You are on page 1of 5

2017 UKSim-AMSS 19th International Conference on Modelling & Simulation

Sunshine Duration-Based Models for Predicting Global Solar Radiation

Alhassan Ali Teyabeen Ali Elseddig Jwaid


Center for Solar Energy Research and Studies Department of Computing and Technology
Tripoli, Libya Nottingham Trent University
Alhassan.teyabeen@gmail.com Nottingham, UK
ali.jwaid@ntu.ac.uk

Abstract—Measured solar radiation data are the best source for the six Algerian cities was provided by cubic and
for proper knowledge of global solar radiation. However, it is quadratic regression models. F. Bannani et al. [5] estimated
not possible to measure global solar radiation in many areas the monthly average global solar radiation using a linear
due to cost and maintenance of the measuring equipment. At equation based on based on measured global solar radiation
this point, the relationship between global solar radiation and collected in eleven stations in Libya during eight years. The
sunshine duration plays an important role. This paper presents results indicated that linear regression equation for ten
a new model, as well as eight models which presented in the stations out of eleven holds very well. H. Li et al. [6]
literature for estimating global solar radiation on the proposed a new empirical model for estimating daily global
horizontal surface using only sunshine duration. This new
solar radiation on a horizontal surface at nine stations in
model is called reverse model. The estimation of global solar
China. The performance of the models is evaluated using
radiation is based on measured global horizontal irradiance
data collected in Tripoli (lat. ૜૛‫ ܗ‬ૡ૚ᇱ, long. ૚૜‫ ܗ‬૝૜ᇱ), Libya
statistical error tests. Their results showed that the new
during 2015 and measured every 1 minute. The performance of proposed model provides a better estimation. H.
the models is evaluated using statistical error tests. The results Khorasanizadeh et al. [7] introduced the best model for
show that the linear and reverse models presented favourable predicting the monthly mean global solar radiation over six
efficiency, and logarithmic model showed the least accurate major cities of Iran. They used eleven models categorised
model. into three categories. The results showed that for all cities,
the best model was either from the category of the sunshine
Keywords-Solar energy; Global solar radiation; duration only or function of sunshine duration, relative
Sunshine duration; Regression analysis; 3erformance humidity and ambient temperature.
evaluation. The main objective of this paper is to compare between
common nine mathematical models to find out which is the
I. INTRODUCTION most efficient for modelling of global solar radiation on the
horizontal surface.
Solar energy as one of the renewable energy sources is
considered as most important and great natural source
[1] [2]. It is also the most promising renewable to
supply a significant part of the world’s energy demand
[3]. Solar radiation data are a fundamental input for TABLE I. REGRESSION MODELS PROPOSED IN THE LITERATURE
solar energy applications such as photovoltaic and solar Mod.
thermal systems. Accurate design of solar energy Model Expression Source
#
systems requires proper knowledge of global solar ‫ܪ‬ ݊ [1, 4, 5,
1 Linear ൌ ܽ଴ ൅ ܽଵ ቀ ቁ
radiation falling on the earth’s surface. Solar radiation ‫ܪ‬଴ ܰ 8–14]
measurements are not always available due to the cost ‫ܪ‬ ݊ ݊ ଶ [1, 4, 8–
2 Quadratic ൌ ܽ଴ ൅ ܽଵ ቀ ቁ ൅ ܽଶ ቀ ቁ
of the measuring equipment. Thus solar radiation models ‫ܪ‬଴ ܰ ܰ 10, 13]
are needed to predict the solar radiation. ‫ܪ‬ ݊ ݊ ଶ ݊ ଷ [1, 4, 8,
3 Cubic ൌ ܽ଴ ൅ ܽଵ ቀ ቁ ൅ ܽଶ ቀ ቁ ൅ ܽଷ ቀ ቁ
Several studies have been reported in the field of ‫ܪ‬଴ ܰ ܰ ܰ 9, 13]
‫ܪ‬ ݊ [4, 9,
estimation of solar radiation. A Teke et al. [1] estimated 4 Logarithmic ൌ ܽ଴ ൅ ܽଵ Žቀ ቁ
‫ܪ‬଴ ܰ 13]
the global solar radiation in Eastern Mediterranean ‫ܪ‬ ௡
ቀ ቁ
Region (EMR) using three models namely: linear, 5 Exponential ൌ ܽ଴ ൅ ܽଵ ݁ ே [4]
‫ܪ‬଴
quadratic and cubic. The results showed that the cubic Linear ‫ܪ‬ ݊ ݊
model gives the more accuracy than others. They 6 ൌ ܽ଴ ൅ ܽଵ ቀ ቁ ൅ ܽଶ Žቀ ቁ [9]
Logarithmic ‫ܪ‬଴ ܰ ܰ
recommended using cubic model for EMR. M. Mecibah et Linear ‫ܪ‬ ݊ ௡
ቀ ቁ
al. [4] introduced the most accurate model for predicting 7 ൌ ܽ଴ ൅ ܽଵ ቀ ቁ ൅ ܽଶ ݁  ே [9]
Exponential ‫ܪ‬଴ ܰ
the monthly average daily global solar radiation on a ‫ܪ‬ ݊ ௔భ
8 Power ൌ ܽ଴ ቀ ቁ [4]
horizontal surface for six Algerian cities. The obtained ‫ܪ‬଴ ܰ
results confirmed the results of the previous studies, ‫ܪ‬ ݊ ݊
9 Reverse ൌ ܽ଴ ൅ ܽଵ ቀ ቁ ൅ ܽଶ ൗቀ ቁ [15]
which indicate that the sunshine-based models are ‫ܪ‬଴ ܰ ܰ
generally more accurate than ambient
temperature-based models. In addition, the best
performance

