RESs) such as PVs and wind power in the context of peer-to-peer (P2P) energy trading on transactive energy markets (TEMs) located between day-ahead markets (DAMs) and real-time markets (RTMs). This strategy used bi-level risk-constrained stochastic programming with interval coefficients (BRSPIC) to exploit price differences between P2P and real-time transactions. Initially, DAM price uncertainties were managed through scenario-based decision making. Subsequently, P2P competition in energy trading was modeled as a non-cooperative leader–follower game in which MGs maximized profits at the top level while social welfare was optimized at the bottom level. To manage uncertainties in RESs, load, and TMR prices under real-time approaches, interval coefficients were used, and conditional value-at-risk (CVaR) was applied to mitigate the risks of profit variability. Using Karush–Kuhn–Tucker (KKT) conditions, BRSPIC was transformed into a single-level optimization, then linearized and solved with a mixed-integer linear programming (MILP) solver. Testing of the model on a test system demonstrated an increase in profits of over 3.1% for MGs using this arbitrage strategy. However, integrating CVaR into a fully risk-averse approach reduced MG’s profits by 27%, highlighting the trade-off between profit and risk aversion.

Raina et al. [71] evaluated the performance of bifacial PV modules compared to monofacial modules in various shading scenarios using Jaipur, India, as a case study. The research involved a three-phase approach: module-level performance analysis, simulations to determine the optimal spacing between rows to reduce shading, and dynamic grid reconfiguration to mitigate shading effects. The results showed that bifacial PV modules consistently outperformed monofacial modules due to their ability to absorb light from the rear side, resulting in lower shading losses and higher power output. Specifically, bifacial modules showed lower negative voltage under shading conditions, and optimal spacing was found to effectively mitigate mutual shading. Furthermore, research on a 3 × 3 array revealed that dynamic array reconfiguration (DAR) significantly reduced the fill factor (FF) to 18.9% under uniform shading compared with over 40% under non-uniform shading. The study highlighted the potential of bifacial technology to improve electricity production in PV systems, despite its slower adoption rate. It also highlighted the need for detailed research into shading effects and array mismatches to optimize the performance of bifacial modules, particularly in space-constrained installations typical of countries such as India. This research provided valuable information for industries considering two-sided PV installations, proposing solutions for maximizing production and profitability in suboptimal conditions.

Ghaebi et al. [72] examined the evolution of energy markets influenced by emerging technologies and the growing presence of RESs and demand response programs (DRPs). Traditionally dominated by conventional power plants, these markets include flexible loads such as plug-in electric vehicles (PEVs) and heating/cooling systems. EV parking lots serve as short-term battery storage, while cooling/heating systems can adjust loads without sacrificing comfort. The study compared short-term energy regulation markets, focusing on bidding constraints and payment mechanisms, using a reconfigurable urban microgrid with various energy sources, including an EV fleet, wind turbines, and rooftop PV panels. The results indicated that coordinated microgrids improved flexibility and cost-effectiveness and that diversified energy portfolios enabled more active and cost-effective market participation. The research, which used real market data from the US and Europe, emphasized that reducing the gap between market closure and real-time operations and relaxing certain market constraints can benefit RES participation, particularly in markets like Germany, France, and the UK.

Yang et al. [73] addressed the challenge of low power conversion efficiency (PCE) in vertically oriented 2D perovskites by introducing a novel crystallization pathway (RCP) to mitigate defects caused by rapid self-assembly. The approach involved controlling the adsorption of an ammonium halide additive on different crystal planes of the perovskite, specifically targeting the suppression of (111) plane growth while promoting the formation of larger grains on 202 planes. As halogen-induced deprotonation progressed, the dominance of 111 plane growth was restored, resulting in a vertically oriented 2D perovskite film with improved homogeneity, lower trap density, and enhanced carrier transport and collection properties. In addition, solar cells fabricated with these RCP-treated 2D perovskite films achieved a reproducible and stable PCE of 18.5% and a high fill factor of 83.4%. This represents a significant advancement in optimizing 2D perovskite films for superior PV performance by effectively balancing defect orientation and reduction.

Addressing the challenge of improving the performance of perovskite solar cells (PSCs) at high temperatures, Li et al. [74] proposed integrating composite phase change materials (CPCMs) with a porous structure with the aim of improving the thermal management and efficiency of PSCs. A pore-scale lattice Boltzmann phase change model was developed to analyze the interaction between the temperature dependence of the PSC and the temperature control behavior of the CPCM. The study revealed that neglecting the temperature dependence of PSCs led to underestimates of surface temperatures and overestimates of power conversion efficiency (PCE), while glass transmissivity affected these measurements inversely. Under variable solar irradiation, PSC temperatures and efficiency showed different trends compared to constant conditions, highlighting the significant impact of temperature dependence and glass transmissivity. The results indicate that PSC temperatures increased by 3.7% and efficiency decreased by 4.2% when temperature dependence was taken into account, while improved glass transmissivity led to lower temperatures and higher efficiency.

Kaner et al. [75] investigated the linear photo galvanic effect (LPGE) in a lead-free quasi-2D organic–inorganic hybrid perovskite, (CH₃)₂NH₂SnI₃ (DMASnI₃), using first-principles calculations. The research revealed that DMASnI3 exhibited a significant shift photocurrent (SPC) of around 430 μAV-2 in the ultraviolet (UV) range, which was around eight times higher than that of its three-dimensional counterparts. The enhanced SPC was attributed to the electronic properties of Sn and I, in particular their band-edge-influencing p-states. The orientation of organic cations affects the band gap and electronic properties, slightly shifting the LPGE photocurrent spectrum. The results suggest that DMASnI₃, with its strong photocurrent responses, potential for reduced recombination, and long carrier scattering lengths, is a promising lead-free alternative for nonlinear optics and advanced optoelectronic applications.

2.2. Electric Vehicles

Alzahrani et al. [76] focused on real-time energy optimization in smart homes by integrating inflexible loads (TV, computer, lights) and flexible loads (EV, HVAC, water heater) alongside RESs such as PVs and wind power. Using the Lyapunov optimization technique (LOT), the researchers developed an algorithm to optimize total cost, minimize thermal discomfort, and manage charging and discharging of batteries and electric vehicles without the need for anticipatory system parameters. Simulations of various scenarios and weather conditions validated the effectiveness of the algorithm, demonstrating superior performance to existing models in terms of computational efficiency and optimization results. Future research aims to extend this approach to commercial and industrial buildings, highlighting its scalability and wide applicability in real-time energy optimization in various sectors.

To optimize coordination between energy hubs (EHs) and electric vehicle aggregators (EVAGGs) within integrated energy systems (IESs), a novel model-free multi-agent deep reinforcement learning (MADRL) approach was proposed by Zhang et al. [77]. EHs improve the efficiency, flexibility, and reliability of IESs, while EVAGGs facilitate electricity exchanges with the grid. However, achieving optimal coordination is challenging due to privacy issues, parameter uncertainties (e.g., EV charging behaviors, load demands, wind power, and PV generation), dynamic energy flows, continuous decision spaces, and non-convex multi-objective functions. The proposed solution used a long-term memory module (LSTM) to predict future uncertainties and formulated the coordination problem as Markov games solved by an attention-activated MADRL algorithm, where each EH or EVAGG is modeled as an adaptive agent focusing only on relevant pieces of state information. The MADRL approach involved centralized offline training to learn optimal strategies and decentralized execution, allowing agents to make online decisions based on local measurements. A safety network ensured the balance between supply and demand. Simulation results showed that this method performed comparably to traditional model-based methods with perfect knowledge of the system but with significantly shorter computation times—at least two orders of magnitude faster. The proposed MADRL method outperformed other MADRL approaches and the concurrent method, achieving lower energy costs of 10.79% and 3.06%, respectively, and higher aggregation benefits of 17.11% and 6.82%, respectively. In addition, the violation of electrical equality constraints averaged only 0.25 MW per day, indicating minimal and acceptable deviations.

To address the challenges of optimal microgrid planning with a focus on improving energy efficiency and environmental benefits while managing uncertainties related to renewable energy, load, and energy prices, Guo et al. [78] proposed a model integrating EV parking, a demand response program, and high penetration of RE (wind and solar) to improve financial and environmental objectives. Using conditional value at risk (CVaR) as a measure of risk, the model assessed the impact of uncertainties such as variations in PV and wind energy, real-time market prices, load fluctuations, and the behavior of EV drivers. Formulated as a two-stage linear stochastic mixed-integer model and simulated on a 33-bus smart microgrid, the results showed that integrating EVs and demand response reduced total operating costs by 9.97% and emission pollution by 12.34% and increased RE by 8.49%. The study also revealed that under a risk-averse strategy, higher levels of risk led to lower expected operating costs as the operator tended to minimize his dependence on the real-time market. The study also suggested that the synchronous use of EVs and demand response programs significantly benefited cost reduction, emissions control, and the efficient integration of renewables into smart microgrids, while suggesting future research into microgrid operations in island mode and under emergency conditions.

Huang et al. [79] presented an innovative method for accurately measuring the carbon emissions of connected and autonomous vehicles (CAVs) powered by RESs and non-RESs in urban environments. Using the SUMO agent-based simulation platform, researchers developed an intelligent driver model and cooperative adaptive cruise control modules to track different types of vehicles, including gasoline-powered vehicles (GVs), human-driven vehicles (HDVs), EVs, and CAVs. In addition, a lifecycle electric carbon emission model was constructed integrating EV energy consumption data with carbon emission factors from different energy sources. The model was validated by a case study in Suzhou, China, revealing that EVs can reduce carbon emissions by 70–90% compared to GVs at peak times, while CAVs can further reduce emissions by 35–50% compared to HDVs. The study also found that carbon emissions from non-renewable energy sources exceeded those from renewable sources. Although comprehensive, the study acknowledged certain limitations, including the need for real-world validation, the inclusion of more vehicle types, and the simulation of more complex traffic situations.

To address the challenges posed by the widespread adoption of EVs, particularly the pressure on charging infrastructure and the power grid, Ahmadi et al. [80] presented a multi-objective optimization framework for intelligent EV smart charging (EVSC) based on the dynamic hunting leadership (DHL) method. The purpose of this framework was to improve the grid voltage profile, eliminate voltage violations, and ensure that energy is supplied to EVs without interruption. It took into account residential EV chargers and parking stations, simulated realistic EV charging behaviors, and solved the issue as a mixed-integer nonlinear programming (MINLP) problem. Tested on a distribution network with different levels of EV penetration, the DHL algorithm proved effective in balancing conflicting objectives and improving the network voltage profile while respecting operational constraints. The study provided practical advice to aggregators and EV owners on optimizing charging processes to minimize technical impacts on the network. The results indicated that the framework can effectively manage EV charging schedules; avoiding voltage violations, blackouts, and peak load increases; and guarantee zero energy not supplied (ENS). The study highlighted the importance of smart charging for the sustainable deployment of EVs. It suggested future research into business models and vehicle-to-grid (V2G) technology for value-added services and additional revenue streams.

