You are on page 1of 7

Solar Energy Vol. 47, No. 3, pp. 223-229, 1991 00384)92X/91 $3.00 + .

00
Printed in the U.S.A. Copyright © 1991 Pergamon Press plc

EVALUATION OF MODELS FOR ESTIMATING


SOLAR IRRADIATION ON VERTICAL SURFACES
AT VALENCIA, SPAIN
M. P. UTRILLAS,J. A. MARTINEZ-LOZANO,and A. J. CASANOVAS
Departamento de Termodinfimica, Facultat de Fisica, Universidad de Valencia 46100 Burjassot,
Valencia, Spain

Abstract--Houdy irradiation data recorded on vertical surfaces at north, east, south, and west orientations
during the winter period going from December 1989to March 1990 in Valencia,Spain, have been compared
with estimated solar irradiation from several tilted-surfacemodels. The isotropic-, Temps' and Coulson's-,
Klucher's-, Hay's-, Skartveit'sand Olseth's-,Gueymard's- and Perez' (simplified)modelshave been considered
for this comparison. Root-mean-square-difference(RMSD), mean-bias-difference(MBD) and mean-absolute-
difference (MAD) estimators have been used to measure the departure of models from experimental data.
Modeled values are evaluated with the original coefficientsproposed by the authors. Results of comparison
show that the south orientation is less prone to modeling errors, Perez' 25° circumsolar simplified model
being the less affected by errors in north, east and south orientations; for west orientation Klucher's model
gives a lesser error. These results are consistent with those reported in previous studies.

!. I N T R O D U C I I O N 2. E X P E R I M E N T A L S E T - U P

Accurate modeling of solar irradiation on tilted surfaces The location where the measurements have been
at several orientations is needed for most of the prac- performed (Campus of Burjassot, Burjassot, Valencia)
tical applications of solar energy in active and passive is situated 40 m above the sea-level at a latitude of
systems. Horizontal solar global radiation is the most 39.5°N. Obstructions above the horizon are in general
widely measured parameter; most of the previous val- less than 4 °, except in a small zone in the northwest
idated work on solar irradiation modeling has therefore (Fig. I ). The data used in this study have been collected
been done in estimating diffuse and beam components during the period comprised between December 15,
of solar global radiation on a horizontal surface[l-4 ]. 1989 and March 15, 1990. Data acquisition continues
Modeling of solar irradiation on non-horizontal sur- to obtain longer measuring periods.
faces is more complex due to the effect of configuration The experimental set-up comprises two parts: the
factors and to diffuse radiation anisotropy over the sky first one provides the standard input of solar irradiation
dome, and needs therefore additional information, models and includes an Eppley Normal Incidence
provided, in most cases, by beam irradiation at normal Pyrheliometer (NIP), to measure beam irradiation, and
incidence. Some models use alternative input data, but a Kipp-Zonen CM-6 pyranometer, to measure global
Hay and McKay reported higher errors in these kind horizontal irradiation. The second one measures global
of models [ 5 ]. irradiation on vertical surfaces facing north, east, south,
In the last years a number of models have been and west. The instruments used in this case are Kipp-
proposed to estimate solar radiation on tilted surfaces; Zonen CM-11. The back of each of these pyranometers
most of them can be found in the reviews of Hay et has been protected with a white-painted wood panel.
a/.[5-7]. Later works however show discrepancies These instruments are provided with artificial horizons.
about the accuracy of models [ 8,9]. New models have Previously all the pyranometers used in this study have
been also proposed[10]. been compared with an Eppley Precision Spectral Pyr-
In the present work several models are compared anometer (PSP) on a horizontal surface.
with experimental data obtained in Valencia, Spain. Measured data are averaged and recorded every 10'
This location is situated at the seaside in the east of period on a battery-operated LI-1000 LI-COR data-
the Iberian peninsula. The data measured are of global logger. Once a week the data are uploaded from LI-
solar irradiation on vertical surfaces oriented north, 1000 to a mainframe computer where comparison of
east, south, and west, global solar irradiation on a hor- data with modeled values are done. Only hourly values
izontal plane and beam solar irradiation at normal in- have been used in this work.
cidence. For the measurement of global solar irradia-
tion on vertical surfaces the pyranometers are protected 3. S E L E C r E D M O D E L S
against ground albedo by an artificial horizon. The aim Hourly total irradiation on a E-angle tilted surface
of this comparison is to find the best model to be used of Ap azimuth is given by:
to evaluate solar irradiation contribution in the thermal
behaviour of buildings. No attempt has been made in ITaAp = Iba.4p + Iaaap + IraA, ( 1)
this paper to fit the models to the Mediterranean cli-
mate of Valencia: modeled values are evaluated with where the first subindex refers to the kind of irradiation
the original coefficients proposed by the authors. ( T, total, b, beam, n, normal to solar rays, d, diffuse
223
224 M.P. UTR1LLAS, J. A. MARTINEZ-LOZANO, and A. J. CASANOVAS

