You are on page 1of 5

Solar Energy, Vol. 53, No. 4, pp.

353-357, 1994
Copyright 1994 Elsevier Science Ltd
Printed in the USA. All rights reserved
0038-092X/94 $6.00 + .00

Pergamon
0038-092X(94)00055-7

THE EFFECT OF TIME ERRORS ON THE ACCURACY OF


SOLAR RADIATION MEASUREMENTS
H. SUEHRCKE
Department of Mechanical Engineering, James Cook University of North Queensland,
Townsville, Qld 4811, Australia
Abstract--There is a need to focus on the accuracy of the time that is attached to solar radiation measurements.
In this paper the effect of time errors on the calculation of the clearness index is investigated. An approximate
expression for the error in clearness index due to time errors is derived. Using instantaneous solar radiation
measurements from Townsville, Australia, the effect of time errors is illustrated with cumulative probability
distribution curves. It is shown that even relatively small time errors ( 10 min ) can substantially change the
distribution curves. Time errors can be caused by drifts or inaccuracies of the data recording device's clock.
1. INTRODUCI'ION

T h e m e a s u r e m e n t of solar radiation is subject to m a n y


sources of error. C o m m o n l y the accuracy of a solar
radiation m e a s u r e m e n t is understood as the accuracy
o f the m e a s u r e d radiation q u a n t i t y [e.g., see Nast
( 1983)]. However, for m a n y purposes it is also required
to record the t i m e of day w h e n the radiation measurem e n t was taken. For example, for the d e t e r m i n a t i o n
of the a t m o s p h e r i c t r a n s m i t t a n c e of solar radiation or
the calculation of horizontal surface b e a m radiation
from pyrheliometer m e a s u r e m e n t s it is necessary to
record b o t h the radiation value a n d the time. In such
a case the accuracy of the derived quantity also depends
on the accuracy of the recorded time.
T h e availability o f computerized data acquisition
in recent times has enabled the recording o f detailed
i n s t a n t a n e o u s or hourly insolation values. However,
the a u t o m a t e d data recording of such detailed insolation values requires one to keep track o f the time
over extended periods. This paper discusses errors in
the calculation of the clearness index resulting from
errors in the time that is attached to radiation measurements. Motivation for this paper is that time errors
are believed to exist in some current radiation data
a n d that the accuracy of the time is of f u n d a m e n t a l
i m p o r t a n c e for the solar radiation m e a s u r e m e n t .

It is a p p a r e n t from eqn ( l ) that a n erroneous value


of Go will produce an erroneous clearness index value.
T h e relative error in the clearness index due to time
errors can be defined as:

{relative error of k } =

Go(t + At ...... )

Go(t)

(2)

Go(t)
where t is the solar time a n d Ate..... the time error. A
first order approximation for the error in extraterrestrial

dGo

radiation Go is T

At ...... (see also Fig. 1 ). Using this

a p p r o x i m a t i o n in eqn (2) gives:

G
dGo
Go + - - ~ Ate

G
Go

.....

{ relative error o f k } =

G
Go

1
1+

(3)

dGo
-Ate .....
dt

Go

2. ANALYSIS
Solar radiation m e a s u r e m e n t s are often presented
in terms of the clearness index which for instantaneous
solar radiation is defined according to Duffle a n d
B e c k m a n ( 1991 ) as:
k -

Go

(1)

F u r t h e r simplification can be achieved by multiplying n u m e r a t o r a n d d e n o m i n a t o r of the first term

dGo

in eqn ( 3 ) by 1 - - - ~ At ...... /Go.

{ relative error of k } =

dGo
dl Ate.....

where G denotes the i n s t a n t a n e o u s solar radiation per


square m e t e r horizontal surface area and, Go, the extraterrestrial solar radiation. T h e reason for the use of
the clearness index is t h a t it enables c o m p a r i s o n s between the radiation at different zenith angles a n d that
it removes insolation variations due to changes in sunearth distance.

l353

Go
dGo
\2
--d7 At ...... }
/

dGo
dt At .....
1x

Go

(4)

354

H.

