2.1 MiniStor system description

The MiniStor system is made up of four key sub-systems that work together to provide both thermal and electrical energy. First, it includes photovoltaic-thermal (PVT) and solar thermal collectors, supported by a buffer tank, which are responsible for generating the necessary heat. Second, it features a TCM reactor that operates with an ammonia cycle to store and release thermal energy. The third sub-system combines a heat pump with PCM storages—one for hot and one for cold thermal energy—ensuring efficient temperature management. Lastly, the system incorporates a battery that stores electricity produced by the PVT panels via an inverter. This battery supplies power to all essential electrical components of the system such as pumps and compressors. An overview of all system components of the MiniStor system and their integration into the building’s HVAC system is presented in Figure 1.

Figure 1 Overview of MiniStor system components.

At the heart of MiniStor lies the TCM reactor, a novel unit designed for dense heat storage using a solid-gas sorption process. The underlying chemistry involves reversible reactions between ammonia and calcium chloride, with the salt initially in the form CaCl₂·4NH₃. These reactions enable significant thermal energy storage and release due to the high sorbate absorption per mass of salt. During the winter months, the MiniStor system provides heating through two operating modes. In the charging mode, solar energy captured by the solar field is primarily used to supply the heat needed for desorbing ammonia from the thermochemical salts. When solar heat is insufficient, a 2-kW electric backup heater supplements the heat input. Once the TCM reactor reaches the required temperature, the pressure rises, initiating the endothermic decomposition reactions that release ammonia gas. This gas is then compressed, condensed, and stored in liquid form. To meet the building's heating needs, the heat released during condensation is further upgraded by a heat pump to reach temperatures around 63 °C. In the discharging mode—which takes place during winter nights, cloudy days, and in the summer—the stored liquid ammonia is allowed to evaporate when the pressure in the reactor drops below the ammonia's evaporation pressure. This causes ammonia to flow to the evaporator, where it vaporizes at ambient-dependent temperatures. When the gaseous ammonia enters the reactor, it undergoes an exothermic reaction that releases heat at 57–63 °C. This thermal energy is then either used directly for space heating or stored in the hot PCM. Additionally, during the summer, the cooling effect from ammonia evaporation is used to partially meet the building’s cooling demand. The resulting cooling output is stored in the cold PCM, which has a melting point of 11 °C.

The solar array driving the system consists of ten photovoltaic-thermal (PVT) panels and five flat-plate collectors (FPCs), all installed at an optimal tilt facing south. Each PVT panel combines high-efficiency monocrystalline silicon cells with thermal absorbers and offers an aperture area of 1.55 m², delivering a maximum electrical output of 260 W. Their optical and thermal loss characteristics are also optimized for dual energy generation. The FPCs, each with a gross area of 2.37 m², an optical efficiency of 0.823 and heat loss coefficients α₁=3.36 (W/m²K) and α₂=0.013 (W/m²K) respectively. To prevent overheating, an external heat exchanger cools the heat transfer fluid if necessary.

A small buffer tank and the backup heater are integrated into the system to stabilize the heat supply, especially important for meeting the TCM reactor's temperature threshold of 56 °C. An automated control system regulates the heat recirculation pump based on the solar output and buffer tank temperature. Meanwhile, electricity generated by the PVTs is managed by a hybrid inverter that allocates power between the MiniStor system, building demands, and the electrical grid. Excess power can be stored in lithium-ion batteries, which offer 7.68 kWh of capacity with high efficiency. A more detailed description of the MiniStor operation including temperature and pressure ranges may be found in previous work [19,20].

2.2 Replication Potential

Section 2.2 outlines a detailed methodology for assessing the replication technical potential of the MiniStor system. This methodology includes the selection of EU regions with harsh climatic conditions and the evaluation of suitable building typologies. The system's performance under different environmental conditions is assessed using simulation results based on the MATLAB/ASPEN model described by Zisopoulos [11,19].

2.2.1 Site Selection

The first step involves identifying sites with climatic conditions that cover most of the EU. Given that extreme temperatures, precipitation, and wind events can significantly influence HVAC system performance, sites with particularly harsh climatic conditions were also included. In the context of MiniStor, harsh climates are defined as those with cold winters and moderate solar radiation potential.

