“Electricity Supply to Africa and Developing Economi es …. Challenges and opportunities.”

Technology solutions and innovations for developing economies Advances in Power Quality Requirements for RPPs

MG BOTHA1, BB PETERSON 2, H MOSTERT2, UJ MINNAAR 2 1 North-West University, 2 Eskom SOC

South Africa

SUMMARY The power quality requirements for RPPs created significant debate when the first RPPs started connecting to the grid. The purpose of this paper is to capture some of the debates around the harmonic emission limits and assessment methods as well as the resulting update to the RPP grid code. KEYWORDS Power quality, harmonics, unbalance, flicker, RPP, Grid Code, grid code compliance

1. INTRODUCTION Since 2011, South Africa has seen a steady increase in independent power producers (RPPs) connecting to the national grid. The South African REIPPPP process has been hailed as a major success for the country [1]. Nevertheless, several teething issues did occur, with the power quality aspects of RPPs being one of the bigger learning curves, especially harmonics. The purpose of this paper is to provide background to the harmonic emission requirements, discuss the issues that were encountered and the solutions provided to date. The changes to the requirements are provided with a view of providing some background to the decisions and further clarification of some requirements. The differences of interpretation of the requirements by NSPs and RPPs and the risks to both parties are highlighted. The primary goal is to marry these differences and minimise the real risks, whilst ensuring fairness to all customers connected anywhere in the network. Whilst criticism remain to the requirements, several RPP installations have already applied the principles successfully to prove compliance and achieve full operational status, most without the need to resort to expensive harmonic filters or other mitigating options.

2. BACKGROUND When the first South African Grid Code for Renewable Power Plants (RPPs) was published in November 2012 [2], the power quality (PQ) requirements were based on the processes used for traditional load customers. These were largely derived from the existing Eskom practices at the time, since Eskom had the most experience in dealing with larger customers. The base premise to manage PQ levels is to provide emission limits to customers for the different power quality parameters. The Network Service Provider (NSP) then has the responsibility to manage the resulting PQ, which would be a combination of network strengthening or other network mitigation interventions as well as enforcing contractual limits when PQ problems arise. The emission limits are set based on the IEC/TR 61000-3-6/7/13 [3][4][5] documents (as updated), taking local network conditions into account. In terms of harmonic emissions, a key factor is the expected harmonic impedance of the network as seen by the customer. This harmonic impedance affects the resulting voltage harmonics due to harmonic currents injected by customers’ loads. The default selection of a three-times base harmonic impedance as an envelope created both confusion and conflict. Consultants started engaging with the Eskom PQ departments when the first installations finalised grid code compliance reports. Two aspects became clear: 1) that the PQ requirements were unclear to consultants and 2) difficult to prove and achieve from an RPP point of view. Therefore, a “Harmonics Working Group” was established under the South African Grid Code secretariat to address the concerns and draft a guideline for consultants and RPP owners. This working group evolved into a PQ working group as more clarity on the harmonic emission requirements were found and additional questions posed around the other parameters.

3. REQUIREMENTS The basic PQ requirements provided for in the grid code are summarised as follows:

1. Which parameters are to be regulated, 2. That the assessment shall be done at the POC, 3. That the NSP (utility) shall calculate appropriate emission limits, along with defining

the network conditions under which these requirements shall be met, e.g. PQ levels, fault levels and the infamous three-times impedance envelope,

4. That the RPP generator shall ensure the RPP is appropriately designed and operated.

The majority of these requirements were already provided for in the Eskom contracts, the so-called Connection and Use of System Agreement (CUOSA). Firstly, it was required that RPPs be treated the same as traditional customers as far as possible. These were published as part of the grid code as power quality forms part of NSPs’ licensing conditions and to ensure transparency in that these rules are provided upfront to any prospective RPP.

