A Dual PCM Thermal Battery Delivering More than 80 kW of Discharge Power Despite the Low Thermal Conductivity of Organic PCM

Abstract

The increasing deployment of decentralized renewable energy systems requires sustainable, affordable, and high-energy-density storage technologies. While thermal energy storage using phase change materials (PCMs) represents a promising solution, its widespread adoption remains limited by the low thermal conductivity of most PCMs, which restricts (dis)charge power. This work demonstrates that high discharge powers can be achieved through optimized heat-exchanger design without enhancing the intrinsic thermal conductivity of the PCM. After sizing simulations for individual homes, a thermal storage demonstrator integrating water heat exchangers immersed in two PCM reservoirs for domestic hot water and space heating was designed and experimentally investigated. The prototype achieved discharge powers exceeding 80 kW, sustaining 65 kW once transit water had fully cleared the tank, in a compact PCM heat exchanger volume of 0.79 m³, despite the low PCM thermal conductivity of only 0.16 W·m⁻¹K⁻¹. This performance is enabled by a heat-exchanger design providing 203 m² of heat-transfer area while limiting the maximum PCM-to-fin distance to about 1.5 mm. Thermosiphon-driven natural convection was also observed during melting and may further enhance heat transfer. These results demonstrate that heat-exchanger geometry, rather than enhanced PCM conductivity, is the key design parameter for achieving compact, high-power PCM thermal energy storage.

1. Introduction

Since 2005, massive investments have been made in solar energy production to address climate change; however, solar generation is temporally mismatched with the thermal energy demand of buildings, which accounts for 80% of their energy consumption [1]. A study by Li et al. (2022) [2,3] evaluated how combining photovoltaic (PV) technology with thermal energy storage (TES) could alleviate this mismatch in a typical single-family house with a heat pump (HP). Figure 1a shows that PV sized to match annual building consumption in Frankfurt (Germany) reduces grid energy use by about 20%. This reduction increases to 56% with the additional integration of 2 m³ TES battery.

To reduce the volume of thermal energy storage (TES) systems, phase change materials (PCMs) can be used owing to their high energy storage density for space heating [5,6,7,8], domestic hot water (DHW) [9,10,11,12], or combined applications [13,14]. In this work, two PCM containers are integrated into the insulated storage concept shown in Figure 1b to simultaneously supply space heating and DHW. The stored thermal energy enables night-time and winter operation using solar energy collected during periods of photovoltaic (PV) production. By operating the air-to-water heat pump during daytime periods of PV generation, when outdoor temperatures generally allow a higher coefficient of performance (COP), the proposed TES system increases solar energy utilization and PV self-consumption, as also reported by Li et al. [2,3].

Nair et al.’s review [13] on PCM in-building applications highlights their potential for space heating and DHW, while noting that most research remains focused on materials or component-level performance; together with other reviews of PCM systems [6], they identify low thermal conductivity, particularly for organic PCM, as one of the most frequently cited drawbacks, limiting heat transfer rate and (dis)charge thermal power. This is reflected in recent experimental studies in the 20–70 °C range relevant to space heating and DHW: most lab-scale prototypes report discharge powers of only 1–5 kW [15,16,17,18], and even the highest reported field-validated result—a 180 kWh tube-and-shell PCM store operated at a district heating substation—reached 40 kW discharge power, by design, to cover morning consumption peaks [19].

This power ceiling is a likely cause of the limited uptake of PCM for building-scale active storage, where solar overproduction coupled with heat pumps can generate several kW to several tens of kW for single- and multi-family homes, respectively [2,3], and instantaneous DHW demand can substantially exceed the steady-state power that a heat pump can continuously deliver. Under EN 12831-3, the European standard for DHW design heat load [20], Switzerland’s national implementation (SIA 385/2) recommends sizing the additional heat-generator power for water heating at only 2 W/m² (single-family) to 3 W/m² (multi-family) of reference energy area [21], while even a single draw-off (e.g., a shower at ~10 L/min, ΔT ≈ 35 K) already requires ~24 kW. PCM systems must therefore provide sufficient heat transfer capacity between the (dis)charge circuit and the PCM to absorb and deliver that power on demand. In this work, we address this heat-transfer bottleneck by integrating tube-and-fin heat exchangers, with round tubes and continuous plate fins, into each of the two PCM containers of the dual-PCM storage system introduced above (Figure 1b), and experimentally demonstrate a discharge power of 80 kW—twice the highest previously reported value for a PCM system in this temperature range—validated at both the component and building scale. The results show that this compact PCM system (<1 m³) delivers high thermal power and can significantly reduce grid energy demand, facilitating the cost-effective and incremental decarbonization of residential buildings.

