A Model-Based Approach to Assessing Operational and Cost Performance of Hydrogen, Battery, and EV Storage in Community Energy Systems

Abstract

Community energy systems are expected to play an increasingly important role in the decarbonization of the residential sector, but their operation depends on how different electricity and heat storage technologies are configured and used. Existing studies typically examine storage options in isolation, limiting the comparability of their operational roles. This study addresses this gap by developing a decision-support framework that enables a consistent, operation-focused comparison of battery energy storage, hydrogen storage, and electric-vehicle-based storage within a unified community-scale hybrid energy system. The model represents electricity and heat balances in a hub formulation that couples photovoltaic and wind generation, a gas engine, an electric boiler, thermal and electrical storage units, hydrogen conversion and storage, and an aggregated fleet of electric vehicles. It is applied to a stylized Polish residential community using local demand, generation potential, and electricity price data. A set of single-technology and multi-technology scenarios is analyzed to compare how storage portfolios affect self-sufficiency, self-consumption, grid exchanges, and operating costs under current electricity market conditions. The results show that battery and electric vehicle storage primarily provide short-term flexibility and enable price-driven arbitrage, as reflected in the highest contribution of battery discharge to the electricity supply structure (5.6%) and systematic charging of BES and EVs during low-price hours, while hydrogen storage supports intertemporal shifting by charging in multi-hour surplus periods, reaching a supply share of 1.4% at the expense of substantial conversion losses. Moreover, the findings highlight fundamental trade-offs between cost-optimal, price-responsive operation and autonomy-oriented indicators such as self-sufficiency and self-consumption, showing how these depend on the composition of storage portfolios. The proposed framework, therefore, provides decision support for both technology selection and the planning and regulatory assessment of community energy systems under contemporary electricity market conditions.

1. Introduction

Global economies and energy policy are increasingly shaped by the objective of reducing greenhouse gas emissions in line with the Paris Agreement and subsequent net-zero roadmaps [1,2]. In the European Union (EU), this objective is being pursued through a rapid expansion of variable renewable electricity generation, particularly from wind and solar resources, alongside the electrification of end-use sectors such as heating and transport [3,4]. While these developments are essential for decarbonization, they fundamentally alter the temporal structure of energy systems and intensify the mismatch between electricity supply and demand. At the local and community scale, this transformation translates into a growing need for flexibility delivered through different forms of energy storage operating under real-world market and regulatory conditions. Despite broad agreement on the importance of storage for integrating variable renewables, there is still no commonly adopted, operation-oriented modeling framework that enables a consistent comparison of different storage technologies—such as batteries, hydrogen systems, and electric vehicles—within a single community energy system. Consequently, the operational trade-offs, complementarities, and system-level value of alternative storage portfolios remain insufficiently understood.

Energy storage technologies differ markedly in their technical characteristics and operational roles. Battery energy storage systems (BES) offer high efficiency and rapid response, making them well suited for short-term balancing and price-driven arbitrage [5,6]. Electric vehicles (EVs) represent a growing, distributed storage resource whose availability is constrained by mobility patterns but can support local balancing through smart charging or vehicle-to-grid operation [7]. Hydrogen-based systems, by contrast, enable energy shifting over longer time horizons but incur substantial conversion losses and require coordinated operation of electrolyzers, storage tanks, and fuel cells [8,9]. Assessing these technologies in isolation obscures their fundamentally different temporal functions and masks how they interact when deployed simultaneously within a community-scale portfolio [10].

The residential sector provides a particularly relevant setting for examining these interactions. Households account for roughly one quarter of final energy consumption in the EU, with space and water heating dominating demand [11]. The ongoing electrification of heating, widespread deployment of rooftop photovoltaics, and rapid diffusion of EVs are therefore reshaping residential load profiles and increasing the need for coordinated management of electricity and heat at the local level [12,13]. However, many existing assessments remain technology-specific or rely on aggregated indicators, without explicitly modeling the operational interactions and trade-offs between different storage technologies within a single community-scale hybrid energy system.

New organizational forms such as energy communities, cooperatives, and energy clusters further amplify this challenge. Enabled by smart grid infrastructure, these entities allow groups of consumers and prosumers to jointly invest in generation and storage assets, share surpluses, and organize local balancing [14,15]. In such settings, storage technologies are not operated independently but as part of an integrated portfolio serving electricity supply, heating demand, mobility needs, and grid interaction simultaneously [16,17]. This portfolio perspective underscores the need for modeling approaches that go beyond single-technology optimization and instead capture interactions across multiple storage options and energy carriers.

These issues are particularly pronounced in EU Member States undergoing a rapid transition away from coal-based power generation. Poland constitutes a highly informative case, combining a coal-dominated power system with one of the fastest recent expansions of distributed renewable energy in the EU. The number of residential photovoltaic micro-installations exceeded 1.5 million by late 2024, with total installed capacity surpassing 12 GW [18], while parallel support schemes actively promote battery storage, electrification of heating, and EV adoption [19,20,21]. At the same time, the Polish regulatory framework governing prosumers, energy cooperatives, and energy clusters introduces specific settlement rules and constraints that directly shape the operational value of different storage technologies, making Poland a particularly relevant test case for comparative assessment under real-world conditions [22].

In such an environment, hydrogen storage, battery systems, and EVs are not competing as independent solutions; they must be assessed as complementary or alternative options for providing flexibility and increasing the use of local renewables. Their suitability depends on technical characteristics such as efficiency, power and energy ratings, as well as on economic factors such as investment and operating costs, price volatility on wholesale and retail markets and compensation mechanisms for exported electricity [10,23]. For Polish community energy systems operating under national rules on prosumers, energy cooperatives and energy clusters, regulatory provisions on grid access, settlement of surpluses and required renewable shares add further constraints and incentives [24,25].

Against this background, there is a clear need for model-based tools that can represent the main techno-economic features of hydrogen, battery and EV storage in a consistent way, and provide quantitative insight into their operational and cost performance in community settings. Such tools should be able to capture the interaction between electricity and heat supply, the coupling between local generation and storage, and exchanges with the public grid under realistic price signals and regulatory conditions. They should also support a systematic comparison of alternative storage portfolios, including purely battery-based systems, EV-dominated configurations, and systems with hydrogen as a long-duration option.

The present study responds to this need by developing and applying a model-based framework for community energy systems that integrates hydrogen, battery and EV storage within a unified operational optimization model. The framework is used to quantify how these technologies, alone and in combination, influence energy flows, self-sufficiency, self-consumption and operating costs under Polish electricity market conditions. By doing so, it provides decision support for planners and stakeholders involved in the design and operation of residential energy communities seeking to expand their storage assets in a cost-effective and resilient manner.

1.1. Related Research

The increasing role of hybrid energy systems that combine electrical and thermal storage has intensified research into methods that improve their technical and economic performance [26]. As noted in Section 1, this interest is particularly strong for community-level systems, which must coordinate storage technologies operating over different time horizons while managing the variability of renewable energy resources. Consequently, a broad body of literature has emerged that applies optimization-based methods to identify cost-effective operational strategies under technical and economic constraints [27,28].

Although a wide range of optimization techniques has been applied to hybrid energy systems, most studies rely on linear and nonlinear programming formulations or on heuristic and metaheuristic algorithms such as genetic algorithms (GA), particle swarm optimization (PSO) and simulated annealing [27,29]. These approaches remain the most established and widely adopted because they are robust and computationally tractable while capturing essential operational features. However, their application is often guided by methodological requirements rather than by a systematic comparison of how different modeling choices affect the representation of storage technologies operating over multiple time scales. This limitation is particularly relevant at the community scale, where short-term and long-term storage must be coordinated within a single operational framework. These considerations motivate the use of modeling approaches that can balance tractability with a consistent and comparative representation of different storage technologies.

Within this methodological landscape, mixed-integer linear programming (MILP) continues to be a central approach for the optimization of energy systems across scales, from individual buildings to regional and national networks [30]. At the community scale, existing MILP-based studies show that such formulations can represent both short-term and long-term storage dynamics and support cost-minimizing operation. For example, ref. [31] developed an MILP model for an off-grid microgrid equipped with PV, batteries, and hydrogen storage based on an electrolyzer and a fuel cell. The formulation explicitly accounted for degradation effects of both batteries and fuel cells and demonstrated that demand response with controllable loads can reduce operational costs. However, the analysis was restricted to an islanded configuration and did not include heat demand or grid interaction, limiting its applicability to grid-connected community energy systems where multi-energy coupling and market interaction are critical.

A similar MILP formulation was presented by [32] for a German multi-apartment building operating under Tenant Electricity Law. The model jointly optimized the operation and sizing of PV, combined heat and power units, heat pumps, and battery storage. The results showed that configurations combining PV, CHP, and a heat pump achieved high levels of energy self-sufficiency, while batteries were not cost-effective under the applicable settlement rules. However, the analysis was limited to individual multi-family buildings considered independently and to a specific regulatory framework, restricting the generalizability of the findings to broader community settings.

More complex system configurations have also been investigated using MILP-based approaches. For example, ref. [33] optimized a hybrid microgrid integrating battery and hydrogen storage with solar and wind generation, including component sizing and net present value maximization, and demonstrated that combining short-term and seasonal storage improved profitability. At a larger spatial scale, ref. [34] applied an MILP model to the planning of distribution networks with PV–battery integration, reporting reductions in network losses, peak demand, and annual energy costs. Nevertheless, these studies focused on specific system layers or objectives and did not explicitly account for regulatory conditions or multi-energy interactions relevant to community-scale operation.

