You are on page 1of 6

Solar Energy,Vol. 53, No. I, pp.

21-26, 1994
Copyright 0 1994 Elsevier Science Ltd
Pergamon Printed in the USA. All rights reserved
0038-092X/94$6.00 + .OO
003%092X( 94) E0020-D

THERMAL ANALYSIS OF FIVE OUTDOOR SWIMMING


POOLS HEATED BY UNGLAZED SOLAR COLLECTORS

B. MOLINEAUX, B. LACHAL, and 0. GUISAN*


Groupe de Physique AppliquCe and Centre Universitaire d’Etude des Probltmes de l’Energie,
University of Geneva, Switzerland

Abstract-We have analysed measurements from five outdoor swimming pools located in Switzerland and
heated by unglazed solar collectors. The main contributions to the daily energy balance of the swimming
pools are evaluated. They include the active and passive solar gains, as well as the heat losses related to
radiation, evaporation, convection, and water renewal (in order of importance). Coherent results arc obtained
using multilinear regressions in order to determine the best fitting values of the empirical parameters involved
in the thermal equations.

1. INTRODUCTION The cooling effect of fresh renewal water is evaluated


This paper presents the thermal analysis and the energy by either estimating the fresh water temperature or
balance of five outdoor swimming pools located in considering it as a free parameter. Radiation losses are
Switzerland and heated by unglazed solar collectors. computed using modelled sky temperature data
The collector analysis, the collector performances, and (Ineichen et al., 1984). Convection and evaporation
complementary information concerning the measure- losses involve a convection exchange coefficient which
ments and the sites are given in a companion paper depends on the wind speed and which can be consid-
(Molineaux et al., 1994). ered either as a known, or a free parameter. By selecting
only night periods, we eliminate solar gains and focus
on the pool losses. By considering days only, or days
2. SWIMMING POOLS, DATA, AND ANALYSIS and nights, the global pool energy balance can be in-
Figure 1 illustrates a schematic diagram valid for vestigated. We have tried all these different approaches
the five systems under study. Table 1 gives various in order to get a coherent description of and a satis-
characteristics concerning each swimming pool. Table factory explanation for the observed data.
2 presents a list of the main parameters related to the Considering Tables I and 2, various studies are
meteorological conditions and the swimming pool, possible. For sites 1 and 2, we analysed separately the
specifying for each site, what information is, or is not, night losses and the pool energy balance for full day
available. (24 h) periods, For site 3, we have no data concerning
Some important parameters were not measured on the auxiliary heating used only at night, which influ-
site but obtained from nearby meteorological stations ences the pool temperature; the analysis of the pool
or evaluated by other means, thus limiting the accuracy energy balance is therefore restricted to daytime periods
of the analysis. Having not participated in the mea- from 9 am to 4 pm. For site 4, we are lacking a reliable
surements, these are not described here nor are the measurement of the pool temperature, no analysis is
corresponding uncertainties. Simple quality control possible. For site 5, data concerning the solar gains
tests showed, however, that the measurements are quite from the collectors are unreliable, only information
reliable as confirmed by the coherence of the results. about pool cover effects can be deduced from the anal-
The lack of published material on the subject also mo- ysis of night losses.
tivated us to extract as much information as possible Results are presented in the following sections with
from these limited, but existing data. complementary explanations.
Let us see now, by looking at Tables 1 and 2, how
to consider and evaluate the main contributions to the
energy balance of the swimming pool. The active solar
gains from the collectors are measured. The passive
solar gains involve an absorption factor which can be
estimated at approximately 80% or they can be con-
sidered a free parameter. Over any period of time, the ,
variation of the pool’s internal energy, which is related $ ’ m
Lf h!%---c
,Rd
to the thermal capacity of the pool, is known from the
pool’s temperature variation. The thermal pump con-
tribution is evaluated from the electricity consumption. .
‘(auxiliary) water

Fig. I. Schematic diagram for the 5 systems under study. A-


* ISES Member. F: position of the temperature sensors.

21
22 B. MOLINEAUX, B. LACHAL, and 0. GUISAN

Table 1. Swimming pool characteristics

1 2 3 4
Site (location) (Soleure) (Burgdorf) (Tenero) (Uetendorf)

No. of pools* I 2 3 1
Area [m’] 1320 2200 1360 3250 3130
Volume [m”] 3300 3200 2850 2000 4Y50
Auxiliary heating+ no gas heat pump no no
Night cover [W] 0 0 12 0 -90
Flow through filter
[m3 h-I]$” 660 270 620 1500
Pump power [kW]’ 32.2 15.5 4s

Solar collector characteristics are given in Molineaux et al. (1 YY4).