2. Materials and Methods

2.1. Simulations for Sizing the PCM Battery Prototype

Prior to designing and building the residential PCM battery prototype, the energy storage system must be optimally sized. To accelerate this process, a simplified MATLAB (R2025b) simulation tool was developed in-house to determine the thermal battery size that maximizes the ratio of stored and reused solar energy to storage volume, used here as a first-order proxy for cost-effectiveness, as overall system cost scales with volume. In other words, it finds the most cost-effective tank size—the one that delivers the most energy savings per cubic meter of storage.

A cylindrical thermal battery was chosen to exploit axial symmetry, reducing the problem to a 2-D (radius–height) finite element (FE) formulation of the energy equation, discretized over annular cells of the PCM, the stainless steel container, and the insulation. The code performs explicit temporal integration, updating temperature-dependent material properties at each time step: specific/latent heat, thermal conductivity, density, and viscosity for the PCM, and specific heat, thermal conductivity, and density for the steel and insulation. Only conduction between adjacent cells is simulated, neglecting PCM convection; the boundary condition is applied at the outer surface of the insulation, where convective heat exchange occurs with the external air of the technical room at 15 °C.

The new MATLAB tool was validated against Ansys Fluent (2025 R2), used here as the reference Computational Fluid Dynamics (CFD) software, applying the heat transfer and melting-and-solidification models on a refined 3-D CFD mesh, with material properties identical to those used in the MATLAB code, including the specific and latent heat curves of CrodaTherm 37 and 53. A reference dual-PCM tank (0.25 m diameter, 1 m height, 15–80 °C) was simulated under charge/discharge loads of 2, 5, 7, and 10 kW, comparing the average tank temperature and the local temperature of individual FE cells predicted by each tool. Since MATLAB cells were ten times larger than Ansys cells in each direction, the corresponding Ansys cells were mass-averaged for direct comparison at the MATLAB cell scale.

The average tank temperature predicted by MATLAB closely matched the Ansys reference, with a marginally faster rise (~5%), a precision considered sufficient for the sizing objective. Larger local discrepancies, up to 8 °C, occurred during phase change: the large MATLAB cells (5 cm × 5 cm) contain substantial PCM mass, so a complete phase change within a cell takes considerable time, producing an extended temperature plateau. After mass-averaging over the corresponding MATLAB cell volume, the much smaller Ansys cells (0.5 cm × 0.5 cm × 0.5 cm) transition rapidly to a fully solid or fully liquid state, so no comparable plateau appears. These local discrepancies did not affect overall thermal performance: total stored energy differed by less than 6% between MATLAB and Ansys over a complete charge or discharge cycle, confirming the validity of the MATLAB tool for the intended sizing optimization. This is particularly remarkable given that the MATLAB-based simulations run roughly 20,000 times faster than the CFD simulations.

The benefits of integrating a thermal battery were evaluated for a reference two-apartment, six-occupant building in Fribourg, Switzerland, complying with the Swiss Minergie-P standard, over the 2022–2023 winter heating season (October–February). The building is equipped with 90 m² of south-facing PV panels (17.4 kWp). Solar production was obtained from the EU Photovoltaic Geographical Information System (PVGIS) database, and hourly external temperature data were obtained from the Open-Meteo API.

Building thermal demand was calculated using LESOSAI (2025) [22], the software widely used for building certification in Switzerland and Luxembourg. Based on the building’s envelope properties, LESOSAI computes the corresponding average U-value—0.15 W·m⁻²·K⁻¹ for the building considered in this study—and dynamically determines the hourly heating demand required to maintain the indoor setpoint temperature (20 °C in this case), accounting for outdoor temperature and solar radiation. Adding the DHW consumption of six occupants yields the total hourly thermal demand. The solar heating capacity was obtained by multiplying the PV production by a fixed HP COP of 3.