In order to meet the energy needs of households, including those equipped with EVs, Knežević et al. [81] examined the operation of an isolated microgrid that integrated RESs, such as wind and solar power, with energy storage systems. The system comprised a storage solution combined with batteries, hydrogen storage, and a control system optimized for efficient storage utilization. The stored energy was converted into electricity via fuel cells when renewable generation was insufficient. The study described the principles of electrolysis, modeled the system components, and described a control mechanism that improved storage lifetime by deciding on optimal storage utilization. The economic analysis calculated the discounted cost of stored energy and the net present cost of storage systems. A simulation using hourly Serbian data over one week evaluated system performance. The results indicated that the microgrid can operate independently and maintain stability using EVs for additional nighttime storage and hydrogen storage to balance the variability of renewables. The study concluded that such a hybrid system could hold promise for developing countries, with future improvements potentially involving advanced control technologies and predictive algorithms to improve energy production and consumption forecasts.

Aghdam et al. [82] explored the optimal operation of a multi-carrier virtual energy storage system (VESS), which included batteries, thermal energy storage (TES), power-to-hydrogen (P2H), hydrogen-to-energy (H2P), and EV technologies. VESS was analyzed for its ability to store surplus energy and supply energy in the event of a shortage with high efficiency and reliability. In addition, a demand response program (DRP) for both electrical and thermal loads was incorporated, as it functions similarly to physical energy storage systems. The research involved three market players—electric, thermal, and hydrogen—and addressed price uncertainties using stochastic programming. A two-level formulation was proposed in which independent system operators (ISOs) managed congestion at the top level while VESS operators focused on financial targets at the bottom level. This model was tested through four case studies on the IEEE 33 bus system using CPLEX in GAMS. The results showed that excluding certain components (thermal support, hydrogen system, and DRP) generated profits of $2781, $892, and $2927, respectively, while including all components increased the profit to $3119. Hydrogen had a significant impact on VESS revenues due to its higher price, while DRP had the lowest investment cost despite its minimal financial impact. In addition, optimal network reconfiguration effectively reduced line loads and managed congestion. The study concluded that VESS offers substantial financial and technical advantages and suggested that future research should include various uncertainties and the co-optimization of VESS owners, production units, and ISO on real-time markets.

The integration of vehicle-to-grid (V2G) technology with integrated energy systems (IESs) offers a promising solution for decarbonizing the energy and transportation sectors. Wei et al. [83] presented an optimization-based planning framework that integrated V2G with IESs, simulating the stochastic behavior of EV fleets. Using an improved NSGA-II algorithm for multi-objective optimization, the research assessed the economic and environmental impacts of V2G on IES design. Through six case studies covering three cities with different climates and two functional zones—residential and commercial—the results highlighted that the commercial case of Beijing generated the greatest mutual benefits. EV charging behavior aligned with time-of-use energy tariffs during transition seasons but not in winter. The sensitivity analysis revealed that electricity and gas prices significantly influenced system design. As the number of EVs reached 300, the economic and environmental benefits leveled off at 1.3% and 1.8%, respectively. The study demonstrated that integrating V2G with IESs improved both cost-effectiveness and emissions reduction, providing valuable information for IES planners, V2G service providers, and policymakers. Future research should explore the multi-agent benefits and advanced data processing methods to better model the behaviors of EV users and their interactions with IESs.

The rise of EVs has led to the emergence of EV aggregators (EVAs) that manage multiple charging stations, providing charging services to EV users with variable pricing based on time-varying electricity prices. However, existing studies on EVA strategies often focus on tendering systems, pricing, or single-period coordination without managing to integrate these strategies or make full use of the spatio-temporal change characteristics of EVs. Wang et al. [84] presented a new multi-period joint bidding and pricing strategy for EVAs that took into account interactions with distribution system operators (DSOs) and EV users. A two-level stochastic joint auction and pricing optimization model was developed, simulating the market equilibrium process with base load uncertainties at the lower level. To estimate the dynamic charging behavior of EV users under differentiated prices, a robust semi-dynamic traffic assignment model (SDTA) was built considering the coupling effect under traffic restrictions. An iterative method based on fixed-point theory was designed to achieve the optimal auction and pricing strategy, deriving an equilibrium solution by sequentially solving the two-level stochastic subproblem and the SDTA robust subproblem. This approach was shown to be effective in achieving a cost-effective and stable EVA strategy when integrated into the practical decision-making processes of DSOs and EV users.

Bamdezh et al. [85] explored a new hybrid thermal management system (TMS) for Li-ion batteries in EVs, which combined cooling water channels (active TMS) and a composite of kerosene phase change material (PCM) and aluminum foam (passive TMS) surrounding each cylindrical 18,650 lithium cell. The role of foam anisotropy, considering axial, radial, and tangential directions, was investigated using 3D numerical simulations in worst-case scenarios. The results revealed that the hybrid TMS operated in two phases: PCM recovery and cell cooling. Initially, water cooling mainly solidified the PCM from top to bottom, reducing its effectiveness in cooling the cell. Improving tangential thermal conductivity significantly improved control of average cell temperature and PCM solidification, while improving axial thermal conductivity positively affected the management of maximum temperature differences. This research highlights the importance of foam anisotropy in optimizing hybrid TMS performance for EV batteries.

Zhang et al. [86] examined the impact of the increase in EVs and the increase in renewable electricity on the midcontinent independent system operator (MISO) grid in the Midwestern United States from 2019 to 2039. Uncontrolled EV charging is expected to significantly increase peak loads by up to 10%, or 8 GW, and exacerbate the need for ramping. However, controlled charging, particularly unidirectional (V1G) and bidirectional (V2G) charging, can mitigate these problems. V1G recharging can avoid peak load increases and reduce ramp rates, while V2G recharging offers even greater flexibility, significantly reducing peak loads up to 23 GW and ramp rates. The study also highlighted the potential for multi-day optimization behavior in scenarios with over 3 million EVs, suggesting that new load planning tools could help to optimally plan resources with a higher share of intermittent renewables. The results indicated that by 2039, several million EVs could have a significant impact on net electrical loads if charged uncontrollably, with peak charging occurring between 5 p.m. and 8 p.m. Controlled charging, particularly V2G, could effectively reshape electrical loads, with V1G reducing peak loads by almost 7 GW. The study highlighted the importance of considering multi-day optimization for effective load management, as more than half of the months in 2038 showed full multi-day behavior. In addition, the study showed a linear relationship between the number of EVs and the average variation in net charge achievable with controlled charging. Future research will incorporate forecasting errors and combine peak shaving and ramp mitigation algorithms to further refine optimization approaches and assess costs compared to conventional solutions.

Ghadbane et al. [87] proposed a new strategy using the self-adaptive bonobo optimizer (SaBO), a metaheuristic optimization algorithm, applied to a Shepherd model for EVs. SaBO efficiently identified battery parameters, reducing the overall voltage error to 4.2377 × 10⁻³ and obtaining a root-mean-square error (RMSE) of 8.64 × 10⁻³. Compared with other optimization methods, SaBO demonstrated superior efficiency, with an optimization efficiency of 96.6%. SaBO’s performance was compared with other algorithms, including northern goshawk optimization (NGO), the zebra optimization algorithm (ZOA), RIME-based optimization, the arithmetic optimization algorithm (AOA), Chernobyl disaster optimizer (CDO), and dandelion optimizer (DO). SaBO outperformed these methods, achieving the lowest voltage error and the highest efficiency. The results indicated that SaBO is highly efficient for parameter extraction, offering significant improvements in accuracy and robustness. These results help to advance battery management systems, improve performance optimization, and develop sophisticated EM strategies for EVs.

Consumer attitudes towards RE technologies and awareness of energy consumption play an essential role in the adoption of sustainable energy systems in various sectors, including electricity, heating, and transport. Behavior change strategies are essential to accelerate the transition to carbon neutrality by reducing household energy consumption, promoting energy-saving technologies, and optimizing consumption patterns. Brown et al. [88] examine domestic consumer awareness on the island of Ireland, a unique case due to its single electricity market with two regulatory regimes. Conducted in March 2022 among 1373 respondents, the survey covered socio-demographic attributes, willingness to adopt new technologies, and eco-certifications. The results revealed that while many consumers favored EVs and expressed concern about carbon footprints and fossil fuel dependency, actual adoption of EVs and low-carbon heating systems remained low, hampered by factors such as cost and infrastructure. The island’s energy system, jointly regulated by the Republic of Ireland and Northern Ireland, offers flexibility and access to renewable energy, benefiting consumers through competitive prices and reduced emissions. Understanding consumer decision making is essential for a fair energy transition. Effective strategies include multi-level customer engagement, financial support for vulnerable groups, and targeted policies to encourage energy-saving behavior. Recommendations include promoting behavioral change, discouraging inefficient technologies, providing financial incentives for heating system upgrades, and offering subsidies for low-carbon technology installations to encourage energy savings and efficiency.

In the same context, lithium-ion batteries, commonly used in EVs due to their advanced cell chemistry, require efficient cooling systems to maintain optimum performance as energy demand increases. High temperatures can degrade battery performance, necessitating the development of various cooling techniques. Wankhede et al. [89] explored the use of nanofluids, in particular aluminum oxide (Al₂O₃), titanium dioxide (TiO₂), and copper oxide (CuO), as coolants in a liquid cooling method for lithium-ion batteries. The nanofluids, which feature improved thermal conductivity and low viscosity at higher temperatures, were compared with traditional coolants such as water and ethylene glycol. Using ANSYS CFX simulations, nanofluids significantly improved heat dissipation and absorption rates. Among the nanofluids tested, Al₂O₃ demonstrated the highest temperature reduction, followed by CuO and TiO₂. The study indicated that incorporating nanoparticles into the base fluid improved battery cooling efficiency, with nanofluids delivering a 14.7% greater temperature reduction than water. In addition, this modular cooling system can be easily integrated into EVs, offering a more efficient solution for battery thermal management. Future research should investigate the impact of different nanoparticle concentrations, coolant flow rates, and mass effects on cooling performance.

The integration of EVs into energy systems, particularly via intelligent parking facilities (IPLs), presents significant opportunities and challenges. Tostado et al. [90] proposed a two-stage optimal planning framework for energy communities that exploited IPLs as large collective storage systems via vehicle-to-grid capabilities. The framework used a stochastic model to represent the state of charge (SOC) of IPLs, taking into account uncertainties related to EV behavior, while upstream energy price uncertainty was managed using information gap decision theory (IGDT), allowing for strategic risk aversion. In addition, it was designed as a mixed-integer linear programming (MILP) model, guaranteeing efficient solvency. A case study demonstrated the effectiveness of this approach, showing that optimally managed MILPs can improve both the economic and operational efficiency of energy communities. However, the study also revealed that increasing the number of charging stations can significantly increase operating costs, underscoring the need for careful planning. Future research should focus on developing planning tools for IPLs, particularly as they develop within energy communities, to fully realize their potential as virtual generators in future energy systems.

Abrantes et al. [91] focused on the accurate determination of mutual inductance in inductive power transfer (IPT) systems for EV charging, which is essential for optimizing power transfer and battery charging efficiency, especially given vehicle positioning and misalignment. Using AI, in particular an artificial neural network (ANN), the study estimated mutual inductance using MATLAB/Simulink^® R2021b. The ANN was trained on a dataset representing 1% of the available data, and the results showed a maximum estimation error of around 3% at the lowest load power, indicating high accuracy. Notably, the ANR was fed with features that remain indicative of changes in vehicle position regardless of charging power. The study also revealed that excluding the amplitude of the third harmonic from the reactive power improved the ANN’s performance, increasing accuracy from 4% to 16% over different load power levels. Furthermore, the ANN demonstrated its robustness when trained with clean and noisy signals, maintaining error rates below 3%. These results suggest that the proposed ANN can efficiently estimate mutual inductance, thus improving power transfer optimization in IPT systems by adjusting for misalignment and varying load power conditions.

