Efficient energy management represents a major challenge, characterized by the complexity derived from multiple factors, including the variability of natural resources, diversification of energy sources, fluctuating demand, and the increasing integration of renewable technologies and Electric Vehicles (EVs) as highlighted in [1]. Similarly, [2] discusses how the transition to sustainable energy systems is compounded by similar challenges, with an emphasis on the integration of renewable energy sources and the need for advanced energy management strategies. In particular, the transition to a low-carbon energy system, as observed in the Nordic-Baltic region, exemplifies the complex interplay between different energy sources, demand variability, and geopolitical contexts [3]. Consequently, the need for Microgrids (MG) and smart grids has become imperative as the power grid and the electricity market undergo a gradual transition from a centralized to a more distributed model to meet the current challenges [4]. Despite the advances made with the implementation of MG, ranging from coordinated control strategies [5] to improvements in modelling [6], energy efficiency [7] and resource and demand management [8], significant opportunities remain for innovation in terms of system optimization, both in energy management and in equipment control, planning, and facility design [9].

In the field of strategies to improve these systems, methods have been investigated to optimize the planning of the combination of generation sources, with a significant emphasis on the search for scenarios of zero greenhouse gas emissions. For instance, a study on Benin's energy sector [10] demonstrates how integrating renewable energy sources, such as solar, wind, and hydropower, can significantly reduce CO2 emissions while achieving higher renewable energy penetration targets. Similarly, the evolution of power generation mixes globally, analyzed through system dynamics models, reveals the potential of clean energy adoption in meeting carbon neutrality goals [11]. Furthermore, the case of Ghana illustrates that increasing renewable energy penetration in the electricity sector can lead to substantial greenhouse gas emissions reductions, aligning with climate change mitigation objectives [12]. These studies thoroughly analyse the impact of incorporating renewable energies into the energy matrix, which is important for guiding policies and implementing strategies to promote the transition to more sustainable energy systems more resilient to climate change. Research such as [13] shows the economic viability of using renewable technologies when considering the optimal combination of energy generation and storage systems, such as pumped hydroelectric storage, lithium batteries, and EV batteries. Meanwhile, from an energy perspective, [14] emphasizes that energy savings play a fundamental role in enhancing hybrid renewable systems, with energy efficiency serving as a primary objective. This involves optimizing energy use to reduce waste and maximize system performance. Additionally, the importance of implementing energy efficiency measures in residential buildings has been extensively underscored, particularly in Mediterranean regions where climatic conditions, such as hot summers and mild winters, significantly influence energy consumption patterns [15]. These studies highlight the need for energy optimization packages tailored to specific climates to maximize efficiency and reduce costs in the residential sector.

The current literature reflects significant interest in applying computational intelligence technologies to address MGs' operation, optimization, and energy control challenges. For instance, Artificial Neural Networks (ANN) combined with gravitational search algorithms improve load and price forecasting accuracy [16], hybrid neuro-evolutionary methods enhance wind power output prediction [17], and ANNs optimize catalytic processes for energy transitions [18]. Bio-inspired algorithms, based on processes and patterns observed in nature, have emerged as effective metaheuristics in optimization and prediction in a variety of contexts. Researchers have explored their application to improve the performance of renewable technologies compared to traditional approaches, as in the case of Maximum Power Point Tracking controllers using the Grey Wolf Optimizer (GWO) with lower curling effect, faster settling time for each irradiation level, and improvement of the system response [19]. Furthermore, comparative analyses have been performed between different optimization methods based on nature-inspired algorithms, such as Particle Swarm Optimization (PSO) and GWO, in the modelling of lithium-ion batteries [20], revealing remarkable enhancements in the model through the terminal voltage compared to the non-optimized model, with superior performance through the GWO. Also, [21] compares the performance of three bio-inspired algorithms, GWO, PSO, and Genetic Algorithm (GA) in the tuning of DC-DC Boost Converter PID Controller, with good overall performance after evaluating the system response under different input voltage and load changes. Other studies have explored optimal energy management in smart microgrids, particularly in scenarios involving high penetration of EVs and distributed renewable sources, utilizing hybrid approaches such as GAs and analytic hierarchy process [22].