Three operating rules govern battery charge and discharge: (1) charge using excess solar power and the HP; (2) discharge when solar production is insufficient, at a rate equal to the thermal demand minus the available solar power; (3) activate the HP when both the battery is depleted and solar production is insufficient. These rules were implemented in the MATLAB code described above to compute hourly battery charge/discharge from October 2022 to February 2023, allowing the total spared energy—the solar energy stored and subsequently reused—to be quantified.

Six dual-PCM configurations of different sizes were simulated, each with CrodaTherm 53 filling the upper half of the container and CrodaTherm 37 the lower half, and 12.5 cm of insulation. Container dimensions (radius, height) were (0.23 m, 0.74 m), (0.31 m, 1.0 m), (0.31 m, 2.0 m), (0.44 m, 2.0 m), (0.67 m, 2.0 m), and (1.02 m, 2.0 m), corresponding to total volumes with the insulation of 0.47, 0.86, 1.52, 2.49, 4.86, and 9.85 m³, respectively. The grid energy spared per unit volume is illustrated in Figure 2 for the 6 PCM containers.

Annual space heating and DHW demand are 7570 kWh and 1950 kWh, respectively (9520 kWh in total). Direct use of solar energy already reduces the grid energy requirement to 5150 kWh. The simulation quantifies the additional reduction in grid demand achievable as a function of battery storage capacity and used volume, revealing little additional spared energy at larger volumes (Figure 2): a seasonal battery of 542 kWh (10 m³) spares about 4150 kWh—80% of the remaining grid demand—but is not cost-effective given its volume. By comparison, a 25 kWh battery (0.86 m³) saves 1720 kWh, corresponding to 2000 kWh/m³—4.75 times higher than the seasonal battery on a per-volume basis (Figure 2). This demonstrates that thermal batteries below 1 m³, charged with PV solar energy, can substantially reduce grid energy demand for the thermal needs of Minergie-P buildings with one or two apartments. In this example, combining PV panels with a 25 kWh battery reduces grid energy consumption by two-thirds, from 9520 kWh to 3430 kWh. Based on the simulation results shown in Figure 2, a prototype with capacity between 25 and 50 kWh was selected for construction because it maximizes the energy spared from the grid per unit volume (Figure 2).

2.2. Prototype Design and Experimental Setup

Based on the simulation results, we designed the PCM battery shown in Figure 1b that fits in a 1 m³ IBC retention tank (type 1000 K 150.50-UN). A lower tank containing CrodaTherm 37, a PCM with a melting temperature (T_m) of 37 °C, for space heating was separated by a 5 cm insulating layer from the upper tank containing CrodaTherm 53, a PCM with T_m = 53 °C for domestic hot water production. The external 30 cm-wide insulation was made of the same material (PBF 030 Saint-Gobain) with a thermal conductivity l of 0.03 Wm⁻¹K⁻¹. Both ~50 cm high tanks were equipped with four aluminum heat exchangers (HX) featuring inlet and outlet manifolds immersed in the PCM baths, which had melting temperatures T_m of 37 and 53 °C, respectively. The distance between the HX boundary and its container was defined as 1 cm, using the PCM’s low thermal conductivity as a first layer of insulation. Conversely, to enhance thermal transfer within each HX, metallic fins immersed in the PCM bath were separated by only 3 mm.

The control and measurement setup is displayed in Figure 3. A water heater/chiller, shown in Figure 3a, could heat or cool the thermal battery at power levels of 2 kW and 0.5 kW, respectively. For other measurements, it could be replaced by a plate heat exchanger (PHX) or, in the case of large power discharges by directly introducing the domestic cold water from the building’s network via valve V4 and evacuating it through V3. Four three-way valves could direct the water flow (coming from V4) towards each PCM tank in an upstream or downstream direction. The flow could cross a single tank or both in a serial or in a parallel configuration. EV020R2 ultrasonic flow meters (Belimo, Hinwil, Switzerland), combined with two UF04 400 50P Pt1000 sensors (sensortec, Morat, Switzerland) of the same color, allowed the measurement of the (dis)charging power dissipated in each tank for upstream or downstream flows; integrating these values yielded the system’s total storage capacity. The calibrated Pt1000 sensors had an accuracy of ±0.1 °C. Based on the manufacturer’s datasheet information, the average differential temperature accuracy could be extrapolated as follows: ± (0.12 °C + 0.01 · ΔT). The EV020R2 ultrasonic flow meters had an accuracy of 2% from 5 L/h to the nominal flow of 2500 L/h, according to EN 1434 class 2 [23]. As power is proportional to the product of ΔT and flow, the worst-case relative power accuracy was obtained by adding both relative errors. The same approach, with time uncertainty considered negligible, was used to propagate the accuracy of the reported energy values. Accuracy is reported in the publication for all measured flow, power, and energy values.