Miletić et al. [92] examined the benefits of household automation, different electricity pricing options, and participation in demand response programs promoted by the European Directive 2019/944. Using a mixed-integer linear programming model, they found that automated households saved more on electricity bills with real-time pricing (RTP) compared to time-of-use (TOU) pricing. Suppliers’ revenues were negatively affected by local power generation from PV installations, particularly when flexible technologies such as PVs and EVs were widely adopted. The study highlighted the importance of diversifying consumer portfolios and integrating batteries to stabilize supplier revenues. Furthermore, while dynamic pricing and the involvement of aggregators can reduce supplier profits, they offer significant advantages in managing consumption patterns. Further research is needed to generalize the results beyond the German context and idealized household behaviors.

Another study proposed by Kandoal et al. [93] explored smart charging for plug-in EVs, highlighting its potential for significant economic benefits through demand response (DR) programs. Unlike previous work, which assumed that EVs always respond to smart charging signals without problems, this research acknowledged uncertainties such as random EV mobility, volatile battery characteristics, and component failures. To address these, a feedback loop was proposed to predict EV charging behavior and adaptively adjust control signals. In addition, a distributed DR algorithm was introduced for optimal EV scheduling under load uncertainties while preserving user privacy and achieving fast convergence. Numerical simulations demonstrated the effectiveness of these methods in a low-voltage distribution network with variable EV penetration. The results suggested that feedback integration guarantees continuous operation and optimal performance even under unpredictable conditions and supports the evolution towards decentralized DR frameworks due to improved user privacy and reduced communication requirements.

Zhou et al. [94] presented a new methodology for quantifying the lifecycle carbon intensity of electrochemical batteries, which is essential for assessing their role in sustainable energy supply and grid integration. The research used a comprehensive multi-stage, multi-scale analytical framework that encompasses different stages such as battery materials (anode, cathode, and electrolyte), operational behaviors, cascading use, recycling, and replication. Through a case study in Guangzhou, the study revealed that the carbon intensity of EV batteries ranged from +556 kg CO₂,_eq/kWh under fossil fuel-based grid conditions to −860 kg CO₂,_eq/kWh when powered by solar and wind energy. In addition, mandatory grid recharging slightly increased carbon intensity to −617.2 kg CO₂,_eq/kWh, while the inclusion of embodied carbon from renewable sources raised it to −583.8 kg CO₂,_eq/kWh. The proposed platform enables a detailed and integrated analysis of battery sustainability, highlighting that the lifecycle carbon footprint is strongly influenced by operational factors and the energy sources used.

2.3. Power Systems and Microgrids

Huang et al. [95] explored the planning of power systems and microgrids, focusing on the role of energy hubs. They presented a multi-objective model integrating economic factors and pollution mitigation for microgrids, focusing on RESs such as wind and solar as well as advanced energy storage systems such as e-fuel. The study used sophisticated probabilistic modeling techniques to manage the uncertainties associated with renewable resources and included a demand-side management program to reduce costs at peak times. The model was transformed into an optimization problem solved using an innovative algorithm inspired by the foraging behavior of African vultures. A comparative analysis with established models revealed the ability of this approach to simulate market behavior and optimize profit strategies. Simulation results for four scenarios showed the method’s impact on operating costs and emissions, with Case 4 achieving the most significant reductions of 6% in costs and 7% in emissions. The study highlighted the model’s effectiveness in optimizing energy hub management, suggesting future research into scalability, real-world implementation, and integrating emerging technologies to improve energy hub optimization further.

Addressing the challenge of maintaining transient stability in power systems with an increasing number of RE plants, Carletti et al. [96] introduced a modified implementation of the virtual synchronous generator (VSG). The proposed method improved system stability during short-circuit events by virtualizing the behavior of resistive superconducting fault current limiters (SFCLs) through adaptive control, which adjusts the armature resistance of the VSG. Theoretical analysis using the surface equality criterion, PSCAD simulations, and experimental validation with a hardware-in-the-loop (HIL) test bench demonstrated the effectiveness of the method. The results indicated that adaptive control increased the critical compensation time (CCT) in proportion to the virtual resistance values, thus improving the system’s transient stability margin. This approach shows that virtualization of SFCL dynamics in VSG operations can significantly improve voltage response and fault stability, all without requiring modifications to the system topology.

Ibrahim et al. [97] addressed the challenge of forecasting electricity demand in the context of demographic uncertainties, industrial changes, and irregular consumption patterns by integrating machine learning with system dynamics. The proposed framework exploited advances in deep learning, in particular convolutional neural networks (CNNs), to improve the accuracy of demand forecasts. Three CNN variants (multi-channel CNN, CNN-LSTM, and multi-head CNN) were evaluated, with multi-channel CNN achieving the best performance (R² of 0.92 and MAPE of 1.62% for forecasting peak demand one month ahead). This model processed four input entities separately, generating a combined feature map. In addition, a system dynamics model was used to simulate the dynamic behavior of different demand sectors, successfully reproducing historical data with an R² of 0.94 and an MAPE of 4.8%. The framework aimed to improve energy managers’ decision making by providing accurate forecasts and strategies for managing variations in supply and demand, as demonstrated by a case study that also examined the impact of COVID-19 on power systems.

To improve power system restoration after disasters by leveraging the flexibility of internet data centers (IDCs) in IT load migration, Liu et al. [98] presented a novel two-level optimization framework. Unlike conventional methods, the proposed approach used Gaussian process regression to model IDC responses to power system decisions, thereby preserving IDC confidentiality and improving computational efficiency without sacrificing accuracy. The framework uniquely took into account load-side operations and different marginal load values, thus aligning more closely with real-world scenarios. Two case studies validated the model’s effectiveness and superiority in terms of computational efficiency over traditional methods, particularly as problem complexity increased. The results suggested that this method significantly improved the optimality and efficiency of restoration solutions, paving the way for a new direction in two-level collaborative optimization in power systems. Future work will extend the framework to incorporate RESs, on-site batteries, and generators into CDIs.

Liu et al. [99] addressed the stability problems caused by high proportions of RE generation, particularly wind power, which introduces non-linear elements into power systems. Traditional linear systems theory is insufficient to analyze these nonlinear influences. Therefore, the study used the description function (DF) method to evaluate the stability and power oscillations of a wind power system with a grid-connected permanent magnet synchronous generator (PMSG), incorporating a complete model that included a wind turbine, generator, machine-side converter (MSC), grid-side converter (GSC), and a weak grid. The study specifically modeled and analyzed the non-linear element of maximum power point tracking control in the power loop. The main findings revealed that MSC significantly affected the amplitude of oscillations, while GSC mainly influenced the frequency of oscillations. High bandwidths in the MPPT and speed control loops increased oscillation amplitude, while a fast MSC current loop improved stability. In addition, the bandwidth of the GSC voltage loop was critically important in determining the oscillation frequency, and a poorly tuned phase-locked loop (PLL) adversely affected system stability. It is also interesting to note that a stronger grid did not necessarily improve stability in the presence of non-linear elements. The DF method proved more accurate and practical than conventional linear methods, and the study’s findings were validated by MATLAB/Simulink simulations. Future work will extend the DF method to study other types of non-linearity, analyze high-order harmonics, and consider the impact of several non-linear elements on system stability.

Krismanto et al. [100] examined the influence of RE-based microgrids (MGs) on power system stability, focusing particularly on small-signal stability and modal interactions in the context of uncertain RE energy injections. The inherent variability of sources such as wind and solar power introduced unpredictability into the dynamic behavior of power system modes, potentially causing oscillatory instability, although MGs can also mitigate stresses on generators and transmission lines, thereby improving stability. Using Monte Carlo simulations and eigenvalue analyses, the study showed that critical modes behaved randomly under RE uncertainties but that cumulative distributions indicated that MG power injections generally enhanced stability. Presenting the modal interaction index (MII), the research evaluated the interactions between local, inter-zone, and MG control modes, finding that MGs can attenuate resonance and modal interactions. Case studies revealed that MG location affected local mode behavior, with higher ER injections improved the damping rate of critical modes. Analytical methods, including eigenpaths and cross-participation factors, confirmed modal interactions. Increased MG penetration tended to reduce local mode interactions but may introduce new interactions between zones and MG control modes at higher levels. The study highlighted the need to identify and manage these interactions to ensure stable power system operation as MG penetration and RE variability increase.

As a solution to the need for accurate, stable, real-time forecasts in smart grids, a new hybrid approach integrating feature engineering (FE) and modified firefly optimization (mFFO) with support vector regression (SVR) was developed by Hafeez et al. [101], called the FE-SVR-mFFO prediction framework. This framework aimed to overcome the challenges of computational complexity and parameter tuning in SVR, guaranteeing high accuracy, stability, and convergence rates in electric load forecasts. FE eliminates irrelevant features to improve computational efficiency, while mFFO optimizes SVR parameters to avoid local optima and achieve accurate results. The performance of the proposed framework was evaluated using half-hourly load data from five Australian states, demonstrating superior results compared to reference frameworks such as EMD-SVR-PSO, FS-TSFE-CBSSO, VMD-FFT-IOSVR, and DCP-SVM-WO. The FE-SVR-mFFO framework achieved lower mean absolute percentage error (MAPE), MAPE standard deviation (STD), and network response time, underscoring its effectiveness. These results underline its potential for ensuring reliable and secure power system operations, making it a valuable tool for political decision makers in the energy sector.

Electricity consumers face challenges in selecting optimal energy-saving plans, especially when integrating RESs that can destabilize the grid. Onile et al. [102] enhanced grid reliability and demand response goals by integrating battery energy storage system (BESS) technologies with reinforcement learning control and recommendation systems. A novel aspect was the use of separate active controllers for BESS technologies and user comfort loads. The adaptive demand-side recommender provided targeted recommendations, achieving a peak load reduction of 24.5% and improving comfort by 94%. Testing with a microgrid scenario showed a 24.5% reduction in energy consumption. This study proposed an energy-efficient behavior model for smart grid applications using an e-commerce demand-side recommender system (including rating prediction and top-N recommendations) and a multi-agent reinforcement learning control model. A new set of integrated energy services recommendations based on three categories (BAT_R, EFF_R, and REC) supported the continuous and sustainable operation of BESS. A vertical prototype design incorporated these recommendations alongside a classic demand-side recommender system, advising microgrid consumers on battery-friendly usage for hybrid PV-BESS systems. The proposed system was tested using a simple case scenario of nine households on a microgrid network, achieving up to a 24.5% reduction in overall energy consumption. The multi-agent reinforcement learning model adapted to increasing data availability, supporting a cost-effective demand response. Future work will address data limitations, dynamic environmental conditions, and the integration of vehicle-to-home systems.

To test and teach the control behavior of AC microgrids, Wang et al. [103] presented a new low-cost emulation system. Unlike traditional approaches, such as dynamic simulation testing and Power-Hardware-In-Loop (PHIL) configurations, which are costly and bulky, this system used a simple circuit configuration comprising two DC-AC converters connected to face-to-face and passive loads. This configuration can flexibly emulate various control schemes, including voltage control, droop control, current control, and secondary control. The emulation system featured a user interface on a host computer for real-time operation and measurements, facilitating the testing process. The proposed approach made it possible to test real controllers and loads, and complex microgrid configurations could be achieved by connecting several experimental platforms with different control schemes in parallel. Although the current system successfully validated typical control schemes, further research is needed to recreate more complex control behavior in multiple converters.

