Busbar Classification Review

Busbar Classification Review Updated Credible Methodology and Trial Results Transpower New Zealand Limited February 2017

Version Date Change 1 01/02/2017 Pubilshed for feedback IMPORTANT Disclaimer The information in this document is provided in good-faith and represents the opinion of Transpower New Zealand Limited, as the System Operator, at the date of publication. Transpower New Zealand Limited does not make any representations, warranties or undertakings either express or implied, about the accuracy or the completeness of the information provided. The act of making the information available does not constitute any representation, warranty or undertaking, either express or implied. This document does not, and is not intended to; create any legal obligation or duty on Transpower New Zealand Limited. To the extent permitted by law, no liability (whether in negligence or other tort, by contract, under statute or in equity) is accepted by Transpower New Zealand Limited by reason of, or in connection with, any statement made in this document or by any actual or purported reliance on it by any party. Transpower New Zealand Limited reserves all rights, in its absolute discretion, to alter any of the information provided in this document. Copyright The concepts and information contained in this document are the property of Transpower New Zealand Limited. Reproduction of this document in whole or in part without the written permission of Transpower New Zealand. Contact Details Address: Transpower New Zealand Ltd 96 The Terrace PO Box 1021 Wellington New Zealand Telephone: +64 4 495 7000 Fax: +64 4 498 2671 Email: Website: system.operator@transpower.co.nz http://www.transpower.co.nz

Contents Bus Classification Project... 1 Updated Credible Methodology and Trial Results... 1 1 Executive Summary... 4 2 Background... 6 2.1 Need for Review... 6 3 Methodology... 7 3.1 Overview... 7 3.2 Changes to the methodology... 8 3.3 Methodology... 9 3.4 Identification of s... 9 3.5 Assess s... 9 4 Trial Results of Analysis and s... 15 4.1 Otahuhu 110 kv Busbar... 15 4.2 Huntly 220 kv Bus... 20 4.3 Bunnythorpe 220 kv Busbar... 23 4.4 Islington 220 kv Busbar... 28 4.5 Clyde 220 kv Busbar... 31 4.6 Manapouri 220 kv Busbar... 34 5 Busbar reconfiguration and SPS use... 37 6 Classification Summary... 38 6.1 Summary of Annual s... 38 6.2 Recommended Classification... 39 7 Appendix 1 Generation and Load Duration Curves... 40 7.1 Clyde GDC... 40 7.2 Roxburgh GDC... 40 7.3 Waipori GDC... 41 7.4 Mangahao GDC... 41 7.5 Patea GDC... 42 7.6 Zone 1 (Grid Zone 1 and Grid Zone 2) LDC... 42 7.7 Grid Zone 14 LDC... 43 7.8 Bunnythorpe GXPs LDC... 43 8 Appendix 2 Policy Statement Electricity Industry Participation Code... 44 8.1 Definition of CE, ECE and Other... 44 8.2 Controls available for CE, ECE and Other... 44 8.3 Quality Levels for CE, ECE and Other... 44 3

1 EXECUTIVE SUMMARY Transpower as system operator is responsible for identifying events which pose credible risks to the New Zealand power system and for managing those risks to deliver the principle performance obligations (PPOs). Transpower carries this responsibility out through a Credible Review (CER) process every five years. The last CER was carried out in 2014. 4 Transpower decided to review the methodology used in the CER following a double circuit outage at Manapouri in 2015 for which pre-event management was undertaken by Transpower. Transpower wished to consider what event management improvements could be made taking mitigation and market costs into account (when currently they are not). Specifically, Transpower reviewed the assumptions and methodologies used in the 2009 CER when it proposed that core grid buses be classified as Extended (ECE) risks. From the review Transpower identified improvements to the assumptions and methodology previously used and trialed the new approach on the following buses; Huntly 220 kv, Manapouri 220 kv, Bunnythorpe 220 kv, Otahuhu 110 kv, Islington 220 kv and Clyde 220 kv. Historical busbar fault statistics over the period 2011 to 2015 were analyzed to obtain bus fault probabilities and durations which were used in the economic assessment of different classifications. Load duration curves for 2015 were analyzed to find the probability that load would be high enough for issues to occur following a bus section contingency. These curves were also used to calculate the average load likely to be affected. This document sets out the new assumptions and methodologies used in the Credible Methodology for industry consideration and feedback. The focus of this review relates specifically to voltage management and not to frequency management. Key Changes The methodology was updated to include assessment of outage conditions, only allow for existing mitigation measures to be considered, and account for wider cost of classification of events. This meant that impact of planned bus section outages were analysed, an SPS would only be used if it was already in place and generation costs to mitigate issues would be costed. A key shift in the approach was to move away from a single blanket classification for all buses and towards classification at individual bus level as the consequences and mitigations can vary at each bus. A deterministic approach for the power system analysis combined with a probabilistic approach for the economic analysis of feasible mitigation measures has been used for this trial classification. Mitigation measures that are operationally feasible are identified and costed based on managing these busbars as (CE), ECE or Other. Generation Duration Curves for 2015 were also analysed to find the expected generation likely to be available and to calculate the additional generation response that would be required to mitigate issues. Results of the trial Normal Conditions No significant issues from the loss of a single bus section when all other bus sections are in service (N-1 considerations) were identified for Huntly 220 kv, Manapouri 220 kv, Bunnythorpe 220 kv, Islington 220 kv and Clyde 220 kv. There are issues for Otahuhu 110 kv under N-1 condition which may be managed through the use of a special protection scheme.