* Number of connected pools. treated as one pool; ’ during the period of measurements, site three is the only site in which
the auxiliary heating was used.
t Evaluated from design values or manufacture specifications; Bcontinuous flow,: ’ no investigation carried out, values not
usable in this analysis.

3. SWIMMING
POOLEiwRGYBALANCE Infrared radiation losses are given by:
For the sake of comparison. all heat or energy PIR = t(~( T;f - T:) [W mm’] (2)
transfers are expressed in [ W me2 1, i.e. the energy flux
per unit of exposed pool area. When considering only t = water emissivity averaged over IR spectrum f
the most significant contributions, the energy balance 0.95:
of a swimming pool for a given period corresponds to: 0 = 5.67. IO-’ [W mm2 Km“],
T,,., = pool, sky temperature [K]
gains - losses = internal energy variation (AU),
gains = active solar (P,,) + passive solar ( PpS) + pump Table 2. Measured data, relevant parameters
contribution (PO) (+ auxiliary), and information
losses = IR radiation ( PIR) + convection (PC) + evap-
Site: 1 2 3 4 5
oration ( Ptw) + water renewal (Par),
METEOROLOGY
or, in other terms:
Global solar radiation* T T T T T
Outdoor temperature+ M M M M M
PO,+ P,,, + P, - P,R - PC- PC,>- P,., = AU (1)
Wind velocity’ M M M M M
Ambient relative humidity T T T T T
Obviously, active and passive solar gains do not apply IR radiation EV EV EV EV EV
when considering night periods. Rainfall T T T T T

SWIMMING POOL
3. I Gains and losses
The active solar gains from the collectors (Pop), are Temperatures:
-outlet to filter (A) M 0 M 0 M
computed from temperature and flow rate measure-
-return after renezal (B) § 0 M M 0 0
ments at points C & D of Fig. 1 as seen previously by -fresh renewal water 0 E 0 0 0
Molineaux et al. ( 1994). Volume of fresh water D 3D D D D
Passive solar gains (P,,,) are evaluated by applying Use of pool cover (if any) - - o-o
Number of entries” 0 D 0 D 0
an absorption coefficient for the swimming pool (LYE)
Solar gains from collectors M M M M M
to the solar radiation reaching the pool. According to Auxiliary heating - - 0” - -
Talwar (1989), Sheridan (1989), and Govaer and
Zarmi ( 198 1 ), the coefficient q, lies in the range of M: measured every 12s. sum or average recorded every 5
min: T: hourly values taken from nearby meteorological sta-
0.75 to 0.9. A numerical model was used to evaluate
tion (~30 km); 0: no measurements, no recorded information:
the value of ap by summing multiple reflections of D, 3D: one daily record, three daily records (morning, noon.
beam radiation in an infinite layer of 2 m of water over evening); EV: evaluated by use of meteorological models
blue tiles, for all wavelengths and angles. We obtained [Ineichen ef u/. (1984)]; E: estimated to vary linearly from
a mean daily value of 0.8 1 when considering full days 10°C in May to 14’C in July; (M): measured as M, but mean-
ingless because of problems in solar loop (leaks, air bubbles).
in summer time, and instantaneous values around * global horizontal radiation corrected for shading by use
noon slightly higher than the mean daily value. ap may of Iisheye photos and solar radiation models. Correction factor
also be considered a free parameter when fitting the ~5%; ’ measurement close to collector location. Also consid-
energy balance of the pool. ered significant for swimming pool analysis; * measurement
close to collector location and - 10m above pool surface;
The pool or filter pump contributes a non negligible
t letters A and B refer to Fig 1: ’number of entries to the
amount of heat to the pool (P,), an estimated 75% of public swimming pool. No record of number ofpeople actually
the rated electrical pump power. in the water; IIheat pump operating only during the night.
Swimming pool heating 23

Convection losses are usually given by: the heat is lost through breathing. Evaporation is
also increased in an agitated, crowded pool. In site
PC = h,,AT [W m-‘1 (3) 1, where the data is available, the only correlation
found between the number of entries to the pool and
AT = temperature difference between pool and am- the measured temperature variation is linked to the
bient [K], increased volume of water renewal on crowded days;
h, = convection heat exchange coefficient [W m-* ?? effect of rain: Rain is scarce during the Swiss summer,