Power was measured continuously throughout the entire (dis)charge process. One particularly important value is the power recorded at the breakthrough time (BT)—the time required for water to travel from the tank’s inlet to its outlet. Before BT, the water leaving the tank was still residual water that was already present inside it, so outlet temperature readings did not yet reflect heat exchange with the PCM; after BT, they did. The power measured at BT therefore represented the earliest value that could be attributed to genuine heat exchange with the PCM.

Figure 3b shows a top view of the upper HX of the lower tank. Five vertical orifices (yellow circles) and 20 Pt1000 sensors were used to measure the PCM temperature of the four HXs, with one sensor per HX at each orifice location. These locations were either at the center of the HX (orifice 3), close to the inlet manifold (orifices 2 and 5), or near the outlet manifold (orifices 1 and 4).

Each tank comprised four stacked aluminum heat exchangers, each containing 47 rectangular water tubes (95 mm high, 8.2 mm thick), spaced 8.2 mm apart, as shown in Figure 3b. Heat transfer was performed through 0.4 mm-thick corrugated fins brazed between the water tubes; the resulting 3 mm gap between fins was filled with PCM, giving a maximum PCM-to-fin distance of 1.5 mm, since the PCM was bounded by fins on both sides. The heat exchanger followed a tube-and-fin design, characterized by tube spacing and corrugated-fin geometry as the relevant geometric parameters. The total heat transfer area—including all fin and tube surfaces in contact with the PCM—was 203 m², split between 115 m² for the CrodaTherm 53 tank and 88 m² for the CrodaTherm 37 tank. With 188 water tubes and two manifolds per tank, the thermal battery also held a substantial volume of water: 140 L in the CrodaTherm 53 tank and 120 L in the CrodaTherm 37 tank. The global system is illustrated in Figure 4 with the PCM battery system in its insulation, the heater/chiller, the PHX, the controller, and the panel with valves and flow meters.

3. Results

3.1. Thermal Losses, Temperature Profile Evolution, and Cycle-to-Cycle Repeatability

Heat losses were measured under steady-state conditions by maintaining the entire thermal battery at 60 °C, corresponding to 40 °C above ambient. Heating was provided by circulating hot water through both the upper and lower tanks. The measured heat loss was 170 W, corresponding to an overall heat loss coefficient of 4.25 W/K.

The cycle stability of the PCM battery has been tested through 100 charge−discharge cycles performed on the CrodaTherm 37 tank, revealing variations of 1% in power and 2.2% in energy [4]. Figure 5 describes the water flow through the thermal battery, which can operate upstream (from B to A) or downstream (from A to B) and compares the evolution of the PCM temperature in the lower tank heat exchangers for upstream and downstream heating. We observed for upstream heating that the PCM at the lower heat exchanger (blue curve) warms up one hour later than the PCM of the upper heat exchanger (red curve). This delay does not occur for downstream heating, which shows a more homogeneous temperature distribution. Consequently, downstream heating and upstream cooling were selected as the preferred methods for battery heat or cold charging, respectively.

The same measurements were performed in the upper tank filled with CrodaTherm 53. The PCM temperature evolution with preferential heating and cooling configuration is displayed in Figure 6. The phase transition is uniform in all heat exchangers, with the time lag for complete melting being less than one hour and the time lag for complete solidification less than two hours. Note that the temperature plateaus during the phase transitions at the heat exchanger outlet (B for melting and A for solidification) are observed at 51.5 °C rather than 53 °C. The small temperature difference of 0.5 °C at the inlet of the heat exchangers (A) observed during melting at time t = 19 h could be explained by slightly more efficient heat transfer in the upper heat exchanger (in red) during downstream heating.

