Executive summary

Solar intermittency and grid integration are two fundamental barriers to the uptake of large-scale solar power in Australia and around the world.

Whilst much is said about the effect of intermittency on electricity networks, the information shared and views expressed are often anecdotal, difficult to verify and limited to a particular technical, geographical or social context. There is surprisingly very little real-world data on how intermittency, particularly solar intermittency, affects electricity networks.

This report provides an in-depth analysis of worldwide research and practical results on renewable generation intermittency, examining what common conclusions can be drawn from other efforts in this area, and how these may apply in the Australian context.

This project, Characterising the Effect of High Penetration Solar Intermittency on Australian Electricity Networks, produced several critical findings that help to understand the challenges and opportunities behind intermittency and grid integration. These are listed below.

Key Finding 1:
Intermittency could stop the adoption of renewable generation
Australia is already facing the situation where, in some network areas, the installation of additional renewable generation has been stopped. This is a conservative response to a lack of information about network problems intermittent renewable generation might cause and/or concerns about the mitigation measures required to address them, including cost and availability.
This needs to be urgently addressed, through rigorous analysis of both network simulations and trial deployments in the context of Australian electricity transmission and distribution systems.
Key Finding 2:
Existing research has conflicting outcomes, suffers from a lack of quality data and consequently often overemphasises anecdotal evidence
Some studies report significant cost savings can be achieved by displacing generation fuels, primarily natural gas, with renewable energy sources, supported by accurate forecasts. Others conclude that increased penetration of intermittent renewable generation (IRG) will actually increase system costs due to the required upgrade of conventional generation equipment to achieve increased system flexibility.
In studies analysing wind and solar variability some sources report wind to be less variable than solar at the second timescale, while others show wind to be more variable for the same timeframe. In other studies, this contradiction is repeated at minute and hour timescales and even when wind and solar generation are considered in aggregate.
Key Finding 3:
There is considerable intermittency in the existing electricity system
The existing electrical power system already incorporates significant load intermittency which is managed through generator dispatch and ancillary services mechanisms. As solar penetration levels increase, additional measures may be needed (for example, additional ancillary services). There is no uniform view on the level at which this will become significant, the requirements for additional ancillary services and how they can be met (possibly through more advanced control of the renewable generation itself). It is critical that this be determined.
Key Finding 4:
The effect of solar intermittency is not uniform
The effect of solar generation intermittency on the power system is context-specific and (PV), concentrating solar thermal (CST) and wind systems have different characteristics and currently must be considered on a case-by-case basis. Intermittency exhibited by photovoltaics consequently have different network impacts.
There is an opportunity to develop a more generalised approach to network assessment, ameliorating the need for detailed modelling of individual systems.
Key Finding 5:
The amount of high penetration solar generation that can be integrated is application specific
The amount of solar generation that can be integrated into utility power systems without compromising power quality, stability and reliability varies widely. The penetration level is dependent on assumptions about how the electrical system should operate, what additional measures are acceptable and what the wider system will look like in future. Re-evaluating network power quality standards (including voltage regulation for generation and load, and flicker requirements for generation) can significantly impact the costs of managing intermittency. Standards vary widely in Australia and around the world. A breach of a restrictive standard in one region may be no worse than complying with a relaxed standard in another.
The appropriateness and adequacy of power quality standards, and how they can be cost-effectively met through the coordination of network, load and generation control needs to be evaluated.
Key Finding 6:
Solar intermittency can be managed
A number of mechanisms can be employed to manage the impact of intermittency on electricity networks. Some of these include:
  • using short-term energy storage systems
  • strengthening the electricity network so that intermittency effects are not as localised
  • controlling loads in response to network requirements
  • deploying additional ancillary services (using conventional generators)
  • curtailing the output of renewable generators.
Depending on the effectiveness of solutions that combine these mechanisms, managing intermittency may require additional generating units for regulation duty or additional fast-response generators. The choice of measures will also affect the economics of solar generation. The effectiveness of these mechanisms both individually and cooperatively requires further investigation via modelling and experimental analysis as well as real-world Australian trials.
Key Finding 7:
Accurate solar forecasting is essential
Accurate forecasting is vital for the successful integration of large amounts of solar generation. Intermittency can be planned for and managed most cost-effectively with appropriate long (years), medium (months/days) and short term (minutes/seconds) forecasts. This is needed for network planning, and grid and market operation, including accurate generator unit commitment scheduling.
In order to support this work, there is an immediate need for high resolution solar data from both large-scale solar systems and large numbers of small-scale solar systems aggregated. This will additionally support further investigation into the effects of temporal variances on the Australian electricity network.
Key Finding 8:
Research and demonstration work is required in Australia
There is an immediate need (expressed at this project’s industry forum) to develop tools that enable the impact of small scale PV on distribution networks to be assessed. Of particular concern to utilities is the perception that PV causes over-voltage problems. This issue needs to be resolved in the context of Australian network configurations (which tend to be sparse and have higher impedance than observed in overseas studies).
In assessing future high solar penetration scenarios, it is necessary for analysis carried out in other countries to be performed within an Australian context. Local research will need to determine the type of ancillary services required and whether existing mechanisms are sufficient for intermittency compensation. These impacts need to be assessed for both large-scale and small-scale solar systems. Future work should include:
  • Development of evaluation tools for DNSPs to assess the impacts of, and develop appropriate mitigation responses for coping with, increasing levels of PV within the distribution network
  • Reconciliation of conflicting information in scientific literature on the impacts of intermittent renewable generation
  • Undertaking a large-scale assessment of the characteristics of generation, load and networks in Australia to determine the applicability of international results and the extent to which Australian networks might require special consideration
    • Consequently, the requirement for intermittency mitigation measures (for example network storage, load management, generation curtailment or additional ancillary services) and the most cost-effective approaches to meeting these at different penetration levels can be assessed
  • Collection of high resolution (temporal and spatial) solar data to support:
    • Development of accurate solar forecasting tools, both for long-term planning and short-term network management
    • Assessment of different large and small scale PV architectures
  • Making detailed case studies and investigations publicly available, about specific intermittency issues and situations (both on actual networks and via modelling and experimental analysis), for assessing the issues and required solutions
    • These should be relatively detailed and could be performed by research groups in conjunction with industry. The modelling, experimental analysis and investigation of case studies would be required at all levels and timescales (e.g. distribution through to system level, and short through to long timeframes).
  • Maintaining industry engagement as initiated through the intermittency workshop and stakeholder perspectives survey undertaken in this project to ensure:
    • Research is relevant and appropriate to Australian industry and the Australian context, including appropriateness to the existing systems and regulatory environment
    • A shared vision which fosters greater renewable generation penetration.