4. POINTS OF CONFLICT

4.1. RPP DEVELOPER CONCERNS Based on the first reports received, as well as interactions between developers and Eskom, the following concerns were evident:

1. The grid code requirements were unclear, since few consultants had sufficient power quality background;

a. Added to this was the fact that the first set of connection agreements contained only a selection of voltage harmonic emission limits explicitly, with a paragraph explaining how to obtain the other voltage harmonic and current harmonic emission limits;

2. The requirements were different to those of other countries with established renewable energy sectors e.g. European countries;

a. As an example, the IEC 61000 apportioning method is not followed in some countries;

3. The compliance is based on the rms harmonic currents; therefore, it is possible that an installation would absorb harmonic voltage and current (and have a beneficial impact on the network) whilst exceeding the allocated current emission limits (i.e. appear non-compliant);

4. The three-times impedance was deemed unrealistic.

4.2. NSP CONCERNS The NSP is legally responsible for the power quality at each customer point of supply (updated as point of connection for power producers). Therefore, the NSP has to manage the power quality, which can be done via the following steps:

1. Limiting the impact of any customer on power quality (generally via contracts such as the CUOSA);

2. Coordinating the combined impact of customers at a point of common coupling (PCC);

3. Monitoring the power quality at all PCCs; 4. Limiting the network’s impact on power quality provided to customers; 5. Communicating to customers on all relevant aspects.

4.3. CONTEXT OF CONCERNS Whilst Eskom had reasonable experience in dealing with customers and power quality aspects, it was not adequately prepared to deal with the number of new challenges brought by the RPPs, since only a handful of specialists had sufficient experience in contracting. It should also be noted that most municipalities lack the staff complement and network information to do detailed studies. Therefore, the processes have to take into account the lack of knowledge in South Africa and be easy to implement by inexperienced staff. Furthermore, there are major differences between South African and European networks, e.g. base load generation location, transmission line lengths, overhead vs. underground (affecting system capacitance), use of shunt capacitors and interconnections to other countries. These factors affect the typical system strength and harmonic impedance. It was also found that many of the generators were connecting at relatively weak locations in the network – many of the RPPs being connected at a short-circuit ratio as low as 3 times. It is

unrealistic to assume that European-based rules and regulations can be implemented as is in South Africa.

5. UPDATING THE REQUIREMENTS It was therefore clear that the requirements had to be updated. However, some aspects appear contradictory, e.g. reducing the harmonic impedance envelope as well as increasing emission limits.

5.1. A BRIEF HISTORY OF POWER QUALITY APPORTIONING The Merriam-Webster dictionary defines apportioning as “to divide and share out according to a plan” and “to make a proportionate division or distribution of”. Power Quality apportioning is, therefore, the equitable sharing of power quality levels amongst customers and network operators, according to a plan. Power quality apportionment is based on the IEC technical reports, IEC/TR 61000-3-6/7 and 13 [3][4][5]. It means that there is no international consensus on the processes, but it contains the current best practice, philosophies and guidelines that are available. The guiding philosophy is that all customers, as well as the network, contributes to the total PQ levels and appropriate limits should be set for each party’s contributions. Edition 1 of these documents considered that emission limits had (most likely) not been allocated before. Therefore, PQ measurements were done to determine the difference between the background levels and planning levels. The appropriate share per customer is based on the size (i.e. notified maximum demand) relative to the available capacity of the connection point. At the time of compiling the RPP grid code, it was understood that several European countries still used this method. Eskom has adopted the latest editions (2008), which proposes a more relaxed apportioning of emission limits than edition 1. For edition 2, the entire planning level is allocated proportionally between all customers already connected, those wanting to connect now as well as likely future customers. This places a heavier burden on NSPs, since customers connected previously may already use more than the fair share as per this calculation. This is especially true in the case of legacy customers, who were connected before such allocations were done. Despite many criticisms of these methods, at this stage, no better options exist. At the time of writing, work on updated apportioning methods is underway and the apportioning may be updated in future when better options exist.