15 Idt~,4p = Ido0 ( 1 4- COS /3)/2 (4)

the diffuse irradiation being therefore independent of


12 plane's azimuth in this model.
ID
,-ff 3.2. Temps' and Coulson's model
.o Measures of diffuse solar irradiation taken by Temps
tO and Coulson [ 14 ] showed that diffuse irradiation is an-
>
isotropic over the sky dome. Diffuse irradiation in-
creases in the near horizon and in the circumsolar zone.
They proposed the following correction factors for the
near horizon zone

1 + sin3(/3/2) (5)
8 9~ 188 278 36~
azimuth(deg.) and the circumsolar zone,

Fig. 1. Actual horizon of the experimental site. 1 + cos20 sin3(90 - 3") (6)

giving the following expression for diffuse solar irra-


from the sky and r, diffuse from the ground), the sec- diation,
ond to the tilt angle/3 and the third to plane's azi-
m u t h Ap. ldaAp = ld00[(l + COS /3)/2][1 + sin3(/3/2)]
The beam irradiation term is given by
X [1 + cosE0 sin3(90 -- 3")]. (7)
IbaAp = I, COS 0 (2)
This is a clear-days-model based on the Robinson's
0 being the angle between the solar rays and the normal model [ 15 ]. Equation ( 7 ) does not reduce to (4) for/3
to the surface. In the case of a vertical surface (2) re- = 0 and O = 90 - 3".
duces [ l 1] to
3.3. Klucher's model
lb{,~/2)Ap = /,COS 3" Cos(A~ - Ap). (3) Klucher[16] considers isotropic model valid for
overcast skies and Temps' and Coulson's model valid
The diffuse from the ground contribution is not con- for clear skies. The Klucher formulation of diffuse solar
sidered in this paper and all the measures on nonhor- irradiation is given by
izontal surfaces are therefore made including artificial
horizons. laaAp = Iaoo[( 1 + cos/3)/2][1 + F sin3(/3/2)]
The differences between models are due to the dif-
X [1 + Fcos20 s i n 3 ( 9 0 - 3')] (8)
ferences in modeling the sky-diffuse irradiation. In the
following the diffuse irradiation models employed are
briefly described. where the modulating factor F is defined as

3.1 Isotropic model F = 1 - (Iaoo/Iroo) 2. (9)


The isotropic model considers an isotropic distri-
bution of diffuse solar irradiation over the sky Equation (8) reduces to isotropic model and Temps'
d o m e [ 12,13]. The diffuse solar irradiation on a tilted and Coulson's model for extreme values of Idoo/Iroo.
plane is given by the diffuse solar irradiation on a hor- Temps' and Coulson's, and Klucher's models, do
izontal plane corrected by a geometrical configuration- not consider a specific division of sky d o m e in zones
factor, of anisotropy.