The Köppen-Geiger climate classification system [21] was employed to categorize the selected sites. This system divides climates into five main groups—tropical (A), arid (B), temperate (C), cold (D), and polar (E)—with further subdivisions based on precipitation and temperature characteristics. For instance, a Cfc classification denotes a temperate climate with a cold summer, whereas Cfb indicates a temperate climate with a warm summer.

Locations including Larnaca, Athens, Rome, Porto, Paris, Krakow, Berlin, Hamburg, and Bergen were selected to ensure broad geographical distribution, distinct climate characteristics, and strong market potential based on residential building stock. According to the EU Building Stock Observatory (2024) [22], Germany, France, and Italy together represent over 50% of the EU’s residential floor area. Cyprus was also included to examine system performance under hot, semi-arid conditions. Collectively, the selected seven countries (Cyprus, France, Germany, Greece, Italy, Poland, Portugal) represent over 60% of the EU’s residential building stock (18,408 million m²). Norway was also incorporated to evaluate performance in extreme cold climates. The selected cities and their key climatic parameters are listed in Table 1. The necessary data for temperature and degree day calculations were sourced from the Meteonorm database [23].

Table 1 Selected sites for investigating the applicability of the Ministor system under different climatic conditions and characteristics of their climate.

Heating Degree Days (HDD) and Cooling Degree Days (CDD) are essential metrics for estimating heating and cooling demands. They are calculated based on the mean daily temperature relative to a baseline—15 °C for heating and 24 °C for cooling—according to the following formulas [24,25]:

where, Tⁱ_m = the mean daily temperature of day i. This is determined as the arithmetic mean value of the daily maximum (T_max) and the daily minimum temperature (T_min) T_m = (T_max + T_min)/2 [26].

2.2.2 Building Selection

To assess the replication potential of the MiniStor system, its applicability was evaluated across different sites and climatic conditions by selecting suitable building typologies for each of the cities and countries identified in the previous Section. The TABULA Web Tool [27,28] was utilized for this purpose. Developed under the projects “TABULA” (2009) and “EPISCOPE” (2013), the tool provides detailed national residential building typologies across Europe, enabling experts to access standardized data on construction methods, thermal transmittance values, common building systems, and energy consumption patterns.

For the replication analysis, a single-family house (SFH), either detached or semi-detached, was chosen as the representative building typology for each site. This selection was guided by the similarity to the pre-pilot building (the Smart Home of CERTH/ITI [18] in which Ministor has been installed and monitored, and the practical benefits of single-family homes, such as the availability of space for system components.

Using the TABULA database, a suitable recent construction typology was selected for each city. Key criteria included building size, construction materials, architectural style, and energy efficiency features, aiming to align closely with the characteristics of the Smart Home. For each typology, data on total (whole building) U-values, which incorporate both transmission and ventilation losses, were extracted, alongside reference floor area and period of construction. These are summarized in Table 2.

Table 2 TABULA category and period of construction for each selected site location along with basic energy consumption data.

It should be noted that no TABULA data exist for Portugal; therefore, equivalent building information was sourced from EPBD Portugal [29].

The heating and cooling loads for these buildings were estimated based on their U-values, reference areas, and the local climatic conditions determined previously. For time-dependent calculations, the following equations were applied:

where A_ref is the reference floor area, i= [1,t_end] is the time step, U_m is the average whole building U-value considering transmission and ventilation losses, T_i is the indoor design temperature, T_b is the external temperature at time step i.

For a simplified annual estimation, Heating Degree Days (HDD) and Cooling Degree Days (CDD) were used with the following formulas:

where Q_H and Q_C are the annual heating and cooling loads respectively. The resulting annual heating and cooling loads for each building typology are presented in Table 2.

To evaluate the broader technical replication potential, the share of single-family dwellings in each country was considered. According to the European Building Stock Analysis by Eurac Research [30], the share of single-family and terraced houses varies between 29% and 91% across the studied countries: 29% in Italy, 48% in France, 68% in Greece, 71% in Germany, 76% in Poland, 86% in Portugal, and 91% in Cyprus. Further focusing on recently built dwellings (post-2010) to align with the TABULA typologies, the share of newer constructions ranged from 4% to 12%, 4% in Germany, 5% in Greece, 7% in Portugal, 10% in Cyprus, and 12% in Italy, France, and Poland. Based on these data, it is estimated that the assessed typologies collectively represent approximately 504 million m² of floor area where the MiniStor system could potentially be deployed.

