10 Model to examine likely impacts of intermittency on Australian electricity networks
In order to understand and investigate the likely effects of various penetration levels of intermittent generation on different segments of Australian electricity networks, we consider four scenarios:
- A weak (e.g. remote and relatively small) network with a small amount of
For a weak network, intermittency effects will be more pronounced than in a larger network with a concordant potential capacity for stability issues. However, as only a small number of customers rely on the network, detrimental effects will be limited to a small number of consumers. If we consider an extreme case of a small isolated power system, such as a microgrid, even a small amount of intermittency may result in an imbalance of power production and consumption on short time scales
- A weak (e.g. remote and relatively small) network with a large amount of PV generation
For a weak network, the PV generation will have a strong effect upon the system stability
- A strong (e.g. urban and relatively large) network with a small amount of PV generation
Here, the PV intermittency will be largely absorbed by the strong network. However, the small amount of variation can cause local voltage problems. The typical case is a PV rooftop array where local voltage swings caused by the PV intermittency will not bring down the network or cause network level problems, but may cause voltage variation locally. The most likely detrimental outcome is that the PV inverter will detect voltages outside its allowed operating range and shut down. The effect of this will be that the consumer may not notice anything wrong but the system will operate at less than its planned production efficiency and as a result consumer revenue will be decreased.
- A strong (e.g. urban and relatively large) network with a large amount of PV generation
This situation has the potential to affect a large number of customers.
A simulation model developed at CSIRO was used to examine the likely impacts of output power fluctuations seen at the Desert KnowledgeSolar Centre (DKASC), Alice Springs, on various types of electricity networks in Australia.
Figure 132 shows actual data recorded from the DKASC PV array over a 15 minute period on a partly cloudy day. As the insolation changes over the course of 15 minutes, the power supplied from the array can be seen to vary from a high of 167kW to a low of 20kW. At the same time, the voltage measured at the Solar Centre varies between 248V and 253V, giving a range of approximately 4V.
The four scenarios mentioned at the beginning of this section were simulated using data from the DKASC as shown above in Figure 132. The model used is a small-signal (time series) model developed at CSIRO for studying the effects of renewable generation attached to the grid. As shown in Figure 133, the grid is represented as a generation source, with series impedance representing the lumped impedance of the generators and the line connecting it to the PV array at DKASC. In this way, it is possible to represent a strong or weak grid simply by changing the values of the lumped impedance specified as L and R in the diagram. The load can also be varied in the model to show the effects of different penetration levels at the point of coupling to the network. The DKASC array is represented as a current source with a nominal maximum (nameplate) capacity of 200kW.
The simulation is run for approximately 15 minutes of actual data obtained from DKASC, and the effects of intermittency on voltage and network stability are observed.
The first scenario modelled is a strong network with low penetration, that is, the nameplate rating of the PV array is 10% of the local load. The load is assumed to be constant - load changes are not considered, andis set at 0.75 p.u., in accordance with the values used by Lasseter and Piagi in . As the PV array at the DKASC is rated at approximately 200kW, we used a load of 2MW to approximate a 10% penetration. In line with , we set the lumped grid impedance to 0.01 p.u. based on the load of 2MW. Consistent with the assumption that line reactive impedance is much larger than resistance, we set R in the diagram to be one tenth of the impedance of L. Only the 240V side was considered. It is assumed that standard voltage regulation is installed and operational in order to correct for voltage drops expected at the load. Results for the simulation with low PV penetration and a strong grid are shown in Figure 134. The voltage variation is only 0.14V. However this kind of grid situation would probably only apply in an urban area close to a substation.
The second scenario also assumes a low PV penetration of 10%, but in a weaker grid. A weak grid by US and European experience would be 0.1 p.u. impedance, but we have chosen to simulate a rural grid under Australian conditions, so have chosen an impedance of 0.2 p.u. The load is 2MW, representing a low penetration of 10% as before. Results are shown in Figure 135. Now the voltage can be seen to sag as power from the PV array drops, and has a range of 0.8V. This is still within the allowable range for voltage and would probably go unnoticed by customers.
The third scenario treats a strong grid with high penetration. As before, the strong grid is modelled as a lumped impedance of 0.01 p.u., and high penetration has been set at 40% of load, so the load used was 500kW, again with 0.75 p.u. reactive load. As shown in Figure 136, the voltage changes very little, only with a range of 0.35 V in spite of the high penetration, because the PV array is attached to a strong grid. This shows the importance of intermittent renewable generation being attached to a strong grid if this is possible.
In the final scenario, we examine a high penetration of PV on a rural feeder. Here we assume a long feeder, typical of Australian conditions, where the grid impedance is considered to be 0.2 p.u. The load is 500kW, giving a penetration of 40% at peak PV production. The results are shown in Figure 137. Here the voltage swings by 4V, which is similar to the actual data recorded from the DKASC array - compare with Figure 132. The voltage swing is still within the allowed range, but such variations can be of concern when the voltage is close to the upper or lower limits.
Power generated by PV arrays is by nature intermittent, and can experience large swings in power output over only a few seconds or minutes even for an array of 200kW like the one at DKASC. When the penetration is low and it is attached to a strong grid this is not an issue, however, when attached to a rural feeder where the grid is not strong, an increase in penetration caused an increase in the voltage swings observed. If the penetration were increased on this type of feeder, the voltage swings would begin to impact on the operation of this part of the network. It is likely that PV inverters would trip off with these voltage swings, causing larger power fluctuation and therefore worsening the voltage swings.