As a global approach to long-term microgrid planning, Mohseni et al. [104] focused on the integration of demand response (DR) programs and RE technologies. The proposed model used non-cooperative game theory and Stackelberg leadership principles to analyze the strategic behavior of energy utilities, DR aggregators, and consumers. It optimized trade-offs between imported electricity and available DR resources, determined cost-effective resource allocations, and promoted technologies such as power-to-gas and vehicle-to-grid. Applied to a 100% RE microgrid in Ohakune, New Zealand, the model demonstrated significant cost savings—around 21%, or $5.5 million—compared to traditional approaches. It used a new two-stage demand-side management market design, improving the accuracy of long-term investment planning by incorporating customer comfort and behavior into projections. The research highlights the model’s ability to improve the economic viability and flexibility of energy systems with high RE penetration, enabling a 32% reduction in total system costs. The study highlighted the importance of accurately predicting consumer participation in DR programs, particularly in remote areas, to ensure the success of fully renewable microgrids. Ultimately, the results suggested that this approach can lead to greater energy independence, security, and sustainability in rural and semi-urban areas.

To address the need for improved system stability and power regulation in microgrids based on virtual oscillator control (VOC), a comprehensive analysis of the dynamic stability of VOC inverters was carried out to guide parameter design by Peng et al. [105]. This study developed a dynamic phasor (DP) model for VOC-based single-phase microgrids, incorporating several control loops, and analyzed the stability regions of VOC parameters. The analysis revealed a strong coupling between VOC dynamics and wide-bandwidth control loops, including internal control loops and virtual impedance, distinguishing VOCs from droop-based microgrids. The results suggested that VOC operation in resistive mode offers a wider stability region for low-voltage applications, with a small η value, a large α value, and the inclusion of a virtual resistor recommended for robust operation. Based on these results, an improved VOC strategy with a hybrid power regulation structure was proposed to improve power sharing, fast voltage recovery, and stability margins. The effectiveness of this strategy and the recommended parameter design were validated by simulation and hardware-in-the-loop experiments, demonstrating superior stability performance and accurate power regulation in single-phase VOC-based microgrids.

In the same context, a bi-level programming model for real-time pricing (RTP) in demand-side management was presented by Yuan et al. [106], which aimed to balance electricity supply and demand while promoting efficient energy dispatch in microgrids. In the top level of the model, the main grid set prices and production to maximize supplier profits. The lower level addressed multiple microgrids operating in different modes with RESs, storage systems, and loads and optimized energy use for maximum user welfare based on dynamic prices. The model used a genetic algorithm (GA) for the top level and a branch-and-bound algorithm (BBA) for the bottom level, combined in a new distributed GA-BBA algorithm to ensure user privacy and computational efficiency. Simulation results using data from Guangxi, China, showed that RTP reduced the peak-to-average ratio from 1.6547 (fixed price) and 1.4513 (hour of use) to 1.2545 and reduced the peak-to-valley difference by 14.33%. In addition, total social welfare increased by 145.53 kCNY. These results showed that the proposed bi-level scheduling model effectively reduced peak loads and improved social welfare, providing a more realistic and reasonable approach to EM. Future research will explore multi-objective optimization in a multi-level electricity market and address the volatility of RESs using advanced mathematical techniques.

Shariati et al. [107] examined the thermodynamic characteristics of an innovative combined poly-generation plant that integrates cooling, hot water, and power generation load cycles fueled by a mixture of natural gas and biomass, where syngas enhances overall plant performance. Key subsystems included a gasification and combustion chamber, a double-acting absorption chiller, and a supercritical carbon dioxide cycle. Energy and exergy analyses reveal efficiencies of around 35% and 78%, with a net power of 7422 kW, hot water production of 14.71 kg/s, and an absorption chiller COP of 1.209. The gasifier and combustion chamber were identified as the least efficient components due to their high energy destruction rates. Parametric studies highlighted the positive impact of gasification temperature and combustion pressure on plant performance. Future research will focus on economic and environmental assessments and system optimization to improve applicability for urban and industrial use.

Sallam et al. [108] presented a detailed numerical model for direct steam generation in parabolic trough solar collectors, a promising technology for improving the performance of solar power plants but difficult to implement due to the liquid–vapor fluxes in the solar field. The model, validated against experimental data, was based on the conservation laws of liquid–vapor mixing and the energy balances of the receiver tube and lid. A parametric analysis examined the effects of inlet conditions (mass flow, pressure, and fluid temperature) on thermal-hydraulic behavior. The results indicated that a mass flow rate greater than 0.4 kg/s was essential to avoid stratification and that moderate pressure (10 MPa) balanced pressure drops and flow stratification. Simulations under the meteorological conditions of Ouarzazate, Morocco, suggested that regulating inlet flow according to solar flux and combining once-through and recirculation modes could ensure high-quality steam production. The study underlined those appropriate operating conditions can prevent deterioration of the receiver tube walls, and the developed model helped optimize and control parabolic trough collector technology.

Ketelsen et al. [109] explored the thermal behavior of electro-hydraulic compact drives (ECDs), focusing on their linear actuation potential in applications in the 5–10 kW power range. ECDs are renowned for their plug-and-play capability, energy efficiency, and compact size, making them promising alternatives to traditional valve-driven linear drives. However, as ECD applications expand to higher power ranges, passive cooling may become insufficient, requiring a deeper understanding of their thermal dynamics. To address this issue, the study presented a comprehensive thermal-hydraulic model based on the principles of conservation of mass and energy. This model was experimentally validated with data from an ECD prototype, demonstrating good accuracy in predicting steady-state and transient temperatures. The research indicated that while a more complex thermal resistance network offered slight improvements in accuracy, a simplified model was often sufficient for most applications, particularly during the early design phases, when limited system information was available. This simplified model can help system design engineers develop efficient thermal designs, supporting the wider application of ECD technology. The results of the study suggest that simplified thermal-hydraulic models are valuable tools for ECD design, contributing to their successful integration into higher-power applications.

2.4. Energy Storage

Due to the high temperature of their heat sinks and their ability to use existing industrial components, high-temperature heat pumps, particularly Brayton heat pumps, are essential for integrating renewable energies into power grids. As these systems are ideally suited to industrial applications and energy storage, demonstration plants have been developed to assess their performance and flexibility. The main challenges have been adapting to different load conditions and responding rapidly to load changes. Pettinari et al. [110] examined the transient behavior of Brayton heat pumps in thermal load management using a transient model of a prototype system. Two scenarios were evaluated: a sudden change in the desired heat sink temperature and a sudden change in the heat sink mass flow rate. The study analyzed the effectiveness of two control methods: compressor speed variation and fluid inventory control. The results showed that adjusting the compressor speed gave a response time of 8 to 20 min for changes in heat sink temperature, with the control time being limited by the maximum thermal stress on the heat exchangers. For changes in well mass flow rate, compressor speed adjustment provided faster response times (2 to 5 min) compared with stock control (15 min), although stock control achieved higher coefficients of performance (COP) and better stability under part-load conditions. This study highlighted the potential of a hybrid control strategy combining speed and inventory control to improve both temperature stability and COP during transient phases, offering a promising direction for future research into high-temperature heat pump systems.

Compressed air energy storage (CAES) systems are becoming more useful for utility-level uses because they can respond quickly. This makes them perfect for dealing with unpredictable renewable energy resources (RERs) and changing demand. Bafrani et al. [111] highlighted the potential of CAES to improve reserve provisioning capabilities, guarantee reserve deliverability, and manage operational uncertainties while maintaining economic efficiency. The authors proposed a mathematical two-stage optimization model for the optimal daily operation of generation units and CAES in the energy and reserves markets, incorporating a stochastic approach. The principal aspects of the model included constraints to guarantee the reserve capacity of the CAES in six operating modes and state-of-charge (SOC) limitations. In addition, generator reliability was taken into account in the planning of thermal generation units, responding to common monitoring. The model used information decision theory (IGDT) with a risk aversion (RA) strategy to handle the stochastic nature of demands and RERs, helping the independent system operator (ISO) to manage uncertainties. The formulation of the mixed integer nonlinear problem (MINLP) was solved using the CPLEX solver in the GAMS software. Applied to a six-bus test system, the model demonstrated its effectiveness, with results showing a reduction in total reserve of 21.34 MW and an increase in operating costs of $434.54 when taking into account the constraints of multi-hour reserve deliverability and an increase in total operating costs of $434.54, approximately 10% when including the reliability index.

Poli et al. [112] developed a detailed techno-economic model to assess the profitability of vanadium flow batteries (VFBs), using data from large-area multicell batteries and actual trends in financial markets. The model evaluated economic performance indicators such as capital cost, operating cost, levelized cost of storage (LCOS), and net present value (NPV). The analysis, based on a 20-year lifetime and daily charge/discharge cycles, showed that although current VFBs are not cost-effective in all energy and power duration scenarios, future technological and commercial advances could make them competitive. The study revealed that, under more favorable future conditions, including lower discount rates and higher electricity prices, VFBs could achieve profitability with investment costs as low as €260 kWh⁻¹ for an energy/power duration of 10 h. The results suggested that VFBs could become a viable option for long-term energy storage, particularly as RESs develop.

Silva et al. [113] explored the role of energy storage systems (ESSd) within active communities, highlighting the need for cleaner energy systems in response to climate change. Using mixed-integer linear programming optimization, the study planned distributed generation (DG), demand response (DR) programs, and ESS to manage uncertainties introduced by smart grid technologies. The focus was on analyzing the impact of ESSs using a clustering method to identify optimal ESS profiles. Using the silhouette method, the authors used the k-means clustering algorithm and determined the optimal number of clusters, finding that two clusters were optimal in various ESS datasets. The results highlighted that ESS status aligns with higher PV production periods and that strategic charging and discharging of ESSs during different price intervals can maximize revenues. An aggregator can mitigate the erratic nature of PV generation by integrating DR programs and ESSs, thereby reducing dependence on external suppliers and fossil fuels. The study highlighted the importance of ESSs in balancing the energy demands of active communities and improving financial and environmental results.

Chi et al. [114] addressed the inefficiencies of energy storage in China and the need for predictive models for heating and cooling loads in residential buildings, which are sensitive to climatic variations. The research focused on creating mesh-based mathematical models to predict these loads on a national scale. The methodology included establishing a digital mesh system for residential building zones, capturing air temperature data to create a hierarchy of thermal load intensity, and developing quantitative correlations between air temperature and thermal load intensity. The study revealed that annual heat loads generally increased from south to north in China, with significant increases in cooling loads in the south and decreases in heating loads in the northeast and west. Key equations were formulated to convert air temperature data into heat load intensities and predict total heating and cooling loads. These models aimed to reduce energy waste by enabling better planning and preparation of energy supply, thus improving the efficiency of energy use in residential buildings.

Liu et al. [115] explored the strategic involvement of virtual energy storage (VES) in electricity markets, focusing on its potential to offer flexibility and mitigate the challenges of RE integration. Common in China, VES involves large industrial electricity consumers with thermal generation sources that can self-supply and interact with the grid. The study introduced a bi-level stochastic optimization model to maximize VES profits in the day-ahead energy market with RE uncertainty. An all-scenario-feasible stochastic (ASFS) method was used to guarantee the decision-making feasibility of all scenarios. The model was tested on two illustrated systems and a practical case in Gansu Province, China. The results demonstrated the ability of VES to manipulate market power and take advantage of price arbitrage, revealing its effectiveness in different operating modes. The study found that VES can improve market flexibility, reduce the curtailment of renewables, and benefit both VES traders and society by optimizing the coordination of power generation.