Outage Conditions Loss of a Huntly 220 kv bus section during a planned bus section outage in winter is likely to result in a loss of generation and circuits at Huntly except for Huntly Unit 5 and the Drury-Huntly 220 kv circuit. The limiting issue in this case is the voltage stability limit into the Upper North Island (Grid Zone 1 and 2). Management as a CE would require demand response reduction in UNI load. If classified as Other, loss of UNI load is possible should the event occur. Loss of a Manapouri 220 kv bus section during a planned bus section outage will result in loss of the whole station. The limiting issue is the voltage stability limit in Southland (Grid Zone 14). Management as a CE would require demand response reduction in Southland. If classified as Other, loss of Southland load is possible should the event occur. Loss of a Bunnythorpe 220 kv bus section during a planned bus section outage is likely to result in the remaining 220/110 kv interconnector, connected to the third busbar, overloading. Management as a CE would require demand response reduction at some Grid Exit Points in the Bunnythorpe Region. If classified as Other, loss of these GXPs load is possible should the event occur. 5 Loss of an Otahuhu 110 kv bus section during normal and planned bus section outage is likely to result in the Bombay-Hamilton 1 and 2 110 kv circuits overloading and low voltages at Wiri. Management as a CE would require demand response reduction at Bombay and Wiri. This could be managed as an ECE via an existing SPS, Bombay 110kV system split inter trip scheme, which would open Bombay-Otahuhu 1 and 2 110 kv circuits and remove the overload with loss of Wiri load. If classified as Other, loss of Bombay and Wiri loads are possible should the event occur. Loss of an Islington 220 kv bus section during a planned bus section outage is likely to result in the in the remaining 220/66 kv interconnector overloading. Management as a CE would require demand response reduction in Zone 3 (Upper South Island). If classified as Other, loss of Zone 3 load is possible should the event occur. If prior to the planned outage, all three 220/66 kv interconnectors were switched to be on different bus sections then the impact of a following bus section contingency would be minimized. Loss of a Clyde 220 kv bus section during a planned bus section outage is likely to result in the in Naseby-Roxburgh 220 kv circuit overloading. Management as a CE would require demand response reduction in Grid Zone 14 (Southland importing) or generation response reduction (Southland exporting). If classified as Other, loss of Grid Zone 14 load is possible should the event occur. If prior to the planned outage, the remaining bus sections each carry one circuit north and one circuit south then the impact of a following bus section contingency would be minimized. Recommended classifications Based on the costs of mitigation measures available and the probability of these events happening, the classification for Huntly, Manapouri and Bunnythorpe 220 kv bus should be Other. The classification for Otahuhu 110 kv bus should be ECE if the existing Bombay 110kV system split inter trip scheme is offered by grid owner. If the SPS is not offered then the classification should be Other. The classification for Islington 220 kv and Clyde 220 kv bus should be Other. However consideration should be given to busbar reconfiguration during planned outage conditions to minimize impact on the system following a bus section contingency.

2 BACKGROUND 2.1 NEED FOR REVIEW The Policy Statement in the Electricity Industry Participation Code (the Code) states Transpower, as the system operator (SO), should identify potential credible events 1 on the power system as a result of asset failure that may result in cascade failure. One such credible event is the loss of a busbar section. 6 The 2009 Credible Review found the loss of a busbar section under normal conditions, with the subsequent overload and tripping of additional assets, would result in unplanned load shedding in a significant number of cases. Introduction of planned load shedding, where reasonable and possible, would allow post event parallel equipment overloads to be managed and thus avoid unplanned load shedding associated with subsequent asset trippings. A cost benefit analysis comparing load shedding options under different event classifications indicated it was beneficial to classify and manage the loss of these busbar sections as an ECE; it would oblige Transpower to plan for these events and employ planned post event load shedding, as appropriate, to manage the risk of unplanned load shedding (shedding would have much higher consequences and costs). The 2009 CER recommended the loss of a 220kV or 110kV or 66kV busbar directly connected to the core grid should, under normal conditions, be classified as an ECE 2. This was incorporated into the Policy Statement in 2010. The relevant clauses of the Policy Statement are in Appendix 2. However, operationalising the change in policy has proven to be very difficult. In May 2015 a planned double 220 kv circuits outage at Manapouri raised concerns that a busbar fault during the planned outage could result in loss of the whole station and cause system wide voltage and frequency stability issues. It was decided pre contingency measures should be established during the planned outage as there were no post contingency measures in place to manage a busbar fault under outage conditions. While this reflected an attempt to apply the new policy its application created significant industry discussion as well as operational complexity. An interim busbar policy was put in place in October 2015 stating Transpower would not manage precontingently the loss of a core grid busbar during normal and outage conditions, with one exception. Only the Manapouri bus would be managed during both normal and outage conditions for ECE risks relating to frequency only. The interim busbar policy would stay in place until an assessment of how core grid busbars could be managed under different classifications was carried out, taking into account both economic rationale and operational requirements. This assessment would form the basis of a permanent busbar policy. 1 Clause 12 Policy Statement, Electricity Industry Participation Code 15 May 2014 2 2009 Security Policy Review: Credible Management Summary of Findings

3 METHODOLOGY 3.1 OVERVIEW The analysis process is illustrated in Figure 3-1 below. Differences between the 2014 Credible Review and 2016 Bus Classification Methodology are listed in Table 3-1 below. Figure 1 Analysis Process Communicate 7 Establish the context: Objectives Criteria Identify events: What can happen How often it happens Analyse : Consequences and impact; Mitigation options Assess Evaluate : High level cost of event; High level cost of mitigation options Conclusion: Select response Develop policy and plans Apply/Implement treatment Monitor and Review

3.2 CHANGES TO THE METHODOLOGY Table 1 Differences between 2014 Credible Review and 2016 Bus Classification Methodology 2014 Credible Review 2016 Bus Classification Methodology Rationale 8 Classification Planned outage conditions Blanket - one classification is applied to all core grid buses. Individual each core grid bus is assessed and classified, classification may be different. The consequences of bus faults vary by bus and therefore it is consider prudent and efficient to vary the classification by bus. Not considered. Considered and assessed. The system operator must meet the PPOs during all grid conditions. Demand 2019 peak load. 2015 peak load. Powerflow analysis are based on existing grid configuration; peak loads experienced with this configuration will identify issues without requiring new generation to meet higher peaks. Demand response is the only mitigation measure considered, priced at $10,000/MWh. Demand response and generation response are the mitigation measures considered. Demand response priced at $700/MWh. Generation response priced at short run marginal costs. The cost of generation is considered to be material following the Manapouri double circuit outage in 2015. Market costs Not considered. Considered. Generation response is included and costed. Sensitivities around offer prices for demand and generation offer prices are calculated. See response to. Extended Assume new Special Protection Scheme (SPS) will be built for every bus requiring it as a mitigation measure. Demand shed priced at $10,000/MWh. If no SPS exists at a bus then it cannot be used as a mitigation measure. Demand shed is priced at $25,000/MWh. The classification can only rely on available mitigations as the time to install alternatives (SPS) can be significant. Busbar Reconfiguration Not considered. Considered where existing circuit breakers or disconnectors can be used to connect to different bus section. This reflects Transpower s normal outage planning procedure.

