Spatial forecast of production and operation parameters from wind power plants using multi-semivariogram method

The meaning of word of rüzgar (wind) is cycle, turmoil and world. Its meaning in science of meteorology is simply moving air. From ancient times, people used wind as a source of energy although not knowing the source of the wind. In this respect, first practices of using wind was grain grinding and sailing boats. Although the ancient Greeks and Romans used to float the sailing boats, it was not done by them to take advantage of the wind power in other activities of the community. Using wind energy for ordinary people has started in Central and East Asia. Windmills first seen in Iran, Afghanistan, Pakistan, Tibet and East Asia and China. Windmills, especially in Iran and Syria, were taken to the West civilizations along with the Crusades. The use of wind energy by the society, contrary to what is thought, is not in Western Civilizations but Eastern Civilizations. It is known that Asian Civilizations are used in China, Tibet, India, Afghanistan and Iran. The first written information about wind turbines (WT) It is written about horizontal axis windmills from 200 to 300 years (BC). It is known from concrete evidence that the Iranians used vertical axis windmills in the year 700 (BC). Wind power has passed from Asia to Europe around the 10th century. The first indication of this transition is the use of windmills in England during the 11th and 12th centuries. In the year 1190, the German Crusaders took windmills from Syria to their countries. The windmills used since the Middle Ages have been used to draw water from the wells and grind the grain. Many examples in the Netherlands and the Mediterranean can be noted. Historians say that in the 1700s BC the Babylonians used windmills for irrigation around Mesopotamia. With the industrial revolution, the emerging world of steam machines in the 18th century began to take advantage of machines based on thermodynamic processes to energize the world. Especially, the use of fossil fuels such as coal, oil and gas has become more advantageous because they provide energy source when requested. Today, wind energy is increasing day by day, even though it cannot be produced in quantities that can cover energy deficit. The main reasons for this are the low operating costs and no need for raw materials. As of the end of 2017, approximately 500,000 MW of wind power capacity connected to grid in the world. By June 2018, when the dissertation was written, it had 550,000 MW installed power. Planning such a large installed production and transmission system planning is crucial. Especially the European countries have started to work 1990s in the subject. Short-term wind energy forecasting from wind power plants (WPP) and energy production planning are becoming very important. One of the most important features of liberalized energy markets lies in the planning of distribution of electricity generation and electricity. Companies that supply electricity to the majority of EU countries have to meet a certain amount of energy required by the legislation in their country from renewable energy sources. Today, the use of renewables is mostly through wind energy. When planning for this purpose, companies want to be able to make energy plans for at least 0-48 hours in order to be able to fill the portfolio by focusing on wind power plants. While there is no such problem in power plants operating with traditional energy sources, this situation poses a serious problem in renewables That's the reason most companies in Europe work with wind energy forecasting models. Regarding the subject, the Wind Power Monitoring and Forecasting Center (RITM) was established in 2010 within the scope of YEGM in our country. The theme of this dissertation is based on the differences between the points studied from the wind power plants (WPP) using the semi-variogram technique and the other points. To investigate the effects of other points; production, temperature, gnerator temperature, vibration, reactive power, height, on a discussed parameter. The results of the missing points between the resulting curves and points can also be calculated. The semi-variogram technique is an important measure of geo-statistics and represents the proportion of area change along a location. Field modeling of wind and variables is a new subject. In general, meteorological studies concentrated grid and mapping methods are mostly used in field studies. The distance between the points and the spatial relationship are not taken into consideration. An important innovation brought about by the thesis is the use of Multi-variogram technique by improving semi-variogram technique. All measured variables are accounted for the calculations. As the data used, it is a significant advantage in terms of the power plant operators if it does not require an external data source other than the plant. The main reasons for such a thesis work can be listed as follows; The fact that our country is going to liberalize the energy market and the forecasting of production from WPPs becomes important, Completion of incomplete data of parameters analyzed from WPP, Production planning of WPP, The application of multi-variogram technique by developing semi-variogram concept. Another important aspect of the prediction of production from WPPs is environmental awareness. Electricity distribution companies using renewable resources as much as possible while giving priority to WPPsin planning. One of the most important aspects of using short-term forecasting is the spot market concept that emerges with liberalized energy market. Making the energy planning in the spot market and the market structure emerging through the bilateral agreements made for instant or short time increases the importance of forecasting. As can be seen in dissertation, these models have been developed by universities and institutes. The forecast of wind energy production