Journal of Advanced Research in Fluid Mechanics and Thermal Sciences 118, Issue 2 (2024) 87-100

Journal of Advanced Research in Fluid

Mechanics and Thermal Sciences
Journal homepage:
https://semarakilmu.com.my/journals/index.php/fluid_mechanics_thermal_sciences/index
ISSN: 2289-7879

Statistical Analysis of Wind Resource Assessment for Different Locations

in South-Western Thailand
Attapon Chumapan1, Paramin Neranon2, Juntakan Taweekun2,*

1 Energy Technology Program, Faculty of Engineering, Prince of Songkla University, Hat Yai, 90110, Songkhla, Thailand
2 Department of Mechanical and Mechatronics Engineering, Faculty of Engineering, Prince of Songkla University, Hat Yai, 90110, Songkhla,
Thailand

ARTICLE INFO ABSTRACT

Article history: Weibull parameters have been widely used to evaluate wind energy potential. In this work
Received 25 January 2024 presents wind resource assessment by statistical analysis with a Weibull distribution
Received in revised form 27 May 2024 model for Krabi, Phuket, and Ranong weather stations in south-western Thailand. Ten-
Accepted 8 June 2024 minute intervals include wind speed and wind direction of 10m from four-year records
Available online 30 June 2024
obtained by the Thai meteorological department. Four numerical methods, namely
empirical method, graphical method, energy pattern factor method and maximum
likelihood method are examined to estimate the Weibull parameters. The Weibull
distribution obtained from each method is compared with the observed wind speed
distribution by the performance tests using root mean square error, mean percentage
error, and chi-square error to select a suitable method for the station area. The results
revealed that the maximum likelihood method was the most accurate for Krabi and
Phuket stations, and the energy pattern factor method was the most accurate for Ranong
station. At a hub height of 80m, the highest mean wind speed and mean wind power
density found in Krabi station were 3.25 m/s and 44.84 W/m2. The most probable wind
Keywords: speed value in three stations had a range from 1.80 to 2.50 m/s. The maximum wind
Weibull distribution; Weibull speed carrying maximum energy found in Krabi station was 5.53 m/s. The operating
parameters; Empirical method; probability of a wind turbine in Krabi station was 49.61%, followed by Phuket station was
graphical method; energy pattern 46.80%, and Ranong station was 37.84%, respectively. In conclusion, all three stations had
factor method; maximum likelihood wind power potential classified as wind class 1 and can be sorted as follows: Krabi, Phuket,
method and Ranong stations.

1. Introduction

Population growth, economic progress, and technology advancement result in an increased

global demand for energy. Fossil fuel is the main resource utilized for energy. These resources
influence the environment and result in climate change. Climate change is one of the leading
problems around the world [1]. In present, renewable energy is continuously expanding. Wind energy
is a clean and inexhaustible source of renewable energy resource. Wind energy can be used to
generate electricity through wind farms and wind turbines. ln 2021, the globally installed wind
*
Corresponding author.
E-mail address: juntakan.t@psu.ac.th

https://doi.org/10.37934/arfmts.118.2.87100

87
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

capacity already amounted to 850 GW [2]. For Thailand, wind energy is one of the alternatives to
electricity generation. The Alternative Energy Development Plan 2018, Thailand has set a target for
total wind power generation of 3,000 MW by the end of 2037 [3].
The assessment of wind energy resources is an important step before deciding about a wind
turbine station or wind farm [4]. The Weibull distribution function is one of the most widely accepted
statistical functions for wind energy assessment [5]. The Weibull distribution function has two
adjustable parameters, namely k for shape and c for scale. ‘k’ shape parameter represents the
characteristic of the wind wave for a particular wind site and ‘c’ scale factor indicates the potentiality
of the wind power that site [6]. There are various methods for determining the k and c parameters.
Mahmood et al., [7] used the maximum likelihood method (MLM) to determine Weibull parameters
that described the characteristics of wind wave in the Al-Salman site in Iraq. Kaplan [8] presented the
power density method for determining Weibull distribution parameters for different location in
Turkey. Al Buhairi [9] used the standard deviation method for determining Weibull distribution
parameters on Taiz in the southwest of Yemen. In addition, several studies have studied comparative
methods for estimating Weibull parameters using statistical analysis. Alsamamra et al., [10]
compared five methods for estimating Weibull parameters and used the root mean square error, the
mean absolute percentage error, and the chi-square error to compare the accuracy of the five
methods. The results showed that the empirical method (EM) and the method of moment (MoM)
were the most accurate to approximate wind speed distribution. Azad et al., [11] used three
numerical methods consist of the power density method, the least square method, and the modified
maximum likelihood method for determining the Weibull parameters for three different sites in
Bangladesh. Chang [12] compared six numerical methods estimating the Weibull parameters for
three wind farms in Taiwan. Kang et al., [13] compared twelve numerical methods for estimating the
Weibull parameters on Maldo Island and Saemangeum Seawall in the Republic of Korea. Many
studies revealed that each location around the world had a different suitable method for estimating
the Weibull parameters.
Thailand is located near the equator and has low to moderate wind speed [14]. The previous
studies for wind energy in Thailand, Ratjiranukool and Ratjiranukool [15] explored wind energy
potential for electricity generation in Thailand by regions. The result showed that southern and
northeastern Thailand had sufficient wind speed for electricity generation. Niyomtham et al., [16]
assessed the wind energy resource in the central region of Thailand for wind power generation.
Werapun et al., [17] compared five numerical methods: the empirical method, the energy pattern
factor method, the maximum likelihood method, the modified maximum likelihood method, and the
graphical method for estimating the Weibull parameters on Phangan Island in Thailand. The south-
western part of Thailand is an interesting region to study for wind energy potential. It is located near
the Andaman Sea, has a total area of 17,689 km2, and the topology is coastal mountains and coastal
[18]. However, finding a suitable Weibull distribution model using statistical methods to evaluate
wind resources in southern-western Thailand is still less.
The aim of this work is to select a suitable method to estimate Weibull parameters using the
statistical methods and evaluate wind resources using the Weibull distribution model for Krabi,
Phuket, and Ranong weather observing stations in southern-western Thailand. Four numerical
methods, namely the empirical method (EM), the graphical method (GM), the energy pattern factor
method (EPF) and the maximum likelihood method (MLM) are examined to estimate the Weibull
parameters. The Weibull distribution obtained from these methods is compared with the observed
wind speed distribution by the accuracy tests using root mean square error (RMSE), mean percentage
error (MPE), and chi-square error (𝑥 2 ) to select a suitable method for the station area.

