6.2 Accommodation of HP-IRG in Australia
Various articles in the existing literature discuss the work required to facilitate the integration of high penetration intermittent generation. The recommendations include:
- Develop models representing solar behaviour and net load. Accurate models are needed to quantify the performance and economic impacts on regulation and load following according to  and for reliable forecasting 
- Determine the required flexibility of non-variable generation to manage the variability introduced by HP-IRG. Again, accurate models on solar are necessary to determine this .
- Simulations to determine the impact of HP-IRG on power quality at the distribution level 
- Investigate how flexibility can be introduced (or variability mitigated) into electrical networks through the utilisation of and load participation. This leads to thinking about centralised control concepts and Systems (EMS).
For successful integration of HP-IRG into the Australian electricity network, it is necessary for this work to be performed within an Australian context. For example, to determine the extent of flexibility required in the Australian electricity network, studies which consider Australia’s generation mix and the individual capabilities of all conventional generators need to be performed. To determine the impact of the introduction of storage for a certain feeder with large amounts of(PV), the local load profile and grid characteristics have to be known.
The need for site specific studies is either mentioned explicitly or implied in the literature. It is stated in  that “to understand how regulation will be impacted, data will need to be gathered at high resolutions of up to 10 sec. Data will also need to be synchronised with load data to give a clear picture of the net impact of varying load and generation.” Obviously the solar and load data to be synchronised in this case needs to be sourced from the same geographical location. Another statement made in  is “Site specific data is required to characterise the spatial variability of solar irradiance over areas of 1km2 or less. This would enable optimised site selection ofplant improving performance. This data could also be utilised to develop site specific forecasting”. From , “Weather systems can cover very large areas, reducing the advantages gained from an increased load balancing area. Studies are required to determine the correlation of weather across the spread of the electricity grid - ensuring links between areas with weak correlation” and “Detailed load profile data for individual homes would allow for accurate modelling on the impact of PV and storage. The results of such modelling would be a reliable basis on which to gauge the economic viability of PV and storage”.
This section of the report presents studies completed in other countries which are considered necessary to assist in the integration of HP-IRG. Findings from some of these studies can be applied in Australia, but ultimately similar studies will need to be carried out in an Australian context, taking into consideration unique Australian conditions.
6.2.1 System level studies
The studies presented in this section are all considered system level studies and are mainly related to the flexibility of an electrical network. These studies would assist in identifying vulnerabilities in Australia’s generation mix in relation to required system flexibility to manage HP-IRG, and help determine the level of intermittent renewables penetration Australian electricity networks can manage.
The inputs used to complete these studies can be categorised as:
- System load profile – A representation of the demand over the course of differing time intervals for differing sample rates
- Renewable generation profile – A representation of the power generated from renewable energy sources over the course of differing time intervals for differing sample rates.
- Generation mix – The mix of conventional generation (coal, gas, hydro and nuclear) which makes up the generation set for the system
- System flexibility – Can be described as the ability of the aggregated set of generators to respond to the variation and uncertainty in net load . Flexibility can be introduced into electrical networks by using energy storage and load participation (e.g. load shedding). Measures of system flexibility are :
- Minimum load – The lowest power output the generation set can manage without being forced to turn off. The lower the minimum load the less likely generation will be forced to shut down during periods of light net load brought on by high solar generation.
- Start/stop speed – The time required for generators to start up and shut down
- Ramp rates – The rate at which generation can increase or decrease output
- Load balancing area – The area over which generation output is matched to demand. For the Australian National Electricity Market (NEM), this is the entire east coast of Australia.
188.8.131.52 Estimation of net load
For a generation set designed to meet net load requirements as opposed to peak load, its flexibility is the measure of its adequacy. To determine the necessary flexibility of the generation set, an accurate estimation of net load for a given penetration level of renewable generation is required. Models characterising net load need to clearly represent the expected variability at the timeframes of interest. In the NEM, generators are scheduled and dispatched into production to match supply every five minutes , as can be seen in Figure 56. NEM need to work out how much generation flexibility is required to manage an expected increase in variability from the introduction of renewable generation. To assist in this process, the amount or extent of variability in net load the network might experience in a five-minute interval needs to be determined.
Figure 9 from Section 3.1.2 is an example of a presentation method to characterise the variability in net load. This was for a study performed on the Californian grid where actual historical data was obtained for wind, load and solar. This data was projected forward for modelling 20% renewable penetration by 2010 (2010T) and 33% renewables by 2010 (2010X) and 2020 (2020). Figure 9 shows the expected variability of net load for the 2010X scenario, 33% penetration of renewables. Also shown is the variability of original load (i.e. without intermittent generation) for comparison. This study could be replicated for the NEM at five-minute intervals instead of hourly, which would enable system planners to calculate the flexibility gap in its generation set. For the NEM to commission work of this nature, it would require a renewable generation output profile scaled up to the penetration level of interest as well as the load profile. An accurate scaled up profile of renewable generation for the NEM would ideally be a combination of:
- existing generation output data scaled up to some degree
- estimation of generation at new sites based on a wind and solar resource model.
