Storing Heat Where the Floor Space Is Scarce
The conversion of the UK housing stock to heat pumps carries a spatial cost that is rarely modelled at the outset. A heat pump produces heat at a modest rate over many hours, whereas a household draws hot water in short, intense bursts, and the difference is bridged by storage. The conventional answer is a vented or unvented hot water cylinder, commonly 150 to 250 litres, with a footprint of roughly half a square metre and a height of up to two metres, in addition to any hydronic buffer vessel specified to support defrost and prevent compressor cycling. In the flats, terraced houses and extensions where retrofits are most needed, the cupboard that housed a compact combination boiler cannot accommodate these volumes, and the lack of space has become a recognised barrier to heat pump deployment.
Phase-change material (PCM) thermal batteries offer an alternative. By storing heat as the latent heat of a solid-to-liquid transition, a PCM can hold several times more energy per litre than water within a given temperature window. The principle is sound and the products are real, but the engineering behind them is considerably more demanding than that of a tank of water, and the marketing claims of density multiples need to be read against the physics of the material. This audit sets out the thermodynamics of the comparison, examines the chemistry and crystallisation behaviour of sodium acetate trihydrate (SAT), the most widely used salt hydrate at domestic temperatures, analyses the heat-transfer constraints that govern the discharge power, and discusses the long-term degradation mechanisms over a ten-thousand-cycle horizon. It is intended for architects, building services specifiers and mechanical contractors who must decide whether a compact thermal battery is appropriate for a particular project.
Sensible versus Latent Thermodynamics
A conventional water store accumulates energy as sensible heat, in which the temperature of the mass rises as heat is added: Q = m × c × ΔT, where m is the mass, c the specific heat capacity (4.18 kJ per kilogram per kelvin for water) and ΔT the temperature change. The stored energy is proportional to the temperature window through which the store is cycled. A PCM stores sensible heat in the same way in its solid and liquid states, but in addition absorbs a quantity of energy at an almost constant temperature while it melts: Q = m × ΔHf, where ΔHf is the specific enthalpy of fusion. The total energy stored between a lower temperature T1 in the solid state and an upper temperature T2 in the liquid state is the sum: Q = m × [cs × (Tm − T1) + ΔHf + cl × (T2 − Tm)], with Tm the melting temperature.
Volumetric Comparison
For water, at a density of approximately 988 kg/m3 at 55 degrees Celsius, cycled between 45 and 58 degrees Celsius, a window of 13 K, the specific energy is 4.18 × 13 = 54.3 kJ/kg, and the volumetric energy density is approximately 53.7 MJ/m3, which is 14.9 kWh per cubic metre or 14.9 Wh per litre. For SAT, with a melting point of approximately 58 degrees Celsius, a latent heat of the order of 264 kJ/kg, specific heat capacities of the order of 2.7 to 3.0 kJ per kilogram per kelvin, and densities of approximately 1,450 kg/m3 in the solid and 1,280 kg/m3 in the liquid state, an effective average density of 1,350 kg/m3 gives a latent contribution of approximately 356 MJ/m3, or 99 kWh/m3, and a sensible contribution over the same 13 K window of approximately 14 kWh/m3. The material-level total is of the order of 113 kWh/m3, about 7.6 times that of water in the same window.
A material-level ratio is not a system-level ratio. The PCM must be contained, the heat exchanger that couples it to the water circuits occupies a substantial share of the volume, and the whole must be insulated. If the PCM occupies 60% of the unit's internal volume, the packaged density is of the order of 68 kWh per cubic metre, against about 10 kWh per cubic metre for an insulated water vessel in which 70% of the volume is water, a ratio of 6.5. To store 5 kWh within the 45 to 58 degree window requires about 336 litres of water, but only 44 litres of SAT at the material level, or roughly 74 litres of packaged unit. This is the basis of the often quoted figure of four times the density, which is plausible for a hydronic buffer in a narrow temperature window.