Shabani et al. [116] addressed the challenges of lithium-ion battery storage in energy systems, focusing on complex degradation behavior, short life, high costs, and electricity market volatility. They proposed two operational planning strategies based on next-day battery behavior to maximize profitability and longevity in grid-connected residential applications with dynamic electricity tariffs. The first scenario focused on short-term profitability by optimizing charging and discharging times at three different rates (high, moderate, and low), with daily charging and discharging schedules as decision variables. The second scenario aimed to simultaneously maximize revenue and minimize degradation costs, incorporating decision variables such as cycle frequency, charge/discharge times, and durations per cycle. Both scenarios estimated battery performance, calendar and cyclic capacity degradations, remaining useful life, and internal states under real-life conditions until the battery reached its end-of-life criteria and was evaluated economically. The first scenario showed that the low charge/discharge rate extended battery life to 14.8 years but was the least profitable (−3 €/kWh/year), while the high and moderate rates generated positive profits (8.3 and 9.2 €/kWh/year) but shorter lifetimes (10.1 and 13.6 years). The high-rate strategy also resulted in a payback period of 1.5 years shorter than the moderate rate. In contrast, the second scenario achieved the highest profit (€18/kWh/year), the shortest payback period (7.5 years), and a considerable lifetime (12.5 years), demonstrating the importance of balancing revenues and degradation costs for sustainable profitability. These findings provide valuable information for decision makers to make informed strategic choices for effective battery management.

The development of light-rechargeable photo-batteries responds to the intermittent nature of solar energy by integrating energy recovery and storage into a single battery electrode. In this context, Pujari et al. [117] investigated the thermal effects contributing to the capacity increase of V₂O₅ and LiMn₂O₄ photocathodes under galvanostatic and photo-charging conditions. By implementing an improved experimental design and temperature-controlled measurements, it was found that capacity enhancements under 1 solar irradiation were largely due to thermal effects. Operando reflection spectroscopy revealed that the optical properties of photocathodes changed with their charge state, showing in particular that V₂O₅ lost its band gap after significant intercalation of zinc ions, rendering it ineffective as a photocathode below a certain voltage. The study highlighted the need to differentiate between thermal and optical contributions to capacity improvements by proposing a new cell architecture to separate these effects. These results are crucial for the rational selection and evaluation of materials for next-generation photo-batteries, ensuring accurate reporting of their performance under illumination.

Saxena et al. [118] examined the thermal management of lithium-ion batteries (LIBs) for electric and hybrid vehicles, focusing on a 3D numerical analysis of a battery module combined with a phase-change material (PCM) and a copper metal foam (MF) composite. The research modeled a single prismatic cell and a 7S1P battery pack under conditions of rapid discharge, realistic driving cycles, and thermal abuse, comparing thermal equilibrium and non-equilibrium states. The results revealed that an 8-mm-thick MF-PCM composite with a porosity of 0.95 effectively reduced the maximum cell temperature and maintained uniform temperature gradients, significantly outperforming natural convection cooling. The composite limited temperature rise and eliminated hot spots, demonstrating superior thermal performance in a variety of tests, including continuous rapid discharge and aggressive duty cycles. The study highlighted the PCM module’s excellent heat absorption capacity, effective temperature control under different short-circuit resistances, and optimal performance in a 7S1P configuration, suggesting that this approach can improve battery thermal management in practical applications.

Chayambuka et al. [119] presented a P2D gravimetric intermittent titration technique (GITT) model coupled with grid search optimization to accurately determine the solid-state diffusion coefficient (D1) and electrochemical kinetic rate constant (k) at different state-of-charge (SOC) points for a sodium-ion battery (SIB) cathode. The model, validated against GITT experimental steps, provided accurate information on the intercalation dynamics and rate capability of the SIB. Unlike traditional methods, the P2D GITT model eliminated unrealistic assumptions, provided a more accurate description of porous electrodes, and allowed for longer GITT current pulses, thereby reducing experimental run times. The model’s ability to simultaneously derive and validate D1 and k parameters from a GITT dataset improved its reliability and applicability. The approach is adaptable to any porous battery electrode material, making it a valuable tool for improving electrode design and performance modeling in battery research.

The study presented by Sharma et al. [120] aimed to advance sustainable waste-to-energy solutions and address global challenges in energy production and plastic waste disposal. It examined technologies for converting plastic waste into energy, focusing on the integration of renewable sources such as solar power to improve sustainability. It included a simulation model to assess the electrical performance of these renewable systems under varying environmental conditions. The main techniques covered included pyrolysis, which efficiently converts plastic waste into energy, char, gas, and oil, and hydrothermal carbonization, which produces hydrochar that can be used in supercapacitors and batteries. The study highlighted the need to advance technology and use cost-effective methods while balancing environmental and health impacts. It also highlighted the importance of life cycle assessment (LCA) in guiding plastic waste management and policy decisions.

Molina et al. [121] explored the potential for energy recovery from human body movements, focusing on an electromagnetic system designed to capture energy from lower limb joints during walking. The system was modeled and simulated with scenarios involving the hip, knee, and ankle joints using kinematic data from walking studies. Applying a state-space representation and recurrence plots, the research evaluated the interaction between the electromagnetic recuperator and the power conditioning circuit. The results indicated that, at a constant walking speed of 4.5 km/h, the energy generated varied from 1.4 mW at the ankle joint to 90 mW at the knee joint without the use of multiplier gear. The results confirm the feasibility of an electromagnetic transducer for converting angular kinetic energy into electrical energy suitable for low-power devices. Experimental data showed that the knee joint offers the highest power output, while the other joints vary considerably in terms of energy production. The system stabilized within 10 to 15 s, and recurrence plots provided valuable information on its dynamic behavior. The study suggested that refining the control strategy of the power conditioning circuit could improve device performance, given that a fixed PWM duty cycle was used in this research.

Arulprakasajothi et al. [122] examined the use of silicon dioxide microparticles embedded in paraffin wax in a compact salinity-gradient solar pool to improve heat storage capabilities during the winter. Silicon dioxide, known for its cost-effectiveness, abundant availability, and excellent heat transfer properties, is used to improve the performance of solar heating systems. The research involved evaluating the thermal behavior of a solar pond with and without silica microparticles. Initial experiments revealed that without any heat storage medium, temperatures in the three zones of the solar pond were 31.2 °C, 32.8 °C, and 35.7 °C. When paraffin wax was used alone, the temperature in the lower convection zone was 4.2% lower due to less efficient heat storage. Incorporating silica microparticles into the paraffin wax improved thermal performance, raising the temperature in the lower convection zone to 56 °C, a 1.5% improvement over using paraffin wax alone. The optimization analysis revealed that the design of the solar pond played a more important role in temperature control than temporal factors. In the end, the study showed that the addition of silica microparticles to phase-change materials significantly improved the heat storage and thermal performance of compact salinity-gradient solar ponds, suggesting potential applications in various thermal systems such as vapor absorption refrigeration, building heating, air conditioning, and small power plant heat exchange.

2.5. Renewable Energy

Bottecchia et al. [123] presented the multi-energy systems simulator (MESS), a new open-source energy system model designed to simulate energy systems without searching for optimal solutions. In contrast to many existing models, which focus on finding optimal solutions within given constraints, MESS allows non-optimal solutions to be studied, offering a more realistic representation of energy systems, particularly at the urban level. Through a comparative analysis with Calliope, a state-of-the-art optimization model, the article highlighted the differences and potential advantages of the simulation approach. The results showed that MESS and Calliope produced consistent annual results despite their different methodologies, suggesting that simulations can effectively analyze urban energy systems. MESS was particularly advantageous for rapidly exploring several alternatives and scenarios, thus enhancing transparency and participatory processes in energy planning. However, the study also recognized that optimization models may be more appropriate for the design of technical systems. The authors’ future work will focus on expanding the MESS technology library, integrating spatial dimensions, and analyzing the impacts of various energy policies to further improve urban energy system planning and analysis.

The challenges of planning and configuring a coal-fired power-plant-based renewable-power-to-gas system (CP-PtG), which integrates electrolytic hydrogen production and carbon capture to cope with fluctuating renewables, were investigated by Zuo et al. [124]. In this context, a two-layer stochastic optimization method was developed to handle the uncertainties and slow dynamics of the system. The top layer used a genetic algorithm to determine the optimal system capacity, while the lower layer used mixed-integer linear programming (MILP) to minimize daily operational costs, considering thermal dynamics as constraints. In terms of optimization results, the consideration of uncertainty increased the capacity of key equipment and thus improved operational stability, while dynamic considerations significantly reduced tracking errors and thus ensured economic and stable operation. Future work should incorporate more dynamic features and explore intelligent optimization algorithms for greater computational efficiency.

Rapidly increasing electricity consumption requires effective EM strategies to meet demand while minimizing costs and peak loads. Ismail et al. [125] presented SPEMS (sustainable parasitic energy management system), a multi-agent system (MAS) that optimizes energy costs, consumption, peak-to-average ratio (PAR), and user discomfort. Using a host–parasite model, SPEMS plans the use of smart appliances to balance conflicting objectives through a symbiotic agent relationship. The model takes into account environmental factors such as user presence and weather conditions without requiring daily user feedback and incorporates a “sampled Hall of Fame” to maintain and use the best planning solutions from previous days. Experimental data from a real home over a year and a half demonstrated that SPEMS was IT-efficient, significantly reducing costs and energy consumption while preserving user comfort. This was achieved by shifting device usage to off-peak hours and adapting to user behavior over periods of 5, 10, 15, and 20 days. The system reduced high-cost peaks by up to 60% compared with non-optimized schedules and kept user discomfort within a range of 5–35% despite variable weather conditions. Future work will integrate RESs, expand the dataset to include more households, refine time slot splits, and explore SPEMS scalability. In addition, the development of distributed SPEMS will enable smart neighborhoods to manage energy collaboratively, improving overall efficiency and automation through advanced smart technologies.

Cruz et al. [126] examined the role of energy communities in the transition to sustainable energy by looking at residential electrical load profile datasets and consumer survey results to identify controllable loads and develop tailored energy consumption models. The analysis exploited these datasets to improve demand-side management (DSM) systems. Patterns of device behavior, particularly for high and flexible loads, were identified and validated in three use cases to support efficient DSM in various contexts, including the presence and absence of RE supplies and bill-saving scenarios. A genetic algorithm was used to optimize flexible demand reallocation and community load profiles by integrating time-varying tariffs. Experiments revealed that managing controllable and modifiable appliances can reduce average peak load by up to 29%, increase renewable self-consumption, and achieve energy bill savings of 9%. However, the study also highlighted the limitations of existing load and consumption datasets, which currently hinder the effective design of demand-side management and demand response programs in energy communities.

In the same context, Ur Rahman et al. [127] focused on the control of microbial electrolysis cells (MECs), a promising RE technology for the production of hydrogen from biomass. Given the non-linear dynamics of MECs, effective feedback control is crucial for optimal hydrogen production. The authors developed a nonlinear dynamic model for MECs, linearized it, and derived a linear time-invariant transfer function. They proposed a robust fixed-structure controller that achieved fast settling time, zero overshoot, and zero steady-state error, demonstrating its effectiveness in the presence of parametric uncertainties, measurement noise, and disturbances. In addition, they introduced an anti-windup control strategy to avoid integral liquidation errors, which had not been addressed in previous research. The proposed controller outperformed existing models, offering a faster and more robust response. Future work will explore decentralized control of continuous MEC processes and adaptive linearization using AI.