Table 1. Error estimating global solar irradiation for a vertical surface facing north at Valencia, Spain
Absolute error (Win-2) Relative error (%)
ERROR ESTIMATORS MBD RMSD MAD MBD RMSD MAD
MODEL:
ISOTROPIC'S 1.2 24.1 18.1 2.2 43.5 32.6
TEMPS' & COULSON'S 21.1 42.4 30.9 38.3 76.4 55.6
KLUCHER'S 12.1 31.6 21.9 21.7 57.0 39.6
HAY'S -13.1 29.3 23.5 -23.6 52.8 42.4
SKARVEIT'S& OLSETH'S -18.4 28.1 21.7 -33.1 50.7 39.1
GUEYMARD'S -0.4 19.4 13.3 -0.8 34.9 24.1
PEREZ' (0°) -10.5 21.0 14.0 -18.9 37.8 25.3
PEREZ' (25°/ 0.5 14.1 10.0 0.9 25.5 18.1
Estimating solar irradiation on vertical surfaces at Valencia, Spain 225

Table 2. Error estimating global solar irradiation for a vertical surface facing south at Valencia, Spain
Absolute error (Wm-2) Relative error (%)
ERROR ESTIMATORS MBD RMSD MAD MBD RMSD MAD
MODEL:
ISOTROPIC'S -52.7 70.1 54.9 -13.7 18.3 14.3
TEMPS' & COULSON'S -8.3 45.1 36.9 -2.2 11.8 9.6
KLUCHER'S -29.2 42.7 31.9 -7.6 11.1 8.3
HAY'S -21.7 41.1 30.6 -5.7 10.7 8.0
SKARVEIT'S&OLSETH'S -26.9 43.6 33.6 -7.0 11.4 8.8
GUEYMARD'S -59.8 78.1 60.7 -15.6 20.4 15.8
PEREZ' (0°) 2.7 42.7 27.3 0.7 11.1 7.1
PEREZ' (25°) -9.2 34.6 22.7 -2.4 9.0 5.9

3.4. Hay's model Idt~.4e = 1dO0{ k(cos 0/sin 3') + Z cos 3


H a y [ l , 5 ] considers two different contributions to
diffuse radiation: a circumsolar one for the near-the- +(l-k-Z)(l +cos f)/2
sun sky-dome and an isotropic-distributed one for the - ~(1 - k - Z)WiCOSOi/Tv} (13)
rest of the sky-dome. Their relative weight depends on
atmospheric transmittance through the anisotropy in- with zero values for cos 0, cos 0i and Z if negative. The
dex k defined by Hay as last term in eqn (13) evaluates the diffuse irradiation
screened by obstacles in the horizon, o:, and 0~ being
k = I, II+,. (10) the solid and incidence angles for the i-th-sector of the
actual horizon.
Hay assumes linearity on the isotropic and circumsolar
contributions to diffuse irradiation on a tilted plane, 3.6 Gueymard's model
G u e y m a r d [ 18 ] considers that the irradiance for
Id~p = ldoo[ k( cos 0/sin 3') partially cloudy skies is a lineal combination of values
for overcast skies Ra~ and for clear skies Rdo,
+(1 -k)(1 +cos3)/2], (11)
Id~.4p = 1do0[( l -- N ) Rao + NRd, ] (14)
The first term corresponds to the circumsolar zone
contribution and the second one is the isotropic con- where N is the cloud opacity. N has been evaluated in
tribution. Equation ( 11 ) reduces to the isotropic model this work by using
for k = 0.
N= max{min(Y, 1),0} (15)
3.5. Skartveit's and Olseth's model
where Yis a function Ofldoo/Iroo. Yhas been obtained
Skartveit and Olseth [ 17 ] pointed out that for over-
by Gueymard from Montreal data and is given by
cast skies (k near zero) a substantial part of diffuse
radiation comes from the zenith. Experimental values
Y = 6.6667(Idoo/Iroo) -- 1.4167
showed that about 30% of the diffuse radiation of full
overcast days comes from the zenith, decreasing (Idoo/Iro0 < 0.227) (16)
quickly to negligible relative contribution if the cloud-
Y = 1.2121 (laoo/lroo) - 0.1758
iness factor is about 0.7 to 0.5. They proposed to mod-
ify Hay's model by introducing a Z correction factor, (Idoo/Iroo > 0.227). (17)

