1. Introduction
The cost of the industrial revolution is a massive increase in the use of fossil fuels, which include coal, natural gas, oil, and gasoline, resulting in highly infected air. Due to this, global climate is changing and is affected severely; high amounts of greenhouse gases are emitted due to the mismanaged use of these fuels in a vast range of applications that can be found in producing and consuming energy. Carbon dioxide (CO
2) is one of the most common of these gases and is the leading cause of these changes in the behavior of the earth’s atmosphere. However, even this intensive use of resources cannot account for the high amounts of energy consumption of today’s age. Therefore, alternate resources need to be used to generate environmentally sustainable energy [
1,
2]. The most economically active energy resources are the solar, biomass, wind, and geothermal power. Among these resources, the wind power is the most affordable, efficient, and commonly used by many developed and developing countries and hence is considered as the first choice of energy resources [
3,
4]. Thirty-five percent of the world’s energy demand can be fulfilled if all the available wind power is consumed [
5]. The installed wind energy capacity all over the globe increased to 539 GW by the end of 2017, which indicates the need and the worth of this energy source for the future [
6]. The Global Wind Energy Council (GWEC), responsible for wind power industry trade throughout the world, is making efforts to make wind power a dominant resource for generating energy. According to the reports from the GWEC, China is the most dominant in making efforts to enforce changes and making a switch to new energy resources. Achieving 34% of the annual market, China installed 19.66 GW in 2017 and is currently the market leader with 188 GW power generated from the wind resources. The United States is in second place and Germany is in third with 89 GW and 56 GW, respectively [
7]. With Pakistan’s negligible oil and gas reserves and high costs required for nuclear energy production, the use of alternate energy sources is essential to fulfilling the energy deficit of the country. The Pakistan Meteorological Department (PMD) conducts surveys to locate optimal locations for the wind and other alternative energy resources throughout the country. According to these reports, a potential corridor for good wind energy generation is about 1100 km, on the coastline of the Sindh and Baluchistan provinces [
8].
In order to obtain the maximum efficiency and minimize investment risk of a wind project, it is imperative to assess the feasibility, operation cost, and energy potential before installing the wind energy system [
9]. Many researchers have used several statistical descriptors and probability distributions to test and analyze the characteristics of wind data. For wind data and its analysis, Weibull and Rayleigh distributions provide the best fit of the data collected [
10,
11,
12,
13]. Weibull has become a standard due to its simplicity and flexibility, and, according to the international standard of the International Electrotechnical Commission (IEC: 614-00-12), two-parameter Weibull is a proper approach to analyze wind characteristics. These parameters—Weibull shape and scale parameters—are easily computed by using several methods [
14,
15].
Many researchers have studied various methods to estimate and optimize the potential of wind energy at various locations around the world [
16,
17,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27]. Others have evaluated the wind energy potential of different areas. One study showed wind data of the Alacatı region in Izmir, which was investigated by using the Weibull statistical distribution [
28]. The measurements were collected on three various altitudes, respectively, 70, 50 and 30 m, in 10 min time intervals continuously for five years, and mean wind speed was found equal to 8.11 m/s for the complete data set. Another study analyzed four different locations in Iran where Weibull distribution was used to study the wind data collected in Mirjaveh, Zabol, Zahedan, and Zahak [
29]. Wind power and energy density were used to analyze the cost for the wind fields at these locations. Assessments for on- and off-grid sites show that Mirjaveh is suitable for off-grid generation and can be used for water pumping and battery charging.
Besides specific technical measurements and statistics of wind energy resources, economic analysis is also quite significant from an investment point of view to avoid investment failure risks. Similar studies have been conducted in Zahedan in Iran, Kutahya in Turkey, selected areas in Jordan, Johannesburg in South Africa, some onshore locations in China, the Kiribati Islands, and many others worldwide [
24,
30,
31,
32,
33,
34,
35,
36,
37,
38,
39,
40,
41].
