Abstract | Od svih vrsta prenapona, za distribucijska električna postrojenja su neusporedivo najopasniji prenaponi atmosferskog porijekla. Oni pripadaju kategoriji vanjskih prenapona i njihova je vrijednost potpuno neovisna o pogonskom naponu mreže. Atmosferske utjecaje na distribucijsku mrežu je teško izbjeći, što ima za posljedicu da je daleko najveći broj kvarova i smetnji u distribucijskim postrojenjima uzrokovan upravo atmosferskim prenaponima. U doktorskom radu analizirane su mogućnosti primjene podataka o atmosferskim pražnjenjima - udarima munja, dobivenim iz sustava za lociranje atmosferskih pražnjenja za unaprjeđenje planiranja i pogona srednjenaponskog elektroenergetskog sustava. Izvršena je korelacija između kvarova i zastoja u elektroenergetskoj mreži srednjeg napona i udara munja. Analiza je provedena na tipičnim nadzemnim srednjenaponskim dalekovodima sa stvarnim podacima o udarima munja i kvarovima u mreži, prikupljenim u razdoblju od 2009. - 2015. godine, pri čemu je promatrano vremensko i prostorno poklapanje podataka. Na osnovi provedenih istraživanja razvijen je postupak za statističku analizu i predviđanje ispada na temelju povijesnih i trenutnih podataka o udarima munja i pogonskih događaja te odgovarajuća metoda za određivanje vjerojatnosti utjecaja munje na pogon (ispad) specifičnog nadzemnog dalekovoda u srednjenaponskom elektroenergetskom sustavu. Razvijen je postupak koji, primjenom matematičke metode najmanjih kvadrata, te korištenjem programskog sustava Matlab, daje aproksimiranu funkciju uvjetne vjerojatnosti ispada tipičnog nadzemnog srednjenaponskog dalekovoda zbog djelovanja munje. Opisane su bitne komponente sustava kojim se razvila procedura za procjenu vjerojatnosti utjecaja munje na pogon, odnosno, ispad specifičnog nadzemnog dalekovoda u srednjenaponskom elektroenergetskom sustavu: • sustav za lociranje atmosferskih pražnjenja, • sustavi za prikupljanje podataka o događajima u elektroenergetskom SN sustavu i • sustav geoprostornih podataka o elektroenergetskim SN objektima. |
Abstract (english) | By far the most dangerous overvoltages for power distribution systems are those of atmospheric origin. They belong to the category of external overvoltages and their values are completely independent on the power network operating voltage. Atmospheric overvoltages in the distribution network are difficult to avoid. As a consequence, by far the largest number of failures and disturbances in the distribution systems are caused by lightning overvoltages. The lightning is the most common cause of failures on overhead lines and as such it has a significant impact on reliability and quality of electricity supply. Outages of overhead power lines due to lightning strokes are one of the main causes of shortages of electric supplies and economic losses of power utilities. The doctoral thesis describes the application of the lightning strokes data from the Lightning Location System (LLS) to improve the planning and operation of the medium voltage (MV) power system. It gives a brief overview of the issue of the impact of lightning strokes on MV power network. Furthermore it also describes the essential components of the system, developed in this thesis, for estimating the probability of the lightning impact on operation or outage of a specific MV overhead line (OHL): - lightning location system - system for data acquisition in MV power system and - system of geospatial data on MV power facilities. Nowdays lighting location systems are typically used for lightning activity tracking. Such systems have been developed and improved for more than twenty years. In developed countries worldwide, lightning location systems are used in networks and systems deployed in large areas such as power systems, TC networks, broadcasting transmitter networks, networks of oil and gas pipelines, security systems, military installations, meteorological services, transport infrastructure, systems of protection against forest fires and more. For example, in the United States the National Lightning Detection Network (LDN) covers almost the entire geographical area of the United States, while in Europe the EUCLID and LINET systems are developing. Lightning strokes may seriously affect the reliability of transmission and distribution lines. Due to lightning overvoltages, the most frequent is the flashover of insulation that cannot withstand the strong electric field as a result of these overvoltages. The flashover of insulation can result in a permanent or temporary outage and failure. Such events in the power network are recorded and collected by the SCADA or similar systems for supervisory, control and data acquisition from the medium voltage power network. In this thesis the emphasis is given to the application of a correlation between failures and outages in the medium voltage distribution network and lightning. The analysis is carried out on typical overhead medium voltage power lines with real data on lighting strokes and failures in the power network collected in the period from 2009 to 2015. Spatial event correlation is performed with real lightning data and failures in the power network collected in the period January 1, 2009 (00.00h UTC - Universal Time Coordinated) - December 31, 2012 (24.00h UTC). Before performing the correlation analysis, it is necessary