3.3 METHODOLOGY Each bus is analysed separately to arrive at a classification; it was also acknowledged that it was possible that a bus could be classified differently under normal conditions compared to planned outage conditions. The methodology considered three possible classifications for each bus: CE, ECE and Other. Further detail is provided in Appendix 2. 9 Mitigation measures for a CE must be arranged pre-contingency and include constraining on/off generation and use of demand response. These measures would be set up to take effect if load is high enough for issues to occur. Mitigation measures for an ECE take effect post-contingently. They consist of shedding load postevent via a SPS or an automatic load shedding scheme. If an SPS does not exist at a bus it is assumed an ECE classification is not practical. Post-contingent generation response is not considered. For events in the Other classification, equipment overload is noted with subsequent tripping of overloaded equipment and load lost. No measures are taken either to prevent the consequences precontingently or reduce the consequences post-contingently, other than what is normally available (such as Automatic Under Frequency Load Shedding Scheme). Instead, the system operator relies on its restoration plans to bring the grid back with least delay. 3.4 IDENTIFICATION OF EVENTS This review has considered loss of a bus section during both normal and planned outage conditions. During a normal condition all equipment is in service. A planned outage condition considers only the planned outage of a bus section at the bus being assessed. For each bus with all equipment in service the impact of a single bus section contingency is analysed. A bus section is then taken out of service, its connections moved to another bus section if possible; the impact of a single bus section contingency at the same bus is then analysed. Where connections can be spread over the remaining bus sections (busbar reconfiguration) so the impact of a bus section contingency is minimized then this configuration is considered. 3.5 ASSESS EVENTS System fault interruption reports from 2011 to 2015 were analysed to calculate the probability of a bus section fault and the average duration of a bus section fault by voltage.

3.5.1.1 Busbar Tripping Statistics Table 2 Busbar Failures Summary (2011-2015) Busbar Failure Summary 10 Voltage Year 220 kv 110 kv 66 & 50 kv TOTAL (by Year) 2011 4 2 0 6 2012 2 0 2 4 2013 6 1 1 8 2014 5 3 1 9 2015 3 3 0 6 Average 4 1.8 0.8 6.6 Table 3 Busbar Statistics (2011-2015) Busbar Statistics (2011-2015) Voltage 220 kv 110 kv 66 & 50 kv TOTAL Total (by kv) 20 9 4 33 Average No of Failures per year 4 1.8 0.8 6.6 No of elements in set 150 138 20 303 0.027 0.013 0.04 0.022 Average Duration (hours) 2.395 3.176 1.108 2.452 The event risk factor is calculated by the average number of failures per year associated with each element divided by the number of elements in a set. It is assumed each bus section will have 1 planned outage per year which last for 8 hours, defined as the equipment request hours. For an N-1-1 condition, (i.e. during a planned outage) the duration for CE classification is therefore the equipment request hours, as pre-contingency measures must be actioned regardless of whether the contingency occurs during the planned outage. The duration for ECE and Other classifications is the average duration of a bus section fault multiplied by the additional probability a bus section fault will occur in the 8 hours period over the whole year. Power flow analysis of peak load conditions is used to identify overloading or voltage issues, under both normal and planned outage conditions, for a single bus section contingency. Where issues are identified mitigation measures are assessed to see if issues can be resolved satisfactorily. The amount of demand and generation response required is quantified. The impact of no mitigations being in place is also assessed.

3.5.3.1 Summary s for mitigation measures are determined for each bus. Load duration curves (LDC) at the relevant GXPs are used to calculate the probability load would be high enough for issues to occur following a bus section fault. These LDCs are also used to calculate the average amount of load affected and the demand response required. Demand response required to mitigate issues is priced at the cheapest demand response price over the last 12 months. However, a sensitivity using the highest demand response price over the last 12 months is also calculated. Generation duration curves are used to calculate the amount of generation that would be available at 90 th percentile and the additional generation response required. 11 Where generation response is required the short run marginal cost is used to price this response. A sensitivity using the highest offer price over the last 12 months for generation is also calculated. Where load is lost or shed involuntarily, a VOLL of $25,000/MWh is used to price the cost of these loads. 3.5.3.2 Annual Equations Normal conditions (N-1) CE classification with pre-event mitigation measures available: Annual (CE) = Generation Response (MW) Generation Unit ($ per MWh) Duration (hours) Load + Demand Response (MW) Demand Unit ($ per MWh) Duration (hours) Load ECE classification with post-event mitigation measures available: Annual (ECE) = Load Shed (MW) VoLL ($ per MWh) Fault Duration (hours) Load This assumes that generation response is not available post contingency. Other classification with no mitigation measures possible or no action is taken: Annual (Other) = Load Lost (MW) VoLL ($ per MWh) Fault Duration (hours) Load Outage conditions (N-1-1) CE classification with pre-event mitigation measures available: Annual (CE) = Generation Response (MW) Generation Unit ($ per MWh) Equipment Request (hours) Load + Demand Response (MW) Demand Unit ($ per MWh) Equipment Request (hours) Load ECE classification with post-event mitigation measures available:

Annual (ECE) = Load Shed (MW) VoLL ($ per MWh) Fault Duration (hours) Load Equipment Request This assumes that generation response is not available post contingency. Other classification with no mitigation measures possible or no action is taken: Annual (Other) = Load Lost (MW) VoLL ($ per MWh) Fault Duration (hours) Load Equipment Request 3.5.3.3 Annual cost variables 12 Generation Response 3 Generation Response is the difference between what our analysis indicated is needed at specific generation sites to mitigate issues and the likely generation at these sites, based on historical data. It is assumed that generation response has to be contracted for pre-event mitigation but is not available for post-event mitigation i.e. generation response can be used for CE but not for ECE. Generation response is based on the short run marginal cost for different fuel types. Constrained on hydro is priced at $20/MWh and constrained off hydro at $50/MWh. A sensitivity using the highest offer price within the last 12 months at the generation site is provided to show the impact on annual cost. It was suggested that a single high offer price of $5,000/MWh be used to price generation response at all sites but this would have unreasonably penalised the CE option and given erroneous market signals. This was not pursued further. See the Generation Duration Curve section for how the quantity of generation response is calculated. Demand Response Demand Response is the load reduction our analysis has indicated is required to mitigate issues. This demand response would be required pre-event for CE. For CE this is based on the lowest demand response price ($700/MWh) over the last 12 months. A sensitivity using the highest demand response price ($1600/MWh) is provided to show the impact on annual cost. See the Load Duration Curve section for how the quantity of demand response is calculated. Load Shed Load Shed is the controlled load reduction our analysis has indicated is required to mitigate issues. This reduction in load would be required post-event for ECE. Load Shed value used is $25,000 per MWh. This is based on the cost of involuntary load shedding and is obtained from the Electricity Authority s Value of Lost Lost (VoLL). See the Load Duration Curve section for how the quantity of load shed is calculated. Load Lost Load Lost is the total load that will be disconnected (uncontrolled lost) should this event occur and no mitigation is undertaken. It is assumed that once the load is lost the system will return to a stable 3 The generation and demand response required and the amount of load shed or lost is likely to be different for N-1 and N-1-1 conditions.

condition with no overloading of equipment. Load Lost_1 refers to N-1 condition and Load Lost_2 refers to N-1-1 condition. Load Lost value used is $25,000 per MWh. This is based on the cost of unplanned load shedding and is obtained from the Electricity Authority s VoLL. Duration When considering CE classification during normal conditions (N-1) condition the duration is the maximum number of hours the load constraint may be applied, which is 8760 hours in a year. When considering CE classification during outage conditions (N-1-1) the duration is the duration of the planned equipment request (i.e. 8 hours). Fault Duration 13 When considering ECE classification during normal conditions (N-1) and outage conditions (N-1-1) the fault duration is calculated from historic statistics. When considering Other classification during normal conditions (N-1) and outage conditions (N-1-1) the fault duration is calculated from historic statistics. Equipment Request Hours The Equipment Request Hours is the duration of the planned equipment release i.e. the duration of a planned bus section outage. It is assumed there will be 1 planned bus section outage per year with an average 8 hour duration. Equipment Request The Equipment Request Hours is the proportion of the year that the planned equipment is on outage. This factor is used for the ECE and Other classifications under outage conditions (N-1-1). The equipment request factor is calculated by dividing the equiprment request hours by the number of hours in the year (8/8760). This is to account for the risk only being present for a short period of time i.e. during the equipment request spread over the whole year. is the probability of a busbar failure if managed as CE the is 1 as the constraint would be applied all the time i.e. this is independent of the fault probability statistics if managed as ECE or Other, the is calculated from fault probability statistics for both N-1 and N-1-1 conditions. Load The Load is the percentage of time where load is likely to be above the point at which issues will occur, this is known as the Load Limit. See the Load Duration Curve section for how this factor is calculated.

3.5.3.4 Load Duration Curve The Load Duration Curve is used to calculate several factors such as Load, average load at risk, the demand response or load shed required. An example of a Load Duration Curve is shown below. Point A is the highest load on the LDC, Points B and c are calculated from power system analysis for N-1 and N-1-1 conditions. Figure 2 example of a Load Duration Curve 14 N-1 condition Load Limit_1 (Point B) is the load above which issues are expected under N-1 condition Load _1 is the percentage of time where load is likely to be above the Load Limit_1 Average load for N-1 condition is the average load between points A and B on the LDC. The demand response or load shed required is the load difference between the average load and point B. N-1-1 condition Load Limit_2 (Point C) is the load above which issues are expected under N-1-1 condition Load _2 is the percentage of time where load is likely to be above the Load Limit_2 Average load for N-1-1 condition is the average load between points A and C on the LDC. The demand response or load shed required is the load difference between the average load and point C. 3.5.3.5 Generation Duration Curve The generation duration curve is used to calculate the MW available from a generator. The MW at 90% of the time is assumed to be the likely generation from that generator.

4 TRIAL RESULTS OF ANALYSIS AND COSTS 4.1 OTAHUHU 110 KV BUSBAR Region: Auckland Probability of bus faults in a year: 0.013 Normal configuration: Four busbars Average fault duration: 3.176 hours OTA 110 kv bus has four bus sections with 4 interconnectors and 7 circuits connected. Normal configuration is as follows; busbar A1 has OTA T5, MNG-OTA-1 and 2; busbar A2 has OTA T3, OTA- ROS-1 and 2; busbar B1 has OTA T4, BOB-OTA-1 and 2 and busbar B2 has OTA T2 and OTA-PEN- 2. OTA-PEN-2 is normally open. Busbar A1 and A2 are normally split from busbar B1 and B2. 15 The worst bus section contingency is busbar B1, this would open the OTA ends of the BOB-OTA 1 and 2 110 kv circuits. The loss of OTA-WRT segments may cause low voltages at Wiri if its load is above 45 MW. This can be mitigated by switching in Bombay capacitors. Loss of OTA-WRT segments may cause BOB-HAM-1 and 2 to overload if the combined BOB and WIR load is above 140 MW. To mitigate overloading of BOB-HAM-1 and 2 circuits, enable BOB 110kV system split inter trip scheme. This scheme monitors loading on BOB-HAM-1 and 2 and ARI-BOB 110 kv circuits; on detection of overload it will open the BOB-OTA circuits at BOB. WIR load will be lost. If the BOB 110kV system split inter trip scheme is not offered it is assumed BOB and WIR loads will be lost. 4.1.1.1 Winter Peak Scenario Assumptions: Winter Peak, BOB+WIR load = 140 MW Post-event: Loss of OTA busbar B1 and connected equipment: Loss of Otahuhu-Wiri Tee segment of BOB-OTA 110 kv circuit 1 Loss of Otahuhu-Wiri Tee segment of BOB-OTA 110 kv circuit 2 Loss of Otahuhu 220/110 kv T5 Consequence: Constraint: Low voltages at Wiri and overloading of Bombay-Hamilton 1 and 2 110 kv circuits. Analysis of the 2015 combined BOB and WIR loads indicated that they would exceed 140 MW for 3% of the time. The average load above this limit is 145 MW. The average demand response required is 5 MW with a Load of 0.03. The average WIR load for the above 3 % of the time is 76 MW. Min generation: This is not applicable.