Mixed-integer linear programming has also been applied to grid-connected community energy systems. Existing studies show that MILP formulations can effectively capture battery operation under dynamic tariffs and grid constraints, leading to reduced peak demand and improved utilization of connection capacity. For example, ref. [35] optimized battery charging and discharging in an on-grid microgrid at the single-facility level, while [36] developed a day-ahead scheduling model for a community of approximately 500 households with PV installations and battery storage, including grid interaction and local energy trading. Similarly, ref. [37] demonstrated that collective battery storage combined with PV can reduce operating costs and limit energy exchange with the main grid when optimized using real demand and generation data. However, all three studies focused exclusively on battery storage and did not consider hydrogen-based systems. Taken together, these studies confirm the value of coordinated battery operation under grid constraints, but they do not address how battery storage compares to or interacts with hydrogen-based and mobility-based storage options within a unified community-scale framework.

Several studies have extended optimization-based analyses to explicitly account for interactions between electricity and heat. These works demonstrate that multi-energy formulations can improve system performance by enabling additional flexibility through energy conversion and storage across carriers. For example, ref. [38] analyzed a system combining a heat pump, a boiler, and a bi-directional electric-to-thermal conversion and storage unit, showing that limited electricity-to-heat and heat-to-electricity conversion reduced total system costs by approximately 12.3% compared to a battery-only configuration under Beijing-based conditions. In a related contribution, ref. [39] developed an MILP model for seasonal hydrogen-based energy storage using reversible solid oxide cell technology for an office building. Although the model accounted for PV variability and electricity price signals, the absence of battery storage constrained short-term operational flexibility, highlighting trade-offs between long-term and short-term storage representations in multi-energy systems.

Other authors have adopted heuristic and metaheuristic algorithms to address nonlinear or high-dimensional formulations that are difficult to capture using purely linear approaches. These methods are particularly suited to neighborhood-scale systems with multi-objective or highly nonlinear behavior [27]. Existing studies demonstrate that hybrid heuristic–optimization frameworks can improve operational performance in specific applications. For example, ref. [40] applied a GA-based approach for real-time electric vehicle charging control combined with a day-ahead MILP scheduler, increasing the average-to-peak demand ratio from 61% to approximately 74%. Similarly, ref. [41] combined PSO with branch-and-bound techniques to model a PV–hydrogen system with a nonconvex representation of the electrolyzer and hydrogen tank, while [42] compared cascaded MILP and accelerated PSO–MILP workflows for a university microgrid with electric vehicles. Despite considering multiple technologies, these studies do not enable a systematic and operation-oriented comparison or joint integration of batteries, electric vehicles, and hydrogen storage within a unified modeling framework.

Despite significant progress, several research gaps remain. Heuristic and metaheuristic methods do not guarantee global optimality and are sensitive to parameter selection, which affects reproducibility. Based on the works analyzed above, only a limited number of studies incorporate fully integrated multi-energy configurations, and most focus on short-term electrical storage [32,34,35,36,37,38,40,42] or on hydrogen storage alone without batteries [39]. Many contributions are restricted to single-building analyses or highly specific local conditions, which limits their generalizability [39,42]. In addition, grid interaction is often simplified or omitted entirely, as in [31], even though most communities do not achieve full energy independence without substantial investment. A structured comparison of representative studies highlighting system scope, modeled storage technologies, treatment of multi-energy interactions, and key limitations is provided in Table 1.

Beyond community-scale operation, a growing body of literature has examined the participation of energy storage systems in electricity markets, where storage units act as market participants responding to price signals or submitting bids in organized markets. These studies typically formulate the storage operation problem as a single-level MILP under price-taker assumptions and focus on energy arbitrage, participation in day-ahead and intraday markets, or the joint provision of energy and ancillary services. For example, ref. [43] optimized the spot market participation of PV–battery systems using a stochastic MILP framework, while [44] developed an MILP-based bidding strategy for storage participation in joint electricity markets. A smaller subset of contributions extends this perspective by explicitly modeling strategic interactions or financial entities through bilevel formulations, as illustrated by the risk-aware approach proposed in [45]. While these studies provide important insights into storage behavior at the market level, they primarily address bidding and price-response strategies rather than the integrated operational coordination of multiple storage technologies within community energy systems.

From a geographical perspective, although the existing literature offers a broad methodological and application-oriented context for community energy systems and storage technologies, it provides limited insight into the specific research gap addressed in this study. In particular, analyses explicitly focusing on community energy systems under Polish market and regulatory conditions remain scarce and typically address individual technologies within narrow system boundaries. For instance, ref. [46] analyzed residential PV–battery systems under variable tariffs, while [47] examined hydrogen storage for surplus PV energy in Polish households without considering alternative storage options or community-scale coordination. Similarly, ref. [48] assessed PV–battery systems under Polish tariff structures, but did not address multi-technology storage portfolios. As a result, integrated and operation-oriented assessments that systematically compare battery, hydrogen, and electric vehicle storage at the community scale are largely absent for Poland, as well as for other coal-transition economies in Central and Eastern Europe.

Overall, the reviewed literature demonstrates the potential of advanced optimization approaches for improving the operation of community energy systems with storage. At the same time, important gaps persist, including incomplete integration of different storage technologies, fragmented treatment of electricity and heating sectors, and limited consideration of realistic grid interaction constraints. Consequently, existing studies provide only partial insight into how batteries, hydrogen systems, and electric vehicles interact when deployed simultaneously within grid-connected community energy systems. Addressing these limitations requires decision-support tools that enable a consistent, operation-oriented assessment of alternative storage portfolios under real-world market and regulatory conditions.

Table 1. Comparison of representative optimization-based studies on community and building-scale energy systems with storage, highlighting system scope, modeled technologies, and key limitations.

Study	System Scale	Storage Technology	Multi–Energy System	Grid Interaction	Method	Main Limitation
Marocco et al. (2021) [31]	Community (off-grid)	BES, H2	No	No	MILP	Islanded system; no heat demand; no grid interaction
Braeuer et al. (2022) [32]	Building	BES	Yes	Simplified	MILP	Single-building focus; batteries not compared with alternative storage technologies
Superchi et al. (2026) [33]	Community	BES, H2	No	Limited	MILP	Focus on profitability; limited regulatory and market representation
Vanegas et al. (2025) [34]	Distribution network	BES	No	Yes	MILP	Network planning focus; no community operation or sector coupling
Sigalo et al. (2021) [35]	Building	BES	No	Yes	MILP	Single facility; battery-only storage
Dinh et al. (2022) [36]	Community	BES	No	Yes	MILP	Battery-only storage; no long-term or multi-energy storage
Barkouki et al. (2023) [37]	Community	BES	No	Yes	MILP	No hydrogen or EV storage; electrical domain only
Gao et al. (2024) [38]	Building	BES (ETU)	Yes	Limited	MILP	No hydrogen storage; limited scalability to communities
Lindholm et al. (2020) [39]	Building	H2	Yes	Yes	MILP	No battery storage; reduced short-term flexibility
Nahid et al. (2025) [40]	Community	EV	No	Yes	GA + MILP	Focus on EVs only; no comparison with other storage technologies
Yang et al. (2024) [41]	Community	H2	No	Limited	PSO + MILP	Nonconvex formulation; no battery or EV integration
Kassab et al. (2025) [42]	Campus	BES, EV	No	Yes	MILP/ PSO+MILP	No hydrogen storage; no unified comparison of storage portfolios
Goryl et al. (2025) [46]	Building (Poland)	BES	No	Yes	MILP	Single prosumer; battery-only; no community coordination
Olczak et al. (2025) [47]	Building (Poland)	H2	No	Yes	MILP	Hydrogen-only storage; no alternative technologies
Benalcazar et al. (2024) [48]	Building (Poland)	BES	No	Yes	MILP	No multi-technology storage comparison
This study	Community	BES, H2, EV	Yes	Yes	MILP	–

Table 1. Comparison of representative optimization-based studies on community and building-scale energy systems with storage, highlighting system scope, modeled technologies, and key limitations.

Study	System Scale	Storage Technology	Multi–Energy System	Grid Interaction	Method	Main Limitation
Marocco et al. (2021) [31]	Community (off-grid)	BES, H2	No	No	MILP	Islanded system; no heat demand; no grid interaction
Braeuer et al. (2022) [32]	Building	BES	Yes	Simplified	MILP	Single-building focus; batteries not compared with alternative storage technologies
Superchi et al. (2026) [33]	Community	BES, H2	No	Limited	MILP	Focus on profitability; limited regulatory and market representation
Vanegas et al. (2025) [34]	Distribution network	BES	No	Yes	MILP	Network planning focus; no community operation or sector coupling
Sigalo et al. (2021) [35]	Building	BES	No	Yes	MILP	Single facility; battery-only storage
Dinh et al. (2022) [36]	Community	BES	No	Yes	MILP	Battery-only storage; no long-term or multi-energy storage
Barkouki et al. (2023) [37]	Community	BES	No	Yes	MILP	No hydrogen or EV storage; electrical domain only
Gao et al. (2024) [38]	Building	BES (ETU)	Yes	Limited	MILP	No hydrogen storage; limited scalability to communities
Lindholm et al. (2020) [39]	Building	H2	Yes	Yes	MILP	No battery storage; reduced short-term flexibility
Nahid et al. (2025) [40]	Community	EV	No	Yes	GA + MILP	Focus on EVs only; no comparison with other storage technologies
Yang et al. (2024) [41]	Community	H2	No	Limited	PSO + MILP	Nonconvex formulation; no battery or EV integration
Kassab et al. (2025) [42]	Campus	BES, EV	No	Yes	MILP/ PSO+MILP	No hydrogen storage; no unified comparison of storage portfolios
Goryl et al. (2025) [46]	Building (Poland)	BES	No	Yes	MILP	Single prosumer; battery-only; no community coordination
Olczak et al. (2025) [47]	Building (Poland)	H2	No	Yes	MILP	Hydrogen-only storage; no alternative technologies
Benalcazar et al. (2024) [48]	Building (Poland)	BES	No	Yes	MILP	No multi-technology storage comparison
This study	Community	BES, H2, EV	Yes	Yes	MILP	–

1.2. Research Aim, Questions, and Contributions

The literature reviewed above demonstrates substantial progress in the optimization of hybrid energy systems with storage across a wide range of system scales and configurations. At the same time, it reveals several persistent limitations that directly motivate the present study. Most existing contributions focus on a limited set of storage technologies, frequently examining short-term electrical storage or hydrogen-based systems in isolation. As a result, batteries, electric vehicles, and hydrogen storage are rarely assessed as a coherent and interacting portfolio. In addition, multi-energy interactions between electricity and heat are only explicitly represented in a subset of studies, while grid exchanges are often simplified or omitted altogether. Heuristic and metaheuristic approaches further limit systematic comparison, as they do not guarantee global optimality and are sensitive to parameter choices. These limitations are particularly pronounced for community energy systems operating under specific national market conditions, where operation-oriented and comparative assessments remain scarce. Together, these gaps highlight the absence of modeling frameworks that can consistently represent different storage technologies, sector coupling, and grid interaction within a unified community-scale setting. Addressing this gap, the present study provides a unified, operation-oriented framework that enables a systematic comparison of battery, hydrogen, and electric vehicle storage within community energy systems under realistic market conditions.