5.2. IMPEDANCE ENVELOPE IEC 61000-3-6 [3] recommends that the ratio between the voltage and current harmonic emission limits be set based on simulated harmonic impedance at the point of connection. However, the document also allows for simplified approximations, such as the examples provided in Annex A [3] for MV systems. No generic extrapolation is provided for higher voltage levels. A comparison of the recommended 11 kV and 33 kV envelopes are shown in Figure 1 along with the South African proposal of three-times base impedance. The

IEC 61000-3-6 also states that such generalizations are usually not possible at the higher voltage levels.

Figure 1: Proposed harmonic impedance envelopes However, a broad generalization has been made in South Africa to deal with industrial and other large customers since the late 1990’s. Given limited skills, resources and network information at the time, an assumption had been made – that was generally accepted as reasonable – an upper resonance limit of three-times the base impedance was adopted at all points of supply and all voltage levels. In the spirit of treating RPPs the same as traditional load customers (i.e. as customers of the wires business), the three-times base impedance was also adopted in the RPP grid code. Note that this three-times impedance envelope is intended to cover both existing network conditions as well as future changes to the network. Therefore, it manages the risk of existing resonances under network normal and contingency situations, as well as looking forward when the network changes. It means that the NSP also has to manage resonances on the network and the risk to customers due to higher resonance conditions is reduced. The three-times impedance envelope was one of the main factors criticized by RPPs. It is important to note that, while the treatment of RPPs and load customers are the same as far as possible, there is a significant regulatory difference when it comes to PQ requirements: RPPs must prove compliance to all requirements before a license will be approved. Compliance is, therefore, required by using the forward-looking three-times impedance and measured current harmonics, irrespective of existing voltage harmonics at the time of connection. Traditional load customers have to show reasonable assurance of compliance and future problems may be addressed as and when they arise. An RPP is, therefore, required to install adequate filters for problems that may not arise over the 20-year Power Purchase Agreement (PPA) horizon. However, once an RPP has shown compliance and received its license, any future harmonic problems become the responsibility of the NSP, unless there is a change in the RPP’s plant and/or equipment, which may be responsible for the harmonic exceedance.

In order to see what harmonic impedance envelope would be reasonable, a simulation study was undertaken to evaluate the harmonic impedance in the Eskom network, since widescale measurement of harmonic impedance would be impractical. Digsilent simulations were done for all busbars in the Eskom system, using case files updated for harmonic simulations1. The harmonic impedance at each busbar was normalized to the 50 Hz impedance for comparison to the three-times envelope. For a system-wide comparison to the three-times envelope, the normalized impedance values of all busbars were aggregated per frequency. Figure 2 shows the 95th percentile plots for MV and HV nodes respectively vs. linear impedance plots for system normal. MV nodes included 11 kV, 22 kV and 33 kV, whilst 66 kV, 88 kV and 132 kV nodes were included for the HV calculations. It appears that the system impedance at most sites are reasonable at the lower harmonic frequencies for normal operating conditions, however, significant resonances appear at the higher order frequencies for both MV and HV networks.

Figure 2: 95th percentile plots of MV and HV system impedance for System Normal. These simulations ignored the potential impact of contingencies, which can be quite significant. Two scenarios are possible, namely that the existing resonance frequency is shifted or that additional resonance frequencies arise. Based on a series of simulations at specific nodes, it was estimated that a typical band of 300-400 Hz exist around resonant frequencies. By using a 600 Hz (300 Hz in either direction) bandwidth around the 95th percentile curves, envelopes for the HV and MV impedance are derived as shown in Figure 3. From this, it appears that the three-times envelope may be a reasonable assumption as an interim solution. Both RPPs and the NSP bear additional risks due to this assumption. Several options were discussed for more representative envelopes or even site-specific envelopes. However, consensus could not be reached between the NSP representatives, consultants and the GCAC. Further research is in progress by Eskom to identify more

1 Base case files were used for these simulations. Whilst the case files were updated for harmonic simulations, no other accuracy checks or operational variations were done, e.g. normally open points, shunt capacitor values, generation patterns, line type data, loading etc. The purpose was to obtain an indication of the harmonic impedance envelope, i.e. statistical analysis of simulated results.

appropriate impedance envelopes and the RPP grid code will be amended once such proposals are accepted by all stakeholders.

