1. Research question

Photovoltaic systems often deliver their highest output at times when a building’s immediate electricity demand is lower. A surplus lasting several hours can be exported to the grid, curtailed, stored electrically or used by a load whose operation has been shifted in time. Heated buildings offer another possibility: electrical energy can be converted into heat and stored temporarily in the existing building thermal mass.

The title ‘Store PV Electricity – Without a Battery?’ is therefore deliberately phrased as a question. Thermal storage in walls, ceilings or floors is not a complete substitute for an electrical battery. A battery stores electrical energy and can supply a wide range of loads later. Building mass, by contrast, stores heat. Once the conversion has taken place, the energy is essentially available only for thermal purposes.

The technically relevant question is therefore not whether a building can ‘replace’ a battery. Rather, if surplus PV electricity is available for four to six hours and heat will be needed later, which pathway allows the resulting heat to be stored most effectively in the existing mass?

Five levels must be distinguished: electrical energy conversion, thermal heat generation, the heat-input pathway, short-term activation of building mass and subsequent usability of the stored heat. This distinction is fundamental to the assessment.

2. It is not only the mass that matters – but the pathway into it

Massive buildings have high calculated heat capacity. Yet that total capacity says little about the fraction that can actually be accessed within a limited charging window. Heat must first pass from its source through surfaces, or enter directly within a structure, before reaching the material. This process takes time.

If a room is heated through its air alone, the air temperature rises first. Heat is then transferred to surfaces by convection and long-wave radiative exchange, from which it penetrates building components over time. If, instead, heat is radiated directly onto surfaces or introduced within screed or a wall, thermal activation begins closer to the storage mass.

The heat-input pathway thus becomes a technical parameter in its own right. For short-lived PV surpluses, not only the theoretical storage capacity matters, but also the rate at which heat can penetrate the usable mass before room-air or surface temperatures reach a desired operating limit. This time-dependent perspective is consistent with the wider research on building energy flexibility: building mass can shift loads in time, but the available flexibility window depends on the building, heating system, controls and comfort limits [L1–L4].

3. Physical basis: the verified six-hour analysis

The principal quantitative basis of this article is the separately published study ‘How Much Heat Can a Building Actually Store in Six Hours?’. It compares three idealized heat-input pathways within the same 20 m² reference room. The initial air and component temperatures are 21 °C, and the analysis covers six hours. Heat losses to the outside, ventilation, solar gains and other internal loads are excluded from this basic calculation.

The common comparison limit is no more than +0.5 K additional room-air temperature at the end of the six hours. Importantly, however, the three cases do not have the same temperature profile. In the air model, the +0.5 K increase is present from t = 0 and maintained for six hours. In the radiation and screed models, the room-air temperature evolves through heat transfer and reaches +0.5 K only after six hours. Thus, the results compare defined dynamic limiting cases rather than cases with identical comfort conditions over time.

The reproduced reference values after six hours are approximately 0.552 kWh for heat introduced via room air, 1.503 kWh for idealized direct irradiation of internal surfaces and 1.820 kWh for direct heat input into the modelled screed layer. These values originate from the documented 58-node thermal-network model and are carried over here unchanged [P1–P4].

The figures expressly do not answer which heating appliance is ‘more efficient’. They describe only the additional heat stored after six hours in the masses considered, under the defined boundary conditions.

4. PV surplus as a time-limited charging window

A realistic application is a sunny day with usable PV surplus over several hours. A window of approximately four to six hours may be assumed for this conceptual discussion. This is neither a daily nor a universal figure. Roof size, season, cloud cover, household demand, battery system, heat pump and grid conditions all affect the duration and magnitude of surplus generation.

Without additional use, surplus electricity can be fed into the grid, technically curtailed, stored in an electrical battery or consumed directly by other loads. Thermal activation adds an option for sector coupling at building level: surplus electricity is converted into heat, allowing heat demand to be shifted in time relative to electricity consumption.

IEA EBC Annex 67 explicitly identifies building mass—including walls, floors, ceilings and furnishings—as a source of energy flexibility. The duration of potential load shifting depends on storage mass, heat losses, the HVAC system and comfort requirements [L2]. Fraunhofer ISE likewise classifies thermal use of building mass as one option for grid-supportive or flexible building operation [L5].