Against this background, the aim of this study is to develop and apply a model-based approach that enables a consistent, operation-oriented assessment of community energy systems integrating hydrogen, battery, and electric vehicle storage under Polish electricity market conditions. The proposed framework is designed to explicitly capture the techno-economic characteristics of different storage technologies, their coordinated operation across multiple time scales, the coupling between electricity and heat supply, and interactions with the public grid under realistic price signals and regulatory constraints.

The study is guided by the following research questions:

• How should a model-based framework represent the techno-economic characteristics of hydrogen, battery, and electric vehicle storage in order to enable a robust assessment of operational and cost performance in community energy systems?
• How does the choice and joint integration of hydrogen, battery, and electric vehicle storage technologies influence the operational behavior and cost performance of community energy systems?
• How can community energy systems evaluate and compare alternative storage portfolios under current Polish electricity market conditions?

In this context, the main contributions of the study are as follows:

• Develop a unified optimization-based framework for community energy systems that jointly represents hydrogen, battery, and electric vehicle storage, while explicitly capturing electricity–heat interactions and grid exchanges.
• Demonstrate the framework on a stylized community energy system operating under Polish electricity market conditions, reflecting the ongoing expansion of distributed renewable generation and storage assets.
• Quantify how different storage technologies and their configurations affect self-sufficiency, energy cost outcomes, and the utilization of local renewable resources, thereby providing decision support for planning and operation of community-scale storage portfolios.

In line with the presented contributions, the remainder of this study is organized as follows. Section 2 outlines the modeling approach, including the hub-based system representation, the treatment of electricity and heat balances, the operational roles of hydrogen, battery, and EV storage, and the modeling of grid exchange within a rolling-horizon setup. Section 3 introduces the stylized community energy system used as the case study and summarizes the underlying data, assumptions, and storage scenarios reflecting Polish market conditions. Section 4 presents the main results, covering annual supply–demand structures, performance indicators, short-term operational dynamics, price-driven storage behavior, and cost outcomes across configurations. Finally, Section 5 concludes the study by synthesizing the key findings and their implications for community-level planning and future research.

2. Methods

This section presents the model-based approach developed to analyze the operational and cost performance of community energy systems integrating hydrogen, battery and EV storage technologies. The model provides a consistent representation of the main techno-economic characteristics of these storage options, and of their interaction with local generation, demand and the public grid. It is structured to quantify the resulting energy flows, associated operating costs and performance indicators such as self-sufficiency and self-consumption, and to enable alternative storage portfolios to be evaluated under different market and other local conditions, including those relevant to Poland.

2.1. System Representation and Energy Balances

The decision-support tool represents a community-scale hybrid energy system as a single electricity and heat hub at the point of common coupling (PCC). All relevant assets—local generation units, storage technologies and the public grid—are connected to this hub. This abstraction enables different storage portfolios to be compared on a common basis, while keeping the level of detail needed to capture their techno-economic characteristics. The overall modeling workflow of the proposed approach is presented in Figure 1.

Figure 1. Modelling workflow of the community energy system, including data inputs, model and outputs.

The configuration is sufficiently generic to accommodate a wide range of component portfolios. In particular, it can represent conventional dispatchable units (for example, gas engines), non-dispatchable renewable units (such as photovoltaic and wind installations), short-term electrical storage (lithium-ion batteries), hydrogen-based conversion and storage (electrolyzer, fuel cell and tank), aggregated electric vehicles, and, where relevant, electric boilers and thermal storage for space and water heating. All these technologies are modeled as interacting through the PCC, which serves as the accounting point for energy flows and costs.

At each time step, an electrical energy balance ensures that the sum of all electricity supplied to the hub equals the sum of all electricity withdrawn. On the supply side, the model accounts for the output of dispatchable generators, usable production from renewable sources after potential curtailment, discharging from batteries and fuel cells, and net imports from the grid. On the demand side, it gathers the aggregated community load, charging of electrical storage and electric vehicles, electricity consumed by the electrolyzer and electric boiler, and exports to the grid. All flows are expressed on the alternating-current (AC) side; conversion efficiencies for inverters, storage devices and hydrogen technologies are applied so that the power balance already reflects conversion losses.

A companion thermal balance captures the heating subsystem whenever thermal technologies are present. In each hour, the sum of heat delivered by the electric boiler and thermal storage discharging must satisfy the community heat demand; charging of thermal storage is treated as an additional thermal load on the boiler. Slack variables with large penalty coefficients are introduced for both balances to signal any infeasibility and to strongly discourage unmet demand. In line with established practice for operational optimization of hybrid energy systems, the temporal resolution is hourly and the optimization horizon is chosen to be long enough to capture typical variability in demand, renewable generation and electricity prices, while remaining computationally tractable.

This hub-based representation is central to the research questions of the paper. It provides a robust way to track energy flows associated with each storage technology and to compute performance indicators such as self-sufficiency, self-consumption and residual exchanges with the grid, under different combinations of hydrogen, battery and EV storage.

2.2. Modeling Assumptions and Simplifications

To ensure computational tractability and enable a transparent comparison of alternative storage configurations at the community scale, the proposed modeling framework adopts several simplifying assumptions commonly used in operation-oriented optimization studies. These assumptions are summarized below to support reproducibility and interpretation of the results.

First, electric vehicles are represented as an aggregated and flexible storage resource rather than individually modeled units. The model captures collective EV charging and discharging behavior through time-dependent availability windows and power limits, without explicitly representing individual mobility patterns or user heterogeneity. This aggregation allows EV flexibility to be incorporated at the community level while limiting model complexity.

Second, battery energy storage systems are modeled without explicit degradation effects. Battery aging, cycle-dependent wear and calendar degradation are not included, which is consistent with many short- to medium-term operational studies. As a result, long-term cost impacts related to battery degradation are not captured and considered when interpreting economic results.

Third, the optimization relies on deterministic forecasts of electricity demand, renewable generation and electricity prices. Uncertainty in weather conditions, load and market prices is not explicitly represented, implying idealized foresight and potentially optimistic performance outcomes.

Finally, the analysis focuses on operational decision-making and assumes fixed installed capacities for generation, storage and conversion technologies. Investment decisions and endogenous capacity sizing are outside the scope of the study, and distribution network constraints are not explicitly modeled. Grid interaction is represented through aggregate import and export variables under prevailing market rules.

These assumptions define the intended scope of the framework as a comparative, operation-oriented assessment of alternative storage portfolios under consistent market conditions.

2.3. Optimization Model and Decision Variables

The problem is formulated as a deterministic mixed-integer linear program (MILP). This choice is consistent with recent work on community energy systems and microgrids that uses MILP to minimize operating costs while enforcing detailed technical constraints and obtaining globally optimal solutions under linear approximations. In the present context, the MILP formulation is used as a tool to represent the techno-economic characteristics of hydrogen, battery and EV storage in a manner that directly supports the operational and cost assessment posed in the research questions.

Continuous decision variables represent, for each hour of the optimization horizon:

• Power outputs of dispatchable units and electric boilers;
• Usable output from photovoltaic and wind units and the associated curtailed energy;
• Power imported from and exported to the grid;
• charging and discharging power and state of charge of electrical and thermal storage units;
• Electricity consumption in the electrolyzer, power generation in the fuel cell and the hydrogen inventory in the tank;
• charging and discharging power and aggregate state of charge of the electric vehicle fleet;
• Slack and surplus variables associated with unmet electricity and heat demand, renewable curtailment and deviations from target storage levels.

Binary decision variables capture discrete operating modes and on/off decisions, including:

• Commitment, startup and shutdown status of dispatchable generators and electric boilers;
• Mutually exclusive charging, idle and discharging states for each storage technology;
• Mutually exclusive import and export modes in the grid interconnection, preventing simultaneous buying and selling in the same time step;
• Mutually exclusive electrolyzer and fuel cell operation in the hydrogen subsystem;
• Mutually exclusive charging and discharging modes for the aggregated electric vehicle fleet.

These binary variables are linked with continuous flows through standard unit-commitment constructs (technical minimum and maximum output levels, ramping limits, minimum up- and down-times) and big-M-type formulations that deactivate flows in incompatible modes. The combination of continuous energy flows and discrete operational states is what makes the model an MILP rather than a purely linear dispatch formulation.

Auxiliary continuous variables measure the net power exchange with the grid and the change in this exchange from one hour to the next. They are used to implement alternative objectives where smoothing the net grid profile is desirable, without introducing non-linear terms into the optimization problem. Together with the variables described above, these constructs enable a detailed accounting of how different storage portfolios translate into changes in dispatch patterns, grid exchanges and operating costs.