On the other hand, proper energy management is necessary in the progress towards the next stage of the energy transition. In this sense, it will be necessary that the household sector adapts flexibly to align energy consumption with the highly variable production patterns inherent to renewable energies [23]. Bio-inspired algorithms emerge as dynamic tools capable of addressing the complexity of energy systems. Along these lines, several researchers have explored their application in Hybrid Renewable Energy Systems (HRES) networks, developing methodologies that integrate ANNs optimization to improve system reliability and efficiency through PSO [24]. In turn, other studies have used these algorithms to estimate the optimal amount of biomass required in gasification plants to produce the necessary synthesis gas and cover the energy demand [25]. Even the efficiency of these approaches has been compared with [26] after implementing an evolutionary game-theoretic approach for an energy management model, with better results for [27].

Despite the progress made, there are still unresolved challenges in this field. For instance, most current approaches focus on optimizing the management of generated energy, neglecting the consideration of domestic demand and other key loads such as EVs, without establishing a bidirectional adaptation. The growing role of EVs in HRES must be considered, especially after policy interventions in the European Union aimed at boosting the transition to electric technology [28].

Therefore, an issue of utmost importance today in the scientific community is to achieve energy management capable of adapting to the most uncertain factors present in HRESs: natural resources, user demand, and the electricity market. Therefore, this paper develops an innovative strategy to address complex problems in energy management, proposing the hypothesis of integrating the GA to determine optimal strategies for both electricity demand (household load and EV charging) and energy use, minimizing grid costs. For this purpose, a case study of a household in Valencia, Spain, is selected to validate the proposed methodology, using real data to increase the applicability and validity of the results. The household consists of a grid connection, a demand, and a PV installation to which an ESS and an EV are added to evaluate the proposed hypothesis's feasibility further. The proposed model is evaluated through a sensitivity analysis under various electricity market cost scenarios, including geopolitical conflicts, pandemic conditions, the Business as Usual (BAU) status, and future projections. Also, two cases, referred to as Stable and Critical, are considered to rigorously assess the model's robustness. Likewise, recent works also address the impact of external factors, such as global events like the COVID-19 pandemic, on household energy consumption patterns, demonstrating that lockdowns led to significant increases in energy use at home due to lifestyle changes [29]. Such studies highlight the need for adaptive energy management models that can respond to shifts in demand caused by extraordinary circumstances.

The implementation of the proposed model shows significant improvements in the operational efficiency of the MG, especially in the utilization of solar energy and batteries during Stable cases, as well as in the adaptability of the algorithm in Critical cases. This points towards smarter energy management that not only reduces costs but also improves the stability of the energy supply by dynamically adapting to variable situations and reducing the carbon footprint. Hence, the main objective of this work is to develop and validate an approach based on GAs through a sensitivity analysis of electricity market costs, which addresses these shortcomings to improve the operational efficiency of MG and advance the transition towards a cleaner and more sustainable energy system. The proposed work is organized as follows: section 2 presents the methodology with the system under study, and a description of the scenarios and cases considered, section 3 provides the results, section 4 refers to the discussion of this application, and section 5 outlines the main conclusions and future work.

The following section provides a detailed overview of the methodology used in this study. Different software programs have been employed to conduct the research: FusionSolar has been used to collect experimental data from the real installation, System Operator Information System (ESIOS) has been employed for obtaining electricity market prices, Iberian Energy Market Operator - Portuguese Hub (OMIP) has been used for future price projections, while MATLAB has been employed to model the system under study and implement the GA.

In this section, the characteristics of the MG are presented in detail, the various systems analysed, and the process of obtaining the experimental data. The configuration and adjustment of the GA parameters are also described. Finally, the scenarios and cases simulated, and the evaluation criteria of the proposed systems are explained, including analysis of power flows, costs, CO2eq emissions to the atmosphere, and computation time. It is worth mentioning that the case studies are based on a household consisting of a grid connection, a demand, and a PV installation to which an ESS and an EV are added to evaluate the proposed model's potential further.

Figure 1 shows the workflow corresponding to the proposed hypothesis, which starts with the first data input step (stage 1), followed by the execution of the genetic algorithm (stage 2), and concludes with the model evaluation stage (stage 3).

Figure 1.

Methodology for the proposed optimized energy manager model

The inputs and outputs of the model are explained as follows.

