The energy sector, which until recently has been very conservative, is now facing increased disruption from increasing penetration of modern variable renewable energy sources. Maintaining the frequency remains a critical metric, but use of existing regulating standards and security mechanisms in place have been deemed ineffective and outdated.
The energy sector has historically been immune to Moore’s law, until recently. Current technology disruptors such as variable renewable energy (VRE) lead the way as operators and regulators struggle to catch up. Consequently, frequency control ancillary services (FCAS), which have historically been used to counter frequency instability, are now deemed ineffective and outdated.
According to the Australian Energy Market Commission (AEMC), in 2016 the NEM had spent more time outside of the conventional frequency band than both years prior [1]. Simultaneously, the Australian Energy Regulator (AER) reported an addition of 2 GW in generation capacity from 2012 to March 2017, of which 92% came from asynchronous renewable sources. While over the same period, 5,3 GW of synchronous coal base load capacity was decommissioned [2].
Therefore, this indicates that the recent rise in the NEM rate of change of frequency (ROCOF) and the decrease in the contingency frequency nadirs are caused mainly by the removal of system inertia due to the substitution of synchronous generation by asynchronous. Thus, with basis on the South Australia case study, this research investigated the potential impact of the aggressive renewable energy target (RET) for Queensland on its frequency stability and evaluated a battery energy storage system (BESS) as a potential solution.
Fig. 1: South Australia inertia decrease [4] (upper graph) ; South Australia ROCOF increase [5] (lower graph).
The described frequency instability scenarios are best illustrated by analysing the South Australian grid. The red line at the top of Fig. 1 shows the continuous removal of system inherent inertia led to the downward trend of the overall available system inertia [3], which is a contributing factor to blackouts. For example, in the June-July period, just months before the 2016 South Australian general blackout, out of the peak amount of 19 000 MW available inertia, as little as 2000MW was operating [4]. Consequently, the ROCOF in South Australia has also shifted from an average of under 1 Hz/s before 2011 to an average of over 3 Hz/s after 2015 (Fig. 1).
The Queensland literature void
Although South Australia presents a very interesting case study, the aggressive RET defined by the State of Queensland is set to rival that attention due to the proximity of its future energy mix to the State. When comparing the different energy mix bar graphs in Fig. 2, it is evident that the current scenario for the region (left-hand column in Fig. 2) opposes that of Queensland (centre column), with Queensland still depending on over 85% of base load synchronous generation and South Australia having over 40% of its energy generation from wind power.
However, when looking into the projected 2030 Queensland energy mix with the RETs (right-hand column), a much more homogeneous relationship is seen [5]. Since a lack of research, focused on future Queensland energy mix scenarios, has been identified, this article explores the potential challenges presented by RETs to the frequency stability in the Queensland grid.
Fig. 2: Queensland (Qld) and South Australia (SA) energy mix comparison in 2016; and Queensland projection for 2030 [5].
Having identified the potential issue, a feasible solution is proposed. The NEM ROCOF increase due to the removal of system inertia allows increasingly less time for the security mechanisms to respond. When a frequency instability event occurs, the first line of defense is the primary frequency response (PFR) service. The current timeframe of this service in the NEM ranges from six to sixty seconds from the time of the contingency.
However, a faster ROCOF following a contingency means this service would need to be activated even earlier, sometimes even within milliseconds. For example, a 1 Hz/s ROCOF would cause the frequency to drop into the under-frequency load shedding (UFLS) zone (49 to 47 Hz) within a second, long before the PFR can start responding. Thus, fast-frequency response (FFR) capabilities would be a good alternative.
The Australian Energy Market Operator (AEMO) defines FFR as: “Any type of rapid active power increase or decrease by generation or load, in a timeframe of less than two seconds, to correct supply-demand imbalances and assist with managing frequency.” [6 ] Battery energy storage systems (BESS) is the preferred FFR technique because it can provide synthetic inertia response within a 10 – 20 ms full activation timeframe to replace the lost system inherent inertia from coal- and gas-fired generators. Moreover, the power industry has also shown a preference for this option given the recent installation of the world’s largest battery (100 MW/129 MWh) in South Australia to provide an FFR solution.
Fig. 3: BESS frequency controller: Proportional droop control (Black) [8]; synthetic inertia control (Blue) [9].
Bess component modelling
We selected the BESS model described in reference [7] because of its high level of detail, its compatibility with the software and the published application example [8]. The BESS model has a very simple setup comprising a DC power storage component (Battery), a DC to AC power inverter and a controller to manage their interactions. The BESS controller was divided into three parts: Frequency, charge and PQ power controllers. However, the focus of this study is on the control of frequency, so only the first will be detailed.
The frequency controller integrates two control philosophies. The first was found in example [8] and represents a conventional proportional droop controller containing a small dead-band (see top of Fig. 3). This emulates mainly the governor controller found in most synchronous generators to provide primary response capabilities. The second integrated philosophy is found in reference [9] and represents a synthetic inertia controller (see bottom of Fig. 3),as is commonly seen in wind turbines. It emulates, with a certain degree of error, the inherent inertia response delivered naturally by synchronous generators, which can be varied by changing the inertia constant (Hsyn). There is also an added delay step in order to simulate the latency limitations of current technologies.