978-1-5386-2735-8/17 $31.00 © 2017 IEEE 168


DOI 10.1109/UKSim.2017.32

Authorized licensed use limited to: UNIVERSITE DE MONASTIR. Downloaded on October 11,2022 at 20:13:55 UTC from IEEE Xplore. Restrictions apply.
TABLE II. MEASURED AND ESTIMATED MONTHLY MEAN GLOBAL SOLAR RADIATION
ഥ (MJ/‫ܕ‬૛ )
Estimated ࡴ
ഥ૙
ࡴ ഥ (MJ/‫ܕ‬૛ )

Month ഥ (h)
ࡺ ࢔ ഥ
ഥ Τࡺ Lin- Lin-
(MJ/‫ܕ‬૛ ) Measured Linear Quad. Cubic Log. Exp. Power Reverse
Log Exp
Jan 19.628 10.103 0.565 0.825 0.822 0.861 0.864 0.878 0.792 0.837 0.862 0.859 0.828
Feb 24.262 10.825 0.474 0.970 0.977 1.028 1.028 1.071 0.932 1.005 1.028 1.058 0.986
Mar 30.475 11.793 0.614 1.394 1.396 1.449 1.450 1.481 1.339 1.417 1.451 1.470 1.406
Apr 36.291 12.829 0.761 1.911 1.911 1.927 1.930 1.942 1.889 1.921 1.928 1.931 1.918
May 40.000 13.695 0.784 2.089 2.089 2.115 2.105 2.137 2.055 2.107 2.117 2.119 2.103
Jun 41.374 14.126 0.776 2.126 2.127 2.155 2.167 2.169 2.096 2.142 2.159 2.152 2.137
Jul 40.583 13.923 0.916 2.270 2.269 2.277 2.271 2.271 2.268 2.277 2.277 2.270 2.277
Aug 37.542 13.173 0.805 1.908 1.909 1.941 1.944 1.940 1.885 1.930 1.944 1.925 1.925
Sep 32.377 12.172 0.729 1.593 1.592 1.624 1.618 1.640 1.558 1.603 1.626 1.631 1.599
Oct 26.041 11.140 0.606 1.156 1.152 1.204 1.193 1.216 1.119 1.185 1.206 1.196 1.177
Nov 20.622 10.282 0.594 0.904 0.904 0.939 0.934 0.937 0.885 0.918 0.941 0.928 0.913
Dec 18.171 9.872 0.567 0.756 0.756 0.776 0.771 0.808 0.729 0.770 0.775 0.795 0.763