Kl. 254 mm fell in site 1 in 5 months corresponding to


a cooling power of 1 W m-* ;
Evaporation losses are evaluated according to:
?? heat losses through conduction to the ground are
more substantial. A simple model for a semi-infinite
P, = (UC,). -L.(Xs - X,) [W m-*1 (4)
wall submitted to a constant pool temperature of
22°C was used [ Sacadura ( 1982)], based on an ini-
h, = convection heat exchange coefficient [W m-* tial ground temperature of 1O’C. Applied to the walls
K-‘I, of the pool in site 1, a resulting heat loss of 7 W m-*
C, = specific heat of air [J kg-‘K-’ 1, over 4 months was determined. The losses through
L, = latent heat of vaporisation of water [J kg-‘], the underground pipes are considered negligible in
X, = humidity content for saturated air at pool tem- all sites but site 4. No evidence was found of cor-
perature [kg/ kg], relation between the ground losses and the pool en-
X, = humidity content of humid air at ambient tem- ergy balance.
perature [kg/ kg].
Equation (4) was derived with the assumption that the 3.2 The convection heat exchange coeficient (h,)
air flux involved in eqn (3) gets saturated with hu- The coefficient h, concerns convection and conse-
midity. quently evaporation, IR radiation has to be considered
The quantity X is directly related to partial pressures separately. The coefficient h,, upon first approximation,
(p) and molar masses (M): is a linear function of the wind speed, v (in ms-’ ).

X=pw.- M, h, = A + Bv [W m-* K-‘1 (6)


(5)
PafPw Ml
Referring to previous studies, assigning values to pa-
where the sub indexes a or w refer to air or water and rameters A and B is rather confusing. The ASHRAE
where X refers either to X, or X, . The convection and Handbook has adopted an expression, apparently first
evaporation losses depend strongly on the convection suggested by Jurges ( 1924) and concerning measure-
coefficient h,, which will be discussed in the next sec- ments on a small vertical copper plate:
tion.
Cooling losses linked to water renewal (P,,,,) are re- h, = 5.7 + 3.8~ [W m-* K-‘1 (7)
lated to the flux and the temperature (T,) of fresh re-
newal water. The amount of fresh water added to the It is suggested by Watmuff ( 1977) and Duffie and
pool is recorded at least once a day. The fresh water Beckman ( 199 1) that eqn ( 7 ) also includes the effects
temperature is estimated: It can also be considered as of radiation and that this equation should be (for con-
a free parameter when fitting the energy balance of the vection alone) :
pool under study.
The internal energy of the pool is directly related h, = 2.8 + .3.Ov [W m-* K-‘1 (8)
to the pool temperature, which is measured at point
A of Fig. 1 for all sites but sites 2 & 4. In site 2, pool The Australian standards ( 1989) are also close to eqn
temperature is derived from the temperature mea- (8):
surement at point B of Fig. 1, taking into account the
effects of the pump and water renewal. In site 4 tem- h, = 3.1 + 4.1~ [W me2 K-‘1 (9)
perature measurements are only available at points E
& F, no value of pool temperature could be deduced. Also subject to discussion is the height at which the
Stratification effects are assumed negligible due to the wind speed should be measured. This has a direct in-
important and permanent flow through the filter. fluence over the value of parameter B. According to
The other heat exchanges which were not taken into Touma ( 1977) and Sheridan ( 1972), the wind speed
account in the pool energy balance are the following is related to the height Z above ground level by the
(given in units of average heat flux over the given pe- approximation:
riod):
?? effect of swimmers: 1000 swimmers a day spending v, = k-Z” (10)
an average f hour each in the pool will release an
average 6 W m-* over 24 hrs at site 1. Average num- where k is a constant and the value of the parameter
bers of swimmers over the measured period were (Y being evaluated around -0.14-0.2 for suburban
inferior to this figure in all sites. In addition, part of conditions.
24 B. MOLINEAUX, B. LACHAL, and 0. C&ISA\