Several studies have been carried out in the past in different areas of Pakistan where different researchers have analyzed various aspects of wind data collected from different periods; an example of such analysis was conducted in Hawksbay, Karachi [
42]. Another study in Karachi was conducted, and wind energy potential was investigated [
43]. Other investigations were conducted in Keti Bander [
44], Gharo and Jhimpir [
45], Jamshoro [
46], and Babur [
47]. Techno-economic evaluation was carried out for Hawksbay and Babur [
42,
47]. These studies show that, to estimate wind power potential precisely, it is vital to analyze the wind speed data in detail.
In this work, the potential of wind energy is evaluated for the city of Hyderabad in Sindh province located in the southern part of Pakistan. The wind speed data were measured at the height of 10 m in 10 min time intervals over two years (May 2015 to April 2017), and data were obtained from the combined project of the Government of Pakistan and the World Bank Group, funded by the Energy Sector Management Assistance Program. However, in this study, the authors used the Weibull and Rayleigh distribution methods to understand the wind characteristics. Root mean square error (RMSE), the determination of coefficient (R2), and the mean bias error (MBE) were computed to assess the performance of the distribution functions. In addition to the mean speed of the wind, the maximum energy carrying wind and the most probable wind speeds were calculated. Moreover, the wind energy and wind power density were predicted. Conclusively, economic analyses for standalone wind systems were carried out considering commercially available wind turbines in the selected areas so as to determine whether assessments are economically adaptable. It is worth saying here that, for the Hyderabad area, this kind of detailed study is rarely carried out. It is hoped that this study contributes to understanding the characteristics of wind in this area. The outcomes of the study can help government officials and potential investors to make efficient energy plans for the proposed area.
2. Site Description and Data Collection
Hyderabad is the second largest city of the southeastern province in Pakistan, located at longitude 68.3578° E and latitude 25.3960° N with an elevation of 29 m. The city is part of one of the major geographical regions in Pakistan and situated at the east bank of the Indus River plain. The provincial capital Karachi is almost 150 km away, which has the largest seaport of Pakistan. The Hyderabad is considered to be in the wind corridor of Pakistan and with a good average speed virtually throughout all the seasons. The location and elevation topography view of Hyderabad city is shown in
Figure 1a,b.
Figure 1b shows an elevation map of the Hyderabad area. Altitudes range from 16 to 35 m, and warmer colors indicate higher altitudes. In a recent study, a Geographic Information Systems (GIS)-based map showed that areas in the southern part of Sindh province are most suitable for installing wind turbines [
48].
The wind tower was installed on land with flat terrain and a wind site easily accessible by various kinds of vehicles. All instruments/sensors were tested by connecting them to the data logger before assembling to the site, and correct functioning of all the instruments during commissioning was also verified. A team of maintenance engineers and observers visited the site every month for data collection and to inspect the apparatus, and sensors were cleaned weekly. Monthly measured data were analyzed by the experts to ensure the quality of the collected wind data. The site terrain is flat with no major obstacles or roughness in the area surrounding the wind tower, except for some medium height trees and buildings with a maximum of approximately 3 m; these are about 200 m away from the tower. The coastal area and southern areas of Sindh province have land with flat terrain, with no major obstacles and roughness. Therefore, the surface roughness is between 0.03 to 0.04 m, and the roughness class for these areas is between 1.0 and 1.2 [
49,
50,
51]. The technical specification of meteorological sensors is given in
Table 1. Wind speed (m/s), wind direction (degrees), atmospheric temperature (centigrade), and pressure (hPa) were measured in 10 min time intervals at the site.
4. Results and Discussion
In this study, two years of wind data, from May 2015 to April 2017, collected at 10 m AGL at Hyderabad in Pakistan, were analyzed. Weibull and Rayleigh distribution functions were used to investigate the probability distribution of the wind speed data. Statistical descriptors such as the mean and standard deviations were used to compute the power and energy density for the output. Wind directions were analyzed to yield an optimum evaluation of the respective wind field. Finally, economic analysis was conducted to determine complete feasibility of the wind profile in the area. The following section is a discussion of the results obtained.