to make the appropriate preparation of the collected data on outages of MV overhead distribution lines (due to the effects caused by direct or indirect lightning strokes in the vicinity of the line) as the input data set for correlation with relevant data from the appropriate lightning location system. The results of the correlation analysis may be used by distribution system operators as a basis for decision making on possible investments in an additional overvoltage protection. These statistical analysis can help the distribution system operators to prioritize and justify the investment (additionally overvoltage protection) based on the risk of exposure of each MV distribution line. The overvoltage protection has significant importance not only to prevent the damage of expensive components and equipment within the power distribution network but also to ensure the power quality. The investment in high quality equipment that can withstand significantly higher voltages than standard equipment is not acceptable due to high costs. It is therefore necessary to select a proper overvoltage protection. The properly implemented overvoltage protection and insulation coordination are the basic preconditions for high quality of the delivered electricity in all aspects of quality and reliability of supply through the overhead power network. To avoid typical overvoltages that can occur on MV distribution lines that are protected only by conventional surge arresters, it is recommended to install long flashover surge arresters which may mitigate some adverse parameters of overvoltage on the line. Long Flashover Arresters (LFAs) were shown in many papers and practice to be effective for lightning overvoltage protection of medium voltage overhead lines, such as those rated to 10 kV. Pole-top metal oxide arresters can protect distribution lines against induced overvoltages, but they can be destroyed in case of direct lightning stroke. LFAs are developed and used successfully for this purpose as they are resistant to direct lightning strokes by conducting the discharging current externally along its surface. However initial installation costs for long flashover surge arresters are significantly higher comparing to the conventional solutions of overvoltage protection. It is therefore necessary to conduct a statistical analysis that will help the distribution system operators to prioritize and justify the investment based on the risk of exposure of each MV distribution line in order to ensure more reliable supply of electrical energy. Temporal and spatial event correlation is analyzed with real lightning data and failures in the power network collected in the period April 12, 2012 (00.00h UTC) - February 2, 2014 (24.00h UTC), for 14 MV OHLs passing through the typical terrain configuration. Based on the conducted research it is developed a new method for calculating the probability of the impact of lightning on the operation/outage of the specific overhead line in medium voltage power system. The developed procedure provides a cumulative distribution function (CDF) in the form of 3D curve as a function of the distance of lightning from the line (m) and lightning current (kA). Such the function can be used for the prediction of outages due to lightning as a function of the distance of lightning from the overhead line (m) and lightning current (kA). The resulting 3D curve is calculated using least square method within Matlab software environment. By correlating the data of lightning discharges and the information on events in the power system one can obtain a valuable information necessary to decide about the measures to improve the reliability of electricity supply. The main objective of this thesis is to build a mathematical model/function for the prediction of the outages caused by the lightning strokes as a function of the distance of the lightning from the overhead line and the lightning current. Such the model is intended to be used dominantly in the distribution line route planning phase to obtain the optimal line route with the respect to the predefined economic based cost. In that regard in this thesis it is developed the procedure for estimating the probability of outage of a typical 10(20) kV overhead line due to lightning. Using the theory of probability and statistics, and on the basis of current and historical data from the LLS (Lightning Location System) and SCADA (Supervisory Control and Data Acquisition) system, a 3D surface as a joint probability function was obtained depending on the distance of lightning from the line and lightning current. There will be used the method of data correlation on lightning strokes and events in MV network with the actual outages data of MV distribution OHLs obtained from the SCADA system, correlated with the data from the LLS, by spatial-time correlation for the period April 12, 2012 - February 2, 2014. In order to asses the quality of the developed model for predicting line outages due to lightning the model is validated on a separate dataset consisting of the actual data obtained from the SCADA system and the LLS for the period February 3, 2014 - February 3, 2015. As a first step in the correlation procedure the acquired data need to be properly preprocessed by linking the data from the lightning location system to those on the