Sei sulla pagina 1di 13

Solar Energy Vol. 68, No. 1, pp.

7789, 2000
1999 Published by Elsevier Science Ltd

Pergamon PII: S0038 092X( 99) 00054 7 All rights reserved. Printed in Great Britain
0038-092X/ 00/ $ - see front matter
www.elsevier.com/ locate/ solener
A CRITICAL REVIEW ON EQUATIONS EMPLOYED FOR THE
CALCULATION OF THE EVAPORATION RATE FROM FREE WATER
SURFACES

ERNANI SARTORI
Universidade Federal da Paraba, Av. Silvino Lopes 425/ 404, 58039-190 Joao Pessoa PB, Brazil
Received 2 December 1998; accepted 29 April 1999
Communicated by ERICH HAHNE
AbstractA critical review on several well-known empirical and theoretical equations that calculate the
evaporation rate from large free water surfaces has been carried out. Since up to now there had not been a
consensus on which equations were better to employ, a large scattering of evaporation rates has been found.
Several theoretical criteria have been undertaken, the scattering has been eliminated and a consensus has been
reached. It has also been shown that the consensual equations account for average roughness over the free
water surface length and tend to better predict the actual evaporation rate. Pan evaporation may generate
considerable differences in relation to a large water surface evaporation, depending on the surface length and
the rate of mass transfer. 1999 Published by Elsevier Science Ltd. All rights reserved.
1. INTRODUCTION can be attributed to the complexity of the evapo-
ration process.
Since the late 1800s when the rst empirical
Nevertheless, for solar applications where
investigations were published after Daltons
energy and economic requirements are involved,
(1802) work, almost a countless number of papers
an increasing accuracy of predictions of the
and equations on evaporation have been written to
evaporation rate is necessary. Thus, the present
date. Due to this profusion of equations, mostly
review is an attempt to discover among the
empirical, and the inexistence of a consensus, the
existing well known working formulas, basically
formulas have been taken from the literature and
utilized in the solar literature, those that more
indiscriminately employed without criteria to
accurately predict actual evaporation rates or, if
several applications around the world. Thus, a
possible, to reach a consensus on a single equa-
large scattering of evaporation rates has been
tion.
found and the initial results suggested that it
might be impossible to determine a generally
applicable equation for evaporation.
2. A PREAMBLE
Of the total heat loss from a free water surface,
much more than 50% is due solely to evaporation. The present work aims to take into account
Therefore, whether an equation severely overesti- only the evaporation due to forced convection
mates or underestimates the actual values, it over free water surfaces. For evaporation due to
generates an error of great proportions. In the case free convection, the reader is referred to corre-
of solar applications, such as swimming pools, sponding articles, e.g., Adams et al. (1990),
solar ponds, roof ponds, this would compromise Navon and Penn (1971), Sparrow et al. (1983).
the energy balances, too. As known, accuracy is This paper also does not consider the evaporation
mandatory for most processes, however, for em- from saline water bodies. For this area it is
pirical equations on evaporation this accuracy has recommended to consult the corresponding publi-
not been too rigorous. On the other hand, the cations, e.g., Adams and Harleman (1987), Calder
lesser number of theoretical working formulas in and Neal (1984), Salhotra et al. (1985), Sartori
relation to empirical ones was expected and this (1991). In Sartori (1991), for example, it is seen
that salinity alone (considering salinities up to
3.5%, the average sea water concentration) has a
minor effect on the evaporation rate, which

reduction is equivalent to the salt concentration


ISES Member. Fax: 155-83-226-3388; e-mail:
solar@openline.com.br value. However, the bottom reectance from a
77
78 E. Sartori
shallow evaporator due to salt precipitation can free water surface take place according to two
make its operation disastrous. mechanisms: the heat and mass transfer by the
Most of the equations employed so far for the molecular motion (diffusion) and the heat and
calculation of the evaporation rate from several mass transfer by the gross motion of the uid over
applications are empirical, i.e., those equations the water surface (advection). Near the water
resulting from regression analysis after a large surface, where the uid velocity is low the
number of experiments in order to get a more advection becomes negligible and the molecular
general validation. Nevertheless, these equations motion or diffusion is predominant being the sole
continue depending or being valid for only par- mechanism at the surfaceuid interface. In a
ticular systems and climates similar to those when very thin layer of air immediately above the water
the measurements were made. Empirical equa- surface, vapor is present which is regarded as
tions also depend strongly on the differences in being due to the action of molecular diffusion.
analysis techniques. Because of these conditions, With forced convection, the evaporation is caused
we could say that practically one empirical equa- by a combination of advection with diffusion,
tion is needed for each class of climate (humid, being the dominant component of the mechanism
windy, sunny, arid, day or night, etc.) existent for of heat and mass transfer generally made by the
each water surface. Due to the immense number bulk or gross motion of the uid.
of such specic equations, it is impracticable to Below, a critical review is made on papers and
consider all them, even for a review paper. Since corresponding equations, either empirical or theo-
the equations normally differ by the values of the retical, which have dealt with the evaporation rate
a and b coefcients, only those most common from free water surfaces.
in the literature are used for this analysis. How- According to Brutsaert (1965), Dalton (1802)
ever, the present paper can serve as a guide for started the empirical hydrodynamic approach to
comparisons with equations that have not been the evaporation problem. In 1801, in a lecture to
considered here. the Manchester Society, Dalton (who also de-
Although frequently cited in the literature, termined the law of partial pressures) stated that
some equations that have shown certain restric- evaporation is proportional to the difference in
tions or reported inaccuracies are not considered vapor pressure at the surface of the water and in
in this paper. Bruin (1978), for example, after the air and that the velocity of the wind affects
experimental tests concluded that the Priestley this proportionality. Subsequently, numerous re-
and Taylor (1972) equation concept is basically searchers started to investigate evaporation based
empirical and that their equation has been tested on Daltons description. Among others, Stelling
mainly in summer and only at high latitudes (1882), Fitzgerald (1886), Carpenter (1889),
(46608 N). Still according to Bruin, their re- Meyer (1915), Rohwer (1931), Penman (1948),
ported parameter a 51.26 is certainly not con- Marciano and Harbeck (1952), proposed several
stant even on an hourly basis and caution should equations on the basis of large amounts of ex-
be taken because of the sensitivity in its varia- perimental data, which were all of the general
tions. This is conrmed by the Sene et al. (1991) form:
measurements, too.