2.3.1. Dispatchable and Renewable Generation

Dispatchable generation units are modeled with explicit technical and temporal constraints. Their output is limited by minimum and maximum power levels whenever they are committed. Hourly ramp-up and ramp-down limits restrict changes in power output between consecutive hours to reflect mechanical and thermal capabilities. Minimum up- and down-times ensure that, once started or shut down, a unit must remain in that state for at least a specified duration. The initial state and operating history at the beginning of the horizon are parameters of the model and ensure consistent behavior across successive optimization runs.

Non-dispatchable renewable units (photovoltaic and wind) are characterized by exogenous availability profiles derived from measured, simulated or forecasted generation for the locations under study. In each hour, the available generation is split into a usable portion and a curtailed portion. Curtailment is allowed but penalized in the objective to reflect either lost feed-in revenues or the foregone opportunity to offset local demand. This structure allows the model to curtail renewables only when storage is saturated, export limits are binding or market conditions make curtailment preferable. The resulting time series of renewable utilization under different storage configurations feeds directly into the performance indicators examined in the results.

2.3.2. Electrical and Thermal Energy Storage

Lithium-ion battery storage is described by its rated power, energy capacity, charging and discharging efficiencies and standing losses. The state of charge follows a recursive relationship that accounts for net charging and discharging during each hour, subject to lower and upper bounds representing depth-of-discharge limits and usable capacity. Charging and discharging powers are bounded by the converter rating and by the maximum charge and discharge durations. Binary mode variables enforce mutually exclusive charging and discharging, so that the device cannot inject and absorb power simultaneously. In the present formulation, battery degradation is not modeled explicitly; instead, round-trip losses and market prices already discourage unnecessary cycling over the studied horizons. This level of detail is sufficient to represent the operational and cost implications of battery storage in the comparative analysis.

Thermal energy storage, if present, is modeled analogously in terms of stored thermal energy. The state of charge is bounded by minimum and maximum thermal capacities, while charging and discharging powers are limited by the thermal equipment ratings and conversion efficiencies. As in the electrical case, exclusive charging and discharging modes are enforced. Unserved heat demand is strongly penalized in the objective to avoid comfort violations in the heating supply.

2.3.3. Hydrogen Conversion and Storage

The hydrogen subsystem consists of an electrolyzer, a fuel cell and a pressurized hydrogen tank. Two mutually exclusive operating modes are defined for each hour: hydrogen production through the electrolyzer and hydrogen consumption through the fuel cell. The hydrogen inventory in the tank evolves by adding electrolyzer output and subtracting fuel cell input, while remaining within minimum and maximum storage limits expressed in mass or energy units.

Electrical coupling is achieved by linking electrolyzer operation to electricity consumption at the hub and fuel cell operation to electricity production, through appropriate conversion efficiencies. Startup and shutdown costs can be assigned to transitions in electrolyzer and fuel cell operating states to reflect additional wear or auxiliary consumption during transients. Upper bounds on hourly hydrogen production and consumption represent power limits of the conversion units, while bounds on the tank inventory represent pressure or volume constraints.

In the context of the research questions, this representation of hydrogen technologies allows the model to capture their role as a long-duration storage option and to quantify how they shift energy across hours and influence operating costs compared with batteries and EVs.

2.3.4. Aggregated Electric Vehicles

Electric vehicles are represented in an aggregated fashion, typically by one or several homogeneous fleet segments. For each segment, the model tracks the state of charge and charging and discharging powers over time. Rated charging and discharging powers limit the instantaneous exchange with the hub. Mutually exclusive charging and discharging modes are enforced whenever vehicles are connected. Time-dependent availability profiles specify the hours when each segment is plugged into the community infrastructure; outside these windows, the fleet cannot exchange power with the hub, but the state of charge is still tracked conceptually.

Additional constraints can enforce minimum state-of-charge requirements at certain times of day to preserve mobility needs. Deviations from these targets are allowed but penalized in the objective, ensuring that the optimization model uses vehicle-to-grid capabilities as a flexibility resource without systematically compromising user-defined requirements. This structure enables the contribution of EV storage to self-consumption and cost reduction to be assessed alongside that of stationary batteries and hydrogen.

2.4. Grid Interaction and Operational Objectives

Interaction with the public grid is modeled at a single PCC. For each hour, the model decides whether the community is in import mode or export mode through a pair of binary variables; this prevents simultaneous buying and selling and avoids artificial arbitrage within the same time step. When import mode is active, the power imported from the grid is bounded by the contracted connection capacity and is costed using time-varying prices, typically derived from day-ahead market data. When export mode is active, the power exported to the grid is limited by technical or contractual constraints and yields revenues based on time-varying feed-in or market prices. If needed, specific hours can be fixed to minimum import or export levels to reflect regulatory obligations or bilateral agreements.

The primary objective of the model is to minimize the total operating cost of the hybrid energy system over the optimization horizon. The cost function aggregates:

• Fuel costs of dispatchable generators;
• Fixed and variable operation and maintenance costs of all technologies;
• Startup and shutdown costs of dispatchable units and, where applicable, hydrogen conversion devices;
• Costs of purchasing electricity from the grid and revenues from selling electricity to the grid;
• Penalties associated with renewable curtailment, unserved electricity and heat demand, and deviations from target states of charge for storage units and electric vehicles.

The design of the objective function and grid interaction constraints is consistent with the intended application of the decision-support tool to specific market and regulatory contexts. In particular, price time series, tariff structures and export rules can be parameterized to reflect local conditions, which is essential for answering how community energy systems can identify suitable storage mixes under the current national framework.

2.5. Rolling-Horizon Operation and Implementation

The framework operates in a rolling-horizon (receding-horizon) mode, similar to model predictive control. In the baseline configuration used in this study, each optimization window covers 168 consecutive hours, corresponding to one week. At the beginning of a window, the model is initialized with the current state of the system, including the commitment status of dispatchable units, states of charge of all storage devices, hydrogen inventory and the state of charge and availability of electric vehicles. Forecasts of electricity and heat demand, renewable generation and electricity prices over the 168 h look-ahead horizon are supplied as input.

The MILP is then solved for the entire 168 h horizon, but only the first 24 h of the resulting schedule are retained. These 24 h are used both for the analysis of operational and cost performance and for updating the system state at the end of the day. The horizon is then shifted forward by 24 h, new forecasts for the subsequent week are ingested, and the optimization is repeated. This iterative procedure allows the framework to combine a relatively long planning horizon with a shorter implementation step, capturing intertemporal storage effects while remaining computationally tractable. The length of the look-ahead horizon and the implementation interval can be modified if needed, for example to reflect different data resolutions, forecast availability or planning requirements.

The model is implemented in GAMS (General Algebraic Modeling System) and solved using Gurobi (13.0.0). Input data are structured in tabular files that specify sets, parameters and time series for all technologies and exogenous signals. For typical horizons and system capacities, solution times remain compatible with day-ahead or intraday planning. Post-processing scripts extract and aggregate the main decision variables and cost components, enabling detailed analysis of dispatch patterns, storage utilization, grid exchanges and the contribution of each storage technology to the overall performance of the community energy system.

Figure 2 summarizes the MILP decision-making process for representative winter (January) and summer (July) conditions. The optimization model maps time-independent and time-varying inputs (demand, RES availability, prices, and constraints) into hourly dispatch decisions, so seasonal differences in operation reflect changing inputs and binding constraints rather than different model structures.

Figure 2. Conceptual illustration of the MILP decision-making process.

3. Case Study

The applicability of the model is demonstrated through a case study of a stylized local energy community, constructed to represent a plausible configuration of consumers, generators, and energy storage technologies. The study comprises two large multi-family buildings (24 dwellings each) and 100 single-family houses. All elements of the hybrid energy system: photovoltaic installations (PV), small wind turbines (WT), a gas engine (GE), an electric boiler (EB), a thermal energy storage tank (TES), fuel cells (FC), electrolyzers (EL), battery energy storage systems (BES), electric vehicle batteries (EV), hydrogen tanks (HT), as well as electricity exchange with the grid are aggregated in the model into single instances with cumulative capacities (e.g., instead of modeling six 8 kW WT units, a single 48-kW turbine is represented). The considered case constitutes a multi-energy system in which both space heating and electricity demand are represented by annual hourly time-series tailored to the characteristics of the Polish residential sector. The policy context also incorporates electricity prices corresponding to the hourly values declared on the Day-Ahead Market for the year 2024, including negative price occurrences. A simplified schematic representation of the analyzed community energy system is provided in Figure 3.

Figure 3. Simplified schematic of the analyzed hybrid energy system.

The selection of the case study in terms of system scale is aligned with the European trend of citizen-led and decentralized energy development, according to which, in 2024, approximately 9000 energy communities were operating across the EU, involving more than 1.5 million citizens [49]. In Poland, several national initiatives responding to the EU Renewable Energy Directive II (RED II) have been implemented [50,51]. Currently, Poland has 217 energy cooperatives listed in the registry maintained by the Director General of National Support Centre for Agriculture (NSCA) [52], as well as 54 energy clusters operating across the country [53]. The establishment of such collective structures enables local investments in renewable energy installations and facilitates collective energy balancing (e.g., sharing surplus PV generation), thereby reducing dependence on external suppliers and generating both economic and environmental benefits for members [54]. Nevertheless, the operation of energy communities in Poland requires meeting a number of regulatory criteria depending on the organizational form. For example, to obtain the status of an energy cooperative, according to the Polish Renewable Energy Sources Act, at least 70% of the electricity balanced within its area of operation must originate from renewable sources, while the total installed capacity may not exceed 10 MW of electrical capacity (or 30 MW of thermal capacity) [24]. Energy clusters are likewise expected to demonstrate a high share of renewable energy and meet efficiency-related objectives [24]. The coordination of systems that must additionally comply with such regulatory requirements becomes increasingly challenging; therefore, a modeling approach capable of testing hybrid energy solutions can support emerging community-based initiatives.

