See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225532612

The effective sky temperature: An enigmatic concept

Article  in  Heat and Mass Transfer · September 2011

DOI: 10.1007/s00231-011-0780-1

CITATIONS READS

15 4,357

4 authors:

Omemah Gliah Boguslaw Kruczek

University of Ottawa University of Ottawa
1 PUBLICATION   15 CITATIONS    64 PUBLICATIONS   844 CITATIONS   

SEE PROFILE SEE PROFILE

Seyed Gh. Etemad Jules Thibault

International ASET Inc. (International Academy of Science, Engineering and Tech… University of Ottawa
53 PUBLICATIONS   2,182 CITATIONS    228 PUBLICATIONS   4,122 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modelling of Mass transfer in Mixed Matrix Membranes (MMMs) View project

Undergraduate Chemical Engineering Thesis View project

All content following this page was uploaded by Jules Thibault on 28 May 2014.

The user has requested enhancement of the downloaded file.

 THE EFFECTIVE SKY TEMPERATURE: AN ENIGMATIC CONCEPT

Omemah Gliah1, Boguslaw Kruczek1, Seyed Gh. Etemad2 and Jules Thibault1

1
Department of Chemical and Biological Engineering, University of Ottawa
Ottawa, Ontario, Canada K1N 6N5
2
Department of Chemical Engineering, Isfahan University of Technology
Isfahan 84156-83111, Iran

Abstract: This investigation is driven by the desire to find an alternative method of

obtaining fresh water due to condensation of the atmospheric water vapour on a
surface maintained below the due point by radiative heat loss from the surface to night
sky. To properly simulate the condensation process and design a water recovery
system, an accurate estimation of the effective sky temperature is required.

To estimate the effective night sky temperature, an experimental system consisting of a

series of metal plates embedded on a heat transfer panel, a weather station and a
control system was used. In addition to the effective night sky temperature, the
experimental system was designed to estimate the convective heat transfer coefficient
between the plates and ambient air as well as the surface emissivity of the plates. The
results obtained from theoretical analyses and preliminary experiments are presented
and discussed. A special emphasis is given to a mathematical solution of the system of
equations that need to be solved to obtain the effective sky temperature. It is shown
that although the system of equations works well for the direct heat transfer problem,
there is a serious difficulty to solve the inverse heat transfer problem to retrieve the
desired parameters.

Keywords: water extraction, sky emissivity, sky temperature, radiation

1. INTRODUCTION
1.1 Water extraction from air

Water is a basic human necessity for survival as important as air and food. The lack of fresh water is currently
one of the main obstacles to the economic development of many countries across the globe. Furthermore, annual
water consumption continues to grow. An additional source of water that is now attracting significant interest is
obtaining water from ambient air (Sciperio, 2005). The extraction of water from air can be accomplished using
several methods: cooling of air (Nebbia, 1961) and absorption of water into a desiccant followed by the
subsequent separation of the water from the desiccant (Sultan, 2004). For the former method, night radiative
cooling can be used to obtain small quantities of water. In this approach, water vapour condenses on the upper
surface of a surface of a plate, cooled by radiative heat loss, when its temperature becomes lower than the dew
point temperature.

To properly design a water collection system, which relies on radiative cooling, an accurate estimation of the
heat transfer parameters is required with a special emphasis on the emissivity of the emitting surface and the
effective sky temperature. Unfortunately, as it will be shown in this paper, available correlations to predict the
effective sky temperature show significant variability.

1.2 The determination of the effective sky temperature

An accurate estimation of the energy emitted by a surface and the surface heat transfer characteristics are
necessary for the design of radiant cooling system and atmospheric water collection system. This information
can also be used in the design of buildings which would remain cool without mechanical conditioners (Berdahl
and Fromberg, 1982). The net radiation loss from the outer surface of the bodies exposed to the sky can be
expressed by:
4
Qrad p A Tp TS4
(1)
where, εp, σ, Tp, and TS are the emissivity of the emitting surface, the Stefan-Boltzmann constant, the surface
temperature, and the effective sky temperature, respectively. The effective sky temperature is commonly
estimated using the following equation (Mills, 1999):
1/ 4
TS Ta4
(2) S

where, εS is the sky emissivity and Ta is the air temperature in the vicinity of the emitting surface. All
temperatures are in degrees Kelvin.

