Statistik över svenska blixtströmmar : Analys av data från SMHI:s blixtlokaliseringssystem

Sammanfattning: Lightning constitutes a serious threat to the electrical grid. If not protected against, lightning can cause comprehensive damage, harm to humans and lead to disruption in the electricity supply. Modeling of lightning protection is hence an important part of operating the grid. The amplitude of the lightning current is a decisive factor in the induced overvoltage in electrical systems. Knowledge about which peak current amplitudes one can expect in a lightning event, and how often such events occur, form the basis in modeling and evaluating the lightning protection system. Today, the Swedish authority Svenska kraftnät models the lightning protection system based on current distributions recommended by CIGRE. The aim of this thesis is to examine lightning data gathered by the Swedish lightning location system (LLS), owned, and operated by the Swedish Meteorological and Hydrological Institute (SMHI), to derive statistical distributions for the lightning peak current. This is done at a national and a regional level for the years 2006-2020, the focus being on the period 2015-2020. The results are then compared with the distributions recommended by CIGRE, to evaluate whether the protection levels used today are justified. Lightning is a complex phenomenon whose properties need to be understood to measure its features and interpret the measurement results. Similarly, the quality and the shortcomings of the LLS must be addressed and considered when analyzing the results. The peak current estimation done by the LLS tends to be an underestimate, as the system is only validated for the subsequent strokes of negative lightning. Nevertheless, the LLS offers a large data set which, while individual measurements are not fully accurate, provides a representative statistical distribution exhibiting the tendencies of lightning peak current amplitude. The results show that the current distribution fits a lognormal and generalized extreme value distribution, the latter yielding a slightly better representation. A small regional variation in median peak current is seen, where somewhat higher amplitudes are found in Northern rather than Southern Sweden. The national analysis shows a significantly lower median peak current in Sweden compared to the reference median peak current recommended by CIGRE. Calibration of the LLS can be assigned a portion of this difference, the remaining is thought to represent a difference in lightning peak current. Unfortunately, there is no reliable way of knowing how large the calibration error is without further studies comparing data from the LLS with direct measurements. Consequently, deviating Swedish lightning conditions cannot be stated, although tendencies towards lower peak current amplitudes are seen.