q 5(a 1bV) (P 2P )
ew w a
Also well quoted in the literature, Harbeck
(1962) proposes an empirical equation that de-
In this paper we are mainly interested in the
pends not only on V but also on the area of the
evaporation rate, which can be obtained by:
lake. Even with this procedure the standard error
of his equation is reported to be 16% (Sweers,
~ m5q /h 5h (P 2P ) /h (1)
ew w e w a w
1976). Moreover, according to Ryan et al. (1974)
or
this approach slightly improved the Marciano and
Harbeck (1952) formula, since the predicted
~ m5q /h 5h (P 2fP ) /h (1a)
ew w e w a w
evaporation rate was signicantly low and the
actual heat loss from Lake Hefner was consistent-
Bowen (1926), in his classical work, for pre-
ly underestimated. Brutsaert and Yu (1968) also
dicting losses from water surfaces eliminated the
included the area of the lake in their equations.
need for huge and laborious measurements of the
evaporation and wind velocity in relating the
convective to the evaporative heat transfer rates
3. THE EQUATIONS AND THEIR USES
through the following equation:
In a very succinct description, the heat and
mass transfer processes by evaporation from a q /q 5[B(t 2t ) /(P 2P )]P/ 760
cw ew w a w a
A critical review on equations employed for the calculation of the evaporation rate from free water surfaces 79
This equation was derived considering an anal- is the Australian Standards equation. de Winter
ysis on a differential control volume of a uid (1978) has also used Eq. (5) but the h and the
c
element (wind) blowing over a lake. For such vapor partial pressures are from a graph, where
analysis, he assumed that the velocity of horizon- the relative humidity is considered. Govind and
tal diffusion is negligible in comparison with the Sodha (1983) employed Eq. (5) but the corre-
wind velocity measured parallel to the main wind sponding coefcient is equal to 0.013, the relative
direction over the lake. In his paper, Bowen also humidity is considered and the h is obtained
c
showed that the heat losses by convection and by from Eq. (9).
evaporation follow the same laws. From Szeicz and McMonagle (1983) the mass
Sutton (1934), based on the theory of momen- of water evaporated from a swimming pool can be
tum and mass transfer in a turbulent medium, obtained by the formula:
deduced the law of variation of wind with the
~ m5(r C /g )(P 2P ) /(r )(h ) (7)
a p w a a w
height above the earths surface. This theory is
extended to the problem of evaporation, on the
where
assumption that the diffusion of mass follows the
2
r 5[ln (z/z )] /kV
same laws as the diffusion of momentum. A main
a o
expression found from this analysis shows the rate
Calder and Neal (1984) utilized Eq. (7) for a
of evaporation as being proportional to the 0.78
saline lake, with slight modications due to the
power of the wind velocity. This equation pre-
salinity.
sented good agreement with experimental data.
Kishore and Joshi (1984) employed an equa-
This power of 0.78 is very close to the 0.8 one
tion, taken from Perry and Chilton (1973), to
from the traditional turbulent convection theory
calculate the heat loss by evaporation from a solar
which has been applied to the evaporation process
~ pond where m is:
(Sartori, 1989).
Sweers (1976) did a review on empirical ~ m5[(h M ) /(C PM )](P 2fP ) (8)
c w w a w a
equations for lakes and found that there is a large
where
scattering of results, particularly those prior to
1958. He also shows that such variability can be
h 55.7 13.8V (9)
c
reduced by converting the measured wind speeds
Yadav and Rao (1983) utilized Eqs. (8) and (9)
to a common height and also by converting the
for estimating the evaporation from a roof pond,
different water surface areas to a common size.
but did not include the relative humidity.
Even with these adjustments, many equations still
Sartori (1987) modied Bowens (1926) equa-
showed poor agreement. After doing an extensive
tion resulting in BowenSartoris equation:
review, Sweers concludes that McMillans equa-
tion is the most reliable. Since in Sweers paper
q /q 5[B(t 2t ) /(P 2P )]P/ 760
cw ew w a w d
there are many empirical equations, which are
also of interest for the present review, some of where P 5fP . This modication allowed the
d a
them are listed below, in SI units: simulation of the three cases of mass ow that can
occur when a free water surface is exposed to the
WMO (1966) used in the USSR:
air. That is, we can now calculate the evaporation
h 50.0369 10.0266V (2)
when t .t and when t .t as well as the e
w a a w
condensation of the air (dew) on the water surface
WMO (1966) used in the USA:
when t .t . The latter had never been simulated
d w
before. These three situations cannot be correctly
h 50.0372V (3)
e
simulated using Bowens original equation, ex-
cept the rst case, i.e., t .t but only if the air is
McMillan (1971): w a
saturated (f 51.0), which is not a common
h 50.0360 10.0250V (4)
e
condition in the atmosphere. Because of this
modication, Sartori (1987; 1996) also developed
Czarnecki (1978), based on Czarnecki (1963),
a graph where the relationships among the water
calculated the heat transfer by evaporation from
temperature, the wind velocity, the relative
~ swimming pools where m is obtained by:
humidity and the (t 2t ) difference inuencing
w a
evaporation are plotted. The theoretical model ~ m50.0163h (P 2P ) /h (5)
c w a w
showed close agreement with the experimental
and
results and all of the above three cases were
h 53.1 14.1V (6) experimentally observed on a pan of one square
c
80 E. Sartori
meter. The resulting equations for the convective sented in Section 4), it is known that the applica-
and evaporative heat transfer rates are propor- tion of such an equation will behave poorly
tional to the 0.5 power of the wind velocity, producing inaccurate and low values of the evapo-
according to the laminar boundary layer theory. ration rates. Their equation that allows the obtain-
~ In Sartori (1989), the laminar analysis from ment of m and does not include the wind effect is
Sartori (1987) is extended for the turbulent ow given by
case. Both fully turbulent and transition regimes
~ m50.0144(P 2fP )[(t 2t )
w a w a
(mixed boundary layer) are considered in that
article. The wind velocity effect on evaporation 1(t 1273)(P 2fP ) /(268900
w w a
according to the ow regime is also shown. Such
1 / 3
2P )] / h (12)
w w
results are in agreement with the general convec-
tion theory where for low uid velocities the
Alagao et al. (1994), for calculating the mass
laminar ow prevails and vice-versa. For the
of fresh water evaporated from a solar pond used
current work, the following equation derived from
an equation in the form of Eq. (1a) and for the h
e
rst principles by Sartori (1989) and applied to
they adopted the Penmans (1948) empirical
the transition regime is chosen and presented in a
equation, which in SI units becomes:
different way:
h 50.074 10.040V (13) 0.8 20.2 21
e
~ m5(0.00407V L 20.01107L )
3(P 2P ) /P (10)
w d
Almanza and Lara (1994) utilized the Ryan and
Harlemans (1973) equations for estimating the
Since normally for a water surface there is a
heat losses by evaporation from a swimming pool.