Table 1 presents the main correlations that were proposed in the literature to estimate the sky emissivity and
then to calculate the effective sky temperature. The most commonly used variables in these correlations are the
dew point temperature Td in oC, water vapour partial pressure Pv in mbar, and the atmospheric pressure P in
mbar. A comparison of the results obtained by applying these correlations, under realistic conditions, was
performed to clearly show their significant variability, which unequivocally justify designing an experimental
system to determine the effective sky temperature and eventually a better correlation.

Table 1 The main correlations to estimate the effective sky emissivity.

Formula Author Reference
1 S 0.806 0.236exp 0.092Pv Angstrom Berdahl and Fromberg, 1982
1/ 2
2 S 0.605 0.048Pv Sellers Berdahl and Fromberg, 1982
1/ 2
3 S 0.55 0.056Pv Brunt Berdahl and Fromberg, 1982
0.5
Pv
4 S 0.55 1.8 Brunt Mills, 1999
p
5 S 0.21 0.22ln Pv Elsasser Dhirendra et al., 1995
6 S 0.8004 0.00396Td Bliss Dhirendra et al., 1995
7 S 0.787 0.0028Td Clark and Allen Berdahl and Fromberg, 1982
8 S 0.734 0.0061Td Berdahel and Fromberg Berdahl and Fromberg, 1982
9 S 0.77 0.0038Td Burger et al. Dhirendra et al., 1995

Figure 1 illustrates the variation of the effective sky temperature, using the estimated sky emissivities of Table
1. When the relative humidity increases, the sky emissivity and the effective sky temperature increase. This
increase is caused by the increased opacity of the atmosphere to infrared radiation as the quantity of water in the
atmosphere increases. Figure 2 presents the range of variation of the water condensation rate as a function of the
relative humidity using the two correlations that display the greatest difference in the predicted sky emissivity
values. These correlations are Eqs. (1) and (6) of Table 1.
284 60

Eq. 6
Condensation rate (g/m2h)

282
Eq. 5
Sky Temperature (K)

Eq. 8
280
7 9 40 Eq. 1
2
278

276 3 Eq. 6
4
Eq. 1 20
274

272

270 0
60 65 70 75 80 85 40 60 80 100
Relative Humidity (%) Relative Humidity (%)

Fig. 1. Effective sky temperature versus relative Fig. 2. Effect of the relative humidity on the water
humidity for correlations of Table 1 for Ta of 293 K. condensation rate for two correlations of Table 1.

In this study, an experimental system was designed to determine the effective sky temperature in view of
deriving a correlation that would allow to confidently predicting its value under various atmospheric conditions.
As an additional benefit, the experimental system was designed to allow deriving correlations to estimate the
convective heat transfer coefficient and to estimate the emissivity of the plate surface.
 2. EXPERIMENTAL SYSTEM FOR THE EVALUATION OF THE EFFECTIVE SKY TEMPERATURE

Tsky is a key parameter in the prediction of the condensation rate due to the night radiation cooling. In this
section, the experimental system that was designed and built for the measurement of Tsky is presented. Consider
a thin horizontal plate exposed to a night sky. Suppose that its temperature could be controlled by attaching a
heating element to the bottom surface of the plate. By manipulating the power input to the heating element, it is
possible to control the plate temperature. In practice, for a given Ta, Td depends on the relative humidity of air.
Therefore, to prevent condensation on the plate at any atmospheric conditions the power input to the plate
should be such that Tp > Ta, and in this case the energy balance is given by Eqn. (3). Note that, the plate is
loosing the heat by convection and radiation from the top surface. Losses from insulated edges and bottom were
taken into account by correcting Qext using an equation that was obtained in separate experiments.

p A Tp4 Ts4 hA Tp Ta Qext (3)

In principle, knowing the plate emissivity, the external power input and the resulting plate temperature, Tsky
could be evaluated by solving Eqn. (3) if h were known. Knowing Tp, Ta and wind velocity, h can be estimated
from appropriate empirical correlations. However, these correlations are associated with a certain degree of
uncertainty, which in turn would be transferred into the calculated effective sky temperature Ts.

