Power Electronic System for Multi-MW PV sites.

The work presented in this thesis addresses the optimization potential of large PV power plants with respect to energy production during periods of moving clouds. Presently the number and size of utility scale Photo Voltaic (PV) power plants in the megawatt range is increasing and the market for solar inverters is under a severe pressure regarding cost reduction. The main topic of this thesis is the investigation of the potential advantage of applying string inverters with multiple Maximum-Power-Point-trackers (MPPT) in large PV plants compared to the use of one large central inverter with one MPPT. The analysis is based on a comparative investigation of non-uniform irradiation events caused by moving clouds during a period of one year. A set-up for the long term recording of data from the 2.1 MW Danfoss Solar Park in Nordborg (DK) and the 62 kW PV plant at the University of Southern Denmark in Sønderborg (DK) was implemented. A total of 17 PV-inverters have been monitored during a period exceeding one year and the recorded data constitutes the basis of this investigation. A part of the 2.1 MW PV plant was reconfigured to emulate the behavior of a central-inverter and solar panels distributed over a distance of 160 m. In parallel a string based inverter configuration was established with solar panels at the same locations. An analysis of irradiation data recorded during the test period showed that non-uniform irradiance due to moving clouds is expected to influence the PV plants for less than 4 % of their operational time. The resulting difference in energy production between a system with 3 MPPT and a system with 1 MPPT is calculated for all days where both systems had comparable operational conditions. It can be concluded that both the estimated and the calculated difference in the annual production of energy is in the range <0.3 % and around the limit of what can be registered in the PV plants. It has further been shown theoretically as well as experimentally that landscape variations result in energy production losses. Two other methods were investigated by applying the recorded data from the 2.1 MW plant. The simulation of a dynamic string allocation concept for fast reallocation of PV strings in parallel show the potential of an increase in annual energy production of up to 0.6 %. The concept allocates PV strings in parallel during periods of low irradiation to avoid low inverter efficiency at low power levels. The effect of a sorting of the panels in the 2.1 MW park has been simulated to show the potential gain by applying PV sorting during the construction of a large PV plant. A sorting of the mounted PV panels is estimated to increase the annual energy production by approximately 0.4%. A portable IV-scanning instrument for the fast long term characterization of solar panels was developed as part of the project. Each second a sweep of the IV-characteristics of a solar panel is performed and the result stored for later analysis. The instrument is based on an active load, is optimized for field use, is battery operated and has been applied for the characterization of a solar panel over a period of 6 months at the ESTER Outdoor PV monitoring station in Rome, Italy.