Pesisir Barat Lampung Offshore Wind Characteristic in Local Renewable Energy Development Studies

Renewable energy is one of the most researched topics. One renewable energy source is wind energy. As with some other energy sources, there are 2 locations for wind energy conversion, onshore and offshore. Using offshore wind energy requires more accurate wind data due to the higher risk. The characteristics of the wind at sea are one of these data. This study used data sets from ECMWF and BMKG from 2001 to 2020 (20 years). This research aims to examine the wind characteristics in the offshore of Pesisir Barat Lampung. The wind characteristics being examined are wind speed, wind direction, wind blowing duration, and wind speed distribution. The maximum hourly mean wind speed in these waters is 16.35 m/s. The dominant wind direction is blowing from the South East (SE). The maximum average wind-blowing duration to SE direction is 11 hours based on ECMWF and 14 hours based on BMKG. Based on the Chi-Square and Kolmogorov-Smirnov distribution fitting methods, the wind speed distribution in these waters is best fitted by the Weibull distribution for the ECMWF and the Lognormal distribution for BMKG. Using these two data sources, on average, ECMWF provides wind speeds 5% to 20% higher than BMKG. In analyzing wind potential, BMKG data will certainly provide more critical results than ECMWF data. In the analysis of offshore support structures, ECMWF data will certainly produce more critical structures. While the calculation of potential is better using BMKG data. Wind power potential in Pesisir Barat offshore is in the range of 44.6 to 82.9 W/m2.


Introduction
The Indonesian government is committed to reducing the use of fossil power plants to participate in reducing greenhouse gas emissions, which is targeted to reduce the greenhouse effect to 314 million tons by 2030 [1].Indonesia is currently starting to replace the use of fossil energy with renewable energy, the Indonesian government has released several guidelines for the development of eco-friendly renewable energy [2].With Indonesia's commitment and potential, it is expected to achieve 17% by 2025 [3] and 50% by 2050 [4] in replacing fossil fuels with renewable energy.
Wind, as one renewable energy source, is clean and always available in nature [5].Indonesia's estimated wind power potential is about 60.6 GW, with wind speeds ranging from 3-6 m/s and installed facility for only 0.15 GW in 2020 [6], [7].With other methodologies, the Energy Sector Management Assistance Program (ESMAP) released offshore wind technical potential in Indonesia is The offshore wind turbine is very popular in many countries and increasingly developed because the industry is very promising.For many years Northern Europe has been developing offshore wind turbines for various reasons.The greater wind speeds at sea and no land use are the top reasons [9].
In the case of Indonesia, offshore wind energy research has not made significant progress because the majority is still in the form of potential studies in several areas, but those will make good progress for the development [10]- [16].Although this industry is very promising, many challenges must be resolved such as the large costs incurred for its study, development, and construction [17].Primary data become a challenge in offshore wind energy development because of its high costs for maritime countries like Indonesia.But now, secondary wind data can be provided from satellites that conduct measurements of wind [18].Due to renewable energy's high risk [19], secondary data could only be used for initial research.Research has already done for Southeast Brazil, China, Korea, North Europe, West Mediterranean are using that kind of data [20]- [24].This implies that the secondary data could be valuable in evaluating the offshore wind energy potential from wind characteristics.
The western and southern waters of Lampung are two potential offshore wind energy areas in Indonesia with an average wind speed range of 5-7 m/s [8].Western Lampung has an average wind speed of about 6 m/s which was obtained from the MTSAT and Lognormal Distribution model [25].With the knowledge of the wind characteristics and potential, it can provide good information for the development of offshore wind energy.Each region has specific and different wind characteristics.Some wind characteristics are wind speed, wind direction, wind blowing duration, and wind speed distribution.

Statistical Distribution
Wind data statistic is being analyzed in this research.The primary statistical parameter being used in determining the potential of wind energy is the data distribution, in this case, wind distribution.Wind distribution is fitted by theoretical distribution.The fitting process uses using probability density function of 3 theoretical distributions, Normal, Lognormal, and Weibull.The best-fit distribution can be used for further analyses in the wind energy sector for Pesisir Barat offshore.Another result of that analysis is the extreme value of wind speed.That can be an input for wind turbine structural analyses.Maximum and average wind duration blowing from a direction is summarized in this research.That can be an input for wind energy analyses.

Probability Density Function
The probability density function is shown in equations (1) for Normal distribution, (2) for Lognormal distribution, and (3) for Weibull distribution.Each parameter for Lognormal and Weibull is described in equation ( 4) to (7).

Distribution Fitting
The distribution fitting methods used in this research are Kolmogorov-Smirnov and Chi-Square.The Kolmogorov-Smirnov method uses the cumulative density function.Whereas, the Chi-Square method uses the probability density function described above.Kolmogorov-Smirnov result value is calculated using equation (8).Chi-Square result value is calculated using equation (9).

Extreme Value
Extreme value is calculated depending on the best-fit distribution to the data.Equation (10) gives the general equation that determines the extreme value.