events in the power network (SCADA system) and geospatial data on power facilities. The data on the events in the power system collected by the SCADA system contain, among others, the information on the exact time of the event and the information on the facility (line or system) where the incident occurred, and possibly other necessary data for correlation with the data from the LLS. It is necessary that both systems, SCADA and LLS, are time-synchronized by the GPS, which in this case is realized by using the data at UTC. The LLS data were linked with the data on the events in the power network (SCADA system) and geospatial data on the 14 MV 10(20) kV OHLs. The relevant data were written in the form of the Gauss-Krueger (GK) coordinate system and it was necessary to switch all relevant route log points to a reference coordinate system WGS84 supported by the predicted LINET LLS. The starting criterion for temporal correlation of the events of activation of the relay protection devices on the observed distribution lines and lightning is the exact time of the failure. For precision, for the correlation of activation of relay protection devices on distribution lines there were used the exact times of failures by excitation signals from the relay device as recorded in the lists of events in the archives of the SCADA system. To make the temporal correlation possible, it is necessary to have synchronized measurements in the LLS and SCADA system. In the correlation analysis only the events from SCADA system that happened up to 1000 ms after lightning stroke are taken into account. The objective is to develop the procedure for estimating the probability of outage of a typical 10(20) kV overhead line due to lightning. Based on current and historical data from the LLS and SCADA system, using the theory of probability and statistics a 3D surface representing a joint probability function of the distance of lightning from the line (m) and lightning current (kA) is obtained. The data on the lightning events from the LLS are temporally and spatially correlated with the data from the SCADA system on the observed outages and the overall probability function is obtained as a function of the distance of lightning stroke from the line for all observed MV overhead lines. The procedure is implemented in Matlab/Simulink environment. The procedure is further extended to obtain a joint probability function of the line outages as a function of the lightning location distance r and the lightning current i. Applying this procedure to calculate the joint probability function leads to the results that describe probability of outage of MV OHL as a function of the distance of lightning from the axis of line (m) and the lightning current (kA). The distance parameter values are discretized at every 100 m in the range from 0 to 2000 m and lightning current values are discretized at every 10 kA in the range from 0 to 100 kA. The objective in this thesis is to determine an approximative probability function of the line outages. This function relates the outage probability (%), distance r (m) and current i (kA). The form of the approximation function is selected by taking into account theoretical and empirical findings on the lightning effects, i.e. the increase of the lightning current increases the probability of the overhead line outage and with the increase of the distance of lightning stroke from the line decreases the probability of the overhead line outage. To determine the coefficients of the approximation function, it is necessary to find their values at which the minimum approximation error is attained. By the virtue of developed approximation function it is possible to determine an approximation of the probability of lightning in a range of the distances and in a range of the currents value. To achieve a more accurate approximation, it introduces more degrees of freedom or higher degrees of polynomials since a higher degree polynomial contains all members of smaller potencies. Testing of the approximation function of probability of medium voltage distribution overhead line outages is performed with real data from the LLS and SCADA systems collected in the period since February 3, 2014 till February 3, 2015. In order to check the quality of the model, it is necessary to check the model on the data that are not used during the learning process. A comparison between the empirically calculated probability outage function and approximate one of the fifth order is conducted. The test results show that the approximation function describes well the influence of the lightning strokes to the analyzed overhead lines. Root Mean Squared Error (RMSE) factor confirms that the results obtained by using the approximation function are in a good agreement with the measurement data for the analyzed medium voltage distribution overhead lines. The main benefit of the proposed method of calculating the outage probability lies in the ability to incorporate the probability function in the line route planning process. Once calculated this outage probability function allows for determining the optimal line route with respect to the overall long term cost that encompasses the technical and economic aspects through the possibility to assess the long-term maintenance and outage related costs for the analyzed routes. |