The ratio is smaller when hot water is the application. A domestic cylinder is typically discharged from around 60 degrees Celsius down to the incoming mains temperature of about 10 degrees Celsius, a range of 50 K, so that water stores 57 kWh per cubic metre of material, or about 40 kWh per cubic metre packaged. A PCM unit used over the same range, melting at 58 degrees Celsius and cooled to 10 degrees Celsius in the solid state, stores latent heat plus the sensible heat of the solid, 264 + 2.7 × 48 = approximately 394 kJ/kg, or about 153 kWh per cubic metre of material at a density of 1,400 kg/m3, and perhaps 92 kWh per cubic metre packaged. The ratio in this case is about 2.3 to 2.7. The multiple that can be claimed depends on the temperature window and on what is included in the volume, and a specifier should ask for the stated density to be accompanied by the temperatures, the packaged or material basis and the delivered hot water temperature.
Delivery and Charging Temperatures
An isothermal medium imposes constraints on both charging and discharging that a water store does not. To melt SAT at 58 degrees Celsius, the charging fluid must be hotter than the melting point by an approach temperature to drive heat into the material, which in practice requires a flow temperature from the heat pump of 62 to 65 degrees Celsius. A water cylinder is commonly charged at 55 to 60 degrees Celsius, and the 5 to 8 K difference reduces the heat pump efficiency; with a rule of thumb of 2% to 3% per kelvin the penalty is of the order of 10% to 25% on the proportion of the energy that passes through the store. Similarly, the heat extracted on discharge is delivered at a temperature below the melting point, so the delivered hot water is limited to about 50 to 55 degrees Celsius at best, rather than the 60 degrees Celsius that a cylinder can provide. These are acceptable for most domestic uses, but they interact with the requirements for the control of legionella, which for stored water call for a store above 60 degrees Celsius. Where a PCM battery heats the domestic water in a flow-through heat exchanger and so stores no potable water, the legionella exposure is reduced, though the regulatory classification of such a unit, including any requirement relating to unvented hot water systems, should be checked for the product and project.
Salt-Hydrate Chemistry and Crystallisation Mechanics
Sodium acetate trihydrate has the formula CH3COONa·3H2O. The solid is a crystalline lattice in which each formula unit of the salt is associated with three water molecules of crystallisation. The molar mass of anhydrous sodium acetate is approximately 82.03 g/mol and that of the trihydrate 136.08 g/mol, so that the stoichiometric composition contains 60.3% sodium acetate by mass. On heating to approximately 58 degrees Celsius the crystal lattice releases the water of crystallisation and the salt dissolves in it, absorbing the latent heat, and on cooling the process reverses.
Incongruent Melting and Phase Segregation
The difficulty is in the details of the phase diagram. At 58 degrees Celsius the solubility of anhydrous sodium acetate in water is slightly less than the 60.3% required by the stoichiometric trihydrate composition, so that a stoichiometric sample does not melt fully into a homogeneous solution, and a residue of anhydrous salt remains as solid. This is incongruent melting. The anhydrous crystals are denser than the saturated solution and settle to the bottom of the container. On the subsequent cooling, the settled anhydrous salt is separated from the water that it would need to re-form the trihydrate, and a portion of the material cannot recombine. The lower layer is anhydrous and does not contribute latent heat on the plateau, and the upper layer is a dilute solution that is lower in latent heat per unit volume. Repeated cycling builds a stratified structure in which a growing fraction of the mass is inaccessible, and the capacity decays.
Several measures combat segregation. The first is to adjust the composition by adding a small excess of water, typically of the order of 1% to 5% by mass, to dissolve the anhydrous residue at the melt, at the cost of a small reduction of the latent heat per unit mass. The second is to add thickening or gelling agents, such as cellulose derivatives, polyacrylate super-absorbent polymers, fumed silica or clays, which increase the viscosity of the melt and suspend the anhydrous crystals so that they cannot settle, keeping the water and salt in intimate contact. The third is to limit the geometry, using shallow, narrow cells or a fin matrix in which the vertical settling distance is short. The fourth is to provide mechanical agitation, which is rarely practical in a sealed domestic unit. The thickener must be chosen with care, since it also reduces the diffusion that carries ions to the growing crystal front and so slows the discharge.
