# Modeling of Photovoltaic System with Power Smoothing for Integration in Power Systems

[Examensarbete på avancerad nivå]

Solar PV was the second largest installed power capacity in the EU 2014 and the rate at which PV is installed in Sweden has doubled each year for the last four years. Both the intermittent nature of PV power and the fact that PV systems do not provide any system inertia can cause a reduced frequency quality in the power system. Due to the fast growth of the PV market, it is of interest to develop methods which reduce PV systems’ negative impacts on the power system frequency quality. In this project, a power smoothing algorithm is developed and implemented in a modeled PV system. The PV system is connected to a simplified power system model in which the peak PV power constitute 10 % of the total power production and the remaining power is produced by a hydro power plant. An analysis of the frequency quality with and without smoothing has been conducted when logged irradiance data from the month of April 2015 is used as input to the PV system. The analysis shows that the smoothing algorithm improves the frequency quality in the simplified power system model. The mean- and maximum positive frequency deviations are improved (reduced) by approximately 40 % and 50 % respectively. Furthermore, the mean- and maximum Rate of Change of Frequency (ROCOF) are improved (reduced) by 20 %. The time during which the frequency goes above 50.9 Hz is reduced from 207 seconds to zero seconds. The time during which the frequency is above 50.5 Hz is reduced from 8.8 % to 0.4 % and the time during which the frequency goes below 49.5 Hz is reduced from 8.9 % to 5.5 %. The energy loss due to the use of the smoothing algorithm was 0.27 %. This can be compared to the average degradation of PV systems, which is around 0.5 % per year. Furthermore, an existing method to characterize a PV panel using measurements is further developed in this project. The further developed method results in a PV panel model that, when compared to the measured PV panel characteristics, gives a residual sum of squares (RSS) of 1.3. Another method, which uses a European standard to find model parameters, has also been tested. However, the difference between the model and the measured PV panel characteristics is higher using this method, with the RSS of 3.17. Thus, the method further developed in this project gives a more accurate fit in terms of RSS error.

**Nyckelord: **Photovoltaic system modeling, power smoothing, ramp rate control, ramp limit, frequency control, frequency quality, PV panel modeling, PV system, smart grid, distributed generation

Publikationen registrerades 2015-09-29. Den ändrades senast 2015-09-29

CPL ID: 223356

Detta är en tjänst från **Chalmers bibliotek
**