SUEHRCKE

where the last approximation is only justified when


dGo
--fir- At~.....
,~ I. Although this approximation does
Go
not hold close to sunrise or sunset, eqn (4) correctly
predicts that the relative error of the clearness index
can approach infinity as sunrise or sunset are approached. Following Rabl ( 1985 ), the extraterrestrial
solar radiation can be expressed as:

C
,O
m

true soler noon

ee.

m
,m
t_

.eir m o s s

L_
t-

W
!
@

Go = GoncOs Oz

t - Solor Time

'

Z~terror

= Go.(COS 6 cos 4~ cos w + sin 6 sin q~)


where Go, is the extraterrestrial radiation normal to
the sun and 6, ~b, and w are the declination, latitude
and hour angle, respectively. Differentiating gives:
dGo
dt -

dw
Go.(COS 6 cos 4~ sin ~0) -~-

The use o f e q n (4) is illustrated with an example.


Consider the relative error for an equinox day (6 = 0)
at 16:00 due to a time error in solar time of +10 min.
dGo
- - Ate .....
dt
Go
dw
- G o , ( c o s 6 cos 4~ sin o~) - ~ At . . . . . .
Go,(cos 6 cos 4~ cos ~o + sin 6 sin 40
= tan w ~

At. . . . . .

tan

i-2 60

= +0.08 ( = +8%)
This error is greater than errors usually accepted for
solar radiometers. The result also provides a check for
dGo
the assumption that

At ...... = 0.08 ,~ 1 in this


Go

case.
The above analys~s is incomplete without an example that shows how time errors can affect the results
of solar radiation studies. The example used below is
the cumulative probability distribution which forms
the basis for many solar energy calculations.

3.

EXPERIMENTAL

DATA

Horizontal surface measurements of global (beam


+ diffuse) solar radiation were taken with a Kipp &
Zonen CM 11 pyranometer. The instrument was placed
on top of a water tower adjacent to the Mechanical
Engineering building at James Cook University in
Townsville, Australia (latitude 19.3S, longitude
146.8E) to ensure shade free measurements.

Fig. 1. The effect of a time error on the calculation of


extraterrestrial solar radiation.

The measurements were acquired with a National


Instruments NB-MIO-16X data acquisition card and
a Macintosh llci computer using NB LabDriver routines. The pyranometer output was transferred to the
computer via a 35 m long shielded signal cable and an
Analog Devices 5B Series signal conditioning module.
The 5B Series module provided signal pre-amplification and optical isolation to protect the sensitive computer equipment from high voltages. Uninterrupted
data recording was achieved using an uninterruptable
power supply.
The measurements of instantaneous solar radiation,
G, were taken at l-minute intervals for the period December 1990-December 1991. Each solar radiation
measurement was stored on computer disc together
with the local time of day. The time attached to each
radiation measurement was kept accurate to within 15
seconds.
In order to reduce high frequency noise picked up
by the signal cables, the cable shield was connected to
the ground at the computer. Remaining high frequency
noise was removed by averaging 200 successive measurements, taken at a rate of one measurement per 100
ttsec, over ~0 sec to give one instantaneous radiation
measurement. The pyranometer was new at the beginning of the measurement period and its glass d o m e
was cleaned every 2-3 days.
The measurements of global solar radiation, G, were
converted to instantaneous clearness index values (k
= G/Go) using a solar constant, Gs,, of 1367 W / m 2
and sorted in to three air mass intervals as defined in
Table 1. For the calculation of the clearness index the
local time of day was converted to solar time using the
equation lsolar = l l o c a / - - 4(Ls, - LI,~ ) + E for longitudes
east of Greenwich.
The extraterrestrial radiation, Go, was calculated
using Go = Gs,( 1 + 0.033 cos(21rn/365 )). cos 0z, where
n is the day of the year (1-365) and 0z zenith angle as
defined by Rabl ( 1985 ). The air mass, m, on the other
hand, was approximated by m ~ 1/cos 0z = (cos ~b
cos 6 cos ~ + sin ~b sin 6) -1 , which according to
Robinson (1966) may be used for zenith angles less
than 70 .