certain laminar portion before the fully turbulent
Interesting to note that the four authors propose
ow is reached, as would be the case for a
two different equations for such calculations, one
swimming pool with no wind breaks, the equation
for natural temperature and the other when the
for the transition regime is here selected. When
water is heated by a boiler because several
transition occurs sufciently upstream of the rear
experimental eld tests have shown that free and
edge (x /L #0.95), the surface average coef-
forced convection are important for the evapora- cr
cients are inuenced by conditions in both the
tion, when the water is articially heated. Such
laminar and turbulent boundary layers (Incropera
considerations are completely misleading since
and Witt, 1985). For the fully turbulent ow, as
the free or forced convection are always im-
would be the case for lakes or when L 4x , the
portant, independent of whether the water is cr
21
term 0.01107L in Eq. (10) becomes negligible.
heated or not and of whether the heating is by
These elections are also substantiated by the
articial or natural means. In fact, those are only
conclusions from Wu (1969) who experimentally
empirical equations as a function of the tempera-
veried that the air ow near the water surface is
ture. It is also difcult to know whether the water
aerodynamically rough or in the transition region
is natural mainly when we consider it being
except at very low wind velocities (V,3 m/ s).
heated by the sun. However, according to the
From Taga et al. (1990) the empirical formula
authors, when the water is articially heated, the
in Eq. (11) may be used to predict
~ m can be mathematically got by:
1 / 3
~ m5(0.088403 10.001296V) (P 2fP ) /h ~ m5[0.027(t 2t ) 10.031V] (P 2P ) /h
w a w wv av w a w
(11) (14)
for a solar pond, where the expression within the where t and t are virtual water and air
wv av
rst parentheses is the Carrier (1949) equation. temperatures calculated through empirical equa-
Subhakar and Murthy (1993), to calculate the tions not given here. Almanza and Lara consid-
heat loss by evaporation from the free water ered the relative humidity in the computation of
surface of a solar pond submitted to forced the partial pressure while Oroud (1994) used Eq.
convection, employed Coopers (1973) equation (14) for the calculation of the evaporation rate
for solar stills, which is for free convection! from fresh water, but did not include the relative
However, for the convection heat transfer and for humidity.
this same wind ux and solar pond they utilized Hahne and Kubler (1994) monitored and simu-
Eq. (9), which is for forced convection! Thus, lated the thermal performance of two outdoor
even previously to our calculations (results pre- swimming pools at two different sites. To simu-
A critical review on equations employed for the calculation of the evaporation rate from free water surfaces 81
late the evaporative losses they tested different Sodha et al. (1981) and Ali (1989) employed Eq.
empirical correlations from the literature (e.g., (19) for solar ponds, but took into account the
Rohwer, 1931) as well as obtained another one relative humidity. According to ASHRAE (1991),
from their tted data. Although their a and b Eq. (19) is for an active pool and for other
coefcients were tted better for the pools than conditions such an expression should be reduced
those from the literature, the resulting solar by about 50%. Smith et al. (1994) conducted the
absorptance coefcient (a 50.56) from simula- experiments mostly at night in an inactive outdoor
w
tions disagreed greatly with the normal solar swimming pool, articially heated to 298C. After
absorptance (a .0.85) for swimming pools. One a linear t with their experimental results, they
w
adjustment effort created a disagreement. The adjusted the above coefcients, resulting in the
evaporation rate from their equation may have the empirical equation (in SI units):
form:
~ m5(0.0638 10.0669V) (P 2P ) /h (20)
w a w
~ m5(0.0850 10.0508V) (P 2fP ) /h (15)
w a w
where the expression within the rst parentheses
4. RESULTS AND DISCUSSION
corresponds to the Rohwers (1931) equation
The analyses that follow are based on simple (obtained from laboratory experiments with wind
comparisons among the equations in order to tunnel) or
determine if the evaporation rates accurately
~ m5(0.0803 10.0583V) (P 2fP ) /h (16)
w a w
respond to stimuli of water and air temperatures,
wind velocity and relative humidity. It is sup-
whose coefcients are the Hahne and Kublers
posed that the rate of evaporation occurs from a
best tted a and b parameters (with a 5
w
free water surface with temperatures ranging from
0.90).
10 to 408C and greater 58C in relation to each
Molineaux et al. (1994) evaluated evaporation
ambient air temperature, wind velocity of 3 m/ s
heat losses from ve outdoor swimming pools
and a relative humidity of 45% or 100% for each
~ where m can be given by:
air temperature. Thus, the comparisons are carried
~ m5(h /C )(X 2X ) (17)
c a n s
out for all equations under identical conditions of
water and air temperatures, relative humidities
assuming that the air ux gets saturated with
and wind speed.
humidity and where:
In order not to introduce a variable in the
X 5[P /(P 1P )](M /M )
(n or s) w w a w a
results, the water vapor partial pressures for the
present work have been calculated through a
For h they tested three known literature corre-
c
unique preliminary selected equation. Ten differ-
lations, i.e., Eqs. (6), (9) and one suggested by
ent equations for calculating this pressure have
Watmuff et al. (1977), as follows:
been found in the literature surveyed. The calcula-
tions were carried out for the temperature range
h 52.8 13.0V (18)
c
noted above and the results were compared. The
pressures from the equation by Keenan and Keyes
The authors converted the a and b co-
(1936) employed by Czarnecki (1978) and Sartori
efcients of the above equations to those adapted
(1987; 1989; 1991; 1996) presented the least
to their measurements of the wind speed 10 m
average in relation to all other equations results
above the pool. Nevertheless, they obtained most
and because of this the formula was chosen and
coherent a and b tted coefcients for their
applied to all calculations.
experimental data. Their tted correlation:
Another matter of concern is the height at
h 53.1 12.1V
which the wind velocities are measured. In the c
numerous works surveyed, it is observed that
is used in Eq. (17).
there is not a consensus on which elevation is the
Smith et al. (1994) experimentally investigated
best for measurements of the wind velocity.
