Short-term Offshore Wind Speed Forecasting using Buoy Observations and Regional scale Wind Resource Assessment based on Scatterometer Data

Abstract

Offshore winds are valuable source of renewable energy. To recognize the potential of area it is essential to assess the available resource and understand the sporadic nature of winds. Wind Resource Assessment (WRA) coupled with short-term forecast of winds will aid in establishing the confidence for undertaking offshore wind farm development. Wind speed forecasting is important for estimating power generation capacity of turbines. The knowledge of availability of the winds in future time steps will be pivotal in planning and improving the efficiency of energy production. Buoys are the fundamental source of in situ atmospheric parameter observations. One of the primary objectives of the present research is to determine suitable technique for short-term forecasting of offshore winds. So, the present study focuses on assessing accuracy of the ANFIS hybrid model for short-term wind speed forecasting. In addition, the Arabian Sea belongs to tropical humid climate zone and therefore the influence of Relative Humidity (RH) on the ANFIS model to estimate offshore wind speed was investigated. In the study, two buoys with id– AD07 and CB02 apart approximately by 500 km were selected. Two models (model 1: 5 inputs, 1 output and model 2: 4 inputs, 1 output) and two scenarios (scenario 1: estimate wind speeds and scenario 2: forecasting wind speeds) were developed for the study. From scenario 1, it was found that at both the buoy locations the model 1 outperformed model 2 in estimating observed wind speeds and RH had noticeable influence on the model performance. Persistence Method (PM) was chosen as base method for comparing the wind speed forecasts. From scenario 2, at AD07, model 1 forecasts were accurate than other two models and at CB02, the PM forecasts were most accurate. However, it was found that the model 1 forecasts at CB02 were closer to PM. Altogether, the model 1 performance was higher than model 2 indicating the error in forecasts due to absence of RH observations. The study concludes that the model performance was enhanced by incorporating RH observations as an input to the ANFIS model. The RMSE of forecasted wind speeds up to three time steps, at AD07 and CB02 would be approximately lower by 37% and 14% respectively.ii Further, the study examines the performance of ANFIS and Wavelet-ANFIS (WT+ANFIS) hybrid techniques to forecast wind speeds for multiple time steps at the same buoy locations (AD07 and CB02) in the Arabian Sea. The forecast accuracy of ANFIS and WT+ANFIS were compared with PM. The RMSE for the testing dataset at AD07 and CB02 using ANFIS model was found to be 1.3 m s-1 and 1.26 m s-1 for 1st (t+1) time step respectively. The RMSE for WT+ANFIS model at AD07 and CB02 was obtained as 1.5 m s-1 and 1.20 m s-1 for 1st (t+1) time step respectively. It was observed at CB02, the WT+ANFIS model forecast was closest to PM. At AD07, an ANFIS and WT+ANFIS model performance was almost similar and found to be better than PM. In general, the WT+ANFIS model outperformed ANFIS and PM for multiple time steps. Thus, the analysis establishes that WT+ANFIS hybrid method has the potential to be a complementary tool in obtaining short-term offshore wind speed forecasts. In the offshore region the scarcity of in situ wind data in space proves to be a major setback for wind power potential assessments. Satellite data effectively overcomes this setback by providing continuous and total spatial coverage. The satellite data needs to be validated at the study area before conducting WRA study. Hence the work centers on estimating the performance of Oceansat–2 scatterometer (OSCAT)– derived wind vector using in situ data from buoys (id– AD02 and CB02) at different locations in the Arabian Sea. For the validation of OSCAT winds, the buoy winds are required to be extrapolated to height of 10 m and are known as Equivalent Neutral Winds (ENW). A comparative study among three methods- power law, logarithmic and Liu– Katsaros–Businger (LKB) method for estimating the ENW for buoys is carried out. OSCAT winds were closest to ENW estimated by the Liu–Katsaros–Businger (LKB) method. The spatial and temporal windows for comparison were 0.5° and ±60 minutes, respectively. The monsoon months (June–September) of 2011 were selected for the study. The root mean square deviation for wind speed is less than 2.5 m s−1 and wind direction is less than 20°, and a small positive bias is observed in the OSCAT wind values. From the analysis, the OSCAT wind values were found to be consistent with in situ-observed values. Furthermore, wind atlas maps were developediii with OSCAT winds, representing the spatial distribution of winds at a height of 10 m over the Arabian Sea. Satellite-based regional scale offshore wind power resource assessment was carried out for the Karnataka state, which is located on the west coast of India. OSCAT wind data and GIS based methodology were adopted in the study. The real time ship based observations is considered in the present work, to assess the accuracy of OSCAT wind data. The INCOIS Realtime All Weather Station (IRAWS) data provides greater spatial coverage than conventional buoy setup. Probably, this is the first attempt to validate OSCAT data using IRAWS dataset, which offered greater number of collocated observation points and hence provided better assessment. Wind speed maps at 10 m, 90 m and wind power density maps using OSCAT data were developed to understand the spatial distribution of winds over the study area. Bathymetric map was developed based on the available foundation types and demarking various exclusion zones to help in minimizing conflicts. The wind power generation capacity estimation performed using REpower 5 MW turbine, based on the water depth classes was found to be 9,091 MW in Monopile (0-35 m), 11,709 MW in Jacket (35-50 m), 23,689 MW in Advanced Jacket (50-100 m) and 117,681 MW in Floating (100-1000 m) foundation technology. In Indian scenario, major thrust may be given for wind farm development in Monopile region. Therefore, as first phase of development for 10% of the estimated potential in this region, 116% of energy deficit for FY 2011-12 could be met. Also, up to 79% of the anticipated energy deficit for the FY 2014-15 of the Karnataka state of India could be achieved.