Pesisir Barat Offshore Wind Characteristic
The characteristics of wind in Pesisir Barat Lampung offshore are shown by the average wind speed map, windrose by the season, and occurrence table.This research analyses the wind data from 2 reanalysis satellite data, the European Centre for Medium-Range Weather Forecasts (ECMWF) and Badan Meteorologi, Klimatologi, dan Geofisika (BMKG).Much research has been done using one of those data sources for another specific area [24], [26], [27].The data points are shown in Table 1.Every point is being analyzed in this research for a duration of 20 years of hourly 10 m wind data (2001 to 2020).The average wind speed map for BMKG and ECMWF data is shown in Figure 1.BMKG average wind speed data has ranged from 3.5 to 5.5 m/s.BMKG's average wind speed is greater the further southeast.ECMWF average wind speed data has ranged from 3.5 to 6.0 m/s.This data has the same pattern as BMKG, but there is an anomaly in point 4 that it is the smallest among all points.Secondary reanalysis data does tend to have quite large values when compared to onshore field data [28].
Turbines have a minimum cut-in speed to operate.Referring to the example of the small-sized turbine with rotor diameter of 9.8 m, the cut-in speed is 2 m/s [29].The average wind speed as obtained by both BMKG and ECMWF in this site is sufficient to rotate the reference turbine.It shows that in ECMWF data, the wind blows dominantly from the SE direction with an average duration of 11 hours and a maximum duration of 279 hours.However, the BMKG data is more diffuse than the ECMWF data.According to BMKG, the maximum wind-blowing duration is 720 hours from the S direction.The average wind duration is 14 hours from the SE direction, while this direction also has a huge max duration of 687 hours of wind blowing.BMKG data also gives more direction with a huge wind-blowing duration (more than 500 hours), SSE, WNW, and WSW.We can't see that from ECMWF data that dominantly gives SE the center of interest.

Pesisir Barat Offshore Wind Distribution
Maximum wind speed in a year or annual maximum wind speed of all points of interest in this research is shown in Figure 6 for ECMWF data and Figure 7 for BMKG data, while the right side of the figure shows the Weibull distribution of annual maximum wind speed data of all points of interest.
From those figures, the annual maximum BMKG data looks more uniform for all points.That is because BMKG gives the data in units of knots.In ECMWF data, point 4 has the smallest annual maximum data among all points.That makes the distribution, in the right figure, of points 4 plotted very left among all points in ECMWF data.In the distribution of BMKG data, all points are in a similar position.The only similarity between BMKG and ECMWF annual maximum wind speed data is that they range from 9.0 to 16.0 m/s.Maximum wind speed for ECMWF data is 16.35 m/s in point 6 in 2018.The maximum wind speed for BMKG data is 15.43 m/s in point 1 in 2001.
For comparison, this offshore maximum wind speed data is far bigger than onshore data in South Lampung, Pontianak, and Banyuwangi, 2.5-3.5 m/s [30]- [32].Of course, offshore winds have more potential than onshore winds.For further study, data validation with filed data is a must.Data validation could be done by measuring the wind speed in the field at a certain period.As done by Novrita [27], the difference is that for offshore wind potential, of course, it needs to be done offshore.That data validation will compare and then correct the satellite wind speed data for all periods.Based on the things that had been learned, the 2 data sources provide different things when analyzed annually, but not when analyzed at the overall statistics.
To get the best distribution fit to the data distribution, goodness of fit analyses were conducted on both data.The result of the analyses is shown in Table 2 for Kolmogorov-Smirnov analyses and Table 3 for Chi-Square analyses.The result varies depending on the point being analyzed.Weibull distribution has the best fitting to the ECMWF data distribution with all points average value of 0.112 for Kolmogorov-Smirnov and 0.265 for Chi-Square.Whereas, the Lognormal distribution has the best fitting to the BMKG data distribution with all points average value of 0.109 for Chi-Square and 0.136 for Kolmogorov-Smirnov.
With the value limit of Kolmogorov-Smirnov value of 0.304 and Chi-Square value of 5.991, all the results, don't exceed the limit.The other research with various locations has the same result that wind data is best fit for Weibull distribution [10], [18], [34], [35].A study using BMKG data in South Lampung before has the same result, that the wind data distribution is best fit for Weibull [30].
Based on the theoretical distribution best fit to the data distribution above, the wind speed extreme value is calculated.The calculation for BMKG data uses Lognormal distribution.Whereas, ECMWF data uses Weibull distribution.Extreme value wind speed is calculated for 1-year and 100-year return periods for each data.The results are shown in Figure 8 for 1 year return period and Figure 9 for 100 years return period.The result showed no difference from the average wind speed data before.ECMWF result has a greater extreme wind speed value than BMKG result, but 1 year and 100 years return period.ECMWF results for 100 years return period have ranged from 14.5 to 16.1 m/s for all points.Whereas, BMKG results for 100 years return period only range from 14.3 to 15.6 m/s for all points.ECMWF result is 5.2% greater than BMKG result for 100 years return period.ECMWF results for 1 year return period have ranged from 9.5 to 10.7 m/s for all points.Whereas, BMKG results for 1 year return period only ranged from 7.9 to 8.9 m/s for all points.ECMWF result is 20.0%greater than BMKG result for 1 year return period.This extreme wind speed value may not be used in the calculation of wind energy potential.But it will be used in the design process of wind turbine support structures.