4.1 PV power varies even within the surplus window

A four-to-six-hour surplus window must not be mistaken for a block of uniform power. Passing clouds can cause substantial changes in PV output on timescales of seconds and minutes; on timescales from minutes to hours, that variability becomes relevant for load following and storage operation. IEA PVPS expressly describes such changes across timescales of seconds, minutes and hours [L10].

At individual-building level, additional fluctuations arise: different module orientations shift the generation profile over the day, while household appliances, domestic hot-water preparation or charging activities alter self-consumption. The surplus actually available is therefore a time-dependent power signal, P_PV,surplus(t), not merely an energy quantity in kWh.

This introduces a second dynamic requirement for thermal storage: the system must not only have sufficient remaining capacity over several hours, but must also be able to absorb the power being offered at each particular moment. Such power matching is not yet described by the six-hour end-point energy balance alone.

5. Four technical pathways

At least four basic conversion and storage chains can be distinguished for PV surplus. They cannot be ranked categorically as ‘better’ or ‘worse’, because they differ in energy form, efficiency, infrastructure and possible end uses.

Path A – PV → electric direct heating or convective heating → room air → building components. At the point of use, electrical energy is converted almost entirely into heat. However, the heat initially enters predominantly the air, and only subsequently the building components via their surfaces. In the six-hour reference model, precisely this indirect heat-input pathway is the limiting mechanism.

Path B – PV → electrically operated radiant heating → surfaces → building components. Here, too, electricity is converted directly into heat. A greater proportion can reach surfaces before the room air warms. The reference model indicates that this idealized pathway can store more heat in the mass over the defined period than air heating alone. A real infrared heater, however, also has convective components depending on its design, temperature and installation.

Path C – PV → heat pump → floor, wall or building component activation → thermal mass. This combines two distinct levels: for each electrical kilowatt-hour, a heat pump can provide several kilowatt-hours of heat, while a surface heating system can introduce that heat directly into or close to the storage mass. This combination makes heat pumps coupled with thermally activated building mass particularly relevant for demand response [L3, L4, L6].

Path D – PV → electrical battery → later use of electricity, for example by a heat pump. A battery preserves the energy in electrical form. This offers substantial flexibility but requires separate electrical storage hardware. Only when the stored electricity is used later is it determined whether it supplies, for example, a heat pump, household appliances or other loads.

6. Conversion efficiency and thermal storage effectiveness are different quantities

A key source of error in discussions of PV-powered heating is the conflation of heat generation and heat storage. These two levels must be evaluated separately.

6.1 Conversion efficiency

In electric direct heating and electric infrared heating, electrical energy is converted almost entirely into heat at the point of use. One kilowatt-hour of electricity therefore becomes approximately one kilowatt-hour of heat in the building under consideration. This does not mean, however, that the entire kilowatt-hour is stored within the building mass at the desired time.

A heat pump operates differently: most of the heat it delivers is not generated by electric resistance heating but is transferred from an environmental heat source to a usable temperature level. Its thermal output can therefore be significantly greater than its electrical input. COP is the ratio of delivered heat to electrical input power. It is not fixed. The heat source, outdoor or source temperature, required supply temperature, operating load, defrost cycles and system design affect it. Efficiency typically falls as the required temperature lift increases [L6, L7].

6.2 Thermal storage effectiveness

The second question is: how much of the generated heat can actually enter the building mass during the available period before the selected room or surface temperature limit is reached? This storage effectiveness depends on the heat-input pathway, heat-transfer coefficients, material properties, component thickness, active area and control strategy.

A system can therefore be electrically very efficient yet fail to exploit its short-term thermal storage potential fully because heat is transferred to the mass unfavourably. Conversely, a heat-input pathway can activate the mass very directly without the upstream generation process being especially efficient in electrical terms. For an overall assessment, the two levels must be combined multiplicatively or evaluated as a chain—not set against one another.

6.3 Storage capacity and thermal charging power

Two quantities must be distinguished for a thermal store. Remaining capacity in kWh describes how much additional heat can still be absorbed before defined temperature or comfort limits are reached. Instantaneous charging power in kW, on the other hand, describes how much heat can actually be transferred into the storage system at a particular moment.

This charging power is not a fixed material property. It changes during charging and depends on the current temperature field, active area, heat-transfer mechanisms and selected heat-input pathway. A building can therefore retain considerable thermal storage capacity while being able to absorb only limited additional thermal power at the present moment.