The clear sky's irradiance Rdo is given by sum of a


Z = 0.3 - 2k (12) circumsolar and a hemispherical term,

where k is Hay's anisotropy index and Z = 0 for k


Rdo = exp(ao + alcos 0 + a2cos20
>- 0.15. The expression for the diffuse irradiation in
the Skartveit's and Olseth's model is given by + a3cos30) + F(fl)G(3") (18)

Table 3. Error estimating global solar irradiation for a vertical surface facing east at Valencia, Spain
Absolute error (Wm-2) Relative error (%)
ERROR ESTIMATORS MBD RMSD MAD MBD RMSD MAD
MODEL:
ISOTROPIC'S -22.3 45.1 31.3 14.4 29.1 20.2
TEMPS' & COULSON'S 3.0 41.9 32.7 2.0 27.1 21.1
KLUCHER'S -9.1 36.8 25.7 -5.8 23.8 16.6
HAY'S -27.1 43.6 33.1 -17.6 28.2 21.4
SKARVEITS& OLSETH'S -32.3 44.2 33.9 -20.9 28.6 21.9
GUEYMARD'S 4.7 97.7 47.8 3.1 63.2 30.9
PEREZ' (0") -14.7 32.7 22.8 -9.5 21.1 14.7
PEREZ' ~25") -12.6 25.4 17.7 -8.1 16.4 11.4
226 M. P. UTRILLAS,J. A. MARTINEZ-LOZANO,and A. J. CASANOVAS

Table 4. Error estimating global solar irradiation for a vertical surface facing west at Valencia, Spain
Absolute error (Win-2) Relative error (%)
ERROR ESTIMATORS MBD RMSD MAD MBD RMSD MAD
MODEL:
ISOTROPIC'S -14.6 36.5 25.1 -8.9 22.3 15.3
TEMPS' & COULSON'S 15.8 38.2 30.2 9.6 23.3 18.4
KLUCHER'S 0.9 30.7 22.1 0.5 18.7 13.5
HAY'S -6.3 36.4 26.6 -3.8 22.2 16.3
SKARVEIT'S & OLSETH'S -11.5 36.7 26.3 -7.0 22.4 16.1
GUEYMARD'S 17.2 121.5 56.1 10.5 74.2 34.2
PEREZ' (0°) -2.6 52.7 29.0 -1.6 32.2 17.7
PEREZ' (25°) -0.2 43.7 22.4 -0.1 26.7 13.7

a~ being functions o f 3'. T h e overcast sky irradiance a n d modeling process can be very tedious a n d the same
Rdj depends only o n the plane tilt-angle a n d o n a cor- a u t h o r s [ 1 0 ] proposed a simplified version, where eqn
rection factor b ( 1.0 -< b < 2.0) a n d is given by ( 2 1 ) is reformulated as

Rd, = Rai - { r - 1 ( / 3 COS/3 - sin/3) IaeAp = I d ~ { ( ~ i , ~ / ~ ) + (la, MaM&~&<.,)


+ ½(1 - cos/3)}/(1 + 3/2b)} (19) + (Iahoza~ho=/~hoz)}. (23)

Ra~ being the configuration factor, By writing

Rdi = ( 1 + cos/3)/2. (20) Ia,.jIa,~ = ( a / c ) (24)

3.7 Perez' simplified model IahojIahoz = ( b / d ) (25)


Perez' model [ 19,20 ] divides the sky d o m e into three
l a , ~ h / ~ = F~ (26)
different zones, each o n e with a characteristic diffuse
irradiance: a h o r i z o n zone, a circumsolar zone a n d an ~h,,:/~ = F~ (27)
isotropic zone. T h e diffuse irradiation given by Perez'
m o d e l is we obtain a lineal expression in the reduced coefficients
of the model F'j a n d F~,
IdSAp = Iaoo[0.5( 1 + cos/3) + a(O)(Fl -- 1)
laeAp = Idoo[0.5(l + COS ~)( 1 - F'I - F~)
+ b(B)(F2 - 1)]/