In a more detailed context concerning the development of individual prosumer technologies, Poland also implements several measures that encourage investment decisions aligned with sustainable development goals. The National Fund for Environmental Protection and Water Management (NFEPWM) operates the long-running (2024–2027) “My Electricity” program, which provides financial support primarily for residential PV micro-installations, but also for battery storage systems [19]. Additional financial incentives are directed towards investments aimed at electrifying building heating systems, as exemplified by the “Clean Air” program, which subsidizes the replacement of solid fuel boilers with electric heating technologies [20]. Further governmental incentives target prospective electric vehicle owners, who by participating in a program such as “Our Car” can reduce the financial burden of vehicle acquisition and actively contribute to the electrification of the transport sector [21]. The effectiveness of these measures is reflected in recent statistics. The number of PV micro-installations in Poland has been continuously increasing and exceeded 1.5 million by the end of 2024 [55]. The year 2024 was also marked by a pronounced rise in the adoption of residential battery storage systems, with prosumers installing nearly 47 thousand units [56]. Interest in electric vehicles has likewise been growing: by the end of August 2025, more than 100 thousand fully electric passenger and light-duty vehicles had been registered in Poland—an increase of nearly 30 thousand compared to the end of 2024. Furthermore, the cumulative growth observed in the first eight months of 2025 was 74% higher than during the same period of the previous year [57]. Among the analyzed technologies, hydrogen-based residential storage remains the least widespread solution in Poland. Despite ongoing research activities [47,58], no statistical data have been identified that would indicate any significant current or anticipated deployment of household-scale hydrogen tanks within the Polish residential sector.

Given that the present work aims to investigate how the presence and type of electricity storage technologies influence operational costs and energy balances, this study evaluates the performance of the system under multiple configurations that differ in the adopted electricity storage strategies. The key system parameters and modeling assumptions used in the construction of the case study are summarized in Section 3.1, while the characteristics of the analyzed storage scenarios are presented in Section 3.2.

3.1. Data Sources

All input data were organized in accordance with the hub-based system representation outlined in Section 2, and the dataset comprises technological parameters, operational costs, and annual hourly time series. The nominal power and energy capacities of all system components are summarized in Table 2, where, following the adopted methodology, each technology is represented as an aggregate reflecting the cumulative capacity of the entire community energy system. The adopted capacity levels do not represent an optimized investment design, but are selected to reflect typical orders of magnitude reported for residential energy communities and collective prosumer installations in recent European studies. In particular, community-scale configurations comprising approximately 20–200 households and total renewable electricity capacities in the range of 100 kW to 1 MW are commonly reported [59,60]. This ensures that all storage technologies are evaluated on a comparable energetic basis, and that differences in model outcomes arise from their operational characteristics rather than from differences in storage capacity.

Table 2. Rated power and storage capacities of all technologies included in the community energy system.

Parameter	Unit	Value
PV capacity	kW	300
WT capacity	kW	48
GE capacity	kW	5
TES capacity	kWh	200
EB capacity	kW	400
BES capacity	kWh	126
BES max. charge/discharge time	h	2
HT volume	L	640
EL power	kW	63
FC power	kW	63
EV battery capacity	kWh	120
EV battery power	kW	15

The capacity of the PV and WT installations was determined based on representative values typical for small-scale prosumer systems—6 kW for PV and 3 kW for WT [61,62]. It was assumed that half of the single-family houses are equipped with PV installations, whereas the system includes sixteen small wind turbines (aggregated as a single unit). The power rating of the GE (after accounting for efficiency) corresponds to the average hourly electricity demand of one of the two multi-family buildings, ensuring an adequate representation of minimum load coverage requirements. The BES capacity was set using a ratio of 1.4, which corresponds to the relation between the average battery capacity and the average PV installation capacity in Polish prosumer systems [55,56]. This empirically grounded proportion was combined with the assumption of a two-hour storage duration, which is commonly applied in residential-scale analyses [63]. For the hydrogen-based technology, the storage tank volume was set to 640 L so that its usable energy content would be comparable to the adopted BES configuration under the conditions represented by the parameters in Table 3. For the same reason, the nominal capacities of the EL and FC units were both set to 63 kW.

The parameters of the aggregated EV storage unit are based on representative producer specifications [64] and on the assumption that its usable energy capacity should be comparable to that of the BES and HT. Ultimately, a battery capacity of 120 kWh and a charging/discharging power of 15 kW were adopted, corresponding to three electric vehicles for the entire community. The power rating of the EB and the TES storage volume were aligned with the peak thermal demand of the multi-family buildings, while for the remaining buildings it was assumed that heating needs are met by an external district heating system, which is not included in the energy or cost balances of the model. Table 3 and Table 4 summarize the technical parameters—including device efficiencies—and the unit operational costs of all components, which determine the system’s optimal behavior, particularly decisions related to charging and discharging of storage technologies and interactions with the electricity grid.

Table 3. Technical performance parameters and efficiency assumptions applied in the model.

Parameter	Unit	Value	Source
GE efficiency	%	35	[65]
TES roundtrip efficiency	%	96	[66]
TES holding efficiency	%	99	[67]
EB efficiency	%	95	[68]
BES roundtrip efficiency	%	96	[69]
BES holding efficiency	%	99	[70]
EL efficiency	%	65	[71]
FC efficiency	%	40	[71]
HT max. pressure	bar	200	[72]
HT min. pressure	bar	40	[72]
HT temperature	K	293	[72]
H2 LHV	kWh/kmol	67.2	[72]
EV battery roundtrip efficiency	%	81	[73]

Table 3. Technical performance parameters and efficiency assumptions applied in the model.

Parameter	Unit	Value	Source
GE efficiency	%	35	[65]
TES roundtrip efficiency	%	96	[66]
TES holding efficiency	%	99	[67]
EB efficiency	%	95	[68]
BES roundtrip efficiency	%	96	[69]
BES holding efficiency	%	99	[70]
EL efficiency	%	65	[71]
FC efficiency	%	40	[71]
HT max. pressure	bar	200	[72]
HT min. pressure	bar	40	[72]
HT temperature	K	293	[72]
H2 LHV	kWh/kmol	67.2	[72]
EV battery roundtrip efficiency	%	81	[73]

Table 4. Economic parameters and variable operating costs applied in the model.

Parameter	Unit	Value	Source
PV variable cost	€ct/kWh	0.3	[74]
WT variable cost	€ct/kWh	0.5	[74]
Gas cost	€ct/kWh	4.6	[75]
GE variable cost	€ct/kWh	1.2	[76]
EB variable cost	€ct/kWh	4.6	[77]
EL variable cost	€ct/kWh	3.0	[78]
FC variable cost	€ct/kWh	1.4	[79]

Table 4. Economic parameters and variable operating costs applied in the model.

Parameter	Unit	Value	Source
PV variable cost	€ct/kWh	0.3	[74]
WT variable cost	€ct/kWh	0.5	[74]
Gas cost	€ct/kWh	4.6	[75]
GE variable cost	€ct/kWh	1.2	[76]
EB variable cost	€ct/kWh	4.6	[77]
EL variable cost	€ct/kWh	3.0	[78]
FC variable cost	€ct/kWh	1.4	[79]

Part of the input data is provided to the model in the form of hourly time series. These data include dynamic electricity prices in the form of day-ahead market prices for each hour of 2024 [80], as well as the electrical and thermal loads. The electrical load profile was constructed using typical hourly consumption patterns for tariff group G11, published by the distribution system operator as representative profiles for each month and day type [81]. The annual consumption level is based on the most recent data from Central Statistical Office (CSO) concerning energy use in households [82]. The same data source was used to define the total annual volume of heat supplied to the buildings. The thermal load profile was generated analogously to the method presented in [83]. Additional time series used in the model represent capacity factors of PV and WT technologies, sourced from the Renewables Ninja database [84] for a location in Southern Poland and the year 2024. Together, these time-dependent inputs provide a localized context for this study, aligning the simulated conditions with the characteristics of the Polish energy market, the renewable energy potential, and average consumption patterns. EV availability in terms of connection to energy sources and loads was adopted based on the data reported in [85]. The final set of model parameters concerns the interaction with the electricity grid, namely the limits on energy import and export. The admissible export volume in each hour was set equal to the corresponding electricity demand, following the approach presented in [86].

3.2. Scenarios

The assessment of how the choice and configuration of energy storage technologies affect the operation of the modeled community energy system is carried out for seven operational scenarios. The scenario set is designed to enable a systematic comparison of different storage technologies with varying temporal and operational characteristics, and to examine whether their joint operation leads to complementary or competing flexibility provision. Each scenario represents a distinct combination of storage devices, allowing both individual effects and interaction effects to be evaluated under identical market and demand conditions. The configurations considered in each scenario is provided in Table 5.

Table 5. Configuration of storage technologies across all scenarios.

Technology\Scenario	Single-Component			Two-Component			Full Integration
	BES	H2	EV	BES + H2	BES + EV	H2 + EV	BES + H2 + EV
BES	✓			✓	✓		✓
H2		✓		✓		✓	✓
EV			✓		✓	✓	✓

The first group consists of single-component scenarios in which the system relies on only one type of storage device: a stationary battery storage system (BES), a hydrogen-based storage system comprising an electrolyzer, a tank, and a fuel cell (H₂), or an electric-vehicle-based storage system (EV). These scenarios enable the evaluation of the operational characteristics of each technology in isolation, considering differences in their suitability for short-term or seasonal energy balancing, performance limitations, and availability constraints (in the case of EVs).

The second group includes hybrid scenarios composed of two storage technologies capable of storing electrical energy: BES + H₂, BES + EV, and H₂ + EV. These scenarios are motivated by prior studies on hybrid energy storage systems that discuss the potential benefits of combining storage technologies with different temporal characteristics, without providing a consistent, operation-oriented assessment at the community scale [87,88,89]. Accordingly, the hybrid scenarios are designed to examine whether such complementarities emerge under identical system and market conditions.