4.1. Wind Speed Analysis
Figure 3 shows the diurnal wind speeds for the two years, 2015 to 2017, for 24 h (hourly change), and both years and mean wind speed show a consistent pattern. It can be seen that higher wind speeds occur from 7 a.m. to 6 p.m. in the daytime. The wind speeds remain above 6 m/s for more than 16 h and above 5.5 m/s all day. A maximum wind speed of 7.3 m/s was found at 5 p.m. The wind speed is relatively lower during the night and reaches a minimum of 5.2 m/s at 6 a.m., although a constant wind speed from 10 a.m. to 2 p.m. can be seen, ranging from 6 m/s to 6.5 m/s, followed by an increase to a maximum speed of 7.3 m/s, which occurs at 5 p.m. It can be seen from the graph that the average wind speed throughout the day is very steady, within a range of 6 m/s and the wind speed in the year 2016–2017 is slightly higher compared to the year 2015–2016.
Figure 3 does not show the variations in the wind speed on a daily or monthly basis; this comparison is concluded in the following section.
In
Figure 4 and
Figure 5, the averages for daily maximum and minimum wind speeds are shown along with the mean for the years 2015–2016 and 2016–2017, respectively. It can be seen that the wind speed fluctuations are very high in the mid part of both years (the data were collected from May 2015 to April 2017; Day 0 in both figures represent 1 May of the respective year), where the maximum speed increases to 16 m/s, while the minimum speed is around 2 m/s. The wind speed is comparatively steady at the end and start of the year. The minimum and maximum wind speeds range between 2 and 13 m/s for the duration of both years. As a consequence, the mean wind speed ranged from 2.5 to 10 m/s for most of the time in both years. The wind speeds of 2016–2017 follow the same pattern, impartially decreasing the maximum occurrence of wind speed to a maximum value of 15.5 m/s.
As evident from the previous analysis, the first half of the two years has higher wind speeds, while the other half shows more steady but lower wind speeds. This trend is supported by
Figure 6, where monthly averaged wind speeds are shown for the two years along with their mean. The highest wind speeds were observed in August 2015 and June 2016 with values of 9.67 and 10.16 m/s, whereas the lowest was found in December for both years with values 3.46 and 3.63 m/s. Overall, the monthly average wind speed over the two years stayed between 3.5 and 10.2 m/s.
Figure 7 represents the seasonal average wind speeds for the two years and their mean of the site. The whole year is divided into four seasons: summer (from May to August), autumn (September to November), winter (December to February), and spring (March to May). It can be seen that summer is the most suitable, the highest wind speed and mean wind speed being 8.8 and 8.3 m/s for years 2015–2016 and 2016–2017, respectively, followed by spring and autumn. Winter has the lowest wind speed (approximately 4 m/s) in both years.
4.2. Wind Speed Frequency Distribution Analysis
The monthly and annual mean wind speed, standard deviation (St.dev.), turbulence intensity (Tr. I), Weibull parameters, i.e., k and c calculated using Equations (4) and (5), and specific wind characteristics (
Vmp &
VmaxE) of the site for both years (2015–2017) are summarized in
Table 2, and seasonal measurements are shown in
Table 3. The average value of the shape parameter for the two years is 2.097, and this parameter is almost the same for both years individually. Maximum shape parameter was in August 2015, where its value was 5.4, while the minimum was in February 2016 with a value of 2.031. The values of k for the whole data set were above 2, which indicates that the wind speed was moderate steady at a 10 m height at the candidate site. The scale parameter for the two years was 7.1 m/s, which is almost the same as that for the two years independently. The maximum scale parameter was found to be 11.115 m/s in June 2016 and 10.6 in July 2015, while the minimum was recorded to be 3.911 in December 2015 and followed by December 2016 is 4.1 m/s. The average most probable wind speed (V
mp) was found to be 5.17 m/s for the two years—6.94 m/s and 5.078 m/s for 2015–2016 and 2016–2017, respectively. The V
mp values ranged from 3.11 to 10.55 m/s. The average maximum carrying energy wind (V
maxE) of the two years was estimated at 9.703 m/s. Averages of 9.64 m/s for 2015–2016 and 9.76 m/s for 2016–2017 were found; V
maxE values ranged from 5.06 to 12.03 m/s.