System under study

The MG consists of a solar PV installation, domestic demand, an ESS, an EV, and a grid connection that allows sending surpluses and receiving when it is not possible to meet the demand. Figure 2 shows the energy flow directions in the MG.

Figure 2.

MG under study

In the current installation, the loads (household demand and EV) are supplied in the following priority order:

1. Demand is met using the energy generated by the PV installation.

2. If the energy produced by the PV installation is insufficient to cover both demands, the ESS is called upon.

3. If both the PV installation and the ESS fail to meet the demands, energy is imported from the grid.

4. Situations involving surplus energy generated by the PV panels are first directed to the ESS. The surplus is injected into the grid for sale if the ESS is fully charged.

This process ensures a continuous balance in energy supply and demand.

Electric vehicle model

An EV has been integrated into the MG to further evaluate the feasibility of the strategy proposed as a hypothesis. For the EV simulation, the vehicle used in [30], the Nissan Leaf, which is a leading choice among EVs, especially in the European market where it has secured its position as one of the best sellers, has been taken as a reference. Since its introduction in 2010, global sales have exceeded 300,000 units, with 68,000 units sold specifically in Europe [31]. The EV battery pack architecture comprises two battery cells connected in series, which in turn are connected in parallel with two other cells, forming a battery module. A total of 48 battery modules are connected in series to create the battery pack. Each battery cell has a nominal capacity of 32.5 Ah, with a nominal voltage of 3.75 V and a maximum voltage of up to 4.2 V. Considering the number and configuration of the battery modules, the total rated voltage and capacity are 360 V and 24 kWh, respectively. The complete battery pack is divided into three sections. One section contains 24 modules positioned centrally within the pack, while the other two sections hold 12 modules each connected in series, located on either side of the central section.

Eq. (1), eq. (2), and eq. (3) were employed to model the EV battery charging system. The EV discharge profile was developed based on the EV battery discharge behaviour analysis presented in [26] and the driving cycles described in [27]. Two distinct driving cycles, as illustrated in Figure 3, were selected for analysis. The first driving cycle, shown in Figure 3 (a), represents a weekend case characterized by recreational use. This cycle involves an extended journey of 46,210.07 meters, typical of non-working days. In contrast, the second driving cycle, depicted in Figure 3 (b), corresponds to a weekday case and represents a commuting journey. This cycle involves a shorter and more routine distance of 14,085.53 meters, reflecting typical work-related travel patterns on working days. This approach provides more variability to the scenarios studied.

Figure 3.

Implemented driving cycles:(a) 'MODEM Hyzem motorway_total, and (b) LDV_PVU commercial cars road_total

Sensitivity scenarios

This section provides a comprehensive overview of the energy demand and climatic conditions considered in this study, as well as an analysis of the various electricity cost scenarios derived from market fluctuations.

Cases description.

This study considered two simulated cases regarding total demand consumption and resource availability: Stable (sunny, characterized by higher availability of solar energy resources and lower energy demand) and Critical (cloudy, associated with lower availability of solar energy resources and higher energy demand). Regarding energy consumption, the daily average for the Stable case is 127 kWh, whereas for the Critical case, it is 164 kWh.

Concerning the classification of solar generation data into "sunny" and "cloudy" conditions was established through an analysis of total energy generation values over a specified week. The data indicate that the total solar generation for sunny days ranged from 203.28 kWh to 264.66 kWh, while cloudy days exhibited generation values between 72.21 kWh and 201.80 kWh. Notably, the highest generation values were consistently associated with sunny conditions, whereas the cloudy conditions demonstrated significantly lower output levels.

Electricity cost scenarios description.

By analysing diverse scenarios, this study aims to observe the potential implications of fluctuating electricity costs on the operation and optimization of hybrid energy systems. Particularly, four relevant and different electricity cost scenarios for the Spanish market are considered. These scenarios were defined as follows:

• War: This scenario reflects the period during which the impact of geopolitical conflicts significantly affected electricity costs in Spain (that is, the war between Russia and Ukraine). During this time, costs surged to unprecedented levels, highlighting the vulnerability of energy markets to external shocks.

• COVID: This scenario represents the period in which the COVID-19 pandemic had a profound impact on electricity costs in Spain. In contrast to the War scenario, costs decreased substantially, reaching notably low levels.

• BAU: This scenario represents a week of September 2024, during which a pronounced reduction in electricity costs is observed, particularly during the midday hours when solar generation peaks. This trend reflects the growing influence of renewable energy sources on market prices.