Figure 3: 95th percentile plots of MV and HV system impedance with 600Hz contingency bandwidth.

5.3. NETWORK INFORMATION PROVIDED A parallel debate by a different working group was had around network information provided to RPP developers. The network information provided by the NSP is provided in Appendix 12 of the RPP grid code version 2.9. It is noted that harmonic system impedance is required for two aspects, namely ensuring that the RPP installation does not cause excessive resonance and for harmonic filter design should filters be required. It is not appropriate to use the three-times harmonic impedance envelope for such simulations. It is possible that any reduction of the network results in significant inaccuracies when simulating the harmonic impedance. An example of the differences between the harmonic impedance based on the full model and some reduced models are shown in Figure 4. Given the differences seen, it was decided that the NSP will be responsible to provide harmonic impedance simulations for both network normal and a range of system contingencies. No simplified model provided as per Appendix 12 can be used for harmonic impedance simulations.

Figure 4: Frequency sweep of base network impedance showing impact of reducing network.

5.4. ASSESSMENT CLARIFICATIONS

5.4.1. Basic Power Quality Assessment The NSP manages power quality in terms of measured voltage parameters. The assessment parameter also takes the statistical nature of PQ into account, i.e. situations may arise, usually outside the control of the NSP, where the performance may be significantly worse than expected, albeit for limited time periods. PQ is, therefore, typically assessed as a 95th percentile aggregated over a sliding weekly scale [3][6]. An RPP, therefore, need to measure for at least one week as per NRS 048-2 [6] to prove compliance. A realistic assessment will also be representative of typical plant performance [3], e.g. power production profiles, equipment switching etc.

5.4.2. Harmonic Voltage vs. Harmonic Current According to the IEC, power quality is defined as “characteristics of the electric current, voltage and frequencies at a given point in an electric power system, evaluated against a set of reference technical parameters” [7]. These reference technical parameters in IEC 61000-series document are provided in terms of the voltage. However, current can be seen as the driving force of deviations from the ideal reference, due to the voltage drop caused across the system impedance. Given that the network may change over the timespan of the PPA, the three-times impedance envelope provides the means to assess the RPP’s performance over this timespan. Depending on the harmonic impedance at the time of connection and assessment, the assessed voltage harmonic emission may be more or less than the corresponding limit – whereas the current emission limit will indicate the likely future performance. The current harmonic emission limits are, therefore, the critical assessment criteria.

5.4.3. CT and VT Accuracy IEC/TR 61000-3-6 recommends a minimum emission limit of 0.1 % for the voltage harmonics, but no recommendation for current harmonic emission limits. However, the emission limits for both voltage and current harmonics calculated for many of the RPPs were less than 0.1 %. Although the NSP could qualify the potential risk to the network, standard transducers are less accurate and the requirement is regarded as unreasonable. The typical CT used has an accuracy specification of 0.2 % and this is specified only at the fundamental frequency. VTs are also known to exhibit potential internal resonance at higher frequencies, typically above the 25th harmonic. The better solution would be to install specialised transducers with very high accuracies for PQ assessment. However, such costs would be prohibitive and it was agreed that measurements will be done using standard transducers. The minimum emission limits were updated to 0.1 % for voltage emissions and 0.1 A for current emissions respectively. It is expected that future revisions of the RPP grid code will update the harmonic current emission minimum level to 0.1 % as well.