As well as in-depth analysis of worldwide intermittency research and practical results carried out in this report, high-resolution data was obtained from three Australian PV plants of different sizes and analysed to evaluate output power fluctuation ramp rates. An investigation of ten months of solar data from the Desert Knowledge Australia Solar Centre (DKASC) in Alice Springs, Australia, at 10-second resolution, found that high ramp rate events occur more frequently at smaller timescales, with observed power output reductions exceeding 66% of plant rating within a ten second period. Further data collected and analysed came from a 22 kW PV system at the CSIRO Energy Centre in Newcastle with 5-second resolution, representing a small-scale PV system, and from a 1.22 MW PV system at the University of Queensland (currently Australia’s largest flat panel PV system) with 1-minute resolution to represent a large-scale PV system.

Models were also developed to examine the likely impacts of fast output power fluctuations on different types of Australian electricity networks and to estimate the power output of PV plants using solar irradiance and plant size information. The network impacts model expanded an existing CSIRO model and was used to examine the likely impacts of output power fluctuations seen at the DKASC on various types of Australian electricity networks at different penetration levels. Four different scenarios comprising weak and strong grids with high and low level penetration were modelled. The model showed that if solar generation, PV in this case, is attached to a rural feeder where the grid is likely to be weak, increased penetration would likely cause an increase in voltage swings observed. This could have adverse impacts on the network’s stable operation. The four scenarios presented here are general representations of common Australian electricity networks, but detailed information on network characteristics, such as feeder impedance, is needed to evaluate the likely specific impacts of solar intermittency.

The PV output power estimation model allows estimation of the output power ramp rate probability density function. This can be used to study the effect of a particular PV array on the local network. Through simulation, and using historical or predicted irradiance levels, the developed model can be used to predict the number of short-term fluctuations, the magnitude of ramp events and the overall AC and DC output power of an existing or proposed PV plant. One feature the model considers is that the larger the collector area, the longer it takes for large passing clouds to cover an entire PV plant.