The highest seasonal shape parameter k value of 3.336 was in summer 2015, and the lowest k value was in winter 2016. Seasonal scale parameter c estimates ranged from 4.318 to 8.985 m/s. The highest c value of 9.985 m/s occurred in 2016 summer and the lowest value of 4.318 m/s was in winter 2016. Seasonal Vmp values were between 3.303 and 8.985 m/s, and the highest and lowest values were 8.985 and 3.303 m/s in summer 2016 and winter 2016, respectively. The highest seasonal VmaxE value of 11.478 m/s was in summer 2016, and the lowest value was 5.760 m/s during the winter season of the year 2016.
A comparison of Weibull and Rayleigh probability density functions and cumulative density functions with real data histograms is presented in
Figure 8 and
Figure 9 for 2015–2016 and 2016–2017, respectively. For the Rayleigh distribution, the value of k is fixed, i.e., 2. The k calculated using Equation (4) for the Weibull distribution is not constant and was found greater than 2 for both years. This trend can be seen in the related figures.
Equations (20)–(22) were used to calculate the errors and verify the accuracy of the results computed from the Weibull distribution. The errors—RMSE, R
2, and MBE—are given in
Table 4. The values of all these errors are found in an acceptable range for the candidate site and verify the better fit of the Weibull distribution. Although the errors are small, the most probable wind speed was predicted to be 5 m/s by the Weibull distribution for both years, compared to around 6 m/s for the measured data.
4.3. Wind Power and Energy Density
It is significant to discuss the output generated by the wind in the form of power and energy density. The monthly values for air density (
) were calculated using Equation (16) and ranged from 1.131 to 1.205 kg/m
3. The average annual
of the proposed site was found to be 1.161 kg/m
3.
Table 5 illustrates the average monthly and yearly trends in power density and energy density calculated using the measured data and the Weibull and Rayleigh estimations. The Rayleigh function overestimates the power density for months with high wind speeds, and shows a relatively better fit for months with low wind speeds. The Weibull functions for months with low wind speeds show a slight underestimation of power density. In [
21], researchers found that the Weibull function can predict the wind speed data better in comparison with the Rayleigh function. Almost similar results were found in this study, and this trend can be seen in
Table 5.
For the years 2015–2016 and 2016–2017, the annual mean power density based on actual data was found to be 256.3 W/m
2 and 259.1 W/m
2, while the energy density was found to be 2245.88 kWh/m
2 and 2265.01 kWh/m
2, respectively. The monthly maximum average power density was observed to be 626.44 and 587.03 W/m
2 in July and August 2015 and for the year 2016–2017, and 695.42 W/m
2 and 539.11 W/m
2 in June and July 2016, respectively. The monthly minimum average power densities were 41.38 W/m
2 and 49.11 W/m
2, observed in the December of both years, whereas the highest and lowest values of energy density were found in the above-mentioned months for both years. In order to scale these outputs, the following wind classification at a 10 m height was used [
29].
Fair | |
Fairly good | |
Good | |
Very good | |
Furthermore, there is another wind power class (WPC) at 10 m height that can be used for the assessment of wind resources based on wind speed and power density [
26,
67]. According to this WPC, the regions classified as Class 3 or higher are considered suitable for wind power production. The calculated results of annual power density (>250 W/m
2), which are given in
Table 5, show that the Hyderabad area is in Class 5 and suitable for installation of wind turbines. It can be seen in
Figure 10 that the higher wind speeds in the summer season resulted in a higher wind power density for both years, calculated to be more than 500 W/m
2. According to the defined scale for the wind power class, the summer season is in Power Class 7. The spring season is in Class 5, as the calculated power density is around 260 W/m
2 for both years. Finally, based on the annual estimate for the power density, the power output is relatively good for both years.
4.4. Energy Production
Total energy production and capacity factor are fundamental aspects of a wind power project. To determine the optimum energy output, it is essential to select the right turbine for a location. The wind data used in this research were measured at the height of the 10 m AGL. The power law was used to compute the average wind speed for different hub heights given in Equation (1). The considered hub heights were between 40 and 95 m, whereas the average wind speeds were between 7.5 and 8.5 m/s. The average wind speed was around 6.2 m/s for the whole data set, and the average maximum energy carrying wind speed (VmaxE) was about 9.7 m/s by Weibull estimation at 10 m height. The average VmaxE ranging from 9.7 to 13.4 m/s was computed up to the height of 80 m. For estimation of VmaxE at higher altitudes, the values for k and c were calculated using Equations (8) and (9). For this purpose, 18 wind energy conversion systems of different manufacturing companies (such as GE, DeWind, Enercon, Nordex, Vestas, Suzlon, and EWT DW) were selected to find the wind power, the annual energy production, and the capacity factor of the investigated site. The selected WTs’ power rating ranged from 0.33 to 2.75 MW.