The Australian Energy Market Operator (AEMO) uses the Australian Wind Energy Forecasting System (AWEFS) which provides a wind forecast every five minutes. In the case that there is insufficient recorded data at existing and proposed wind generation sites, data from the AWEFS could be used instead. Unfortunately there is no equivalent resource for solar yet. It may be necessary to begin accruing irradiance data at existing and proposed sites at five-minute intervals and combine this with historical satellite data to develop an adequate solar model.
The influence large amounts of PV have on the net load profile is looked into in more detail in . Thegovernment has set a target of 53,000MW of PV to be installed by 2030. Given the fluctuating nature of PV generation, a proper evaluation on net load variability is deemed necessary. Nagoya City is the focus of the study; the city is broken up into a grid pattern with the resultant net load calculated for 8000MW of installed PV with three different distributions. PV output is calculated using measured insolation for five sites located within Nagoya City and the load data sourced from recordings by the local electric power company in 2000/2001. The three cases are:
Case 1: Most concentrated – 50 blocks with 160MW of PV per block. The blocks are located in more densely habituated regions of the city.
Case 2: 150 blocks of 53MW of PV per block.
Case 3: Least concentrated – 500 blocks of 16MW of PV per block.
Figure 57 illustrates the fluctuation levels of net load for each of the abovementioned case in comparison to the case with no PV, showing how the impact varies according to distribution type. It is seen that the net load fluctuations are greater when PV generation is more centralised. The top left chart of Figure 57 shows the most variability in net load due to its greater concentration. The standard deviation of net load with PV is at times 200MW greater than without. The significant increase in apparent load fluctuations with the integration of PV would require significant measure for load frequency control . Note the difference in variation between the most concentrated and the least concentrated (bottom chart) cases, with the increase in net load fluctuations being minor for the lower concentration case. The peak load for Nagoya City is around 25 GW, making 8 GW of PV correspond to a penetration level of 32%.
At the system level, this study demonstrates the variation in influence PV has on net load according to its geographical orientation. Ifwas to have a target of high penetration installed PV, similar studies would need to be performed during the planning stages. For the NEM, findings from such studies are unlikely to be similar. The NEM covers the entire east coast, far greater than the area of Nagoya City, so the area over which the solar resource needs to be defined is also far greater with a greater diversity in PV generation across the network. The density of load is also far greater in Japan than in the NEM, meaning these are two very different electrical systems. The NEM, covering the entire east coast of Australia, supplies 25 GW on a typical day , the same as for Nagoya City, which is approximately 325 km2 in size.
184.108.40.206 Displacement of conventional generation
It is important to understand that variability and uncertainty are inherent characteristics of all power systems including the Australian NEM. Loads, power lines, and generator availability and performance all have a degree of variability and uncertainty. Regulations, standards, and procedures have evolved over the past century to manage variability and uncertainty to maintain reliable operation while keeping costs down. In general, system operators and planners use mechanisms including forecasting, scheduling, economic dispatch and reserves to ensure performance that satisfies reliability standards in a least cost manner.
A study in  looked into how the integration of PV at penetration levels of 10%, 20% and 30% would impact on the existing generation mix. Figure 16 from Section 3.3 shows the load duration curve from CAISO. The graph is based on 2007 load and PV data projected into 2010 for scenarios of 10%, 30% and 50% PV penetration. Overlaying Figure 16 over the generation mix gives Figure 17 and shows how the increased levels of solar displace generation in the existing fuel mix. Looking at the US fuel mix, the majority of the solar displaces gas fired generation with coal fired generation being displaced into for short periods. Figure 58 compares the Australian fuel mix for the NEM with the US. It can be seen that the US utilises approximately 50% coal fired generation whereas the NEM uses more than 80% and if this fuel mix was laid over Figure 16, there would be a major change in the Australian generation mix.
High penetration intermittent generation creates a need for increased system flexibility. A study can be performed locally to determine feasible options that could be implemented or integrated into the different Australian electricity markets to cope with this requirement. Various ways in which the system flexibility in various Australian electricity networks can be enhanced need to be investigated.
- load control
- ancillary services
- spinning reserve
- energy storage
- renewable generation curtailment
- upgrading of conventional generation mix.