ଵ ு೎೔ ିு೘೔
II. GLOBAL SOLAR RADIATION ‫ ܧܲܣܯ‬ൌ σ௞௜ୀଵ ቚ ቚ ൈ ͳͲͲ (5)
௞ ு೘೔
Nine models using only the sunshine duration for
predicting the global solar radiation on a horizontal surface where ‫ܪ‬௖௜ and ‫ܪ‬௠௜ are the ݅ ୲୦ calculated and measured
are examined in this paper and given in Table I. For the values of global solar radiation, respectively, and ݇ is the
models presented in Table I, ‫ ܪ‬is the global solar radiation number of data points.
( Ȁଶ ), ‫ܪ‬଴ is the extraterrestrial radiation ( Ȁଶ ), ݊ is The RMSE (  Ȁଶ ) measures the average mismatch
the sunshine duration (h), ܰ is the day length (h), and ܽ଴ , ܽଵ , between of calculated and measured values. It is given by
ܽଶ and ܽଷ are empirically determined regression constants. [20]:
The extraterrestrial radiation on a horizontal surface, ‫ܪ‬଴ , can ଵ
ܴ‫ ܧܵܯ‬ൌ ට σ௞௜ୀଵሺ‫ܪ‬௖௜ െ ‫ܪ‬௠௜ ሻଶ (6)
be calculated by the following equation [3, 8, 9, 11, 16]: ௞

ଶସൈଷ଺଴଴ீೞ೎ ଷ଺଴஽
The correlation coefficient describes the correlation of
‫ܪ‬଴ ൌ ቀͳ ൅ ͲǤͲ͵͵…‘• ቁ two data sets. It is given as [20]:
గ ଷ଺ହ
గఠೞ
ൈ ቀ…‘•‫߱‹•ߜ•‘…׎‬௦ ൅ •‹‫ߜ‹•׎‬ቁ (1) ଵ ഥ೎ ሻሺு೘೔ ିு
ሺு೎೔ ିு ഥ೘ ሻ
ଵ଼଴ ܴൌ σ௞௜ୀଵ (7)
௞ିଵ ఙಹ೘ ఙಹ೎
where ‫ܩ‬௦௖ is the solar constant (1367 Ȁଶ ), ‫ ׎‬is the where ‫ܪ‬ ഥ௖ and ‫ܪ‬ ഥ௠ denote the mean of calculated and
latitude of the site (in degree), ߜ is the solar declination (in measured values, respectively. ߪு೎ , ߪு೘ denote the standard
degree), ߱௦ is the sunrise hour angle (in degree), and ‫ ܦ‬is the
deviation of calculated and measured global solar radiation,
number of days of the year starting from first January
respectively.
(‫ ܦ‬ൌ1,2, …, 365). ߜ and ߱௦ can be calculated as following
The RE represents the relative difference between
[1, 3, 8, 9, 11, 16]: ഥ௠ and
monthly averaged measured global solar radiation ‫ܪ‬
ଶ଼ସା஽
ߜ ൌ ʹ͵ǤͶͷ•‹ ቀ͵͸Ͳ ቁ (2) monthly average global solar radiation predicted by the
ଷ଺ହ
model ‫ܪ‬ഥ௖ . It is given as [21]:
߱௦ ൌ …‘• ିଵ ሺെ –ƒ‫ߜƒ– ׎‬ሻ (3) ഥ೎ ିு
ு ഥ೘
ܴ‫ ܧ‬ൌ ቚ ഥ೘
ቚ ൈ ͳͲͲ (8)

The daily maximum possible sunshine duration is given as
[1, 8, 9, 11, 16]: IV. RESULTS AND DISCUSSION