88
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

2. Methodology
2.1 Wind Data Collection

Thailand has 128 Meteorological Department weather stations classified into six regions. In this
study, the wind data of three stations in the southern-western part of Thailand, consisting of Krabi
stations, Phuket station and Ranong station, are adopted for statistical analysis with Weibull
parameters estimation and assessment of wind energy potential. The wind data including wind speed
and wind direction were provided by the Thai Meteorological Department. The collection period of
raw wind data was every 10 minutes from 2019 to 2022 (4 years) at 10m above ground level. The
geographical coordinates of the selected weather stations are shown in Table 1. The overview
procedure of this study is shown in Figure 1 and the wind power class is shown in Table 2.

Table 1
Summarization of study area sites
Station name Latitude (°) Longitude (°) Altitude (m a.s.l) Zone Measurement period
Krabi 7.884 98.392 30 47 2019 - 2022
Phuket 8.103 98.975 8 47 2019 - 2022
Ranong 9.955 98.634 39 47 2019 - 2022

Fig. 1. Overview procedure of the study

Table 2
Wind Power Class at height 10m and 80m [19]
Wind At height 10m At height 80m
class Density (W/m2) Speed (m/s) Density (W/m2) Speed (m/s)
1 < 100 < 4.4 < 240 < 5.9
2 100 – 150 4.4 – 5.1 240 – 380 5.9 – 6.9
3 150 – 200 5.1 – 5.6 380 – 490 6.9 – 7.5
4 200 – 250 5.6 – 6.0 490 – 620 7.5 – 8.1
5 250 – 300 6.0 – 6.4 620 – 740 8.1 – 8.6
6 300 – 400 6.4 – 7.0 740 – 970 8.6 – 9.4
7 > 400 > 7.0 > 970 > 9.4

2.2 Methods to Estimate Weibull Parameters

The Weibull distribution has been widely used to describe wind speed frequency distribution and
to estimate wind energy potential [20]. The Weibull distribution can be divided into probability
density function 𝑓(𝑣) and cumulative distribution 𝐹(𝑣). The Weibull probability density function for
fitting the wind speed is shown in Eq. (1) and the cumulative distribution is shown in Eq. (2)

89
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

𝑘 𝑣 𝑘−1 𝑣 𝑘
𝑓(𝑣) = ( 𝑐 ) (𝑐 ) 𝑒𝑥𝑝 [− (𝑐 ) ] (1)

𝑣 𝑘
𝐹(𝑣) = 1 − 𝑒𝑥𝑝 [− (𝑐 ) ] (2)

where (𝑣 > 0; 𝑘, 𝑐 > 0), 𝑣 is wind speed (m/s) and c is a scale parameter and k is a shape parameter.
There are various methods to estimate the Weibull parameters, shape k and scale c, the four
different numerical methods were used to estimate the Weibull parameters, which are explained
below [21].

2.2.1 Empirical method (EM)

The empirical method is considered a special case of the moment of methods, where k and c
parameters can be determined using mean wind speed and standard deviation as follows Eq. (3) and
Eq. (4) [22].

𝜎 −1.086
𝑘 = (𝑣) (3)

𝑣
𝑐=( 1 ) (4)
𝛤(1+ )
𝑘

where, 𝑣 is mean wind speed (m/s), 𝜎 is standard variation of observed data (m/s), 𝛤 is gamma
function.

2.2.2 Graphical method (GM)

The graphical method is to convert the cumulative function into a linear equation by using the
logarithmic function [23]. The cumulative function is shown in Eq. (2) and taking natural logarithms
twice gives Eq. (5).