For this reason, energy-flexibility research uses power- and time-related indicators alongside energy quantities. Flexibility is not only ‘how many kWh’, but also ‘how many kW, from when and for how long’ [L11].

6.4 Why charging power can decrease during charging

Heat transfer requires a driving temperature difference. In simplified form, heat transfer can be expressed as Q̇ = U · A · ΔT, or locally as Q̇ = h · A · ΔT. As surfaces and near-surface layers warm, ΔT relative to a heat source at constant temperature decreases. The instantaneous rate of heat transfer can therefore fall even though colder mass—and thus remaining capacity—still exists deeper within the component.

Real systems have additional limits: thermal conductivity and layer structure, active area, pipe spacing, volume flow, permissible surface and supply temperatures, and hydraulic or electrical power limits. Thermal ‘charging power’ is therefore a system state, not a constant.

7. Infrared radiation: more direct surface heating, but no efficiency miracle

In the idealized radiation case of the six-hour analysis, the additional power is distributed entirely over internal surfaces. Heating of the near-surface component layers therefore starts immediately; the room air subsequently warms through convection and radiative exchange.

Under the defined model conditions, approximately 1.503 kWh of additional heat is stored after six hours, compared with approximately 0.552 kWh in the air-only case. This does not mean that an infrared heater is ‘three times more efficient’. Electric resistance and infrared heaters convert electricity into heat on fundamentally the same order of magnitude. The model difference concerns the short-term location and time profile of heat input.

The model case itself is also idealized: radiant power is allocated in prescribed fractions to walls, floor, ceiling and furnishings. No geometrically calculated view factors are used. A real infrared panel also has a rear side, housing losses and a convective heat component. Obstructions and geometry may substantially change the distribution depending on the installation location.

The potential technical benefit of direct radiation is therefore not a magically higher energy yield, but a different time-dependent coupling between heat source, surfaces and storage mass.

7.1 Direct radiation in the context of short-term PV fluctuations

For dynamic PV utilization, electrical response capability must be considered in addition to the spatial distribution of heat. In principle, an electrically operated radiant surface can change its output quickly while directing a high proportion of the heat initially toward surfaces. This pathway therefore combines two distinct technical characteristics: rapid electrical controllability and direct introduction of heat into near-surface building mass.

This makes direct radiation an interesting supplementary pathway for brief PV power peaks, but it does not establish a ranking above heat pumps. The six-hour analysis evaluates only the thermal heat-input pathway; it contains neither a comparison of actual appliance efficiencies nor an optimization of control-response speed.

8. Heat pump plus building component activation

The combination of a heat pump with surface heating or component activation is particularly relevant for making use of PV surplus. It combines efficient heat supply with a heat-input pathway situated close to the storage mass.

In the third reference model, heat is introduced in an idealized manner at a depth of approximately 3 cm within the 6 cm screed layer. After six hours, approximately 1.820 kWh of additional heat has been stored in the overall system under consideration, most of it in the screed. Room air reaches +0.5 K only at the end. This illustrates the importance of where heat is introduced; it is not a simulation of a real underfloor heating system, let alone of a specific heat pump.

In a real hydronic system, heat absorption is limited by pipe spacing, screed construction, heat transfer from the water, volume flow, maximum supply temperature, pump output and surface temperature. Deliberately charging a component to a higher temperature can also increase the required supply temperature and thereby reduce heat-pump COP. For a heat pump, keeping the required temperature level of the heating system as low as possible is therefore particularly favourable. The appropriate difference between supply and return temperatures depends on thermal output, volume flow, heat delivery and system operation.

Research on heat-pump demand response confirms that the structural mass of a building can decouple electricity consumption and heating demand over several hours. Sperber, Frey and Bertsch model a demand-response potential for German building types with activation periods typically lasting several hours [L3]. A review by Lind et al. concludes that building mass enables short-term load shifting and that hydronic underfloor heating can be advantageous for exploiting this flexibility [L4].

This establishes a clear research task for real systems: measure not only heat-pump COP, but also how much heat enters floors, walls and the remaining building mass during a defined PV surplus window, and how much of that heat subsequently actually displaces heating demand.