Fig. 4: Dynamic analysis: WECC model – Hsyn variation (upper graph); WECC model without BESS – PV penetration variation (lower graph).
Result one: Relating low system inertia to high VRE penetration
The upper graph in Fig. 4 shows the results of the WECC system with a BESS, while the inertia constant (Hsyn) is increased and the other parameters are fixed. As the Hsyn increases, so does the total inertia (synthetic) the BESS provides to the system. This leads to a more effective arrest of the system frequency drop due to the contingency event and the slope of the frequency curves becoming less steep (a lower ROCOF). Additionally, the lowest point of each curve (nadir) rises as the inertia increases. This confirms that a higher synthetic inertia provided to the system shortly after a contingency event results in a lower ROCOF and a higher frequency nadir, as expected.
The lower graph in Fig. 4 shows the increase in PV penetration percentage while all other variables are fixed. The results show a worsening arrest of the system frequency drop following a contingency event, since the ROCOF increases at the same pace as the frequency nadir decreases.
When making a comparison between both graphs it is evident that they have a directly negative relationship. While the increase of system inertia results in lower ROCOFs and higher nadirs, the increase of PV penetration results in the exact opposite. Therefore, it can be concluded that an increase in PV penetration results in a decrease of system inertia. It is also clear that the faster a mitigation solution is applied, the earlier the frequency drop is stopped, which indicates that FFR services can be very effective in such conditions.
Fig. 5: Dynamic frequency responses: Comparison of PV penetration variation with and without BESS WECC model.
Result two: Comparing FFR with and without BESS
Fig. 5 shows a comparison of the overlapping curves from the WECC system tested with and without BESS. It proves the effectiveness of the BESS system in quickly arresting the frequency drop. As the graph shows, the corrective impact is so vast that a total deviation of 1,77 Hz is observed, as the lowest nadir for the system without BESS nears 48 Hz and the system with it nears 49,8 Hz. Also, BESS FFR requires less energy to be injected into the system to reinstate the system’s equilibrium.
Result three: Comparing Queensland’s 2020 and 2030 RETs
The graph shown in Fig. 6 illustrates how the aggressive increase of PV penetration, proposed by the RET, could potentially lead to the weakening of the system’s frequency stability, since the simulation indicates that the frequency nadir drops while the ROCOF rises.
However, before the graph is analysed, some considerations must be mentioned regarding the Australian NEM Frequency Operation Standard (FOS) [2]. The green area shown in the graph indicates the “no contingency” zone (i.e. above 49,85 Hz); the yellow area represents the “normal operating frequency excursion band” (i.e. above 49,75 Hz) and the red area indicates the “contingency” zone (i.e. below 49,75 Hz). The first zone only uses normal regulation FCAS services, while the two later areas require FCAS frequency rise services to keep system frequency stability.
Fig. 6: Dynamic frequency responses: Queensland RETs model.
When analysing the curves, the base case scenario (black line) already dropped the frequency into the “contingency” band after the event, which means FCAS’ raise services were required. However, when the 2020 (red line) and 2030 (green line) scenarios were tested, the curve dropped even lower into the “contingency” band, in each case with a deeper nadir and faster ROCOF. According to NEM regulation this event requires proper protection, thus BESS FFR services would be needed to guarantee proper frequency stability recovery, given the need for an increasingly faster balancing power injection.
Conclusion
Results 2 and 3 above reinforce the notion that further studies on the use of BESS FFR in Queensland are required. This is to ensure adequate system security schemes are implemented to accommodate the planned increases in VRE. It is imperative that the experience gained in South Australia is correctly interpreted and adapted to Queensland to prevent financial and social losses. Queensland is projected to have a similar VRE penetration rate as South Australia as both states experience similar extremely high temperatures, but Queensland also has extreme weather which is expected to increase in severity as climate change progresses. This situation makes the requirement for system security even more pressing.
Acknowledgement
This article was first published in the June/July 2019 edition of APT’s Transmission & Distribution and is republished here with permission.
References
[1] “Annual Market Performance Review 2016”, Australian Energy Market Committee (AEMC).
[2] “State of the Energy Market – May 2017”, Australian Energy Regulator.
[3] “Update to Renewable Energy Integration in South Australia”, Australian Energy Market Operator (AEMO), 2016.
[4] J. Riesz: “The future power system security programme – frequency control”, FPSS roadshow presentation, Australian Energy Market Operator (AEMO), 2016.
[5] “Credible pathways to a 50% renewable energy target for Queensland”, Queensland Renewable Energy Expert Panel, 2016.
[6] “Fast frequency response in the NEM”, Future Power System Security Programme, Australian Energy Market Operator (AEMO).
[7] F Gonzalez-Longatt and S Alhejaj: “Investigation on grid-scale BESS providing inertial response support”, IEEE International Conference on Power System Technology, 2016.
[8] “Battery Energy Storage System (BESS)”, Digsilent Power Factory.
[9] F Gonzalez-Longatt: Large Scale Renewable Power Generation, p. 193-231, 2016.
Send your comments to energize@ee.co.za