Supercooling and Nucleation
The second characteristic behaviour of SAT is supercooling. After melting, the liquid can be cooled well below the freezing temperature of 58 degrees Celsius, to room temperature and below, without crystallising, because the nucleation of the solid phase needs an energy barrier to be overcome. According to classical nucleation theory the work of forming a critical nucleus is ΔG* = 16πγ3 / (3 × ΔGv2), where γ is the interfacial energy between the nucleus and the melt and ΔGv the volumetric free energy released by the transition, which is approximately ΔHf × ρ × ΔTsc / Tm for an undercooling ΔTsc. The nucleation rate is proportional to exp(−ΔG* / kT). The barrier falls steeply with increasing undercooling, so that for a clean, homogeneous melt the nucleation does not occur until a large undercooling is reached. The presence of a foreign surface that is structurally similar to the crystal lowers the effective interfacial energy by a geometric factor f(θ) = (2 + cos θ)(1 − cos θ)2 / 4, where θ is the contact angle, and so permits crystallisation at a small undercooling.
Supercooling can be exploited deliberately. A supercooled liquid retains its latent heat indefinitely with no standing losses, since it stays at ambient temperature, and the heat is released on demand by triggering crystallisation, which is the principle of heat-on-demand storage and of hand warmers. Where a domestic store is intended to act as a conventional isothermal buffer, however, the supercooling is a hazard: if the material does not crystallise reliably as the store is discharged, the latent heat is not released at the plateau and the delivered hot water temperature falls to that of the sensible heat of the liquid, which may be much lower. To ensure that crystallisation occurs at or near 58 degrees Celsius, designers add nucleating agents, such as compounds with similar crystal structures, disodium hydrogen phosphate and related salts being commonly reported, and design the unit so that a small quantity of solid crystals is retained as seed at all times, for example by limiting the maximum charging temperature or by providing a cold region where the material never fully melts. Overheating well above the melting point can dissolve the seeds and the nucleators' effectiveness, and a charge limit is therefore a protection against loss of nucleation. The reliability of nucleation across thousands of cycles is one of the principal points to verify in the manufacturer's test data.
Containment and Expansion
SAT expands on melting. With a solid density of about 1,450 kg/m3 and a liquid density of about 1,280 kg/m3, the volume increases by approximately 13%. The container must accommodate this expansion without pressure build-up, usually through a flexible membrane or bladder, an air space or compliant cells. The container must also be sealed against water loss: the water of crystallisation is an essential part of the composition, and its slow loss through seals or through vents shifts the composition towards the anhydrous side, accelerating segregation. The solution of sodium acetate is mildly alkaline, which affects the choice of materials: copper and stainless steel are generally compatible, while aluminium can be attacked and galvanic couples must be avoided.
Heat Transfer Limitations and Internal Exchanger Geometry
The thermal conductivity of SAT is low, in the range of 0.4 to 0.7 W per metre per kelvin, depending on the phase and the additives, which is comparable to water without the benefit of convection in the solid state. This has important consequences for the power with which the store can discharge. During discharge, heat is withdrawn through a heat exchanger surface and the PCM solidifies on that surface. The solid layer grows outward from the surface into the liquid, and because the solid conducts heat poorly, the layer acts as a growing thermal insulator between the remaining liquid PCM and the exchanger. This is the classical Stefan problem.