Table 3. Convection heat exchange coefficient (h,) and ( 11). Table 3 gives the values of parameters A
and B adapted to a measurement of wind speed 10 m
h,.= A + Bv r&rencc~: ASHRAE Australia
above pool. In addition, A and B can be considered
A [W m-* K-r] 5.1 3.1 free parameters in fit procedures.
B* [J mm3K] 3.8 4. I
“reference height” of wind speed [m] surface+ .3
resulting value of B with wind speed As seen previously, all gains and losses can be more
measured 10m above pool* .2-.44 .5-.61
or less well evaluated. The evolution of the resulting
* value of B as given in the reference with wind speed, v, pool temperature can be compared to the observed
measured at “reference height”: ’ value stated in the reference one. This is the basic test of our analysis.
but eqn (10) gives v = 0 at pool surface. therefore an arbitrary The comparisons are carried out on both an hourly
reference height of 0.03m was chosen. also more adequate for
and a daily basis. By fitting daiiy data, we get reliable
the wind speed measurement; * value of B determined from
eqn (I 0) for the purpose of this study. The range is due to the and coherent results. Fitting hourly data does not im-
estimated value of 01in eqn (10). (Y= 0.14 to 0.2. prove the situation. we are limited by the large thermal
capacity of the pool and by the accuracy of the cor-
responding temperature measurement. For site 1~con-
If, in eqn (6), B is defined for a given reference sider. for instance. that a temperature variation of
height (Z,) and if we have to deal with wind speed data 0.1 “C in 24 h corresponds to an energy flux of 12 W
corresponding to another height (2,)) parameter B has m ‘. By selecting night periods, we eliminate solar gains
to be corrected according to eqns ( 10) and (6): and concentrate on the other contributions (the con-
vection coefficient /I, for instance). Considering 24-
= B,.j’(%,) (11) hour periods, we focus on passive solar gains and the
corresponding absorption coefficient ( LY,,).By following
such ideas. various linear and multilinear regressions
where B, and B, correspond to 2, and Zr. were performed.
In our analysis of swimming pools, the wind speed The most important results. as well as the most
was measured at a height Z of about 10 m above pool. signihcant parameters, of such fits are shown in Table
Different values of the parameters in eqn (6) were con- 4. Each line of this table corresponds to a least squares
sidered in this study. According to eqns (7), (9), ( 10) tit over the given periods. The value of a parameter. if

Table 4. Various fits of swimming pool energy balance

Site/days/time
period+

I / I37d/night only 18.7 + .6 3.5 i I 2.7 + .5 27.7


(9am to 4pm) 2 17.0 f .5 3.1 2.1 29.0
3 26.5 f .9 5.7 I .4 5Y. I
4* 18.0 3.1 2.1 - 30. i
5 18.0 3.1 2.5 29.7
6 18.0 5.7 74 64.4
7 18.0 5.7 1.73 65.9

l/l 37d/full day 8 19.7 k .7 2.6 + 3 1.1 + .j .70 i .03 27.0


(Oam to 12pm) 9* 18.0 3.1 3.1 .80 28.X
10 18.0 5.7 .74 .80 67.8
11 18.0 5.7 .-I4 I .oo t .O? 41.1

Z/SXd/night only 12 2.9 i .l 2.2 t .s 18.6


(9am to 4pm) 13* 10-14 3.1 2.1 18.7
14 (see Table 2) 3.1 2.5 19.9
I5 5.7 .74 73.4
16 5.7 1.72 80.5

2/58d/full day 17 2.3 +- .3 2.9 i- .j .73 i .03 14.9


(Oam to 12pm) 18 10-14 3.1 2.1 .79 f .Ol 15.5
19 (see Table 2) 5.7 .74 I .oo +-.02 28. I
20* 3.1 2.1 .80 15.8
21 5.7 .74 .80 52.4

3/73d/daytime 22 19.9 1 2.9 5.7 r_t6 4.4 .* I.6 .Y9 I? .03 48.3
(9am to 4pm) 23* 15.0 3.1 2. I 80 58.5
24 15.0 5.7 .74 .80 87.3