• OMIP 2030: This scenario incorporates future projections based on the OMIP trends for Spain in 2030. It reflects anticipated market dynamics, including evolving energy supply and demand patterns, policy changes, and advancements in technology.

Since the COVID scenario occurred before 2021, the corresponding electricity tariff is 2.0DHA, while the remaining scenarios correspond to the 2.0TD tariff. The graphs representing electricity costs are presented in Figure 4. The data for the first three scenarios analysed in the sensitivity assessment come from the ESIOS platform of the electricity grid operator in Spain [32]. In contrast, the cost projections for the OMIP 2030 scenario were obtained from the Spanish market operator [33].

Figure 4.

Comparison of electricity costs across different scenarios

This section begins by showing the costs obtained, as this is the main objective of the study and the sensitivity analysis. Then, the results related to energy consumption and CO2 emissions reduction are presented, to conclude with an analysis of the computation time of the optimization processes.

The implementation of the optimization algorithm in this study yielded substantial cost reductions across all scenarios. In the Stable case, the energy system operates with fewer constraints on renewable energy availability, allowing the optimization algorithm to function more efficiently by shifting consumption to less expensive periods and reducing reliance on more costly energy imports. For example, in the War scenario, which reflects geopolitical instability and its associated market disruptions, the optimization process resulted in a 26% reduction in electricity costs. Specifically, the net weekly costs before optimization were approximately 30.40 €, and they decreased to 22.54 € after the optimization was applied, that is, a weekly cost reduction of 7.86 €. This represents a significant saving for just one week and highlights the optimization model's ability to manage energy resources effectively, even under circumstances where external market forces might otherwise inflate costs. Similarly, in the BAU scenario, representing contemporary energy market conditions, a cost reduction of 29% was also observed. This consistency across different scenarios under the Stable case suggests that the algorithm has a robust capacity to optimize energy usage by redistributing energy consumption to less expensive periods, even when the external market forces are less extreme compared to the War scenario. In this case, costs decreased from 8.15 € before optimization to 5.79 € post-optimization. This reduction, while lower in absolute terms compared to the War scenario, is still significant, particularly considering that the BAU scenario operates within a more predictable market environment. The algorithm’s effectiveness in both the War and BAU scenarios indicates that the benefits of optimization are not confined to periods of market volatility, but also extend to more stable and regular market conditions.

However, while the results under Stable cases are notable, it is under Critical cases that the optimization process showcases its full potential. Critical cases simulate scenarios where renewable energy availability is severely constrained, often due to adverse weather patterns or other external pressures that limit the system's ability to rely on clean energy sources. In such situations, the reliance on imported electricity increases, and without optimization, the system would face significantly higher costs. Under these more challenging conditions, the optimization algorithm proved to be exceptionally effective, yielding even greater cost reductions than under the Stable case. Concerning, the COVID scenario, for instance, which represents a period marked by unprecedented global disruptions in energy demand and supply chains due to the pandemic, the optimization algorithm delivered a reduction of nearly 59% in peak period costs. Before optimization, peak period costs stood at 5.97 €, but after applying the optimization model, these costs dropped to 2.43 €. This reduction shows the model's capacity to adapt to highly constrained energy environments. Moreover, the Critical cases also presented a unique challenge in the War scenario, where renewable energy is less accessible, and electricity import costs are higher. In this scenario, the optimization algorithm reduced peak period costs by 52%, lowering them from 24.74 € to 11.78 €. This reduction, though slightly lower in percentage terms than the savings achieved in the COVID scenario, is still highly significant given the much higher baseline cost in the War scenario. Additionally, the OMIP 2030 scenario, which projects future energy pricing trends based on market forecasts, presents another interesting case for analysis. Under Critical cases in this scenario, a reduction of 49% was observed in peak period costs, with costs falling from 3.37 € to 1.73 €.