In the quasi-steady approximation for a planar layer, the thickness s of the solid grows as s = (2 × k × ΔT × t / (ρ × ΔHf))1/2, where k is the conductivity of the solid, ΔT the temperature difference between the melting point and the exchanger surface, ρ the solid density and t the time. The heat flux through the layer is q = k × ΔT / s. With k = 0.6 W/m·K, ΔT = 5 K, ρ = 1,450 kg/m3 and ΔHf = 264 kJ/kg, the solid layer is approximately 3 mm thick after 10 minutes and 7.5 mm after an hour. The corresponding heat flux falls from approximately 1,000 W/m2 at 10 minutes to approximately 400 W/m2 after an hour. A peak domestic hot water draw, such as a shower at 8 to 10 litres per minute heated from 10 to 40 or 45 degrees Celsius, requires a thermal power of roughly 17 to 24 kW. At a flux of 1 kW per square metre, this requires 17 to 24 square metres of heat transfer area, delivered from a unit that may occupy 50 to 100 litres.
Fin Density and Composite Enhancement
The solution is to make the conduction path in the PCM short, by embedding it in a dense matrix of high-conductivity fins. If copper or aluminium plates are spaced at a pitch of 5 mm, the maximum distance from any point in the PCM to a conductive surface is 2.5 mm, which bounds the solid layer thickness and so holds the flux near its initial high value; the area density of such a stack is about 400 square metres per cubic metre, and a volume of 0.05 cubic metres then provides 20 square metres. The fins are thin, a few tenths of a millimetre, to limit the volume of metal. A fin thickness of 0.4 mm at a 5 mm pitch occupies 8% of the volume, which reduces the PCM fraction and thus the density, and a design trade-off exists between the discharge power and the stored energy per litre. Smaller pitches improve the power but reduce the capacity, while larger pitches do the reverse. Composite approaches, in which the PCM is impregnated into an expanded graphite matrix, raise the effective conductivity to several watts per metre per kelvin and offer an alternative to metal fins, with its own trade-offs in cost, density and long-term stability.
The charging process is generally less limiting, because the melted liquid layer adjacent to the exchanger can develop natural convection, which enhances heat transfer, and because charging takes place over hours. The discharge, which must respond to an instantaneous demand, is the design constraint. A specifier should therefore require the manufacturer's data on the sustained discharge power as a function of state of charge, since the power available when the store is 20% charged may be much lower than when it is full, and a unit that meets a stated peak at full charge may not meet a second shower immediately after the first.
Long-Term Cyclic Lifespan and Structural Fail-Safes
A ten-thousand-cycle horizon corresponds to roughly 27 years at one full cycle per day, or about 14 years at two full cycles per day, which is a plausible duty for a thermal battery that serves both hot water and a heating buffer or that is charged twice daily on a time-of-use tariff. The question for the specifier is how much of the initial latent capacity remains at the end of that period. The mechanisms of loss are the segregation of anhydrous salt described above, the loss of nucleation reliability, the loss of water through the seals, the degradation of the thickener and the mechanical fatigue of the container and the exchanger.
A simple phenomenological model captures the form of the segregation loss: the inaccessible fraction s(N) after N cycles approaches a limiting value s∞ exponentially, s(N) = s∞ × [1 − exp(−N / N0)], where N0 is a characteristic number of cycles. The role of the additives is to reduce s∞ and to increase N0. For an unthickened and unadjusted SAT, laboratory reports commonly show a significant decay of the latent heat within the first hundreds of cycles, and a retention after ten thousand cycles in a wide range, which for illustrative planning purposes might be taken as 55% to 75%. For a composition with extra water, thickener and nucleator, retention figures above 90% are the target, and the products on the market are designed to that goal. The figures in this paragraph are illustrative, and the manufacturer's accelerated-cycle test data, performed on the full product and not on small laboratory samples, should be preferred to any generic model.