𝑙𝑛[−𝑙𝑛 {1 − 𝐹(𝑣)}] = −𝑘𝑙𝑛(𝑐) + 𝑘𝑙𝑛(𝑣) (5)

where, 𝐹(𝑣) is a probability of observing wind speed, 𝑣 is wind speed (m/s) k is a shape parameter
equals the slope of the line, and c is a scale parameter obtained from −𝑘𝑙𝑛(𝑐) = the intercept with a
y-axis.

2.2.3 Energy pattern factor method (EPFM)

The energy pattern factor method can determine the Weibull parameters following on the ratio
of the wind power energy to the third power of the average wind speed values [24]. Eq. (6) presents
the energy pattern factor method. Eq. (4) and Eq. (7) present the Weibull parameters estimation.

1
𝑣3 ( ∑𝑛 3
𝑖=1 𝑣𝑖 )
𝑛
𝐸𝑝𝑓 = ( 3 ) = 1 3 (6)
𝑣 ( ∑𝑛 𝑣)
𝑛 𝑖=1 𝑖

90
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

3.69
𝑘 =1+( 2 ) (7)
(𝐸𝑝𝑓 )

Where, 𝐸𝑝𝑓 is the energy pattern factor, 𝑣 is mean wind speed (m/s).

2.2.4 Maximum likelihood method (MLM)

The maximum likelihood method is a mathematical-based expression known as a likelihood

function of wind speed data presented in time-series, it is widely utilized in the literature to estimate
the Weibull parameters, Eq. (8) and Eq. (9) present the maximum likelihood method [25].

−1
∑𝑛 𝑘
𝑖=1 𝑣𝑖 𝑙𝑛 (𝑣𝑖 ) ∑𝑛
𝑖=1 𝑙𝑛(𝑣𝑖 )
𝑘=[ ∑𝑛 𝑘 − ] (8)
𝑖=1 𝑣𝑖 𝑛

1
∑𝑛 𝑘 𝑘
𝑖=1 𝑣𝑖
𝑐=( ) (9)
𝑛

where, 𝑣𝑖 is wind speed measured at the interval 𝑖 (m/s), 𝑖 is the measurement interval, n is the
number of non-zero values.

2.3 Performance Tests

Three different indicators were used to assess the performance of methods for estimating
Weibull parameters. These statistical tools were widely used to compare wind speed data-fitting [10].
The suitable method for estimating Weibull parameters for a study area must have the lowest error
ranking among these three statistical tools. Three different indicators show in Eq. (10), Eq. (11) and
Eq. (12).

2.3.1 Root mean square error (RMSE)

1
1 2
𝑅𝑀𝑆𝐸 = [𝑛 ∑𝑛𝑖=1(𝑦𝑖 − 𝑥𝑖 )2 ] (10)

where, 𝑦𝑖 is the actual wind speed value, 𝑥𝑖 is the predicted wind speed value using the Weibull
distribution, and n is the number of records in the wind speed data. The result is close to zero,
indicating good performance because it has little error [26].

2.3.2 Mean percentage error (MPE)

1 𝑦𝑖 −𝑥𝑖
𝑀𝑃𝐸 = 𝑛 ∑𝑛𝑖=1 ( ) ∗ 100 (11)
𝑦𝑖

where, 𝑦𝑖 is the actual wind speed value, 𝑥𝑖 is the predicted wind speed value using the Weibull
distribution, and n is the number of records in the wind speed data. The result is close to zero,
indicating good performance because it has little error [27].

91
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

2.3.3 Chi-square error ( 𝜒 2 )

(𝑦𝑖 −𝑥𝑖 )2
𝜒 2 = ∑𝑛𝑖=1 (12)
𝑦𝑖

where, 𝑦𝑖 is the actual wind speed value, 𝑥𝑖 is the predicted wind speed value using the Weibull
distribution. The chi-square error returns the mean square of the measure value and calculate value
for the distributions [10].

2.4 Variation of Weibull Parameter with Height

The Weibull parameters value can be adjusted to another hup height by following Eq. (13), Eq.
(14) and Eq. (15) [28].

𝑧
𝑘𝑎 [1−𝑘𝑎 0.088 𝑙𝑛( 𝑎 )]
10
𝑘𝑧 = 𝑧 (13)
[1−0.088 𝑙𝑛( )]
10

𝑧
𝑐𝑧 = 𝑐𝑎 (𝑧 )𝑛 (14)
𝑎

[0.37−0.088 𝑙𝑛(𝑐𝑎 )]
𝑛= 𝑧 (15)
[1−0.088 𝑙𝑛( 𝑎 )]
10

where, 𝑘𝑧 is Weibull shape parameter at height Z, 𝑐𝑧 is Weibull scale parameter at height Z (m/s), 𝑘𝑎
is Weibull shape parameter at anemometer height 𝑐𝑎 is Weibull scale parameter at anemometer
height (m/s), n is power law exponent.

2.5 Wind Power Density Base on Weibull Distribution Function (𝑃𝑊 )

The Weibull parameters can be used to determine the wind power density (wind power per unit
area) [29]. The calculation of wind power density using the Weibull parameters method is expressed
as Eq. (16).