To overcome potential problems, h should also be treated as an unknown. In this case, there are two unknowns,
h and Tsky, in Eqn. (3). Another serious problem associated with this approach is the error associated with an
uncertainty in the exact values of surface emissivity of the plates. While the surface emissivity is a material
property, it strongly depends on the finishing of the surface. The latter is subject to changes due to surface
oxidation, fouling, etc. If the surface emissivity was considered to be an unknown in Eqn. (3) there would be
three unknowns and only one equation, and Tsky could obviously not be determined. Consequently, it is
necessary to introduce more independent equations to match up the number of equations with the number of
unknowns. That can be achieved by employing three pairs of plates having three different surface emissivities,
which are divided into two groups; each group consisting of three different plates operated at Tp1 and Tp2.
Accordingly, the above system represents a set of six independent equations with six unknowns, 1, 2, 3, h1 , h2 ,
Tsky. Solving this system of equations simultaneously allows the determination of all unknowns, including Tsky.
The only measurable parameters in the above system of equations are the power inputs to the plates required to
maintain them at their specific set temperatures, the actual plate temperatures and the air temperature. In
addition to having the same surface area, each plate should have the same characteristic length for heat transfer.

Each metallic plate has two thermocouples, inserted in groves on the upper surface, to continuously record the
surface temperature. Beneath each metallic plate, there is a heating mat to maintain the temperature of the plate
at its set-point. Each of the 6 metallic plates can be set to individual temperatures above the air temperature. To
ensure that the heat provided to the plates by the heaters is dissipated only through their top surface, the bottom
surface and the sides of each plate are insulated with a 5 cm-thick Styrofoam sheet. In addition, in case of forced
convection, in order to have identical characteristic lengths the longer edge of the heat transfer panel is oriented
perpendicular to the wind direction using a step motor. The motor is attached at the pivot point, which is located
on a tri-pod stand beneath the heat transfer panel. The direction of wind is determined by Vaisala Weather
Transmitter, (Model: WXT510), which also records wind speed, ambient temperature, and the relative humidity.
The signals from the weather station are sent to the data acquisition system. Wind direction is used to orient the
panel in the desired direction. The schematic diagram of the six-plate system is presented in Figure 3.

Fig. 3. Schematic diagram of the six-plate system.

 3. RESULTS AND DISCUSSION

This section presents the results that were obtained throughout this investigation. A large number of experiments
were performed with the six-plate experimental system in the laboratory. It was the intention, following this
experimental validation to install the system on the roof of a building to perform the final series of experiments
that would have allowed measuring the effective sky temperature. Unfortunately, as it will be shown, the
inherent difficulty in solving the set of heat balance equations prevented to complete the initially proposed
experimental plan.

It is interesting to note that most heat conduction problems, such as determining temperature profiles and/or heat
fluxes in a system subject to prescribed initial and boundary conditions are called direct problems. Inverse
problems are usually concerned with finding boundary conditions and heat transfer parameters which give rise
to a measured temperature distribution. While direct problems are characterized by stable unique solutions (i.e.
they are well-posed), inverse problems are ill-posed and typically very sensitive to small variations in the input
data. A theoretical analysis on the mathematical solution of the system of equations in an attempt to find a
suitable solution the problem observed will be presented.

Six energy balance equations were performed for the six plates. Each equation has three unmeasured
parameters: the plate emissivity, the convective heat transfer coefficient, and the effective sky temperature. The
input energy is the heat provided by the heater that was corrected for heat losses whereas the output energy
corresponds to convection and radiation heat transfer. This simple energy balance is expressed by the following
equation with m being the mass of the plate and cP its thermal capacity:
dT p
mc p Qext p A Tp
4
TS4 hA T p Ta (4)
dt

3.1 Steady state energy balance under natural convection

Steady state energy balances were preformed for each plate. Under steady state, the power required to maintain
a given plate at a constant temperature must equal the total of the energy lost by radiation and convection as
expressed by Eq. 3.

All experiments, performed inside the laboratory using the six plates simultaneously and the roof acting as the
sky, led to a system of six equations and six unknowns. In these experiments, three plates of a different surface
finish were set at a constant temperature and the other three plates were set at a different constant temperature.
The power required to maintain each plate at the desired temperature was measured and corrected for heat
losses. It was therefore possible to write a heat balance (Eq. 3) for each plate. Solving the nonlinear system of
six equations with six unknowns was unfortunately not successful. The system of equations was found to be
very sensitive to small errors in measured variables.