is of great importance for electric power transmission and distribution companies. In addition, for the WPP operator, the production forecast for the WPP's maintenance planning is very important. It is desirable to perform maintenance in the period when production is minimum. For this purpose, 4 WPPs operational data were used belonging in various regions of Turkey. Generally speaking, in meteorological literature, grid and mapping methods are mostly used in field studies. The distance between the points and the spatial relationship are generally not considered. Variograms are the basic approaches to geo-statistics. It examines the distance and direction change ratio of the parameter or variable examined. Two variables close to each other are expected to show more similar properties and values. That is, the correlation between nearby variables is greater than the correlation between two distant variables. As the distance increases, the correlation between the two variables diminishes and there is such a point that this correlation becomes zero. Thus, the variogram graph also shows the distance at which the correlation between the variables examined will be zero. One of the most important assumptions of the semi-variogram proposed by Matheron (1963) is the field relation, which depends on stationarity and equal distances. Here, a pointwise relation is given to a more uniform distribution of spatial relations. In the absence of stability, the relationship between randomly distributed points does not give accurate results. In order to overcome this problem, the point total semi-variogram (PTSV) method in the study by Şen (1989), the lack of stability of the examined variable and the random distribution of the examined points, PTSV, which is based on the variogram of the calculations and acceptance used, was mostly used in earth science studies. Xr is the reference point and h1 ... hn is the expected value of the change point, which is the distance between this reference point and the other points E[(X_r-X_hi )^2 ]≥0 i=1,………n (1) and thus semi-variogram; Y(h_i )=1/2(n-1) ∑_(i=1)^n▒(X_r-X_hi )^2 (2) In the field estimation calculations, the weighted averages of the influence coefficients are used together with the data depending on the distance. Therefore, in order to make estimations; A=w(h_(st,1) )+w(h_(st,2) )+ ………+w(h_(st,n) ) (3) 〖 X〗_T=1/A ∑_(i=1)^n▒〖X_i w(h_(st,i) ) 〗 (4) can be formulated. A represents the weighted average effect of the influence coefficients, while XT yields the area-based weighted average predicted value. The following steps must be followed in order to calculate NTY; 1. A reference point should be selected and distances between this selected point and other points should be calculated. If there are n points, the number of distances will be n-1. These points must be aligned from small to large, 2. Each value must be divided into halves so that consecutive sums of the squares of differences of the values between the selected reference point and other points can be taken and the semigram can be obtained, 3. The NTY values corresponding to the respective distance values (on the X axis) are marked on the Y axis, 4. Obtain the distances on the X axis of the obtained function and the NT values on the Y axis. To use these in point-domain association calculations, the values at each point are divided by their largest values into standard and unit-free. 5. According to the objective analysis method, the standardized values are subtracted by 1 and the greatest value is divided and the SAB function is found. 6. For the SAB function found for each point examined, weighted averages are taken according to the distances between the points and spatial estimates are made. 7. Semi-variogram technique was developed and estimates were made on multi-semivariogram calculations. In the dissertation, all semi-variograms were related to the variables analyzed by drawing calculations and graphs. Here is the development of a spatial formulation using all the changes studied. The field dependency coefficient for the examined variable and the standardized value of the examined variable are multiplied and divided by the field dependency sum. For the examined wind turbine, the effect of the variables in the total area was calculated and the effect coefficient was found. In the thesis, the relations between the variables analyzed and the variables investigated by the semi-variogram method were found. A spatial formulation has been developed using all the changes studied. The field dependency coefficient for the examined variable and the standardized value of the examined variable are multiplied and divided by the field dependency sum. For the examined wind turbine, the effect of the variables in the total area was calculated and the effect coefficient was found. This calculation is done with a multi-semivariogram technique. That is, a coefficient has been developed for each turbine and for each variable examined. This thesis used 4 wind power plants commissioned İstanbul, Aydın, Balıkesir and Amasya provinces. A total of 31 wind turbine data were analyzed. 15 units of 2 MW; and 16 units of 2.5 MW wind turbine data examined. The turbine distance, wind speed, production, temperature, generator temperature, vibration and reactive power data are worked. Each of the wind turbine data examined was divided into 70% educational and 30% test data and the multi-variogram was processed with 30% of the field coefficients from 70% area and the results obtained were compared. The results show that the prediction accuracy approximates the measured value between 85% and 95%. The forecast range given for the WPP projects is a good approach. The developed multi-semivariogram method hs been tested with multiple regression approach. In addition, the same methods were applied for 1 month data. 1 month data study has shown that proposed multi-semivariogram method had better forecats results than multiple regression method.