1 3
𝑃𝑊 = 2 𝜌𝑐 3 𝛤(1 + 𝑘) (16)

where, 𝜌 is the air density (often adopted as 1.225 kg/m3), 𝛤 is gamma function, k and c are the
Weibull parameters.

2.6 Most Probable Wind Speed (𝑉𝑚𝑝 ) and Wind Speed Carrying Maximum Energy (𝑉𝑚𝑎𝑥 𝐸 )

𝑉𝑚𝑝 represent the most frequently occurring wind speed for the wind speed probability
distribution. 𝑉𝑚𝑝 is expressed in Eq. (17). 𝑉max 𝐸 represent wind speed carrying maximum energy,
which is important for the designed wind speed of a wind turbine [30]. 𝑉max 𝐸 is expressed in Eq. (18).

1 1
𝑉𝑚𝑝 = 𝑐(1 − 𝑘)𝑘 (17)

92
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

2 1
𝑉𝑚𝑎𝑥 𝐸 = 𝑐(1 + 𝑘)𝑘 (18)

where, k and c are the Weibull parameters.

2.7 Operating Probability of Wind Turbine (𝑃𝑜𝑝 )

In the wind power industry, wind turbines are used to convert wind energy into electrical energy.
In general, wind turbines have two specific characteristics of wind speed: cut-in speed and cut-off
speed. The cut-in speed is the minimum wind speed at which the wind turbine will operate to
generate power. The cut-off speed is the wind speed when the turbine stops working. The operating
probability of wind turbines is an important factor in assessing the cost-effectiveness of installing a
wind turbine [28]. (𝑃𝑜𝑝 ) is expressed as Eq. (19).

𝑣 𝑘 𝑣 𝑘
𝑃𝑜𝑝 (𝑣1 < 𝑣 < 𝑣2 ) = 𝑒𝑥𝑝 [− ( 𝑐1) ] − 𝑒𝑥𝑝 [− ( 𝑐2 ) ] (19)

where, 𝑣1 and 𝑣2 represent cut-in and cut off wind speed, k and c are the Weibull parameters.

3. Results
3.1 Statistical Analysis

This section describes in detail the Weibull parameters estimation. In this study, the Weibull
distribution model was used to assess wind energy potential. The Weibull distribution model had two
significant parameters: k and c. The k and c parameters are obtained from four numerical methods:
EM, GM, EPFM, and MLM. Appropriate methods for each study area were selected using
performance tests to assess accuracy, including RMSE, MPE, and 𝑥 2 , by comparing the Weibull
distribution obtained from each method with the observed wind speed distribution.
Table 3 shows the assessment accuracy of Weibull parameters estimation from different methods
for Krabi station by year and the whole year. The results reveal that in 2019, the energy pattern factor
method gave the lowest error values for both RMSE, MPE, and χ². In 2020, The maximum likelihood
method had the most accuracy from the other methods. In 2021 The maximum likelihood method
gave the lowest error values for both RMSE, MPE, and χ². In 2022 The maximum likelihood method
had the most accuracy of the other methods. The whole year shows that the Weibull parameters
were k = 1.941 and c =1.9057 m/s for EM, k = 1.2740 and c =1.3183 m/s for GM, k = 1.4917 and c
=1.9053 m/s for EPFM, and k = 1.4730 and c = 1.9042 m/s for MLM, which indicated that the
maximum likelihood method had the best performance of each method for Krabi station. Figure 2
shows the Weibull frequency distribution according to each method and frequency of measured wind
speed for Krabi station for four years (2019 – 2022).

93
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

Table 3
Assessment accuracy Weibull parameters four methods for Krabi station
Year Method Weibull parameters Performance tests Ranking
k c (m/s) RMSE MPE (%) χ²
2019 Empirical 1.5060 1.8948 0.0149 2.6556 0.0616 3
Graphical 1.2635 1.4191 0.0636 12.9643 0.6165 4
Energy pattern factor 1.4960 1.8933 0.0142 2.4827 0.0547 1
Maximum likelihood 1.4977 1.8971 0.0146 2.5561 0.0581 2
2020 Empirical 1.4673 1.8829 0.0165 3.8269 0.1104 2
Graphical 1.2398 1.3580 0.0739 11.9610 0.7037 4
Energy pattern factor 1.4719 1.8836 0.0168 3.8934 0.1157 3
Maximum likelihood 1.4353 1.8779 0.0144 3.3676 0.0777 1
2021 Empirical 1.5902 2.2297 0.0174 4.6527 0.1376 2
Graphical 1.3229 1.8366 0.0516 9.3488 0.3971 4
Energy pattern factor 1.5995 2.2310 0.0179 4.8022 0.1494 3
Maximum likelihood 1.5473 2.2227 0.0150 3.9847 0.0903 1
2022 Empirical 1.4940 1.7606 0.0156 3.0532 0.0722 3
Graphical 1.2234 1.2397 0.0747 13.9317 0.7824 4
Energy pattern factor 1.4866 1.7595 0.0151 2.9396 0.0664 2
Maximum likelihood 1.4803 1.7612 0.0149 2.8753 0.0634 1
Whole Empirical 1.4941 1.9057 0.0149 3.0525 0.0830 3
Graphical 1.2740 1.3183 0.0797 13.3542 0.9130 4
Energy pattern factor 1.4917 1.9053 0.0148 3.0180 0.0808 2
Maximum likelihood 1.4730 1.9042 0.0136 2.7661 0.0659 1