Effect of time errors


Table 1. Air mass intervals
Nominal

Actual

1.2
1.5
2.0

1.125 < m < 1.275


1.400 _< m < 1.500
1.800 _< m < 2.200

hibit a step separating the clear and cloudy radiation


values. The clear radiation values are associated with
moments in time when no cloud is in front of the sun,
whereas the cloudy radiation values correspond to
moments in time when a cloud obscures the sun. The
bimodal clear/cloudy radiation character is very similar to the one observed in Perth, Western Australia,
by Suehrcke and McCormick (1988), even though
Townsville has a different latitude and different climate.
Figure 3 shows cumulative probability distribution
curves that were evaluated from the same data as the
curves of Fig. 2. However, in this case a progressive
time error of 5 sec/day was deliberately introduced to
the measurements of radiation. The error in the time
was linearly increased from zero in December 1990 to
approximately 30 min (365 ~) in December 1991
according to Robinson ( 1991 ). This accumulating time
error simulates a speeding clock of the data recording
device attaching the incorrect values of time to the
radiation measurements.

Following the procedure described above, the radiation values were sorted into the air mass intervals
shown in Table 1 and put into ascending order to obtain cumulative probability distribution curves.

4. CUMULATIVE PROBABILITY DISTRIBUTION OF


INSOLATION VALUES

Figure 2 shows cumulative probability distributions


of instantaneous horizontal surface solar radiation for
three different air mass values for Townsville, Australia.
The cumulative probability curves in Fig. 2 clearly ex-

Townsville
Dec90toDec91

1.0

355

m~//]

1.0

D:wnsvtlle

rn= 1.5 1

X
CP
"D
m

CP

0.5

0.5

(.J
L
=~

v a l u e s27004/

(b)

(a)
J

0.0

O.O

0.5
f-

Fraction

0.0

1.0

0.0

0.5

of Time

f-

1,0- Townsvllle
Dec9 0 ~

Fraction

of Time

m :2.0 J

0.5

22808values
(e)

0.0

0.0

0.5
f - Fraction

I
1.0

~"

1.0
of Time

Fig. 2. Cumulative probability curves of instantaneous horizontal surface solar radiation for Townsville,
Australia: (a) m = 1.2, (b) m = 1.5, (c) m = 2.0.

356

H. SUEHRCKE

1.0

Townsville

m = 1.2

1.0

Townsville
Dec 90 to Dec 91
Time

m = 1.5

J
I

~P
I

~P

0.5

0.5

k.
~P
Q}

~P
w

3 3 2 4 8 values

27001

I
=~

f
0.0
0.0

S {b )

(a)
,

0.5
f - Fraction

0.0
0.0

1.0

0.5

of Time

1.0 -

f-

Townsville
Dec 90 to Dec 91

Fraction

1.0
of Time

m = 2.0 }

T i m e drift 5 sec/day
X

L_
=

0.5

(c)
0.0

0.0

0.5
f - Fraction

1.0
of

Time

Fig. 3. The cumulative probability curves of Fig. 1 with a deliberately introduced time error increasing from
zero in December 1990 at a rate 5 sec/day to 30 rain in December 1991: (a) m = 1.2, (b) m = 1.5,
(c) m = 2.0.

It is a p p a r e n t from Fig. 3 that the time errors have


little effect for the curve of air mass 1.2. O n the other
hand, for the curve of air mass 2.0, the time errors
have a significant effect. The bimodal radiation character, clearly visible in Fig. 2c, has completely disappeared in Fig. 3c. Moreover, the time errors have increased/decreased the clear sky clearness index values
o f the m = 2.0 probability curve by approximately
10%. For example, Fig. 3c suggests a clear sky clearness
index of 0.80 at f = 0.85, whereas the accurate curve
in Fig. 2c only shows a clearness index of 0.72 at this
point.
W h y are the cumulative curves in Fig. 3 different
from Fig. 2? The curves in Fig. 2 and Fig. 3 were derived
from the same data except that the second data set
c o n t a i n e d erroneous values o f the time of day. These
time errors caused different radiation values to be selected from the second data set when the radiation
data were sorted into air mass intervals. This is illus-