2.2.3 Simulation analysis

This section outlines the methodology applied to assess the technical potential for replicating the MiniStor system. The methodology is structured in five key steps:

(1) Site Selection:

The selected locations for the applicability study of the MiniStor system are presented in Table 1.

(2) Building Typology Identification:

Representative building typologies were identified for each location, as described in Section 2.2.2. Based on these typologies, heating and cooling loads were calculated using Equations 5 and 6.

(3) Definition of a Representative Time Period:

Since the MiniStor system simulations are limited to representative days due to computational constraints, a typical period covering a full cycle of charging and discharging (typically two to three days) was determined for both winter and summer conditions.

(4) Input Data Definition:

Additional inputs related to the solar field configuration and system operation such as collector layouts, mass flow rates between system components, and operating pressures were defined based on prior system specifications. Table 3 summarizes the main input and output variables considered in the current simulation analysis, along with their respective units.

Table 3 Main input and output variables considered in the simulation analysis.

(5) Simulation and Optimization:

At first, the analysis assumes a conservative worst-case scenario where both the TCM and PCM storage tanks are initially empty. However, in practical applications pre-charged storage would enhance system performance. Therefore, an optimized discharging strategy where the discharge initiates first is also examined to demonstrate MiniStor's full capability in meeting heating and cooling demands. In this scenario, the TCM reactor is discharged at night when residential heating demand is highest, the PCM tank is discharged to meet early morning heating loads and the TCM reactor is recharged using solar heat, while a heat pump covers heating demand and simultaneously charges the PCM tank. The proposed optimization is based on optimizing the control and operation of the existing system without any changes in its hardware and on typical strategies for thermal storage systems as seen in the literature [31–33].

The entire procedure, including the MiniStor model, has been incorporated into a custom MATLAB tool, facilitating the replication of the MiniStor system for any location with available meteorological and building typology data. Since Steps 1 and 2 are detailed in Sections 2.2.1 and 2.2.2, the remainder of this section focuses on the procedure for defining the typical time periods.

Following the approach described in [11,19], representative periods for winter and summer were identified based on outdoor temperatures, Heating Degree Days (HDD), Cooling Degree Days (CDD), and available solar radiation:

(1) For each day of the Typical Meteorological Year (TMY), daily HDD, CDD, and solar radiation on a tilted surface were calculated. TMY data was sourced from the Meteonorm database, known for its high reliability based on station-derived measurements.

(2) A variable was calculated for each day, representing deviations in HDD, CDD, and solar radiation from their respective seasonal averages.

(3) The variable values were summed over three-day periods, corresponding to a full MiniStor charge/discharge cycle.

(4) The three-day period with the lowest variable value was selected as the representative period for the simulation.

This procedure identified typical heating and cooling periods for all cities. Generally, the maximum length for a representative cycle was three days. The full methodology, including the steps for finding the representative periods, is depicted in Figure 2. The heating and cooling durations, for each city, are presented in Table 4. The heating and cooling profiles were further refined by considering solar radiation and ambient temperature values, interpolated from hourly TMY data to 15-minute intervals to match the required time step of the MiniStor model.

Figure 2 Methodology flowchart for the MiniStor system replication analysis.

Table 4 Typical heating and cooling period for each city.

For Step 4, system parameters—such as the thermal and electrical capacities of the solar field—were adopted from the Thessaloniki pre-demonstration site and may be found in [19,20]. In addition, the system model incorporates several essential parameters: the PI controller settings (Table 6 in [11]), including temperature and pressure setpoints; the initial input values and operating conditions of the thermochemical material (TCM) reactors; and the phase change material (PCM) storage tank characteristics, such as phase change temperature and energy density (Table 7 in [11]). The complete list of system parameters is not included here for brevity; however, all parameters are available in [19]. The default solar field layout consists of three rows: two rows with five solar photovoltaic-thermal (PVT) hybrid panels each, and one row with five solar flat plate collectors (FPCs). The system's total installed thermal capacity is 16 kW_th, with a peak electrical output of 2.65 kW_e.