8.1 Underfloor heating as an example of state-dependent charging power

For a hydronic heating circuit, the thermal power currently transferred can be approximated by Q̇ = ṁ · c_p · (T_supply − T_return). If return and supply temperatures approach one another at unchanged volume flow, the instantaneous thermal power released by the water decreases. This does not automatically mean that the entire building mass is ‘fully charged’. It may only mean that, under the prevailing conditions, the transfer pathway currently in use can deliver little additional power into the floor.

Additional charging power would then be possible only by changing the boundary conditions, for example by raising the supply temperature, increasing the active area, using a different heat-input pathway or providing an additional store. Each option has its own limits. In particular, a higher supply temperature may increase the temperature lift required of the heat pump and thus reduce its efficiency.

8.2 Heat pumps can modulate – but are not ideal fast-acting controllers

Modulating heat pumps can adjust their output within a defined technical range. This range has both upper and lower limits; below the minimum output, cycling or other part-load operating modes may occur, depending on the device and system. Starts and stops, and unfavourable part-load operation, can cause efficiency losses. The relevance of these effects depends strongly on the particular unit, hydraulics and controls [L12, L13].

This does not establish a universal minimum run time for heat pumps, nor does it mean that heat pumps are inherently unable to track short-term PV fluctuations. The engineering trade-off is more specific: it may be unfavourable to make the compressor track every individual cloud-induced ramp or brief change in self-consumption if this repeatedly drives operation to modulation limits or causes frequent starts and stops.

8.3 Rapidly controllable electric heating loads

Electric resistance loads such as immersion heaters can change their electrical input very quickly. Electrically operated radiant heating surfaces can also, in principle, be switched or modulated rapidly. For PV load management, this responsiveness is an advantage in its own right: short-lived surplus peaks can be absorbed without requiring a compressor to change its operating point at the same rate.

The energy-related distinction remains: with direct electric heating, 1 kWh of electricity yields approximately 1 kWh of heat at the point of use. Under suitable conditions, a heat pump can provide more useful heat from the same electrical energy. Rapid controllability and conversion efficiency are therefore separate assessment criteria.

9. Multiple thermal loads rather than a single charging pathway

Where PV surplus fluctuates strongly, a combination of loads and storage pathways may be technically appropriate. The individual components then perform different roles, without implying a universal order of priority.

A heat pump can provide a high thermal output per electrical kilowatt-hour and is particularly suited to smoothed or sustained surplus periods. Floor or wall activation makes large masses available for storage, but the instantaneous rate of heat absorption depends on their state and the heat-transfer pathway. A domestic hot-water tank with an immersion heater can absorb electrical power quickly but has limited thermal capacity and temperature constraints. Electric radiant surfaces can respond over short periods and direct heat onto surfaces. A battery, meanwhile, preserves electrical energy for a much broader range of subsequent uses.

9.1 Slow and fast control channels

A cascade consisting of a slow and a fast channel can therefore be envisaged as a technical concept. The slower channel—for example, a heat pump with component activation—responds to smoothed PV surplus, its expected duration, heating demand and the thermal charging state. Faster electric loads can absorb or release short-term differences in demand.

This concept is not a finalized control strategy, nor does it prescribe optimal switching times. It merely describes one possible system architecture: not every load has to handle the same dynamics. This may be particularly relevant when PV power fluctuates faster than a sluggish thermal-hydraulic system can follow in a technically or economically reasonable manner.

9.2 An extended understanding of thermal storage

It is therefore not useful to characterize a building thermal store solely by theoretical capacity in kWh. Practical operation requires at least the current thermal state, remaining capacity, instantaneous charging power, active area, permissible temperature limits and response time of the heat-input pathway.

As a cautious analogy to a battery, remaining capacity may be understood as one aspect of a thermal ‘state of charge’, while instantaneous heat absorption capacity is closer to an allowable charging power. But the analogy has clear limits: a building is not an electrochemical store; its temperature fields are spatially distributed, its storage limits depend on comfort and component constraints, and its charging power changes with its thermal state.

10. Battery as reference – different energy form, different purpose

An electrical battery stores high-grade electrical energy. Once discharged, that energy can still supply electrical loads or be converted into heat using a heat pump. Building mass, by contrast, stores low-temperature heat. This heat cannot readily be converted back into electrical energy of equivalent quality.

A direct comparison based solely on ‘kWh of storage capacity’ would therefore be misleading. At a minimum, energy form, charging and discharging losses, useful storage period, thermal comfort limits, power, controllability, existing infrastructure and necessary additional equipment must be considered.