Fig. 2. Weibull distribution generate by 4 methods and

measure wind speed as histogram for Krabi station for 4
years

Table 4 shows the assessment accuracy of Weibull parameters estimation from different methods
for Phuket station by year and the whole year. The results reveal that from 2019 to 2022, The
maximum likelihood method had the lowest values of the root mean square error, the mean
percentage error, and the chi-square error, followed by the energy pattern factor method and the
empirical method, which had a similar Weibull parameters value, and the graphical method showed
the worst performance of the four methods. The whole year shows that the Weibull parameters were
k = 1.7102 and c = 1.7679 m/s for EM, k = 1.3706 and c =1.1626 m/s for GM, k = 1.7068 and c =1.7677
m/s for EPFM, k = 1.6720 and c =1.7639 m/s for MLM. Figure 3 shows the Weibull frequency
distribution according to each method and frequency of measured wind speed for Phuket station for
four years (2019 – 2022).

94
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

Table 4
Assessment accuracy Weibull parameter four methods for Phuket station
Year Method Weibull parameters Performance tests Ranking
k c (m/s) RMSE MPE (%) χ²
2019 Empirical 1.6938 1.7539 0.0151 4.9662 0.1277 2
Graphical 1.4125 1.1924 0.1013 17.9688 1.0522 4
Energy pattern factor 1.6994 1.7543 0.0154 5.0461 0.1335 3
Maximum likelihood 1.6516 1.7493 0.0127 4.3389 0.0876 1
2020 Empirical 1.6311 1.6230 0.0140 3.9418 0.1026 3
Graphical 1.2948 1.1036 0.0891 16.0598 0.9373 4
Energy pattern factor 1.6242 1.6224 0.0136 3.8594 0.0964 2
Maximum likelihood 1.5989 1.6201 0.0123 3.5584 0.0755 1
2021 Empirical 1.9589 1.9422 0.0117 6.1239 0.1446 3
Graphical 1.6389 1.7463 0.0350 7.9246 0.1896 4
Energy pattern factor 1.9583 1.9422 0.0116 6.1151 0.1440 2
Maximum likelihood 1.9141 1.9372 0.0101 5.3498 0.1060 1
2022 Empirical 1.7278 1.8560 0.0117 4.0494 0.0887 3
Graphical 1.3758 1.4382 0.0647 12.9292 0.6211 4
Energy pattern factor 1.7174 1.8553 0.0112 3.9204 0.0804 2
Maximum likelihood 1.6955 1.8530 0.0101 3.6229 0.0641 1
Whole Empirical 1.7102 1.7679 0.0123 3.9490 0.1119 3
Graphical 1.3706 1.1626 0.0994 17.7717 1.2622 4
Energy pattern factor 1.7068 1.7677 0.0121 3.9110 0.1088 2
Maximum likelihood 1.6720 1.7639 0.0104 3.4920 0.0787 1

Fig. 3. Weibull distribution generate by 4 methods and

measure wind speed as histogram for Phuket station for 4
years

Table 5 shows the assessment accuracy of Weibull parameters estimation from different methods
for Ranong station by year and the whole year. The results reveal that in 2019, the maximum
likelihood method gave the lowest error values for both RMSE, MPE, and χ². In 2020 – 2022, the
energy pattern factor method gave the lowest error values for both RMSE, MPE, and χ². The whole
year shows that the Weibull parameters k = 1.4020 and c =1.5286 m/s for EM, k = 1.1399 and c =
0.9855 m/s for GM, k = 1.3944 and c =1.5286 m/s for EPFM, k = 1.4139 and c =1.5355 m/s for MLM,
which indicated that the energy pattern factor method had the best performance of each method
for Ranong station. Figure 4 shows the Weibull frequency distribution according to each method and
frequency of measured wind speed for Ranong station for four years (2019 – 2022).

95
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

Table 5
Assessment accuracy Weibull parameter four methods for Ranong station
Year Method Weibull parameters Performance tests Ranking
k c (m/s) RMSE MPE (%) χ²
2019 Empirical 1.4026 1.5943 0.0175 3.3442 0.0855 2
Graphical 1.1533 1.0322 0.0879 15.0033 0.9018 4
Energy pattern factor 1.4074 1.5951 0.0179 3.4194 0.0901 3
Maximum likelihood 1.3886 1.5944 0.0166 3.1452 0.0747 1
2020 Empirical 1.3908 1.4646 0.0195 1.9701 0.0636 2
Graphical 1.1504 1.0835 0.0533 11.8419 0.4706 4
Energy pattern factor 1.3763 1.4621 0.0187 1.9736 0.0577 1
Maximum likelihood 1.4197 1.4755 0.0220 2.1962 0.0837 3
2021 Empirical 1.4520 1.5669 0.0175 2.1419 0.0545 2
Graphical 1.1967 1.0397 0.0840 18.2373 0.8770 4
Energy pattern factor 1.4356 1.5644 0.0164 1.9105 0.0465 1
Maximum likelihood 1.4652 1.5739 0.0190 2.4723 0.0660 3
2022 Empirical 1.3874 1.4966 0.0184 1.9364 0.0607 2
Graphical 1.1252 1.0491 0.0643 13.5101 0.6272 4
Energy pattern factor 1.3782 1.4950 0.0178 1.8262 0.0560 1
Maximum likelihood 1.4085 1.5058 0.0204 2.2989 0.0774 3
Whole Empirical 1.4020 1.5286 0.0169 2.0515 0.0603 2
Graphical 1.1399 0.9855 0.0792 14.8207 0.9073 4
Energy pattern factor 1.3944 1.5286 0.0164 1.9328 0.0558 1
Maximum likelihood 1.4139 1.5355 0.0182 2.2927 0.0718 3