The main difference in this analysis compared to the MiniStor model described in [19] is the approach used to estimate photovoltaic (PV) energy generation. Instead of relying on the MiniStor model, annual solar electricity generation was calculated using the Photovoltaic Geographical Information System (PVGIS) [34]. Developed by the Joint Research Centre (JRC) of the European Commission, PVGIS is a widely recognized tool for estimating solar radiation and PV performance across Europe, offering detailed data on solar irradiance, temperature, and meteorological conditions. PVGIS has been extensively used by researchers to assess PV performance across diverse climates and by policymakers for strategic renewable energy planning [34,35]. In this study, PVGIS data and MiniStor system parameters were combined under the following assumptions:

(1) Optimal panel tilt for each location,

(2) 13% system energy losses,

(3) Data sourced from the PVGIS-SARAH2 database, and

(4) PVT panel specifications.

PVGIS was preferred over the MiniStor model due to its ability to provide full-year energy estimates, whereas the MiniStor simulations are limited to short 2 to 3-day cycles. For evaluating year-round feasibility, PVGIS offers significantly higher accuracy. Finally, the MiniStor model simulations were performed for all cities using their corresponding heating and cooling periods and covering the worst-case as well as the optimized scenario. The results are presented in the next Section.

3.1 Analysis during the heating and cooling period

This section presents the MiniStor system results for the cities analyzed in Section 2. The system configuration follows [19], with the solar field setup modeled after the Thessaloniki pre-demo site. Results are reported for Athens, Bergen, Berlin, Hamburg, Krakow, Larnaca, Paris, Porto and Rome. The heating and the cooling season correspond to the average winter and summer period respectively, defined in Table 4.

Figure 3 depicts the heating operation performance with and without auxiliary electricity, including the heat load coverage for a full cycle. In Bergen, Berlin, Krakow, Hamburg, and Paris, insufficient solar radiation prevents full charging, resulting in minimal heat load coverage by solar energy and the need for auxiliary grid electricity. Conversely, in Athens, Larnaca, Porto, and Rome, solar radiation is sufficient for completing the cycle, with Athens achieving the highest solar heat coverage, followed by Larnaca and Porto. Rome shows lower coverage due to longer cycles and higher heating demands. Limited auxiliary electricity could further enhance system coverage in these cases.

Figure 3 Heat load coverage for buildings in selected cities for replication with and without using auxiliary electricity.

Charging depends on solar availability and auxiliary electricity, while discharging depends mainly on building load profiles. In colder climates (e.g., Berlin, Krakow, Paris), solar heat coverage ranges from 0.07 to 0.19, with the system having a higher reliance on auxiliary electricity.

Table 5 presents results for the cooling operation. It includes total cooling demands per cycle, cooling load coverage covered by solar energy, and cycle completion time. Charging during the cooling season relies exclusively on solar energy. Results show that higher solar radiation accelerates charging and shortens cycle durations. Cooling coverage also varies, with Rome achieving the highest (0.78) and Larnace the lowest (0.46). In Rome, higher coverage is due to a favorable balance of strong solar radiation and moderate cooling loads, while Larnaca's cooling needs are much higher.

Table 5 Cooling load coverage for buildings in selected cities for replication.

3.2 Annual performance analysis

To evaluate MiniStor's ability to meet annual heating and cooling demands using solar energy, the annual heating and cooling data from Table 4 and the simulation results from the previous Section are utilized. First, the hourly coverage rate of MiniStor during a typical full charging/discharging cycle was calculated and then applied to the annual demand to estimate the total annual covered heating and cooling loads. Annual electricity production values were obtained from PVGIS 5.2.

Table 6 presents the annual energy performance under the typical (or the worst-case) scenario. In Athens, Larnaca, Rome, and Porto, MiniStor’s solar field can charge the TCM reactor without auxiliary electricity. The RES coverage ranges from 53% to 88%. Electricity production during the year is similar across Athens, Rome, and Porto, reflecting similar solar field configurations, solar radiation, and heating/cooling periods.

Table 6 Annual heating and cooling demand coverage by RES - typical scenario.

Results for Paris, Krakow, Berlin, Hamburg, and Bergen are also included. In these cases, TCM reactor charging relies on both solar energy and auxiliary electricity. Lower solar contributions are noted in harsher climates, with Paris achieving 60% RES coverage, Berlin and Hamburg achieving 39% and 36% RES coverage respectively.

