The sky should not only be seen as a potential source of energy in the form of sunlight, but also as a low-temperature reservoir for cooling. This cooling can be obtained through a longwave radiative heat exchange between a radiator, located on the surface of the earth, and the cold air masses above it. Passive cooling methods such as radiative cooling can significantly reduce dependence on carbon-based fuels.
Air-conditioning produced by a traditional vapor compression cycle is an energy demanding operation. Today 10–15% of the world’s electricity is used for cooling and this percentage is increasing; for instance, air-conditioning energy needs in Finland are expected to grow tenfold by 2050.
This electricity is mostly generated from carbon-based fuels and thus contributes to climate change via the greenhouse effect. By using passive cooling methods is it possible to avoid, or at least reduce, the need for vapor compression cooling. A significant amount of cooling could be obtained by using the sky as an alternative thermal reservoir.
Radiative coolers serve diverse cooling needs
Radiative coolers can achieve savings not only for air-conditioning, but also for the refrigeration of foodstuffs. The radiative coolers could also be used as solar collectors when sufficient cooling is not needed or attainable. Radiative cooling, yet to be exploited for cooling in northern Europe, can be used for air-conditioning and for cooling to even lower temperatures.
With the use of meteorological data obtained from the Finnish Meteorological Institute, the heat exchange from a flat plate radiator has been modelled.
Figure 1 presents the cumulative frequency distribution at varying radiator temperatures. It describes for what length of time, a specific heat transfer could have been achieved, for a defined radiator temperature during the two-year measurement period. It shows, for example, that a heat transfer of 50 W/m² could have been attained 50% of the time at a radiator temperature of -12°C. The results are comparable to the cooling obtained from a PV-cell driving a refiregerator. The biggest differences are that the availability of radiative cooling is higher and that lower temperatures can be reached with the PV-cell hooked to a refrigerator. 
There are already commercial products such as paints and roofs that utilize a radiative cooling effect. However, the cooling effect of skylights has been largely overlooked.
Figure 1. Cumulative frequency distribution of radiative heat exchange in W/m2 at different radiator temperatures in Helsinki
Double-glazed windows control radiative heat flux
With the use of a double-glazed window, of which the inter-glazing space can be filled with a greenhouse gas, it is possible to control the radiative heat flux from a room (or buildings in general) to the skies. Calculations show that expanding the wavelength range for window material transmittance increases heat ﬂuxes through the system while using greenhouse gasses gives an insulation effect.
However, since cold winters are common in northern Europe, there is a need for a certain amount of insulation. One option would be to pump out the greenhouse gas from the skylight whenever cooling is required and to pump it back in when insulation is again needed. As this pumping is somewhat impractical, an alternative was developed.
Figures 2A & 2B. Conceptual design of the skylight
Controllable skylight concept
The controllable skylight has two different operating modes, one for cooling and one for insulation. It consists of two windows made of a transparent material for longwave heat radiation and one of regular silica (opaque to longwave heat radiation).
By filling the volumes between the glass sheets with a so-called participating (greenhouse) gas that both absorbs and emits heat radiation it is possible to provide a room with cooling or thermal insulation when needed in a manageable fashion. In the skylight such a gas works as the heat carrier of the system, when the skylight is in cooling mode (See Figure 2A), and as a heat barrier, when in the insulating mode (See Figure 2B).
This means that when in cooling mode the gas found in gas layer position one will absorb heat from the space (room) located below it. As the gas temperature increases, the gas density increases and it flows to gas layer position two. Here, the gas is cooled down by radiative cooling from the sky, which in turn increases the density of the gas and thus makes it flow back down to gas layer position one. When the skylight is in its insulating mode, the thicknesses and temperature differences are so small that convective swirls will not be formed; consequently, forming an insulating barrier. 
Selection of materials critical
One of the most critical points in this work was the selection of materials. Gases used in similar applications have not worked or are detrimental to human health and/or the environment. Therefore, another gas is needed that can be used for radiative cooling.
To address this demand, the list kept by the Agency of Environmental Protection (EPA) of gases that possess a global warming potential (GWP), is used. It lists the GWP of different gases with their related atmospheric lifetime. The GWP compares a gas’ heating effect in the atmosphere to that of CO₂.
The gas preferred for the skylight should have a high GWP but disintegrate rapidly if subjected to the elements outside the skylight. From this evaluation, HFC-125 stands out as being suitable for use in windows, but it will, however, presumably also be prohibited in the future. 
As regular windows are opaque to longwave heat radiation other materials were needed. A low-density polyethylene (LDPE) film was used in preliminary experiments as it is partially transparent to heat radiation at the required wavelength interval and cheap.
The problem with an LDPE window is that it needs to be thin to be transparent and, therefore, is too fragile to be used in real-life applications. Zinc sulphide (ZnS) was identified as an interesting material with promising properties and thus investigated further. The material was measured to have a transparency of 64% in the 8 to 14 µm interval (the so-called atmospheric window) when 4 mm thick.
Figure 3. The shutter-type skylight
Concept prototype of the skylight
Optimizing the skylight design
Adverse convective cells were observed in skylight simulations for both the cooling and heating cases of the model presented in Figure 2.  These cells decrease both the cooling and insulating capacity and should therefore be minimized.
The solution was a redesign of the skylight that used a genetic algorithm to optimize the dimensions of the skylight into a shutter-type skylight as presented in Figure 3. This design enables an assembly of the skylight in different sizes without losing either cooling or insulating properties.
The aforementioned optimization has two conflicting objectives; for summer use the cooling needs to be maximized, corresponding to the maximum heat transfer through the window, while for winter use the insulation properties need to be maximized, corresponding to a minimal amount of heat transferred. The variables studied were the skylight width (W) and height (h) as shown in Figure 3. More variables could be considered; however, this would make the computation time unmanageable.
The results from this optimization are presented in Figure 4, where every point is a separate window with its dimensions. The circles form a Pareto front, which represents the trade-offs between the conflicting optimization problems.
Due to calculation limitations, the radiative heat transfer was initially excluded from the optimization. Later the radiative heat transfer was calculated for the different designs of the Pareto front. The results indicated that the cooling effect varied between 47 and 53 W/m² and the insulation effect varied between 22 and 32 W/m². 
The work was finalized after an experimentation stage. The research results will be published in the coming months in the sixth and final paper of the thesis.
Figure 4. The results of the optimization. The Pareto front is noted with “o” and the inferior solutions are marked with “+”.
 Radiative cooling in northern Europe for the production of freezer temperatures, M. Fält, R. Zevenhoven, Proc. of ECOS’ 2010, Volume III, Cycles and Buildings, Lausanne (Switzerland), June 14–17, 2010, paper (peer-reviewed) 208, 413–419
 Radiative cooling in northern Europe using a skylight, M. Fält, R. Zevenhoven, Journal of Energy and Power Engineering 5 (2011) 692–702
 Optimizing a design for a cooling or isolating skylight, M. Fält, R. Zevenhoven, Applied soft computing, Submitted 2014
 Thermal radiation heat transfer: including wavelength dependencies into modeling, R. Zevenhoven, M. Fält, L.P. Gomes, International Journal of Thermal Sciences 86 (2014) 189–197
 Combining the radiative, conductive and convective heat flows in and around a skylight, M. Fält, R. Zevenhoven, Journal of Energy and Power Engineering 6 (2012) 1423–1428
 Experimentation and modeling of an active skylight, in preparation, M. Fält, R. Zevenhoven.
The original text was published in our 2/2014 Top Engineer magazine