H. Tyralis, Spatial and temporal analysis of electricity demandin Greece, MSc thesis, 95 pages, July 2016.
In this study we analyse the electricity demand in Greece for the time period 2002-2014 and we simulate the electricity demand in Greece and Thessaly. We visualize the electricity demand in Greece for the time period 2002-2014. We search the relationship between the energy demand and the Gross Domestic Product (GDP) and the temperature. The analysis shows that in general the conclusions of Psiloglou et al. (2009) are still valid. However, since 2010, when the GDP began to decline, the pattern of the energy demand has changed in the winter. Moreover, we investigate the effectiveness of the energy demand forecasting system of the Independent Transmission System Operator (IPTO). The results of the analysis are useful for forecasting the energy demand and for generating synthetic time series. The data and the code of this work are available as supporting material. We analyse the spatial patterns of the energy demand. We present patterns of energy demand for various uses and combinations of energy demand variables for various uses with variables such as the GDP, the population, the area and the population density for the time period 2008-2012. We perform clusters and outliers analysis, hot spot analysis and grouping analysis. The most important results of these analyses are presented in the main body of work, while the total of 1 125 Figures produced during the work, are provided as supporting material. We believe that the results are useful to understand issues related to the spatial distribution of the energy demand in Greece and the developmental orientation of the state, to improve previous works, which use socioeconomic variables. We simulate the total energy demand in Greece and Thessaly, after the removal of the energy demand for agricultural use. Furthermore, we simulate the energy demand for agricultural use in Thessaly. The simulation concerns a long time period. It is performed to assess the long-term properties of an energy project. Therefore, it neglects features such as the dependence on initial conditions. The simulation of energy demand is performed after the removal of the annual averages and the observed periodicities. The synthetic series is produced using a bootstrap method with blocks of random length derived from a geometric distribution. The removed elements are added to the synthetic time series. Summarizing the results of the analysis in the time domain, we observe during the day two local maxima at approximately 12:00 and 20:00, and two local minima at about 04:00 and 16:00. We observe daily maxima of the energy demand on Wednesdays or Thursdays and daily minima on Sundays. The shape of the daily energy demand is almost independent of the month. We observe local maxima of the monthly energy demand in January and July and local minima in October and April. The December monthly energy demand increased significantly after the hydrological year 2011. We observe less variation in energy demand in the period between December and April. A linear model for the relationship between the energy demand and the GDP seems reasonable. Moreover, we observe that the GDP increase results in increased energy demand (or vice versa). Nevertheless, the GDP increase in the hydrological year 2008 was followed by a decrease in energy demand. We observe a global minimum of the energy demand for temperatures at about 17-18° C, and local maxima of temperatures at about 3° C and 32° C. A regression line for the energy demand (the y axis) and temperature (the x axis) would be a convex curve. The IPTO’s daily predictions usually overestimate the energy demand. The results of the investigation will be useful for the construction of a short and a medium-term energy demand forecasting model. Regarding the spatial analysis, we show some Figures from the support material which present significant results. It seems that in Greece’s middle there is an area which is classified as industrial but also agricultural. The Greek islands are mainly characterized as commercial, while the area around Athens is characterized by high values of household energy use. Regarding its development model Greece is divided to the main continental part, which includes areas in the middle and North, to Central Greece and to island regions. The island regions are similar to the Athens area and Peloponnese in specific cases. These results may be useful for the management of the electricity grid and for finding optimal policies for the development model of the state. In regard to the simulation, we did not remove the industrial energy demand use from the data. The industrial use is characterized by different periodicities compared e.g. with the household use. The result of interest was the production of a synthetic time series to simulate a power system and assess its long-term performance. For this specific application it was sufficient to examine only the time series electricity demand. However, if we are interested in forecasting we must take into account the forecasts of temperature, GDP etc. In this case it is necessary to correlate the electricity demand with other variables.
Our works that reference this work:
|K. Mavroyeoryos, I. Engonopoulos, H. Tyralis, P. Dimitriadis, and D. Koutsoyiannis, Simulation of electricity demand in a remote island for optimal planning of a hybrid renewable energy system, Energy Procedia, 125, 435–442, doi:10.1016/j.egypro.2017.08.095, 2017.