CE Approach: Constrain generation off and load off pre-event Pre-event measures: 5 MW demand response required for 3% of time. Duration is 8760 hours as this would have to be managed all year. = 1 as this is managed pre-event. Post-event: Otahuhu and Wiri loads are secured. Annual : Scenario Location Demand Response (MW) Unit ($/MWh) Duration (h) Load Annual 16 Base OTA & WIR 5 $700 8,760 1 0.03 $918,800 Sensitivity OTA & WIR 5 $1,600 4 8,760 1 0.03 $2,102,400 ECE Approach: Post-event load shedding Available mitigations: BOB 110kV system split inter trip scheme enabled. Post-event: Loss of 76 MW load at Wiri for 3 % of the time. Annual : Scenario Location Demand Response (MW) Unit ($/MWh) Fault Duration (h) Load Annual Base WIR 76 $25,000 3.176 $6m 0.013 0.03 $2,353 Sensitivity WIR 76 $25,000 3.176 $6m 0.05 5 0.03 $9,051 Other Approach: Rely on mitigations already in place (reserve, AUFLS, and restoration plans) Post-event: OTA and WIR loads are likely to be lost. Average load for 3% of the time is 145 MW. Annual cost: Scenario Location Load Lost at GZ (MW) Unit ($/MWh) Fault Duration (h) Load Annual Base OTA & WIR 145 $25,000 3.176 $11.5m 0.013 0.03 $4,490 Sensitivity OTA & WIR 145 $25,000 3.176 $11.5m 0.05 5 0.03 $17,269 4 Sensitivity using highest demand response cost 5 Sensitivity using an event risk factor higher than from fault statistics

Planned Outage: OTA 110kV busbar B1&B2 Probability of bus faults in a year: 0.013 : Loss of OTA 110kV busbar A1 Average fault duration: 3.176 hours During a planned outage of busbar B1 and B2, T2 and T4 are taken out. Busbar B1 circuits will be moved to busbar A1. Busbar A1 will now have OTA T5, MNG-OTA-1 and 2, BOB-OTA-1 and 2. This is the worst bus section contingency. The loss of OTA-WRT segments may cause low voltages at Wiri if its load is above 45 MW. This can be mitigated by switching in Bombay capacitors. The loss of OTA-WRT segments may cause BOB- HAM-1 and 2 to overload if the combined BOB and WIR load is above 140 MW. To mitigate overloading of BOB-HAM-1 and 2 circuits enable BOB 110kV system split inter trip scheme. This scheme monitors loading on BOB-HAM-1 and 2 and ARI-BOB 110 kv circuits, on detection of overload it will open the BOB-OTA circuits at BOB. WIR load will be lost. 17 If the BOB 110kV system split inter trip scheme is not offered it is assumed BOB and WIR loads will be lost. Loss of OTA T5, MNG-OTA-1 and MNG-OTA-2 should have no impact as the system is still connected via OTA-T3, PEN-T6, PEN-T10, HEN-T1, HEN-T5 and HOB-T12. 4.1.2.1 Winter Peak Scenario Assumptions: Winter Peak, BOB+WIR load = 140 MW Post event: Loss of OTA busbar A1 and connected equipment: Loss of Mangere-Otahuhu 110 kv circuit 1 Loss of Mangere-Otahuhu 110 kv circuit 2 Loss of Otahuhu-Wiri Tee segment of BOB-OTA 110 kv circuit 1 Loss of Otahuhu-Wiri Tee segment of BOB-OTA 110 kv circuit 2 Loss of Otahuhu 220/110 kv T5 Consequence: Constraint: Low voltages at Wiri and overloading of Bombay-Hamilton 1 and 2 110 kv circuits. Analysis of the 2015 combined BOB and WIR loads indicated they would exceed 140 MW for 3% of the time. The average load above this limit is 145 MW. The average demand response required is 5 MW with a Load of 0.03. The average WIR load for the above 3 % of the time is 76 MW. Min generation: This is not applicable.

CE Approach: Constrain generation off and load off pre-event Pre-event measures: 5 MW demand response required for 3% of time. = 1 as this is managed pre-event. Post-event: Otahuhu and Wiri loads are secured. Annual : Scenario Location Generation or Demand Response (MW) Unit ($/MWh) Equipment Request (h) Load Annual 18 Base OTA & WIR 5 $700 8 1 0.03 $840 Sensitivity OTA & WIR 5 $1,600 4 8 1 0.03 $1,920 ECE Approach: Post-event load shedding Available mitigations: BOB 110kV system split inter trip scheme enabled. Post-event: Loss of 76 MW load at Wiri for 3 % of the time. Annual : Scenario Location Demand Response (MW) Unit ($/MWh) Fault Duration (h) Load Equipment Request Base WIR 76 $25,000 3.176 $6m 0.013 0.03 0.0009132 $2.15 Sensitivity WIR 76 $25,000 3.176 $6m 0.05 5 0.03 0.0009132 $8.27 Annual Other Approach: Rely on mitigations already in place (reserve, AUFLS, and restoration plans) Post-event: OTA and WIR loads are likely to be lost. Average load for 3% of the time is 145 MW. Annual : Scenario Location Load Lost at GZ (MW) Unit ($/MWh) Fault Duration (h) Load Equipment Request Annual Base OTA & WIR 145 $25,000 3.176 $11.5m 0.013 0.03 0.0009132 $4.10 Sensitivity OTA & WIR 145 $25,000 3.176 $11.5m 0.05 5 0.03 0.0009132 $15.77

The classification recommendation from using the proposed methodology for the Otahuhu 110 kv bus is set out in the table below. Table 4 Classification recommendation for Otahuhu 110 kv bus Bus & Conditions OTA 110 kv (N-1) OTA 110 kv (N-1-1) Base Scenario Annual Sensitivity Scenario Annual Classification Extended Other Extended Other $919,800 $2,353 $4,490 $2.1m $9,051 $17,269 ECE* $840 $2.15 $4.1 $1,920 $8.27 $15.8 ECE* 19 * The classification proposed requires the availability of a SPS. Before classifying this as an ECE the system operator would need to consult with the grid owner on its expected operation, its suitability for this purpose and require an offer for the SPS use. Summer scenario was not analysed as summer peak load is expected to be lower than winter peak load.