Building mass has a significant practical advantage: it already exists in massive buildings. Using it for storage nevertheless requires an appropriate heat source and control strategy. An electrical battery requires additional storage hardware, but provides considerably greater electrical flexibility. The appropriate solution therefore depends on the specific objective.

For the application ‘PV surplus at midday, heating demand in the evening’, thermal storage may provide an additional option. For ‘PV electricity at midday, electric mobility at night’, only electrical storage or the grid fulfils the equivalent function.

11. Thermal comfort is a second system boundary

The +0.5 K room-air temperature limit in the six-hour analysis is a simple comparison boundary, not a full comfort model. People exchange heat not only convectively with the air, but also through long-wave radiation with the surrounding surfaces. Surface temperatures and mean radiant temperature therefore affect thermal sensation.

Directly heating a floor or wall can change surface temperatures more strongly than heating room air alone. This can affect comfort, but must not automatically be interpreted as an improvement. Excessively high or asymmetric surface temperatures can also be undesirable.

Comfort effects are addressed separately in the I[n]solation article series, particularly in ‘Radiant Comfort in Front of the Window’ and in a related discussion of why it may feel colder near a window even when room temperature appears adequate. For the storage question, the important point remains that air temperature alone is an incomplete limit, and real experiments should therefore also consider surface and operative temperatures.

12. Time is the decisive storage parameter

For PV surplus, the practical question is not ‘How much heat can this building store altogether?’ Instead, it is ‘How much heat can enter the mass during the surplus window actually available?’

Over a very short charging period, thermal penetration depth remains limited. With more time, deeper layers can be activated, while temperature differences and therefore heat flows also change. One-hour, three-hour, six-hour and longer charging periods consequently represent physically different states.

The present publication uses only the verified six-hour values. Although complete time series exist from the original model, this publication version does not derive new one-hour or three-hour performance figures for publication. Such an addition may be valuable later, but would have to be transparently documented as a new evaluation of the archived original model and checked again.

Energy-flexibility research addresses precisely this time dimension: flexibility is not a constant storage capacity but a function of duration, power, building state, comfort limits and external signals [L1, L2].

The six-hour end values alone therefore do not fully describe charging capability during the surplus window. For future dynamic PV applications, two curves should be considered together: the storage energy already absorbed, E(t), and the thermal power that can still be transferred at that moment, P_aufnahme(t). High remaining capacity does not automatically imply high instantaneous charging power.

The archived original model contains time-dependent states, but this publication version does not silently extend it with new power-related indicators. A quantitative evaluation of P_aufnahme(t) would be a new analysis of the existing model and must be documented and reviewed separately.

13. Comparison per 1 kWh of PV electricity – why no universal number is given

For a future comparison graphic, a starting point of 1.0 kWh of available PV electricity would be illustrative. With direct electric and infrared heating, that electrical energy would appear approximately as 1 kWh of heat at the point of use. With a heat pump, the heat delivered would depend on the actual COP at the respective operating point. With a battery, charging and discharging losses would arise before a later load used the energy.

A numerical ranking is deliberately not presented in this publication version. It would require specific, verifiable assumptions about heat-pump COP, battery-system efficiency, standby losses, supply temperature and operating state. Those assumptions would determine the result more strongly than the storage physics of the available six-hour analysis.

A two-stage assessment is therefore methodologically more appropriate: first, analyse the electrical-to-thermal supply chain; then determine separately what fraction of the heat generated can be stored in building mass during the available window and used later.

14. What research says about flexible building mass

For years, scientific literature has considered building mass a potential source of demand-side flexibility. IEA EBC Annex 67 defines energy flexibility as a building’s ability to shift its demand or supply in response to external signals without violating comfort or technical limits [L1, L2].

Arteconi et al. studied thermally activated building systems in the context of demand-side management. The central function of TABS is to integrate heating or cooling into the building structure so that the components themselves participate as stores in load management [L8].

For German building types, Sperber et al. show that heat pumps can shift electrical loads over time through activation of structural mass; the usable flexibility window is limited and depends on building standard and heat-distribution system [L3]. Lind et al. summarize research on thermal mass in residential buildings and emphasize its suitability for short-term load shifting [L4].