Fig. 4. Weibull distribution generate by 4 methods and

measure wind speed as histogram for Ranong station for 4
years

Table 6 shows summarization of suitable method for estimating Weibull parameters. The
maximum likelihood method was suitable for Krabi and Phuket station. The energy pattern factor
method was suitable for Ranong station. The calculation about Weibull shape parameter k varied
from 1.40 to 1.70 and the range of Weibull scale parameter c was 1.50 to 2.00 m/s. The scale
parameter c is highly consistent with the mean wind speed. The maximum scale parameter c was
found in Krabi station, and the lowest scale parameter c was found in Ranong station. Figure 5 shows
a comparison of the Weibull distribution model for Krabi, Phuket and Ranong station for four years
(2019 – 2022).

96
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

Table 6
Summarization of suitable method for estimating the Weibull parameters of 10m
Station Method estimates Weibull Weibull parameter
parameters k c (m/s)
Krabi Maximum likelihood 1.4730 1.9042
Phuket Maximum likelihood 1.6720 1.7639
Ranong Energy pattern factor 1.3944 1.5273

Fig. 5. Weibull distribution model for three stations of 10m

3.2 Evaluate of Wind Resource

This section describes in detail the assessment of wind potential by the Weibull model. The
standard wind turbine installed hub height is in the range of 80 to 100 meters [31,32]. In this study,
a reference hup height of 80m was used to assess the wind energy potential.
Figure 6 plots the Weibull distribution model for three stations of 80m. Table 7 shows adjusting
the parameters k and c to the hub height of 80m and assessing wind energy potential. The Weibull
parameters k value was 1.8029, 2.0465, and 1.7067, and the parameter c value were 3.6533 m/s,
3.4317 m/s, and 3.0508 m/s for Krabi, Phuket, and Ranong stations, respectively. The mean wind
speed was 3.25 m/s for Krabi, 3.04 m/s for Phuket, and 2.72 m/s for Ranong station. The mean power
density was 44.84 W/m2 for Krabi, 32.13 W/m2 for Phuket, and 28.15 W/m2 for Ranong station. The
wind speed carrying maximum energy found in Krabi station was 5.53 m/s. In terms of the most
probable wind speed, all three stations had a range from 1.80 to 2.50 m/s, which is in the range of
low wind speed. In this study, the selected cut-in speed is 3 m/s, which is generally the minimum
wind speed for wind turbine working, while the selected cut-out speed is constant at 25 m/s. The
operating probability of a wind turbine at Krabi station was 49.61%, followed by Phuket station was
46.80%, and Ranong station was 37.84%, respectively.

Table 7
Wind resource assessment from the Weibull distribution model of 80m
Station Weibull parameters 𝑣 𝑃𝑊 𝑉max 𝐸 𝑉𝑚𝑝 𝑃𝑜𝑝 (3 < v < 25)
k c (m/s) (m/s) (W/m2) (m/s) (m/s) (%)
Krabi 1.8029 3.6533 3.2486 44.8403 5.5269 2.3325 49.6068
Phuket 2.0465 3.4317 3.0402 32.1349 4.7883 2.4727 46.7909
Ranong 1.7067 3.0508 2.7213 28.1521 4.8059 1.8199 37.8423

97
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

Fig. 6. Weibull distribution model for three stations of 80m

4. Conclusions

This study presents an estimation of Weibull parameters using statistical methods and evaluates
wind resources using the Weibull distribution model for Krabi, Phuket, and Ranong stations in
southern-western Thailand. The main outcome presented as follows:
The maximum likelihood method was the most accurate for determining Weibull parameters for
Krabi and Phuket stations. The energy pattern factor method was the most accurate for determining
Weibull parameters for Ranong station. The graphical method was the worst performance.
At a hub height of 80m, the maximum mean wind speed and the maximum wind power density
were found in Krabi station, which are 3.25 m/s and 44.84 W/m2. The wind speed carrying maximum
energy found in Krabi station was 5.53 m/s. The most operating probability of a wind turbine found
at Krabi station was 49.61%. All three stations had wind power potential classified as wind class 1 and
can be sorted as follows: Krabi, Phuket, and Ranong stations.
This study does not include wind turbine installation simulations and economic calculations for a
better assessment and exploitation of the wind energy potential in Thailand. More studies are still
needed in the future, such as using another statistic function besides the Weibull function to evaluate
the wind speed profile and using numerical simulation by computational fluid dynamics (CFD) to
determine the spatial distributions of wind speeds over a weather observing station, which can offer
valuable information for wind turbine installation.