5.4.4. Assessment Methods The guideline is not intended to prescribe the methods that may be used to assess the PQ performance of the installation, but some recommendations were provided. Should the assessed harmonic current level exceed the emission limit, the following recommended analysis methods were recommended:

• The impedance slope method, as described in Cigré Technical Report 468, section 4.2.2: Harmonic Emissions Level Compliance Assessment [8];

• Negative correlation between harmonic current and voltage (i.e. an increase in harmonic current coinciding with a decrease in harmonic voltage and vice versa);

• Zero harmonic voltage measured throughout the measurement period, allowing for up to 300% of the current emission limit;

• The impedance scaling method, where the measured harmonic voltage is scaled according to the ratio of the three-times impedance envelope vs. the simulated system impedance;

• Group harmonic distortion levels less than group harmonic distortion limits, where individual harmonic emissions may exceed the emission limit by up to 50%.

The list is not exhaustive and additional methods may be employed to prove compliance to the emission limits. Further details can be found in the PQ guideline, Appendix 13 of the RPP grid code [2].

5.5. GUIDELINE The guideline, included as Appendix 13 of the RPP grid code version 2.9, contains details of both the simulation and measurement requirements to prove grid code compliance. The guideline includes the steps to be taken, including flowcharts of the typical process. The concepts around harmonic emissions and the three-times impedance, as discussed in this paper, were included.

6. CONCLUSIONS An update to the power quality requirements (clause 9 of the RPP Grid Code) as well as a guideline for power quality compliance was published in version 2.9 towards the end of 2016. Key improvements are the clarification of the requirements, high-level guidance for simulating and measuring compliance as well as relaxation of emission limits under certain conditions. Amongst others, emission limits for rapid voltage changes and inter-harmonics have been removed and a group harmonic emission limit was introduced with allowance up to 150% of individual emission limits when the group harmonic emission limit is met. The preferred analysis methods from the CIGRE C4.109 brochure are discussed, without restricting the analysis methods that may be used. The process was a significant learning curve for both the NSPs and RPP consultants. The guideline is now available to everybody and it is hoped that RPPs will in future be better able to plan appropriately for PQ aspects. It is acknowledged that the PQ requirements and guideline are not yet perfect. Further research and analysis is being done and updates will be published as better solutions are found.

7. REFERENCES

[1] Creamer Media Reporter, “Tallying the benefits of South Africa’s Renewable Energy Power Producer’s Procurement Programme, Engineering News”, 9TH MAY 2016, online, accessed 2017/10/01, http://www.engineeringnews.co.za/article/tallying-the-benefits-of-south-africas-renewable-energy-power-producers-procurement-programme-2016-05-09/rep_id:4136

[2] NERSA, “Grid Code Requirements for Wind Energy Facilities Connected to Distribution or Transmission Systems in South Africa”, version 2.6 - 2.9, July 2012 - July 2016.

[3] IEC 61000-3-6, Electromagnetic compatibility (EMC) - Part 3-6: Limits - Assessment of emission limits for the connection of distorting installations to MV, HV and EHV power systems, Edition 2.0, 2008.

[4] IEC 61000-3-7, Electromagnetic compatibility (EMC) - Part 3-7: Limits - Assessment of emission limits for the connection of fluctuating installations to MV, HV and EHV power systems, Edition 2.0, 2008.

[5] IEC 61000-3-13, Electromagnetic compatibility (EMC) - Part 3-13: Limits - Assessment of emission limits for the connection of unbalanced installations to MV, HV and EHV power systems, Edition 2.0, 2008.

[6] NRS 048-2, “Electricity Supply – Quality of Supply: Part 2: Voltage characteristics, compatibility levels, limits and assessment methods”, Edition 4, 2015.

[7] IEV Online, “Electropedia: The World's Online Electrotechnical Vocabulary”, IEV ref 617-01-05, http://www.electropedia.org/iev/iev.nsf/display?openform&ievref=617-01-05.

[8] Arlt, D. et al, Working Group C4.109, “Cigré Technical Report 468: Review of Disturbance Emission Assessment Techniques”, June 2011, ISBN: 978-2-85873- 158-9.