the performance of the Carrier (1918) empirical
However, it has been observed that this height
equation, widely used by the ASHRAE Hand-
sometimes does not depend on personal choice
books, for predicting the evaporation rate from
but on the corresponding instruments already
swimming pools, which in SI units is represented
installed near the test site. Such measurement
by:
heights vary from 0.3 to 0.5 m, extending to 2.0
~ m5(0.0888 10.0783V) (P 2P ) /h (19) and 3.0 m and sometimes to 8.0 and 10 m above
w a w
82 E. Sartori
the water surface. The most common height found Sometimes, such an empirical equation is ap-
in the literature is 2.0 meters. In Molineaux et al. plied the relative humidity. This procedure, how-
(1994), a formula relates the wind speed to any ever, can cause additional errors: if the coef-
height above ground level. Sutton (1934) de- cients are originally high this modication will
veloped such a formula. produce still higher values of the evaporation
In this paper, however, such conversion is not rates. Even if the a and b parameters were
performed because frequently these heights are corrected for the original elevation, it is necessary
not published and, as discussed later, this should to also correct them for the new wind velocity and
not be considered the equations central issue, relative humidity ranges and combinations. We
since there are more relevant factors for determin- should remember that such coefcients depend
ing their performance. For these heights, the also on the statistical methods utilized for their
differences in the wind velocities are not suf- adjustments. Additionally, intentional alterations
ciently large to drastically change the results. In have been made in order to t the coefcients for
Sweers (1976), it can be seen that the variation of some particular situation, as is the case of Ryan
the coefcients is 5% between the heights of 2 m and Harleman (1973) who reduced the coef-
and 3 m and 12% between 3 and 8 m. Further- cients of the Lake Hefner by 18% in order to
more, these equations are frequently taken from improve agreement with measurements (Adams et
the literature and employed as they are, without al. 1990).
modications. Therefore, it is also intended to Because of all these variables and uncertainties,
verify the equations performances as they are normally it is necessary to make a new tted
normally applied. correlation for each particular system. This is the
The equations from Section 3 considered for same as saying that it is practically dispensable to
the present analysis can be divided in two groups: employ any original empirical equation for any
(a) does not take into account the relative humidi- other case without precise information on its
ty; (b) takes into account the relative humidity. development. Hahne and Kubler (1994), for ex-
We can immediately say that those of the class (a) ample, veried that correlations from literature
have inaccurately predicted the evaporation rate describing evaporative heat losses deliver results
from free water surfaces for humid, non-saturated, that differ by a factor of 22.5 in relation to the
air. In those equations that do not take into actual ones.
account the relative humidity implicitly it is Fig. 1 presents a simple sketch showing the
assumed to be equal to 1.0 or 100%. This, main parameters involved in a free water surface
however, is not a regular or common condition in considered for this paper. The curves from the
the atmosphere. Such equations use the partial above two groups of equations are presented in
pressure of the ambient temperature for a satu- Figs 2, 3 and 4. Fig. 2 shows the values of the
rated state, not taking into account its correction evaporation rate from equations which do not
due to the relative humidity. The lower the consider the relative humidity and then they
relative humidity, the greater the evaporation rate implicitly predict only the evaporation rate for the
becomes and for the same ambient temperature saturated state (f 51.0). Figs. 3 and 4 present the
there are different relative humidities and thus corresponding values for the equations that take
different partial pressures. into account the relative humidity, that is, con-
In the case of empirical equations, case (a) is sider the humid air condition (f ,1.0). In the
valid only when such correlations are applied
exclusively to situations close to the original
experimental conditions, since these correlations
have been attained for evaporation rates depen-
dent on the wind velocity, relative humidity and
on other psychrometric parameters prevailing at
the time and place of the measurements. That is,
many environmental factors inuenced the origi-
nal evaporation, but only the wind velocity was
considered as a variable. Moreover, the correla-
Fig. 1. A simple sketch showing the main parameters involved
tions were tied to the P . Hence, when an
a
in the evaporation from a free water surface. TA5air tempera-
empirical equation is taken out of its reference
ture; TW5water temperature (surface); L5characteristic
conditions to be used at other place and situation,
length in the wind direction; V5wind velocity; f5relative
the equation tends to become invalid. humidity.
A critical review on equations employed for the calculation of the evaporation rate from free water surfaces 83
Fig. 4. Evaporation rate for low relative humidity (RH545%)
Fig. 2. Evaporation rate for the saturated state (relative
and for high relative humidity (saturated state) according to
humidity5100%) according to water temperatures. Equation
water temperatures. Equation numbers are listed at the curves:
numbers are listed at the curves: 25WMO for the USSR;
85KishoreJoshi; 105Sartori; 115Carrier (1949); 125
35WMO for the USA; 45McMillan; 55Czarnecki; 65
SubhakarMurthy; 185Watmuff et al.
Australian standards; 75SzeiczMonagle; 8*5YadavRao;
95from DufeBeckman; 195Carrier (1918); 205Smith et
al.
the humid air. In addition, the condition of t 5t
w a
exists. For equations that consider only the P or
a
latter category some equations are included even do not take into account the relative humidity, the
if the original authors (e.g., Carrier, Penman, evaporation rate is zero for this condition. How-
Rohwer) did not consider the relative humidity ever, this is wrong, since in reality evaporation
but other authors inserted it, as seen in Section 3. exists even when t 5t . Performing simple
w a
For the sake of comparisons and tests, Eqs. (6) calculations with any of the above equations, we
and (9) have been applied in the form of Eq. (1), can see that the rate of evaporation for t 5t 5
w a
included in Fig. 2 and treated as independent 258C and f 50.45, is about the same as that for
equations. The same has been done for the Eqs. t 5408C, t 5358C and f51.0, which is not
w a
(9) and (18), but they have been used in the form negligible. The evaporation is zero only when
of Eq. (1a) and included in Figs. 3 and 4, t 5t and f5100%.
w a
respectively. Of particular interest is the equation employed
The equations that predict the evaporation rate by the WMO/ US, which does not contain the a
only for the saturated state may generate large coefcient. Empirical investigators (e.g., Brut-
differences in relation to the evaporation rate for saert, 1965; Sweers, 1976; Sene et al., 1991)
consider this parameter as the parcel to take into
account the evaporation when V50 (which is
inaccurate). However, its exclusion by the WMO/
US doubles the evaporation when the wind ve-
locity is duplicated. This is not accurate, since the
evaporation is not related in such a way with the
wind velocity (e.g., Hahne and Kubler, 1994;
Sartori, 1987; 1996; Smith et al., 1994).
The above arguments are sufcient to discard
all equations that do not consider the relative
humidity, however, instead of simply rejecting all
of them, lets be more prudent. Observing Fig. 2,
we can see the results from Eq. (9), which, as
suggested by Watmuff et al. (1977) and by Dufe
and Beckman (1980), includes the effects of
Fig. 3. Evaporation rate for low relative humidity (RH545%)
radiation. Then, it is noted that Eqs (5), (7), (19)
and for high relative humidity (saturated state) according to
and (20) overestimate the evaporation for the
water temperatures. Equation numbers are listed at the curves:
saturated state. This same result would occur if
95from DufeBeckman; 135Penman; 145RyanHar-
these equations had been applied to the relative
leman; 155Rohwer; 165HahneKubler; 175Molineaux et
al. humidity.