Acknowledgement
The authors would like to thank the Thai Meteorological Department (TMD) for sharing valuable wind
data and their gratitude to the Energy Technology Program Faculty of Engineering at Prince of
Songkla University, Hat Yai, Songkhla 90110, Thailand, for their support.

References
[1] Shang, Yunfeng, Ding Han, Giray Gozgor, Mantu Kumar Mahalik, and Bimal Kishore Sahoo. "The impact of climate
policy uncertainty on renewable and non-renewable energy demand in the United States." Renewable Energy 197
(2022): 654-667. https://doi.org/10.1016/j.renene.2022.07.159
[2] Global Wind Energy Council. "Global Wind Report 2022." Global Wind Energy Council GWEC. April 4, 2022.
https://gwec.net/global-wind-report-2022/.
[3] Department of Alternative Energy Development and Efficiency. "THAILAND: Alternative Energy Development Plan
2018-2037 (AEDP 2018-2037)." Ministry of Energy (TH), Bangkok, 2020.

98
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

[4] Dayal, Kunal K., John E. Cater, Michael J. Kingan, Gilles D. Bellon, and Rajnish N. Sharma. "Wind resource assessment
and energy potential of selected locations in Fiji." Renewable Energy 172 (2021): 219-237.
https://doi.org/10.1016/j.renene.2021.03.034
[5] Khahro, Shahnawaz Farhan, Kavita Tabbassum, Amir Mahmood Soomro, Lei Dong, and Xiaozhong Liao. "Evaluation
of wind power production prospective and Weibull parameter estimation methods for Babaurband, Sindh
Pakistan." Energy Conversion and Management 78 (2014): 956-967.
https://doi.org/10.1016/j.enconman.2013.06.062
[6] Azad, Abul Kalam, M. G. Rasul, M. M. Alam, SM Ameer Uddin, and Sukanta Kumar Mondal. "Analysis of wind energy
conversion system using Weibull distribution." Procedia Engineering 90 (2014): 725-732.
https://doi.org/10.1016/j.proeng.2014.11.803
[7] Mahmood, Faleh H., Ali K. Resen, and Ahmed B. Khamees. "Wind characteristic analysis based on Weibull
distribution of Al-Salman site, Iraq." Energy Reports 6 (2020): 79-87. https://doi.org/10.1016/j.egyr.2019.10.021
[8] Kaplan, Yusuf Alper. "Performance assessment of Power Density Method for determining the Weibull Distribution
Coefficients at three different locations." Flow Measurement and Instrumentation 63 (2018): 8-13.
https://doi.org/10.1016/j.flowmeasinst.2018.07.004
[9] Al Buhairi, Mahyoub H. "A statistical analysis of wind speed data and an assessment of wind energy potential in
Taiz-Yemen." Assiut University Bulletin For Environmental Researches 9, no. 2 (2006): 21-33.
https://doi.org/10.21608/auber.2006.150274
[10] Alsamamra, Husain R., Saeed Salah, Jawad AH Shoqeir, and Ali J. Manasra. "A comparative study of five numerical
methods for the estimation of Weibull parameters for wind energy evaluation at Eastern Jerusalem, Palestine."
Energy Reports 8 (2022): 4801-4810. https://doi.org/10.1016/j.egyr.2022.03.180
[11] Azad, Abul Kalam, M. G. Rasul, Rubayat Islam, and Imrul R. Shishir. "Analysis of wind energy prospect for power
generation by three Weibull distribution methods." Energy Procedia 75 (2015): 722-727.
https://doi.org/10.1016/j.egypro.2015.07.499
[12] Chang, Tian Pau. "Performance comparison of six numerical methods in estimating Weibull parameters for wind
energy application." Applied Energy 88, no. 1 (2011): 272-282. https://doi.org/10.1016/j.apenergy.2010.06.018
[13] Kang, Sangkyun, Ali Khanjari, Sungho You, and Jang-Ho Lee. "Comparison of different statistical methods used to
estimate Weibull parameters for wind speed contribution in nearby an offshore site, Republic of Korea." Energy
Reports 7 (2021): 7358-7373. https://doi.org/10.1016/j.egyr.2021.10.078
[14] Chingulpitak, Sakkarin, and Somchai Wongwises. "Critical review of the current status of wind energy in Thailand."
Renewable and Sustainable Energy Reviews 31 (2014): 312-318. https://doi.org/10.1016/j.rser.2013.11.038
[15] Ratjiranukool, Sujittra, and Pakpoom Ratjiranukool. "Wind speed projections for electricity application over
Thailand." Energy Procedia 79 (2015): 423-429. https://doi.org/10.1016/j.egypro.2015.11.513
[16] Niyomtham, Lattawan, Jompob Waewsak, Chuleerat Kongruang, Somphol Chiwamongkhonkarn, Chana Chancham,
and Yves Gagnon. "Wind power generation and appropriate feed-in-tariff under limited wind resource in central
Thailand." Energy Reports 8 (2022): 6220-6233. https://doi.org/10.1016/j.egyr.2022.04.068
[17] Werapun, Warit, Yutthana Tirawanichakul, and Jompob Waewsak. "Comparative study of five methods to estimate
Weibull parameters for wind speed on Phangan Island, Thailand." Energy Procedia 79 (2015): 976-981.
https://doi.org/10.1016/j.egypro.2015.11.596
[18] Niyomtham, Lattawan, Charoenporn Lertsathittanakorn, Jompob Waewsak, and Yves Gagnon. "On the wind
resource assessment along the western coast of Thailand." Energy Procedia 138 (2017): 1190-1195.
https://doi.org/10.1016/j.egypro.2017.10.387
[19] Oh, Ki-Yong, Ji-Young Kim, Jae-Kyung Lee, Moo-Sung Ryu, and Jun-Shin Lee. "An assessment of wind energy
potential at the demonstration offshore wind farm in Korea." Energy 46, no. 1 (2012): 555-563.
https://doi.org/10.1016/j.energy.2012.07.056
[20] Oyewole, J. A., F. O. Aweda, and D. Oni. "Comparison of three numerical methods for estimating weibull parameters
using Weibull distribution model in Nigeria." Nigerian Journal of Basic and Applied Sciences 27, no. 2 (2019): 8-15.
https://doi.org/10.4314/njbas.v27i2.2
[21] De Andrade, Carla Freitas, Hely Falcão Maia Neto, Paulo Alexandre Costa Rocha, and Maria Eugênia Vieira da Silva.
"An efficiency comparison of numerical methods for determining Weibull parameters for wind energy applications:
A new approach applied to the northeast region of Brazil." Energy Conversion and Management 86 (2014): 801-
808. https://doi.org/10.1016/j.enconman.2014.06.046
[22] Kaoga, D. Kidmo, R. Danwe, S. Doka Yamigno, and N. Djongyang. "Performance analysis of methods for estimating
Weibull parameters for wind speed distribution in the district of Maroua." Journal of Fundamental and Applied
Sciences 6, no. 2 (2014): 153-174. https://doi.org/10.4314/jfas.v6i2.3