4.2 HUNTLY 220 KV BUS Region: Hamilton Normal configuration: Two busbars HLY 220kV bus is a breaker and half station with 6 x 220 kv circuits; with everything in service a single bus section contingency will result in no loss of generation or circuits. There are no issues following a single bus section contingency. 20 Planned Outage: HLY 220kV busbar A Probability of bus faults in a year: 0.027 : Loss of HLY 220kV busbar B Average fault duration: 2.395 hours During a planned outage of busbar A all connections will be on busbar B. Busbar B contingency will result in only Huntly Unit 5 left generating via Drury-Huntly 220 kv circuit. Loss of local supply transformers T5 and T6 will result in Units 1, 2 and 6 tripping. 4.2.2.1 Winter Peak Scenario Assumptions: Winter Peak, Grid Zones 1 and 2 load = 2100 MW Post event: Loss of HLY busbar B and connected equipment: Loss of Huntly Units 1, 2 and 6 Loss of Huntly-Ohinewai 220 kv circuit 1 connected to Unit 2 Loss of Huntly-Ohinewai 220 kv circuit 2 connected to Unit 1 Loss of Huntly-Te Kowhai 220 kv circuit 1 Loss of Huntly-Stratford 220 kv circuit 1 Loss of Huntly-Takinini-Otahuhu 220 kv circuit 1 Loss of Huntly supply transformers T21 and T22 Loss of Huntly local supply transformers T5 and T6 Consequence: Constraint: Min generation: Voltage stability load limit of 1850 MW into Grid Zones 1 and 2 for winter and summer. From the Load Duration Curve, GZ 1 and 2 would exceed 1850 MW for 5% in 2015 so Load is 0.05. The average GZ1 and 2 loads above the voltage stability limit is 1961 MW. The demand response required is 111 MW (1961-1850) in GZ 1 and 2. This is the minimum generation required to survive an event if the event is not managed. However, there is no other significant generation in GZ 1 and 2. Analysis indicates the amount of Huntly Unit 5 generation is not as critical as the fact Unit 5 provides voltage support in the region. It is assumed there will be sufficient generation from Huntly Unit 5.

CE Approach: Constrain generation on and load off pre-event Pre-event measures: 111 MW demand response required. = 1 as this is managed pre-event. Post-event: Grid Zones 1 and 2 load is secured. Annual : Scenario Location Demand Response (MW) Unit ($/MWh) Equipment Request (h) Load Annual Base GZ 1 & 2 111 $700 8 1 0.05 $31,080 Sensitivity GZ 1 & 2 111 $1,600 4 8 1 0.05 $71,040 21 ECE Approach: Post-event load shedding Available mitigations: No post-event measures are currently available so this option was not considered. Other Approach: Rely on mitigations already in place (reserve, AUFLS, and restoration plans) Post-event: Grid Zones 1 and 2 loads are likely to be lost. Average load above the voltage stability limit is 1961 MW. Annual cost: Scenario Load Lost at GZ 1&2 (MW) Unit ($/MWh) Fault Duration (h) Load Equipment Request Annual Base 1961 $25,000 2.395 $117.4m 0.027 0.05 0.0009132 $145 Sensitivity 1961 $25,000 2.395 $117.4m 0.05 5 0.05 0.0009132 $268 4.2.2.2 Summer Peak Scenario Summer peak scenario is not an issue as summer peak load is below the 1850 MW voltage stability limit.

The classification recommendation from using the proposed methodology for the Huntly 220 kv bus is set out in the table below. Table 5 Classification recommendation for Huntly 220 kv bus 22 Bus & Conditions HLY 220 kv (N-1) HLY 220 kv (N-1-1) Base Scenario Annual Sensitivity Scenario Annual Classification Extended Other Extended Other - - - - - - Other $31,080 - $145 $71,040 - $268 Other

4.3 BUNNYTHORPE 220 KV BUSBAR Region: Bunnythorpe Normal configuration: Three busbars BPE 220KV bus has 9 x 220kV circuits and 3 x interconnecting transformers connected. These circuits and transformers are connected to 3 bus sections; busbar A1, A2 and B. With everything in service, busbar B normally has 4 x 220kV circuits and 1 interconnecting transformer connected. Loss of busbar B is the worst single bus section contingency. There are no issues following a single bus section contingency. 23 Planned Outage: BPE 220kV busbar B Probability of bus faults in a year: 0.027 : Loss of BPE 220kV busbar A1 Average fault duration: 2.395 hours During an outage of busbar B, all its connections will be moved to either busbar A1 or A2. For these scenarios most connections will be moved to busbar A1. 4.3.2.1 Winter Peak Scenario Assumptions: Winter Peak, GZ 8 load = 640 MW Loads at GXPs DVK, HWA, MHO, MTN, WDV, WGN, WPW, WVY = 148 MW. HVDC North Flow = 810 MW. Te Apiti generation = 0 MW. Post-event: Loss of BPE busbar A1 and connected equipment: Loss of Bunnythorpe-Wilton 220 kv circuit 1 Loss of Bunnythorpe- Linton 220 circuit 1 segment 1 Loss of Bunnythorpe-Paraparaumu-Haywards 220 kv circuit 1 Loss of Bunnythorpe-Brunswick 220kV circuit 1 Loss of Bunnythorpe-Tokaanu 220kV circuit 2 Loss of Bunnythorpe- Tangawai 220kV circuit 1 Loss of 220/110 kv interconnectors T1 and T3 Loss of Bunnythorpe supply transformer T10 Consequence: Constraint: Bunnythorpe 220/110 kv interconnector T2 overloaded From the Load Duration Curve, the GXPs would exceed 148 MW for 0.27% in 2015 so Load is 0.0027. Average load for this LRF is 152 MW. 30 MW generation at MHO, 20 MW generation at PTA and 5 MW demand response at GXPs above is required to relieve the overloading on BPE T2. Analysis of Generation Distribution Curves over 2015 indicates there is a 90% chance of 0 MW of generation at Mangahao and 0.5 MW at Patea. Min generation: A minimum generation of 30 MW at Mangahao and 20 MW at Patea is required to avoid overloading the 110 kv circuits and maintain voltages within limits should BPE T2 trip.