84 E. Sartori
The predictions of empirical equations that Yadav-Rao results are a little bit higher than the
consider solely the case for f51.0 are only trials double of Subhakar and Murthys values but are
for intermediate values of evaporation rates gener- comparable among them and still very low,
ated by low or high relative humidities. Their producing an efciency of only 25%. All their
coefcients obviously having been adjusted for efciencies would be still lower if other h
c
the actual conditions prevailing during the experi- correlation had been used instead of the super
ments, i.e., probably for f ,1.0. The usual dimensioned Eq. (9).
period for computing the evaporation rate by In Fig. 3, it is veried that Eq. (17) delivers
empirical investigators is one week or one month. slightly lower results than those of Eq. (9). This
In this case, the calculations are rough estimates also tends to the overprediction of the evaporation
of the actual evaporation, with the accuracy rate. In order to conrm this observation lets get
depending on uctuations of the atmospheric another insight on the equations performance.
parameters. Table 1 represents a simulation of the evaporation
Thus, as before, in Fig. 3 we can see that Eqs. from a free water surface during 24 hours a day.
(13), (15) and (16) when compared to the results This evaporation is supposed to occur during 10 h
from Eq. (9) also overpredict the evaporation rate. of daylight with t 5308C, t 5258C, V53 m/ s
w a
Supplementing this reasoning, lets consider the and f50.45 plus 14 hours of the night with
evaporation rate obtained using Rohwers (1931) t 5258C, t 5208C, V52 m/ s and f51.0. These
w a
Eq. (15), for 308C, f545%, as an example. Such data correspond to low and high relative
26 22 21
value corresponds to 275310 kg m s (0.99 humidities, which together with the other parame-
22 21
kg m h ), which is an excessive evaporation ters represent approximate values of conditions
rate. This number equals a loss of heat of 669 that may generally occur. The evaporation rate in
22 22 22 21
W m . If we consider G5700 W m , this heat Table 1 is given in kg m day and not in
22 21
~ loss would represent an efciency (h 5mh /G) kg m s because for this very particular case
w
of 96% of the total, only for evaporation! If we only the daily evaporation has meaning for the
take now t 5358C, f545%, the heat lost by comparisons intended here.
w
evaporation due to Eq. (15) is equivalent to 881 From the previous ponderings, we now have
22
W m ! Corroborating to the general non-valida- Eqs. (10), (11), (14), (17) and (18) that take into
tion of the Rohwers equation, Ryan et al. (1974) account the relative humidity. Although the equa-
mention that experiments by Rohwer (1931) tions that do not consider the relative humidity
were generally performed in the laminar range have erroneously predicted the evaporation rate
and are thus not applicable to eld cases. for humid air, Eqs. (2) and (4) presented close
On the other hand, we can verify in Fig. 4 the agreement for the saturated state with equations
curves from Eqs. (8) and (12). As commented in from this article that up to the moment deserve
Section 3, Subhakar and Murthys equation, Eq. consideration. In this same way, Eq. (6) although
(12), was expected to produce low values of the giving slightly lower results than those from Eq.
evaporation rate. The results conrmed these (9), may overestimate the evaporation rate and
expectations. Their results are really too low. will be also included in the above list. Thus, in
22
Considering again G5700 W m , the evapora- order to know whether Eqs. (2), (4) and (6) are
tion for t 5308C would correspond to an ef- denitively valid or not, they will be employed in
w
ciency of only 12%. Both the Kishore-Joshi and the form of Eq. (1a). Hence, Eqs. (2), (4), (6),
a
Table 1. Daily evaporation rates and efciencies of evaporation for several equations
AS CARR MCMIL MOLIN RYAN SART WATM WMO
(6) (11) (4) (17) (14) (10) (18) (2)
Day 6.42 3.85 4.63 6.19 5.87 4.83 4.93 4.87
30258C
f50.45
Night 1.93 1.56 1.47 1.97 1.90 1.38 1.50 1.54
25208C
f51.0
Total 8.35 5.41 6.10 8.16 7.77 6.21 6.43 6.41
22 21
(kg m d )
Efciency 81 52 59 79 75 60 62 62
(%)
a
Equation numbers are listed below the authors names
A critical review on equations employed for the calculation of the evaporation rate from free water surfaces 85
(10), (11), (14), (17) and (18) are included in Therefore, we have now Eqs. (2), (4), (10),
Table 1. (11) and (18) to consider for analyses. From these
As can be seen from Table 1, our previous expressions, Eq. (11) should be examined separ-
suspicions that Eqs. (6) and (17) overestimate the ately. Average efciencies around 60% are ex-
evaporation rate are conrmed now. We can see pected for evaporation processes and thus we can
that these equations produce very high values for see that Eq. (11) produces the lowest efciency
the daily evaporation and much greater than is and evaporation rate from the above group of
expected for a day with those conditions. From equations. Although having nothing too serious
the experience and intuition, daily evaporation for its rejection at the moment, it seems that the
22 21
higher than 7.0 kg m day is not common and results of the Carriers (1949) equation employed
not expected with such atmospheric conditions. as is by Taga et al. (1990) are low and produce
Sartori (1987), for example, experimentally ob- lower rates of evaporation than the actual ones.
tained average water temperatures of 308C (day), Finally we consider the equations by the WMO
22 21
228C (night) and 6.1 kg m day using a real (1966) for the USSR, McMillan (1971), Watmuff
shallow pan of one square meter, very well et al. (1977) and Sartori (1989). In order to
thermally insulated, with the pan bottom painted visualize the close agreement among these four
black, operating 24 hours a day with a daily equations, their results are issued in Fig. 5. As can
22
average solar radiation of 670 W m and an be seen, the scattering seen in Figs. 24 has been
average relative humidity of 76%. eliminated, showing that the results from these
It is also observed that the RyanHarlemans four equations coincide. As noted, this scattering
equation (Ryan and Harleman, 1973), Eq. (14), suppression has not been achieved by visual
generates very high values of the evaporation rate. selection but rather based on theoretical consid-
By the way, Adams et al. (1990) show that the erations. McMillans equation, which has been
Ryan and Harleman (1973) equation tends to dened for the elevation of 3 m, has generated
overpredict evaporation by about 1525%. Still slightly lower values than the other equations.
according to Adams et al. (1990), most re- Such difference is of the order of 5% or less. For
searchers have found that pure addition of forced the sake of verication, when converting its
and free convection overpredicts evaporation, as coefcients to the height of 2 m, the results
is the case of the Ryan and Harleman (1973) become very close to the other ones. This also
equation. Normally, a comparison between the suggests that the wind speed for the other equa-
Reynolds and Grashoff numbers to decide tions work well for the height of 2 m. We also
whether forced or free convection should be used believe that the best elevations for wind speed
is adopted. Applying a 20% reduction on the 7.77 measurements for evaporation purposes is from
22 21
kg m day from the RyanHarlemans equa- 0.3 m up to 2.0 m above the water surface. These
22 21
tion, we will get the exact 6.2 kg m day four equations have also been tested for distinct
obtained by Sartoris equation, Eq. (10), for wind velocities and the agreement among them is
example. It is also veried from Table 1 that Eqs. similar.
