Small, grid-connected PV systems are becoming increasingly popular and represent a significant market for system developers. However, as the density of distributed generation sources increases, so too do the potential impacts on the local distribution system. Hermann Laukamp and colleagues reveal the surprising results of a housing estate case study in Germany that shed light on the issues concerned.
With rapid penetration of photovoltaics (PV) and other distributed generation capacity, the impact of PV on the grid and vice versa is becoming increasingly important. Major concerns are the maximum tolerable installed capacity and possible voltage rises on the grid from the increasing penetration of distributed generation.An extensive theoretical study in 2002 by J. Scheffler from the Technical University of Chemnitz had shown that in German urban areas, increases in voltage due to reverse power flow are the limiting factor for penetration of PV on the low voltage grid. This effect limits PV capacity in residential settlements to about 3.5 kWp per household.
However, in the city of Freiburg a development completed in 2006 featuring a PV system on every house allows these theoretical results to be compared with a real example comprising about 440 kWp across some 60 PV systems.
THE SOLARSIEDLUNG SCHLIERBERG PROJECT
The Solarsiedlung Schlierberg project is in a housing estate in Germany’s Freiburg. Built from scratch on the site of a former garrison, it is located in the southern outskirts of the city with older residential areas to the east and north. In the south are new apartment blocks, fed by the same transformer as the PV project, and to the west stands a newly developed city quarter.
The whole development is privately funded, initiated and pushed by the architect Rolf Disch. It features highly efficient buildings with an annual heating demand below 20 kWh/m2 (in line with the ‘Passive house standard’).
The development serves as a showcase to investigate the mutual interaction of a high PV capacity and the low-voltage grid. Crucially, it allows an assessment of the upper limit of PV capacity which can be connected to the grid without exceeding the standards set out for the distribution network.
Rows on the eastern side were built in 2000-2004, western rows and the business block ‘Sonnenschiff’ from 2004-2006.
Featuring 60 PV systems comprising a total of around 160 inverters, about 440 kWp have been installed, some 6 kWp per household.The last PV systems were connected to the grid in the autumn of 2006. Since 2003, when the first grid stability measurements were conducted, some 20 apartments, 4700 m2 of office/shopping area, some 200 kWp of PV arrays and around 40 apartments outside the solar development, but which are supplied through the same transformer, were added to the development.
THE ELECTRICAL NETWORK
The electrical network has been designed as it would have been without any consideration for the PV systems. A transformer of a nominal power of Sn = 400 kVA feeds several feeders that use 150 mm2 aluminium cables. Short circuit power at the transformer – a measure for the network impedance – is 11 MVA.A voltage drop or rise depends on the network impredance and that is reflected in the short circuit power Sn. Distribution cabinets, where all flats are individually connected to the main feeders by 35 mm2 aluminium cables, are located along the roads.
The guidelines for connecting distributed generators to the grid were issued by the German association of Electricity Companies or VDEW.This guideline allows a voltage rise of 2% of the nominal voltage of 230 V from local generators, that is 4.6 V. For the transformer at the Solarsiedlung development this corresponds to about 220 kVA inverter power.
Voltage quality at customers’ premises is described by the European standard EN 50160, which defines several quality criteria and corresponding limits. It requires that 95% of data taken at 10 minute intervals are within the respective limit. Due to the transition of the nominal voltage level from 220 V to a common 230 V,the upper voltage limit was set to 106% until the end of 2007. In 2008 the upper limit is set at 110%. However, the 106% limit, corresponding to 244 V, has been used here.
PV systems are constructed using single-phase systems as a building block. Depending on the available roof area, defined by factors such as the width of an apartment, a single house system comprises between two and four single phase systems. Nonetheless, the PV arrays feature true building integration with the modules constituting a part of the building skin, covering the south facing roofs and providing shade to upper balconies in summer.The inverters are also mounted outdoors as part of the architectural concept.
All the inverters were manufactured by the German company SMA. Earlier systems employ SB 2000 and SB 2500 type inverters. Later systems in the westerly rows and in the Sonnenschiff block mainly employ SB 5000 as well as some SB 3300 and SWR 3100.
PV systems and loads are connected in parallel at the distribution cabinet of each apartment. Main cabling is thus loaded with the balance of residential generation and consumption.