If this minimum generation is not available an average 152 MW of load will be lost. CE Approach: Constrain generation on and load off pre-event Pre-event measures: Constrained on 30 MW at Mangahao and 16 MW at Patea. 5 MW demand response required. = 1 as this is managed pre-event. Post-event: Bunnythorpe regional load is secured. 24 Annual : Scenario Location Generation or Demand Response (MW) Unit ($/MWh) Equipment Request (h) Load Base Mangahao 30 $20 8 1 0.0027 $13 Patea 19.5 $20 8 1 0.0027 $8 GXPs 5 $700 8 1 0.0027 $76 Total $97 Sensitivity Mangahao 30 $350 6 8 1 0.0027 $227 Patea 19.5 $389.71 6 8 1 0.0027 $164 GXPs 5 $1,600 4 8 1 0.0027 $173 Total $564 Annual ECE Approach: Post-event load shedding Available mitigations: No post-event measures are currently available so this option was not considered. Other Approach: Rely on mitigations already in place (reserve, AUFLS, and restoration plans) Post-event: Bunnythorpe 220/110 kv interconnector T2 will overload and trip. 152 MW of load will be lost as minimum generation is higher than base generation assumed for Mangahao and Patea. is obtained from bus fault statistics. Annual : Scenario Load Lost at GXPs (MW) Unit ($/MWh) Fault Duration (h) Load Equipment Request Annual Base 152 $25,000 2.395 $9.1m 0.027 0.0027 0.0009132 $0.61 Sensitivity 152 $25,000 2.395 $9.1m 0.05 5 0.0027 0.0009132 $1.12 4.3.2.2 Summer Peak Scenario Assumptions: Summer Peak, GZ 8 load = 475 MW Loads at GXPs DVK, HWA, MHO, MTN, WDV, WGN, WPW, WVY = 129 MW. HVDC North Flow = 630 MW. Te Apiti generation = 0 MW. 6 Sensitivity using highest offer price for generation response

Post-event: Loss of BPE busbar A1 and connected equipment: Loss of Bunnythorpe-Wilton 220 kv circuit 1 Loss of Bunnythorpe- Linton 220circuit 1 segment 1 Loss of Bunnythorpe-Paraparaumu-Haywards 220 kv circuit 1 Loss of Bunnythorpe-Brunswick 220kV circuit 1 Loss of Bunnythorpe-Tokaanu 220kV circuit 2 Loss of Bunnythorpe- Tangawai 220kV circuit 1 Loss of 220/110 kv interconnectors T1 and T3 Loss of Bunnythorpe supply transformer T10 Consequence: Constraint: Min generation: Bunnythorpe 220/110 kv interconnector T2 overloaded From the Load Duration Curve, the GXPs would exceed 129 MW for 5.69% in 2015 so Load is 0.0569. Average load for this Load is 136 MW. 30 MW generation at MHO and 20 MW generation at PTA is required to relieve the overloading on BPE T2. Analysis of Generation Distribution Curves over 2015 indicates there is a 90% chance of 0 MW of generation at Mangahao and 0.5 MW of generation at Patea. Therefore the cost analysis has assumed availability of a base generation of 0 MW at Mangahao and 0.5 MW at Patea. A minimum generation of 20 MW at Mangahao and 0 MW at Patea is required to avoid overloading the 110 kv circuits and maintain voltages within limits should BPE T2 trip. If this minimum generation is not available an average load of 136 MW will be lost. 25

CE Approach 1: Constrain generation and/or load pre-event Pre-event measures: Constrained on 30 MW at Mangahao and 19.5 MW at Patea. No demand response required. = 1, as this is managed pre-event. Post-event: Bunnythorpe regional load is secured. Annual : Scenario Location Generation or Demand Response (MW) Unit ($/MWh) Equipment Request (h) Load Annual 26 Base Mangahao 30 $20 8 1 0.0569 $273 Patea 19.5 $20 8 1 0.0569 $178 Total $451 Sensitivity Mangahao 30 $350 6 8 1 0.0569 $4,780 Patea 19.5 $389.71 6 8 1 0.0569 $3,459 Total $8,239 ECE Approach: Post-event load shedding Available mitigations: No post-event measures are currently available so this option was not considered. Other Approach: Rely on mitigations already in place (reserve, AUFLS, and restoration plans) Post-event: Bunnythorpe 220/110 kv interconnector T2 will overload and trip. 131/136 MW load will be lost as minimum generation is higher than base generation assumed for Mangahao and Patea. is obtained from bus fault statistics. Annual : Scenario Location Load Lost at GXPs (MW) Unit ($/MWh) Fault Duration (h) Load Equipment Request Annual Base GXPs 136 $25,000 2.395 $8.14m 0.027 0.0569 0.0009132 $11.42 Sensitivity GXPs 136 $25,000 2.395 $8.14m 0.05 5 0.0569 0.0009132 $21.16

The classification recommendation from using the proposed methodology for the Bunnythorpe 220 kv bus is set out in the table below. Table 6 Classification recommendation for Bunnythorpe 220 kv bus Bus & Conditions BPE 220 kv (N-1) BPE 220 kv (N-1-1) Season Base Scenario Annual Sensitivity Scenario Annual Classification Extended Other Extended Other All - - - - - - Other Winter $97 - $0.61 $564 - $1.12 Other Summer $451 - $11.42 $8,329 - $21.16 Other 27