3.2. Thermal Storage Capacity and High-Power Discharge Performance of the Dual-PCM Battery

The charge/discharge cycles presented previously were performed at thermal powers of 2 kW and 0.5 kW, corresponding to the nominal powers of the heater and chiller, respectively. To evaluate the maximum discharge performance of the thermal battery, a different experimental setup was used. Domestic cold water from the building’s network at temperatures between 9 and 18 °C was circulated through the PCM battery after the latter was heated to either 50 or 60 °C. The large temperature difference (>40 K) and high water flow rates (>600 L/h) minimized the thermal resistance on the water side, allowing the intrinsic discharge power of the PCM battery to be evaluated. It should be noted that using the building’s network has the drawback of not allowing full control over the flow rate; while it was kept constant within each measurement, slightly different flow levels were observed between measurements. Precise flow and temperature measurements, however, allowed the thermal power to be determined accurately.

The individual tanks were first tested separately. The lower tank, filled with PCM (T_m = 37 °C), was heated to 50 °C and discharged with water at 9 °C flowing at 780 ± 16 L/h, yielding a maximum discharge power (P_max) of 29.8 ± 1 kW. Although the thermal storage system delivered its maximum power at the beginning of the discharge, the cooling water did not yet fully pass through the heat exchangers. It took 9 min for the water to travel through the CrodaTherm 37 tank and reached breakthrough time (BT). The corresponding power at breakthrough (P_BT) was found to be 28.6 ± 0.9 kW. The upper tank, filled with PCM CrodaTherm 53, was heated to 60 °C and discharged with water at 18 °C at a flow rate of 725 ± 14 L/h, resulting in P_max and P_BT values of 32.2 ± 1.1 and 29.2 ± 1.0 kW, respectively.

For the tests involving both tanks, two discharge configurations were investigated: (1) a serial configuration, with cold water flowing successively through the CrodaTherm 37 and CrodaTherm 53 tanks, and (2) a parallel configuration, with the flow divided between the two tanks. The results are presented in Figure 7. In the serial configuration, with a water flow rate of 730 ± 15 L/h, the CrodaTherm 37 tank, the CrodaTherm 53 tank, and the complete thermal battery achieved maximum discharge powers (P_max,ser) of 35.6 ± 1.2, 16.6 ± 0.05, and 35.7 ± 1.2 kW, respectively (Figure 7a). Their breakthrough times were 10, 21, and 21 min, corresponding to discharge powers at breakthrough (P_ser,BT) of 17.3 ± 0.6, 14.0 ± 0.5, and 29.0 ± 1.0 kW, respectively. Integration of the discharge power curves yielded thermal storage capacities (Eth) of 41.5 ± 1.4 kWh for the complete thermal battery, 18.6 ± 0.6 kWh for the CrodaTherm 37 tank, and 22.9 ± 0.8 kWh for the CrodaTherm 53 tank.

In contrast to the 730 ± 15 L/h used in the serial configuration, the parallel configuration showed a higher total cold-water flow rate of 1500 ± 30 L/h, with 840 ± 17 L/h directed to the upper tank (CrodaTherm 53) and 660 L/h to the lower tank (CrodaTherm 37)—likely attributable to the doubled flow section and the resulting reduction in pressure drop. Under this configuration, the PCM battery system achieved a maximum discharge power P_max,par of 80.1 ± 2.6 kW (Figure 7b). At the breakthrough time of 10 min, the discharge power P_par,BT reached 65.0 ± 2.1 kW. The discharge power distribution between the CrodaTherm 37 and CrodaTherm 53 PCM modules closely follows their mass distribution, with mass fractions of 43% and 57%, respectively, which also correspond to the distribution of HX tubes in the two tanks. The maximum discharge powers P_max,par reached 36.1 ± 1.2 kW for CrodaTherm 37 and 44.6 ± 1.5 kW for Crodatherm 53. Similarly, discharge powers at breakthrough P_BT,par of 26.2 ± 0.9 and 38.8 ± 1.3 kW were obtained for CrodaTherm 37 and 53, respectively.

Integration of the discharge power curves resulted in thermal storage capacities Eth of 41.7 ± 1.4 kWh for the entire battery, 19 ± 0.6 kWh for the CrodaTherm 37 tank, and 22.7 ± 0.7 kWh for the CrodaTherm 53 tank, which are consistent with the capacities calculated from the serial configuration.