The power curves of all selected WTs (curves obtained and compared using Windographer software) are shown in
Figure 11 and other technical specifications are given in
Table 6. To operate with its optimal capacity, the wind turbine design parameters, for example, rated power, cut out and cut in velocity, rated velocity, and hub height, were selected according to the wind characteristics of a particular site. The operating range of wind turbines was typically between 2.5 and 14 m/s, which can affect the capacity factor and cost of energy, as well as changes depending on the size of the turbine; the average maximum energy carrying wind speed (
VmaxE) for this location was almost the same as the rated wind speed of the wind turbines. The power produced (P), the annual energy delivered (E), and the corresponding capacity factor using these WTs are summarized in
Table 7. The maximum energy generated (E) per year by the GE 2.75-103 wind turbine was 11227.16 MWh followed by Suzlon S95-2.1 (8948.70 MWh) and Vestas V90-2.0 (8580.12 MWh), while the minimum energy was produced by Enercon E-33/330 (1056.82 MWh). The maximum capacity factor (CF) was 0.56 for the GE 1.6-100 wind turbine, whereas the minimum value was 0.33 for Nordex N80/2500 turbine. Overall, the selected turbines showed a good amount of energy production. Some showed a low output but still had a good CF.
4.5. Windrose Diagrams
In order to determine the maximum outputs, an optimum configuration for the wind farm is required; wind direction plays an integral part in obtaining this optimization. Wind directions can be analyzed using rose diagrams, where 0° is north and a 15-degree arc divides the whole into 24 sectors.
Figure 12 and
Figure 13 show the average annual wind rose diagrams for the two years. Windographer was used to construct the wind rose diagram to find the overall wind direction frequency of the selected site. It can be seen from the rose graph that more than 45% of the wind is directed between 195° and 240° degrees clockwise. The critical fact is that, during the spring and summer seasons, the most productive seasons, the wind blows in the same direction.
4.6. Cost Prospect Analysis
An economic assessment of wind energy systems depends on many factors that vary in the regions. For wind energy projects, the fuel is free but requires a very high initial investment. The site is suitable for most wind turbine applications, from small standalone turbines to large wind power plants. The cost of substantial investments needed for the wind farm requires a detailed analysis of wind farm cost. Advantages and optimization of the wind farm layout are beyond the scope of this article, but an approximate economic analysis of standalone or large systems for powering the local community is described below. Wind turbines with different hub heights were selected. Hub height turbine analysis can be a benefit and provide an initial approach for the government and local investors to set up a wind farm.
The cost analysis of the selected wind turbines was calculated using Equation (27). The cost of wind turbines with power ratings above 200 kW ranged from 0.7 to 1.6 M
$/MW, and an average specific cost of 1.15 M
$/MW was considered [
41,
68]. The cost of a wind turbine in this study was estimated by assuming an initial price of 1.2 M
$/MW. The initial investment cost was considered 30% of the wind turbine cost and the real interest rate was considered 10% [
47]. The lifespan (s) of a wind turbine was supposed to be 20 years, whereas operation and maintenance cost were taken as 2% of the wind turbine cost [
29,
47]. By considering the above assumptions, the cost of energy (COE) per MW generated by the preferred wind turbines was computed, and the outcomes of these calculations are given in
Table 7. To simplify the calculations, other losses, such as grid and distribution losses, transformer losses, wake losses, and cables losses were neglected and are beyond the scope of this study. The cost of energy (COE) per unit ranged from 19.27 to 32.80
$/MWh. The highest cost was 32.80
$/MWh for the Nordex N80/2500 wind turbine, and the lowest was 19.27
$/MWh for the GE 1.6-100 wind turbine.