[! +c(Oz)(Fa - 1 ) + d ( F 2 - 1)] (21) + F't(a/c) + F;(b/d)]. (28)

being all factors in e q n (21 ) geometrical except for the T h e horizon zone is considered of zero width a n d
anisotropy coefficients o f the model F~ = L ' / L and/72 the circumsolar zone alternatively 0 ° ( p o i n t source)
= L " / L . E q u a t i o n (21 ) reduces to the isotropic model or 25 ° wide to simplify the evaluation of F~ a n d F~.
ifF~ = F 2 = 1. T h e n u m b e r o f fits is reduced to only eight. For the
F, a n d F2 are functions o f zenith angle 0~, diffuse p o i n t source circumsolar model we obtain
horizontal irradiation Idoo a n d of sky clearness p a r a m -
eter ~, defined as Id#Ap = Id00[0.5( 1 + cos/3)( 1 - F'l)

= (Idoo + I~)/Idoo. (22) + F'~ (cos 0~/ cos 0) + F~ sin/3]. ( 29 )

Intervals o f these three parameters define 240 different T h e 25 ° circumsolar Perez' model is less simple but
categories in Perez' model. F o r each category Fj a n d m o r e accurate. Both versions have been applied in this
/72 m u s t be fitted from experimental data. T h e fitting- paper.

Table 5. Rank of models (I: Isotropic; T&C: Temps and Coulson; K: Klucher; H: Hay; S&O: Skartveit
and Olseth; G: Gueymard; P0°: Perez simplified-point-source; P25°: Perez simplified 25 ° circumsolar)
for the four vertical surfaces
VERTICAL NORTH VERTICAL SOUTH VERTICAL EAST VERTICAL WEST
RMSD MAD RMSD MAD RMSD MAD RMSD MAD
~5 ° P25° P25° P25° P25° ~5 ° K K
G G H ~o ~° ~° H P250
P0° ~° P0° H K K 1 I
I I K K T&C I S&O S&O
S&O S&O S&O S&O H T&C T&C H
H K T&C T&C S&O H ~5 ° ~°
K H I I I S&O P0° T&C
T&C T&C G G G G G G
Estimating solar irradiation on vertical surfaces at Valencia, Spain 227