The serial discharge test was repeated using the plate heat exchanger (PHX) shown in Figure 3a and Figure 4. A domestic cold-water flow of 660 ± 13 L/h was circulated through the secondary circuit of the PHX, extracting heat from the primary circuit, which consisted of a 730 ± 15 L/h water flow circulating through the PCM thermal battery. The maximum discharge power P_max,PHX,ser calculated from the temperature difference between the inlet and outlet of the PHX secondary circuit and the secondary circuit water flow rate was 28.8 ± 1.0 kW (Figure 8), while the discharge power at breakthrough, P_PHX,ser,BT, was 24.5 ± 0.8 kW.

Table 1. Main characteristics and performance of the double-PCM thermal battery. For clarity, accuracy values are omitted. The internal dimensions of the PCM tanks define the volume for PCM and HXs. The entire battery corresponds to the total system illustrated in Figure 1 without external insulation.

Parameter & Configuration			Container
			Entire Battery	CrodaTherm 37	CrodaTherm 53
Container role			Retention tank with a drain valve	PCM tank	PCM tank
Contained elements			PCM tanks + internal insulation	4 HX in a PCM bath	4 HX in a PCM bath
Tank dimensions	Length × Width × Height		1.2 m × 1 m × 1.16 m	0.8 m × 0.8 m × 0.54 m	1 m × 0.8 m × 0.54 m
	Volume		1.4 m3	0.35 m3	0.44 m3
PCM weight			290 kg	120 kg	170 kg
Energy storage capacity (15–60 °C)			41.6 kWh	18.8 kWh	22.8 kWh
Single tank		Pmax	------	29.8 kW	32.2 kW
		PBT	------	28.6 kW	29.2 kW
Storage change after 100 cycles * Power change after 100 cycles *			------ ------	−1% +2.2%	----- -----
Serial configuration		Pmax,ser	35.7 kW	35.6 kW	16.6 kW
		Pser,BT	29.0 kW	17.3 kW	14.0 kW
Serial configuration via PHX		Pmax,PHX,ser	28.8 kW	-----	-----
		PPHX,ser,BT	24.5 kW	-----	-----
Parallel configuration		Pmax,par	80.1 kW	36.1 kW	44.6 kW
		Ppar,BT	65.0 kW	26.2 kW	38.8 kW

* Cycling stability tested only for the CrodaTherm 37 tank; see Discussion for justification: the most important results of the entire battery are shown in bold.

summarizes the main characteristics of the multi-PCM storage system, including its geometry, PCM inventory, thermal storage capacity, maximum discharge power ($P_{max}$), and power at breakthrough ($P_{BT}$) for the different hydraulic configurations.

4. Discussion

One of the principal motivations for using PCM rather than conventional sensible heat storage is the higher energy storage density achievable within the same volume [5]. A direct comparison between the built PCM prototype and a water-based storage of the same dimensions was performed by calculating the storage capacity of the battery when the PCM and heat exchangers are removed and both tanks are filled with water: For the temperature difference relevant to customer operation (ΔT = 15 °C, with the upper and lower tanks at 55 °C and 40 °C, respectively), the PCM battery reaches 28.2 kWh compared with 13.8 kWh for an equivalent water storage, corresponding to a 2.2× higher storage density. This value remains below the theoretical ratio of 3.5 derived from the PCM latent heat and the specific heat of water over the same volume and ΔT. This gap mainly comes from the PCM occupying only half of the tank volume, with the rest filled by the heat exchangers, their water tubes, and the oversized 7 cm-wide water manifolds—elements a commercial design would narrow to increase the PCM volume fraction.

A cost-benefit analysis further supports this positioning. Hot water tanks remain the most economical thermal storage option: for the same storage capacity as the prototype at ΔT = 15 °C, a conventional combined storage tank (CST)—corresponding to 45–60 °C for DHW and 30–45 °C for building heating—costs approximately 4000 CHF, compared to a planned cost of 12,000 CHF for the proposed PCM battery, including materials, PCM, fabrication, and installation. PCM storage therefore cannot compete with hot water storage where sufficient space is available. However, a market segment opens where hot water storage cannot be accommodated—frequently the case in renovation projects—particularly when combined with local thermal or PV solar production. This aligns with the positioning of PCM-battery manufacturers such as COWA, who target the same niche market: Cowa Thermal Solutions AG, “What is PCM? FAQ,” [Online]. Available: https://www.cowa-ts.com/en/was-ist-pcm-faq (accessed on 25 August 2026).