Recent work on TABS also confirms that controls and system design are decisive. High thermal inertia provides flexibility but requires predictive control; technical storage capacity alone does not determine practical value [L9].

15. Scope and limitations

The quantitative core of this article derives from a single 20 m² reference room. The six-hour basic calculation includes no heat loss to the outside, no ventilation, no solar gains and no further internal loads.

Long-wave radiative exchange is calculated in an energy-conserving but simplified manner; geometrical view factors are not determined. The power distribution in the idealized radiation case is prescribed.

Direct screed activation is an idealized introduction of heat within the structure. Specific heat-pump hydraulics, pipe geometry, supply and return temperatures, volume flow and substructure are not simulated.

No fixed COP assumption is used for heat pumps. The existing six-hour values describe thermal energy stored, not electrical energy consumed.

The three reference cases have different temperature profiles. In the air case, +0.5 K is present from the beginning; the radiation and screed cases reach that limit only after six hours.

No universal storage metric for buildings is derived, no annual energy balance is calculated, and no economic assessment of a battery, infrared heater or heat pump is performed.

In particular, the higher six-hour storage figure for the idealized radiation case does not imply that infrared heating is generally more efficient than a heat pump.

16. Technical implication: three questions rather than one

Three questions should be answered separately when using PV surplus for space heating.

First, how efficiently is the available electrical energy converted into usable heat? With a heat pump, this depends on the real operating point and COP. With direct electric and infrared heating, the electrical energy becomes heat almost entirely at the point of use.

Second, how effectively does this heat enter the storage mass during the available window? The relevant factors are location of heat input, surface, temperature difference, component structure, active area and time.

Third, how much thermal power can the selected storage pathway actually absorb in its current state? Remaining capacity, active area, temperature differences, flow rate, permissible temperatures and dynamics of the heat-transfer pathway are relevant here.

Only the combination of these three levels describes the overall benefit for PV self-consumption and load shifting. A discussion confined to appliance efficiency is just as incomplete as stating only the theoretical heat capacity of a building.

17. Conclusion

After conversion into heat, surplus photovoltaic energy can be stored temporarily in building mass. This makes it a potential building-level flexibility resource, but not a functionally equivalent replacement for an electrical battery.

The verified six-hour analysis shows that, in the defined reference room, the storage energy that can be activated in the short term depends strongly on the heat-input pathway: approximately 0.552 kWh for heat introduced via room air, 1.503 kWh for idealized direct irradiation of surfaces, and 1.820 kWh for direct activation of the modelled screed layer. The common limit is +0.5 K additional room-air temperature at the end, with expressly different temperature profiles.

This highlights a second level beyond heat generation: not only how much heat a system provides per electrical kilowatt-hour, but also how quickly that heat enters the building mass determines how useful a short PV surplus window can be.

Direct activation of surfaces or building components can reach more of the mass within a limited period before the room air reaches a predefined temperature limit. Which technology is energetically appropriate also depends on heat-generation efficiency, real comfort limits, controls and available infrastructure.

A clear next step for research follows: do not reduce the heat pump or other heat source and component activation to a single performance figure. Instead, measure heat generation and heat input separately, then evaluate them together as a complete chain.

For fluctuating PV surplus, a third question arises: how much thermal power can the selected storage system absorb right now? Remaining capacity and instantaneous charging power are not the same. A component may still have considerable remaining capacity while the current heat-transfer pathway can transmit only limited additional power because temperature differences have become smaller.

Future descriptions of thermal use of PV surplus should therefore include at least two time-dependent storage quantities: stored energy E(t) and instantaneous charging power P_aufnahme(t). The six-hour end values used here address the first question at the comparison time, but not the full time profile of the second.

18. Position within the existing article series

This publication builds on ‘How Much Heat Can a Building Actually Store in Six Hours?’ without replacing it. The six-hour article remains the detailed physical foundation for the three heat-input pathways.

Additional related articles include ‘The Wall as Thermal Storage – What Dynamic Thermal Protection Changes’, ‘From the Component to the Building Energy Balance’, ‘Window Heat Losses and Solar Gains’, ‘Radiant Comfort in Front of the Window’, and articles on energy balances. Final internal links will be added only when the website is published and the URL structure is unambiguous.

Related reading

• Energy Balance
• Why It Feels Colder Near the Window – Even When the Room Temperature Is Right