3~ mard's and Perez' (25 ° and 0 ° circumsolar) models


caic.O,4m-2 ) improve the isotropic model (MAD and RMSD are
lower). The best relative MAD is for Perez' 25 ° model
(18%), Temps' and Coulson's model obtaining the
highest MAD (56%). The same models give the ex-
treme values for RMSD: 26% Perez' 25 °, and 76%
Temps' and Coulson's model. MBE values show that
150 Hay's, Skartveit's and Olseth's, and Perez' 0 ° models
:.....:::i/"
• .....~.,.,r. •
underestimate the global irradiation values, whereas
the Klucher's and Temps' and Coulson's models clearly
-!~'i:'" overestimate global irradiation on north plane.
In the south plane, on the contrary, all models ex-
cept Gueymard's improve the isotropic model values,
okm.Chlm-2) on a RMSD and MAD basis. The best relative MAD
0 150 300 is for Perez' 25 ° model (6%), Gueymard's model ob-
taining the highest MAD (16%). The same models
Fig. 2. Perez simplified25 ° circumsolar model calculated vs.
observed hourly values of total solar irradiation incident on give the extreme values for RMSD: 9% Perez' 25 °, and
a vertical surface facing north at Valencia, Spain. 20% Gueymard's model. All models, except the iso-
tropic and Gueymard's ones, introduce relative MAD
values lower than 10%. Therefore any of these models
4. MODEL ERROR ESTIMATORS seems to be adequate to evaluate irradiation on vertical
The statistical estimators root mean square differ- planes facing south. As other authors have pointed
ence (RMSD), mean absolute difference (MAD) and out[10,21] all the models, except the Perez' 0 ° one,
mean bias difference (MBD) are common model error subestimate the hourly global irradiation values.
estimators used to evaluate the accuracy of the models. In the east plane the best relative MAD is for Perez'
The three estimators are different in nature and can 25 ° model (11%), Gueymard's model obtaining the
give different results. We therefore report the three es- highest MAD (31%). The same models give the ex-
timators for each of the compared models and orien- treme values for RMSD: 16% Perez' 25 °, and 63%
tations. RMSD gives more weight to points far away Gueymard's model. The Perez' 0 °, and Klucher models
from the mean than MAD. RMSD and MAD give give similar errors slightly higher than Perez' 25 °
positive values. MBD can give positive or negative val- model. Temp' and Coulson's, Hay's, and Skartveit's
ues, corresponding the positive ones to an overesti- and Olseth's models give similar errors as the isotropic
mation of the model. Usually in solar irradiation papers model. Only Temps' and Coulson's, and Gueymard
absolute values have been used as statistical estimators. models overestimate slightly the experimental values;
We evaluate also the relative values to obtain com- the rest of the models underestimate global irradiation
parable results. on the east plane.
In the west plane the best relative MAD is for Perez'
25 o and Klucher's models (about 13.5% ), Gueymard's
5. RESULTS AND DISCUSSION
model obtaining the highest MAD (34%). Extreme
The reported results compare hourly experimental values of RMSD are 19% for Klucher's (27% for Perez'
global irradiation with the corresponding model eval- 25 °) and 74% for Gueymard's model. Perez' 25° model
uated value.
Horizontal diffuse irradiation values are needed for
evaluating models. Diffuse irradiation has been eval- 1000
uated by the sustractive method. We can therefore in- calc.(~Im-2) "
troduce high uncertainties in the evaluation of diffuse
solar irradiation in clear days. In these conditions the
absolute error of instruments is the highest and we can
/ .i!:i
• ,,r.
have, at same time, a reduced value of diffuse irradia- ..~: ;:
tion. Overcast days on the contrary give relative errors
of diffuse irradiation close to the 5% low limit. The 500
average value of relative error in the experimental
evaluation of diffuse irradiation in our measures is $'¢..r',. :
about 13%, being the low limit 5%. In any case, the
uncertainties of horizontal diffuse irradiation influence
all models in the same way.
Results are summarized in Tables 1 to 4 that provide
a summary of the statistical estimators for each one of 0 5e0 1 ~
the eight models and four orientations. Fig. 3. Perez simplified25° circumsolar model calculated vs.
The north plane is receiving basically only diffuse observed hourly values of total solar irradiation incident on
radiation during the measured period. Only the Guey- a vertical surface facing south at Valencia, Spain•
228 M. P. UTRILLAS,J. A. MARTINEZ-LOZANO,and A. J. CASANOVAS

relative MBD is only -0.1%, less than Klucher's model 1000


0.5%. Therefore Perez' 25 ° model gives values more cale.(14m-2 ) " /
scattered about the mean. All the models, except the
Gueymard one, give very similar relative MAD values:
therefore any of these models seems to be adequate to
evaluate irradiation on vertical planes facing west. Only
Temps' and Coulson's, and Gueymard models over-
estimate the experimental values; the rest of the models 500