To address this market, a segmentation between single-family (1F) and three-family (3F) buildings was carried out. The current prototype-based solution, with a nominal capacity of 30 kWh (28.2 kWh measured at ΔT = 15 °C), is suited to 3F buildings with integrated solar power requiring high power, with a payback period of 8 years. For 1F buildings and 3F buildings with lower power requirements, a modular solution using extruded aluminum heat-exchanger elements integrated into smaller containers was developed, with planned costs of 5500, 6750, and 8700 CHF for capacities of 27, 39, and 50 kWh, respectively. A 27 kWh unit was identified as optimal for 1F buildings, with a payback period of 6–7 years depending on solar PV area. For 3F buildings, a 39 kWh modular unit achieves payback periods of 3.7 years with 95 m² of PV and 4.2 years with 48 m² of PV. These benefits arise from stored solar energy that would otherwise need to be purchased, and from relying exclusively on low-rate electricity tariffs when solar production is insufficient. Notably, a 3F building without any solar production using the same 39 kWh modular battery still achieves a payback period of 8 years, matching the economics of the larger prototype-based solution while requiring a smaller footprint and lower cost.

Although PCMs are very promising materials for building heating and DHW applications due to their high energy density [6,16] and their nearly isothermal heat storage and release [15], their main drawback is their low thermal conductivity [17]. Other drawbacks are the volume expansion during melting, potentially insufficient thermal parameter stability, and material degradation due to corrosion.

In our approach, we built containers that allow volume expansion, addressing the first drawback. The cycling stability test was performed only on the CrodaTherm 37 tank, due to budget constraints within the research project. The aim of this test was to verify the validity of the RAL Quality 5 certification, which guarantees thermal parameter stability over at least 10,000 cycles; both CrodaTherm 37 and CrodaTherm 53 hold this certification. The measured stability of CrodaTherm 37 over 100 cycles is therefore considered representative of CrodaTherm 53 and, by extension, of the storage and power stability of the entire battery. Assuming approximately 330 full charge/discharge cycles per year—a conservative estimate accounting for periods of lower DHW demand—10,000 cycles correspond to more than 30 years of operation within the certified stability range guaranteed by RAL Quality 5.

In addition to the thermal-parameter stability of the PCM addressed above, long-term degradation due to corrosion of the aluminum heat exchangers was also considered in the design. Unlike systems using open water circuits, the water circulating through the heat exchangers operates in a closed loop, analogous to a thermal battery connected to the DHW (and building heating) circuit(s) via plate heat exchangers, limiting continuous exposure to fresh oxygenated or mineral-rich water. In addition, the circuit pH should be maintained at approximately 8 using an aluminum-compatible corrosion inhibitor, replenished approximately every 7 years, to preserve the stable oxide layer that forms on aluminum and minimizes corrosion risk. While these design choices are expected to limit long-term corrosion, dedicated long-term corrosion testing under real operating conditions was not performed within the scope of this study and remains a subject for future work.

The remaining drawback is still the low thermal conductivity of PCMs. In this respect, the choice of organic PCM such as CrodaTherm 37 and 53 was not ideal, as they have a lower thermal conductivity than salt hydrates and eutectic PCMs. Furthermore, esters have lower conductivity than paraffine and fatty acid. Both used PCMs exhibit thermal conductivities (λ) of approximately 0.16 Wm⁻¹K⁻¹ in the liquid phase and 0.24–0.28 Wm⁻¹K⁻¹ in the solid phase and should therefore be expected to display weak (dis)charge power capacities. However, the prototype achieved a discharge power exceeding 80 kW, with 65 kW sustained at breakthrough. To the best of the authors’ knowledge, this work reports the highest discharge power and the highest discharge power at breakthrough ever demonstrated experimentally for a low-temperature (T_m < 60 °C) solid–liquid PCM system in building heating applications. Previously reported experimental systems generally achieved peak discharge powers between approximately 5 and 15 kW [18,24], whereas the larger sodium acetate trihydrate systems reported by Englmair et al. [25] reached higher transient peak powers of up to 32 kW but substantially lower latent heat discharge powers. Bentivoglio et al. [19,26] reported discharge powers of 40 kW during the solidification of RT70HC PCM in a tube-and-shell storage system containing approximately 1.7 tons of PCM within a heat exchanger volume of 2.65 m³. With a relatively high PCM melting temperature (T_m = 70 °C), the system was designed to store excess heat from the Grenoble district heating network rather than for direct integration into decentralized residential heating systems. By comparison, the present prototype achieved twice this discharge power (80 kW) in a heat exchanger volume of only 0.79 m³—less than a third of Bentivoglio’s 2.65 m³—despite its lower-temperature PCM and its design for decentralized residential use.