To better comprehend this sensitivity problem, a theoretical study was undertaken whereby pure simulation was
used to generate data that was then solved in the same manner. The advantage of simulations is the possibility of
generating data that perfectly satisfy heat balances in the absence of noise and measurement errors. Results in all
cases showed that it is possible to estimate very accurately the effective sky temperature but all the other five
unknowns were far from their original values that were used in the simulation. The reason why the sky
temperature was very well estimated is probably because it appears in all six equations. In theory, this would not
be too problematic because the information that is truly desired is the effective sky temperature. However, under
a realistic situation where inherent noise and measurement errors prevail, the estimation of the sky temperature
becomes impossible as it will be shown in the following paragraphs.

One simulation of the six plates was performed using Eq. (3) and the conditions given in Table 2. For this
simulation, the wind speed was 5 m/s (forced convection), the air temperature was 282 K, and the sky
temperature was set at 260 K. The nonlinear regression was performed 1000 times to estimate the six unknown
variables and, each time, the values of the power required (Qext) to maintain a plate at constant temperature for
all six plates were corrupted by a Gaussian random noise with a standard deviation of 5% of the calculated
values (Table 2).

In Figure 4, values of the sky temperature as low as 215 K and as high as 305 K were obtained. The Gaussian fit
gave an average value of approximately 258 K and a standard deviation of 13 K. It is interesting to note that the
standard deviation corresponds to 5% of the 260 K simulated value, i.e. the identical percentage of the standard
deviation that was used to artificially corrupt the power required for each heater. These results truly show the
sensitivity of the estimate of the sky temperature on the accuracy of the power required by the heaters and the
difficulty to achieve an accurate estimation. The additional five unknowns were also far from their original
values for the 1000 regressions that were performed with the noisy data of Table 2.

Table 2. Conditions for the simulation of the six-plate system.

Plate TP (K) h (W/m2K) Qconv. (W) Qrad. (W) Qext (W)
1 285 0.85 19.1 1.468 2.501 3.969
2 295 0.85 19.2 6.374 3.705 10.079
3 285 0.20 19.1 1.468 0.589 2.056
4 295 0.20 19.2 6.374 0.872 7.246
5 285 0.03 19.1 1.468 0.088 1.556
6 295 0.03 19.2 6.374 0.131 6.505

The objective function that was minimized in order to obtain an estimate of the six unknown variables for the
1000 simulation runs was calculated as the sum the differences between the predicted and simulated power
required to maintain the plate at a constant temperature. Figure 5 shows that, over a wide range of the sky
temperature, the objective function is nearly the same. In that range, a change in the effective sky temperature is
compensated by a change of the other variables involved due to high correlations that exist amongst all the
variables.
40
300
TAvg = 258 K
T = 13 K
Number of occurences

30
(Qpred - Qactual)2 (W2)

200

20

100
10

0 0
200 220 240 260 280 300 320 200 220 240 260 280 300
Tsky (K) Tsky (K)
Fig.4. Number of occurrences of the sky temperature Fig. 5. Plot of the objective function as a function the
for 1000 simulations. estimated sky temperature for the 1000 simulations.

3.2 Steady state energy balance under forced convection

A new experiment was therefore conducted in the presence of forced convection with only one plate. Under the
same experimental conditions, the plate temperature was set in turn to three different steady state temperatures.
For this experiment, it was therefore possible to generate three heat balance equations with three unknowns:
emissivity, heat transfer coefficient, and sky temperature. Since forced convection conditions prevail, only one
heat transfer coefficient applies. Linearizing the heat balance equation, it follows:

hA T p Ta p AT p4 AY Qext (5)
4
With, Y T p S
(6)
However, due to measurement errors, this transformation did not lead to a better solution. Since the system of
equations is linear, it can be theoretically analyzed and, more particularly, it is possible to examine the matrix
conditioning to assess the difficulty in solving this apparently simple problem. For a specific example of this
three-temperature level system, the Eigenvalues are: -0.1943, 14.357 and 2.275×10-11. The huge differences in
these values are a clear indication of the ill-conditioning of the matrix, and point to the same sensitivity problem
noted in the previous discussion.

