Analysis of performance data of wind turbines for polarregions

The Radio Neutrino Observatory in Greenland is situated at Summit Station. Due to its isolatedlocation the main power to support the autonomous radio stations is solar energy. However,during the winter months when there is no sun another strategy needs to be found. This projecthas been done within a...

Full description

Bibliographic Details
Main Author: Serra Garet, Arnau
Format: Bachelor Thesis
Language:English
Published: Uppsala universitet, Högenergifysik 2023
Subjects:
Online Access:http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-501328
Description
Summary:The Radio Neutrino Observatory in Greenland is situated at Summit Station. Due to its isolatedlocation the main power to support the autonomous radio stations is solar energy. However,during the winter months when there is no sun another strategy needs to be found. This projecthas been done within a group at Uppsala University, which aims to use wind turbines to powerthese stations during the long winter. We firstly analysed the performance and power output of SAVANT G turbine, a variation ofa Savonius turbine, for different periods of time and turbine system configurations. From thisanalysis we obtained the relation between the power and wind speed. Subsequently, this powercurve has been used in a model to predict the live time fraction of the radio stations during thewinter months. The model uses a database of one minute meteorological measurements fromNOAA at Summit from 2008 until 2022 and the power curve to calculate one minute powergeneration estimations. Knowing the electronic characteristics of the station, the model predictsthe charge status of the batteries and ultimately the number of hours that the batteries are chargedenough to power the radio station. The predicted live time fraction obtained with a parallel turbineconfiguration is 58%±7%, while for the parallel configuration with dual MPPT is 54%±4%. Theuncertainty was calculated taking the standard deviation of the fitted curve as upper and lowerpower curve limits using a 1% relative error and a 0.5W absolute error for each data point. Thenthese power curve limits were used to compute the live time fraction limits.1