The analysis of energy imports is also important since it shows the model's ability to reduce the reliance on external electricity supplies across all the studied scenarios. First, under the Stable case, the optimization algorithm resulted in notable reductions in energy imports, in situations where renewable energy sources are readily available. In the War scenario, for instance, energy imports were reduced from 643.60 kWh to 477.65 kWh, representing a 26% decrease. Similarly, in the BAU scenario, the optimization model achieved a 27% reduction in energy imports under the Stable case, lowering imports to 469.66 kWh. Furthermore, the results from the COVID scenario also reveal interesting insights into the algorithm’s performance under the Stable case, where energy imports were reduced by 28%, to 465.46 kWh. Although the reductions are less pronounced compared to those observed under the Critical case, this still represents a significant improvement in energy mangegement.

However, the most compelling results were observed under the Critical case, where the availability of renewable energy is more restricted. In these scenarios, the optimization model demonstrated its true potential by significantly reducing energy imports during peak periods, where the cost and demand for electricity are at their highest. For example, in the COVID scenario, energy imports during peak periods were reduced by 59%, dropping from 817.73 kWh to 335.95 kWh. The results in the war scenario under the Critical case further validate the optimization algorithm's adaptability. Peak period imports were reduced by 53%, from 481.47 kWh to 224.83 kWh, a significant drop. It is also worth noting the results in the OMIP 2030 scenario, which projects future energy market conditions. Under the Critical case, the optimization model managed to reduce peak period imports by 50%, from 481.47 kWh to 241.85 kWh.

Beyond the raw cost reductions, it is important to recognize how the optimization algorithm shifts energy consumption patterns, especially in terms of reducing peak period costs and reallocating demand to less expensive times of day. For instance, across several scenarios, it was observed that while peak period costs decreased significantly, there was a slight increase in flat and valley period costs, as the optimization process redistributed energy demand. This reallocation is particularly important as it demonstrates the algorithm’s capability to smooth out energy consumption, preventing spikes during high-cost periods and spreading demand more evenly throughout the day.

Furthermore, the optimization model presented in this study also shows a reduction of CO2 emissions since it is directly linked to the energy imports reduction. In scenarios characterized by the Critical case, the optimization process was able to achieve reductions in emissions by redistributing energy use away from peak demand periods, helping to decrease the overall environmental footprint. In more Stable cases, for instance, in the BAU and OMIP 2030 scenarios, the optimization model continued to demonstrate its relevance. Although the availability of renewable energy was higher, leading to inherently lower baseline emissions, the optimization process still managed to deliver meaningful reductions in CO2 emissions.

Concerning the computational times for the optimization algorithm, significant variations were observed across scenarios, reflecting the complexity of energy management under different cases. In the Critical case, such as the War and BAU scenarios, computation times reached up to 60 minutes due to the need to manage limited renewable energy availability and fluctuating demand. Conversely, under Stable cases, computation times were considerably lower, typically around 23 minutes, as the optimization process benefited from more adaptative energy patterns.

This study presents an analysis of a genetic algorithm framework designed to optimize energy management in hybrid poly-generation systems, focusing on a significant week characterized by fluctuating energy prices influenced by global events. Simulations were conducted across four distinct scenarios: War, COVID, BAU, and OMIP 2030, reflecting a range of market conditions and price dynamics. The analysis was further contextualized by examining both Stable and Critical cases, the former with high resource availability and lower energy demand, and the latter with low energy availability and high energy demand. This procedure highlights the model's adaptability and effectiveness in diverse operational environments.

The results show substantial cost reductions in electricity management, particularly under the Critical case where renewable energy availability was limited. For instance, during the COVID scenario, the optimization algorithm achieved nearly a 59% reduction in costs during peak periods, showcasing its capability to manage energy resources effectively in times of crisis. Its ability to reduce energy imports by shifting consumption patterns away from peak periods and into valley periods contributes to a more sustainable, resilient energy system that is better equipped to handle fluctuations in energy supply and demand. Additionally, the model contributed to significant reductions in CO2 emissions, reinforcing the dual benefits of economic efficiency and environmental sustainability.

The computational performance of the algorithm varied across scenarios, indicating that while it effectively adapts to changing conditions, there remains an opportunity for refinement to enhance efficiency, particularly in constrained scenarios. Future research should aim to accelerate computational processes to enable real-time applications, thus maximizing the potential for timely cost and emissions reductions.

Therefore, the study provided valuable insights into the implementation of advanced optimization techniques in energy management systems. Addressing the complexities of modern energy markets through simulations in significant contexts contributes to the ongoing address on sustainable energy practices and highlights the importance of resilient energy infrastructures against volatility in the electricity market.