The final scenario (BES + H₂ + EV) represents the full integration of all three storage technologies, allowing for the assessment of the maximum flexibility potential of the system resulting from the parallel operation of battery, hydrogen, and mobile storage units. This scenario is included to evaluate the upper-bound flexibility potential within the modeled community energy system and to assess whether the addition of multiple storage options yields incremental operational value beyond pairwise combinations.

For all scenarios, identical input datasets were used, excluding only the parameters of those storage technologies that were not included in a given scenario.

4. Results

This section provides an overview of the results obtained from the implementation of the approach introduced in Section 2 to the case study and research scenarios described in Section 3. The outcomes include both the hourly profiles of individual variables and the aggregated annual indicators. To ensure a clear graphical presentation of the hourly-resolution deliverables, the first week of January and July was selected as two representative samples. The analysis of the results is presented in the subsequent subsections, combining the annual supply–demand structure of the optimized system, yearly technical performance indicators, representative hourly operating profiles, and an assessment of how differences in storage characteristics translate into distinct cost-optimal operational strategies and associated trade-offs between energy autonomy, flexibility, and operating costs.

4.1. Electricity Generation and Demand Structure

The annual electricity supply of the analyzed system for all scenarios is presented in Figure 4a. To support the interpretation of both the annual electricity supply and demand structures, the corresponding numerical values are reported in Table 6. The results indicate that the generation from PV, WT, and GE remains constant regardless of the scenario. Moreover, the combined output of RES, together with imports and storage discharge, is sufficiently high that the contribution of the GE unit is negligible, not exceeding 0.1% of the total supply in any configuration. The volumes of electricity purchased from the grid and withdrawn from storage vary across scenarios, despite the identical levels of electricity and heat demand (the latter derived from electricity-to-heat conversion). This relationship demonstrates how differences in storage efficiency, power capacity, and availability shape the cost-optimal operating strategy of the system. Among the individual storage options, BES primarily operates as a short-term, high-efficiency buffering technology that responds to price signals, whereas hydrogen storage acts as a long-term energy sink whose conversion losses limit the amount of electricity effectively recovered. In contrast, the standalone EV battery provides the smallest contribution to system balancing, accounting for approximately 0.9% of total electricity supply (11.1 MWh), compared with 1.4% (17.4 MWh) for the H₂ variant and 5.6% (71.5 MWh) for BES. The aggregated representation of electric vehicles implies that this contribution reflects average fleet availability, which smooths short-term charging and discharging variability compared with individually modeled vehicles.

Figure 4. System annual electricity (a) supply structure, (b) demand structure for all considered scenarios.

Table 6. System annual electricity supply and demand structure for all considered scenarios (numerical values corresponding to Figure 4).

Scenario	BES	H2	EV	BES + H2	BES + EV	H2 + EV	BES + H2 + EV
Electricity supply components	Supply structure (MWh)
PV	357.5	357.5	357.5	357.5	357.5	357.5	357.5
Wind	95.6	95.6	95.6	95.6	95.6	95.6	95.6
Gas engine	0.2	0.2	0.2	0.2	0.2	0.2	0.2
Grid import	750.3	735.5	753.9	751.3	763.0	748.3	760.4
Battery discharging	71.5	0.0	0.0	60.4	69.8	0.0	61.8
Fuel cell	0.0	17.4	0.0	15.9	0.0	16.3	13.8
EV discharging	0.0	0.0	11.1	0.0	10.2	11.5	11.2
Electricity demand components	Demand structure (MWh)
Electricity demand	382.7	382.7	382.7	382.7	382.7	382.7	382.7
Grid export	64.3	48.5	44.2	70.1	65.3	49.7	70.6
Electric boiler consumption	748.2	708.1	757.6	698.7	736.7	701.9	694.7
Battery charging	79.8	0.0	0.0	68.1	78.1	0.0	69.6
Electrolyzer consumption	0.0	66.8	0.0	61.2	0.0	62.8	53.1
EV charging	0.0	0.0	32.6	0.0	32.5	30.9	28.4

An interesting observation that also contributes to the differences in the supply structure is the interaction between the electricity storage units, the electric boiler, and the thermal energy storage (TES) system. This interaction enables temporal load shifting, ultimately determining the amount of electricity consumed by the electric boiler for TES charging as well as the real-time heat demand. Regardless of the number of storage components considered, scenarios that include BES consistently achieve the highest share of discharged energy within the total supply volume. However, the H₂ and H₂ + EV scenarios exhibit the lowest grid electricity imports, corresponding to approximately 61% of the total supply (735.5 MWh and 748.3 MWh, respectively). This behavior results from the absorption of local RES surpluses by the energy-intensive electrolyzer, which lowers grid imports even though only a limited share of this energy is later recovered through the fuel cell.

The presented relationships are directly reflected in the structure of electricity demand shown in Figure 4b, where clear differences are visible in the electricity used for heating, which accounts for approximately 53.5–62.2% of total demand. This variability arises from the previously discussed interaction between the electricity storage units, the electric boiler, and the TES system—surplus low-cost electricity is routed to the electric boiler and thermal storage, increasing the relative share of boiler consumption. Similar to the variability observed in the discharged energy volumes, the charging volumes also changed considerably across the scenarios. The annual amount of charged energy was lowest for the EV battery and highest for the BES. Hydrogen storage exhibits the largest gap between charged and discharged energy, reflecting conversion losses that distinguish it structurally from BES and EV-based storage. These losses correspond to comparable relative values in the EV and BES scenarios.

Although differences are also visible in the volume of excess electricity sold to the grid, its share in the total outflow from the system remains between 3.6% (44.2 MWh for the EV scenario) and 5.5% (70.1 MWh for the BES + H₂ scenario). The highest exported volume occurs in the most comprehensive configuration, BES + H₂ + EV, amounting to 70.6 MWh over the entire year. The relatively low yet differentiated level of export results from the fact that most of the local generation is first consumed within the system—either directly or via storage charging—and only the residual surplus that cannot be effectively utilized internally, and does not exceed the imposed export limit, is fed into the grid. The highest export volume in the BES + H₂ + EV scenario reflects the greatest overall storage flexibility, which increases the availability of surplus energy once local demand is met, rather than the maximization of electricity exports. A summary of the annual values of key energy flows and operational indicators, enabling assessment of the impact of storage configuration on system performance, is provided in Table 7.

Table 7. Annual operational variables and indicators for all considered scenarios.

Annual Variable/ Indicator	Unit	BES	H2	EV	BES + H2	BES + EV	H2 + EV	BES + H2 + EV
Self-sufficiency	%	33.7	32.6	33.9	30.5	31.8	31.0	29.4
Self-consumption	%	85.8	89.3	90.3	84.5	85.6	89	84.4
Charged energy	MWh	79.8	66.8	32.6	129.3	10.5	93.8	151.2
Discharged energy	MWh	71.5	17.4	11.1	76.3	80.0	27.8	86.7
Energy import	MWh	750.3	735.5	753.9	751.3	763	748.3	760.4
Energy export	MWh	64.3	48.5	44.2	70.1	65.3	49.7	70.6
Peak import	kW	480.0	471.9	475.8	480.0	480.0	475.8	480.0
Peak export	kW	183.5	159.1	173.5	183.5	184.6	173.5	184.6

4.2. Annual Operational Indicators and Storage Technology Performance

The operational performance indicators reported in Table 7 reflect how the model formulates a cost-optimal operating strategy under different configurations of energy storage technologies. One of the key indicators is the community’s energy self-sufficiency, which ranges from 29.4% to 33.9%. Notably, the BES + H₂ + EV configuration—the system with the greatest technical flexibility—exhibits the lowest self-sufficiency (29.4%), as cost-optimal operation favors increased grid interaction, resulting in the highest annual electricity imports (760.4 MWh) despite the availability of multiple storage options. A similar pattern is observed for self-consumption, whose lowest—yet still comparatively high—values occur in scenarios involving BES. This relationship highlights the difficulty of maintaining a high degree of independence from the external grid and maximizing the use of locally generated energy while ensuring low operating costs, even when substantial storage capacity is available. The highest self-consumption levels are observed in the EV (90.3%) and H₂ (89.3%) scenarios, not due to superior flexibility, but because limited storage capabilities constrain exports and force direct local use of electricity. As battery degradation is not represented explicitly, the model may favor more frequent BES cycling than would be obtained under an explicit aging-cost formulation.

A reduction in the maximum grid import power relative to the allowable level occurs only in the scenarios without the BES system, whereas in all remaining variants the maximum peak electricity purchase is reduced by merely 0.9–1.7%. The lack of pronounced differences between scenarios suggests that even substantial storage capacities do not alter the instantaneous maximum import level, as peak demand occurs during hours in which grid imports are cost-optimal or when local generation cannot cover sudden increases in load. The peak power exported to the grid shows a somewhat wider range of 159–185 kW. It is noteworthy that adding a hydrogen tank to the system does not affect this indicator in any of the configurations, since hydrogen storage—being a technology with high conversion losses—is not used for short-term management of power surpluses. Consequently, it does not influence instantaneous export peaks. Thus, the system does not operate in a way that maximizes grid independence, but instead optimizes the timing of imports and exports—making full use of storage technologies where they yield the greatest cost reductions, even if this leads to seemingly lower levels of self-sufficiency in the most advanced configurations. These results indicate a structural trade-off between reducing annual energy imports and managing instantaneous power peaks, as storage technologies are effective for energy shifting and, in this particular case, do not necessarily provide peak-shaving capability.