(6) and (17) yield higher evaporation rates than Interesting to note that the Watmuff et al.
the RyanHarlemans equation.
Penmans (1948) equation, Eq. (13), with the
conditions of Table 1, would produce the absurd
22 21
value of 10.50 kg m day with an incredible
22
efciency of 102% (with G5700 W m )! How-
ever, since Penmans and all other empirical
equations originally did not consider the relative
humidity we can discuss a constant f51.0 for
day and night. Its total for the 24-h period is then
22 21
5.73 kg m day . This apparently acceptable
value conrms that many of the empirical co-
efcients are high only to (involuntary) compen-
sate for the absence of the relative humiditys
inuence in the equations. In a dry region or
Fig. 5. Evaporation rate according to water temperatures after
season, Penmans equation, for example, would
the scattering suppression. Upper curves: low relative humidi-
generate extremely erroneous evaporation rates
ty; lower curves: high relative humidity (saturated state).
with these same temperatures and wind velocity. Authors names are listed at the curves.
86 E. Sartori
equation was not developed from an evaporation transition regime. The results from these laminar
investigation, but from a convection heat transfer analyses are always extrapolated to the entire
problem. In this selection, we have one theoretical water surface, but in reality they remain valid
and three empirical expressions. An empirical only up to few meters in length from the front
correlation is based upon experimental data edge. By analogy to the convection heat transfer
whereas a theoretical equation is based upon the theory, the characteristic length L seen in Sartoris
rst principles of the process. A theoretical equa- equation corresponds to an attenuation of the
tion may contain empirical parameters, but the mass transfer coefcient in the direction of uid
derivation is based upon a theoretical analysis ow. The length L does not mean local values but
which resulting expressions produce a more gen- it generates average values of the evaporation rate
eral validation and accuracy. Empirical equations, for the entire surface length of unit width.
on the other hand, are simple representations of The evaporation rates for different lengths are
the instantaneous effects of physical processes. distinct from each other. This does not happen,
Both methods may eventually originate close however, for the equations whose evaporation
values. rates are considered equal for any length. This can
Despite this excellent agreement, there is still be easily visualized in Fig. 6 where the evapora-
another step to be examined. It has long been tion rates as a function of the surface length are
customary to estimate evaporation from lakes and plotted. The curves are for t 5308C and V53
w
to obtain empirical equations on the basis of m/ s, the upper ones for f50.45 and the lower
observations from pans, generally the US Weather ones for f51.0. We can see in Fig. 6 that the
Bureau circular class A pan (1.2230.254 m), evaporation rate from Sartoris equation decreases
placed in the vicinity. According to Cummings as the length increases, while the results from the
and Richardson (1927), there has no means of other equations remain constant over the entire
showing conclusively that the error thus involved surface length. This constancy simply multiplies
is not excessive. Moreover, according to Brut- the evaporation rate of the rst meter by the total
saert and Yeh (1970) it is impossible to predict length of the water surface. However, this does
lake evaporation from pan evaporation and unless not happen in reality, since there must be a decay
measurements of a number of additional physical of the mass transfer along the distance of the main
factors are made, pan data are useless. Obvious- wind direction. According to the heat and mass
ly, the ow over a pan is laminar while the ow transfer coefcient theory, this decay is propor-
20.2 20.5
over a large water surface is turbulent or in the tional to L for the turbulent ow and to L
transition regime. for the laminar ow. Thus, Sartoris equation, Eq.
The three empirical equations here selected (10), tends to better represent the total evapora-
were developed in laminar ow while the Sartori tion over a free water surface in turbulent ow.
(1989/ present) equation, Eq. (10), has been de- An alternative equation developed during the
veloped and is valid for the turbulent ow or the course of this analysis and based on the Sherwood
Fig. 6. Evaporation rate according to water surface lengths. Upper curves: low relative humidity; lower curves: high relative
humidity (saturated state). Authors names are listed at the curves. McMillans equation converted for the height of 2 m.
A critical review on equations employed for the calculation of the evaporation rate from free water surfaces 87
and Schmidt numbers generates close agreement literature and directly employed to particular
applications. Since up to now there was not a with the results from Eq. (10) and is represented
consensus on which equations were better to in equivalent way in Fig. 6 and by the formula:
employ, a large scattering of evaporation rates has
0.8 20.2 21
~ m5(0.00562V L 0.01529L ) (r 2r )
resulted.
w d
It has been shown that the equations that do not
(21)
take into account the relative humidity inaccurate-
where r 5fr . Wishing to include P as a ly predict the evaporation for the humid air, and
d a
variable, the right hand side of Eq. (21) needs the differences may be large in relation to the
only be multiplied by the standard atmospheric saturated state. It has also been demonstrated that
pressure and then divided by the desired one. the simple inclusion of the relative humidity in
equations does not ensure the accuracy of results. Observing Fig. 6, we can also see that for high
Several procedures, step-by-step, based on theo- relative humidity the decay is lower than that for
retical considerations, have been undertaken in the low one, showing that the decline of the
order to arrive to a consensus and to a harmoniza-
evaporation according to the length is directly
tion among many equations from various elds of
proportional to the mass transfer. By analogy, the
study of the evaporation from free water surfaces.
corresponding decay is directly proportional to the
The scattering has thus been eliminated and the
heat transfer variation over a at plate.
equations developed by McMillan (1971), Wat-
Moreover, since on large water surfaces waves
muff et al. (1977) and Sartori (1989) and that
exist that evidently inuence the rate of evapora-
utilized by the WMO (1966) for the USSR have
tion, Brutsaert and Yeh (1970) based on their
shown effectively coincident values.
studies reported that it is not inconceivable that
Since the empirical formulas by the WMO for
evaporation decreases with increasing rough-
the USSR, McMillan and Watmuff et al. were
ness. They also mention the Easterbrooks
developed in laminar ow, they are valid only up
(1968) work who experimentally observed both in
to few meters from the leading edge of the water
laboratory and on the Lake Hefner that for
surface and these values cannot be extrapolated to
certain combinations of wind speed and well-
the entire water surface length. Sartori (1989)
developed waves, the evaporation rate can be
theoretical equation, presented in the current
smaller than that measured under similar wind
paper in a different way, was developed for the
speed conditions with no waves present. Easter-
turbulent ow and is thus applicable to eld cases,
brook attributed this to the presence of dead or
like for swimming pools, solar ponds, lakes, etc.