/
underestimate global irradiation on the west plane. ~ ~•
We consider that quantitative comparison of results
obtained by different authors in different locations is
better done on a relative RMSD or relative MAD basis.
Unfortunately, only very few papers report relative
-2)
values of RMSD or MAD. A rank of models for the
four orientations is presented on Table 5 for qualitative o 5oo 18o8
comparison. Fig. 5. Klucher's model calculated vs. observedhourly values
If the relative MAD is used as the criterion for of total solar irradiation incident on a vertical surface facing
ranking models, we obtain that Perez' 25 ° model gives west at Valencia, Spain.
the best overall results except in the west plane, where
the Klucher model gives a lesser relative MAD. Except
in the north and west planes, Perez' 0 ° model gives
give good predictions on the south vertical plane (rel-
the second best model. In the north plane Gueymard's
ative MAD less than 10%), whereas on the north ver-
model atypically gives good results, being the worst of
tical plane only the Perez' 25 ° model gives a relative
the compared models in the other planes. Again, we
MAD less than 20%.
can expect that Gueymard's parameters are more
The results obtained for the east and west planes
strongly tied together with location than the other
are about the same, with an appreciable asymmetry
models, due to local characterization of ai parameters.
on the type of models best suited to the plane consid-
The north plane shows the greatest differences between
ered. The isotropic, Temps' and Coulson's, Klucher's,
the models in the first and second position. Modeled
Hay's, and Skartveit's and Olseth's models predict the
values are compared with the experimental data in Figs.
global irradiation on the west plane better than on the
2 to 5. The best model for each plane has been selected.
east plane; the Perez' models, predict the global irra-
These results agree, except for the west plane, with
diation on the east plane better than on the west plane.
those obtained by Hay, Perez, and McKay [ 6 ], and by
This asymmetry has been found also by Hay, Perez,
Perez et a/.[10]. The results obtained with relative
and McKay [ 6 ] for Vancouver and by Perez et al. [ 10 ]
RMSD coincide with the former, being only the Hay's
for Trappes.
model in second place for south and west orientations.
Results in the west plane can be affected by the horizon
obstructions present in the west orientation of location 6. CONCLUSIONS
( 51% of the area of Perez' 6.5 ° horizon zone, Fig. 1).
An overall evaluation of results shows that the Perez'
In the scope of the obtained results we can say that
25 ° model gives the more accurate results. As in the
all models, except the isotropic- and Gueymard ones,
case of Harrison and Combes [ 22 ], it is likely that the
RMSD and MAD estimators decrease if we use pa-
rameters fitted for Valencia data in the Perez model.
Ifl00 In a future paper we will compare the results of using
calc.(14m - 2 )
the parameters fitted to Valencia values relative to the
original ones on models.
In the south plane most of the models give accept-
able results. In fact, most of the models have been de-
veloped to perform well in south planes. Perez' 25 °
• .~":" gives slightly better results than other models, that may
• .. ~,.:7 . not be compensated by a more extended work to fit
the coefficients and to model the diffuse irradiation.
The experimental values used in this paper corre-
spond to a winter period, where the diffuse irradiation
values are relatively higher. As in the case of Ma and
Iqbal [ 21 ] for Woodbridge, a longer measuring period
obs.(Nm -E)
, will probably lower the values of RMSD and MAD.
500 IO00

Fig. 4. Perez simplified25° circumsolar model calculated vs. Acknowledgment--This work has been partially supported by
observed hourly values of total solar irradiation incident on the lnstituci6 Valenciana d'Estudis i Investigaci6(IVEI), Va-
a vertical surface facing east at Valencia, Spain. lencia, Spain.
Estimating solar irradiation on vertical surfaces at Valencia, Spain 229