99
Journal of Advanced Research in Fluid Mechanics and Thermal Sciences
Volume 118, Issue 2 (2024) 87-100

[23] Murthy, K. S. R., and O. P. Rahi. "Estimation of Weibull parameters using graphical method for wind energy
applications." In 2014 Eighteenth National Power Systems Conference (NPSC), pp. 1-6. IEEE, 2014.
https://doi.org/10.1109/NPSC.2014.7103858
[24] Dogara, M. D., H. O. Aboh, P. M. Gyuk, and M. K. O. Onwumere. "The use of energy pattern factor (EPF) in estimating
wind power density." Science World Journal 11, no. 3 (2016): 27-30.
[25] Shaban, Auday H., Ali K. Resen, and Nathalie Bassil. "Weibull parameters evaluation by different methods for
windmills farms." Energy Reports 6 (2020): 188-199. https://doi.org/10.1016/j.egyr.2019.10.037
[26] Kang, Dongbum, Kyungnam Ko, and Jongchul Huh. "Comparative study of different methods for estimating Weibull
parameters: a case study on Jeju Island, South Korea." Energies 11, no. 2 (2018): 356.
https://doi.org/10.3390/en11020356
[27] Kaplan, Yusuf Alper. "Determination of Weibull parameters using the standard deviation method and performance
comparison at different locations." Scientia Iranica 27, no. 6 (2020): 3075-3083.
[28] Shu, Z. R., Q. S. Li, and P. W. Chan. "Statistical analysis of wind characteristics and wind energy potential in Hong
Kong." Energy Conversion and Management 101 (2015): 644-657.
https://doi.org/10.1016/j.enconman.2015.05.070
[29] Promsen, W., S. Janjai, and T. J. E. P. Tantalechon. "An analysis of wind energy potential of Kampot Province,
Southern Cambodia." Energy Procedia 52 (2014): 633-641. https://doi.org/10.1016/j.egypro.2014.07.119
[30] Kaoga, Dieudonné Kidmo, Raidandi Danwe, Serge Yamigno Doka, and Noël Djongyang. "Statistical analysis of wind
speed distribution based on sixWeibull Methods for wind power evaluation in Garoua, Cameroon." Journal of
Renewable Energies 18, no. 1 (2015): 105-125.
[31] Holt, Eric, and Jun Wang. "Trends in wind speed at wind turbine height of 80 m over the contiguous United States
using the North American Regional Reanalysis (NARR)." Journal of Applied Meteorology and Climatology 51, no. 12
(2012): 2188-2202. https://doi.org/10.1175/JAMC-D-11-0205.1
[32] Nedaei, Mojtaba. "Wind resource assessment in Abadan airport in Iran." International Journal of Renewable Energy
Development 1, no. 3 (2012): 87-97. https://doi.org/10.14710/ijred.1.3.87-97

100