ܰൌ ߱௦ (4) In this study, the models presented in Table I are fitted to
ଵହ
measured values of global solar radiation on horizontal
surface, their empirical coefficients are determined using
III. STATISTICAL ERROR TESTES USED FOR least square regression where the points (daily values) of
PERFORMANCE EVALUATION ‫ ܪ‬Τ‫ܪ‬଴ along y-axis versus ݊Τܰ along x-axis were plotted for
The performance of the proposed models is evaluated each month as shown in Fig. 1.
using statistical tests such as mean absolute percentage error For each month the daily values of ‫ܪ‬଴ and ܰ were
MAPE, root mean square error RMSE, correlation determined using equations (1) and (4) respectively, ‫ܪ‬
coefficient ܴ, and relative error RE. The MAPE ( Ȁଶ ) represents the measured values of solar radiation which
represents the mean absolute percentage deviation between recorded per day, and ݊ is the measured sunshine duration
calculated and measured values. It is given by [6, 17–19]: (daily values).

169

Authorized licensed use limited to: UNIVERSITE DE MONASTIR. Downloaded on October 11,2022 at 20:13:55 UTC from IEEE Xplore. Restrictions apply.
January
0.06 February March
0.06 0.06
0.055
0.055 0.055
0.05
0.05 0.05
0.045
0.045 0.045
0.04
0.04 0.04
H/H 0

H/H0

H/H0
0.035 0.035 0.035
0.03 0.03 0.03
0.025 0.025 0.025
0.02 0.02 0.02
0.015 0.015 0.015
0.01 0.01 0.01
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
n/N n/N n/N

April May June


0.06 0.06 0.06

0.055 0.055
0.055
0.05
0.05
0.05
0.045
0.045

H/H0
H/H0

H/H0

0.045 0.04
0.04
0.035
0.04
0.035
0.03
0.035 0.03
0.025

0.03 0.02 0.025


0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
n/N n/N n/N

July August September


0.06 0.06 0.06

0.055
0.055
0.058
0.05
0.05
0.056 0.045

0.045 0.04
H/H0
H/H 0
H/H0

0.054
0.04 0.035

0.052 0.03
0.035
0.025
0.05 0.03
0.02

0.048 0.025 0.015


0.7 0.75 0.8 0.85 0.9 0.95 1 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
n/N n/N n/N

October November December


0.055 0.06 0.06

0.055 0.055
0.05
0.05 0.05
0.045
0.045
0.045
0.04 0.04
0.04
H/H 0

H/H0
H/H0

0.035 0.035
0.035
0.03
0.03
0.03
0.025
0.025
0.02 0.025

0.02 0.015 0.02

0.015 0.01 0.015


0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
n/N n/N n/N

௡ ு
Figure 1. Monthly variation between and
ே ுబ

170

Authorized licensed use limited to: UNIVERSITE DE MONASTIR. Downloaded on October 11,2022 at 20:13:55 UTC from IEEE Xplore. Restrictions apply.
Each model is used for predicting the monthly average
global solar radiation and compared with measured monthly TABLE III. SUMMARY OF RE: MEAN AND STANDARD DEVIATION
averaged global solar radiation, the result is shown in Stand. dev. of
Table II and Figure 2. Math. model Mean of RE% Rank
RE%
The RE is used to determine the error between values of Linear 0.1449 0.2067 1
monthly average measured global solar radiation and Quadratic 2.7062 1.7365 6
predicted by each model, the result is shown in Figure 3. The Cubic 2.5319 1.7439 5
mean and standard deviation of RE is illustrated in Table III. Logarithmic 4.1185 2.9786 9
Exponential 2.3801 1.3354 4
From results shown in Figure 3 and tabled in Table III, it is Linear Logarithmic 1.4092 0.9332 3
clearly seen that the linear model gives the lowest relative Linear Exponential 2.7946 1.7540 7
error followed by Reverse model. Power 3.0728 2.5737 8
The values of the empirical coefficients of two preferred Reverse 0.8047 0.4762 2
models are illustrated in Table IV. The values of MAPE,
RMSE, and ܴ are also presented in Table IV.
3
Linear
2.5 2.5 Quad
Measured Cubic
Measured and predicted radiation (MJ/m )
2

Linear Log.