Wind Potential
The wind potential or energy is often measured and calculated by its power.Wind power per swept area can be calculated using the equation (11). denotes wind speed,  denotes air density with value of 1.23 kg/m 3 , and  denotes swept area.Based on that equation, when the wind blows faster, the greater the power generated quadratically.Furthermore, the swept area only has a linear effect on wind power.Therefore, the development of wind energy relies heavily on finding sites with a strong wind speed.
/ =     (11) It is crucial to use the more reliable wind speed for a more realistic potential estimation, as overestimation is not desirable in this potential calculation.Therefore, the average wind speed from BMKG is selected.BMKG data is a little bit lower than ECMWF for annual average data.The wind potential is shown in Figure 10.From the result, wind power potential in Pesisir Barat offshore is in the range of 44.6 to 82.9 W/m 2 .The potential of this site for small-medium scale utilization is promising.A similar result from a study in Sidoarjo with wind speed more than 5 m/s that produced 89 W/m 2 [33].
Note that the calculated potential concerns only the wind potential at a height reference of 10 m above ground.The wind speed increases with a higher elevation of wind measurement.With a higher wind turbine altitude, the potential or energy produced would exceed that produced in this study.For instance, a study carried out in Jember, which is situated at an elevation of 43.2 m, yielded a power of 277.03 W/m 2 at wind speed of 6.5 -7.5 m/s [10].In the study conducted in Banyuwangi, the same methodology yielded an average of 1.34 W/m2, based on wind speed with average of 2 m/s over 10 days [36].In onshore measurements, Abdullah [26] obtained a potential value of 0.23 -0.80 W/m 2 for the coastal area of North Sumatra with an average wind speed of 1.4 m/s sourced from BMKG.
In line with Burke et al [28], Indonesia has wind potential, especially in Pesisir Barat offshore.This location is just an example of wind potential analysis in the rural area.With many small and outer islands, Indonesia has the potential to develop it even more with an off-grid system.This will help people living on small and outer islands could get green energy not far from where they live.

Conclusion
From all the calculations and analyses done in this research, some conclusions have been obtained are: 1. Wind blows from SE in Transition 1, Southeast Monsoon, and Transition 2 seasons while the wind blows from NW in Northwest Monsoon season for both of the data sources.2. Average wind speed has ranged from 3.5 to 5.5 m/s for BMKG data.Whereas, average wind speed has ranged from 3.5 to 6.0 m/s for ECMWF data.3. From all the data source analyses, the wind blows dominantly from the SE direction with an average wind duration of 11 hours and a maximum duration of 279 hours for ECMWF data and the average wind duration is 14 hours from the SE direction for BMKG data, while this direction also has a huge max duration of 687 hours of wind blowing from.4. The Weibull distribution has the best fitting to the ECMWF data distribution while the Lognormal distribution has the best fitting to the BMKG data. 5. ECMWF results for 100 years return period have ranged from 14.5 to 16.1 m/s, 5.2% greater than the BMKG result.Besides that, ECMWF results for 1 year return period have ranged from 9.5 to 10.7 m/s, 20.0% greater than the BMKG result.6.Using BMKG average wind speed data, wind power potential in Pesisir Barat offshore is in the range of 44.6 to 82.9 W/m 2 .

Figure 1 .
Figure 1.Average wind speed data BMKG (left) and ECMWF (right).The windrose for every season in Indonesia is shown in Figure 2 and Figure 3.The wind blows from SE in Transition 1, Southeast Monsoon, and Transition 2 seasons while the wind blows from NW in Northwest Monsoon season, both BMKG and ECMWF data.

Figure 2 .
Figure 2. Windrose for Northwest Monsoon and Transition 1 season.The incidence percentage of every wind speed class is quite similar for both BMKG and ECMWF data.This implies that both data source has a fairly high similarity in hourly wind speed and direction.The northwest monsoon season covers the months of December to March.The transition 1 season

Figure 4 and
Figure4and Figure5show the average and maximum duration of wind blowing in the same direction for ECMWF and BMKG data for representing Point 6.All the direction data shows the direction of wind blowing from, in 16 directions.

Figure 4 . 6 Figure 5 .
Figure 4.The average and maximum duration of wind-blowing ECMWF data.

Figure 7 .
Figure 7. Annual-maximum wind speed data (left) and Weibull distribution (right) BMKG data.Nonetheless, the trend of those data is different.Annual maximum wind speed ECMWF data was decreased from 2003 to 2006, 2015, and from 2019 to 2020.While BMKG data decreased from 2003 to 2006, and from 2014 to 2017.The annual maximum wind speed increased in 2001, 2007 to 2008, and 2016 to 2018 for some points for ECMWF data.While BMKG data increased in 2001, and from 2019 to 2020.Based on the things that had been learned, the 2 data sources provide different things when analyzed annually, but not when analyzed at the overall statistics.

Figure 10 .
Figure 10.Wind Power Potential per Swept Area in Pesisir Barat Lampung Offshore.

Table 1 .
Coordinate points of interest.