Australia leads the world in the uptake of rooftop solar (PV). As the proportion of energy from PV increases, a point is reached where the natural fluctuation of the output, due to both intermittent and sustained cloud cover, can cause problems on the local distribution network that the PV is connected to.
PV is installed and connected behind the customer’s meter and is connected, via the customer’s connection, to the low voltage network. The energy produced from PV is initially consumed on site and, at this stage, has minimal effect on the grid. If the power supplied by the PV exceeds what is required on site, the excess is put back into the low voltage network. At lower levels of PV penetration in an area, this is normally not a problem because the exported power is consumed by neighbouring loads and, although reduced, the power supplied by the grid is still sufficient to provide voltage stability.
However, as the number of PV systems increases, and the power generated from PVs in a particular area gets close to, or exceeds, the total load in that area at any point in time, the PV generation can cause problems on the local grid.
The problems that can be caused by high levels of penetration of PV systems include reduced understanding of the native load and its interaction with PV because of the lack of visibility of PV generation; voltage fluctuations due to short-term fluctuation in PV generation; increased wear and tear on transformer tap changers; spurious tripping of protection devices; and voltage rises and changes to the voltage profile of the feeder.
A project by The Australian National University (ANU) is seeking to address two of these issues: the lack of visibility of PV generation and voltage fluctuations due to short-term fluctuations in PV generation.
Improving visibility of PV generation
The ANU project, supported by ARENA and the active involvement of 12 of the 15 electricity distribution businesses in Australia, is developing and refining techniques for predicting real-time and future electricity generation from distributed PV at a relatively small geographic level.
The project is comprised of:
Solar forecasting services provided by Solcast which utilise advanced satellite weather mapping at 1km resolution updated every 10 minutes and provide a 0 to 7-day probabilistic forecast of radiation in 30-minute resolution (available at up to 5-minute resolution)
Raw data on the capacity of rooftop PV systems connected at the distribution system asset level as provided by Distributed Network Supply Providers (DNSPs) and organised into databases by ANU
A PV Power Model (Solcast) that combines the information from the two components above into a probabilistic forecast of PV output at the distribution system asset level
An Application Programming Interface (API, Solcast) that allows the DNSP to call for and receive those forecasts at the distribution system asset level
The project is also integrating information being made available from Fronius on the real-time performance of thousands of actual PV systems. This information will allow the project to fine-tune its forecasts, and improve them by taking into account information on the impact of variables like shading, and the tilt and orientation of the PV arrays.
In total, the system provides significant benefits in improving the visibility of the PV generation within the distribution network at a very localised level, including the relationship between weather variables (irradiance, cloud cover and opacity), PV capacity and other installation characteristics, PV output and operational parameters of the distribution network.
At present, the system provides information at the zone substation level, but resolution can be increased to finer scales. Current pilot projects include feeder level modelling, and forecasting for virtual power plants.
Forecasting PV generation
The ANU project, in partnership with Solcast, has demonstrated an ability to forecast PV generation within a small geographic area with a high level of accuracy, even when influenced by transient cloud cover. The modelling uses satellite imaging of clouds from the Himawari 8 Satellite and details from the DNSP on the size, location and connectivity of rooftop PV. Using its forecast of solar radiation and its PV Power Model, the project produces forecasts of PV generation.
In the meantime, DNSPs are reactively responding to the effects of high PV penetration in a number of ways including upgrading the network where required, placing limitations on the size of installations or the level of energy that can be exported, installing synchronous compensators and line drop compensation, and reducing the grid reference voltage.
Having more complete and accurate forecasts of the load that will be met by PV generation within the distribution networks will support increased certainty about the amount of dispatchable generation that will be required to supply the remaining load.