The emergence of citizen energy communities (CECs) within liberalized retail energy markets offers consumers opportunities to improve local carbon neutrality and sustainability through decentralized production. By reducing their dependence on transmission networks, these communities can cut costs and benefit from legislative incentives such as tariff reductions. Algarvio et al. [128] presented a comprehensive model for CECs, including local investment in renewables, competitive tariff designs, and strategic bidding on wholesale markets. A consumer optimization model selected the most cost-effective retail tariff. Using data from Portuguese consumers and the Iberian electricity market, the study revealed that inflexible consumers can reduce costs by up to 29% by adhering to a CEC and by over 50% with demand response strategies that align with local generation and wholesale prices. The results of the study highlighted that CECs can significantly reduce their costs through strategic tariff selection and investment in local generation, although a significant proportion of their expenses still come from trading, balancing, and network costs.

Xiao et al. [129] presented a two-level model of strategic behavior in which the top level focused on profit maximization by implementing strategic constraints and pricing, while the bottom level evaluated revenue outcomes for the day-to-day market equilibrium. This model resulted in an equilibrium problem with equilibrium constraints (EPEC), which revealed a new market equilibrium influenced by these strategies. The results of the study indicated that RE companies are not very sensitive to the costs of private ES, suggesting widespread use of these systems to improve their market position. Imposing strategically convenable strategies could enable RE companies to expand their market share and increase profitability by modifying production limits and capitalizing on price differentials. The study showed that private ES plays an essential role in enabling greater integration of renewables, with significant impacts on social welfare, energy pricing, and overall system functioning. The results highlighted the importance of policy constraints and pricing in shaping market equilibria, with different effects observed under different market mechanisms, such as locational marginal price (LMP) and substitute energy price (SEP). Future research will explore further complexities, including the integration of shared and private ES, geographical constraints, and joint optimization strategies in electricity-carbon markets to advance clean energy goals.

To address the growing need for flexibility of gas-fired units due to the increasing penetration of renewable energies, Jiang et al. [130] reinforced the interdependence between the electricity and natural gas markets. They proposed a two-level strategic bidding model in which a gas-fired unit acts both as a strategic electricity producer on the electricity market and as a strategic gas buyer on the natural gas market. The higher-level problem aimed to maximize the gas-fired unit’s profit, while the lower-level problems managed the equilibrium of the electricity and natural gas markets, interacting through localized marginal prices. The two-level model was converted into a mathematical problem with equilibrium constraints (MPEC), then into mixed-integer second-order conic programming (MISOCP) for an efficient solution. The numerical results showed that this methodology effectively simulated the market equilibrium process and strategic behaviors of gas-fired units in interconnected energy markets. Key findings included a 30.7%, 5.6%, and 2.6% increase in gas-fired unit profits compared to other bidding methods, achieved by modeling the impact of strategic auctions on market prices. In addition, the proposed strategy significantly reduced the second-order cone (SOC) relaxation gap, demonstrating the model’s ability to address the complex interdependencies of energy markets.

Microalgae, as a promising RES, face challenges in biorefinery operations, mainly due to their dependence on climatic conditions such as temperature and sun exposure. Lim et al. [131] addressed these challenges by using a forecasting algorithm to predict daily weather conditions one year in advance, which was then applied in a dynamic metaheuristic optimization framework to determine the most efficient microalgae biorefinery processes. The optimized system achieved an annual margin of $815,716 and reduced greenhouse gas emissions to 1.1 × 10⁷ kg CO₂ equivalent per year. By evaluating various microalgae species and optimal harvesting schedules, the study identified the most suitable species based on climate-related growth patterns. An uncertainty analysis using Monte Carlo simulations revealed that the performance of the proposed configuration was robust to climatic variations, with deviations in the total annual margin limited to 5%. The integration of RESs, including wind (98.7%), solar (0.1%), and biomass (1.2%), further reduced greenhouse gas emissions by 50%. This research provides detailed guidelines for the large-scale deployment of microalgae biorefineries, offering an adaptable model for different regions and conditions, although future work should include a wider range of products and detailed experimental validations.

To meet today’s energy challenges in buildings, emerging technologies such as thermally activated building systems (TABSs) are attracting attention for their energy-saving potential. Guerrero et al. [132] focused on a new prefabricated TABS panel for residential facades, featuring high thermal inertia due to phase-change materials. To evaluate the system’s performance and potential energy savings, a two-stage approach was used. The first step used performance maps derived from CFD simulations to assess system behavior under ideal conditions. The second stage involved integrating a TABS into a simplified building model to estimate monthly energy demand and potential savings in different Spanish climatic zones. The results showed that the system’s high thermal inertia enabled it to discharge energy over several days, even with a limited load, demonstrating its adaptability to the constraints of renewable energies. The study revealed that TABSs could reduce heating demand by over 40%, even under less favorable conditions. The study highlights the system’s efficiency, in particular its compatibility with renewable energies and its ability to maintain high efficiency within standard HVAC ranges. It also noted the importance of phase change material (PCM) proportions and phase change temperature selection, suggesting that the optimum PCM content ranges from 5% to 15% for cost-effective energy results.

Buonanno et al. [133] presented a robust tool designed to generate solar irradiance profiles, essential for optimizing the operation of distributed energy resources (DERs) in smart grids. Due to the uncertain nature of RESs, such as solar energy, it is essential to accurately forecast and integrate these uncertainties into DER planning for optimal performance. The proposed tool used historical solar irradiance data and employed the roulette wheel mechanism to initially create a diverse set of scenarios. It then refined these scenarios using the fast-forward method to retain the most representative profiles while minimizing the computational load for subsequent stochastic optimization. The application of this tool to a case study in Turin, Italy, for January and July demonstrated its effectiveness in generating plausible solar irradiance scenarios. The tool’s flexibility enabled it to model a range of fluctuating solar patterns, improving the accuracy of DER optimization. A sensitivity analysis revealed that key parameters, such as the number of regions and the metrics used for outlier removal, had a significant impact on scenario regularity and variability, providing valuable insights for scenario definition. This method can be adapted to a variety of locations and applications and offers potential as a web service enabling users to generate and query solar irradiance scenarios, which is essential for improving stochastic DER optimization.

Wang et al. [134] presented a deep neural network approach for planning investment decisions in the low-carbon transformation of power grids, which is essential for sustainable development and the fight against global warming. The approach integrated three targeted investment branch models: investment behavior, electricity production and consumption, and investment forecasts for new capacity. These models addressed the challenges of electricity distribution, pricing, carbon allowances, and the feasibility of low-carbon technologies. Using spatiotemporal and recurrent neural networks, the study built a comprehensive investment decision-planning model that integrated existing data on low-carbon technologies. The results showed that the method accurately predicted future low-carbon technology installed capacity, sustainability indices, and investment returns, achieving an accuracy of over 90% compared to actual values over four years. The results underlined that optimal power generation portfolios improve stability and investment returns while reducing the risks associated with RE generation. The study highlighted the need for comprehensive risk assessment in economic investment planning and suggested that future research include diverse regional data and transfer learning strategies for wider applicability.

Soil microbial fuel cells (SMFCs) have emerged as a promising, low-cost, carbon-neutral technology for pursuing a carbon-free future. Within this framework, Dziegielowski et al. [135] investigated the voltage evolution of flat-plate, membraneless SMFCs, examining both anodic and cathodic potentials as well as the impact of organic matter content and soil porosity on voltage dynamics. Four distinct voltage evolution stages were identified, significantly influenced by soil properties. Notably, increasing the organic matter content from 10% to 50% almost tripled the output voltage, with the anode potential reaching −450 mV, triggering an exponential increase in the cathode potential and achieving a power density of 68 mWm⁻². Experimental results informed the development of a new computer model capable of predicting the electrochemical behavior of SMFC in various soil types, thus providing a powerful tool for optimizing power generation. This research laid the foundations for system optimization and the practical application of SMFCs, highlighting the critical role of soil composition and operating conditions in achieving sustainable energy production. Specifically, soils with a higher clay content, which is less permeable to oxygen, facilitated a rapid increase in voltage by rapidly lowering the anode potential. In addition, a well-functioning anode with a potential of −450 mV significantly improved cathode performance, probably due to the formation of catalytic biofilm on the cathode surface, resulting in substantial voltage increases under closed-circuit conditions. Soils with high organic matter content, particularly at 50%, demonstrated superior performance, generating 550 mV compared with 320 mV and 200 mV in soils containing 19% and 10% organic matter, respectively.

Kim et al. [136] investigated the impact of water management on the performance of polymer electrolyte membrane fuel cells (PEMFCs) using a rotating circular spiral channel. Numerical and experimental analyses revealed that water production and behavior, influenced by channel design and rotational speed, are critical to fuel cell efficiency. Three circular flow field models (CASE #1, CASE #2, and CASE #3) were analyzed, showing that increased water content improved ionic conductivity and current density. Experimental results showed that optimal rotation speeds (up to 100 rpm) improved performance by effectively removing excess water through centrifugal force, while excessive rotation (125 rpm) reduced performance due to excessive drainage. The study concluded that controlled water drainage significantly improved PEMFC performance, increasing power density by around 75%.

To optimize the damping parameters of wave energy converters (WECs) to mitigate line forces under extreme wave conditions, accurate predictions of future wave and system states are crucial. For this purpose, Shahroozi et al. [137] presented a sophisticated approach using deep neural networks (DNNs) to predict surface elevation and system state, thereby predicting maximum line strength. A two-stage neural network model was used in which one network predicted wave elevation, another predicted system states based on these predictions, and a convolutional neural network (CNN) estimated maximum line strength. The results indicated that although the DNN model for surface elevation achieved high accuracy, the prediction of the position of the power take-off translator was crucial for the accurate prediction of maximum force, with sensitivity to uncertainties in this prediction. A comparative analysis showed that DNN outperformed traditional autoregression (AR) and Kalman filter methods, notably by adapting to real-time variations and minimizing the need for extensive pre-training. The study highlighted that, despite uncertainties, neural networks offer robust predictive capabilities for WECs, underscoring their advantage over conventional models under dynamic and varied system conditions.

He et al. [138] presented a new three-dimensional yaw wake anisotropic model to improve the prediction of wake behavior behind wind turbines during yaw operations, aiming to improve yaw control strategies that mitigate power loss and structural vibrations. The model incorporated general expressions for wake expansion rates and a deflection term to accurately represent wake evolution and trajectory. Validated by public measurements and wind tunnel experiments using particle image velocimetry (PIV), the model demonstrated superior accuracy in predicting wake characteristics, such as width and maximum deficit, compared with existing models. The model’s versatility was demonstrated by its accurate predictions for high- and low-thrust wind turbines and various downstream locations, capturing the elliptical wake cross-section resulting from yawing operations. Importantly, the model used readily available parameters such as input velocity, turbulence, and thrust coefficient, guaranteeing its generalizability. The wake deflection model, validated against wind tunnel data, accurately predicted the trajectory of the wake centerline, particularly at high yaw angles and in the main forward 7D region of interest. This analytical model represented the first attempt to provide general, anisotropic wake expansion rates, making it a reliable and cost-effective tool for optimizing yaw control strategies in wind farms.