MEASUREMENT CAMPAIGN AND RESULTS
The PV installations had been completed by September 2006 and thus for 2007, regular operation of the whole development could be expected. Measurements were taken in both summer 2006 and summer 2007 and to represent the worst-case scenario in terms of voltage rise, measurement periods included a part of the summer holiday season, when presumably many families are on vacation – and consequently low demand is seen – but while the PV output is at a maximum.
Measurements were taken at three network nodes including at the end of feeders 1 and 2 and at the transformer low voltage terminals (MP1, 2 & 3 respectively) with 10 minute average values recorded, according to the EN 50160 standard.
A number of issues were considered, including power quality related to the EN 50160 standard, the voltage level at the most remote network nodes, power flow across the transformer, and harmonic current injection by the inverters.
Most focus was placed on feeder 1 (MP1), because additional PV systems on this feeder had been connected since earlier measurements and would therefore be more likely to show significant change.
Analysis of the transformer operation gives the reference voltage for the remainder of the electrical system. Voltage distribution at the transformer shows a nearly Gaussian distribution at around 235 V.This is 5 V above the nominal grid voltage of 230 V. Typically, utilities adjust transformer voltage slightly above nominal voltage to ensure a sufficient voltage level at the end of the feeder in the presence of voltage drops along the feeder.
During the week of highest export the maximum observed power delivery was 150 kW, which occurred on a Sunday. Chosen to show the impact of reverse power flow, the 150 kW is far below the rated transformer power of 400 kVA and during the construction phases of the development, the local network operator had watched the ratio of generation and loading on the transformer and added loads proportionally to the progress of PV generation installation.
In an earlier measurement campaign in 2003 some 15 kW of power imbalance had been observed between phases L1 and L3 during power export and by 2006 this imbalance was found to be at about 30 kW. Phase 1 delivers most power and this corresponds with a slightly higher voltage of L1. It is consequently assumed that the inverters are not distributed evenly on the three phases.
While there is no evidence of violation of the standard, some 0.5% of data points for L2 did exceed the voltage limit. However, such events indicate a large number of very short duration voltage excursions. In the standard there is no hard limit given for these short events which last 10 ms-100 ms. Voltage at MP1 increases as a function of delivered inverter power. It should be noted that the observed voltage rise is due to the generation of all the PV systems on that feeder, not only from the power of the last system. To assess harmonics generation from the inverters the Total Harmonic Distortion of the voltage at MP1 was investigated. The distortion level falls well below the standard’s limit of 8% and shows no dependency on PV.
NETWORK FOR SUCCESS
A measurement campaign in an urban PV settlement shows that distributed generation from PV systems with a high capacity only slightly affects the quality of the grid.
Power quality is affected only with regard to voltage levels and measured voltages fall within tolerances given for 2007 and well within those for 2008 and beyond. For the whole area supplied by the transformer, including buildings without PV, there is an average of 4.4 kWp of PV per apartment. The maximum tolerable capacity of PV to a single feeder was found to be an average of 7 kWp per apartment. By reducing the set voltage of the transformer to the nominal system voltage level – 230/400V – an even higher PV capacity could be accommodated with an additional 5 V margin available for voltage rises from local generation. Power imbalance between the phases was found to be increased because of an uneven distribution of inverters over the three phases but this should be easily avoided by integral planning of inverter distribution.
During the peak summer generation periods of 2006 and 2007 grid quality measurements at various network nodes were taken to assess the impact of this high level of installed PV capacity. Certainly, power flow across the transformer is reversed under sufficient irradiance and reached up to 150 kW in July 2006.A phase imbalance at the upper voltage range was also noted, again probably due to uneven distribution of inverters across the phases. An increased voltage level at the furthest house was also observed. Nevertheless, the PV systems at the settlement did not cause any violation of the power quality standard EN 50160. Harmonic injection from the inverters is well below permissible limits and below that of some conventional appliances as well.
These latest results back measurements made during the construction of the settlement and when only a fraction of the PV and load sources were included and which had also passed muster without showing violations of power quality standard.
Hermann Laukamp, together with Javier Diaz, Thomas Erge, and Guenther Ebert, are from the Fraunhofer Institute Institute for Solar Energy Systems. e-mail: Hermann.Laukamp@ise.fraunhofer.de
This work was performed within the European Project PV-UPSCALE.The authors gratefully acknowledge the financial support from to the European Union in the framework of the programme “Intelligent Energy for Europe” (EIEE).