Larnaca records the highest annual production (4,680 kWh/year), while Bergen showed minimal output. Notably, in cities without MiniStor cooling operation (Bergen, Berlin, Hamburg, Krakow, Porto,), surplus electricity during summer is injected into the grid. Conversely, during heating periods, on-site electricity production is generally insufficient to cover MiniStor’s electricity needs, requiring grid supplementation.

Krakow shows the highest heating demand (25,575 kWh/year), followed by Hamburg, while Porto exhibits the lowest (7,946 kWh/year). Cooling demands are highest in Larnaca, while minimal in northern cities like Berlin and Bergen. In some locations, the integration of solar and electric heating significantly boosts heating coverage (e.g., Paris from 2,663 to 5,186 kWh/year), whereas cities like Bergen depend mainly on auxiliary electricity.

As MiniStor primarily targets heating applications, cooling coverage is considered supplementary, enhancing the system’s economic viability. Since cooling loads are minimal or nonexistent in many cities (Table 4), results are presented only for Athens, Larnaca, and Rome. Cooling operation relies entirely on solar heat, with no backup electricity heater needed. MiniStor covers between 33% and 100% of annual cooling needs in these southern cities.

`The results discussed thus far are based on the worst-case scenario, demonstrating that MiniStor performs satisfactorily even under the most adverse conditions. However, in real-world applications, systems typically start with fully charged TCM and PCM storage tanks. Therefore, an optimal operating strategy, where discharging precedes recharging, was also evaluated to better reflect practical performance.

In the optimal scenario, the TCM reactor discharges at night to meet peak heating demands, while the PCM tank addresses early morning loads. Solar heat then recharges the TCM reactor during the day, with the heat pump simultaneously supporting heating demand and charging the PCM tank. This strategy maximizes system efficiency by synchronizing energy storage and demand patterns.

Due to the complexity of the optimization, representative cities—Athens, Bergen, Hamburg, Larnaca, and Rome—were selected for simulation, considering both cases with and without auxiliary electricity. Table 7 compares typical and optimal scenarios across the five cities, focusing on heating demand covered by solar energy, total heating and cooling demand covered, and the overall RES energy coverage ratio. In Athens, optimization increased heating coverage by solar energy from 4,456 to 5,199 kWh/year, and nearly doubled the covered cooling demand, raising the total RES coverage ratio from 53% to 65%. Larnaca exhibited the most significant gains, with heating coverage nearly doubling and cooling coverage rising substantially, resulting in an RES coverage ratio exceeding 100%, indicating periods of energy surplus.

Table 7 Annual heating and cooling demand coverage by RES: typical vs. optimal scenarios.

Bergen and Hamburg showed limited improvement, with RES coverage increasing marginally due to less favorable climatic conditions. In Rome, optimization improved heating coverage from 3,931 to 6,100 kWh/year and cooling coverage from 1,286 to 1,649 kWh/year, increasing the overall RES ratio from 72% to 91%.

In summary, under the optimal operation, heat load coverage improved by 3% to 40%, with Larnaca showing the highest increase. The cooling performance under optimal operation was also assessed for Athens and Larnaca. Discharging began at dawn, aligning with rising cooling demands. In Athens, cooling needs were fully met (100%) through optimized use of TCM and PCM, while Larnaca achieved 87% coverage. Overall, the optimized strategy allowed discharging/charging cycles to complete within a single day, significantly enhancing performance in warmer climates.

Ethics Statement

Not applicable.

Consent for Publication

Not applicable.

Availability of Data and Material

Dataset are available on request from the authors.

Funding

This paper was prepared in the context of the European Union's Horizon 2020 research and innovation program through grant agreement No 869821 (Minimal Size Thermal and Electrical Energy Storage System for In-Situ Residential Installation - MiniStor).

Competing Interests

The authors have declared that no competing interests exist.

Author Contributions

Conceptualization: G.M. and N.N.; Methodology: A.T. and G.M.; Software: A.T.; Validation: A.T. and G.M.; Formal Analysis: A.T. and G.M.; Investigation: A.T. and G.M.; Resources: A.T. and G.M.; Data Curation: A.T.; Writing – Original Draft: A.T. and G.M.; Writing – Review & Editing: G.M. and N.N.; Visualization: A.T.; Supervision: G.M. and N.N.; Project Administration: G.M. and N.N.; Funding Acquisition: N.N.

Acknowledgement

Not applicable.