Relative error (%)


Quad 2 Exp.
Cubic
2 Lin-Log
Log
1.5 Lin-Exp
Exp
Lin-Log Power
Lin-Exp 1 Reverse
1.5 Power
Reverse
0.5

1 0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month

0.5
Figure 3. The relative error of all presented models.
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month

Figure 2. The monthly average of measured and predicted values of


global solar radiation.

TABLE IV. EMPIRICAL COEFFICIENTS AND STATISTICAL ERROR TESTS OF THE NINE MODELS.
Month Model # ࢇ૙ ࢇ૚ ࢇ૛ MAPE RMSE ࡾ
1 0.01905 0.04038 0.40606 0.01373 0.32911
Jan
9 0.02096 0.03792 -0.00010 0.39196 0.01351 0.35169
1 0.01979 0.04320 0.53751 0.02182 -0.44642
Feb
9 0.02118 0.04111 -0.000001 0.52542 0.02171 -0.41788
1 0.02022 0.04166 0.62620 0.01933 -0.11682
Mar
9 0.02195 0.03938 -0.000001 0.59632 0.01890 -0.07098
1 0.02855 0.03168 0.38276 0.01087 0.06308
Apr
9 0.03281 0.02811 -0.00105 0.38283 0.01100 0.05637
1 0.02300 0.03727 0.65835 0.01686 -0.24485
May
9 0.02940 0.03120 -0.00102 0.63350 0.01679 -0.24712
1 0.02376 0.03561 0.56652 0.01248 0.34332
Jun
9 0.02887 0.03095 -0.00097 0.55435 0.01252 0.33478
1 0.03419 0.02372 0.11499 0.00335 -0.31521
Jul
9 0.16078 -0.05140 -0.05276 0.10779 0.00348 -0.37586
1 0.02140 0.03658 0.49052 0.01207 0.08140
Aug 9 0.03461 0.02612 -0.00351 0.47927 0.01236 0.03421
1 0.02284 0.03613 0.70630 0.01667 -0.08390
Sep 9 0.02526 0.03318 -0.00002 0.66353 0.01619 -0.09694
1 0.02259 0.03575 0.46380 0.01619 -0.27125
Oct 9 0.02922 0.02801 -0.00061 0.41593 0.01570 -0.23154
1 0.02114 0.03816 0.49233 0.01589 -0.15755
Nov 9 0.02491 0.03292 -0.00013 0.47219 0.01592 -0.19902
1 0.02214 0.03434 0.45883 0.01583 -0.20409
Dec 9 0.02407 0.03194 -0.00012 0.44618 0.01578 -0.18472