4.3. Operational Dynamics of the System

A deeper understanding of the variability in annual outcomes requires examining how cost-optimal short-term operational decisions and storage constraints translate into the observed system behavior. Given the full time series (8760 h) and the seasonal nature of renewable generation and energy loads in Poland, the presentation of hourly results is limited to two representative weeks—one winter and one summer. The supply and demand structure across the hours of these representative weeks is shown in Figure 5 and Figure 6, respectively. To further support the interpretation of the hourly profiles and to provide a transparent, numeric illustration of the model’s short-term decision-making, representative 24 h operational snapshots for a winter day (7th of January) and a summer day (7th of June) are reported in Appendix A (Table A1 and Table A2). These tables complement Figure 5, Figure 6 and Figure 7 by showing the corresponding hourly values of key inputs and optimized operational outcomes.

Figure 5. Hourly energy supply structure for all scenarios during the first week of January and July.

Figure 6. Hourly energy demand structure for all scenarios during the first week of January and July.

Figure 7. Hourly charging and discharging patterns of storage technologies and electricity price for all scenarios during the first week of January and July.

The obtained characteristics for the January week indicate a dominant share of imported electricity in the supply structure during the winter period. Despite the increased WT generation, the strong seasonal decline in PV output necessitates electricity purchases from the grid. Importing electricity remains a competitive option compared with GE. Although the contribution of storage units to the supply mix during this high-demand period is modest, only the hydrogen tank exhibits no activity, as also visible in Figure 7. Under conditions of low PV generation and elevated winter demand, the system avoids electricity-to-hydrogen conversion losses. The remaining storage technologies are discharged; however, only BES provides substantial temporal and power availability due to its relatively high power rating and the absence of strict charging and discharging constraints, in contrast to EV batteries. This behavior illustrates a clear hierarchy of storage roles in winter conditions: BES addresses short-term supply deficits and daily peaks, EV storage contributes only within limited availability windows, while hydrogen storage remains inactive because its conversion losses cannot be justified under low-surplus conditions. The use of deterministic demand, generation and price profiles leads to clearly defined charging and discharging schedules, whereas forecast uncertainty could result in more dispersed short-term operation. The near-identical daily timing of BES discharge across scenarios indicates that the battery functions as a predictable short-term buffer responding to recurring diurnal deficits in RES generation rather than to scenario-specific system configurations. The highest share of storage discharge in the winter week’s supply structure occurs in the $\mathrm{BES}+\mathrm{EV}$ configuration (and in BES + H₂ + EV, due to the inactivity of hydrogen technologies) and in the BES (and BES + H₂) variants, amounting to 3.6% and 3.14%, respectively.

During the summer period, the influence of the applied storage technologies and their configurations becomes more clearly visible in the energy mix, with the exception of the share of energy discharged from storage, which reaches only about 0.7% in the EV variant. This value is similar to that observed in January, indicating that seasonal fluctuations in supply and demand have a limited impact on the operational capabilities of the electric vehicle battery. The most effective configuration in terms of covering a portion of demand (consistent with the conclusions drawn from the annual results) is the scenario combining all three types of storage. Furthermore, BES provides the largest share of discharged energy in summer, reflecting both the high conversion losses of hydrogen technologies and the battery’s rapid response to changing system conditions. This contrast highlights a seasonal trade-off: hydrogen storage is effective in absorbing sustained summer surpluses, while BES remains superior for rapid, short-term balancing despite lower total energy volumes. This effect is evident through the regular, cyclic operation of the fuel cell in the evening hours and the irregular, temporally dispersed discharge pattern of the BES, demonstrating its ability to flexibly support an economically favorable energy balance over short time horizons.

A similar pattern can be observed when analyzing the demand structure shown in Figure 6, particularly in the July week. Configurations that include BES are characterized by sharper, more pronounced peaks in electric boiler demand, which largely stem from the discharge of the battery storage. Between these demand peaks, short one- to two-hour increases appear, corresponding to BES charging cycles. In the H₂ scenario, the overall demand profile is considerably smoother, and the charging of the hydrogen storage unit typically occurs in multi-hour, stable sequences of continuous electrolyzer operation, reflecting the slow-varying and less flexible characteristics of this technology. During the summer week, approximately 2.8 MWh of electricity was stored in the hydrogen tank—a value more than 50% higher than that stored in the BES. Scenarios involving hydrogen technology as one of the storage components therefore prove highly effective in conditions of substantial RES surplus, despite conversion losses. The implementation of EV storage, on the other hand, enables the utilization of relatively small excess energy volumes (approximately 0.65 MWh when used alone, and 0.44 MWh in the BES + H₂ + EV configuration), including surplus generated by the fuel cell, which in configurations without the EV battery would otherwise be exported to the grid. These patterns confirm that hydrogen storage operates over multi-hour horizons with stable power levels, whereas BES and EV storage exploit short, high-frequency flexibility, leading to fundamentally different impacts on demand profiles.

In January, similarly to the electricity supply profile, the system does not utilize energy surpluses for hydrogen production in any of the configurations. The lack of hydrogen-technology activity during this period results from the fact that the winter generation profile does not provide sufficient surplus energy to cover the high energy cost of electrolysis, leading the model to prefer direct discharge of electrical storage units instead of initiating the long conversion cycle. Any surplus electricity is directed predominantly to the BES and EV batteries, with their combined charging in the $\mathrm{BES}+\mathrm{EV}$ variant amounting to 1.68 MWh, and in the individual configurations reaching 1.20 MWh (BES) and 0.52 MWh (EV), respectively. The prioritization of BES and EV charging reflects their higher round-trip efficiency and lower activation thresholds, confirming that hydrogen conversion is activated only under conditions with sufficiently large and sustained energy surpluses.

4.4. System Response to Electricity Prices

The short-term operation of storage technologies is shaped not only by seasonal demand patterns and renewable generation variability, but also by electricity prices on the Day-Ahead Market (RDN). Hourly prices reflect the system-wide balance of supply and demand, which may differ substantially from local residential conditions. As a result, the energy storage system, responding to price signals, may undertake operational decisions that diverge from intuitive patterns based solely on RES output or local demand levels. Observing the relationship between storage operation and the cost of grid electricity is also crucial due to the objective function of the proposed approach. The results for the first week of both months, illustrated in Figure 7, show that despite being exposed to the same price signal, each storage technology operates differently—a consequence of differences in efficiency, power capacity, temporal availability, and interactions with other system components.

In the case of the BES, its behavior most closely resembles the classical pattern of aligning charging and discharging cycles with prevailing electricity prices. In January, the BES charges during more hours than it discharges, with the average price during charging hours being lower than—or comparable to—the price observed during discharging. In the summer week, the battery operates at a much higher frequency, storing more energy during low-price periods that coincide, for example, with increased PV generation (Figure 5 and Figure 6). These observations, combined with the results presented in Section 4.3, confirm the role of battery storage as a short-term buffer responding to both local supply deficits and market price signals. In configurations composed of multiple storage components (BES + H₂, $\mathrm{BES}+\mathrm{EV}$, BES + H₂ + EV), the BES acts as the primary technology responding to electricity prices, leaving the other storage units to perform more specialized tasks—such as handling large energy volumes or operating within narrow temporal windows. At the same time, in the summer period, a visual assessment of the BES response to price variability becomes more challenging due to the frequent, alternating occurrence of charging and discharging hours, whereas H₂ and EV technologies exhibit a much more block-like operational pattern. This further emphasizes the inherently short-term nature of BES operation.

The hydrogen technology exhibits strong seasonality and a distinctly different price-response pattern compared with the BES. In January, the H₂ storage system remains completely inactive across all configurations due to insufficient energy surpluses under low PV generation conditions and the high energy losses associated with the EL–FC cycle. Under such conditions, direct grid import is more cost-effective for the model than incurring hydrogen conversion losses. In July, however, this storage method becomes more relevant and enables multi-hour charging during sustained surplus periods or grid intake during hours with moderate or slightly elevated prices. These charging blocks occur on every day of the July week, along with shorter 1–3 h sequences. The average price level during charging and discharging hours is less differentiated than in the case of BES. Moreover, charging events at prices somewhat higher than those during discharging occur much more frequently. Hydrogen storage can therefore serve as a technology that does not respond dynamically to short-term price fluctuations but instead accumulates large amounts of energy over longer time horizons, effectively reducing the volume of electricity exported to the grid.

The EV battery, in turn, is characterized by strong constraints related to temporal availability, low power capacity, and requirements for full charging and discharging cycles. These factors limit its ability to align with the prevailing price structure. In January, the EV charges primarily during hours with relatively lower prices and discharges during higher-price periods, reflecting the use of available connection windows to support the system at economically favorable moments. In July, when price variability is greater and other storage technologies exhibit higher activity, the operational pattern of the EV becomes less distinct. Charging and discharging decisions are more strongly driven by vehicle availability. The specific characteristics of the EV scenario therefore confine its role to absorbing only small surplus energy volumes, including those originating from fuel cell operation.

In multi-component configurations, the price signal is addressed sequentially. This ordering emerges because the optimization model assigns flexibility first to the technology with the highest round-trip efficiency and fewest constraints (BES), then uses hydrogen mainly to absorb sustained surpluses (thereby reducing exports) despite conversion losses, and finally exploits EV capacity only when availability windows allow. The BES provides the most accurate response to the price structure, hydrogen storage absorbs surplus energy (particularly long-lasting surpluses), while the EV captures the remaining small portions of power during the hours in which it is connected. These results indicate that the described strategy effectively solves the problem of minimizing the system’s operational costs, but does not yield the highest levels of self-sufficiency or self-consumption. The response of storage technologies to dynamic electricity prices proves to be highly technology-specific, as it is weakened by factors such as low conversion efficiency, insufficient power capacity, or non-technical constraints such as vehicle availability schedules. This sequential allocation of price-driven flexibility reveals a clear trade-off between minimizing operational costs and maximizing energy autonomy, as technologies best suited for market arbitrage do not necessarily enhance self-sufficiency or self-consumption.