* Most coherent fit


+ Local time
r Standard deviation of linear fit
Swimming pool heating 25

assigned, is given without error. The value of a param- [ Wmm2]


eter, if free, is given with its error (standard deviation), 300 - 0
as resulting from the fit. We also give the standard 58 nights= ^ 0
y---l----
RMSE
deviation ( uht) of the fit applied to the global energy
200 ., 18.7 [Wm-‘1
balance of the pool expressed in W rne2. The fits are
very sensitive to parameter A, they are sensitive to ap, \, E
weakly sensitive to T,,and even less sensitive to B.
The fresh water renewal temperature T,is known
or evaluated with an uncertainty ofthe order of f2”C, h, =3.1+2.l~[Wm~~K-‘]
04’ I
which corresponds to f6 W me2 when translated into
0 100 200 300
energy flux units.
Let us discuss in more detail the fits of Table 4. For COMPUTED LOSSES [Wm-‘1
the first fit of each site, the parameters are left free. In
the following fits we assign fixed values to the param- [ Wme2]
eters, in order to test different assumptions. The stan- 300
dard deviation of the fit ( ar,) has to be considered as RMSE = 15.5[Wme2] ,’
an indicator of the quality of the fit and consequently 2 ap = 0.79k.01
of the validity of the corresponding assumptions. 3 200 ‘i &&,f”’ ,~,
For example at site 1, the first fit appears reasonable.
F 0&
-’*j
+
Fits 2 and 3 show clearly that the value of A from z 100 ‘, ‘.$ ;-
Australian standards gives much better results than the ‘5( ,. 58 days
l/ 1
value of A from ASHRAE, considering both ufit and 2 J.’ h, =3.1+2.lv[Wm-‘K-‘1
/’ 4
0
T,.Actually, T,= 26.5”C (fit 3) is not acceptable (far
0 100 200 300 400
from reality) whereas a value of T, around 18°C is
realistic. Comparing fits 2 and 4 shows that they are SOLAR RADIATION [Wme2]
not very sensitive to T,.Comparing fits 4 and 5 or 6
and 7 shows that they are not sensitive to B, because Fig. 2. Swimming pool of site 2. A: Measured versus computed
of the low observed values of wind speed (less than 1 night losses. See eqn ( I ), T, as in Table 2; B: Passive solar
gains versus average solar radiation on a horizontal plane for
m s-’ average in all sites but site 4). Considering full
full day periods (24 h). Passive gains are evaluated according
day periods instead of nights only, confirms the pre- to eqn ( I ). T, as in Table 2.
vious conclusions and gives additional information
about 01~.A value of ap around 0.80 is coherent (see
$ 3.1). ol, = 1.00 as in fit 11 is not plausible. Similar
and compatible conclusions can be drawn from the mean gains and losses were computed over the period
fits of site 2. of measurements and are illustrated in Fig. 3 for sites
Uncertainties are larger for site 3 than for sites 1 I, 2, and 3. The average solar, temperature, and wind
and 2. Nevertheless, the same conclusions apply. Fit conditions encountered in each site over the period of
22 is in favour of ASHRAE for the value of A, but the measurements are also given in Fig. 3.
corresponding values of both T,and LYE are not plau- All gains and losses are expressed in energy flux per
sible. Fits 23 and 24, corresponding to realistic con- unit area of pool surface. Gains and losses do not
ditions again show that the value of A from Australian strictly balance each other due to the non-free param-
standards has to be preferred to the ASHRAE value. eters used in the linear regressions (Table 4). The cor-
As an illustration of fit 13, in Fig. 2A we present responding differences can be considered negligible
the measured losses versus the computed losses for site compared to the overall energy budget and the uncer-
2 and for night periods, according to eqn ( 1) . Measured tainties.
losses correspond to the measured internal energy Figs. 3A and 3B refer to full day periods but different
variation. Computed losses also include the pump periods of the year. They exhibit comparable contri-
contribution. butions for the different losses and the passive gains.
For the same site and for full day periods, in Fig. Fig. 3C exhibits comparable losses but larger passive
2B we present the passive solar gains versus the solar gains balanced by a large increase in internal energy,
radiation for conditions similar to those of fit 18. Pas- as the energy balance refers to part of the day only.
sive gains are evaluated, according to eqn ( I ), where The most important energy exchanges are clearly re-
the active gains and the internal energy variation are lated to passive gains, radiation, and evaporation losses.
measured, where the pump contribution is evaluated Finally, the swimming pool energy balance analysis
and where the losses are computed. The linear relation can be summarized by confirming that all data fit sat-
between passive gains and solar radiation is directly isfactorily with the following values of the involved
related to the pool absorption coefficient ((u,), as it parameters:
clearly appears in Fig. 2B. ?? swimming pool absorption coefficient: ap = 0.80 +
We also indicate in Table 4 the most coherent fits, 0.05,
i.e., the fits corresponding to the most coherent values ?? convection heat exchange coefficient (also involved
of the involved parameters. Using these values, the in the evaporation process):
26 B. MOLINEAUX, B. LACHAL. and 0. GtitS.Ah