2.6. Smart Cities and Rural Communities

Akiyama et al. [139] presented a new edge computing system designed to enhance the robust automation of personal mobility vehicles, a crucial aspect for the advancement of smart cities. While previous efforts focused primarily on pedestrian safety, this system addressed the imperative of ensuring the safety of personal mobility vehicles, particularly in scenarios where on-board sensors may be compromised due to a variety of external and internal factors. By deploying multiple light detection and ranging (LIDAR) sensors similar to roadside or indoor security cameras, the proposed system leveraged advanced computing to facilitate real-time data awareness and analysis. A prototype system was developed and evaluated using real LIDAR units and a mobility scooter, demonstrating the feasibility of the approach, particularly in situations where on-board sensors were disabled. Through experimental testing, it was observed that the system enabled the personal mobility vehicle to continue driving even when its on-board sensor was compromised, with route errors lower than those of conventional systems under certain conditions. Future work will include assessing the system’s robustness to communication delays and losses, highlighting its potential importance for improving the safety and efficiency of personal mobility vehicles in smart city environments.

Lu et al. [140] addressed the energy crisis and environmental pressures by proposing a household EMS involving a smart residential energy hub (SREH) that integrated electricity and natural gas. The system used models of energy-consuming equipment and control strategies based on physical characteristics and user preferences. It formulated a multi-objective optimization problem to efficiently allocate energy supply, aiming to minimize both energy costs and comfort differentials. Through four case studies, the model showed a significant reduction in energy costs and slight improvements in comfort differentials, demonstrating its robustness in the face of uncertain user behavior. The study highlighted the adaptability of the system to different household preferences and the importance of scenario analysis for efficient calculations. Future research will focus on real-time adjustments and the cooperative optimization of several SREHs.

In another context, Shuhan et al. [141] aimed to strengthen the cybersecurity of smart cities through three integrated initiatives. They suggested a CycleGAN-based data architecture to protect against false data injection attacks (FDIAs) by changing the data distribution in a cycle. This kept the data integrity and reliability. Furthermore, they introduced an innovative IoT concept to improve real-time monitoring and contextual threat detection within intrusion detection systems (IDSs). They used IoT-generated data to comprehensively analyze network activity and adaptive threat response. Finally, the researchers developed an IDS hyperparameter tuning system combining biogeography-based optimization (BBO) and the whale optimization algorithm (WOA), inspired by biogeographic principles and whale behavior, to balance exploration and operation in IDS configuration. Empirical evaluations on real smart city data showed that the CycleGAN-based IDS system with hybrid metaheuristic hyperparameters achieved a high FDIA detection rate of 95%, a low false positive rate of 5%, and the fastest response time of 2 min. These results demonstrated the effectiveness of using advanced machine learning techniques and optimization algorithms to improve smart city cybersecurity against evolving threats.

Smart grid systems require sophisticated analysis due to the interaction of various domains and actors, making co-simulation frameworks essential. A new multi-agent co-simulation framework was introduced by Rando et al. [142], offering a test bed for various smart grid strategies, in particular, demand response programs that exploit the thermal behavior of residential buildings. The framework was modular, flexible, and configurable, enabling realistic scenario evaluations. In a case study involving 1000 buildings, the platform effectively mitigated electrical imbalances at the network’s main substation by making minor adjustments to indoor temperature setpoints. By integrating the buildings’ thermal model, physical power grid model, HVAC systems, agent-based communication infrastructure, and demand response strategies, the framework guaranteed maximum flexibility of use through modularity, plug-and-play, and scalability. The smart grid, considered a multi-agent system, enabled distributed infrastructure and problem decomposition without a central coordinator. This platform simulated various urban scenarios and integrated different players in the exchange of electrical energy at the distribution level using real data and archetypes. It supported the addition of external modules, new agents, and the testing of different scenarios and strategies. Tests of three strategies for electrically activated thermal loads demonstrated the effectiveness of slight internal temperature deviations in achieving good balancing results without additional sources of flexibility, such as storage. The test bench framework, scalable to 1000 buildings, maintained detailed physical descriptions of technologies such as heat pumps, building envelopes, domestic distribution, emission systems, and the power grid, providing a robust environment for evaluating flexibility strategies.

Dalla Mora et al. [143] evaluated different methods of estimating space heating energy demand in residential neighborhoods, focusing in particular on a historic area of Italy. By comparing advanced dynamic simulation tools—the energy urban resistance capacitance approach (EUReCA) and city energy analyst (CEA)—with a more accessible quasi-stationary method using Excel, the researchers identified both the strengths and limitations of these approaches. The quasi-steady-state method provided reliable estimates for basic factors such as building geometry and infiltration but failed to take internal loads into account, often underestimating energy demand. Despite this, it remains a practical tool for non-specialists due to its simplicity and lower calculation requirements. The main findings highlighted that solar heat gains have a significant impact on heating demand, with the simplest method overestimating these gains. The study suggested that while dynamic tools offer greater accuracy, the quasi-steady-state method remains valuable for preliminary assessments and planning, with future research focusing on extending these methods to cooling and domestic hot water requirements.

Yu et al. [144] examined the integration of bike sharing with metro stations to solve the first and last mile problem using Beijing as a case study. They explored usage patterns at macro and micro levels, revealing that bicycle and car trips exhibit power-law distributions with different exponents on weekdays and weekends. The study identified that demand for bike sharing follows scale-free behaviors across the city, with variations in usage patterns at different times of the day, such as morning and evening rush hours. This information contributed to understanding the dynamic deployment of bike-sharing systems in conjunction with public transport. However, the study was limited by its reliance on data from a single provider (MoBike) and a short one-week period, which may not fully capture user preferences or be generalizable to other cities. The analysis also faced difficulties in accurately identifying bicycle and car trips due to limitations in buffer size and the lack of explicit data on metro access trips. The study also pointed to the importance of using long-term, multi-source datasets and taking into account additional factors such as land use and transport infrastructure to improve the reliability of the results. It also highlighted the importance of the structure of metro networks and suggested that areas with a low density of metro stations could benefit more from bike sharing as a mass transportation mode. Despite its thorough examination of spatio-temporal travel demand, the study recognized the need to further explore issues such as connection interpretation and metro bridge structures.

In the context of rural communities, Ibrik et al. [145] presented the impact of microgrid solar PV systems in rural areas of the West Bank, Palestine, highlighting their potential to improve rural electrification and development. These systems can improve social services, water supply, agriculture, and environmental sustainability by using PV panels at a decreased cost. Through case studies in Dir Ammar and Al-Birin, the study highlighted how sustainable energy availability promotes access to water, agricultural productivity, and food security. The analysis demonstrated that micro-grid solar PV systems are more cost-effective and environmentally friendly than diesel engines, offering long-term solutions for energy, water, and food security in rural communities in Palestine.

Raza et al. [146] assessed the socio-economic and environmental benefits of PV systems operating high-efficiency irrigation systems (HEISs) in Punjab, Pakistan. Field research indicated that PV-powered HEISs have significantly reduced dependence on diesel, resulting in annual savings of 6.6 million liters of diesel and a reduction in CO₂ emissions of 17,622 tons per year. Farmers experienced a 100% increase in income and water savings of 41%, with the unit cost of the PV system significantly lower than that of subsidized electricity and diesel. PV systems have improved agricultural productivity in remote, water-poor areas, with farmers expressing great satisfaction and recommending continued government subsidies for these systems.

Onu et al. [147] investigated an integrated solution to energy and water scarcity in rural African communities through a design that combined PV power generation with a pumped hydroelectric storage system. The study highlighted the seasonal and geographical challenges of renewable energies and underlined the importance of energy storage systems in mitigating these problems. The proposed system used the natural elevation of a body of water to store hydraulic energy, guaranteeing a reliable power supply and water for irrigation with a 99.9% probability of power supply. The design also aimed to improve the economic sustainability of rural power grids by focusing on agricultural and micro-industrial activities, thereby improving the socio-economic conditions of the community. A cost analysis revealed significant operating and maintenance costs, but these can be mitigated by community training. The study highlighted the potential of this integrated approach to foster economic development and reduce energy poverty in isolated communities.

The main challenges facing rural areas in the Mediterranean are the reduction of agricultural land, the impact of climate change, and the demand for resources such as energy, which is vital for development. Highlighting the crucial role played by rural areas in food security and sustainable development, Abouaiana et al. [148] introduced the concepts of energy communities and agrivoltaics as key strategies for enhancing the positive impact of land use in buildings and farms on the ecosystem. Focusing on Egypt and Italy as representative case studies, the researchers examined two rural settlements—Lasaifar Albalad in Egypt’s Delta region and Pontinia in Italy’s Lazio region—both characterized by agriculture-based land use. Through focus group discussions with various stakeholders, the study analyzed the alignment of these key concepts with national rural and agricultural policies, fostering a new collaborative approach between the two countries. The results offered the first comparative analysis of these contexts, improving understanding, increasing social acceptance, and identifying significant obstacles, paving the way for future micro-scale field practices.

In the same context, Mengi et al. [149] addressed the challenges of climate change and the growing presence of “food deserts” in rural and urban areas, highlighting the need for year-round availability of fresh produce in communities at the end of traditional agricultural supply chains. Indoor vertical farming was presented as a next-generation agricultural method that offers advantages such as reduced water and pesticide use, higher yields, consistent quality, and lower transport and harvesting costs. Unlike industrial greenhouses, small-scale “pod farms” can be deployed in areas where large plots of land are unavailable or too expensive. These pods, usually the size of a shipping container, use hydroponic systems and LED lighting to grow plants, making them fundamentally different from traditional greenhouse farming. Although many of these indoor farming pods claim high energy efficiency, little analysis or optimization has been carried out to validate these claims. To address this, the study developed a genomic optimization and digital twin framework to improve the optical design of vertical indoor agriculture pods. Using ray-tracing methods and a genetic algorithm, the research optimized LED source size, beam propagation, and power requirements to maximize the plants’ energy absorption.

Wang et al. [150] examined the impact of RE adoption on farms in China, highlighting the important role of agriculture in greenhouse gas emissions and the unsustainability of current energy consumption patterns in rural areas. The research, based on a survey of 801 farmers, revealed that over 25% have adopted RE technologies, with factors such as education, farm size, government support, and farmers’ perceptions of RE (its usefulness, profitability, and environmental friendliness) significantly influencing adoption. The study also showed that innovative farmers are more likely to adopt these technologies. Importantly, those who adopted renewable energies experienced a 10% increase in technical efficiency compared to non-adopters, underscoring the need for a transition to more sustainable farming practices. The study suggested that transforming agriculture with RE is essential to meet increasing demands for food and energy sustainably, aligning with global goals such as the Sustainable Development Goals and the Paris Agreement. Policy implications include the need for government involvement in creating an enabling environment, improving access to financing, enhancing the dissemination of information through the media, and developing entrepreneurial skills among farmers to accelerate the adoption of RE technologies.