trapped air regions to the lee of the wave crests,
Moreover, there must be a decay of the mass
which form an effective barrier to the vertical
transfer along the distance of the main wind
transport of water vapor. This decrease due to
20.2
direction, which is proportional to L for the
waves also contributes to the pan evaporation
turbulent ow. Since the Sartori (1989/ present)
being useless in relation to a large water surface
equation includes such decrease as a variable, it
evaporation.
tends to more accurately represent the actual
Although the wave inuence on evaporation
evaporation over the total free water surface
should remain a matter of intensive investigation,
length submitted to turbulent ow.
the equations that include the referred decay may
In the course of this paper, Sartori developed
account for average roughness over the water
another equation based on the Sherwood and
surface length and tend to predict the evaporation
Schmidt numbers, which includes the referred
rate in these conditions better than those which do
decay and showed close agreement with the
not take into account such decay.
Sartori (1989/ present) equation. The evaporation
rate from a water surface with well-developed
waves may be smaller than that measured under 5. CONCLUSIONS
similar wind speed conditions with no waves
In this paper, a critical review on several well-
present. Thus, the equations that include the
known equations employed for the calculation of
mentioned decay, i.e., the Sartoris equations:
the solar evaporation rate from free water surfaces
0.8 20.2 21
has been carried out. Both empirical and theoret-
~ m5(0.00407V L 20.01107L ) (P
w
ical working formulas have been analyzed. It has
2P ) /P
d
been veried that numerous equations, mainly
empirical, have normally been taken from the and
88 E. Sartori
0.8 20.2 21
Almanza R. and Lara J. (1994) Energy requirements for a
~ m5(0.00562V L 20.01529L ) (r 2r )
w d
swimming pool through a wateratmosphere energy bal-
ance. Solar Energy 5321, 3739.
account for average roughness over the total free
ASHRAE (1991). In HVAC Applications Handbook, pp. 47,
water surface length submitted to turbulent ow ASHRAE, Atlanta, GA, USA.
Bowen I. S. (1926) The ratio of heat losses by conduction and
and tend to predict the evaporation in these
by evaporation from any water surface. Phys. Rev. 27,
conditions better than those which do not include
779787.
Bruin H. A. R. (1978) A simple model for shallow lake such decay. For pan evaporation, both the lami-
evaporation. J. Appl. Meteorol. 17, 11321134.
nar / turbulent issue and the absence of waves
Brutsaert W. (1965) Equations for vapor ux as a fully
contribute to considerable differences in the rela-
turbulent diffusion process under diabatic conditions. Bull.
Intern. Assoc. Sci. Hydrol. 10, 1121.
tion to a large water surface evaporation, depend-
Brutsaert W. and Yeh G. T. (1970) Implications of a type of
ing on the surface length and the rate of mass
empirical evaporation formula for lakes and pans. Water
transfer. Resour. Res. 64, 12021208.
Brutsaert W. and Yu S. L. (1968) Mass transfer aspects of pan
evaporation. J. Appl. Meteorol. 74, 563566.
Calder I. R. and Neal C. (1984) Evaporation from saline lakes:
NOMENCLATURE
a combination equation approach. Hydrol. Sci. J. 291,
8997.
Carpenter L. G. (1889) Section of meteorology and irrigation a, b Coefcients of empirical equations
engineering. In Second Annual Report Agriculture Ex-
B Bowen coefcient
21 21
perimental Station, pp. 4976, State Agriculture College,
C , C Specic heat of air and water (J kg K )
a w
Fort Collins, USA. 22
G Solar radiation (W m )
Carrier W. H. (1918) The temperature of evaporation. ASHVE
22 21
h Convective heat transfer coefcient (W m K )
c
Transactions 24, 25.
h Evaporative heat transfer coefcient
e Carrier W. H. (1949). In Fan Engineering, 5th ed, p. 95,
22 21
(W m Pa )
Buffalo Forge Co., Buffalo, NY.
21
h Latent heat of vaporization of water (J kg ) Chemical Engineers Handbook, (1973). 5th ed., Perry R. and
w
Chilton C. H (Eds.), pp. 1222, McGraw-Hill, New York, L Characteristic length (m)
22 21
USA.
~ m Evaporation rate (kg m s )
Cooper P. I. (1973) The maximum efciency of single-effect
M , M Molecular weight of the air and water
a w
solar stills. Solar Energy 15, 205217.
P Atmospheric pressure (Pa)
Cummings N. W. and Richardson B. (1927) Evaporation from
P , P , P Water vapor partial pressure at the air, dew point
a d w
lakes. Phys. Rev. 30, 527534.
and water temperatures (Pa)
Czarnecki J. T. (1963) A method of heating swimming pools
q , q Heat uxes by convection and evaporation
cw ew by solar energy. Solar Energy 7, 37.
22
(W m )
Czarnecki J. T. (1978) Swimming pool heating by solar
r Aerodynamic resistance to the diffusion of water energy. In Technical Report No. Tr 19, CSIRO, Victoria,
a
21
Australia. vapor (s m )
Dalton J. (1802) Experimental essays on the constitution of
t , t , t Air, dew point and water temperatures (8C)
a d w
21
mixed gases; on the force of steam or vapor from water and
V Wind velocity (m s )
other liquids in different temperatures, both in a Torricellian
x Critical length (m)
cr
vacuum and in air; on evaporation and on the expansion of
X Humidity content of humid air at ambient tem-
n
gases by heat. Mem. Manchester Liter. and Phil. Soc. 511,
21
perature (kg kg )
535602.
X Humidity content for saturated air at pool tem-
s Dufe J. A. and Beckman W. A. (1980). In Solar Engineering
21
perature (kg kg )
of Thermal Processes, p. 137, Wiley-Interscience, New
z Height above ground level (m) York, USA.
Easterbrook C. C. (1968) Project evaporation: a study of the z Roughness length for water (m)
o
effects of waves on evaporation from free water surfaces. In
f Relative humidity
21
Cornell Aeronaut. Lab. Rep. Rm 2151p1, Cornell, Buffalo,
g Psychrometric constant566 Pa K
New York.
h Evaporation efciency, %
Fitzgerald D. (1886) Evaporation. Trans. ASCE 15, 581646.
k von Karmans constant50.41
Govind M. and Sodha M. S. (1983) Thermal model of solar
r , r , r Density of water vapor at the air, dew point and
a d w
swimming pools. Energy Conv. Mgmt 23/ 3, 171175.