However, an important question remains as follows: Are such power levels relevant for residential applications? For practical applications, discharge power can be more critical than storage capacity, particularly when DHW is produced instantaneously. High discharge power enables instantaneous DHW production without requiring a dedicated hot water storage tank. Since the thermal demand of a single shower can reach approximately 24 kW when heating inlet water from 10 °C. Figure 8 shows that the prototype provides sufficient power for a single-family home with a crossflow HX that separates the PCM battery from the DHW circuit. Increasing the battery flow rate to 1800 L/h (30 L/min) is expected to extend this capability to three- or four-family buildings.

Two main factors explain the high heat transfer achieved by the prototype. First, the large metal heat-transfer area (approximately 200 m²), combined with the short conductive path through the PCM (about 1.5 mm), minimizes the conductive thermal resistance. Second, the water flow rate reaches up to 1500 L/h, maintaining a high convective heat-transfer coefficient on the water side.

As the heat exchangers incorporate large water manifolds, they contain a significant volume of water. Consequently, the breakthrough times reached 21 and 11 min for the serial and parallel configurations, respectively. During the initial stage of discharge, the high outlet temperature is mainly sustained by the sensible heat stored in this water rather than by heat transfer from the PCM itself. Nevertheless, this energy contributes to the battery’s total stored thermal energy. Once the breakthrough time is reached, the delivered thermal power is predominantly governed by heat transfer from the PCM. Despite this major change, the discharge power remains remarkably high, reaching 81% of the maximum power in both the serial and parallel configurations. This confirms that the optimized heat-exchanger design maintains efficient heat transfer despite the intrinsically low thermal conductivity of the PCM.

The low thermal conductivity of the liquid PCM (λ = 0.16 Wm⁻¹K⁻¹) limits conductive heat transfer, allowing the thermal stratification generated by buoyancy-driven natural convection to be maintained. As a result, a highly stratified temperature profile develops during discharge. After 24 min in the parallel configuration, the stacked radiators in the lower tank exhibited temperatures of 16, 32, 51, and 60 °C from bottom to top. Such a profile is thermodynamically advantageous because the phase-change front remains localized (e.g., between radiators 2 and 3 after 24 min), allowing the remainder of the tank to retain a large temperature gradient and thereby preserving a high exergy potential.

5. Conclusions

Thermal energy storage is a key solution for bridging the time gap between solar energy production and the thermal demand of residential buildings, which accounts for nearly 80% of their total energy consumption [27]. By using PCMs, the developed prototype stores approximately 42 kWh of thermal energy, corresponding to a volumetric energy density of 54 kWh/m³, while exhibiting a low overall heat loss coefficient of 4.25 W/K.

The low operating temperatures reduce storage heat losses and are well suited for heat pump operation, allowing higher COP and potentially extending heat pump’s lifetime. The thermal battery combines two PCMs: a lower tank filled with PCM (T_m = 37 °C) dedicated to space heating and an upper tank filled with PCM (T_m = 53 °C) for DHW production. Measurements performed through a plate heat exchanger demonstrated discharge powers exceeding 28 kW while heating DHW, enabling instantaneous hot-water production without storing potable water and therefore avoiding the risk of Legionella growth.

One of the key findings of this work is that the prototype, in the parallel configuration, delivers discharge powers exceeding 80 kW, while maintaining 65 kW at breakthrough despite the low thermal conductivity of the PCM (λ = 0.16 Wm⁻¹K⁻¹). This high discharge performance can be explained by the combination of a large heat transfer area of 203 m², a maximum PCM-to-fin distance of 1.5 mm, and a water flow distributed in parallel across 8 aluminum HXs reaching 1500 L/h. This mitigates the power ceiling typically imposed by the low thermal conductivity of organic PCMs, making the proposed concept a promising solution for compact thermal energy storage in residential applications. Moving towards a commercial product, the PCM fraction should be increased through narrower manifolds, copper HXs should be used to further reduce conduction resistance, and long-term stability tests of the PCM and HX materials should be performed.