4.4 ISLINGTON 220 KV BUSBAR Region: Christchurch Normal configuration: Six busbars ISL 220KV bus has 6 bus sections connecting 7 circuits and 3 interconnectors. Normal configuration is as follows; busbar A has ISL-TKB and ISL-T7, busbar B has ISL-LIV-1, busbar C has ASB-ISL-1 and ISL-T3, busbar D has ISL-WPR-CUL-KIK-3, ASB-BRY-1 and ISL-T6, busbar E has ISL-WPR- CUL-KIK-1 and busbar F has ISL-KIK-1. 28 There are no issues following a single bus section contingency. Planned Outage: ISL 220kV busbar A Probability of bus faults in a year: 0.027 : Loss of ISL 220kV busbar D Average fault duration: 2.395 hours During a planned outage of busbar A all its connections will be moved to busbar D. Busbar D contingency will result in loss of three circuits and two interconnectors. This can lead to the remaining interconnector overloading. Zone 3 (comprising of GZ 9 Nelson, GZ 10 Christchurch, GZ 11 Canterbury and GZ 12 West Coast) loads would need to be constrained to 830 MW to avoid overloading Islington T3. Possible Busbar Reconfiguration: Prior to busbar A outage, switch ISL T6 from busbar D to busbar B as this is possible via selectable disconnectors. Busbar D would still have three circuits but only one interconnector; this arrangement would allow Zone 3 load to increase to 1090 MW with no issues for a contingency of busbar D with generation in the area at 90th percentile. 4.4.2.1 Winter Peak Scenario Assumptions: Winter Peak, Zone 3 load = 1050 MW Post event: Loss of ISL busbar D and connected equipment: Loss of Islington-Waipara-Culverden-Kikiwa 220 kv circuit 3 Loss of Ashburton-Bromley 220 kv circuit 1 Loss of Islington-Tekapo b 220 kv circuit 1 Loss of Islington 220/66 kv T6 and T7 Consequence: Overloading of Islington 220/66 kv T3. Loss of this interconnector may result in loss of Zone 3. Constraint: Analysis of Generation Distribution Curves over 2015 indicates there is a 90% chance of having 18 MW at Coleridge and 0 MW at Cobb, Argyle and Kumara. Analysis of the 2015 Zone 3 Load Duration Curve indicated load would exceed 830 MW for 37% of the time so Load is 0.37. The average load is 912 MW above this limit.

If generation was constrained on to maximum (COL to 39 MW, COB To 32 MW, ARG to 11 MW and KUM/DIL to 10 MW) this would provide 74 MW of generation response. Demand response required is 8 MW (912 MW-830MW- 74MW). Min generation: This is not applicable. CE Approach: Constrain generation off and load off pre-event Pre-event measures: 8 MW demand response and 74 MW generation response required for 37% of time. = 1 as this is managed pre-event. Post-event: Zone 3 load is secured. Annual : Scenario Location Generation or Demand Response (MW) Unit ($/MWh) Equipment Request (h) Load Annual 29 Base Zone 3 load 8 $700 8 1 0.37 $16,576 Zone 3 gen constrained on 74 $20 8 1 0.37 $4,381 Total $20,957 Sensitivity Zone 3 load 8 $1,600 4 8 1 0.37 $37,888 Zone 3 gen 74 $250 6 8 1 0.37 $54,760 Total $92,648 ECE Approach: Post-event load shedding Available mitigations: No post-event measures are currently available so this option was not considered. Other Approach: Rely on mitigations already in place (reserve, AUFLS, and restoration plans) Post-event: Zone 3 loads are likely to be lost. Average load above limit is 912 MW for 37% of the time. Annual cost: Scenario Location Load Lost at GZ (MW) Unit ($/MWh) Fault Duration (h) Load Equipment Request Base Zone 3 load 912 $25,000 2.395 $54.6m 0.027 0.37 0.0009132 $498 Sensitivity Zone 3 load 912 $25,000 2.395 $54.6m 0.05 5 0.37 0.0009132 $922 Annual

The classification recommendation from using the proposed methodology for the Islington 220 kv bus is set out in the table below. Table 7 Classification recommendation for Islington 220 kv bus 30 Bus & Conditions ISL 220 kv (N-1) ISL 220 kv (N-1-1) Season Base Scenario Annual Sensitivity Scenario Annual Classification Extended Other Extended Other All - - - - - - Other Winter $20,957 - $498 $92,648 - $922 Other Summer scenario was not analysed as summer peak load is expected to be lower than winter peak load.

4.5 CLYDE 220 KV BUSBAR Region: Southland Normal configuration: Three busbars CYD 220KV bus has 3 bus sections connecting 4 generators and 4 circuits. Normal configuration is as follows; busbar A has CYD-ROX-1, CYD-G1, busbar B has CYD-TWZ-1, CYD-ROX-2, CYD-G3, CYD- G4 and busbar C has CYD-TWZ-2, CYD-G2. There are no issues following a single bus section contingency. 31 Planned Outage: CYD 220kV busbar A Probability of bus faults in a year: 0.027 : Loss of CYD 220kV busbar B Average fault duration: 2.395 hours During a planned outage of busbar A all its connections will be moved to busbar B. Busbar B contingency will result in loss of both CYD-ROX-1 and 2 and CYD-TWZ-1 and 3 generators at CYD. This can lead to NSY-ROX overloading. Loss of NSY-ROX may result in loss of Southland GZ14. NSY-ROX would need to be constrained to 230 MW to avoid overloading. Possible Busbar Reconfiguration: Prior to busbar A outage, switch CYD-ROX-2 from busbar B to busbar C as this is possible via selectable disconnectors. Ensure that both busbar B and C have two Clyde generators connected if possible. Busbar C would have CYD-ROX-2, CYD-TWZ-2 and 2 generators; this arrangement would have no issues for a contingency of busbar B. 4.5.2.1 Winter Peak Scenario Assumptions: Winter Peak, Grid Zones 14 load = 951 MW Post-event: Loss of CYD busbar B and connected equipment: Loss of Clyde-Roxburgh 220 kv circuit 1 Loss of Clyde-Roxburgh 220 kv circuit 2 Loss of Clyde-Twizel 220 kv circuit 1 Loss of Clyde Generators 1, 3 and 4 Consequence: Constraint: Min generation: Overloading of Naseby-Roxburgh 220 kv circuit 1. Loss of this circuit may result in loss of Southland GZ14. Analysis of the 2015 GZ 14 load and generation indicated that NSY-ROX loading would exceed 230 MW for 3% of the time when load is higher than generation. The average demand response required is 61 MW with a Load of 0.03. NSY-ROX loading would exceed 230 MW for 69% of the time when generation is higher than load. The average generation constrained off is 278 MW with a Load of 0.69. This is not applicable.