4.5. Cost Performance of the System

The results presented in the preceding sections—covering the supply and demand structure of the analyzed community energy system, the annual performance indicators, the detailed operational dynamics of the storage technologies, and their responses to short-term market signals—directly translate into the total operational costs of the system, which are minimized within the proposed optimization model. The analysis summarized in Table 8 reveals clear cost differentials among the various storage configurations, both in terms of cost structure and overall magnitude, highlighting the importance of technology choice and integration pathways under cost-minimizing operation.

Table 8. Annual operating cost breakdown across storage scenarios.

Annual Operating Cost	Unit	BES	H2	EV	BES + H2	BES + EV	H2 + EV	BES + H2 + EV
RES	k€	1.7	1.7	1.7	1.7	1.7	1.7	1.7
DU	€	44.4	44.4	44.4	44.4	44.4	44.4	44.4
Energy purchase	k€	72.3	71.8	75.0	71.2	73.4	72.8	72.3
Energy sell	k€	6.2	4.3	3.5	7.4	6.3	4.4	7.4
Average purchase	€ct/kWh	9.64	9.77	9.95	9.48	9.63	9.73	9.51
Average selling	€ct/kWh	9.59	8.77	7.93	10.52	9.61	8.82	10.44
H2 technology	k€	0.0	2.6	0.0	1.4	0.0	2.4	2.0
Heat generation	k€	33.1	31.4	33.6	30.9	32.6	31.1	30.8
Total OPEX	k€	101.0	103.3	106.8	97.9	101.5	103.7	99.5

The lowest total operational costs were observed in the BES + H₂ scenario, whose components—although not offering the highest storage capacity or power rating—enable a substantial reduction in the average electricity purchase price (9.48 €ct/kWh) and generate revenue through electricity exports during high-price periods (averaging 10.52 €ct/kWh). The large combined storage capacity, together with highly effective battery scheduling aligned with dynamic price fluctuations, also contributed to lower heating costs. However, this effect does not stem from a reduction in the unit cost of heat, but rather from more efficient TES charging, which resulted in the lowest heat production volume among all seven analyzed configurations (Figure 8). Achieving such low unit costs of electricity purchased from the grid, while simultaneously avoiding hydrogen-related operational costs, offsets both the relatively high import volume (751 MWh) and the lower export revenues compared with configurations including EV storage. It should be emphasized, however, that this favorable economic outcome comes with a trade-off: this scenario exhibits the second-lowest self-consumption and the second-lowest self-sufficiency levels among all examined variants.

Figure 8. Marginal cost of net energy and annual energy volumes by source and scenario.

Among the scenarios employing individual electricity storage technologies, the BES system proves to be the most economically favorable, with total operational costs amounting to 101 k€. The least cost-effective option is the EV-based variant, which is 5.7% more expensive. Although both technologies exhibit high efficiency, the advantage of BES stems from its more than fourfold higher power rating and the absence of temporal availability constraints associated with vehicle use, enabling substantially more flexible and cost-effective energy exchanges with the grid. Hydrogen-based storage could potentially compete with BES were it not for the relatively high variable costs of hydrogen production, which account for approximately 2.5% of the total operational costs in the H₂ scenario. It is worth noting, however, that the difference in operational costs between the H₂ and BES scenarios remains relatively small, despite the lower efficiency of the hydrogen technology and the inclusion of its additional variable costs. This outcome results from several key techno-economic mechanisms: the limited share of hydrogen discharge in total supply, the lower electricity demand of the electrolyzer relative to BES charging, and the dominant role of import and export costs in overall system economics.

The combination of all three energy storage technologies is characterized by only a slightly higher operational cost compared with the most economically efficient configuration—an increase of merely 1.6%. This indicates that the introduction of EV battery storage is an element that, depending on the system structure, may either enhance or reduce overall efficiency. It is the only one of the analyzed technologies (BES, H₂, EV) that introduces additional operational constraints resulting from the imposed vehicle availability schedule. Nevertheless, the impact of adding EV storage appears relatively small—both economically and technically—especially when compared with the differences observed among the remaining system configurations.

Aside from the storage technologies, the model also optimizes the operation of the remaining generation units. Regardless of the scenario analyzed, both the supply of energy from renewable sources (PV and WT) and the generation from the dispatchable unit (GE) remained at comparable levels. Moreover, the gas engine exhibited the lowest—practically negligible—share in total electricity production. This is due to the fact that it has the highest marginal cost among all available generation technologies, which automatically limits its utilization. The lowest marginal cost was observed for renewable sources, which are used first in the merit order. However, their utilization remains below the technically available level because of constraints related to installed capacity, storage capabilities, and imposed limits on electricity exports to the grid.

Across scenarios, configurations that increase short-term flexibility (BES and BES-containing portfolios) achieve lower average purchase prices (down to 9.48 €ct/kWh in BES + H₂) and the lowest total OPEX (97.9 k€), but tend to reduce self-consumption and self-sufficiency (down to 84.4% and 29.4% in BES + H₂ + EV) because cost-optimal operation increases grid interaction when prices are favorable. Conversely, constrained storage options (EV and H₂) limit exports and therefore yield higher self-consumption (90.3% and 89.3%), but at higher operating costs (up to 106.8 k€ in EV) due to weaker price arbitrage and limited ability to shift grid exchanges across hours.

5. Conclusions

This study develops an operation-oriented modeling approach for assessing the operational and cost-related role of energy storage technologies and storage portfolios in community energy systems, including battery storage, hydrogen-based storage, and the use of electric vehicle batteries. The developed decision-support tool enabled a consistent representation of techno-economic characteristics and interactions with local generation, electricity and heat demand, as well as market-based electricity prices. Its application to a stylized community system operating under Polish conditions allowed us to identify how different storage configurations shape energy flows, the utilization of renewable energy sources, and operational costs.

The results indicate that storage technologies exhibit distinct operational characteristics, which may become complementary when multiple storage types are integrated. These findings reflect the underlying operational mechanisms, with BES and EVs providing short-term, price-driven flexibility, in contrast to the intertemporal, surplus-oriented role of hydrogen storage under conversion losses. BES offers the greatest capability for balancing short-term mismatches between supply and demand due to its high efficiency and lack of availability constraints, and therefore plays a dominant role in responding to hourly price variations and short-term system imbalances. EV batteries, despite having similar charging and discharging efficiencies, are limited by power constraints and availability schedules, which makes them primarily suitable as auxiliary flexibility resources. Hydrogen-based technologies are most aligned with system needs during prolonged periods of surplus energy; however, they may become economically unattractive in winter, when renewable generation is limited and conversion losses dominate system behavior. Nevertheless, the energy losses associated with hydrogen conversion reduce the overall cost-effectiveness of this option.

None of the technologies considered is able to substantially increase self-consumption or self-sufficiency while simultaneously minimizing operational costs. Multi-component configurations do not automatically lead to the highest levels of energy independence, as the optimal system strategy is shaped by the interplay between market prices, storage availability, and the flexibility of energy conversion pathways. This highlights a structural trade-off between economically optimal operation and autonomy-oriented performance indicators that are often used to assess community energy systems—a central focus of the analysis presented in this study. However, multi-component scenarios do enhance system flexibility with respect to variable supply, demand, and market conditions, partially offsetting the associated operational costs and conversion losses.

Additionally, the analysis showed that under variable electricity prices, the short-term operational dynamics of storage technologies—particularly stationary BES—play a key role in shaping the system’s annual economic performance. Storage units respond not only to fluctuations in renewable generation and electrical load, but also manage the system in a way that exploits the dynamic character of electricity prices by reducing the average purchase cost and increasing the value of electricity sold. These relationships are revealed in the hourly analysis under different seasonal conditions, confirming that a full understanding of the value of each technology requires examining both annual performance indicators and short-term operational decisions.

The outcomes of the study highlight the complexity of selecting an appropriate configuration of energy storage technologies. The selection process should not focus on maximizing the number of components or the available storage capacity, but rather on understanding their role within the specific market context and the system’s generation and demand structure. No single technology offers a dominant advantage across all dimensions, and the effectiveness of each depends on a combination of factors including local generation, load profile, storage availability, and the applicable energy settlement rules. The results indicate that two- or three-component systems may be particularly effective when simultaneous responses to both seasonal and short-term conditions are required, but also suggest diminishing marginal benefits from adding further storage technologies once short-term flexibility needs are met.

From a planning perspective, the results translate the operational insights of the framework into concrete recommendations for Polish energy communities, regulators, and investors. For energy communities, stationary battery storage provides the most reliable reduction in operating costs under current price volatility, while EV-based storage should be treated as a supplementary flexibility resource rather than a primary balancing option. For investors, configurations combining BES and hydrogen storage are justified mainly where seasonal surpluses are substantial; otherwise, conversion losses weaken their economic performance. At the regulatory level, the findings suggest that uniform prosumer settlement mechanisms do not necessarily promote higher self-sufficiency, and that price signals remain the dominant driver of storage operation. Support schemes that prioritize short-term flexibility and price-responsive behavior over installed storage capacity could, therefore, better align private incentives with system-level objectives.

Although the presented approach offers valuable insights into operational and cost performance, its scope implies several methodological limitations that should be considered when interpreting the results. The analysis is based on deterministic representations of demand, renewable generation and electricity prices, which abstracts from uncertainty and forecast errors that may influence real-world storage operation. Grid interaction is represented through exogenous price signals and fixed import and export limits, without explicitly capturing network constraints or congestion effects. Moreover, the framework focuses on operational optimization and does not incorporate investment costs or endogenous capacity sizing, which constrains its applicability for long-term planning. Individual components, most notably EV storage, are therefore modeled in an aggregated manner that captures system-level behavior but does not reflect user-specific heterogeneity. Future research may extend the framework by incorporating stochastic or robust formulations, joint investment–operation modeling, and validation through application to real-world pilot energy communities operating under Polish regulatory conditions.