171

Authorized licensed use limited to: UNIVERSITE DE MONASTIR. Downloaded on October 11,2022 at 20:13:55 UTC from IEEE Xplore. Restrictions apply.
V. CONCLUSION [13] J. Almorox and C. Hontoria, “Global solar radiation estimation using
sunshine duration in Spain,” Energy Conversion and Management,
In this study, the reverse model is presented and vol. 45, pp. 1529–1535, 2004.
compared with eight mathematical models to find out which [14] K. Bakirci, “Correlations for estimation of daily global solar radiation
is the most efficient for predicting the global solar radiation. with hours of bright sunshine in Turkey,” Energy, vol. 34, pp. 485–
The accuracy of the proposed models is evaluated using 501, 2009.
statistical error tests. From the results of this study it can be [15] S. Chapra and R. Canale, Numerical Methods for Engineers.
concluded: McGraw-Hill Companies Inc. Sixth Edition, 2010.
[16] H. Khorasanizadeh and K. Mohammadi, “Diffuse solar radiation on a
(1) Among the presented mathematical models, the horizontal surface: Reviewing and categorizing the empirical
linear model was found the most accurate model models,” Renew. Sustain. Energy Rev., vol. 53, pp. 338–362, 2016.
followed by a reverse model for predicting global [17] K. Mohammadi, O. Alavi, A. Mostafaeipour, N. Goudarzi, and M.
solar radiation, since they gave the lowest mean Jalilvand, “Assessing different parameters estimation methods of
Weibull distribution to compute wind power density,” Energy
relative error in estimating monthly average global Convers. Manag., vol. 108, pp. 322–335, 2016.
solar radiation. [18] T. Boukelia, M. Mecibah, and I. Meriche, “General models for
(2) The logarithmic model was found the least accurate estimation of the monthly mean daily diffuse solar radiation (Case
study: Algeria),” Energy Convers. Manag., vol. 81, pp. 211–219,
model. 2014.
[19] T. Ayodele and A. Ogunjuyigbe, “Prediction of monthly average
global solar radiation based on statistical distribution of clearness
index,” Energy, vol. 90, pp. 1733–1742, 2015.
[20] A. A. Teyabeen, “Statistical Analysis of Wind Speed Data,” 6th Int.
REFERENCES Renewable Energy Congress (IREC), IEEE, PP. 1-6, 2015.
[1] A. Teke and H. Yldrm, “Estimating the monthly global solar [21] M. Lydia, S. S. Kumar, A. I. Selvakumar, and G. E. Prem, “A
radiation for Eastern Mediterranean Region,” Energy Convers. comprehensive review on wind turbine power curve modeling
Manag., vol. 87, pp. 628–635, 2014. techniques,” Renew. Sustain. Energy Rev., vol. 30, pp. 452–460,
2014.
[2] G. Hassan, M. Youssef, M. Ali, Z. Mohamed, and A. Shehata,
“Performance assessment of different day-of-the-year-based models
for estimating global solar radiation - Case study: Egypt,” J. Atmos.
Solar-Terrestrial Phys., vol. 149, pp. 69–80, 2016.
[3] G. Hassan, M. Youssef, Z. Mohamed, M. Ali, and A. Hanafy, “New
Temperature-based Models for Predicting Global Solar Radiation,”
Appl. Energy, vol. 179, pp. 437–450, 2016.
[4] M. Mecibah, T. Boukelia, R. Tahtah, and K. Gairaa, “Introducing the
best model for estimation the monthly mean daily global solar
radiation on a horizontal surface (Case study: Algeria),” Renew.
Sustain. Energy Rev., vol. 36, pp. 194–202, 2014.
[5] F. K. Bannani, T. A. Sharif, and A. Ben-Khalifa, “Estimation of
monthly average solar radiation in Libya,” Theor. Appl. Clim., vol.
83, pp. 211–215, 2006.
[6] H. Li, W. Ma, Y. Lian, and X. Wang, “Estimating daily global solar
radiation by day of year in China,” Appl. Energy, vol. 87, no. 10, pp.
3011–3017, 2010.
[7] H. Khorasanizadeh and K. Mohammadi, “Introducing the best model
for predicting the monthly mean global solar radiation over six major
cities of Iran,” Energy, vol. 51, pp. 257–266, 2013.
[8] H. Khorasanizadeh, K. Mohammadi, and M. Jalilvand, “A statistical
comparative study to demonstrate the merit of day of the year-based
models for estimation of horizontal global solar radiation,” Energy
Convers. Manag., vol. 87, pp. 37–47, 2014.
[9] H. Li, W. Ma, Y. Lian, X. Wang, and L. Zhao, “Global solar radiation
estimation with sunshine duration in Tibet, China,” Renew. Energy,
vol. 36, no. 11, pp. 3141–3145, 2011.
[10] A. El-Sebaii, F. Al-Hazmi, A. Al-Ghamdi, and S. Yaghmour, “Global
, direct and diffuse solar radiation on horizontal and tilted surfaces in
Jeddah, Saudi Arabia,” Appl. Energy, vol. 87, no. 2, pp. 568–576,
2010.
[11] J. Duffie and W. Beckman, Solar Engineering of Thermal Processes,
Fourth edition, John Wiley & Sons, Inc. 2013.
[12] A. Manzano, M. L. Martín, F. Valero, and C. Armenta, “A single
method to estimate the daily global solar radiation from monthly
data,” Atmos. Res., vol. 166, pp. 70–82, 2015.

172

Authorized licensed use limited to: UNIVERSITE DE MONASTIR. Downloaded on October 11,2022 at 20:13:55 UTC from IEEE Xplore. Restrictions apply.

You might also like