4;29):88-9/17/88, / =18”C, (i=207Wm-’ j:I4188&7!10;88. /ras7ableL. (i=LlLWm


7;, =16,7”C,~=0,23m~‘, T” =21,8”C,&, =5.4”C 7, =149”C,v -0.46ms~‘.? :20.6”C,r,, =4 1°C
i’

l.OSSES (;AINS
208 Wm ’ 203 Wn~’ 22hWm J 127 Wm ’

X,18/88-lO/27,88,/; =15”C, c=394Wn1-


7;t =l88”C,V- 022m’,?; =24,8”C,I;,, :9.4”C ACTIVE SOI.AR

PIJMPS

PASSIVE SOLAR

ci l.R RADIATION

? ?EVAPORATION
??CONVECTION
WATER RENEWAL

1.0SSES ’ GAINS 1 INTERNAL ENERGY


417 Wm ’ 426 Wm.‘

Fig. 3. Overall energy balance in W mm’ for 3 swimming pools. 4: Site I. full days (24 h). B: Site 2. full
days (24 h). C: Site 3, day time periods (9 am to 4 pm), Assigned parameters: Convection heat exchange
coefficient, h, = 3.1 + 2.1~ and pool absorption coefficient alp = 0.80 in all cases: 7;: fresh water re-
newal temperature is also assigned. Average measured conditions: ci: global solar radiation on horizontal
plane; T,: ambient temperature: F: wind speed, T,: pool temperature. T,i; is the mean hourly modelled
sky temperature.

h, = A + Bv [W me2 K], REFERENCES

A = 3 t I [W mm2 KM’], ASHRAE Handbook, HVAC Systems and Applications


B = 2 r I [J mm3 Km ‘1 for v measured at - 10 m (1987).
above pool surface, or B = 4 +_2 [J mm3 K-’ ] J. A. Duffie, and W. A. Beckman, Solar engineering ef’therma:
for v measured at -.3 m above pool surface. processes, Wiley-Interscience. New York, 174 ( 199 I ).
D. Govaer, and Y. Zarmi, Analytical evaluation of direct solar
heating of swimming pools, So/u Energy. 27, 529-533
(1981).
4. CONCLUSIONS
P. Ineichen. J. M. Gremaud. 0. Guisan. and A. Mermoud.
Infrared sky radiation in Geneva. Solar Energy. 32, 537-
The main conclusions of this study can be sum- 545 (1984).
marized as follows: W. H. McAdams. Hmf rransm~.s.sion.3rd Edition, McGraw-
considering the lack of consensus on some funda- Hill, New York ( 1956), and W. Jurges, Gesundheit In-
genieur, 19 ( 1924).
mental parameters involved in the calculations and
B. Molineaux. B. Lachal, and 0. Guisan, Thermal analysts
the fact that we are dealing with an incomplete set of five unglazed solar collector systems for the heating ot
of data, the results are surprisingly coherent; outdoor swimming pools, Solar Ener~~~~. 53( I ), 27-32
the convection heat exchange coefficient (which is (1994).
also involved in evaporation) amounts to -3 W J. F. Sacadura, (coordinator), Initiation aux transfer& ther-
miques. Technique & Documentation, Lavoisier: Paris
m-* K-’ for windless conditions. The corresponding
(1982).
value of 5.7 W mm2 K-l, as proposed by the ASH- N. R. Sheridan, Solar heating systems for swimming pools.
RAE, does not fit our data; Australian standard A53634 ( 1989).
a more complete set of measurements would be use- N. R. Sheridan, The heating of swimming pools, University
of Queensland Research Notes 4 ( 1972 )
ful in the future in order to improve the current
R. Talwar, Solur energ.?:technology handbook B, Marcell
knowledge we have of the thermal exchanges in- Dekker, New York, 15-3 1 ( 1980).
volved in the energy balance of a solar heated swim- J. S. Touma, Dependance of wind profile law on stability for
ming pool. various locations, J. Air Poll. Cont. Assoc. 27(9), 863-
866 (1977).
J. H. Watmuff, W. W. S. Charters, and D. Proctor, Solar wind
Acknowledgements-This work was sponsored by the Swiss induced external coefficients for solar collectors, COM-
Federal Energy Office. PLES2, 56 (1977).

You might also like