23
water temperatures (kg m )
Hahne E. and Kubler R. (1994) Monitoring and simulation of
the thermal performance of solar heated outdoor swimming
pools. Solar Energy 531, 919.
Harbeck G. and (E. (1962). In A Practical Field Technique
REFERENCES
For Measuring Reservoir Evaporation Utilizing Mass-trans-
Adams E. E. and Harleman D. R. F. (1987) The alpha, beta, fer Theory, pp. 272E, US Geology Survey .
gamma of evaporation from saline water bodies. Water Incropera F. and Witt P. D. D. (1985). Fundamentals of Heat
Resoure Res. 2329, 17691774. and Mass Transfer, John Wiley and Sons, New York, USA.
Adams E. E., Cosler D. J. and Helfrich K. R. (1990) Keenan J. and Keyes H. F. (1936). In Thermodynamic
Evaporation from heated water bodies: predicting combined Properties of Steam, p. 14, John Wiley and Sons, New York,
forced plus free convection. Water Resour. Res. 2623, USA.
425435. Kishore V. V. N. and Joshi V. (1984) A practical collector
Alagao F. B., Akbarzadeh A. and Johnson P. W. (1994) The efciency equation for non-convecting solar ponds. Solar
design, construction and initial operation of a closed-cycle, Energy 335, 391395.
salt-gradient solar pond. Solar Energy 5324, 343351. Marciano C. J. and Harbeck G. E. Mass transfer studies,
Ali H. M. (1989) Potential of solar ponds in hot climates. water-loss investigations, Lake Hefner studies. In Tech.
Solar & Wind Technology 622, 137141. Rep. 229, Vol. 1, pp. 4670, US Dept. Int. Geol. Survey.
A critical review on equations employed for the calculation of the evaporation rate from free water surfaces 89
McMillan W. (1971). Heat dispersalLake Trawsfynydd cool- Sene K. J., Gash J. H. C. and McNeil D. D. (1991) Evapora-
ing studies. Symp. on Freshwater Biology and Electrical tion from a tropical lake: comparison of theory with direct
Power Generation, Part 1, 4180. measurements. J. Hydrol. 127, 193217.
Meyer A. F. (1915). Trans. Am. Soc. Civil Eng. 79, 1056. Smith C. C., Lof G. and Jones R. (1994) Measurement and
Molineaux B., Lachal B. and Guisan O. (1994) Thermal analysis of evaporation from an inactive outdoor swimming
analysis of ve outdoor swimming pools heated by unglazed pool. Solar Energy 531, 37.
solar collectors. Solar Energy 531, 2126. Sodha M. S., Kaushik N. D. and Rao S. K. (1981) Thermal
Navon U. and Penn J. B. (1971) Interfacial mass and heat analysis of three zone solar pond. Energy Res. 5, 321340.
transfer during evaporation. AIChE J. 171, 131136. Sparrow E. M., Kratz G. K. and Schuerger M. J. (1983)
Oroud I. M. (1994) Evaluation of saturation vapor pressure Evaporation of water from a horizontal surface by natural
over hypersaline water bodies at the southern edge of the convection. J. Heat Transfer, ASME Trans. 105, 469475.
Dead Sea Jordan. Solar Energy 536, 497503. Stelling (1882) In St. Petersburg, Russia. Stellings complete
Penman H. L. (1948) Natural evaporation from open water, reference has not been got but its name and year are kept
bare soil and grass. Proc. Royal Soc. Lond. A193, 120145. here with the aim at preserving the historical data.
Priestley C. H. B. and Taylor R. J. (1972) On the assessment Subhakar D. and Murthy S. S. (1993) Saturated solar ponds: 1.
of surface heat ux and evaporation using large-scale Simulation procedure. Solar Energy 503, 275282.
parameters. Mon. Wea. Rev. 1002, 8192. Sutton O. G. (1934) Wind structure and evaporation in a
Rohwer C. (1931) Evaporation from free water surfaces. In turbulent atmosphere. Proc. Royal Soc. Lond. A146, 701
U.S. Dept. Agric., Tech. Bull, Vol. 271, p. 96. 722.
Ryan P. and Harleman J. D. (1973) An analytical and Sweers H. E. (1976) A nomogram to estimate the heat-
experimental study of transient cooling pond behavior. In exchange coefcient at the airwater interface as a function
Tech. Rep, Vol. 161, Ralph M Parsons Lab. for Water of wind speed and temperature: a critical survey of some
Resour. and Hydrod, MIT, USA. literature. J. Hydrol. 30, 375401.
Ryan P. J., Harleman D. R. F. and Stolzenbach K. D. (1974) Szeicz G. and McMonagle R. C. (1983) The heat balance of
Surface heat loss from cooling ponds. Water Resoure Res. urban swimming pools. Solar Energy 303, 247259.
105, 930938. Taga M., Matsumoto T. and Ochi T. (1990) Studies on
Salhotra A. T., Adams E. E. and Harleman D. R. F. (1985) membrane viscosity stabilized solar pond. Solar Energy
Effect of salinity and ionic composition on evaporation: 456, 315324.
analysis of Dead Sea evaporation pans. Water Resour. Res. Watmuff J. H., Charters W. W. S. and Proctor D. (1977) Solar
219, 13361344. wind induced external coefcients for solar collectors.
Sartori E. (1987) A mathematical model for predicting heat COMPLES 2, 56.
and mass transfer from a free water surface. In Proc. of the de Winter F. (1978). In How To Design and Build A Solar
1987 ISES Solar World Congress, Hamburg, Germany, Swimming Pool Heater, p. 46, Copper Development As-
Bloss W. H. and Psterer F. (Eds.), pp. 31603164. sociation, New York, USA.
Sartori E. (1989) Prediction of the heat and mass transfer from WMOWorld Meteorological Organization (1966) Measure-
a free water surface in the turbulent ow case. In Proc. of ment and estimation of evaporation and evapotranspiration.
the 1989 ISES Solar World Congress, Kobe Japan, In Tech. Rep., Vol. 83, p. 92, Working Group on Evaporation
Horigome T., Kimura K., Takakura T., Nishino T. and Fuji Measurement, Geneva.
I. (Eds.), pp. 23432347. Wu J. (1969) Wind stress and surface roughness at the airsea
Sartori E. (1991) Evaporation from a free water surface with interface. J. Geoph. Res. 742, 444455.
salt concentration. In Proc. ISES Solar World Congress, Yadav R. and Rao D. P. (1983) Digital simulation of indoor
Denver USA, pp. 23472351. temperatures of buildings with roof ponds. Solar Energy
Sartori E. (1996) Solar still vs. solar evaporator: a comparative 312, 205215.
study between their thermal behaviors. Solar Energy 562,
199206.

Potrebbero piacerti anche