Cetina-Quiñones et al. [151] presented two sensible heat storage materials, limestone and beach sand, from Yucatán, Mexico, that were evaluated in an indirect type of solar dryer with a storage system (ITSD-TSS), comparing their performance with a conventional solar dryer under various environmental conditions. The results showed that limestone outperformed beach sand, achieving a 1.55% higher drying efficiency and extracting 0.0257 kg more water per kWh. Although beach sand stored 1.5 times more energy, its sensitivity to solar radiation made it less suitable for cloudy regions. The use of three storage compartments in the ITSD-TSS further improved drying efficiency by 3 to 4% over conventional dryers. Due to its availability and cost-effectiveness, limestone was recommended for rural communities in southeastern Mexico to improve agricultural drying processes and contribute to sustainable development. Future work will focus on optimizing the dryer for various local products, reducing costs, and promoting sustainable integration through community training and workshops.

2.7. Carbon Emissions

Maradones et al. [152] quantified the efficiency gains of expanding the sectoral coverage of emissions trading schemes (ETSs) using an optimization model calibrated with microdata from Chilean thermoelectric and industrial sources. By simulating independent and integrated ETS scenarios, the research revealed that merging these carbon markets can significantly reduce compliance costs. For instance, with a 30% emissions reduction target, the allowance price would be USD 36.2/tCO₂ for thermoelectric sources and USD 17.4/tCO₂ for industrial sources independently, but only USD 34.4/tCO₂ if integrated, saving USD 30.7 million (9.7% of total costs). The study highlighted that sectoral expansion of ETS improved economic efficiency through lower prices and reduced regulatory costs, with significant savings achieved, particularly for lower emissions targets. Key findings included the different trends in regulatory costs for thermoelectric and industrial sources, the convergence of prices in a common ETS towards higher thermoelectric ETS prices, and the active participation of industrial sources in carbon markets through fuel switching and the sale of surplus allowances. Future research must take into account the impact of ETSs on other pollutants, notably due to the replacement of biomass by industrial sources, and explore the trade-offs in global and local emissions reductions when several ETSs are operating simultaneously.

Ji et al. [153] examined the energy consumption of residential buildings in the city of Changsha, China, focusing on the hot summer/cold winter (HSCW) climate zone. Modeling 32,145 buildings, the study evaluated seven energy conservation measures (ECMs) and the impact of occupants’ energy conservation behavior, particularly concerning air conditioner (AC) use. Using the AutoBPS Retrofit Toolkit, the study analyzed energy efficiency improvements, investment costs, and economic feasibility. The results underscored the importance of customized retrofit strategies based on building type and age. Lighting retrofits were found to be economically viable and energy efficient, while air conditioning upgrades, while highly efficient, posed economic challenges. The results provided valuable information for policymakers and stakeholders in developing effective strategies to reduce carbon emissions and improve energy efficiency in residential buildings.

To evaluate emerging biofuels and advanced combustion modes that reduce emissions, Parker et al. [154] explored the impact of fuel properties on spray development and combustion in diesel engines. Using a constant-pressure flow chamber and high-speed optical diagnostics, three fuels with distinct chemical and thermophysical properties were analyzed under high-pressure and high-temperature conditions. The results showed that while ignition delay times were similar for fuels with comparable cetane numbers, their liquid lengths, ignition locations, and soot formation differed considerably. The study revealed that thermophysical properties, particularly boiling point and density, primarily determined liquid length, which in turn affected vaporized fuel availability, ignition locations, and take-off length. Fuels with lower boiling points and densities, such as n-heptane, had shorter liquid lengths and higher soot formation, despite similar ignition times to fuels with higher boiling points and densities. These results underlined the importance of considering both chemical and thermophysical properties in fuel screening, as traditional parameters such as cetane number may not fully predict combustion behavior and emissions.

Ruiz et al. [155] focused on developing optimal strategies for maximizing electric mode distance and minimizing total emissions, particularly for routes that include green corridors where combustion engines are prohibited. The authors pointed out that, unlike previous studies, this approach focused not only on improving energy consumption but also on environmental benefits and urban quality of life. Two state-of-the-art multi-objective evolutionary algorithms and a new heuristic, GreenK, were used to meet this challenge. Real bus routes M6 in Badalona and 18 in Grudziadz were analyzed, revealing a significant reduction in emissions of up to 21% compared with the GreenK strategy. This translated into 24 kg fewer pollutants emitted daily and a 22.5% increase in electric range compared with the current solution.

Yang et al. [156] introduced a mathematical model for evaluating CO₂ capture using amine-containing facilitated transport membranes (AFTMs), incorporating the effects of water vapor. The model, which employed a tanks-in-series approach, accounted for variations in gas permeance due to changing water vapor content along the membrane module. The results indicated that both temperature and pressure impacted water vapor levels, influencing CO₂ recovery and product purity. Specifically, higher temperatures and pressures accelerated the reduction in relative humidity, which improved CO₂ recovery but reduced purity. The model also revealed that while water vapor permeance had a minimal impact on separation performance, permeating with a higher water vapor content was better suited for secondary treatment stages. This approach offered valuable insights for optimizing membrane CO₂ capture systems, supporting the design and performance assessment of large-scale AFTMs.

3. Challenges and Future Directions

Today, developing energy systems through a system of systems (SoS) approach presents both promising opportunities and significant challenges. Integrating diverse energy sources such as solar, wind, and bioenergy into a coherent and reliable system requires advanced coordination mechanisms and robust communication networks. To ensure the smooth operation of the overall energy system, it is essential to ensure the interoperability of these subsystems, each of which has different operational characteristics and requirements. However, there are challenges in managing these interconnected systems’ complexity, particularly in real-time data processing, decision making, and ensuring network security and resilience. The current studies highlight the importance of interoperability for the systems mentioned above; however, their standardization is limited due to diverse energy sources and geographical regions, communication protocols, data formats, operating procedures, etc. Ensuring continuity of energy flow, real-time decision making, and flexible control or adjustment of subsystems in response to energy demand variations remains a significant challenge.

In addition, the economic and technical feasibility of deploying large-scale SoS approaches remains a key concern, as it requires significant investments in infrastructure, communication technologies, energy storage systems, and harmonization of regulatory frameworks across different regions. In this context, several research projects and initiatives are still at the experimental stage, and large-scale deployment is limited. Although studies have shown that optimizing energy flow and implementing demand-side management lead to cost savings, some long-term returns on investment are difficult to quantify, especially when trying to compare initial capital expenditure with operating savings. The economic challenges are further complicated by the considerable disparity in regulatory environments between countries, meaning that for SoS approaches to be integrated, policies and incentives need to be harmonized to encourage adoption.

Furthermore, cybersecurity and data confidentiality pose significant challenges, particularly as SoS approaches focus on Internet of Things (IoT) technologies and distributed control systems. Typical communication network vulnerabilities are sources of operational disruption that can lead to malicious attacks. Consequently, cybersecurity frameworks that take into account the structural complexity of SoS applications in the energy sector need to be developed and applied. As far as resilience is concerned, since it will never be possible to eliminate external shocks, such as extreme weather events or network failures, this complicates management. Particularly effective examples are timely fault detection, predictive maintenance, and automatic repair, all of which could be made more resilient.

Subsequent directions in the application of SoSs in energy systems are expected to focus on improving the intelligence and autonomy of sub-systems, allowing them not only to adapt independently to changing conditions but also to contribute to the stability and efficiency of the overall system. More advanced control algorithms and machine learning techniques are essential for optimizing the performance of these interconnected systems, ensuring efficient and sustainable management of energy production, storage, and distribution. In addition, innovative policy frameworks are needed to support the integration of SoS approaches, encouraging collaboration between industry players, governments, and research institutions. By addressing these challenges and exploiting the potential of SoS, the energy sector can move closer to a more sustainable, resilient, and adaptive energy infrastructure.

Table 1. Advantages and challenges of system of systems (SoS) applications in key energy sectors.

Field	Advantages of SoS Application	Challenges of SoS Application
Photovoltaic systems	▪ Improved power generation and efficiency through coordination of multiple systems ▪ Optimization of energy flow between subsystems	▪ Real-time monitoring and control of distributed PV systems ▪ Integration with existing grid systems
Electric vehicles (EVs)	▪ Efficient energy usage and smart charging/discharging between grid and vehicles ▪ Enhanced vehicle-to-grid (V2G) interaction	▪ Managing high loads on energy grid ▪ Lack of infrastructure for large-scale EV integration
Energy storage	▪ Optimization of storage systems for balancing supply–demand variability ▪ Efficient energy management between renewable sources and storage	▪ High cost of storage solutions ▪ Reliability and maintenance concerns with multiple storage systems
Renewable energy	▪ Integration of multiple renewable sources for higher sustainability ▪ Increased system resilience and flexibility	▪ Intermittency of renewable energy sources (solar, wind) ▪ Increased system resilience and flexibility
Smart cities	▪ Enhanced energy efficiency through data-driven decision making ▪ Improved coordination of energy services (lighting, HVAC, etc.)	▪ Data privacy concerns ▪ Complex coordination between various urban infrastructures and services
Rural communities	▪ Supply of reliable and sustainable energy to remote areas ▪ Increased energy access through microgrids and local SoSs	▪ High initial investment in infrastructure ▪ Ensuring technical support and maintenance in remote areas

summarizes the main applications of SoS approaches in various areas of the energy sector, including photovoltaics, electric vehicles, energy storage, renewable energies, smart cities, and rural communities. Each specific area is associated with a specific set of benefits, including energy efficiency, improved grid management, and optimized resource utilization. However, the advantages cited are counterbalanced by a variety of challenges, ranging from technological constraints and high implementation costs to the need to support integration frameworks. To this end, the table above seeks to explore the opportunities of SoS for the development of the global energy system, as well as pointing to areas of specific interest that merit further development to reduce existing barriers.

Table 2. Summary of system of systems (SoS) applications in various energy fields.

Field	References
Photovoltaic systems	[64,65,66,67,68,69,70,71,72,73,74,75]
Electric vehicles	[76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94]
Power systems and microgrids	[95,96,97,98,99,100,101,102,103,104,105,106,107,108,109]
Energy storage	[110,111,112,113,114,115,116,117,118,119,120,121,122]
Renewable energy	[123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138]
Smart cities and rural communities	[139,140,141,142,143,144,145,146,147,148,149,150,151]
Carbon emissions	[152,153,154,155,156]

summarizes the SoS applications present in the literature for the various fields surveyed in this review.

4. Conclusions

In this review, we provided an in-depth exploration of the emerging behavior within SoSs and their implications for managing modern energy systems, offering a path toward more integrated, efficient, and resilient infrastructures. The article began with a review of studies focused on SoS control and optimization. Then, by examining the importance of SoSs for energy systems, including photovoltaics, electric vehicles, power systems, energy storage, renewable energies, smart cities, and carbon emissions, the article explored how SoSs can improve energy efficiency.

To this end, the WoS database was explored using a systematic literature review approach to gather relevant studies published between 2020 and 2024. Search criteria focused on keywords such as “systems of systems” and “control or optimization” to ensure comprehensive coverage of the topic.

The analysis provided an overview of the current state of research in this field, highlighting the increasing interest and focus on SoS research in energy systems. It highlighted the potential of SoSs to address the multifaceted challenges of the energy transition, notably through better coordination and optimization of diverse energy sources. However, the successful implementation of SoSs in energy systems depends on the ability to overcome major challenges, including the complexity of real-time data processing, guaranteeing interoperability between subsystems, and establishing favorable regulatory frameworks.

In addition, future research and development efforts must focus on advancing control algorithms, improving machine learning applications, and promoting collaboration between stakeholders to take full advantage of the benefits of SoSs. By meeting these challenges, the energy sector can move towards a more sustainable future, with energy systems better equipped to adapt to dynamic environmental and economic conditions.