trated in Fig. 1 which shows how a time error shifts


the air mass interval windows away from the correct
symmetric position. As a consequence, for each air
mass interval there are two groups of insolation values
which do not belong to the same value of air mass.
Hourly solar radiation m e a s u r e m e n t s are affected
by time errors in a similar m a n n e r to instantaneous
measurements. The procedure of separating hourly
data according to the h o u r o f day or incidence angle
is analogous to separating instantaneous solar radiation
m e a s u r e m e n t s according to air mass. The effects of
time errors for hourly radiation measurements, however, are less obvious. As the bimodal radiation character is not as p r o n o u n c e d for hourly cumulative
probability curves [Suehrcke a n d McCormick ( 1988)]
except at high solar elevations [ Olseth a n d Skartveit
( 1987 ) ], the effect o f time errors m a y only be detected
from spuriously high radiation values close to s u n r i s e /
sunset.

357

Effect of time errors


T h e effect o f time errors on the probability distrib u t i o n of insolation values was discussed above for
progressive time errors as they can be caused, for example, by a speeding clock. This type of error is difficult
to detect a n d even more difficult to correct. Systematic
errors, on the other hand, such as errors in the conversion of local time to solar t i m e can be detected simply by plotting m o r n i n g a n d afternoon values separately. T h e results presented in this section simultaneously provide an experimental result on the statistics
of i n s t a n t a n e o u s solar radiation.
5. CONCLUSIONS
Errors in the time that is associated with solar radiation m e a s u r e m e n t s can have a significant effect o n
derived radiation quantities such as the clearness index.
Even relatively small time errors in the solar time can
cause (two-sided) errors in the clearness index that
would be regarded as unacceptable for solar radiation
measurements. Using the example o f cumulative
probability distribution curves from Townsville, Australia, it has been shown that accurate timekeeping is
particularly i m p o r t a n t for solar radiation measurem e n t s close to sunrise or sunset. The cumulative probability curves provide experimental evidence that ins t a n t a n e o u s solar radiation exhibits a distinct bimodal
character.

Acknowledgment--The financial assistance by the Australian


Energy Research Development Cooperation for the equipment
used in this study is gratefully acknowledged.
NOMENCLATURE
E equation of time. rain, see Duffle and Beckman ( 1991 )
for definition

G instantaneous total (beam + diffuse) solar radiation


on horizontal surface, W/m 2
Go extraterrestrial instantaneous solar radiation on horizontal surface, W / m z
Gon extraterrestrial solar radiation measured on a plane
normal to the sun on the nth day of the year, W/m 2
Gs,. solar constant, Gsc = 1367 W/m 2
f fraction of time
k clearness index for instantaneous solar radiation (atmospheric transmittance), k = G/Go
L longitude, degrees
n day of year
m air mass, m ~ 1/cos 0z for zenith angles from 0 to
70
t time
Ate..... error in solar time
declination
0~ zenith angle
4~ latitude
/
CO

hour angle, ~ - = rr/12 per hour


aft

REFERENCES

J. A. Duffle and W. A. Beckman, Solar Engineering ~?fthermal


processes. 2nd ed., Wiley, New York ( 1991 ).
P. M. Nast, Measurements on the accuracy of pyranometers,
Solar Energy 31,279-282 ( 1983 ).
J. A. Olseth and A. Skartveit, A probability density model for
hourly total and beam irradiance on arbitrarily oriented
planes, Solar Energy 39, 343-351 ( 1987 ).
A. Rabl, Active solar collectors and their applications, Oxford
University Press, New York (1985).
A. M. Robinson, Personal communication, Department of
Electrical Engineering, University of Edmonton, Canada
( 1991 ).
N. Robinson, Solar radiation, p. 51, Elsevier. New York
(1966).
H. Suehrcke and P. G. McCormick, The frequency distribution
of instantaneous insolation values, Solar Energy 40, 413422 (1988).

You might also like