Source: U.S. EPA, eGRID year 2005 data
220.127.116.11 Study – how system flexibility limits intermittent renewable generation penetration levels
A study performed on the Electric Reliability Council of Figure 28 in Section 3.4.3 show how the system flexibility impacted on the amount of wind curtailed. At 50% wind penetration, curtailment dropped from 50% to 20% for a 10% increase in system flexibility. Inputs into this study included the load profile for ERCOT, the local wind generation profile and the theoretical minimum load levels. Based on the system flexibility, this study gives indication of the level of IRG penetration possible; when a percentage increase in IRG results in a similar percentage increase in curtailment then the limit is likely reached. The study also looked at required curtailment levels for various percentage mixes of wind and solar generation. The effect of storage on increasing system flexibility was also investigated in the study, the results of which are shown in Figure 30 in Section 3.4.3. It is seen that a minimum amount of four hours storage reduces the fraction of curtailed IRG from nearly 35% to less than 20% for 80% penetration of IRG.(ERCOT)  looked into the degree of curtailment of intermittent renewable generation (IRG) required for different mixes of wind and solar generation for varying levels of system flexibility. System flexibility refers to the flexibility of the conventional generation fleet which is characterised in terms of parameters such as minimum start-up and shut-down times, minimum stable generation and ramp rates. Some results from the study presented in
This study uses theoretical values for system flexibility, and for it to be applicable in Australia it would need to use the local generation mix and models of solar and wind generation based on local wind and irradiance data. Work of this type would be of great assistance in determining possible levels of intermittent renewables penetration for the system and to identify what particular generation, due to its relative inflexibility, is limiting the penetration level. A study could also be performed to investigate whether introducing storage would be a more feasible option of increasing system flexibility than introducing more flexible generation. This study could also determine the cost savings that could be achieved by introducing storage through alleviating the need to decommission inflexible generation and reducing the curtailment of HP-IRG.
6.2.2 Distribution level studies
The following studies revolve around possible impacts of HP-IRG at the distribution level. The studies look into the impacts of large amounts of centralised and distributed solar on frequency and voltage regulation, power flow, harmonics and voltage fluctuation.
An analysis on the voltage behaviour for a PV demonstration project in Ota, Japan is presented in . The project consists of 553 PV systems, ranging in size from 3-5kW, in a 1km2 area; 80% of homes have PV on their roofs with a total installed capacity of 2.1 MW. The main focus of the project was to look into the effectiveness of the Power Control System (PCS) installed at each inverter. The PCS uses active and Figure 59. The bottom graph is the aggregate PV output for the project for the day, where conditions were particularly sunny with very little fluctuation in output. The top plot of Figure 59 shows the kind of variation in voltage possible from house to house: at worst case there is almost a 4% variation at midday. The study claims higher voltages are observed for PV systems with higher impedance between their output and the nearest pole transformer, demonstrating that the degree of impact PV has on power quality is influenced by grid characteristics. The voltage fluctuation is within tolerance (±6V for Japan at 100V) but this is with PCSs and battery storage. The results are likely to be different for Ota without the power flow management in place. According to the Japan PV2030 roadmap, 50% of residences are forecast to have PV on their roofs by 2030.control in combination with battery storage to manage voltage rise. Of interest is the kind of variability in voltage levels seen at each of the homes, shown in
To facilitate similar penetration levels and gain confidence in the management of increasing solar penetration levels in Australia, demonstration projects like Ota are a necessary preparation and need to occur locally. There are a number of long and ‘skinny’ feeders in Australia, most of which are located in rural networks. For example, Ergon Energy has one of the largest distribution networks in the world, with electricity infrastructure assets across one million square kilometres of regional. Its service area covers 97% of the state and has one of the lowest customer densities of any network in the western world. High penetration of solar power in such networks with high impedance feeders is likely to cause voltage fluctuations beyond acceptable limits due to intermittent solar power output caused by passing clouds. Similar studies to that carried out in Ota, Japan, need to be performed in local areas with high impedance feeders to investigate the impacts of high penetration solar on such networks.
The impact of a 1 MW PV system on a local feeder is analysed in . The system is installed at the Main Stadium of the 2009 World Games in Kaohsiung, Figure 60 is a representation of the electrical system for the stadium. The PV system is connected to Bus 7, labelled ‘PVS’. Simulations using this model and increasing levels of PV were conducted to determine the level of voltage fluctuation introduced into the feeder by the PV system.and connected to a Taiwan Power Company (TPC) feeder. The single line diagram shown in
A probability distribution of PV generation was developed using hourly historical weather data and a model of a solar cell. Figure 61 shows the resultant PV output profile. The sample rate was not provided. As voltage fluctuations are calculated at the same rate as the generated PV output profile, it is important that this rate is high enough to capture the change in PV output due to passing clouds. It is stated in  that the PV output profile is affected by the random passing of clouds. To investigate the impact of the PV the circuit breaker at MF65 is opened (making Bus 7 the end of the feeder) and the feeder tie switch is closed, leaving the PV system at Bus 7 as the only power source at that end of the feeder. The voltage fluctuation at MU67 is then examined through simulation using historical load data and the generated PV output profile mentioned above. The results show voltage fluctuation starts to exceed constraints of 2.5% when the capacity of the PV system is around 3.5 MW. The rating of the feeder is assumed to be around 6 MVA as power is supplied to the loads through three 2MVA transformers, indicating voltage fluctuation constraints are being breached with PV penetration of around 58%.
This study is of interest, but would benefit from measured high resolution irradiance data. The generation profile of Figure 61 also looks to be representative of a fairly sunny day, with not many fluctuations. An impact analysis using a partly cloudy day would likely show more extreme voltage fluctuations. Taiwan is looking to increase PV installation levels to 2 GW by 2025 and studies like these are being conducted to try to better understand the impact of this goal.