Structural fail-safes are as important as chemistry. The container and exchanger undergo a thermal and volume cycle on every charge, with a 13% volume change in the PCM and the differential expansion of the metals, plastics and the PCM. The seals must retain the water over the life of the unit, and any leakage should be detectable. The unit should be fitted with high-temperature protection to prevent overheating the PCM past the temperature at which nucleators lose effectiveness or the container is damaged, and with a controller that monitors the state of charge by temperature and energy metering so that a loss of capacity is visible to the user. The warranty terms, the cycle count to which they apply and the capacity retention they guarantee are useful indicators of the manufacturer's confidence.
| Storage Medium Type | Volumetric Energy Density (within 45-60°C range) | Phase Change Kinetics | Primary Structural Degradation Vector | Lifetime Cyclic Degradation Profile |
|---|---|---|---|---|
| Water (sensible heat), cylinder or buffer vessel | Approx. 15 kWh/m3 for a 13 K window; about 5.7 kWh/m3 for a 5 K window | No phase change; temperature varies continuously; convection mixes the store; discharge power limited only by the exchanger and flow | Corrosion, scale and sludge; anode depletion; legionella management for stored potable water | Stable: negligible loss of capacity over decades if maintained |
| Neat sodium acetate trihydrate (stoichiometric, unthickened) | Approx. 113 kWh/m3 (about 7.6 times water) | Melts at about 58°C; severe supercooling; nucleation unreliable; conduction-limited discharge | Incongruent melting, settling of anhydrous salt, phase segregation | Poor: rapid early loss of capacity; illustrative retention 55% to 75% after 10,000 cycles |
| Formulated SAT (extra water, thickener, nucleating agent) | Approx. 100 to 110 kWh/m3 (slightly reduced by the additives) | Melts at about 57 to 58°C; controlled nucleation near the plateau; discharge set by fin geometry | Water loss through seals, thickener ageing, nucleator deactivation if overheated, container fatigue | Good, subject to verification: target retention above 90% after 10,000 cycles |
| Organic paraffin PCM (reference alternative) | Approx. 50 kWh/m3 (latent about 200 kJ/kg at a density of about 800 kg/m3) | Melts over a small range; negligible supercooling; very low conductivity (about 0.2 W/m·K) | Flammability, volume expansion of 10% to 15%, oxidation at high temperature | Good chemical stability with no segregation, though at a lower density and with fire-safety considerations |
The densities in the matrix are derived from the properties stated in the text and are intended to show the order of the differences. Real products differ in formulation and in the proportion of volume given over to the exchanger and insulation, and the data sheet and the test reports are the proper source for any particular unit.
Specification Guidance for Practitioners
The analysis suggests the following approach. First, establish the real constraint. If the limiting factor is floor space, as it is in many flats and retrofits, a PCM battery can resolve it, and the saving of perhaps 60% of the volume at equal stored energy is a substantial gain. If the limiting factor is cost, a water cylinder is cheaper per kilowatt-hour of storage, and the compact unit carries a premium that must be justified by the space or by other benefits. Second, ask for the density on a stated basis, with the temperatures, the packaged or material basis and the delivered hot water temperature.
Third, verify the heat pump interface. A store that melts at 58 degrees Celsius calls for charging at 62 to 65 degrees Celsius, and the heat pump must be capable of delivering it, with the efficiency penalty accounted for in the energy modelling. Fourth, request the discharge power as a function of state of charge, and check it against the peak demand of the dwelling, remembering the possibility of consecutive draws. Fifth, review the evidence for the cycle life: the cycle count, the capacity retention, the test conditions and whether the testing was performed on the full-scale product. Sixth, confirm the containment, expansion, over-temperature protection and monitoring, and the regulatory status of the unit. Seventh, plan the end of life: salt hydrates are generally benign, but the unit should be recyclable and the manufacturer's take-back arrangements should be known.
Taken together, the comparison between a phase-change battery and a water store is not a matter of one technology being superior. It is a trade between space and simplicity. Water is cheap, forgiving, well understood and unconstrained by crystallisation kinetics, but bulky. A formulated salt hydrate is compact and can be remarkably effective, but it demands higher charging temperatures, is limited by conduction, and depends for its longevity on materials science that the buyer must take on trust unless the supporting evidence is provided. Specifying it well means asking the questions that this article has outlined.