NOMENCLATURE 5. J. E. Hay and D. C. McKay, Estimating solar irradiance


on inclined surfaces: A review and assessment of meth-
Ap Plane azimuth angle odologies, Int. J. Solar Energy 3, 203-240 ( 1985 ).
As Solar azimuth angle 6. J. E. Hay, R. Perez, and D. C. McKay, Addendum and
ai Local parameters of Gueymard' model errata to the paper estimating solar irradiance on inclined
F Klucher's modulating factor surfaces: A review and assessment of methodologies, Int.
F~, F2 Perez' model coefficients J. Solar Energy4, 321-324 (1986).
F~, F~ Perez' simplified model reduced coefficients 7. J.E. Hay and D. C. McKay, Calculation of solar irradiance
Ir Total irradiance/total hourly irradiation (depending for inclined surfaces; Verification of models which use
on context) hourly and daily data, I.E.A. Solar Heating and Cooling
Beam irradiance/beam hourly irradiation (depending Program, TASK IX Final Report, Atmospheric Environ-
on context) ment Service, Downsview, Ont., Canada ( 1987 ).
1,l Diffuse sky irradiance/diffuse sky hourly irradiation 8. M. A. Abdelrahman and M. A. Elhadidy, Comparison
(depending on context ) of calculated and measured values of total radiation on
Diffuse reflected irradiance/reflected hourly irradia- tilted surfaces in Dhahran, Saudi Arabia, Solar Energy
tion (depending on context) 37, 239-243 (1986).
1,, Normal incidence irradiance/normal hourly irradia- 9. B. H. Chowddhury and S. Rahman, Comparative assess-
tion (depending on context) ment of plane array irradiance models, Solar Energy 39,
I~ Solar constant 391-398 (1987).
k Hay's anisotropy index 10. R. Perez, R. Seals, P. Ineichen, R. Stewart, and D. Men-
L Isotropic zone irradiance (Perez' model ) icucci, A new simplified version of the Perez diffuse ir-
L' Circumsolar zone irradiance (Perez' model ) radiance model for tilted surfaces, Solar Energy 39, 221-
L" Horizon zone irradiance (Perez' model ) 231 (1987).
N Cloud opacity factor 11. K. Y. Kondratyev, Radiation in the atmosphere, Aca-
Rao Clear sky's irradiance (Gueymard' model) demic Press, New York (1969).
Rat Overcast sky's irradiance (Gueymard' model ) 12. K. Y. Kondratyev and M. P. Manolova, The radiation
Z Zenith diffuse irradiation correction factor (Skarveit's balance of slopes, Solar Energy 4, 14-19 (1960).
and Olseth's model ) 13. B.Y.H. Liu and R. C. Jordan, Daily insolation on surfaces
tilted toward the equator, ASRHAE J. 3, 53-59 ( 1961 ).
14. R.C. Temps and K. L. Coulson, Solar radiation incident
Greek upon slopes of different orientations, Solar Energy 19,
a Circumsolar zone half-angle width (Perez' model 179-184 (1977).
Plane tilt angle 15. N. Robinson, Solar radiation, Elsevier, Amsterdam
3' Solar altitude ( 1965 ).
Sky clearness parameter 16. T. M. Klucher, Evaluations of models to predict insolation
0 Angle between solar rays and surface normal on tilted surfaces, Solar Energy 23, 111-114 (1979).
0z Zenith angle 17. A. Skartveit and J. A. Olseth, Modelling slope irradiance
~" Horizon zone angle width (Perez' model) at high latitudes, Solar Energy 36, 333-344 (1986).
18. C. Gueymard, An isotropic solar irradiance model for
tilted surfaces and its comparison with selected engi-
REFERENCES neering algorithms, Solar Energy 38, 367-386 ( 1987 ).
19. R. Perez, J. Scott and R. Stewart, An anisotropic model
1. J. E. Hay, Calculation of monthly mean solar radiation for diffuse radiation incident on slopes of different ori-
for horizontal and inclined surfaces, SolarEnergy 23, 301 - entations, and possible applications to CPCs, Proc. oJ
307 (1979). ASES, Minneapolis, 883-888 (1983).
2. M. lqbal, Prediction of hourly diffuse solar radiation from 20. R. Perez, R. Stewart, C. Arbogast, R. Seals, and J. Scott,
measured hourly global radiation on a horizontal surface, An anisotropic hourly diffuse radiation model for sloping
Solar Energy 24, 491-503 (1980). surfaces: Description, performance validation, site de-
3. J. E. Hay and D. C. McKay, Evaluation of selected models pendency evaluation, Solar Energy 36, 487-497 ( 1986 ).
for estimating solar radiation on horizontal surfaces, Solar 21. C.C.Y. Ma and M. lqbal, Statistical comparison of mod-
Energy 43, 153-168 (1989). els for estimating solar radiation on inclined surfaces, Solar
4. R. Perez, R. Seals, A. Zelenka, and P. Ineichen, Climatic Energy31, 313-317 (1983).
evaluation of models that predict hourly direct irradiance 22. A. W. Harrison and C. A. Coombes, Performance vali-
from hourly global irradiance: Prospects for performance dation of the Perez tilted surface irradiance model, Solar
improvements, Solar Energy 44, 99-108 (1990). Energy 42, 327-333 (1989).

You might also like