To continue the analysis, the problem was reduced by one dimension whereby the plate emissivity was assumed
to be known. The energy balance equation required to represent a plate, under the same experimental conditions
but at two temperature levels, is given in Eqn. (7).

hA(T p Ta ) AY QExt. p AT p4 (7)

 The Eigenvalues are: 0.204 and 1.8144×10-9. The Eigenvalues are also very different such the problem cannot
be solved because a small error in the value of the measured power leads to a major difference in the obtained
solution.

Finally, the number of unknowns was reduced to one, the sky temperature, by assuming that the plate emissivity
is known and the convective heat transfer coefficient can be calculated from the correlation presented in
literature. In this case, from the steady state energy balance Eq (3), it is possible to write:

0.25
QExt. hA T p Ta
TS T 4 (8)
p
p A
This equation is also greatly affected by small errors associated with the power input which causes a large error
in the value of the effective sky temperature.

4. CONCLUSION

Water extraction from atmosphere is a potential method to produce fresh water in regions where water sources
are scarce or non-existing. An alternative method was analyzed in this study which makes benefit from the night
cooling phenomenon. At night time, a surface emits energy to a relatively cool sky and its temperature cools
down. If its temperature drops below the dew point temperature of air, moisture will condense on that surface. In
the literature, there exist many correlations to estimate the sky emissivity and then calculate the effective sky
temperature. Applying some of those correlations has led to a large difference in the estimation of the amount of
condensed water.

To design a water collection system, an accurate sky temperature is required. To estimate the effective night sky
temperature, an experimental system consisting of a series of six insulated plates has been built. In addition to
the effective night sky temperature, the experimental system was designed to allow the estimation of the
convective heat transfer coefficient between a plate and ambient air as well as the surface emissivity of the
plates. The experiments performed inside the laboratory using the six plates simultaneously and the roof acting
as the sky led to a system of six equations and six unknowns. Solving the nonlinear system of six equations with
six unknowns (three plate’s emissivities, heat transfer coefficient at two different temperatures, and sky
temperature) was unfortunately unsuccessful. The system of equations was found to be very sensitive to small
errors in the measured variables.

A theoretical analysis of the mathematical solution of the system of equations was performed to find a suitable
solution to the problem. The results obtained from theoretical analyses and preliminary experiments showed that
although the system of equations works well for the direct heat transfer problem, there is a serious difficulty to
solve the inverse heat transfer problem to retrieve the desired parameters. This is due to the inherent error
associated with the measurements and the extreme sensitivity in the evaluation of the sky temperature.

REFERENCES

Burton, K. (1956). A study of the conditions and mechanism of the diphenylamine reaction for the colorimetric
estimation of deoxyribonucleic acid. Biochem. J., 62, 315-323.
Abualhamayel, H.I. and Gandhidasan, P. (1997). A Method of Obtaining Fresh Water From The Humid
Atmosphere. Desalination, 113, 51-63.
Berdahl, P. and Fromberg, R. (1982). The Thermal Radiance of Clear Skies. Solar Energy, 29, 4, 299-314.
Beysens, D. and Milimouk, I. (2000). The Case For Alternative Fresh Water Sources. Pour Les Ressources
Alternatives En Eau., 11, 4, 1-17.
Daniel, B., Irina, M., Vadim, N., Mark, M., and Jacques M. (2003) Using Radiative Cooling to Condense
Atmospheric Vapor: A Study to Improve Water Yield. Journal of Hydrology. 276, 1-11.
Dhirendra, K., Robert, B. and Jack, P. (1994). Effects of Atmospheric Emissivity on Clear Sky Temperatures.
Atmospheric Environment, 29, 16, 2201-2204.
Gandhidasan, P. and Abualhamayel, H.I. (1996). Water Recovery From The Atmosphere. WREC 745-748.
Mills A. F., Heat Transfer, 2nd ed. Prentice-Hall, Inc. 1999.
Nabbia, G. (1961). The Problem of Obtaining Water From The Air. Proc. Of the conf. on Solar and Aeolian
Energy, Part V, 33-46.
Sciperio, A Science Revolution, http: www.sciperio.com/watertech/water-from-air.html consulted (9-17-2005).
Sultan, A. (2004). Absorption/Regeneration Non-Conventional System For Water Extraction From Atmospheric
Air. Renewable Energy, 29, 1515-1535.

View publication stats